key: cord-0034080-eitceew2 authors: nan title: Abstracts of the contributed papers date: 1995 journal: Qual Life Res DOI: 10.1007/bf00518459 sha: 1629a4ff3c90eaa96bb66adf1374834272469ddf doc_id: 34080 cord_uid: eitceew2 nan While many specific instruments have been confronted with generic measures, few generic instruments have been compared head to head. Aiming for conceptual and practical clarification, we compared two generic health status instruments: the Nottingham Health Profile (NHP), and the SF-36 Health Survey . A cross-sectional evaluation of 321 male patients with stable chronic obstructive pulmonary disease (COPD), recruited from four hospital outpatients and one primary care ambulatory settings to represent the whole range of clinical disease severity, was carried out. Patients were clinically assessed by a pneumologist and administered, in random order, the NHP and the SF-36, among other questionnaires. Distribution of item responses and internal consistency Cronbach's a coefficients were calculated. A multi-trait multi-method correlation matrix was constructed with similar dimensions of the two questionnaires. Exploratory factor analysis and Rasch analysis were carried out to assess dimensionality, item ordering and scaling assumptions of the physical function subscales (NHP-Physical mobility and SF-36 Physical functioning). Percentage of missing items was similar for both instruments (range = 0.3%-1.9%). Reliability was higher than 0.7 in all dimensions except for NHP Social isolation (r = 0.52) and SF-36 Social functioning (r = 0.55). Cronbach's a for NHP dimensions were, in general, slightly lower than those for SF-36 dimensions. 'Ceiling effect' was higher for the NHP (range = 17%-62%) than for the SF-36 (range = 3%-54%) as well as 'Floor efiect' (NHP range = l%-16%) (SF-36 range = l%-3%). For both instruments, internal consistency coefficients were higher than correlation coefficients with other subscales. Correlations between the common NHP and SF-36 dimensions were higher than their correlations with other dimensions. A dyspnea scale was slightly more correlated with the SF-36 (-0.46 with Physical function, -0.46 with Vitality) than with the NHP (0.36 for Physical mobility, 0.43 for Energy). Correlations with the FEV, a clinical measure of disease severity, were higher for the SF-36 Physical function (-0.41 ) than with the NHP physical mobility (0.31), but lower with the SF-36 Vitality (-0.23) than with the NHP Energy (0.31). Factor analysis confirmed a unidimensional structure of both physical subscales, although items of the SF-36 Physical functioning explained a substantially higher proportion of the variance than those of the NHP Physical mobility (58% vs. 33%). NHP Physical mobility showed a wider item logits range (-3.2-4. 2) than the SF-36 Physical function (-2.8-3.7) . The NHP Physical mobility dimension showed a higher proportion of items with the expected fit. Also, match between raw score and the logit-based measure was found to be higher for the NHP Physical mobility. The comparison of characteristics of these two measures should be cautiously interpreted since they differ in several ways: conceptual basis, developmental approach, content, and scaling. Results suggest that the concepts that are ad-This research draws on two main bodies of literature relating to local notions of health and well-being-that of health-related quality of life (HRQOL) assessment and social anthropology. It is part of a programme designed to develop or adapt an instrument to be able to measure the impact of disease control programmes on the HRQOL of Kenyans. In the field of HRQOL assessment, issues surrounding language are increasingly being raised. HRQOL instruments are being translated into several languages and international instruments are being developed from the outset. Language is therefore predominantly being used as a medium to translate concepts of health developed in one context to another, or to spread the notion of universal definitions of health around the world. Curiously, at the same time, language has been portrayed as a barrier to the development of instruments as certain items in questionnaires do not translate easily or well. From a social anthropologist's perspective, language is seen as more fundamental to the understanding of concepts of health and well-being. It is seen as a means of gaining an insight into locally specific ways of thinking and feeling. It is interesting to note that the field of Bible translation validates this thinking by requiring missionaries to not only learn the language, but also to live in a new community for several years before attempting any translation. To date, local languages themselves have not been used in the HRQOL field as a means of exploring local definitions and perceptions of health. Language has simply been subsumed into the general approach of the HRQOL field which seeks to explain models of health as a set of universally held values. Even where research groups depart from such explicit views, research methods used tend to rely on a relatively structured approach to asking respondents about their beliefs e.g., people are asked what good/bad health is or how quality of life is described, as a direct question. Part of our initial research strategy in Kenya was to follow the broad approach taken in the development of generic measures of HRQOL based, at least partially, on lay perceptions of HRQOL i.e. SIP, NHP and WHOQOL. We used semi-structured interviews and focus group discussions to ask a series of questions around the definitions of good and bad health, whilst proceeding with participant observation fieldwork. During one of a series of meetings to discuss the findings of these two approaches in different areas, we discovered that words used to describe 'health ', 'well-being' and 'affliction' as well as words which described individuality appear to be significantly different in the two locations. This was not only with respect to the issues that have been raised in the translation of HRQOL instruments, but with respect to more fundamental assumptions of locally specific philosophies. For example, the local resonances of words associated with health appear to be crucially different, and potentially to contradict expectations. In Makueni, where levels of formal education are relatively low, our questions have tended to elicit issues of health which are essentially connected with bodily wellbeing. Spiritual and interpersonal concerns, which appear to be of enormous importance, only tend to arise when asking open ended questions when discussing relationships, and do not appear to be understood directly with the local term generally used for the English word 'health'. In contrast, in Vihiga where there is greater education and exposure to biomedical interventions, the local word for 'health' elicits a wide range of responses relating to disease as well as much more general issues about life. or a heavy smoker. Patients who confessed to smoking tobacco also reported lower levels of concurrent QQL, across a broad spectrum of its interrelated aspects, with particular emphasis on vitality, vigor, life satisfaction, restful sleep, and optimism. The more they smoked, the worse their QOL. Age and marital status were associated with QOL in somewhat confounding ways, but even after adjusting for these factors, the smoking effect remained strong and negative. No dimension of QOL appeared to benefit from smoking-nor to characterize smokers as happier than nonsmokers. In this population, either smoking makes people unhappy, or unhappy people tend disproportionately to smoke, or perhaps both. This picture stands in sharp contrast to cultural images which associate smoking with vivacity, contentment, relaxation, energy and happiness. not gained much attention in the literature. The aim of this study was to analyze the perception of distress in leg ulcer patients. Sample: Seventy-eight leg ulcer patients were identified in community nursing in one Norwegian local authority during a three month period. Sixty-two patients (80%) were willing to participate in the study. 63% of the sample were women. Mean age was 77 years (range = 30-92 years). Forty-six patients had suffered from leg ulcer less than 5 years and 16 patients have suffered from leg ulcer more than 5 years. The respondents were interviewed by the researcher using the Nottingham Health Profile (NHP) part I (38 items) measuring perceived distress in relation to physical mobility, pain, energy level, emotional reactions, sleep and social isolation. Higher scores (optimal 100) indicate more limitations. Cronbach's a for the different domains ranged from 0.56-0.81 for the social isolation and pain domain respectively indicating satisfactory reliability. Correlation analysis showed significant relations between the different domains (absolute values of the correlation coefficient ranged from 0.26-0.63, p < 0.05) indicating construct validity. The profile of the NHP scores ranged from lo-37 with less limitations in the emotional and social domain and most limitations in the physical domain. The results compared to an age matched reference group showed that leg ulcer patients perceived most negative impact in relation to the domains of physical mobility (37 vs. 13) and pain (26 vs. 13 ). Patients suffering from leg ulcer more than five years report more problems than patients suffering less than 5 years (total score 31 vs. 19, p = 0.047). Frequency of social contact with family and friends was significant associated with perception of less distress (total score 27 vs. 16, p = 0.03). Patients living together with a significant other also report fewer problems compared to patients who live alone (total score 27 vs. 15, p = 0.013). In conclusion, leg ulcer patients perceive limitations especially in relation to physical mobility and pain. Suffering from leg ulcer over time seems to increase perceived distress. Social network plays an important role in the perception of HRQOL among the patients. The Short Form 36 (SF-36) survey is used in most studies to provide a common measure of health-related quality of life (HRQQL). Formal evaluation of measures of HRQOL, including the SF-36, are part of the core activities of the Pepper Center. Among relatively healthy older adults these studies include: (1) a comparison of the Sickness Impact Profile (SIP), the Quality of Well-Being (QWB) and three scales of the SF-36; (2) a comparison of the SIP, full SF-36, and 'disability days' measures; and (3) a mailed test-retest of the SF-36. Also underway is a formal study of the validity and test-retest reliability of the SF-36 for relatively non-cognitively impaired nursing home residents. The specific goals of this series of studies are: (1) to test several psychometric properties of these instruments including similarities among dimensions and composite scores of the instruments, (2) to look for possible scale limitations, specifically 'floor' or 'ceiling' effects, and (3) to determine if there are practical administration and response problems among the instruments for specific categories of older adults. To date, the SF-36 has performed relatively well among community-dwelling research subjects. There have been some response problems using a mailed survey; we achieved 60% response with a postcard reminder and twotime mailing, and some scales could not be scored due to missing items (2%-S% of scales). A telephone reminder has been more successful at getting response to a mailed SF-36 and SIP packet (88% response). SF-36 scales measuring physical health and function may suffer from 'ceiling' effects for relatively healthy subjects. The SIP is clearly inappropriate as a measure of HRQOL for the same reason; in one study over one-quarter of subjects received 'perfect' (0%) health scores and the total SIP mean was 3.4% (on a In this report we present the development and outcome of a pilot study carried out with a group of 26 end-stage renal disease patients, 13 of whom were treated with Continuous Ambulatory Peritoneal Dialysis (CAPD) and nine of them treated with In centre Hemodialysis (In centre HD). Both groups were homogeneous regarding sex, age and period on dialysis. This sample belongs as a whole to Galdakao Hospital Collective (Bizkaia, Spain). This project was mainly aimed to give a complete answer to two essential questions: (1) Is there any difference in anxiety, depression and perceived health and objective health in those patients treated with In centre HD and CAPD? (2) If so, what is the tendency of these differences? We have taken measures from the dependent variables previously pointed out through medical reports, personal interviews with each patient and the following structured self-reports: Spielberger State Trait Anxiety Inventory (STAI); Beck Depression Inventory; Nottingham Health Profile. The application of these instruments lasted for an hour session at the hospital. Except in those cases dealing with sight or reading problems, all the self-reports were filled in by the patients themselves. The results that were found on the questions raised are as follows: (1) Health perception is superior in patients treated with CAPD in the following areas: physical mobility, pain, sleep and energy, in fact, showing a statistically significant difference from that of the In centre HD group. (2) The scores in depression are superior in patients treated with In centre HD, though the difference shown is not statistically significant. (3) At the anxiety variable no difference between groups was shown at all. (4) Finally, there was no difference in the biological variables (objective health), though the In centre HD group showed a higher number of symptoms. 12-15 and 8-11 years. The structure of the sixteen-dimensional measure of HRQOL (16D) for children aged 12-15 introduced here is based on the 15D measure developed by Sintonen et nl. for adults. Specialists in pediatrics, child psychiatry, adolescence psychiatry, pediatric neurology and health economics participated in developing the 16D. Based on the suggestions of the expert group and pilot testing, some of the questions of the 15D measure were deleted, others added or reformulated age-appropriate. In its final form, the 16D measure consists of 16 multiple choice questions, each representing one health-related dimension. The subject marks the level (l-5) of each dimension best describing his or her present health status. The 16D score representing the overall HRQOL and ranging from 0 (worst possible) to 1 (best possible)-is calculated by combining the mean dimension importance weights and within-dimension level desirability values obtained from children aged 12-15 attending primary school with the subject's own assessments of his or her current health status. The 16D questionnaire has been completed by several patient groups as well as healthy children aged 12-15 years. The children found 16D easy to use; filling out the questionnaire took 5-10 minutes. As an evidence of construct validity, the HRQOL profiles of the patients differed according to the diagnosis. For example, the 16D score of the five patients from Children's Hospital, University of Helsinki, currently waiting for organ transplantation was significantly lower (mean 0.874, SD 0.076) than that of the 239 healthy controls (mean 0.947, SD 0.049) (p = 0.02, Mann-Whitney U-test). The controls-children attending primary school-had a significantly better status (less problems) on the dimensions of school and hobbies, friends, mobility, eating and elimination than the patients (p < 0.001). Similarly, 21 patients aged 12-15 with skeletal dysplasias had a significantly lower status on the dimensions of school and hobbies, friends and mobility than their controls (p < 0.001). Further, these patients were less satisfied with their physical appearance than the controls (p = 0.03). To assess the reliability of the 16D questionnaire over time, 35 healthy children aged 12-14 completed the 16D twice one Abstracts of the Contributed Papers week apart (test-retest). The repeatability coefficient (the percentage of cases lying within two standard deviations from the mean difference) for the 16D score was 91%. In a similar manner, a HRQOL measure for children aged 8-11 is being validated. However, the questionnaire, the structure of which is similar to 16D, must be administered by a trained interviewer in this age group. We conclude that the subjective HRQOL of school-aged children can be assessed. Further, the 16D measure seems useful in this assessment, even if further studies on aspects of validity and reliability such as sensitivity to change should be performed. Factors associated with quality of life among persons with multiple sclerosis K. J. Aronson Queen's University, and the Multiple Sclerosis Society of Canada Enabling persons with multiple sclerosis (MS) to enhance their quality of life is part of the mission statement of the Multiple Sclerosis Society of Canada. The objective of this study was to determine satisfaction with quality of life among persons with MS, and to determine the relationships between this and demographic, economic and health factors. A cross-sectional survey throughout Ontario was conducted using self-completed mailed questionnaires. Included was a five-point rating scale for satisfaction with quality of life as a whole and six components, including health. A response rate of 83% was obtained, and 662 subjects with MS are included in this analysis. Subjects' mean age is 48 years, 70% are female, and their mean age at diagnosis of MS is 36 years. Fatigue and walking problems are reported as the most prevalent and distressing symptoms. Some type of aid or a wheelchair is required by most subjects for getting around. Friendships receive the highest satisfaction rating, closely followed by family relationships and housing. The lowest rating is for health. In comparison to results from a Statistics Canada survey which asked the same questions, those with MS are less satisfied to a statistically significant degree with health, job or major activity, and life as a whole than are ablebodied and disabled persons in the general population, and they are more satisfied to a statistically significant degree with housing and family relations than are the general population disabled. Quality of life as a whole is very satisfactory for 28% of those with MS. Demographic characteristics such as age and education level are not associated with quality of life, while unemployment and not being in the labour force are associated with decreased quality of life. Any level of household income over $15,000 is associated with increased satisfaction with quality of life. In terms of health factors, there is no apparent association with time since diagnosis of MS. Any severity of MS symptoms worse than mild, and any mobility restriction, are associated with decreased sat-isfaction with quality of life. Experiencing fatigue, and a current disease course or relapsing-remitting/progressive or progressive, are also associated with decreased quality of life. Interference by MS with social activities shows a particularly strong relationship, with increasing interference associated with decreased satisfaction with quality of life as a whole. Considering all variables simultaneously through multivariate regression, the factors most strongly associated with less satisfaction with quality of life are interference by MS in social activities and an attitude other than acceptance towards having MS. Intervention in these areas may help enable those with MS to enhance their quality of life. Constructing a health-related quality of life questionnaire for Italian peptic patients Despite peptic diseases (PD) being a common complaint, in Italy there is not any suitable questionnaire for assessing the Health Related Quality of Life (HRQOL) of PD patients; for this reason we planned to construct a specific questionnaire. Three focus groups were assembled in three different towns and, during each meeting 15 patients were invited to discuss their disease on the basis of a very preliminary questionnaire, made during a previous meeting with a panel of gastroenterologists. Finally, 48 items attributable to six domains (anxiety, daily living, food, pain, social relations, symptoms) were obtained. The process of validation was carried out through an Italian multicentric study called Herqules (Health Related Quality of Life Evaluation Status). One hundred and seventy digestive endoscopists administered the questionnaire, called QPD48, to 1,772 patients. Results reveal the powerful emphases and functions of disease-specific questionnaires. Frequency distributions were calculated for all 48 items, after recoding item one to produce a scalar numerical categorization of reported frequency of stomach upset. Reliability analyses were performed on all 48 items and on a number of subscales. The reliability of the trial 48 item scale was 0.89. Several items showed low item total correlation (they were not contributing meaningful variance to what the QPD was measuring as a whole), so they were removed and a total of 30 items were retained for standardization. Reliability analysis on the 30 remaining items revealed an increase Since item one provides a built-in concurrent validity opportunity because it asks for self report of the frequency of stomach trouble, a one-way analysis of variance was conducted on each of the three subscale scores and on the total QPD30 score. Highly significant differences associated with attack frequency and substantial effect sizes ranging from 0.46-1.27 of a standard deviation were found. The effect sizes offer conclusive evidence of the concurrent validity of our questionnaire. We consider our aim of developing a valid instrument for the assessment of QOL in gastric patients to be largely realized. consists of 40 items measuring 10 different domains-mental health, physical functioning, physical health, cognitive functioning, social functioning, social support, financial status, partner intimacy, sexual functioning, and medical care. These domains were derived from indepth personal interviews with HIV-infected men and women and refined through subsequent psychometric testing. Each domain is measured by four questionnaire items selected on the basis of associations with established scales for that domain. In a sample of 121 HIV+ clients recruited from two clinical sites, Pearson correlations between domain scores and criterion scales ranged from 0.60-0.82. Internal consistency, measured by Cronbach's CI, was 0.74 or higher for eight of the ten domains. An overall index of Quality of Life (QOL) may be computed from the mental health, physical functioning and social functioning domain scores. This weighted index was derived from three global QOL indicators (Cantril's Ladder of Life, Campbell's Life Satisfaction item, and a visual analogue scale). In the clinic sample, MQOL-HIV index scores were strongly related to Whalen's HIV Symptom Index (r = -0.64, p < 0.001). There was a substantial amount of variation within CD4+ count ranges, suggesting that the MQOL-HIV is a useful supplement to measures of immune status. The MQOL-HIV thus serves as a short, easily administered multi-domain measure of quality of life that addresses issues especially relevant for people with HIV or AIDS. Is there any health state worse than death? X. Badia, A. D. Prieto, M. Rue and A. Seaura Institut de Salut Publica de Catalunya, Hospital de Bellvitge, Hospital Part Tauli, Barcelona, Spain Our objective was to discover if there are differences on the assessment of health states, as they are described in the EuroQoL (EQ), using valuations from healthy individuals and patients with different conditions. The health states of the EQ are classified as five dimensions of health: (mobility, self-care, usual activities, pain/discomfort and anxiety/ depression) each one with three levels of severity. The levels of the dimensions are codified as '1' if the individual has a perfect health state, '2' if he/she has moderate health problems and '3' if he/she has many health problems in each dimension. This classification generates 35 = 243 possible health states. We obtained values of a set of 15 health states including the best health state (lllll), the worst health state (33333), unconscious and death. The assessment was made on a Visual Analogue Scale (VAS) by 360 non patients, 240 outpatients, 120 severe chronic patients (diabetes, EPOC, neoplasia) and 100 intensive care unit patients. A bivariate and a multivariate analysis was performed in order to analyze the differences between values assigned to health states. The general hypothesis of the study was that individuals assess in a different way the hypothetic health states depending on their own health state. The specific hypothesis was that there are not health states worse than death in individuals with severe conditions in opposition to healthy individuals. The multivariant analysis showed that the values assigned to health states are affected basically by the seriousness of the condition. Age and education were only associated to 3 of the 15 health states values. Nevertheless, the R* was low (0.05-0.56). Death was the worse state as valued by critical patients and patients with a severe condition. The state unconscious and 33333 was valued as worse than death by healthy individuals. The results suggest that individuals tend to give a higher value to those health states familiar to them. The inclusion of health states worse than death in a preference-based instrument would be only justified when there will be taken decisions about the general population but not in populations with a certain degree of illness because preferences may change. Therefore, different tariffs from standardized health indexes should be generated. between glycemic control, diabetes-related symptoms, and SF-36 scales scores in patients with non-insulin dependent diabetes mellitus glycemic control was associated with better functioning. All six of the diabetes-related symptom scales are significantly associated (p < 0.001) with each of the eight SF-36 scales at baseline. The magnitudes of the correlations range from 0.17-0.58 with most of the correlations greater than 0.3. All correlations are negative indicating that more severe symptoms are associated with worse functioning on the SF-36 scales. These results support the conceptual model, emphasize the Importance of measuring diabetes-related symptoms, and support the validity of the quality of life measures that were used in this study. These analyses were conducted to verify a conceptual model in which treatment would improve glycemic control, alleviate symptoms and, in turn, improve quality of life. A total of 792 patients with non-insulin dependent diabetes mellitus participated in an intervention trial. Results are based on assessments at baseline and change from baseline at 12 weeks and are not stratified by treatment. Three measures assessed glycemic control: Fasting serum glucose, fructosamine and hemoglobin Ais Quality of life was assessed with the acuteversion of the SF-36 Health Survey. Aspart of an effort to develop a set of diabetesrelated symptom scales, patients rated each of 23 symptoms by checking a box if they did not experience the symptom during the last 7 days and by circling a number on a 7-point scale to indicate the extent to which they were bothered by each experienced symptom. The symptoms were selected because they were considered to be related to hypoglycemia and hyperglycemia. Data for each symptom were converted to an 8-point scale. Five symptoms were dropped from scale development because they were redundant with SE-36 items. A principal components analysis with an orthogonal rotation was used to group the 18 remaining symptoms into 6 scales: cognitive/awareness (difficulty concentrating or thinking, confusion or disorientation, bluny vision, headache, dizziness or lightheadedness, apathy or indifference), nervous response (lack of appetite, tremors or shaking, heart pounding, pimples or boils), metabolic (excessive appetite, excessive thirst, frequent urination), sexual interest/problems (lack of interest in sex, problems with sexual function), sleep (problems sleeping well, daytime sleepiness), and itching/burning. Relationships between measures were assessed with Pearson product-moment correlations. Four hundred and seventy-two men and 320 women with a mean age of 58 years entered the study The mean body mass index was 31.1 kg/m* the mean duration of diabetes was 6.9 years, and 75.6% were white. The distribution of treatment screening for randomized patients was diet only, 15.8%; oral medications, 76.9%, insulin with or without oral medication, 5.8%; and none, 1.4%. Mean FSG and HbAi, levels at baseline were 247 mg/dL and 9.3% respectively Correlations between the three measures of glycemic control and the six diabetes-related svmptom scales were assessed at baseline and in terms of change from baseline. Nineteen of these 36 (3 x 6 x 2) correlations are significant (p < 0.05) and all of the significant correlations indicate that poorer glycemic control is associated with symptoms that are more bothersome. Correlations between the three measures of glycemic control and the eight SF-36 scales also were assessed at baseline and in terms of change from baseline. Ten of these 48 correlations were significant and all of the significant correlations Indicate that improved Relating spontaneous adverse experience reports to scores on a questionnaire querying tolerability L. Barber and N. C. Santanello Merck Research Laboratories, West Point, PA, USA In this study, we examined the mlation between spontaneously reported adverse experiences and responses given on the Comparison of Ophthalmic Medications for Tolerability (COMTOL) checklist questionnaire which queries the frequency and bother of specific side effects known to be associated with topical ophthalmic agents used to treat ocular hypertension or open-angle glaucoma, and the impact that the side effects have on health-related quality of life. The study was a 4week, randomized, open-label, twoperiod cross-over clinical trial comparing 2% dorzolamide given twice a day (b.i.d.) and 2% pilocarpine given four times a day (q.i.d) in 92 patients who were also receiving 0.5% timolol b.i.d. for the treatment of ocular hypertension or open-angle glaucoma. Patients completed the COMTOL questionnaire at baseline and at the end of each period and spontaneous reports of adverse experiences (AEs) were collected throughout the study by the investigator. Since there were only three spontaneously reported AEs related to drug treatment while patients received dorzolamide and since COMTOL scores indicated a low level of side effects, the analyses were limited to pilocarpine treatment periods. We discovered that during the pilocarpine treatment periods, a large percentage (94%) of the 47 patients who failed to spontaneously report any adverse experiences, indicated on the COMTOL that they had experienced side effects. Specifically, on the COMTOL, 89% of the 47 indicated that they had experienced ocular symptoms. 70% of the 47 indicated they had experienced vision difficulties, 43% of the 47 indicated they had experienced accommodation difficulties, and 26% of the 47 indicated that they had experienced browache. These discrepancies between the methods of spontaneous reports and a checklist questionnaire are similar to those previously reported in the literature for other drugs. Unlike previous literature, we went beyond identifying discrepancies with the two reporting methods and we looked for possible explanations for why the discrepancies Quality of Lifp Reswch Vol4.1995 exist. We observed that patients who spontaneously reported AEs expressed more bother from these specific side effects on the questionnaire than patients who did not spontaneously report AEs. As well, patients who spontaneously reported AEs and discontinued the drug as a result of the AEs expressed on the COMTOL the greatest bother from side effects. For example, the Bothered by Accommodation Difficulties Domain Scores were 1.04 ('bothered a little') for the group that did not spontaneously report any AEs, vs. 1.97 ('bothered some') for the group who reported AEs but did not discontinue, vs. 2.86 ('bothered very much") for the group who reported AEs and discontinued. This trend of increasing negative impact (as patients reported AEs and discontinued) was also observed with COMTOL global question scores on the impact of side effects on health-related quality of life, the impact of activity limitations on quality of life, satisfaction with medication and compliance with medication Therefore, spontaneous reporting of side effects appears to be detecting the most clinically meaningful side effects. By integrating the psychological, sociological and medical literature on QOL, we developed a model of QOL consisting of three main components: cognitive, affective and behavioural. Although each component is in itself important, the relationships between the components are an essential aspect of QOL as it is experienced in daily life. For example, satisfaction (cognitive component) and mood state (affective component) may vary in relation to the activity engaged in (behavioural component). In the context of a clinical trial, the illness and its treatment could influence different aspects of QOL as conceptualized in this model. This model may be particularly useful in depression because of the known effects of this illness on all three components: cognitive deficits, low affect and impaired behavioural functioning. Although instruments have been developed specifically for measuring QOL in depression, they seldom evaluate all three components. In addition, little is known about the impact of depression on QOL in the context of daily activity and subjective experience. In order to assess the impact of pharmacological treatment on daily life QOL, a time-sampling method (Experience Sampling Method, ESM) was used in a clinical trial. Twenty-one depressed outpatients completed ESM self-reports during daily activities 10 times a day for 6 days, before and again after 6 weeks of treatment with either fluvoxamine or amitriptyline. Results showed changes not only in the separate components of QOL (e.g. increase in positive affect, increase in concentration), but also in the interrelationships between the components (e.g. increase in positive affect during productive vs. non-productive activities, decrease in negative affect in alone vs. not alone situations) following treatment. ESM monitoring may thus be a valuable supplement to more traditional questionnaire methods of assessing QOL in clinical trials. The results are currently being replicated in a larger, placebo-controlled study including a group of normal controls. The cognitive foundations of quality of life assessments (QOLA) I. Barofsky The Johns Hopkins University School of Medicine, Baltimore, MD 21224, USA A QOLA is an activity that we all can do every day. The single item global QOLA is a prototype of this cognitive process. Studies comparing global with multi-attribute QOLA's suggest a high degree of correlation. In addition, the available data suggests that global QOLA's such as the self-assessed health status question, involve an active process of selecting information that is used to generate the judgment. Examination of the phenomenology of QOLA indicates that the following processes are occurring: descriptions of the state a person is in, scaling of these states, statement of the preferences for being in a state, and finally aggregation of preferences to produce an index. Furberg and Kane have shown that all the available QOLA's can be classified as either including a preference system or not, and providing aggregation rules that permit the calculation of an index or not. This section will conclude with the statement that in most instances the available instruments do not reflect the phenomenology of a QOLA, which raises serious ethical and empirical concerns. A second question that can be asked is if available QOLA do not reflect what a person would do themselves, would it make a difference in terms of the information that a QOLA generated? One approach to answering this question is to compare available QOLA's to determine if they generate comparable information. A number of studies which have concurrently administered several QOLA's will be Abstracts of the Gmtrzhted Papers scales discriminated between levels of self-rated overall health. Finally, standardized response means (SRM) were calculated to determine the responsiveness of each instrument, to recovery (baseline to one year follow up). There were 445 men and 487 women in the cohort of 932 injured workers with mean age of 39 years. Five hundred and thirty-five had low back pain, 250 had upper limb conditions and 147 lower extremity conditions. Baseline distributions of disease-specific measures revealed more dysfunction than did the SF-36 Acute. However, the SF-36 showed less dysfunction in the upper extremity group (physical functioning scores in figure 1 ). Spearman rank correlations with pain and functional limitation scores were higher for disease specific measures (0.18 64 years old, 50% were male. The primary cause of kidney disease was interstitial (28%), followed by unknown cause (28%), hypertension (18%), diabetes (14%) and glomerulonephritis (12% The aim of this study was to assess the level of psychological distress and the quality of life in climateric patients. All patients (n = 30) who sought treatment in the menopausal clinic of the University Hospital of Oviedo, Spain between January and March 1995 were included in the study. The interviews included an 'ad hoc' sociodemographic and clinical questionnaire, and the Spanish versions of the SF-36 Health Survey (SF-36) and the Psychological General Well-Being Index (PGWB index). Descriptive statistics were used to describe the sample. Comparisons of psychological distress and quality of life measures were made using the t-test. The mean age was 50.4k4.4 (maximum = 61 years, minimum = 37 years), 66.6% were married, 10% were widowed and 23.3% were single. On the PGWB index, 46.6% of the patients were identified as 'psychologically distressed' (score > 60 The objective of this study was to obtain population norms on pertinent domains of quality of life and health-related quality of life, and the relative importance of these domains to the public in general and to people with specific longstanding illnesses and conditions. The OPCS Omnibus Survey was used as the vehicle for the study. This is a monthly survey of a random sample of 2,000 adult members of the population in Great Britain, based on a random sample of households. The data was collected for a one month period. The response rate to the survey was 77%. Main measures were open ended and coded responses to questions eliciting the five most important areas of life (and priority ordering) and the five most important areas of life tures. The symptoms (e.g. anxiety, phobic avoidance) associated with panic disorder can be as debilitating as the actual panic attacks. Therefore, a treatment's efficacy at reducing not only the number of panic attacks, but also the symptoms associated with panic disorder should be considered. The purpose of our study was to pilot test the feasibility, validity and reliability of a daily patient diary of panic symptoms. We used an adaptation of the Q-TWiST concept to evaluate patients in Phase III clinical trials comparing fluvoxamine to placebo. Patients recorded their daily feelings of panic, anxiety, phobic avoidance, sickness, and general well-being. We analyzed the proportion of time patients spent in each of these health states. We found that diaries captured a portion of patient well-being that is not reflected in the number of attacks or the standard scales used to measure quality of life. The patient diaries were also more sensitive to the number of panic attacks (Pearson's T = 0.74, p < 0.001) than the standard physician-rated scales currently used. In addition, we found that while fluvoxamine compared to placebo had no additional effect on the number of attacks experienced by the patients, it did provide a substantial benefit by reducing patients' symptoms. Fluvoxamine-treated patients had a 25% improvement in well-being over patients treated with placebo. Without our measure of symptomatic time, the efficacy of the drug would be questionable (or rejected) because the number of panic attacks did not differ between the groups. We recommend that future studies adopt a similar strategy to assess treatments' efficacy at reducing the efects of panic disorder. affected by reported longstanding illnesses or conditions (and priority ordering). have not yet been seen as determinants to health-related quality of life. We report results of a longitudinal epidemio-logical study into well-being and function as well as perceived environmental conditions in over 500 teenagers (age 13-15) and their parents, living in airplane-noise exposed and non-exposed urban and rural regions of southern Germany. Assessments are perforrned via telephone interviews, using the 12-items form of the SF-36 general health survey as well as newly developed scales to assess perceptions of environmental conditions. Furthermore instruments to assess perceived locus of control over own health and over enviromnental conditions are included. Analysis of the first wave of assessments will examine if noise exposed teenagers and parents as compared to non-exposed peers, suffer from a decreased well-being especially in terms of mental health which might be dependent on enviromnental annoyance and might be mediated by environmental as well as health locus of control. Results will determine whether and how a community based health intervention programrne can be carried out to alleviate potential negative impacts of airplane-noise, with effect of such intervention being tested in the second assessment wave of the study. Planning quality of life assessments in a clinical trail often involves knowledge about performance of measures especially as concerns the use of generic vs. disease-specific measures. We report the inclusion of generic (SF-36; NHP) as well as a newly developed disease-specific measure (PAVK 86) in a pilot study involving 308 inpatients and outpatients suffering from peripheral arterial occlusive disease (PAOD; Fontaine stages II to IV). Questionnaires were filled out by patients before and one week into surgical or medical treatment with clinical data collected at both measure points. Psychometric analysis at baseline yielded good scale structure indices and reliability for all questionnaires tested as well as validity in terms of differentiation according to clinical severity groups as well as measures intercorrelations. In patients with treadmill tested improved performance all questionnaires detected change over time. The relative precission of the PAVK 86 questionnaire however was higher in several dimensions assessed as were SF-36 or NHI? Results of the present study suggest that psychometric performance of generic and diseasespecific measures in the PAOD patient population are comparable. The disease-specific measure PAVK 86 however is slightly superior in identifying treatment-related change over time. 80 articles could be labelled either experimental studies of specific strategies to improve QOL or results of clinical programmes to enhance QOL. The most common patient groups for nursing studies were community living elderly, nursing home elderly, and cancer patients. The Life Satisfaction Inventory was the most commonly used Instrument for measuring QOL. Of the QOL instruments developed by nurses, only the QOLI (Padilla) was used in more than one study. Specific strategies used to improve QOL included special nursing care, reminiscence, education, exercise, and cognitive approaches. About 50% of the studies yielded statistically significant results. We conclude that, although increased attention is being paid to QOL by nurses, nursing lacks sufficient experimental studies to guide decisions regarding appropriate nursing strategies to improve QOL. We found many experimental studies of nursing strategies that could have used QOL as a valid endpoint but did not. We also noted that nursing research studies seldom used validated QOL instruments developed outside of nursing. The QALY's (quality-adjusted-life years) models have been increasingly suggested to be a useful tool to decision-making public resources allocation in the health sector through the cost-utility approach. Unlike the usual QALY models (based on Rosser and Kind, QWB indexes) which multiply the utility of being in a certain health state by the corresponding life expectancy, excluding therefore the integration of preference for quantity of life, we will focus on the concept of a QALY which would consider preferences for both quality and survival components (Miyamoto and Eraker), expressing trade-off between these dimensions. Based on Keeney and Raiffa's multi-attribute utility theory, our first purpose will be to determine the set of very restrictive conditions which allows such a QALY to modelize the consequences of a health action according to the economic concept of utility measurement of lenglh of survival and of the health attributes (quality of life). We will focus more specifically on the limits that come from the application of the multi-attribute approach to the health scope when taking as example of the health attributes, the HUI (Health Status Index) Mark III classification ones (McMaster University). In a second part, we will discuss the methodological and empirical limits that any QALYs model raises, due to the approximation of the human judgement process they represent, to the preference eliciting methods they mobilize, to the presentation and the construction of situations in which preferences are elicited, to the choice of health states taken as reference, to the level of information given to the individual. The purpose of this research study was to examine the relative impact of health status, meaning of illness events, recurrence, cancer related experiences and uncertainly of illness upon adjustment to disease progression in women with breast cancer. It was hypothezied that meaning of illness would be the most powerful influencing variable and that the relationship between adlustment to breast cancer and meaning of illness would be altered as events of disease progression, as defined by recurrence(s) were experienced. A cross-sectional, exploratory design was used. One hundred and eighty-two women receiving treatment for breast cancer who were newly diagnosed, had a first recurrence or two or more recurrences of breast cancer were accrued from five cancer treatment centres in two Canadian provinces. Participants completed four questionnaires; the Medical Outcomes Study Short Form 20(SF-20), the Meaning of Illness Questionnaire, the Mishel Uncertainty in Illness Scale, and the Psychosocial Adjustment to Illness Questionnaire. Demographic data and disease and treatment-specific clinical data were also collected. Forty-eight variables were used in the analysis. Modelling techniques of multiple regression analyses, analysis of covariance and analysis of variance (ANOVA) were employed. Four variables explained adjustment: perceived physical functioning, meanmg of illness, perceived mental well-being and uncertainty. The physical functioning variable was the best predictor of variation in adjustment. Adlustment was not related to disease progression from a new diagnosis to recurrence(s). The levels of adjustment were strikingly similar for the women who were receiving treatment either at the time of diagnosis or for recurrence(s We explored differences by race and gender in ratings of importance of quality of life (QOL) domains assigned by patients undergoing hemodialysis (HD) or peritoneal dialysis (I'D) in developing a disease and treatment-specific instrument for end-stage renal disease (ESRD). Patients were eligible if they were over age 18 and undergoing dialysis at one of four centres in Baltimore or Boston. Frequency and severity (bother) of 20 QOL domains were assessed: sleep; time; role; social; cognitive; physical and sexual function; work; body image; emotional well-being; energy; freedom; general health; quality of life; recreation; finances; diet; dialysis access; pain and other symptoms. Response categories were from 1-5 in magnitude of frequency or severity. For each domain, we used logistic regression analyses to test the independent association of race and gender on either frequency or severity adjusting for potential confounders such as age, dialysis modality (HD vs. I'D), education, months on dialysis, living situation and comorbidity. Of 136 patients, 50 were white, 67 African-American and 12 another race; 60 were women. Bivariate analyses showed significant (p < 0.05 Wilcoxon rank sum test) differences in the distribution by race in frequency of problems with finances, access, pain and symptoms and in severity of problems with sleep, finances, access, pain and other symptoms. By gender these differences appeared in frequency of problems with role function and body image and in severity of problems with body image and sexual function. In multivariate analyses, African-Americans were 2-4 times more likely than whites to rate finances [odds ratio (OR) 95% CI = 4.1 (1.6-10.6)], symptoms [OR = 3.6 (l&9.7)] and diet [OR = 3.3 (1.2-9.2)] as a frequent problem and sleep [OR = 3.2 (1.2-8.4)] and symptoms [OR = 3.1 (1.1~8.5)] as bothersome. Women were two times more likely than men to rate body image as both a frequent )] and bothersome issue [OR = 2.6 (l.l-5.8)]. Race and gender differences for other domains were mitigated by adjustment for confounders We conclude that QOL research should be sensitive to differences in the importance of domains among racial and gender groups. This should be considered before eliminating domains that may be differentially affected. Comparisons of changes between NHP Part I mean scores at one and five years after surgery were performed using the Wilcoxon signed ranks test for paired data. To check for convergent validity of the measures used, associations between study questionnaire results were assessed using the chi squared test for contingency tables and associations between NHP Part I scores and symptom and activity results were assessed using the Mann-Whitney U-Test. In comparing the 1 and 5 year scores from the NHP, slightly lower mean scores, indicating improvement over time, were seen in the dimensions of pain, sleep, social isolation and emotional reactions. Signs of deterioration were noted in the physical mobility score of 5.6 at 5 years, compared with 4.4 at 1 year and the energy score of 16.4 compared to 12.1 at 1 year. None of the changes in NHP scores between 1 and 5 years were statistically significant. Comparing the 5 year NHP scores with those from a general population sample in the same age group, it was in the dimensions of physical mobility and energy that the patients produced higher scores. At 5 years following coronary artery bypass grafting, 54 out of 84 patients (64%) were working compared to 65 of 89 (73%) at 1 year after surgery. Chest pain was experienced by 34 out of 84 patients (40%) at 5 years compared to 17 of 89 patients (19%) at 1 year (p7) reduced from 54 (42%) for anxiety and 38 (29%) for depression pre-operation to 22 (17%) (p 0.84). In the absence of a gold standard against which to test the validity of MQOL, SIS is used as the best available measure of the patient's QOL. MQOL total score correlated more highly with SIS (rho = 0.73) than the QLI (rho = 0.60) or any MQOL subscale or QLI item. The pattern of correlation of MQOL subscales with Spitzer QLI items suggests that MQOL subscales have convergent validity. People with no evidence of disease (NED) following treatment had significantly less problems caused by physical symptoms (ANOVA, p = 0.0020) and significantly greater physical well-being (ANOVA, p = 0.0029) than people with active disease. Overall QOL was slightly but significantly worse for those with active disease compared to the NED group as measured by MQOL Total score (ANOVA, p = 0.0300) and SIS (ANOVA, p = 0.0244). There is a highly significant difference in the same direction for QLI scores (Mann-Whitney U-test, p = 0.0025). Multiple regression of MQOL subscales on SIS was highly significant (p < O.OOOl), with Rzadlusredof 0.59 and 0.52 for the NED and active disease groups respectively. The physical well-being, psychological, and existential subscales were most important in predicting SIS. Physical symptoms and support were less important. The existential subscale was more heavily weighted for the group with active disease than the NED group. We conclude that the results support the hypothesis that the existential domain is important. This is the first evidence that MQOL is acceptable, has reliable subscales, and is a valid measure of QOL in this patient population. Post-stroke quality of life P. J. Clarke, J. M. Lawrence, S. E. Black and J. I. Williams Health Science Centre, Toronto, Canada The objective of this research project was to determine the factors influencing quality of life after stroke. Stroke is a devastating illness that can affect many facets of everyday life. Not only do survivors experience limitations in their physical and self-care abilities, but family relations and social activities are often affected as well. In fact, limitations in social activities often remain long after survivors have regained independence in activities of daily living. Yet, quality of life receives surprisingly little attention in stroke outcome research. Quality of life is difficult to define, but handicap, as described in the International Classification of Impairments, Disabilities and Handicaps (ICIDH), can be conceptualized as a parallel concept to quality of life. Handicap is defined as a restriction in the meaningful social roles or activities for an individual following incapacitating illness or trauma, and is seen to be a situational result of an interactive process between impairments (which are abnormalities in the structure or functioning of the body), disabilities (restrictions in the performance of activities generally considered essential in everyday life) and environmental factors. This research project used this WHO framework to investigate the extent to which impairments and disabilities after stroke affect handicap, or quality of life, in interaction with environmental factors. In a prospective cohort study of 450 consecutive hemispheric strokes we examined the determinants of handicap in those survivors who were able to report for themselves at 3 months (n = 147) and at 1 year (n = 151) post-stroke. Handicap was assessed using the Reintegration to Normal Living Index (RNL), a self-report measure which asks individuals to rate their satisfaction with various aspects of their physical, emotional and social life. Disability was measured using the Functional Independence Measure (FIM) which assesses an individual's independence in mobility, locomotion, self-care, cognition and communication. Stroke-induced impairment was measured using a standardized hemispheric stroke scale. Information on environmental factors such as use of rehabilitation services, living arrangements, and social supports was also collected. We found that disability and impairment were important determinants of quality of life following stroke. Using oneway analysis of variance (ANOVA) with post-hoc comparison tests, stroke survivors who were more independent and less impaired reported higher RNL scores at three months and at 1 year than those with more severe impairments and disabilities (p < 0.05). However, not all individuals who remained disabled or impaired throughout the recovery period reported persisting handicap. For instance, there was a 75% improvement in quality of life between the two follow-up periods for those who remained dependent in the self care dimension of the FIM (increase in RNL score from 36 to 63 at 3 months and 1 year, respectively). While it is often assumed that the quality of life possible after a stroke is determined solely by the individual's physical disabilities and impairments, our results suggest that stroke survivors find ways to adapt to their persisting deficits in order to participate in life activities in a manner that is acceptable to them. Therefore, although disabilities and impairments are important determinants of post-stroke quality of life, they do not necessarily lead to inferior life quality. Some individuals appear to come to terms with their residual disabilities, or perhaps redefine which activities are necessary to them, so that they can reach some acceptance of their post-stroke existence. We need to gain a better understanding of the factors and mechanisms which influence this adaptation process so that health care providers will have more ability to assist in the restoration of optimal quality of life following stroke. The FLIC was published in 1984 and has been used in a variety of settings since that time. It is a 22-item self-report instrument having five subscales: physical well-being and ability, psychological well-being, hardship due to cancer, social well-being and nausea. Answers use a visual analog response format with verbal anchors at the extremes of the scale. An alternative response format is available for illiterate persons. Nineteen foreign language versions are available. The Flesch-Kincaid grade level is six. The instrument is characterized by its focus on the effect of cancer and its treatments on functional quality of life. It assesses not merely the presence or absence of symptoms but their impact on quality of life. Psychometric properties are well established. There are five published factor analyses other than the original which confirm the factor structure. Internal consistency reliability (Cronbach's a) is over 0.8 for the physical, psychological and nausea subscales and over 0.6 for the other subscales, while a for the entire scale is 0.89. Test-retest reliabilities ranged from 0.68-0.81 for five lmonth intervals on the same patients. Concurrent validity has been established by several authors against batteries of standard tests. As a measure of its acceptance the FLIC has been used to validate other general and specific quality of life instruments developed since 1984. These include the Cancer Rehabilitation Evaluation System (CARES) and the Functional Assessment of Cancer Therapy (FACT). A wide variety of studies have used the FLIC to evaluate the quality of life impact of a variety of different treatments in various types of cancer. Comparisons include high vs. low doses of chemotherapy, less vs. more toxic types of chemotherapy and surgical vs. medical treatments. Several studies have tracked patients over time to assess the longterm effects of disease and treatment: examples include primary breast cancer patients and bone marrow transplant recipients. The FLIC has also been shown to have prognostic value for response and survival out-performing the Kamofsky performance status score. In conclusion the FLIC has demonstrated acceptability to patients, readability, reliability, validity, sensitivity to treatment effects, sensitivity to the time course of disease and treatment, prognostic value for response and survival as well as cross-cultural validity in the specific settings examined so far. Self-reported well-being and functioning in bipolar disorder Clarke Institute of Psychiatry, Mood and Anxiety Division, Department of Psychiatry, University of Toronto, Canada The objective of this study was to assess inter-episodic well-being and functioning in subjects with bipolar (manicdepressive) disorder using the Medical Outcomes Study (MOS) 20-item short form. Brief self-assessment tools are increasingly being used to compare the impact of different medical disorders on well-being and functioning. In the RAND Corporation's Medical Outcomes Study, patients with active major depressive disorder scored in the same range as patients with several chronic medical illnesses on five of six indices of well-being and functioning, but the concurrent depressed mood may theoretically have negatively distorted patients' self-reports. We decided to estimate the impact of bipolar disorder on these indices by giving the MOS 20-item self-report tool to 27 men and 41 women [mean age 35.6k1.2 (s.e.m.) yrs] meeting RDC for Bipolar Disorder. To minimize and correct for the potential distortion in self-report due to depressed mood, subjects rated themselves when they were clinically judged to have been free of acute depression or mania for 2 1 month supported by scores < 10 on the Young Mania Rating Scale and < 16 on the 17-item Hamilton depression scale (HamD). We also administered the observer-rated Global Assessment Scale (GAS) measuring overall functioning and psychopathology. Our patients reported comparable or more severe levels of impairment compared with MOS major depressive and chronic medical patients on all six indices (physical social and role functioning, pain, mental health and overall health). HamD scores (range O-13; x f SD = 3.5ti.5) were significantly (p < 0.01) inversely correlated with social functioning, mental health and overall health perception and with GAS scores. GAS scores also correlated significantly with social and role functioning and mental health (p < 0.01) and with overall health perception (p < 0.05). These results suggest that residual depressive symptoms are an important determinant of bipolar disordered patients' self-reported inter-episode well-being and functioning. However, the correlation between self-reported social and role functioning and the observer-rated GAS suggest that the reported functional impairments were not simply a distortion due to concurrent depressed mood. The impaired social and role functioning reported by our patients is consistent with previous reports of poor social outcome in bipolar disorder. Thus, our results underscore and extend previous findings about the severe impact of bipolar disorder on well-being and functioning. The purpose of this study was to assess the physical and mental health quality of life among a sample of Laotian refugees (n = 126). The sample was selected using snowball sampling techniques. Interviews were performed by three trained Laotian health professionals and were conducted in both Lao and English languages. The most frequent year of arrival in the United States was 1980 and the average year of arrival to the USA was 1981. The SF-36 proved quite reliable (a = 0.85 and 0.91 for mental health and physical health, respectively) The results suggest the two subscales may not measure distinct construct domains between mental and physical health (r = 0.71, p < 0.01). When compared to a population of European-Americans (n = 2272), the Laotian sample evidenced more physical health (t = 11.4, p < O.Ol), but no difference was found for the mental health scale (t = 1.01, ns.). The discussion will also include factors promoting higher SF-36 scores and risk factors to lower mental and physical health quality of life. The current economic climate with its attendant Medicare costs containment endeavors, has encouraged much investigation to examine and quantify the benefits and costs of various medical and surgical interventions. In ophthalmology, cataract surgery is one of these procedures. While costs can be measured accurately with little effort, the investigation of non-economic variables represents a greater challenge to the medical and social scientists involved in research on the quality of life. The purpose of this study was to explore the association between improved visual functioning, and the quality of life satisfaction in patients who undergo cataract surgery with intraocular lens (IOL) implantation. Measurements of quality of life status in patients with bilateral cataract/IO1 surgery were compared to patients with cataract who declined surgery. This study was a prospective cohort study of 31 cataract surgery and 19 non-surgery patients all of whom had baseline and l-year follow-up interviews All surgically removed cataracts were replaced with IOL implants. Routine baseline and l-year follow-up visual acuity testing was obtained in both groups. All patients were pathology free except for cataract. Subjective measures assessing activities of daily living, visual functioning and a quality of life were investigated. Quality of life measures were derived from the Johns Hopkins Results suggest that alcoholic patients' perception of HRQOL is more severe in roles and psychological than physical and functioning dimensions. Moreover, depression may be a major factor of the perception of HRQOL in alcoholic patients. The SF-36 presents good criteria of applicability and acceptable criteria of reliability and validity. Further studies will determine the clinical and research interest of the SF-36 in alcoholic patients. Quality of Life study presented to the American Society for Cataract and Refractive Surgery. Best corrected visual acuity increased significantly (p < 0.0001) following cataract surgery. For the nonsurgical controls there was a nonsignificant trend for decreasing visual acuity over the l-year period. Subjectively, 98.3% of surgery patients indicated improvement in their quality of life. The quality of life indices measured statistically significant improvement in quality of surgery patients' lives (p < 0.04) compared to non-cataract surgery patients (p = 0.62). Change in best corrected visual acuity from preto post-operative was the best indicator of patient satisfaction (p = 0.04, p = 0.63). of health related quality of life (HRQOL) might be a useful tool in the management of alcoholic patients, for instance in determining treatment and measuring outcome. The aim of the study was to measure HRQOL and to examine the applicability, reliability, and validity of the SF-36 in alcoholic patients. The patients included were in-and outpatients with a DSM-III-R diagnosis of alcohol dependence, referred to a specialized alcohol multidisciplinary unit, grouping medical, psychiatric and social services, between January 1993 and March 1994. Validated translations in French of the SF-36, the Hamilton Depression Scale (HAD) and the Severity of Alcohol Dependance Questionnaire (SADQ) were used. The SF-36 was administered twice with an interval of 10 days. Correlation coefficients were calculated to measure the degree of association between different dimensions of the SF-36, test and retest scores, as well as between the different instruments used (concurrent validity). Internal consistency was measured by means of Cronbach's a. In 147 outpatients (mean age = 48, range = 26-78; 77% males) the SF-36 scores were relatively low in the psychological and role dimensions (range = 52-55/100) and higher in the physical and functioning dimensions (range = 61-75/100). Twenty-eight patients (19%) gave at least one inconsistent answer. Average test-retest correlation coefficient of reliability was 0.71 (range = 0.65-0.80) and internal consistency Cronbach's a coefficient was 0.82 (range = 0.70-0.89). Correlation coefficients between different dimensions varied in the range 0.10-0.62. Moderate correlation coeffi-High-tech medicine examinations such as Magnetic-Resonance-Imaging (MRI) cause psychological as well as somatic stress symptoms in many patients. This has negative effects on the patient as well as on the quality of diagnosis. This prospective study aims at clarifying overall tolerance, detecting specifically anxiogenic parts of the whole process and at developing a model for rational patient management geared to the patients' needs. The study sample consisted of 258 consecutive patients undergoing diagnostic MRI. Pre-and post-scan anxiety levels were determined by self-rating (STAI-Xl; Spielberger) and by a visual analogue scale (VAS) rated by the attending nurse. The patients' attitude towards the examination and their subjective feelings were evaluated before and after the scan with specially developed questionnaires (Imaging Distress Questionnaire, IDQ; Dantendorfer). Adverse reactions and motion artifacts were documented. 84.6% of all patients rated the scan as easy to tolerate, 14.2% as unpleasant and only 1.2% as hardly bearable. The anticipation of the forthcoming scan caused a significant elevation of state anxiety. STAI-Xl revealed a significant reduction of anxiety levels once the scan was over. Between pre-scan anxiety levels and the frequency of motion artefacts no statistical connection could be found. Patients who stated to be worried by the big technical apparatus had significantly more motion artefacts impairing the diagnostic quality (11.5%) than patients considering themselves as not worried (3.2%' p = 0.007). More than one in nine patients reacts to MRI scanning with significantly stronger than average emotional distress. This subgroup has a significantly higher incidence of motion artefacts than the non-disturbed group. The subgroup can be identified by the Imaging Distress Questionnaire. cients were found between non-physical dimensions of the SF-36 and the HAD (range = 0.37-0.60), and the SADQ This paper demonstrates how clinimetric and psychometric methodologies can be complimentary in producing a functional status index, that is, satisfying both clinimetric and psychometric criteria of measurement The Toronto Extremity Salvage Score (TESS) is a measure of physical function for patients undergoing limb salvage for bone and soft tissues tumours and The Patient Specific Index (PASI) is specific to patients undergoing total hip arthroplasty (THR). Both scales were developed using patient perceptions, with the priority on the clinimetric methodology. It is possible that these measures, due to the concepts from which they were developed, may have psychometric properties that support their clinimetric underpinnings. To establish the items of the TESS, questionnaires were mailed to 80 upper and 80 lower extremity sarcoma patients who had undergone limb salvage surgery a minimum of 6 months previously. Difficulty and importance were rated for each item and the response options included a 'not applicable' category. In the clinimetric evaluation, difficulty and importance frequencies were calculated for each item. Items rated 'totally unimportant' or 'not applicable' by 30% of the sample were eliminated. Extra items identified 30% of the time were added to the questionnaire. Literature, open-ended interviews with 20 subjects identified items for inclusion in the PASI. Psychometric evaluation included factor analysis and measurement of internal consistency by Cronbach's alpha. The difficulty responses of the content survey of the TESS and difficultly responses (n = 78) of the PAS1 were analyzed with principal component analysis varimax rotation of the factors. Factors with eigen values of 1.0 or greater were retained and items had to have minimum communality estimates of 0.40. Factor loadings of 4.0 or higher were used to locate each item on the factor. Test-retest reliability and validity of the TESS and PAS1 have been evaluated on subsequent patient samples. The intraclass correlation coefficient (ICC) was calculated for test-retest reliability and correlations were calculated for construct validity. Standardized effect sizes were calculated as a measure of responsiveness. Sporting activities were identified as an additional item by 37 and 18 lower (n = 66) and upper (n = 50) extremity subjects respectively meeting the criteria for addition to the questionnaire. No items met the criteria for exclusion based on the importance rating. The open ended interviews of the PAS1 identified four new items: (1) night pain; (2) leg length discrepancy; (3) discomfort during sexual activities; and (4) discomfort during recreational activities. Carrying and independence were added following the survey compo-nent. In the lower extremity TESS Cronbach's a was 0.94. After varimax rotation the items loaded heavily on the first factor, which had an eigen value of 15.00 and accounted for 51.7% of the total variance of the items. Cronbach's a was 0.92 for the upper extremity TESS. Varimax rotation loaded items heavily on the first factor, with an eigen value of 16.39 accounting for 58.5% of the total variance of the items. The difficultly ratings of the PAS1 had an alpha of 0.84 and varimax rotation identified two factors: the first accounting for 54.8% of the variance and including mobility and self-care items and the second accounting for 24.8% of variance and including activities requiring bending. ICC's for the TESS and PAS1 were greater than 0.70 at multiple time points and Pearson's correlation coefficients in the 0.50 range were obtained in comparison to existing clinical measures. Over time the TESS had effect sizes greater than 1.0 in response to surgical excision of extremity tumours. Effect size of the PAS1 is 1.4. Content of two independent functional status indexes were developed using clinimetric principles, however, this method was also able to produce questionnaires that satisfied psychometric criteria for questionnaire content, reliability, validity and responsiveness. The clinimetric approach focuses on the development of measures that are useful in forming clinical judgements. While the clinimetric and psychometric approaches may be viewed as opposing sets of methods, they can also be viewed as independent approaches toward a common goal, the development of a measure for use in climcal research. Quality of life evaluation with children: Validation of a questionnaire A. Dazord, S. Manificat, J. Nicolas, P. Cochat and L. David Script-Inserm, Hopital Saint-Jean de Dieu 69008 Lyon, France We have developed a questionnaire about quality of life which is derived from the concept of 'goals' and which has been broadly used in adults with different diseases and treatments (Gerin I', Dazord A, Boissel JR Chifflet R. Quality of life assessment in therapeutic trials: rationale for and presentation of a more appropriate instrument. Fundamental and Clinical Pharmacology, 1992; 6: 263-276) . This questionnaire covers four main domains of life: functional life, social life, material life, spiritual life. It is modular both for the items that cover the domains, and for the measurements (the questions). The unavoidable measurement is the satisfaction, but the importance attached by the patient and the degree of change expected allow us to obtain interesting results. The questionnaire has been used and validated with about 3,000 patients. It is able to describe some specific changes, but also to explain the mechanism of action of some therapeutic approaches according to the patients' personality, behaviour or expectations. [Dazord A, Germ I', Boissel JI? Subjective quality of life assessment in therapeutic trials: presentation of a new instrument in France (SQLP: subjective quality of life profile) and first results in Quality groups of patients with different disease activity. In order to test for construct validity, the scores of the IBDQ and of the IBDQ subscales (bowel, systemic, social and emotional) were correlated (pmcc's) with disease activity, health care use and medication. We examined the correlations (pmcc's) between IBDQ and four generic indices of quality of life (MOS-24, COOP charts, CES-D, MOS Social Support), in order to further evaluate the construct validity. Dutch IBDQ scores were compared to Canadian norms, in order to assess cross-cultural validity. In all, 271 questionnaires were returned completed (87.7%). The reliability coefficients of the IBDQ subscales were high (0.73-0.87). Patients with higher disease activity had significantly lower quality of life on all IBDQ subscales (p < 0.001). IBDQ scores were significantly correlated with health care use ( Orley J and Kuyken W, eds. 19941. More recently we developed a QOL questionnaire with 27 items, for children from 4-12 years. Two successive samples of children filled up the questionnaire. One sample of 54 children in a hospital setting, filled up a reduced version (nine items) and a second sample of 71 children filled up the 27-item version. Half of them are in good health, while the other half is under medical treatment in hospital (14 HIV positive, 22 with kidney transplants). Average age is 7.2 years. Satisfaction levels are presented using faces expressing different emotional states. In terms of acceptability this questionnaire is very seldom rejected, and understanding of the satisfaction levels proved to be good. Cronbach's a is 0.71, and the children's answers are most of the time in good agreement with the evaluation by the doctor or the nursing staff. Principal Component Analysis with rotation produces four factors (40% of the total variation): one relating to autonomy, one to leisures, one to functions, and one relating to parents. Preliminary data show some specific characteristics in children with severe diseases. Comparing healthy children with sick children we found than some items discriminate better than others: in particular the satisfaction expressed over family meals, sport, the father, school and birthdays is modulated. Society. Reliability of the PDQL was tested by investigating the consistency of results, using Cronbach's CL coefficient. The method of known-groups comparison was used to evaluate the clinical validity of the PDQL. One-way analysis of variance (ANOVA) was employed to test for the statistical significance of differences in PDQL scores between three groups of patients with different disease severity. In order to test for construct validity, the scores of the PDQL were correlated (pmcc's) to generic indices of quality of life (MOS-24, COOP charts, CES-D, MOS Social Support). In all, 450 questionnaires were returned (85 1%). Of these 450 questionnaires, 384 (85.3%) were complete enough to be used for data analysis. The PDQL consisted of four subscales: Parkinson symptoms, systemic symptoms, emotional functioning and social functioning. The reliability coefficients of the PDQL subscales were high (0.82-0.87). Patients with higher disease severity had significantly lower quality of life on all PDQL dimensions The long term treatment of hypertension is directed towards reducing organ damage and associated risk of mortality. However, the lack of symptoms in a large population of hypertensives, the side effects and impairment of Quality of Life (QOL) caused by drugs are the main reasons for patients discontinuing treatment. In 1993 we began the development of an Italian hypertension-specific questionnaire; nine focus group meetings were carried out, using a preliminary 31 item questionnaire designed for use in Italian hypertensive patients. Responses from 117 patients (64 males and 53 females, average age and duration of hypertension respectively 57.1 and 7.6 years) revealed their difficult comprehension of the disease and of their treatments, difficulties of complying with therapeutical regimens, the nature and frequency of drug adverse effects and the burden of lifetime therapy. The most important and distressing clinical aspects of the disease were night awakenings, mood changes, tiredness and constipation. Two psychological dimensions were identified: a general feeling of well being, characterised by not being afraid of high blood pressure or of stopping work through illness, and a low self-esteem, defined by frustrations, worsened family relations, etc. These results gave rise to the categorization of two kinds of hypertensive patients for application in clinical practice. The first, type A, is asymptomatic, self-confident, possibly non-compliant and relatively carefree. Such a patient is likely to respond to appeals to responsibility in compliance. The second, type B, is concerned, constrained, self-stressed and worrie, and probably would benefit from counselling. On the basis of the data collected, a disease-specific tool for Italian hypertensive patients, called Hyper 31, has been prepared and protected with copyright. It consists of 31 items, and it contains a specific personality inventory scale to allocate patients to one of the two hypothetical psychological dimensions. A symptoms check list is part of the questionnaire, which has a Cronbach's CI of 0.81. Only lo%-20% of the resuscitated patients reach this outcome. As survival after CPR can be impaired by neurological disorders, functional health and perceived quality of life (QOL) after CPR are important measures of its definitive success As yet health outcomes after CPR are sometimes reported using clinical judgement, but hardly by clinically well-evaluated indices. The purpose of this study is to evaluate multi-dimensional health status in survivors from in-hospital CPR, using selfrating indices, to support decision making regarding in-hospital do-not-resuscitate orders. CPR survivors were identified from the data base of the resuscitation team in our university hospital. Cognitive impairments were measured with the Mini Mental State Examination index (MMSE); depression with Centre for Epidemiologic studies-Depression scale (CES-D); physical disabilities with the Sickness Impact Profile (SIP) and level of independence was assessed with the Rankin score (RS). Patients rated perceived physical, psychological and social health and overall QOL on visual analog scales (VAS) ranging from 'worst' (0 points) to 'best' (10 points). A total of 447 patients were resuscitated from June 1988 to January 1994. 95 patients (20%) survived to discharge (58% male, median age 69). At follow-up (February 1995) 22 hospital survivors had died; (one-year survival 83%). appeared between life satisfaction and the following QOLfacets: Overall Quality of Life and General Health, Positive Feelings, and Participation in and Opportunities for Recreation/Leisure Activities. Furthermore, it was shown that total WHOQOL-scores were linked to the total score on the GHQ-0 (r = -0.58; p < 0.001) and the six subscales of the PGWB, ranging from r = 0.56 (p -c 0.001) for Anxiety to Y = 0.71 (p < 0.001) for Positive Well-being. The association between WHOQOL-total score and the total score of the PGWB was 0.74 (p < 0.001). Eight patients declined to participate in our study, whereas three were lost to follow-up. Sixty-two survivors were taking part and interviewed at home. Evaluating cognitive impairments, six patients (10%) had MMSE scores < 24, indicating dementia. Depression, as measured by the CES-D, was found in 18% of the patients. SlP-scores contrasted against scores from an open population elderly (age 60-70), showed significantly higher relative scores in CPRsurvivors, indicating poorer health after CPR (fig). A substantial loss of independence (RS13) was found in 11 patients (17%). Perceived physical, psychological, social health and overall QOL was rated 6.4, 7.5, 7.9, 7.1 respectively. Health outcomes in patients after in-hospital CPR indicate some affected cognitive functioning and loss of independence, as well as a decline psychological, functional health compared to an open sample within the same age group. However perceived health and QOL suggest that CPR was still successful, for those surviving. Validating the Dutch WHOQOL J. De Vries Tilburg University, Department of Psychology, Tilburg, The Netherlands Two studies were done to validate the Dutch version of the World Health Organization Quality of Life assessment instrument (WHOQOL). In the first study, the subjects were 176 persons who had been randomly selected and asked by phone to participate. Besides the WHOQOL they filled out the Life Satisfaction Index A (LSIA), the General Health Questionnaire (GHQ-30), and the Psychological General Well-Being (PGWB) index. In the second study, 320 elderly persons aged 60 or older who were still self-sufficient participated. As in the first study, they were recruited by telephone. Questionnaires employed in this study were, apart from the WHOQOL, the Psychological Well-Being scale from the Philadelphia Geriatric Center Morale Scale, a scale measuring Residential Satisfac&on, a standard scale for measuring neighboring, the Senio@&ifestyle Inventory, and the Jong-Gierveld's Loneliness Scale. In both studies, the WHOQOL was correlated with the other questionnaires on facet-, domain-and total score-level . The outcomes of these studies indicated that the WHO-QOL has a good validity. For instance, the LSIA correlated significantly with five out of the six QOL-domains, ranging from r = 0.46 (p < 0.001) for Social Relationships to T = 0.59 (p < 0.001) for Psychological Health. The purpose of this study was to describe the experience of fatigue over time in patients with cancer while receiving treatment with interferon alpha. Fatigue is one of the most prevalent and insidious of symptoms experienced by patients with cancer. Interferon alpha is widely used and known to cause considerable symptom distress and fatigue in cancer patients. The conceptual model of fatigue in people with cancer by Piper et al. is consistent with a quality of life model and guided this study. A descriptive repeated measures design was used. A convenience sample of thirty patients with malignant melanoma was drawn from a comprehensive cancer centre in Southern California. Two instruments were used in data collection: the Symptom Distress Scale (SDS) (McCorkle, Young, 1976 ) and the Piper Fatigue Scale (PFS) (Piper, 1989) . Both of these instruments have established reliability and validity. The data was analyzed using the CRUNCH statistical programme. Ddptive statistics, Pearson's correlations, and ANOVA were used to evaluate fatigue and other symptoms over five points of treatment. Study findings revealed descriptive data on patients' perceptions of the causes and remedies for fatigue while receiving active treatment for cancer. Strong correlations were found between the total SDS and items of fatigue (T = 0.81), discomfort (Y = 0.76), and subscales of temporal fatigue (I = 0.73), and the sensory dimension of fatigue (Y = 0.80). The total Piper Fatigue Scale score demonstrated strong correlation with items of pain (r = 0.64), concentration (r = 0.64), and overall SDS of r = 0.78. The pattern of fatigue was consistent over the five points of treatment with the most extreme fatigue scores in the affective domain, followed by the sensory, temporal, total fatigue, and fatigue severity scores. The purpose of this methodological study is to determine the psychometric equivalence of a Tagalog (Pilipino) version of the RAND 36-item health survey 1.0 as a culturally-valid measure of the health-related quality of life (HRQOL) of Filipino-Americans who will comprise the largest Asian group in the USA in the year 2000. The RAND 36-item Health Survey 1.0 was translated into Tagalog (Pilipino) and then independently backtranslated into English using several levels of iteration The first level of iteration used four independent Filipino-American bilingual translators: the first translator initially translated the 36-item health survey into Tagalog; then the second translator reviewed and modified this preliminary version, resulting in the first Tagalog version. The first Tagalog version, in turn, was backtranslated into English by the third translator. The resulting English backtranslation was also reviewed and modified by the fourth translator creating a second English version. The second level of iteration comprised of three panels consisting of four to five ambulatory bilingual subjects who met the eligibility criteria for the study: (a) able to speak and read English and Tagalog, (b) confirmed diagnosis and treatment of hypertension and/or heartproblems, and (c) 18 years of age or older. The first panel reviewed the equivalence of the first Tagalog version to the original RAND measure, creating a second Tagalog version. The second panel examined the equivalence of the second English version to the original RAND measure, resulting in a third English version. Finally, the third panel separately evaluated the equivalence of the second Tagalog version and the third English version to the original RAND measure, leading to a third Tagalog version and the fourth English version. The last level of iteration involved the final expert review of the equivalence of the third Tagalog version and the fourth English version to the original RAND measure by a state-licensed Filipino bilingual translator resulting in the final Tagalog and backtranslated English versions. Two hundred fifty-seven eligible subjects were recruited in 9 months. They were randomly assigned to initially receive either the final Tagalog version or the original RAND measure. Each subject received the appropriate version of the measure upon mailing back the version first received. One hundred fifty subjects (58%) completed both the final Tagalog version and the original RAND 36-item measure. Data was scored using the RAND method. Agreement in the scores between the final Tagalog version and the original RAND 36-item measure was 68%. Both versions showed an overall 0.95 reliability coefficient (Cronbach's a). Correlations between each item of the two versions varied from 0.38 to 0.79 Except for two items, the item 418 Quality of Life Research VoJ 4 1995 discrimination across eight scales showed item-scale correlations varying from 0.63 to 0.92 in the RAND measure and 0.65 to 0.95 in the Tagalog version. Principal components analysis using orthogonal (varimax) rotation extracted seven factors from the RAND measure and eight factors from the Tagalog version The same items in both versions loaded under role limitations due to physical and emotional problems (Factor I), physical functioning (Factor 2), feelings of vigor, peace and happiness (Factor 3), and symptoms of physical and mental suffering (Factor 4). In the RAND measure, items loaded on general perceptions of health and illness (Factor 5), meaningful physical activities (Factor 6), and comparison of health status (Factor 7). In the Tagalog version, items loaded on meaningful activities (Factor 5), energy/fatigue (Factor 6), general perceptions of health (Factor 7), and general perceptions of illness (Factor 8). On the whole, there is good correspondence among the items that loaded on the factors extracted from the RAND 36-item measure and the Tagalog version. * Pilipmo IS used to refer to persons of PhilippIne ancestry. This term IS preferred by recent immigrants to the USAfor purposes of ethmc identlty in Imewzth thetrsageof thenattonal languugeofthePhilippines. ation the effect of treatments on individuals' health. Conceptual Equivalence. Content equivalence: an analysis of the a coefficient for internal consistency was > 0.75 for the QOLS. Semantic equivalence: backtranslation of the scales showed > 80% of similar meaning in the QOLS. Technical equivalence: the sample of patients interviewed accepted well the structure interview method. Criterion and conceptual equivalence were not measured due to the nature of the study which had as a goal to get the subjective perception of the interviewees in those variables universally represented in the QOLS. Cultural free clinical instruments are rather rare. Content, semantic and technical equivalence was successfully achieved for a modified version of Quality of Life Scale. Criterion and conceptual equivalence were not done due to lack of instruments available to measure similar phenomena in each culture for comparison. Although the evaluation of many experimental treatments and programmes now includes measuring the treatment's effect on additional years of healthy life (YHL), many statistical aspects of YHL have not yet been studied. We investigated some theoretical and practical aspects of YHL, using cohort data from an evaluation of a health promotion programme for older adults as an example. We first derived algebraic expressions for the effect size of YHL measures under different theoretical models of change. We then examined the influence on effect size of the length of the study, the number of measurements per person, the number of patients per group, and the correlation over time of the health measurements. We next explored the relative statistical power of analyses based on YHL vs. analyses based on simple change-scores using the same data. Finally, we used real data to study how the variance and correlation of health measures change over time, whether a discrete measure of health status is more powerful than a continuous measure, and whether a study using a change score is more powerful than a study using YHL. Findings suggest that the number of measurements per person need not be large, that a longer study does not always provide the greatest power, that YHL may not perform as well as a change score, that measuring the health of different people at each time point may have higher power than measuring a cohort of individuals repeatedly, and that a discrete measure of health status may perform as well as a continuous measure. These surprising findings suggest that a variety of study designs and outcome measures should be considered in the evalu- Health-Econ AG, Base1 (R. Dinkel); Gesundheitsschutz der AUDI AG, Ingolstadt (8. Birkholz, R. Tiller); Zyma GmbH, Miinchen (M. Unkauf) Chronic venous disease is a disturbance, where-especially in the first stages-an objective measure of pathophysiological changes is rather problematical; nevertheless the problems and impairments exist and are felt subjectively by the patients. Therefore, the assessment of the subjective perspective of these patients should be of interest. Unfortunately, a specific instrument for the assessment of quality of life (QOL) in patients with chronic venous disease did not exist up to now. We have developed and validated a disease-specific German questionnaire for this purpose, the Venenskala. The questionnaire consists of eight subscales ('Complaints', 'Treatment', 'Functional status', 'Disease-specific fears', 'Mood state', 'Contentedness', 'Social life' and 'Judgement of the disease'), and contains 104 items. It was presented together with two German standard QOL instruments (AL and MLDL) to 100 healthy persons and 100 patients with venous problems in a monocentrical cross-sectional study, and it was tested for its reliability, validity and sensitivity. The results show that the instrument has good psychometric properties, the scale fit was between 90% and lOO%, the test-retest coefficient between 0.63-0.94, and Cronbach's a above 0.7. Only the subscale 'Treatment' did not meet the demanded criteria, but this result was unmistakably due to the fact that the patients tested were not under current treatment for their venous disease. The Venenskala exhibited also a good correlation with the standard questionnaires and an excellent discriminative power. The instrument was sensitive enough to find a difference between healthy persons and patients with CVI stage I, a stage when patients do not yet have real problems (as there is still no edema) but the very first slight skin alterations. It turned also out to be quite practicable: it could be filled out independently by the patients in a short time (< 20 min) and was judged to be easily comprehensible. The results of the validation study let us come to the conclusion that the Venenskala is a suitable instrument for the assessment of quality of life in patients with chronic venous problems. It is comprehensive and specific enough to be used as a sole instrument in studies. ones are short stature, a more or less distinct phenotype and infertility. To analyze quality of life in these patients a complex research design is necessary because of the complexity of physical signs in this chromosomal aberration. The dramatic improvements witnessed in pediatric oncology patients' prognosis have led to the question, Survival at what cost? Quality of life (QOL) during and following treatment has emerged as an important issue for patients, clinicians, researchers and family members. While the assessment of QOL has become accepted practice for adult patients, relatively little work has been conducted with pediatric patients who present a special challenge to those interested in evaluating this construct. Questions which arise in the assessment of QOL in young medical patients are as follows: which factors need to be measured; who is to 'speak for' the child; how can developmental changes be dealt with in repeated measures designs; which instrument should be employed; when and how often should assessment occur? In order to address these questions a study was conducted with 106 pediatric oncology patients who were assessed with the Health Utilities Index (HUI) developed by the McMaster group (Feeny, Barr, Torrance, et al.) and the Child Behavior Checklist (CBCL). 51.9% of the patients were in treatment whereas 48.1% were post-treatment; there was a significant difference in their global HUI scores [Xs and (SDS) were 0.81 (0.17) and 0.92 (0.09), respectively] indicating that patients in treatment had a lower quality of life compared to those post-treatment. Specifically, they had more difficulties in the following attributes: self-care, emotions, and pain. The sample was also subdivided into leukemia patients vs. other types of cancer patients; 51.4% were the former and 48.6% were the latter. While there was no significant group difference for the global HUI, the leukemia group experienced more pain than the other patients. There were no significant group differences on the Internalizing or Externalizing Problem Scales of the CBCL, when the leukemia vs. other cancer patients comparisons were made. There was a significant difference for children still undergoing treatment: they had more internalizing problems compared to those post-treatment. Potential answers to the questions posed by working with pediatric patients will be discussed. For example, pediatric oncology survivors often experience psychosocial late effects and cognitive deficits-these factors must be adequately measured. Parents report more problems than do medical personnel, therefore their input is required. Children over eight years old who do not have cognitive deficits may be consulted, along with their parents; children over 14 years old may be tested directly. To date there is no 'gold standard' in the assessment of quality of life in pediatrics, but progress has been made. To explore the quality of life and its determinants we studied 55 adult patients with UTS and their parents. As adequate methods for the patients we've chosen among others: an interview, a quality of life questionnaire (MLDL), a body image questionnaire (BIQ), a social support questionnaire (SSS), a questionnaire concerning the parental coping with the illness as seen by the patient (FKV), a questionnaire concerning social fear (IAF) and an IQ-test (WIP). The parents had to answer a questionnaire for parents of short statured children, a questionnaire concerning the raising of their child (EPT, ET), and a 'coping with illness' questionnaire (FKV). We will present the results concerning quality of life in women with UTS and its most important determinants. The development of the AGHDA: A measure to assess quality of life of adults with growth hormone deficiency L. C. Doward Galen Research, Manchester, UK The paper will describe the development and testing of a condition-specific measure to assess the quality of life (QOL) of adults who have Growth Hormone Deficiency (GHD). The measure was designed for use in clinical trials and for the routine monitoring of GHD patients. The AGHDA has as its theoretical basis, the needs-based approach to QOL and its content is derived directly from GHD patients. The instrument was developed for use in the UK, Sweden, Germany, Spain and Italy. The initial stages of the work were carried out in the UK alone, with later stages being conducted in all five countries in parallel. In order to obtain the views and experiences of adults with GHD, in-depth qualitative interviews were conducted in the UK with 36 GHD patients. The interviews were tape-recorded and transcribed. Content analysis of transcripts resulted in a pool of potential items for the draft questionnaire. Items were selected on the basis that they related to needs interfered with by GHD and that they were applicable to all respondents. The pool of items were considered by a panel of international quality of life researchers representing the five European countries. Items were conceptually translated for use in each of the languages, following the procedure recommended by the European Group for Health pretation, BMJ 1992, 305 .) The concept of effective and efficient care provisions for people requiring long-term care must address quality of life issues. This study has used the Functional Analysis of Care Environment classification system (FACE) which is a multi axial, problem oriented approach in conjunction with a diagnostic classification system to study eight different residential care provisions for elderly people in England. The care environment profile is assessed against established criteria in six domains: Mental Health; Physical Health; Physical Environment; Compensatory Care; Rehabilitation and Individuality providing a quantitative scoring for each provision as well as a qualitative statement. This empirical study involved 229 people in eight different residential settings including residential, nursing and hospital care in the public, private and voluntary sectors. The population studied were classified as mentally or physically infirm requiring 24-hour care The study involved two main approaches: (1) individual assessments using a criteria form completed by the primary carers following training; (2) evaluation of the care provisions and environment using established criteria in six domains. This was undertaken using semi structured interview and observational techniques by a health care professional and a lay person in each case. The report, commissioned by a health care purchaser, uses analysis of variance to compare the range of individual dependencies The collated data clearly demonstrates that all provisions are caring for people with the full range of both physical and mental disability. The study showed that the physical environments were significantly different as was the provisions of care, however there was no significant difference between the dependency of patients in private nursing home care and long-term hospital care. As many of the patients in private nursing home care are funded by the National Health Service and Community Care provisions there are significant questions to be addressed regarding the quality issues of equity and access against cost comparisons as well as other quality of care issues for this very vulnerable client group. The value of the measure in providing evidence for informed purchasing of care and future development of the service can be evaluated. and Quality of Life Assessment. The draft measure was field-tested for face and content validity in each of the countries. It was found to be acceptable, relevant and easy to complete. Each language version of the AGHDA was tested for reliability and construct validity in a survey conducted in each of the countries. Test-retest correlations showed that reliability for all language versions is high (UK 0.93; Sweden 0.93; Germany 0.89; Italy 0.85; Spain 0.91). The internal consistency of each language version was also assessed and alpha coefficients were found to be consistently high [UK 0.93; Sweden 0.92; Germany 0.89; Italy 0.89; Spain 0.88 (Time 1 values)]. Validity was assessed by correlating scores on the AGHDA with a suitable comparator measure. The Psychological General Well-Being Schedule (PGWBS) was used as a comparator for the AGHDA in the UK, Sweden and Germany. Correlations were 0.77, 0.73 and 0.77 respectively The NHP was used as a comparator for Italy and Spain. Here, a moderately strong relation was found between QOL and the energy level (Italy, 0.52; Spain 0.79), emotional reactions (Italy, 0.70; Spain 0.77) and physical mobility (Italy, 0.49; Spain 0.79) sections of the NHI? The psychometric properties of the AGHDA mean that it is suitable for use in routine clinical practice for monitoring the progress of individual patients. It is further concluded that the new measure can be used in multinational clinical trials to assess the effectiveness of growth hormone replacement therapy. Additional language versions are in preparation for use in the USA, France, Belgium, the Netherlands, Denmark, Norway and Iceland. Quality of life: Quality of careconceptual dissonance in elderly care University of Luton, UK A comparative study of eight residential care environments for elderly mentally and physically infirm people using a new multi axial problem orientated approach in conjunction with a diagnostic classification system. This paper reports a study which seeks to compare different care provisions within the context of quality of life for the people studied. It seeks to develop the body of knowledge in the use of a new measurement tool over a range of disabilities within the field of elderly care. The growth in health economy oriented measurement methods has been fuelled by the development of market forces in health care provision and the rising cost of care; resulting in increasing economic pressure from health care purchasers to identify efficient and effective indicators for care provisions. It has been suggested that measures of health-related quality of life should be selected on the basis of their performance in a specific role, in a given population. Measures of health-related quality of life can be implemented to fulfil three roles; describing a sample, predicting future health or evaluating change in health over time. Rarely is there an opportunity to demonstrate performance of each of these three roles in the same population. The purpose of this study was to describe the psychometric properties of the Short Form-36 (SF-36) Acute as they related to the three roles (describing the sample, predicting and evaluating change) in a cohort of workers with losttime, soft tissue injuries. Nine hundred and thirty-two workers with soft tissue injures were followed for one year from the onset of injury. Telephone interviews were conducted at entry, four weeks, sixteen weeks and one year post injury. In addition to demographic questions and injury descriptors, each subject completed the SF-36 at each assessment. Baseline data was used to assess the instruments' ability to discriminate across self-rated overall health (analysis of variance) and independent ratings of pain and activity restriction (Spearman ranked correlation coefficients). Cox proportional hazard modelling was used to predict future duration of time off work from baseline SF-36 scores. Duration of the first episode of time off work was obtained using Workers' Compensation Board data. Standardized response means (SRM) of those who returned to work and those who had not returned to work were calculated to determine the ability of the SF-36 to evaluate change. The sample of 932 injured workers consisted of 487 women and 445 men with a mean age of the subjects of 39 years. Five hundred thirty-five subjects had low back pain, 250 an upper extremity condition and 147 a lower extremity conditions. The SF-36 scores, with the exception of the emotional role functioning, were able to discriminate across levels of self-rated health (p < 0.05). Correlations with independent ratings of activity and pain scales ranged between 0.004 < r < 0.44. The scores for pain, vitality, physical functioning, social functioning and physical component scores had the highest correlations with the activity and pain scales (0.30 < r < 0.44). The pain, physical functioning and the two summary scores were significant predictors of the amount of time a subject was off work. Hazard rate ratios ranged between 1.005-1.044, indicating less time off work in those with higher SF-36 scores at baseline. The predictive ability was enhanced by including binary variables indicating gender and upper extremity conditions. Using SRM, the SF-36 was able to show that those who returned to work had more improvement in health than those who did not. This was most pronounced in physicallyoriented dimensions (pain, physical function, physical component score); in the psychosocial dimensions, including the mental health summary score this difference w a s more marked at 52 weeks follow up (figure 1). This study demonstrated the ability of the SF-36 to describe, predict return to work and evaluate change in a group of workers after a work-related soft tissue injury. Measurement properties of these three roles, on the same population, has not been described before. Construct validity was confirmed through moderate correlations with theoretical constructs. Predictive ability was optimized only after adjusting for a systematic difference for SF-36 scores in upper limb patients. SRM's confirmed the ability of the SF-36 to detect change over time as indicated by return to work. The SF-36 was able to perform in each of the three roles identified for this type of instrument. Since 1990 we have investigated the psychosocial impact of short stature, growth hormone (GH) insufficiency and response to GH treatment prospectively in children. Assessment of quality of life in children becomes increasingly important in clinical studies. The well-being of children has commonly been assessed by asking the children's parents and teachers. Few studies have been performed that include questionnaires that are completed by the children themselves. Proxy r e s p o n d e n t s and formal b e h a v i o r a l performance tests constitute standard methods. This study reports baseline data for 93 prepubescent children of short stature with various amounts of GH-secretion. Sixty-nine of the children were boys and 24 were girls. Mean age for the total group was 11.7 years and ranged from 8.3-14.9 years at start of the study. All the children and their parents are part of the same psychological examination yearly until they reach final height. A short and simple measure, the Visual Analogue (VA) scale, was used to assess well-being. The measure w a s designed to focus on aspects believed to be of relevance to children of short stature. The measure comprises 39 visual Absfracfs of fhe Contributed Papers for each individual as an endpoint. Dawson and Lagokos (Biometrics, 1993) suggest distribution free tests for a general class of summary statistics that are a scalar function of the data. We will present the results from clinical trials in adjuvant breast cancer and advanced non-small cell lung cancer using these approaches and an additional approach utilizing the weights of the Cox approach with estimates derived using mixed effects models with a focus on effect of missing data on these analysis strategies. analogue scales with bipolar adjectives (e.g., sad-happy). The children's well-being was also assessed with 'I think I am' a Swedish self-esteem questionnaire. The questionnaire includes statements which the child responds to by putting a mark in one of four categories concerning how well he/she thinks that the statement describes himself/herself. The results from 'I think I am', the Swedish self-esteem questionnaire, is expressed as a stanine distribution (a scale ranging from 1-9) The normal population has a mean of 5 and 1 (SD = 2). The study group of short statured children have a mean of 4.9 for the total score, physical abilities 4.2, talents 5.1, well-being 4.9, relationship to family 4.8 and relationship to others 5.2. As a group the prepubescent short children perceive themselves much like the Swedish norm group. The mean values for the children on the VAscale are: total score 68, alertness 71, self-esteem 67, mood 74, elation 52, stability 69 and vitality 69. Compared to a school sample of 342 normal children in the age 9-13 the short statured children's mean values are very close to the normal group. We conclude that this group of prepubescent children with short stature rated themselves within the normal range on the tests used at start of the study. No adverse impact on well-being or self-perception was found in this sample with the methods used. There might though be a decline in psychological adjustment when this group reaches puberty. quality of life (HRQOL) has become an important endpoint in many clinical trials of cancer therapy. In most of these studies, HRQOL is repeatedly assessed over time. One of the analytic challenges is to provide an overall summary of treatment related differences where either early toxicity in one treatment group may be balanced by later improvements in disease control or small but consistent differences may occur over extended periods of time. Traditional MANOVA or mixed effects model derived test statistics focus on differences at specific points in time, but may present a confusing picture of the overall treatment differences. Several summary statistics have been suggested which are potentially sensitive to consistent trends in the same direction across multiple endpoints and control the Type I error rate. O'Brien (Biometrics, 1984) suggest weighted averages of individual t-statistics where the weight are based on the correlations between the endpoints. Pocock et nl., (Biometrics, 1987) The measurement of outcome of treatment is important in clinical practice particularly in relation to treatments of high costs. Traditionally, the measurement of outcomes of treatment have focused on survival rates, test results, return to work figures and clinical judgements of outcome, neglecting patients' perception of their state of health over time. The aim of this study was to analyze the effect of CABG on patients' perception of quality of life before, 1 week after and 3 months after grafting. Ninety-eight consecutive patients were asked to participate in the study at the same time they received the date for operation. Fifty-eight patients (64%) were willing to participate. Mean age of the total sample was 65 years (range 48-80, SD + 7.4). Eighty-five per cent of the sample were men. The patients filled in a questionnaire before, 1 week and 3 months after CABG, consisting of (1) demographic variables; (2) one overall coping item; (3) one global health item; (4) Nottingham Health Profile (NHP) part I (38 items) measuring perceived distress in relation to physical mobility, pain, energy level, emotional reactions, sleep and social isolation [higher scores (max 100) indicate more problems]; and (5) Angina Pectoris Quality of Life Questionnaires (APQLQ) (22 items) measuring psychological and somatic distress, level of physical activity and life satisfaction in relation to heart disease (lower scores indicate more problems). Cronbach's a was 0.87 and 0.91 for the NHP and APQLQ respectively at the first measurement. Before operation the profile of the NHP scores ranged from 7-22 with less limitations in the social domain and most limitations in sleep domain. Results from the APQLQ showed most problems in relation to physical activity. One week after operation there was a significant deterioration in the sleep and physical mobility domain measured by the NHP (p = 0.000). The results of the APQLQ showed significantly more problems in relation to somatic distress and physical mobility (p = 0.000). Three months after operation there was a significant improvement in energy, pain, sleep and physical mobility domains compared to the measurement after 1 week (p < 0.005) and a significant improvement in energy, pain and physical mobility compared to before CABG (p < 0.006). All the disease specific dimensions measured by the APQLQ showed significant improvement both between 1 week and 3 months after CABG and before CABG and 3 months after (p < 0.001). Perception of overall coping and global health improved significantly from before CABG compared to after 3 months (p < 0.001). Improvements were evident in perception of general health status, life satisfaction and disease specific limitations 3 months after CABG indicating that this treatment is effective according to the ratings of the patients. Factor analysis for assessing validity P. Fayers and D. Machin MRC Cancer Trials Office, 5 Shaftesbury Road, Cambridge CB2 2BW, UK Factor analyses, together with the related techniques of principal components, confirmatory factor analysis, structural equation modelling, and multitrait-multimethod analysis are frequently used in quality of life (QOL) research. In particular, these methods are commonly used when exploring the 'construct validity' of a new QOL instrument and, to a lesser extent, to derive scoring algorithms. The current usage of factor analysis is reviewed, and problems inherent in the application of factor analysis techniques to QOL instruments are discussed. In particular, there are many examples of cases in which QOL studies have over-used and misused factor analysis techniques. Practical issues and problems are covered, including sample-size estimation, scaling of item-responses, and the instability of both factors and factor scores. One often overlooked issue is the impact of the definition of the population and the selection of the samples for assessing subscale validity. The importance of this is illustrated using data from several MRC randomized clinical trials in cancer. Technical issues include the determination of the number of factors, the skewness of many QOL items, the discrete nature of many QOL items (frequently four-or five-point responses), linearity and scaling assumptions. Over recent years there have been many advances in the methodology of factor analysis. However, most of these developments have taken place in specialised areas of psychometric test theory, and appear little known to most practitioners of QOL studies. We review the current state of the art and discuss the controversial issues and problems that remain with factor analysis when applied to QOL data. Whilst recognizing that factor analysis can sometimes provide useful insight into the structure of a instrument, we appeal D. Feeny, G. W. Torrance, W. Furlong, C. Goldsmith, S. DePauw, M. Boyle and M. Denton The primary objective is to provide estimates of utility scores of health-related quality of life for the most prevalent health states in the Canadian population as assessed using the Health Utilities Index (HUI) Mark 3 system. The secondary objective is to provide data for assessing the predictive validity of multi-attribute utility functions estimated for the Mark 3 system. The HUI approach provides both a method for assessing health status (in population and clinical evaluative studies) using a multi-attribution classification system, and a method for valuing that health status, using multi-attribute utility theory The HUI Mark 3 is comprised of eight attributes with five or six levels per attribute. HUI-3 defines 972,000 unique health states. A preference measurement survey, involving 248 randomly selected respondents, was conducted in 1994. The survey involved the use of two preference measurement instruments: the visual analogue scale (feeling thermometer), and standard gamble (a new version of the chance board). Standard gamble scores of prevalent health states were measured on a common (but not conventional) scale, perfect health = 1.0 and PITS = 0.0. Perfect health (PH) is defined as the state having the highest functional level on each of the eight attributes and PITS is defined as the state having the lowest level of function on each attribute. PH (utility = 1.0, by definition) is the most prevalent health state, reported by 29.7% of the general population. The five most prevalent health states cover 63.8% of all respondents in the 1991 Canadian General Social Survey. The 12 most prevalent states account for 75.4%. States below the twelfth ranked in prevalence each account for < 1% of the population. In total 950 unique health states were identified. The estimated mean utility scores on the PH/PITS scale (n = 14-247) for the 12 most prevalent health states range from 0.71 to 1.00 (medians from 0.75-1.0). Mean measured utility scores for each of the 74 prevalent health states are presented and range from 0.20-0.86. These directly measured utility scores are being used to construct indices of population health status and will be used to assess the validity of HUI-3 fitted multi-attribute utility functions. The HUI provides a method for linking population health and clinical evaluative studies. Portuguese cultural adaptation and linguistic validation of the Sickness Impact Profile (PSIP) for a more cautious and restrained usage of these techniques, and for a greater recognition of the issues involved. Recently, the interest in health status measurement and, in particular, in health-related quality of life has greatly increased. As a consequence, there is an increasing need for assessment and standardization. Answering to this concern, the European Research Group on Health Outcomes (ERGHO) initiated its activities aiming at establishing the criteria for the selection, adaptation and cross-cultural validation of health status and health outcomes instruments in use at different EU countries and, even in each country, to populations with distinct cultural features. Cultural and linguistic differences produce a deep effect in the way people see the concepts of health, illness and sickness. In different cultures, different words and sentences have different meanings. We also present some examples of difficulties we passed through in our experiences of cultural adaptations. We conclude by describing some strategies to use health outcomes instruments originally developed in other countries and cultures. The awareness of health care professionals of the need for health-related quality of life evaluations has increased during the past decade. Linguistic and cross-cultural issues are of major importance in the developmental process and validation of health-related quality of life instruments. As a consequence, the process of translating from the original to native language of such questionnaires is crucial, since it has to ensure semantic, conceptual and technical equivalence. Here the translation process of the UKSIP into Portuguese was carried out and it s linguistic validation will be reported. The UKSIP (Salek 1990 ) was translated into Portuguese by a postgraduate student, a clinician and a professional translator. Then a backtranslation was carried out by an independent translator employed at the British Council in Portugal. The original version, the translation and the backtranslation were reviewed by two bilingual senior academics in order to produce the development version. This was, in turn, subject to evaluation by a panel (nine clinicians, five senior academics, four final year medical students, three nurses and one priest) in order to rate every single statement of the Portuguese version according to linguistic clarity, completeness and comprehensiveness. According to the panel results, the mean rates for linguistic clarity and comprehensiveness were 4.51 and 4.43, respectively from a scale ranging from 0 (very bad) to 5 (very good). The only chances suggests to the panel to development version was grammatically. Finally this development version was entirely approved by a specialist in linguistics from Coirnbra Universrty. As a consequence, the process of validation of the developed questionnaire was undertaken in order to achieve a reliable and reproducible instrument for health-related quality of life measurements in Portuguese population. A study was carried out in the Maternity ward of the Coimbra University Hospital to measure health outcomes of women in different periods of their pregnancy. The major research objectives were to gain knowledge about the natural history of pregnancy and to study the sensitivity of the Portuguese version of the MOS SF-36 when measuring the first, the second and the third trimesters of the pregnancy, as well as the post-partum phase. Pregnant women constitute a cluster of individuals characterized physiologically and psychologically, experiencing symptoms related to hormonal changes at the endocrine level, the increase of the uterus and emotional state of various types. In this paper we present an empirical study with 930 women in different phases of their pregnancies and where principal components factor analyses were used to validate a Portuguese translation (conceptual equivalence) of the well known health outcome measure, the Medical Outcome Study SF-36. In general, we found the same factor dimensions as the authors. In fact, the factor model for the whole sample allows us to draw similar conclusions to those obtained by Ware and colleagues. However, there is a different behavior for each period, especially between the second and the third trimesters. When we move from one period to the other, we assist to shifts on the cutoff points of the physical functioning from heavy/non-heavy to light/non-light activities. In this paper we present a study aimed at gaining more knowledge about the impact pregnancy has in a women and at evidencing the utility functional health measures have in supporting decisions regarding the care provided for a pregnant woman. This research project was a crosssectional study where we tested and validated the Portuguese version of the Dartmouth Pregnancy COOP Charts, an instrument intended to detect health and social problems during the pregnancy and the post-partum period. After following the translation and cultural adaptation processes, we gave the Portuguese COOP Pregnancy Charts to 930 pregnant women. The age distribution was normal (mean = 26.9 years; SD = 5.3 years), 76% never had any kind of complication, 14% were in their first trimester, 27% in the second and 35% in the last trimester of their pregnancy. To test the instrument we studied its sensitivity by analyzing the per cent of women rating high in the charts, and by comparing the distributions via the Kolmogorov-Smirnov test. Especially between the second and the third trimesters we detected significant differences in the variables related to the dimensions of mental health and impact of pregnancy in the perception of physical well-being. ritis patients enrolled in a 6-week health educationprogramme. In a pre-baseline assessment, patients will complete a number of health status and attitudinal measures that have been suggested as possible predictors of response to health education programmes. These will include patient report of: (1) frequency and intensity of pain; (2) patient global health assessment; (3) patient health state preference measures for current state and improved state; (4) importance and value of improved self-management; (5) patient self-efficacy for specific arthritis self-management; (6) patient health locus of control; and (7) patient symptoms of depression and anxiety One-week test-retest reliability showed high coefficients and the data spread all over the scale. The conceptual equivalence was also tested by using the factor analysis technique and we found the same dimensions has the authors. Finally, the criterion equivalence was tested by comparing the results obtained by this instrument to the results obtained by the utilization of the MOS SF-36 at the same time and by the same women. We conclude by defending the utility of the Portuguese Pregnant COOP Charts in the detection of psychosocial problems during the pregnancy. Our hypothesis is that all of these variables will be strongly associated with positive patient outcomes. The health education intervention will be delivered both by mail and telephone and will be modelled after the Arthritis Self-Management Group Programme (Lorig and Fries, Stanford Medical School, Palo Alto, CA). Patient outcomes to be measured include: patient completion of the programme; patient satisfaction with the course; decrease in patientreported pain; reduction in patient-reported health care utilization; and positive change in patient reported health status. Follow-up will take place at week 3 of the programme, at the close of the programme and at 2 months following the programme. As health care dollars become more scarce, it will become necessary to target these interventions to the patients who are the best responders. While the literature reports numerous assessments of differing rates of positive and negative outcomes for patients enrolled in various patient education and self-help programmes for arthritics, studies such as this will take place more frequently to aid in determining a priori which patients are most likely to respond with positive outcomes. Assessment of the health status in post-polio syndrome patients in developing outcomes for a rehabilitation programme: A preliminary report Using health and attitudinal measures to predict response to health education interventions L. Finch, A. Venturini, D. Trojan and J. Lemoignan Institute and Hospital, Montreal, Quebec, Canada S. K. Fifer Technology Assessment Group As health education and patient self-management programmes become more sophisticated and more widely used, much of their ultimate value will depend upon targetting these programmes to patients most likely to benefit from them. This study will address the potential relationship between health status and attitudinal measures and patient response to educational programmes. We will present preliminary results from a prospective study of approximately 100 arth-Individuals who suffered poliomyelitis can develop new symptoms of weakness, pain and fatigue several years after their polio (post-polio syndrome or PI'S). Knowledge of the functional consequences of the later effects of polio is essential in managing PI'S, as is the recognition of potential psychosocial problems. The routine use of a health measure coupled with clinical performance outcomes and long term follow-up would provide the necessary knowledge to design and monitor therapeutic interventions. The objective of this presentation is to define the overall health status of PI'S patients referred for therapy and determine a specific programme or programmes of intervention. cover the basic dimensions of QOL as well as diseasespecific problems and which are psychometrically well tested and internationally available. We report about the psychometric testing of a questionnaire for the assessment of the QOL of patients with intermittent claudication (CLAU-S). The questionnaire consists of 80 mostly Likertscaled items covering the following dimensions: everyday life, pain, complaints, social life, disease-specific anxiety and psychological well-being including depression, fatigue, vigor and anger. The CLAU-S was carefully translated into four languages (German, English, French, Flemish) and psychometrically tested in cross-sectional trials in each country. About 100 patients were included into each validation study and filled in the CLAU-S questionnaire and the Nottingham Health Profile (NHP The results of the reported cross-sectional validation studies in four countries demonstrated that the CLAU-S is a reliable and valid instrument for the assessment of the quality of life of patients with intermittent claudication in international clinical trials. (Requests for information concerning the CLAU-S should be addressed to S. Comte.) At the initial evaluation, all patients referred to a PI'S outpatient therapy clinic are surveyed with the Short Form Health Survey (SF-36). Thirty-six patients with an average age of 55k9.19 years were surveyed over a 6 month period. Performance measures of mobility, gait speed and walking endurance were available for 14 of the 36. Responses on the SF-36 were analyzed descriptively and the percentage of patients one standard deviation (SD) below the USA population norms in each health survey domain were determined. This level was taken as an indication of an area of concern for potential intervention. Performance measures were analyzed descriptively and Pearson Product Moment Correlations were used to relate performance to the health status scores. Greater than 50% of the PI'S group were one SD below norms in concepts of physical functioning (77%), physical role (61%), bodily pain (58%) and vitality (50%). The concepts of least concern were mental health (22%) and emotional role (36%). However, 47% had difficulty with social functioning and 42% perceived their general health as poor. Only four of the 36 had a 'normal' health status. All of the performance measures except mobility (r = -0.07) correlated with physical domains but not with mental health domains. In addition, physical functioning and bodily pain were highly correlated with vitality (r = 0.76 and 0.77, respectively). The main impact of PI'S is on physical functioning, especially in the area of endurance. A therapy programme has been developed over the last 6 months to address the problems of physical function, pain and vitality. The programme focuses on improvement of work and home environments to reduce fatigue and physical stress, pain control, energy conservation and fitness exercises, with education on factors which limit function and endurance. The SF-36 will be used continuously as an outcome measure to assess the success of this programme. The ongoing use of a health status measure provided the knowledge to develop a therapy programme for a specific PI'S group. This programme needs to be closely monitored and evaluated. Assessing the quality of life of patients with intermittent claudication: psychometric properties of the claudication scale (CLAU-S) We prospectively followed 86 patients with definite systemic lupus erythematosus. We used five validated lupus activity and two damage measures as predictors. Outcome measures were the physical function scale of the HAQ and the eight subscales of the SF-36. All instruments were administered monthly for four months (five visits) except for damage which was only measured at baseline. For the cross-sectional study, laboratory tests were the erythrocyte sedimentation rate (ESR), third and fourth complement fraction and anti-double stranded DNA. At baseline, Pearson's correlation were calculated for each of the health status measures and each clinical and laboratory parameter. To determine predictors of physical function and general health, different combinations of one activity and one damage measure were tested using age, disease duration in months and level of education in years as covariates. Analyses were carried out using BMDP5V based on compound symmetry model for covariance. Wald X-square tests were used to assess significance of the individual predictors. There were 78 (91%) females, mean age was 45.1k13.7, median education level was 11 years and mean disease duration was 12 years f7.1. Means for the subscales of the SF-36 were consistent with those found in serious medical illnesses (PF = 64.33, RI' = 48.53, BP = 45.71, VI = 51.49, SF = 74.11, RE = 69.02, MH = 68.94 and GH = 57.71). Crosssectional study: (1) HAQ correlated significantly with all eight subscales of the SF-36 (p < 0.005); (2) none of the laboratory tests correlated significantly with HAQ or SF-36; (3) lupus activity measures behaved differently one from another with one correlating both with HAQ and SF-36, one with SF-PF only and three with none; (4) the two damage measures correlated with SF-PF only. Predictive study: (1) all five lupus activity measures were associated with most subscales of SF-36 (p < 0.001) and with HAQ (p < 0.01); (2) the two damage measures only predicted physical function as measured both by SF-PF or by HAQ (p < 0.001). Cross-sectional studies reveal differences among lupus disease activity measures. In a prospective study and using repeated measures, change in activity measures reflected change in health status and physical function while baseline damage only predicted change in physical function. Health status measures can be used to validate and test hypothesis in chronic illnesses research. The aim of this research is to develop, or adapt, an instrument to be able to measure the impact of disease control programmes on the health-related quality of life (HRQOL) of Kenyans, using two ethnic groups as case studies, and be able to use the tool in cost-effectiveness analysis. The objectives for the first phase of research are: first, to describe the lay perceptions of HRQOL; and, secondly, to reflect on the appropriateness of using Western based generic instruments in a Kenyan context. The research is being conducted in two regions of Kenya; in Western province (Evojo and Chavakali sub-locations) and in Eastern province (Kambu sub-location). Given the paucity of research in non-Western countries, we have pursued a range of research methods from relatively structured questioning, to unstructured interviews and participant observation fieldwork. Each of these methods has been aimed at providing information m three areas: general perceptions of health and well-being in the community; personal accounts of health, illness and well-being in the individual and family; and general descriptions of the local area such as agriculture, occupations, access to water, etc. The multiplicity of methods, some of which are applied for the first time in the HRQOL field, has enabled us to begin addressing a range of themes which have been largely ignored in the HRQOL literature. We recognize there is still much to accomplish over the next four years of the research, especially in the investigation of meaning in local language, before we can develop techniques to develop/adapt a comparative measure of HRQOL. Nevertheless, research so far indicates it may be productive to explore eight broad areas: key resources, relationships, bodily well-being, access to and confidence in therapy, expectations, time and differentiation. It may appear that the above categories fit 'neatly' into existing patterns of the HRQOL literature. However, the range of ideas being explored under each of these headings is significantly different in certain cases. For example, 'relationships' includes physical contact with friends and relations, links with the spirit world/religious beliefs, the stability and predictability of family relationships, certain social stigmas, customs of avoidance (e.g. of in-laws) and notions of well-being connected with inter-personal concerns about both the living and the dead. This paper will briefly summarize the principal findings and raise a series of questions for discussion. The predominant concern raised as a result of our research to date is that there are fundamental problems with the 'H' in HRQOL. Much of the literature treats 'H' in the context of biomedical disease control programmes. Some of it treats it in a more general way, linking it with the WHO definition. However in practice this does not reflect local conceptions of health, but instead mirrors mtemational thinking about what health 'ought' to be. Local conceptions of health may not relate to the first of these and may contradict the latter. The objective of this study is to investigate the quality of life (QOL) and medication adherence rates of people with a clinical diagnosis of schizophrenia 3-6 months postdischarge in the community. People suffering from chronic schizophrenia were recruited in hospital and interviewed using a semi-structured questionnaire 3-6 months postdischarge in the community. Reporting was validated using information from hospital and community case notes and carers (where possible). The study was carried out in psychiatric wards in four hospitals in the City and East London health district and the local community. Subjects were inpatients aged 18-64 years with a clinical diagnosis of schizophrenia and prescribed an injected (depot) form of antipsychotic medication. Patients excluded from the study were those with a history of organic brain disease, those who were unable to complete the questionnaire due to English language difficulties, those who self-discharged against medical advice and those who were discharged outside the study district. Main outcome measures were DSM-III-R classification of diagnosis, severity of illness using the Global Assessment of Functioning Scale, Lancashire Quality of Life Profile, record of attendance at clinic for depot medication. The follow-up sample mainly consisted of people who experience repeated admissions, i.e. 'revolving door' patients. At follow-up, attendance at clinic for depot medication was measured. About half of the population attended regularly and the rest of the sample attended irregularly with some stopping their medication completely. The subjective global well-being scores were compared between groups: about a third of the regular attenders felt 'mostly satisfied/pleased/best life imaginable' compared to about half of those people who attended the clinic irregularly. Initial analysis of the results suggests that those people who regularly receive depot antipsychotic medication do not report as high subjective well-being scores as those who do not always attend. Mental health assessments in patients with facial paralysis R. Freedman, S. Scott, B. Segal, I. Danys and J. Shiau Unilateral facial paralysis (due to Bell's Palsy, surgical trauma, etc.) is characterized by varying degrees of paresis and involuntary facial movements, which may cause physical discomfort, disfiguration, and sometimes psychological and emotional stress. The objective of the present study was to examine the relationship between degree of facial paralysis and mental health. This was done as part of a larger study which aimed to evaluate the effectiveness of neuromuscular rehabilitation of patients with chronic (0.5-20 yrs) facial paralysis. Patients who were treated at a facial nerve rehabilitation clinic during the previous 5 year period were randomly divided into two groups. In the first group, one assessor interviewed patients and administered two questionnaires, while a second assessor blindly rated degree of facial paralysis. In the second group, the assessors reversed their roles. Thirty-eight of 126 eligible patients agreed to participate. First, patients completed an internationally-used general health questionnaire (SF-36 by the International Resource Centre for Health Care Assessment). Next, patients completed a specifically-designed facial-paralysis questionnaire. Mental health was assessed using the SF-36 mental-health subscore, and using the facial-paralysis questionnaire subscore which rated how often in the past 4 weeks patients had been 'concerned or worried' about their 'facial paralysis'. Patients perception of their paralysis was measured by a second facial-paralysis subscore obtained by marking a scale of O-10 (0 = complete paralysis; 10 = normal). Finally, severity of facial paralysis was blindly rated by the second assessor using the international House grading system (Ofoiaryngof Head Neck Surg 1985; 93: 146-7) , which characterizes facial paralysis on a six point scale (6 = total paralysis; 1 = normal facial function). Results from the questionnaires were analyzed with Spearman correlation analysis, using SAS. Although patients exhibited the full range of severities of paralysis, it was found (1) that the SF-36 mental health subscore for all patients tended to be average (mean = 73; SDk17.0); (2) that this subscore was significantly correlated (Rs = 0 46, p < 0.002) to the facial-paralysis concernworry subscore, suggesting a pronounced influence of facial concerns on a patients mental health; (3) that the SF-36 mental health subscore was significantly correlated (Rs = 0.48, p < 0.001) with a patient's perception of their paralysis, and (4) that the SF-36 mental health subscore was not correlated with a patient's House grade, which rated actual degree of facial paralysis. Thus, it appears that it is a patient's perception of the severity of their facial paralysis, and not the actual severity of their condition, which most strongly influences their mental health. it is important to discuss and promote tools of decision making. As far as economic analysis is concerned, cost-utility analysis is more and more used rather than cost-efficiency analysis. Quality of life measures are involved in cost-utility analysis. The questions are: Which measure of the quality of life? By who (the patient, the physician, the society)? For what choices? Between priorities among health problems like in the case of QALY in the UK or in Oregon, or part of a decision making process between techniques or protocols concerning one single problem? These questions are those of QUALIDOVIE (which comes from QuALIte-DOmieile-VIE: QALY-Home-Life). We have done a measure of quality of life in two populations: people with cancer in a day care hospitalization scheme and people with hip fracture in a rehabilitation medical service at the hospital. The sample is of 46 and 37 persons at the present time. We used four measures of quality of life (Kamofsky index, EORTC QLQ-C30, STAS, EuroQoL) at three times (entry, 20th day, 40th day). So we can compare the value attributed to the quality of life and their evolution, estimated by the doctor and in terms of physical performance (Kamofsky), by the person (EORTC), by the nurse (STAS, which is also or mainly a measure of quality of care), by the society (EuroQoL). The objectives are (1) to compare the results of these measures (level, evolution, correlation between indicators) and (2) to compare what people say about their quality of life with the questionnaires, on the basis of interviews by psychologists at the same moments. Few studies have been made on such a comparison, those existing being generally limited to two indicators. Our results show (objective one) that the levels are different and that some correlations are very weak, like between the value by the physician and by the person herself, but also with the social value. This is discussed from the point of view of the measure itself: person giving the value and specificity of the indicator, difference between health status and quality of life. It is also discussed from the point of view of the economic analysis in a decision making process. This implies to think of the appropriate use of each tool; individual and social choices do not require the same one. Nevertheless the general trend of the evolutions is similar. Concerning the second objective, our results show that about 70% of what people say on their QOL is not in the questionnaires. What do they measure then? How can we reconcile the individual and subjective approach with the aims of the measure for collective or social choices? Using SF-36 to monitor the effect of delisting pharmaceuticals from the Ontario drug benefits (ODB) formulary The objective of this project was to assess the clinical decision to remove all sustained release (SR) non-steroidal anti-inflammatory drugs (NSAIDs) from the ODB formulary. The Pharmacy Medication Monitoring Programme (PMMP) is a post-marketing surveillance network which recruits patients through community pharmacies as they are filling prescriptions. This allows for the monitoring of a broader sample of patients treated by a variety of physicians than found in most clinical trials. Patient health data are collected through structured telephone interviews using computer assisted telephone interviewing software. The Ministry of Health and Drug Quality and Therapeutics Committee critically reviewed the SR drug formulations in the formulary ultimately delisting all SR-NSAIDs in September 1993. PMMP had a network of ten pharmacies which began recruiting patients filling prescriptions for any SR-NSAID at the end of July. A consent form and a demographic form were completed by the patient in the pharmacy. Patients received their first follow-up 1 month post-registration and continued to receive calls monthly until they switched to another therapy or discontinued their SR-NSAID (max six calls). Switchers and discontinuers then received calls at 1,2,3, and 6 months after the change. Outcomes were SR-NSAID status and concomitant medication usage (including non-prescription over the counter drugs), new health problems and hospitalizations, health care usage, specific health problem status and Quality of Life (SF-36). The SF-36 is administered over the phone and takes approximately 5-8 minutes to complete. The entire interview is completed within 20 minutes. There were 135 patients enrolled, 125 (93%) completed their first follow-up interview and 80 (59%) completed all their interviews. Patient cohorts were determined by which drug therapy they were using at the time of the interview; CONT: still using a SR-NSAID; SWD: switched to another NSAID formulation; or D/C: discontinued therapy. Comparisons were performed between CONT-SWD, and CONT-D/C across the eight SF-36 measures. Six of the eight measures (PF, RP, BP, GH, VT, RE and GH) were significantly lower for SWD compared to the CONT (max p = 0.015). Social function (SF) was significantly higher for SWD compared to CONT (p = 0.008). There were no significant differences in any of the eight measures when comparing CONT and D/C (min p = 0.08). Switching from an SR-NSAID to an alternate NSAID is associated with a lower Quality of Life score than when continuing with the SR-NSAID. Discontinuation of therapy is more often associated with resolution of the clinical problem, but with no detectable change in the Quality of Life score compared to patients continuing to use their SR-NSAID. is widely recognized as a critical tool in biomedical research, but the discipline has traditionally focused on outcomes that are dichotomous in nature: the disease is either present or absent, the patient either lived or died. Further, the use of measures such as mortality rates, standardized mortality ratios, and survival analysis for the assessment of life expectancy ignores variations in morbidity that have a direct bearing on quality of life. In recent years there has been a growing consensus for the use of a more broad-based outcome measure. Many problems in epidemiology and public health statistics may benefit from this approach. This paper introduces a project designed to perform extensive analyses in an ongoing epidemiologic cohort of independent living, older adults from a Southern California community using a broad-based outcome measure. Specifically, in this project we will: (1) examine the cross-sectional relationships between the physiological, psychosocial, behavioral, and medical history data obtained in this cohort and health-related quality of life as estimated by the QWB; (2) determine how past physiological, psychosocial, behavioral, and medical history data collected on this cohort predict health-related quality of life; (3) evaluate the feasibility of calculating the quality-adjusted life expectancy of cohort subgroups US: ig the methods described in Healthy People 2000; and (4) st two models to predict QWB scores using other data collected at the concurrent visit of this cohort. We are approaching all surviving members of the original 1972-1974 cohort who are now over age 50 years (n = 3,044). Over 90% of subjects respond to an annual mail questionnaire and agree to telephone follow-up if necessary. The follow-up for vital status has been 100% for this cohort. QWB interviews are obtained by telephone within 7 days of the current clinic visit. All data since 1972, including that of the current visit will be available for analysis. Specific analysis plans by specific aim are included in the m cystic fibrosis (CF) therapy leads to a significant increase in life expectancy. This extensive and time consuming treatment as well as the notion of living with this severe illness poses great demands on the patient and his family Psychosocial counseling and support for these patients are evolving areas for professionals in hospitals and ambulatory care settings. The aim of the study was to investigate if a patient education programme helps to increase quality of life (QOL) for the patient and his family. Is it possible to transfer patient education programmes developed for other chronic diseases to this special group? Are there certain family or coping characteristics that help improve QOL? What kind of psychosocial support is needed for these patients? Do the categories used in systemic family counselling apply to this group or need they be changed? This study is part of a public health research project. Out of the 210 CF patients and their families who have been recruited from four German CF centres (two in the East and two in the West) for this study, 93 families were especially tested with the newly developed test for family structures (FAST). Family structure theory assumes that structural properties of families (measured in dimensions of cohesion and hierarchy) detect balanced, labil-balanced and unbalanced families. Besides the usual situations (typical, conflictual and ideal family situations) a health crisis situation was also tested. Severity of illness was assessed with the Shwachman score, QOL was measured using a German instrument by M. Bullinger (Alltagsleben). In balanced families the severity of illness of the child as measured with the Shwachman score only has minor effects on the QOL of the mother. Only in unbalanced and labil families does the severity of illness of the child influence the QOL of the mother. This influence however is not linear, medium severity leads to a lower overall QOL compared to the severe or mild degrees of disease. There was no significant correlation between QOL of the mother or of the child with any of the tested family structures. Even for the QOL subscales no significant effects could be detected in regard to family structure. Especially families who are freshly confronted with the diagnosis seek counseling and guidance. Stable families are less influenced by the challenges posed by this illness. Counselors however can not advise to a certain family structure as different families cope differently without a substantial change in QOL. 1992; 4: 92-101.). The Lehman interview is a lengthy instrument that provides subjective and objective scores in seven domains. The Frisch interview is much briefer and provides an overall QOL score based on items weighted for importance by individual respondents, QOL interviews were scored according in authors' instructions. Pearson correlations were calculated between Lehman domain scores and relevant Frisch items. Statistically significant correlations were obtained for the subjective domains of living situation, social relations, finances, and global scores, and for the objective domains of social Relations and health. These findings suggest that it may be possible to develop a more brief composite interview consisting of the Frisch instrument, and some objective items from the Lehman instrument. Such an instrument might be better tolerated by people with serious mental illness. In such cases Phase II trials provide an opportunity for empirical testing of several questionnaires. The instument(s) selected should satisfy the following criteria: (1) sensitivity to variation in the clinical endpoint; (2) sensitivity to treatment side effects. In addition the selected questionnaire should be reliable and demonstrate internal validity and independence. At present, there are no QOL studies that have investigated the impact of platelet engraftment after PBPC, on patients' QOL immediately after treatment and during the recovery period. In preparation for PBPC Phase III trials, we pilot tested a number of instruments measuring patients' quality of life in a Phase I/II tnal. Four QOL instruments were tested: the SF-36; the EORTC-QOL-C30; the FACTG-B-BMT; and the EuroQoL. Because the data consists of multiple observations over time on single individuals, panel data analysis was used to control for within-subject variation. The adjusted R* from the regression of QOL on platelet counts was used as the selection criterion. Based on a sample of about 60 patients for a total of 375 observations, the following QOL instruments and domains had the highest R* scores: the general health domain from the EuroQoL (adj R* = 0.71); the FACTG (whole instrument) (adj R* = 0.68); and the general health domain from the SF-36 (adj R* = 0.68). We specified criteria for the selection of the Phase III QOL instrument based on empirical test of several questionnaires. This selection method is performed on the whole sample in a blinded fashion, and therefore does not bias the selection with respect to treatment. Once domains are selected it is important to test for independence. coping; marital and family outcomes; and social activity). Survival curves were generated with the Kaplan-Meier method and were compared by means of the generalized Wilcoxon test. Proportions were compared using the chi-square (Yates correction when indicated). The characteristics of the sample prior to OLTX were: 35.6% of the sample were > 64 yrs old; 58.9% were male; indications for liver transplantation were: cholestatic disorders 19.2%, parenchymal diseases 69.9%, and primary liver malignancy 11.0%. At the time of the operation, 74% were UNOS (United Network of Organ Sharing) class 3 (in the hospital). Of the entire group, 13.7% received two grafts. The actuarial patient survival rate was 67.12% after 4 years. Of the survivors, 68.3% were able to carry on normal physical activity, and daily Living activities were not affected in at least 63.4%. Pain was reported in 27.0%, and sleep complaints in only 7.9%. Overall, emotional as well as psychosocial status was good in almost 70% of the survivors. Survival rates were not influenced by age, gender, nor UNOS status. When the indication for transplantation was considered, the survival was significantly worse (p < 0.001) for primary malignancy. When we compared the survival rate (after 4 yrs) of elderly patients with that of adult patients who had undergone OLTX in the same year, survival was shghtly more favourable in younger patients (75.93%), but the differences were not stahstically signifi- Although psychosocial aspects of skin diseases are well known, no validated questionnaire suitable for use in clinical trials is available to assess the impact of acne on healthrelated quality of life (HRQOL) or to evaluate therapeutic change. Scientific methods were used to develop such an instrument, and included item generation, item reduction and pilot-testing phases. Initially, 168 possible HRQOL items were generated based on feedback from interviewing acne subjects and dermatologists as well as literature review. To reduce the number of items, 165 subjects with acne aged 1339 years were recruited from two dermatology clinics and a work site. Subjects identified which items affected their quality of life and rated the importance of each item on a 5-point scale from 'not very important' to 'extremely important'. The 168 items were reduced to 34 key items by evaluating patient-perceived importance (frequency item was selected multiplied by mean importance rating) and applying principal components factor analysis to identify meaningful domains. After pilot-testing the 34-item questionnaire, further wording revisions were made to improve clarity, reduce redundancy and broaden relevancy of questions, resulting in a 24-question instrument which takes about 10 minutes to complete. Questions in the reduced instrument are those which acne patients identified as important. All questions pertain to how acne affected aspects of the patients' jives in the past week, with response options on a seven-point scale. Four domains were identified: 1 year, l-2 years, and over 2 years). The 24-item QOL-BMT scale has a lo-point format per item. Scores available include a single IO-point measurement of quality of life, subscale scores for physical well being, psychological well being, social well being, and spiritual well being. Overall quality of life scores are calculated by summing all items and dividing by the total number of items, to produce a score ranging from 0 (lowest quality of life) to 10 (best quality of life). Psychometric evaluation of the QOL-BMT revealed test-retest reliability (Pearson's r = 0.71) and internal consistency (Cronbach's 0: total scores r = 0.85). Content validity by experts reached 90% agreement. The mailed questionnaire was returned by 174 patients, who had a mean age of 35 years with a standard deviation of 10.4 years. Allogeneic transplants occurred in 88 (50.6%) and autologous transplants in 86 (49.4%). Twenty-four per cent of the population reported having graft-vs-host disease. Forty-six per cent had not returned to work, and 47.4% reported that the bone marrow transplant had interfered with their personal financial goals. Prior to the BMT 63% were married, 28% single, 5% divorced and 4% had other status. At the time the questionnaire was completed, 63% were married, 23% single, 7% divorced, and 7% had other status. The psychological well-being subscale consisted of eight items, had an o! of 0.72, total mean score of 7.28 with scores ranging from 8.1 for an item on 'how much enjoyment are you getting from life' to 4.4 for 'how distressing is your illness to your family'. The physical well-being subscale consisted of seven items, had an alpha of 0.68, and an overall mean score of 7.7 with scores ranging from a high of 9.4 for 'do you have any bleeding problems' to 6.4 for 'how much fatigue do you have'. The social well-being subscale consisted of five items, had an alpha of 0.56 and an overall mean score of 7.4 with scores ranging from a high of 8.7 for 'is your health interfering with your relationships' to a low of 6.5 for 'how much strength do you have'. The spiritual well-being subscale consisted of four items, had an a of 0.76, and an overall mean score of 7.45 with scores ranging from a high of 8 3 for 'do you sense a purpose for being alive' to a low of 6.4 for 'how important are religious activities'. When autologous and allogeneic subpopulations were compared, no significant differences were found in any subscale or total scale scores. No significant differences were found in time since trans-University of California, Los Angeles, CA, USA and City of Hope, Medical Centre Over the past decade, gender issues have become important topics in health sciences. Gender-related research has focused mainly on the healthy population. In a sample of 162 females and 65 males diagnosed with cancer, gender differences in patient outcomes were investigated. Quality of life (QOL), physical health, and functioning were assessed using the Quality of Life Cancer Scale (QOL-CA), the Kamofsky Performance Status (KPS), and the OARS-Multidimensional Functional Assessment Questionnaire (OMFAQ). Statistical analyses were performed using xsquare and two-sample t-tests. The analyses revealed gender differences in physical functioning, economic resources, and some areas of QOL. Male patients showed significantly lower scores on the KPS than female patients. Compared to women, men reported more functional impairments related to disease and treatment and more limitations in performing activities of daily living (ADL) than females. There were no gender effects in the number of previous and current health problems nor in the impact of previous comorbid conditions on current health. Concerning quality of life, males showed significantly higher scores on the QOL subscale attitude of worry than females. Men were more likely to report financial concerns, although their incomes were significantly higher than those of women. However, no gender differences were found in the QOL subscales: symptom distress, psychosocial-existential well-being, and physical-functional well-bemg. The results indicate gender differences for certain aspects of physical functioning and QOL. Females performed better in ADL, and self-care and had fewer impairments related to disease. The study revealed financial concerns regarding medical care and treatment predominantly in males. ( Trial (DCCT). At one year post-transplant, patients were classified into four groups: K (with graft functioning, n = 64), KP (both grafts functioning, n = 30), KPF (pancreas graft loss-only kidney graft functioning, n = 11) and controls (n = 33). At one year follow-up, 92% of K, 93% of KP, 82% of KPF and only 25% of the control patients reported their health was better than a year ago (x square, rr < 0.001) From baseline to follow-up, K and KP groups had significant improvements in the physical (PCS) and mental (MCS) component scores of the SF-36 (paired t-tests, p c 0.05 PCS, p < 0.001 MCS), and in subscales: vitality (K, KP); social function (K, KP), health perceptions (K, KP); role-physical limitations (K only) and pain (K only). The KPF group had no significant differences between baseline and follow-up. Their PCS was down (worse), their MCS was up (improved) and the changes were slight. The control group had no changes in SF-36 scores. The K and KP groups also had significant changes in symptom and disease-specific measures (reduced depression symptoms, more satisfaction with health and treatment), but no changes in global QOL measures. When compared to K patients at follow-up, KP patients had significantly higher scores on a scale which combined DCCT satisfaction items related to health, diet and health impact on family (p c 0.001). In multiple regression analysis with covariates for age, gender, education and baseline satisfaction, pancreas transplant was the strongest significant predictor of the satisfaction score. Differences were not found between the K and KP groups on generic or global QOL measures. When QOL at 1 year after surgery is compared to pretransplant levels, both Kl' and K recipients have large improvements in their physical and mental health-related QOL. A similar impact is not found for global QOL. The Incremental benefit to QOL of KP over K is detectable only in diabetes-specific assessments. While this is a nonrandomized study, the internal validity of the results is supported by the lack of concurrent improvements in the comparison group of controls. The study was approved by the University's Committee on the Use of Human Subjects in Research. Funding was provided in part, by the Diabetes Research and Education Foundation and grant MOl-RR00400 from the NCRR/NIH. The aims of this study were to test the indexes (scales) of the EORTC QLQ-C30 Quality of Life Questionnaire for item bias (differential item functioning). Here, we define an index as the summation of the scores of two or more items. The formation of an index is often practical. However, the index formation may result in unintentional loss or distortion of information. Item bias analysis (IBA) tests whether this is the case. IBA is especially relevant when groups are compared: it can be used to test whether the information about differences between groups, obtained by the individual items constituting the index, is correctly passed on by the index score. This is not possible by means of the psychometric techniques traditionally used in quality of life research. The EORTC QLQ-C30 of 1,189 breast cancer patients were analyzed. In order to perform cross-validation of results, the sample was divided in two. IBA was done separately in each sub-sample. Only item bias found in both sub-samples is reported. IBA was carried out in three-way tables of item score, index score, and age or treatment (k chemotherapy). Thus, we tested for item bias in relation to age or treatment. In addition, we calculated Cronbach's CL. Statistically significant item bias in relation to age or treatment was found in four cases involving three items from three scales, the Physical Function, Pain, and Cognitive Function scales. Cronbach's CI could not be used to predict the presence of item bias. We found item bias, and thus possibly systematic loss or distortion of information, when testing the indexes of this questionnaire in common analytical situations involving the comparison of groups varying as to age or treatment. The detection of item bias can prevent possible misinterpretations of data. At present, we do not know whether our results can be generalized to other applications of this questionnaire, and we do not know the extent of item bias in other similar questionnaires. The clinical importance of IBA, and thus its place in the psychometric validation of quality of life questionnaires, depends on whether it will be proved that it prevents misinterpretations sufficiently often to justify the extra work involved. The purpose of this study was to adapt a generic quality of life (QOL) measure, the DUKE Health Profile (DHP), cross-culturally and to document the psychometric properties of the French version in a general population sample. The cross-cultural adaptation of the DHP (a 17-item selfreport questionnaire) was conducted through three translations independent of each other, each backtranslated to the original language for the detection of errors. A multidisciplinary expert committee examined this material, and produced a version presenting semantic, idiomatic, experimental and conceptual equivalence with the original. A final version was obtained after pre-testing with a few patients and healthy subjects. A cohort of 963 persons from the general population, volunteers to participate in a prevention trial (mean age = 48.5 yrs, F/M = 1.6) filled in this questionnaire. Nine hundred and three persons (93.7%) filled it in again 3 months later. The internal consistency of the French version was acceptable in all (Cronbach's CL = 0.63-0.81) but social dimension (0.54). The reproducibility in stable subjects (601 subjects unchanged on overall health VAS) was satisfactory (intra-class correlation coefficient = 0.63-0.78). Convergent validity was assessed by a significant correlation with overall health m all (r = 0.39-0.61) but disability dimension. Discriminant validity was assessed in 263 subjects reporting a chronic disease. They showed a significantly lower score in all but social and disability dimensions. The sensitivity to change was assessed by a significant modification of scores in the same direction as overall health change after 3 months in subjects improved (n = 128) or worsened (n = 187) on overall health VAS. A structured methodology was used to cross-culturally adapt a generic QOL measure. The psychometric properties of the French version are close to the original version. This short form proved simple to use in the general population. sought an understanding of how the women were affected by diabetes; how they managed their disease and their daily activities; how they felt about their treatment regimen and how this impacts on the quality of their lives. The study provides a general understanding of the experience of having and managing diabetes from the views of refugee women. The specific objectives of this study were to have refugee women with diabetes describe (a) how they define (perceive) the condition of diabetes; (b) strategies they use to manage their daily demands; (c) the effects of diabetes on their family members and the family unit as a whole; (d) their wishes/hopes and fears/worries in this life; (e) their overall life satisfaction; (f) their quality of life (QOL) with diabetes; (g) the factors which influence their management of diabetes; and (h) how the perception of their environment influences their management of diabetes. The research design included qualitative approach based on grounded theory. Data was obtained over a 3 month period through tape-recorded intensive interviews with 42 Palestinian refugee women with insulin dependent diabetes mellitus. The study was conducted at two Palestinian refugee camps that were established in Jordan in 1948 and 1967. Data collection took place over February to April, 1994. The interviews provided women with the opportunity to describe their experiences of being refugees and their perception of their QOL. All the interviews were transcribed verbatim and translated into English by the investigator. Data were analyzed using the content analysis and Ethnograph. Constant comparison method focused on the ways in which individual refugee women defined and managed their situation. Most women used similar strategies to manage their daily lives. Insulin was viewed as the most important element of management, followed by traditional medicine. Most women felt that diabetes has an impact on the different aspects of their life. However, hopefulness, helplessness and suffering were some of the common themes emerging from these data. Faith in God was seen as the most satisfying and helping element for most women. Other important elements included health, family and children, psychological and spiritual comfort, financial and social relations. Refugee women were satisfied with most of the areas they thought were important in their lives. Diabetes and the refugee status seemed to have a great effect on how refugee women view their lives and manage their diseases. To maintain quality of life and life of Palestinian refugee women in satisfactory metabolic control in Type II Jordan diabetes patients-one and the same? to be an indicator of QOL. It is therefore a challenge for health personnel to include the quality of life perspective in clinical practice to fulfil the intentions of the St Vincent declaration. patients' experience of quality of life (QOL). Much of the attention in daily diabetes care is focussing on satisfactory metabolic control and less attention is paid to QOL perception. However, if metabolic control is not an indicator of QOL then important aspects of patients' lives are neglected in diabetes care. The aim of this study was to analyze patients' perception of QOL compared to metabolic control (HbAlc). A correlation survey design was performed. So far, 110 Type II diabetic patients are included in the study. The sample comprises hospitalized patients (n = 31), patients from an outpatient diabetic clinic (n = 51) and general practice patients (n = 28). Fifty per cent of the sample are women. Mean age of the total sample is 63 years (range, 35-88, SDk12) and mean duration of diabetes is 9 years (range, O-28, SDk6). Forty-two per cent, 51% and 36% of the patients are treated with diet, tablets and insulin injection(s) respectively. Mean HbAlc is 8.6% (range 5.3%-16.4%, SD?2.2%). The patients filled in a l9-item Symptom Checklist. Quality of life was operationalized using the version of the SF-36 now called the Health Status Questionnaire (HSQ 2.0) and the Diabetes Health Profile (DHP II). The validity and reliability of the questionnaires are described previously as satisfactory and this is confirmed in the present study. The most frequent symptoms patients reported were: frequent urination (both day and night), thirst, fatigue and hunger. The HSQ 2.0 profile showed (optimal score 100): mental health (five items) 78.0 (+20.8), bodily pain (two items) 72.0 (*31.1), social functioning (two items) 71.6 (+29.0), physical functioning (IO items) 68.2 (*27.1), role emotional (three items) 62.0 (+38.5), health perception (five items) 54.0 (+25.4), role physical (four items) 51.6 (k43.0) and energy/fatigue (four items) 51.3 (f23.4). This pattern is very similar to reference data from people aged 65 and older. The DHP II profile showed (lower scores indicate less problems): depression because of diabetes (four items) 13.1 (_+ll.I), worry in relation to the results of blood sugar and urine tests (11 items) 31.3 (+17.1), and eating restraints problems (five items) 47.0 (i24.0). There was a significant negative correlation between scores of each scale of the HSQ 2.0 and frequency of symptoms (absolute values of the correlation coefficient ranged from -0.40 --0.59, p < 0.01). A significant correlation between level of HbAlc and frequency of symptoms was found (r=0.32, p = 0.006). Correlation analysis between the different scores of the HSQ 2.0 and DHP II and HbAlc showed a significant relationship between the score of general health perception (f. = -0.30, p = 0.008) indicating that a tendency to believe that personal health is poor is associated with higher levels of HbAlc. Further the amount of worry in relation to knowing the results of their blood sugar and urine tests (I = 0.25, p = 0.03) was associated with higher levels of HbAlc. However, in general these results indicate that the perception of QOL (operationalized as is this study) is not associated with metabolic control. Metabolic control (HbAlc) is associated with symptoms of higher levels of blood glucose. However it does not seem In analyzing the quality of life in UTS patients, short stature has to be taken into account as it is among others (e.g. infertility) the main feature of UTS. The aim of the study was to specify the pattern and intensity of psychosocial burden associated with short stature at different developmental stages and different growth patterns. Therefore we developed a specific parental questionnaire focusing on short stature associated problems which was answered by 177 parents of UTS patients of different age groups (age 6 months-40 years). Nearly all parents reported general short stature associated problems (81% in preschool age up to 95.5% in adulthood). Stress with social relationships and low self-esteem dominated throughout the whole life (24% in preschool age up to 35% in adulthood; 24% in preschool age up to 40% in adulthood, respectively). Main agedependent problems were infantilization (23.8%) in preschool age, stigmatization (19.4%) during primary school, school problems (18.2%) during puberty and managing everyday life (20%) during adulthood. There was no significant difference between patients with distinct phenotype (e.g. pteygium ~11~) and others withoutdistinctphenotype. The early onset of psychosocial problems in UTS points to an isolated stress effect by short stature because at this early age, problems with reduced pubertal development or with infertility do not yet occur. Part of the psychosocial burden is age-dependent and may reinforce basic stress associated with short stature. Therefore individual problems with short stature in these patients must be handled in accordance to the developmental stage. All SGRQ components significantly correlated with the fatigue scale (r = 0.42 -0.57; p < 0.01) and with the MRC Dyspnea Scale (Y = 0.45 -0.76; p < 0.01). Tho shuttle walking test was the only measure which significantly correlated with the SGRQ Activities and Total scores. FEVl (%pred) and the shuttle test were significantly correlated to the MRC Dyspnea Scale. However, CT score did not correlate with any of the QOL measures nor with FEVl (%pred) and the shuttle score. We conclude that in these patients the relationship between anatomical disturbance of the large airways and physiological disturbance is weak, and that exercise tolerance, but not the extent of bronchiectasis, is a good predictor of QOL. Structural abnormality of the large airways in the lungs of bronchiectasis patients is not directly responsible for their reduced QOL, and further studies are required to investigate other possible causes. Long-term care residents' preferences for health states on the quality of well-being scale The Quality of Well-being (QWB) Scale quantifies healthrelated quality of life (HRQOL) with a single number that represents community-based preferences for combinations of symptom/problem complexes, mobility, physical activity, and social activity. The aim of this study was to compare preferences of a long-term care population with the general population, determine if preferences vary by the age of the hypothetical person depicted in the health state case description, and derive weights for new symptom/problem complexes of particular relevance to frail, older individuals. A sample of 38 female and 12 male long-term care residents with an average age of 86 were asked to rate health state scenarios that combined the four health domains of the QWB. This sample rated quality of life approximately of 0.10 units lower (on a O-l scale) than did the general population sample on which the QWB preferences were originally developed. Ratings of the same health state for younger vs. older target persons did not differ significantly (all p's ; 0.05 for f-statistics). Weights derived for 11 new symptom/problem complexes were as follows: disturbed sleep (-0.252), sit-to-stand requires maximal effort (-0. Researchers may have concerns about the distribution of data obtained from quality of life (QOL) scales and impacts of possible non-normality on parametric general linear model (GLM) analyses (regression, ANOVA, etc.). Some scales within QOL instruments are relatively 'granular' measures, having few possible values. In addition, the shape of the distribution of data may be too non-normal for analysis via GLM. Because of these concerns, some investigators have suggested using non-parametric statistical analysis methods for QOL data, however, these statistical methods cannot incorporate important covariates and may therefore be substantially less powerful than GLM methods. Oblectives for this study, then, were (1) to assess the extent to which GLM methods are appropriate for the analysis of QOL data and (2) to evaluate and compare the use of non-parametric with GLM methods for QOL data. QOL data collected during a 16-week clinical trial to assess the effects of study treatment were utilized for this study. QOL was assessed using the Short Form-36 which consists of eight domains: physical functioning, role-physical, pain, general health, vitality, social functionmg, role-emotional 438 Qunlity of Life Research Vol 4 1995 psychological and social well-being but no generally accepted definition has been developed yet. Epidemiology and rehabilitation sciences require the quantification of health outcomes besides traditional disease outcomes like death and recovery. In contrast, to directly observable dimensions a new aspect of health-related outcome measurement is the use of patient-rated scales. In a cross-sectional study, two generic instruments have been compared, both constructed as profiles consisting of SIX dimensions/subscales. The Nottingham Health Profile (NHP, Hunt et al., 1981) contains 38 items answered in a yes/no format measuring the components energy level, emotional reactions, social isolation, pain, physical mobility and sleep. In comparison the genuine German instrument 'Profil der Lebensqualitlt bei Chronischkranken' (PLC, Siegrist et al., 1993) was used. It covers the dimensions of positive and negative mood, ability to relax, physical capacity, social well-being and social functioning (40 items in five-point Likert scales). Scores were calculated as means of the corresponding items in each subscale. To assess the convergent validity several other established instruments were used and correlated (CESD-Depression, PI&S---Coping, Catastrophizing, SES-pain perception, FFBH-functional capacity). One hundred twenty-two patients (81 female, 41 male) of a rehabilitation-clinic with rheumatoid arthritis (n = 60), low back pain (n = 33) or osteoarthritis of the hip or knee (n = 29) were enrolled and examined in the second week of their stay. The average age was 53.9 years and the mean disease duration about 10 years. After a standardized interview concerning pain, sociodemography, medical treatment and the patient's definition of 'quality of life', the questionnaire was completed by the patients within 50 to 70 minutes. Reliability, validity and descriptive data were analyzed with SPSS. Both instruments show a high practicability and acceptability Missing values were below 0.8 % of all responses. The patient groups obtained similar results in all dimensions without significant disease-specific differences. Sex-specific differences were measured in positive mood (p < 0.05) Moderate correlations were observed between physical (Pearson correlation-coefficient r = -0.45, p c 0.01) and psychological dimensions (r = -0.35 to r = -0.58, p < 0.01) of NHP and PLC respectively, whereas the correlating social dimensions ranged from r = -0.26 to r = -0.29. High correlations were noted between the global rating of the health status and the subscales pain (NHI') and the emotional subscales of the PLC. The internal consistency is acceptable for both instruments with average values for Cronbach's a = 0.67 (NHP) and a = 0.74 (PLC). Both instruments seem to be reliable, valid and adequate measures for the assessment of quality of life in musculoskeletal disorders. They are associated in the physical and psychological dimensions but measure different aspects of social well-being. The NHP is likely to be more sensitive in measuring outcome in patients with severe disabilities whereas the PLC seems to be more suitable for considerable well populations. Seventy per cent of the patients associated 'quality of life' with terms such as health, general satisfac-and mental health. For this study, the first objective was addressed by evaluating asymptotic (large sample) criteria via computer simulation and performing a complete residual analysis. The second objective was assessed by applying non-parametric (Kruskal-Wallis) and categorical (Grizzle-Starmer-Koch) analysis methods and then comparing them to results obtained via GLM (ANCOVA: adjusting for baseline QOL, age, gender) methods. Computer simulation assessing asymptotic properties and sample size consideration showed that the parametric analyses approach was asymptotically valid. Linear model estimates of regression coefficients and related quantities and of the variance, are valid for a number of quality of life scales. However, for some QOL scales where granularity and non-normal distribution is a concern, data must routinely be analyzed to evaluate the accuracy of GLM analyses. This study suggests that non-normally distributed QOL data is not necessarily restricted to nonparametric analysis techniques. Rather GLM methods, which will include covariates that affect QOL (i.e. age, sex), can be utilized. Researchers have defined the meaning of quality of life, but this expression is also used in nontechnical, everyday language. The purpose of this presentation is to show that although questionnaires provide a good measure for the technical meaning, they capture only part of the meaning implicit in the common sense interpretation of the word quality of life. Quality of life is defined technically in terms of the patient's subjective experience. However, there are different kinds of subjective experience. Questionnaires measure retrospective recall, in contrast, diary assessments provide a better measure of immediate experience. Research comparing diary and questionnaire assessments is reviewed and the difference between these types of assessment discussed: retrospective recall and diary assessments are both legitimate objects of study in quality of life research. The common sense interpretation of quality of life includes the patient's objective state. Research shows objective and subjective reality often differ. Patients adapt psychologically to chronic illness: 'disengagement' from previously desired goals means that the patient is not conscious of wanting goals that have become unattainable. In relation to objective reality, patients appear to deny the effect of illness. Subjective reports, such as questionnaires (or diaries) provide an important but nevertheless partial account of the lay person's understanding of quality of life. Methods for assessing objective levels of restriction are discussed. Quality of life assessments are used for different purposes. Some of those purposes are best served by generic or disease specific questionnaires, but others are best served by a form of idiographic assessment Recent idiographic assessments (individual quality of life assessments) have provided insight into patient variability in reactions to illness, in particular how patients gain meaning from their illness. Retrospective, subjective recall of health status IS an important component of quality of life, but it is not the only thing that gives life its quality. It is possible to be satisfied with your health because you forget about your problems, you don't expect much in life, or you adopt an idiosyncratic way of turning ill health to an advantage. Although quality of life may be defined precisely by researchers, it is used in applied settings by people who have their own understanding of the term, an under-This study describes the use and validation of a structured asthma-specific quality of life diary in a clinical trial. Asthmatics were randomized to either low dose inhaled steroid combined with the long acting p2 agonist salmeterol (n = 220) or high dose inhaled steroid (n = 206). Outcome was measured by questionnaire (Living with Asthma Questionnaire) and a combined symptom, peak flow and quality of life diary, the latter requiring patients to indicate level of problem in four categories of broadly defined problem content: employment, domestic jobs, social/leisure and missed appointments. Diary QOL was assessed in terms of the proportion of problem free days. The QOL diary correlated with the QOL questionnaire for both cross sectional and longitudinal assessments in ways consistent with expectation. Cross sectional correlations with peak flow were higher with the questionnaire compared with the diary but longitudinal correlations were higher with the diary compared with the questionnaire. Diary assessed quality of life was a better predictor of low peak flow than diary assessed symptoms. Diary assessed QOL was more sensitive to treatment effects compared with the questionnaire and more sensitive than symptom assessment. However, unlike the questionnaire, the QOL diary methodology provided evidence of ceiling or floor effects for 25% of patients. We explored different methods of scoring a diary in terms of effect on correlations and effect size. Quality of life diaries are similar though more complex than symptom diaries. The main advantage of the diary methodology compared with the questionnaire appears to be its sensitivity to change, resulting from a stronger association with longitudinal change in physiology. However, a QOL diary measures only one component of QOL and idiosyncratic forms of patient response lead to ceiling or floor effects. Consequently, when used in a clinical trial, absence of pre-treatment diary QOL problems may be considered as a reason for patient exclusion. In addition, unless electronic diaries are used, retrospective completion of diary data may be undetected. QOL diaries are a promising new form of QOL methodology. hypertension. There are also interactions among personality variables and knowledge of current health status. These factors represent dimensions of disease sensitivity with consequences for patient adaptation, and they are firstorder (Level One), having ancillary determinant capacity only for patients suffering from the one specific complaint under investigation. These are most unlikely to be present in general QOL instruments. Three-level Q theory, we conclude, is a unifying frame of reference for the interpretation of QOL measurements and for their future construction. By adopting this, researchers can avoid recent debates about the use and function of either general or specific QOL instruments and produce shorter ones with more validity University of Plymouth, Plymouth, UK (S. H. Irvine); Glaxo Italy, Verona, Italy (G. de Carli, G. G. Recchia, F. Bamfi, F. Arpinelli) The purpose of this paper is to provide a standard frame of reference for the interpretation of QOL measurements; and it is based on large-scale research in multinational contexts. The history of practical outcomes and publishing output on QOL is visible in bibliographies, but little has been proposed to reconcile the inconsistent claims of, let alone the conflicts among instrument constructs. Hence, little of usable theory has emerged from domain-dominated measures from patients of their percepts of the effects of disease or trauma on their lives. An adequate theory of QOL measurement remains to be derived from empirical findings. The model here advocated for QOL measurement is psychometric and is closely allied to that produced by Spearman in defining human intelligence. We propose a single unifying proposition: In all QOL measurements there is a latent general factor of perceived health status Q, whrch IS inferred from a number of correlated primary determnants Qp, and varieties of disease-specific factors defined as Qs. Structural approaches to definition based on two different approaches are presented. One domain-referenced, is characteristic of general and broad-spectrum inventories. A second, is disease-specific and for the most part is patient-referenced. The results of analyzing general QOL inventories by data reduction methods (factor analysis) in published sources are compared with the results of our analysis of extensive data sets in three different specific and chronic diseases: asthma (UK Home-Care, UK Hospital Outpatients, Italian Standardization Sample (Total II > 700)); gastric patients suspected of having ulcers (1772 Italian patients forming a national sample); and, perhaps the most challenging of all, regional focus groups (n = 90) of asymptomatic hypertensive patients in Italy. Both the review of published research on structural approaches, and the results of our empirical studies converge to the conclusion that a three-level factor theory will serve as the basis of future QOL measurement. In all QOL measures there is variance attributable to a Level 3 general factor of perceived well-being, which we define as Q. Invariably, the domain scores of general or specific instruments produce evidence for far fewer patient percepts of health than the number of specified domains. The domains are convenient for the emergence of Q factors. The most important factors, which are measurable directly, are primary and determinant. The three pervasive primaries (Q$ are perceived physical dlscornfort, constrmnts on daily living activities and psychological distress. These can be designated as correlated second-order (Level Two) group factors present in all QOL instruments, from which the general factor Q at Level Three can be inferred, but not measured directly. These factors are present in all general and disease-specific instruments and they have the same determinant functions. There are, however, factors (Qs) specific to disease, such as awareness of precipitating agents in asthma and gastric complaints, differential disease burdens and side-effects of drug regimens, notably in In epilepsy, as in other conditions, most QOL research to date has focused on adults. Recently, however, the need to address QOL issues in children with epilepsy has been highlighted, and in this paper I will describe a series of studies in which we have attempted to do so. The conspicuous lack of information about how their physical status affects the QOL of children with epilepsy has been said to contribute to the lack of services specially adapted to this population. The lack of psychometrically sound QOL scales appropriate for use in this clinical population has also meant that clinical trials of treatment for epilepsy in children have only rarely included any QOL assessment. Our group has recently been involved in a number of studies which have incorporated a QOL assessment, including a prevalence survey in one UK health region, a randomised study of immediate vs. delayed treatment for early epilepsy, and a clinical trial of a novel antiepileptic. The theoretical framework for QOL assessment in these studies draws on our previous work to assess QOL in adults with epilepsy, and is made explicit in our epilepsy-specific QOL model, which focuses on the physical, social and psychological impact of the condition. While the broad dimensions are fixed by the model, it allows for a flexible approach to selecting scales and items, to ensure that they are appropriate to the particular population and clinical question under consideration. The methodological approach we have adopted has been to use a battery of generic and epiIepsy-specific scales and items, identifying previously validated measures where possible, and developing others de novo where necessary. Questionnaires are completed by the parents rather than by the children themselves, and we have confined our efforts to children of school age. In the prevalence survey and the trial of treatment for early epi-Quality of Li/e Research Vol 4 2995 441 lepsy, the QOL assessment in children includes scales and items about the clinical nature of epilepsy, including its perceived severity; the impact of antiepileptic drug therapy, general health status, psychiatric morbidity and the perceived impact of epilepsy on aspects of daily functioning, including school and social relationships and activities. The nature of the study population for the trial of a novel AED and the constraints of the trial design meant that no appropriate existing measures could be identified, and so we have developed a novel QOL instrument. Although the prevalence survey is now complete, the two trials are still ongoing; but preliminary analysis from pilot studies supports the validity and reliability of the instruments we have developed. In conclusion, we believe that assessing QOL in children as in adults with epilepsy is essential to increasing our understanding of the impact of their condition and of its treatment and to informing the systematic and organised management of their care. a process of translation and backtranslation, and was piloted in a small number of parents whose children were attending special epilepsy centres throughout Europe. The questionnaire is currently being used in a European multicentre clinical trial of the efficacy of a novel anti-epileptic, Lamotrigine, in the treatment of the severe epileptic disorders of childhood. Quality of life for adults with rheumatic disease: Impact of physical and psychosocial changes J. Jeffrey and W. Nielson Quality of life (QOL) is now recognized as an important outcome measure in clinical trials; and its assessment in trials of the treatment of epilepsy may be particularly relevant for children with severe seizure disorders, many of whom also have learning disabilities, in whom the traditional outcome of seizure frequency may be both irrelevant and unhelpful. We have developed a brief QOL questionnaire intended for completion by the parents of children with severe epilepsy and informed by in-depth interviews with parents and discussion with expert clinicians. The questionnaire consists of a series of novel scales and items to measure parental perceptions of seizure severity, adverse drug effects, their child's mood and behaviour, seizurerelated injuries, overall health status and overall QOL. The psychometric properties of the various scales and items incorporated in the questionnaire were examined in a pilot study of 50 children with a severe seizure disorder, all of whom had accompanying learning difficulties. The measures were found to have good face and construct validity and good reliability, as assessed by their internal consistency (a range = 0.71-0.84) and test-retest correlation (r = 0.67-0.84). To ensure its appropriateness for use in a cross-cultural context, the questionnaire was subjected to One purpose of this study was to determine the impact of perceived change in physical symptoms of rheumatic disease and psychosocial response on quality of life for adults who have rheumatoid arthritis (RA) or fibromyalgia (FS). Subjects in this 2-phase study were recruited from the practices of rheumatologists in Southern Ontario and met diagnostic criteria for their respective rheumatic disease. Subjects were adults, aged 18-70, who did not have concurrent musculo-skeletal disease or any condition that impacted their lives more than either RA or FS. In 1991/92, Phase I, 293 subjects completed questionnaires and physical examination. Two to three years later, at Phase II, questionnaires were mailed to participants in Phase I: a) personal characteristics (demographic characteristics and illness history); b) visual analog scales and single item rating of impact of symptoms (pain, stiffness, fatigue, difficulty sleeping); c) Health Assessment Questionnaire-functional ability; d) West-Haven Yale Multidimensional Pain Inventory; e) Multidimensional Assessment of Fatigue; f) Centre for Epidemiological Studies-Depression; and g) Quality of Life Index. At Phase II, subjects were also asked to rate their disease as the same, better, or worse than two to three years ago. To date, 76 adults with RA (RAs) and 60 with FS (FSs) have returned completed questionnaires, with 15% (43) returned undeliverable. Personal characteristics were similar for both groups except RAs were older and diagnosed longer than FSs. There were no significant differences in scores for any measure between Phase I and II, and no differences between respondents and non-respondents at Phase II except for tendency for respondents to have reported higher quality of life at Phase I than non-respondents. At both Phase I and Phase II, subjects with FS reported greater severity and impact of symptoms (pain, stiffness, fatigue and difficulty sleeping) as measured by single item and multidimensional measures (f-tests p < 0.01). Depression scores (CES-D) continued to be higher for FS (m = 22) than RA (m = 12, t = -5.7, p = 0.004). Subjects with FS also continued to report significantly poorer quality of life in general and for all of its domains-health/function, socioeconomic, variables. Favourable internal consistencies for the multi-item scales were observed, with a coefficients ranging from 0.63-0.91. SF-36 scale scores were not influenced by sex or education level. Age was significantly associated with four of the eight dimensions. Indicators of glycemic control were not significantly associated with any SF-36 multi-item scale scores. Subjects with more comorbid chronic conditions had significantly lower SF-36 scores. This research represents a single cross-sectional assessment of quality of life in a sample of diabetic Pima Indians. We believe this is the first such assessment in this population. The results indicate a favourable internal consistency of the SF-36 scales in this sample. If the goal of maximizing health status of American Indians is to be reached, healthrelated quality of life assessment should be used as a means to monitor progress toward that goal. Further evaluation of the SF-36 or other health status/quality of life instruments in this population should be undertaken. psychological, and family (e.g., RA m = 22, FS m = 18.5, f-test p < 0.01). Although not a statistically significant difference, a greater proportion (30%) of FS subjects reported their disease was worse at Phase 11 than at Phase I vs. RA subjects (20%), while 9% FS and 17% RA subjects said their disease was improved and 12% of both groups reported their disease was unchanged. The relationships between both physical symptoms and psychosocial response to disease and quality of life were in the expected direction, using ANOVA. Subjects who rated their disease as worse reported significantly greater severity and impact from physical symptoms, as well as more depressive symptoms and cognitive distortions, and poorer quality of life than those who rated disease as better or unchanged. Change in scores for these measures between Phase I and II were not significantly different for diagnostic group. However, correlations between variables for both groups were in the expected direction-greater depression (r = -0.72), worse symptoms (r = -0.48) and poorer quality of life. Although time alone does not account for change in physical symptoms or quality of life, perception of disease status is a good predictor of overall impact of the illness on life (confirmed with multivariate statistics). FS has a greater impact than RA on quality of life, in part because of differences in physical symptoms of rheumatic disease and psychosocial response to disease. The relative newness of FS as a diagnosis, lack of effective treatment options and lack of knowledge about it may also contribute to perceived quality of life. Health-related quality of life of diabetic Pima Indians The outcome measurement component of the model is known as the Quality of Well-being Scale (QWB). The Scale has been used in a variety of validation studies and clinical research investigations. However, the QWB is not used as widely as other measures of health-related quality of life because it is not self-administered and it does not offer a profile of outcomes. The QWB does produce a Quality-Adjusted Life Year calculation that is required for cost-utility analysis. Most competing measures that provide profiles of outcomes can not be used in cost-utility studies. We are in the process of revising the QWB into a self-administered instrument (QWB-SA) The revised measure would produce not only QALYs but also provide the foundation for a profile of outcomes. This paper describes the first phases of the adaptation of the QWB to a self administered form. A content review of the QWB and other instruments revealed the need to add new symptoms and problems. In particular several new items were added to increase coverage of mental health. The self administered QWB can be completed in about 10 minuets and has been shown to be comprehensible to a wide variety of patients at different levels of education and well-being. Several pilot tests led to revisions in format and wording. Results from field tryouts will be A convenience sample of judges in each country (ranging from 27-117 judges) rated translations of response choices using a variation of the Thurstone method of equal appearing intervals. Judges marked a point on a 10 cm line representing the value of a given response choice (e.g. 'good' relative to the anchors of 'poor' and 'excellent'). Five response scales including 20 response choices were evaluated. The ordinal consistency of response choices within a response scale was determined by the percentage of all possible pairs of response choices that were ordered as hypothesized. To evaluate the equality of interval-level differences between adjacent response choices, the mean judgments of response choices were examined. Analyses of variance (ANOVAs) were conducted for each response scale to determine the amount of variance in judgments that was due to response choice, country, judge within country, and interaction between response choice and country. Results confirmed the hypothesized ordering of response choices across countries. Across countries, the average percentage of ordinal pairs ranged from 94.5-99.3 across all scales, and from 88.7% to a high of 100% within scales. Examination of mean values for the response choices supported the 'quasi-interval' nature of the scales, which is consistent with findings from previous studies of multicategorical rating scales and is the reason why some response scales are scored based on values that are not strictly interval. The results of ANOVAs provided strong support for the generalizability of response choice values across countries, Total models explained nearly all of the reliable variance (between 82%-89%) in judgments. Response choice explained between 72% and 88% of the explained variance in judgments across scales. In contrast, country effects, though statistically significant, accounted for only 2%-4% of this variance. The effect of individual differences in judges within countries (7%-18%) was stronger than the country effect. Most of the total model variance not explained by response choice was due to individual judge. These results support the equivalence across countries/ languages of the five sets of response choices evaluated. One or more of these response scales are used in most widely-used health questionnaires and all are used in the SF-36 Health Survey. The results demonstrate the ordinality of response choices within scales and support the interval properties of most of the response scales with a high degree of reproducibility across countries. Departures from the assumptions of equal intervals, where observed, were consistent across countries and were greatest for the two response scales ('excellent' to 'poor' and 'none' to 'very severe') that have been recalibrated in the SF-36 scale scores to date. The practical implications of these departures from equal-interval assumptions for scoring response choice scales warrant further study. Most widely-used health surveys use standardized response choices that are paired with numbers. Yet, the correspondence in meaning between response choices and the numbers assigned to them has long been a matter of debate. For example, is it accurate to assign a value of '5' to 'extremely', '4' to 'quite a bit' and '3' to 'moderately' in a severity rating? Such assignment assumes: (1) ordinality, i.e., that these choices are ordered in terms of magnitude such that 'quite a bit' would appear between 'extremely' and 'moderately'; and (2) equal intervals, i.e., that the distance between 'extremely' and 'quite a bit' is equal to that between 'quite a bit' and 'moderately'. These assumptions became crucial in evaluating the equivalence of translations of response choices during the International Quality of Life Assessment (IQOLA) Project. We examined the values assigned to five sets of response choices commonly used in health surveys (e.g. 'excellent' to 'poor', ' all of the time' to 'none of the time') for equivalence across 12 countries. Presented here are results for the first nine of those countries (Australia, Canada, Germany, Italy, Japan, the Netherlands, Spain, Sweden and the USA). These results bear on the ordinal and interval properties of the response choice scales as well as the generalizability of scoring algorithms widely used in health surveys across countries. In a cross-sectional study in 48 chronic schizophrenic outpatients with a long-term disease history the participants were asked to fill in a German version of the Lancashire QOL Profile (Oliver, 1991) . This questionnaire covers sublective satisfaction and oblective QOL indicators in various life domains, a self-esteem and an affect-balance scale and a single question on general life satisfaction. Sociodemographic data, psychopathology and neuroleptic side effects were assessed as well. The mean age of the patients was 57.2 yearskl2.9, 60% were female. Patients' ratings of satisfaction with various life domains ranged between moderate scores for financial situation (61.8*11.8 on a scale transformed to a range O-100) and social relations (65.6k9.0) to fairly high scores for safety (8O.ti8.4) and living situation (74.5k10.9). Satisfaction with life in general was usually reported to be lower than A study among 200 staff-members of a German psychiatric hospital was conducted for the purpose of analyzing the impact of organizational and cognitive empowerment on occupational health and worksite quality of life in psychiatric services. The general hypothesis that organizational and cognitive empowerment are positively related to quality of worklife by increasing staff members' ability to reduce job stress and to cope with job stress in an active, problem oriented manner, was tested by means of a LISREL path analysis. The study yielded the following results: organizational empowerment, operationalised as degree of participation, influences the level of perceived stress and also the behavioural reactions to perceived stress. The more frequently and intensively staff members participate in the improvement of their working conditions the less they experience these conditions as stressful and, with increasing degrees of participation, staff members tend to cope with job stress in a more active and problem-oriented instead of a passive emotion-focused manner. Because active, problemfocused coping significantly reduces the frequency of psychosomatic symptoms, as the path coefficient shows, participation has a threefold positive effect on quality of worklife: first through a direct reduction of job stress, secondly, by enabling staff members to cope with job stress in a more adequate form, and thirdly by a significant direct positive effect on general job satisfaction. As a general conclusion it will be suggested that the opening of new channels for participation might be an efficient strategy for the improvement of the quality of life of persons working in psychiatric services. self-rated quality of life finding is contrasted by higher correlations of overall life satisfaction with self-esteem (r = 0.64) and with affect balance (r = 0.60). In a multiple regression analysis with general life satisfaction as dependent and all abovementioned QOL variables as independent variables only self-esteem and satisfaction with social relations remained in the model (p < 0.05, R2 = 0.55). If one accepts general life satisfaction as a proxy for sublective QOL, a necessary requirement for a questionnaire assessing QOL would be that its subscales correlate at least moderately high with this profile. The results of our study will be discussed against this background. Due to the small sample size and the pilot character of the study (first experiences with a German version of the Lancashire QOL Profile) the results are to be considered as preliminary The objective of this study was to obtain health state valuations from a representative sample of the UK population. In this cross-sectional study, face-to-face interviews were conducted in the homes of individuals identified from the national postcode address file. Households were sampled so as to provide geographical coverage of the 14 regional health authorities in England, and those of Scotland and Wales. The initial element of the study protocol required respondents to record their health status using the Euro-QoL, which covers five dimensions (mobility, self-care, usual activities, pain/discomfort, anxiety/depression). ln addition to this descriptive record, respondents also rated their current health status using a 20cm visual analogue scale (VAS). Background information on the respondent's education, employment status and illness experience were also elicited. Three thousand, three hundred and ninety-five interviews were achieved. The characteristics of the study 446 Quality of Life Research Vol 4 2995 sample were virtually identical to those of the general population in terms of age/sex, social class, housing tenure, employment status, marital status, and educational attainment. Nonparametric methods were used to analyse the categorical variables, and analysis of variance was used to analyse the self-rated VAS data. Interaction effects were examined by controlling for age/sex in ANOVA. Levels of reported problem were highest for the pain/ discomfort dimension, with nearly 305 of all respondents indicating some level of problem. Over 20% of respondents indicated a problem with anxiety/depression. Rates for all dimensions increased with age. Significantly higher rates of reported problem were found in social classes IV/V than were encountered for classes I/II. Of the 243 theoretically possible EuroQoL states, 77 were found in the survey population. Ninety per cent of respondents were classified in terms of 12 health states. Fifty-six of the 77 states involved some problem with pain. Mean VAS valuations of current health status have been computed for 5-year age groups, by sex. These show a general pattern in which average levels rise until the mid-40s, and thereafter decline with increasing age. Mean values for health status differ significantly with age, social class, education and housing tenure. The results from this national survey correlate well with 'disability' rates reported in GHS and OPCS surveys, but indicate higher levels than reported previously. This paper demonstrates that the combination of the EuroQoL classification and the VAS, as a measure of selfrated quality of life, has substantial validity The simple methods deployed in this national survey have demonstrated the value of a low-cost, accessible technology with considerable potential for public health medicine and primary care workers. Valuing quality of life: a low-cost alternative to time P. Kind, C. Gudex, P. Dolan and A. H. Williams Centre for Health Economics, University of York, UK The primary objective of this study was to obtain health state valuations from a representative sample of the UK population. In this cross-sectional study, face-to-face interviews were conducted in the homes of individuals identified from the national postcode address file. Households were sampled so as to provide geographical coverage of the 14 regional health authorities in England, and those of Scotland and Wales. Study participants completed a 3-stage valuation procedure in which they (a) ranked a set of 15 health states described in terms of the EuroQoL classification; (b) used a 20 cm visual analogue scale (VAS) to record values for each of those states; (c) determined utility weights for each state using time trade-off (TTO) methods. Three thousand, three hundred and ninety-five interviews were achieved. The characteristics of the study sample were virtually identical to those of the general population in terms of age/sex, social class, housing tenure, employment status, marital status, and educational attainment. Parametric and nonparametric tests were applied to the VAS and TTO data. Regression methods were used to establish the relationship between VAS and TTO data. Variations in health state scores recorded using VAS were not linked with factors such as age, sex, marital status or employment status. Respondents in lower social classes or with poorer educational attainment, gave higher values than other respondents. Current health status also impacted on valuations-respondents with low self-rated health status tending to 'underestimate' compared with those in higher health status. Variations in utility weights elicited using TTO were primarily linked with the age of the respondent. Respondents aged 60 and over gave lower weights to health states than did younger respondents. Significant differences were found particularly in the weighting of the more severe health states. Current illness experience did appear to influence IT0 weights. The average TTO weights were estimated from the VAS scores using a general equation of the form TTO, = aa + at VAS, + a2 VAS?, with mean absolute differences between observed and estimated values of 0.065. Utility-based scaling methods tend to be resource-intensive and may require interviewer mediation. Such methods are favoured by those researchers who consider them to be well-grounded in theory, or who demand that expressions of preference must involve an element of choice. There remains uncertainty about the 'correct' method for use in determining valuations for health states. However, if a technically 'simple' method can yield results which are systematically related to those obtained by a more 'complex' method, the former might be adopted as a costeffective substitute for the latter when resources preclude the use of relatively expensive interview-based methods. Weighting life expectancy for qualitywhose values count? Centre for Health Economics, University of York, UK The measurement of healthy-life expectancy combines information on survival and health status in a single, weighted index. Such measures share a common link with quality-adjusted life years (QALYs) which combine information on the quality and quantity of life, in that both forms of measure require a means of adjusting for variations in quality of life. The investigation of an appropriate means of making such adjustments has been the subject of considerable research activity over the past 25 years. Initial estimates based on disability-free life expectancy (DFLE), took a minimalist approach in which a weight of 1 was given to a year of life free of disability, and a weight of 0 was assigned to a year of life with are disability. Death, too had a value of zero. Such arbitrary weighting systems incorporate only the values of the researchers who propose them, and lack any demonstrable contact with the preferences of society at large. More complex measures, such as that adopted in the OPCS Disability Survey, are based on multidimensional classifications of health status, and incorporate more sophisticated weighting systems, but these nonetheless, reflect the values of a limited, unrepresentative group of judges. Results from a national cross-sectional survey recorded values for self-rated health status, and these were used to compute estimates of Quality Adjusted Life Expectancy (QALE). Information from actuarial life tables for 5-year age bands, was combined with the corresponding mean value of self-rated health status obtained from the survey. Values of QALE were compared with index values for DFLE, computed in like manner, but based on survey data obtained from the General Household Survey. The difference in 'weighted' life expectancy which results from the two approaches amounts to two years for young adults. Those aged 20-24 have a period life expectancy of 54.3 years. DFLE yields an estimate of 42.6 years, compared with the 45.3 years based on QALE. The proportionate 'error' between the two methods is greatest for older cohorts. DFLE and QALE values for those aged 85-89 are 2.9 and 3.7 years respectively, being 41% and 24.7% of the crude life expectancy figure. Definitional issues around the criterion for classifying individuals as 'disabled' have also been examined. In the national UK survey, levels of reported problem were highest for the pain/discomfort dimension, with nearly 30% of all respondents indicating some level of problem. Over 20% of respondents indicated a problem with anxiety/depression. Rates for all dimensions increased with age. Significantly higher rates of reported problem were found in social classes IV/V than were encountered for classes I/II. Respondents in lower social classes simultaneously report high levels of problem while at the same time regarding poor health states as not as bad as others think they are. Values attached to severe health states not directly experienced by respondents also differ significantly by social class. QALY estimates based on the EuroQoL pain/discomfort and anxiety/depression dimensions were additionally calculated to demonstrate its sensitivity to the choice of 'disability' dimension. This paper presents new empirical evidence showing significant variation by social class in the frequency of self-reported 'disability', and in the valuation of respondents' own health status. These findings have serious implications for those who seek to weight life-expectancy for health status. Whose values should be used-those of social classes l/II who give low values to poor health states, or those of social classes IV/V who give higher values to the same health states, or an aggregate set based on a representative sample of the population at large? (2) provide sample size estimates for a range of differences Ten studies which reported mean scores for clinically-defined groups of patients using either the QLQ-C30, or its predecessor QLQ-C36, were included in this review. Mean QOL scores were collated for the following clinical indicators: performance status (F'S = 0, 1,2,3,4), weight loss (WL = 0%, I lo%, > lo%), toxicity (TX = low, medium, high), and extent of disease (local, metastatic (mets)). For each study, and each clinical indicator, the mean difference between adjacent groups was calculated. Sample size calculations were based on a two-tailed t-test at the 5% significance level with 80% power. Estimates are for the number of patients in each group for a study design with two equalsized independent samples. Generally, the largest differences were between performance status groups, and the smallest differences were between groups of patients with local disease and those with metastases. Standard deviations (SD) were very similar between studies. Sample size estimates are: with SF-36 scales. The direction and magnitude of the correlations were as predicted. Based on these results, the IBSQOL is a valid and reliable disease-specific instrument that can be used to assess the impact of IBS on patients' lives and well-being. International Pharmacoeconomic Research, Glaxo Wellcome, Inc The purpose of this project was to develop and validate a disease-specific QOL instrument for Irritable Bowel Syndrome (IBS) sufferers. To date, there is no instrument specifically designed to assess QOL for patients with IBS. The primary characteristic of IBS is abdominal pain/discomfort associated with a change in bowel habits; patients may also suffer from abdominal bloating and distention, as well as diarrhea or constipation. It is believed that these symptoms impact aspects of patients lives including social activities, ability to perform work, mental and emotional well-being, and sexual relationships. A classic development approach was used to construct the instrument. A pool of 46 items, comprising 10 hypothesized sub-scales was generated. Initial face validity of the instrument was provided by clinical experts. Over 200 diagnosed IBS patients from two geographically distinct sites completed the prototype instrument, One group of 108 patients was used to identify the underlying factor structure and to provide the basis for scale construction, scoring and psychometric testing. The remaining 93 patients served as a hold-out sample to verify the scale construction and psychometric properties of the instrument. Factor analysis was used to identify three underlying factors: general, physical and emotional. A multi-trait, multi-attribute methodology was applied to the hypothesized scales. Results indicated that most scales fulfilled the criteria for summated ratings scales, but two scales (health belief and sexual functioning) did not fulfill this criteria. Remedial steps were taken to drop the health belief scale and to correct the other scale. Results from item discriminate validity tests and the factor analyses were used to delete items from scales and to generate a revised instrument, The revised instrument was administered to three separate focus groups of 8-12 patients with IBS. As a result the instrument was shortened to nine scales consisting of at least three items per scale, for a total of 30 items. Listed below are the mean IBSQOL scale scores for the two different sites and a control group drawn from a tertiary care centre: The purpose of this study was to describe the change over a two-year period in the health-related quality of life of disease-free patients treated for cervical cancer with radiotherapy. All patients with cervical cancer referred to the Department of Oncology treated with radiotherapy in a three year period were included. Participation: 118 out of 161 (70%); mean age at diagnosis: 54.5 years. A questionnaire consisting of the EORTC QLQ-C30 version 1.0 as well as 70 supplementary questions was used. The patients filled out the questionnaires three days before the end of radiotherapy and 1,3,6,12,18 and 24 months after radiotherapy. Patients were excluded from further participation if a recurrence was diagnosed. A random sample of Danish women completing the same questionnaire was collected (1,041 persons). From this a normative sample was constructed by a 2-l match with the patients on age and partner status. The EORTC QLQ-C30 as well as other relevant questions were analyzed in three different ways: (1) a comparison between all available datasets with the normative sample; (2) a comparison of 70 patients who had complete sets of questionnaires with the normative sample as well as with the whole sample; (3) a plot of curves showing the scores from sub-groups of patients answering different numbers of questions. On the physical symptoms, the patients have significantly more symptoms at the end of treatment and 1-3 months later. The treatment-specific symptoms i.e., diarrhoea and urinary symptoms show a significantly higher score than the normative sample up to two years after treatment. The psychological and social items show very little difference from the normative sample, often only at the first point of measurement. On the EORTC QLQ C-30 scale on quality of life the patients score significantly lower at the end of treatment one and three months after (Mann-Whitney p < 0.05). The results for the physical symptoms are in accordance with the clinical experience. The lack of difference in psychological distress between patients and the normative sample may also be due to the fact that the EORTC QLQ-C30 was developed to measure signs and symptoms in patients with active disease undergoing treatment and it may not be sensitive enough to detect symptoms in survivors. The evaluation of the health-related quality of life using the EORTC QLQ-C30 and other questions in the same format show that disease-free patients treated for advanced stages of cervical cancer have more symptoms than a normative sample at the end of radiotherapy and one to three months after. Only treatment-specific symptoms like diarrhoea show a difference up to two years after treatment. Measuring qualityof life changes in patients treated for coronary artery disease in Alberta M. Knudtson and C. Norris The clinical benefit of percutaneous transluminal coronary angioplasty (PTCA) as compared with coronary-artery bypass grafting (CABG) or Medical therapy for patients with multivessel coronary artery disease has not been firmly established. To determine the outcomes of these treatments in patients referred for the first time for coronary revascularization, an eight year prospective, observational study, the Alberta Provincial Project for Outcome Assessment in Coronary Artery Disease (APPROACH Prolect), began collecting data in January 1995. The APPROACH Prolect will compare and contrast patients undergoing these three treatment modalities. A QOL study will be concurrently carried out utilizing the APPROACH database. The purpose of this substudy is to describe the QOL changes of a large prospective cohort of Coronary Artery Disease patients treated with medicine, PTCA or CABG. The study is designed as a prospective non-randomized treatment comparison in tertiary care centres in Alberta. Eligible subjects include patients with symptomatic CAD referred for cardiac catheterization. Patients with valvular or congenital disease, or nonischemic cardiomyopathy, will be excluded from the study. Baseline and follow-up QOL data, utilizing the Seattle Angine Questionnaire (SAQ), will be collected. The SAQ is a valid and reliable instrument that measures five clinically relevant dimensions of health in patients with coronary artery disease. These five dimensions include (1) physical limitation; (2) angina1 stability; (3) angina1 frequency; (4) treatment satisfaction and (5) disease perception. It has been shown to be sensitive to clinical changes in patient's coronary disease. The SAQ will be administered to all APPROACH Project patients undergoing coronary angiograms (baseline), and then again, one year later as follow-up. Baseline and follow-up mean SAQ dimension scores will be analyzed by paired t-tests It is anticipated the SAQ scores will demonstrate measurable QOL differences, following treatment for CAD, in the five dimensions identified. If physicians are to make rational decisions in treating patients with coronary artery disease, the importance of issues such as QOL and functional independence must be addressed in more detail and should be seriously considered m the treatment choice. Utilizing a diseasespecific functional status measure will allow for the quantification of the physical and emotional effects of coro- The impact of psychotherapeutic support on survival time and quality of life in patients with gastrointestinal cancer undergoing surgery was studied. A randomized controlled trial was conducted in cooperation with the Departments of General Surgery and Medical Psychology, University Hospital of Hamburg, Germany Two hundred and seventyone consenting patients with a preliminary diagnosis of cancer of the esophagus, stomach, liver/gallbladder, pancreas or colon/rectum were stratified by gender and randomly assigned to a control group that received standard care as provided on the surgical wards, or to an experimental group that received formal psychotherapeutic support in addition to routine care during the hospital stay. Data about all medical and psychotherapeutic interventions were systematically collected. Survivors in both groups completed the EORTC-Quality of Life questionnaire pre-and post-operatively, and at 3, 6, 12 and 24 months following surgery Date of death, if applicable, was also recorded. Unadjusted and adjusted survival analyses were performed. The value for the fifth percentile score of the patient's group was imputed for those who died to evaluate quality of life outcomes. Two-way analyses of variance were then used to compare the groups. Kaplan-Meyer survival curves demonstrated better survival for the experimental group than the control group. The unadjusted significance level for group differences was 0.044 for survival to one year and 0 004 for survival to two years. In a Cox regression model that took tumor staging into account, the significance level was 0.034 for the first year of survival and 0.005 for survival to two years. Secondary analyses found that most of the differences in favour of the experimental group occurred in the patients with pancreatic or colorectal cancer. Following imputation of quality of life scores for those who died, quality of life tended to be better for patients in the experimental group. The results of this study indicate that patients with gastrointestmal cancer, particularly those with pancreatic or colorectal cancer who undergo surgery, bene-Since 1986, patients participating in the Liver Transplant Programme at the University Clinic, Hamburg, have had their quality of life measured before and after surgery in a cooperative study of the Departments of Medical Psychology and General Surgery. The purpose of these evaluations was to obtain data to track the status of patients over time and to allow pre-and post-transplant comparisons in quality of life. The instrument used for evaluation was the EORTC Quality of Life Questionnaire, which includes a specific Liver Transplantation Module developed for this study. Ail patients had a psychosocial interview at the time of first referral to the Transplantation Centre. In addition, some patients were provided with psychotherapeutic support throughout the waiting period and all received it regularly after transplantation, during the time in the hospital. Between January 1984 and December 1993, 167 patients (85 males, 82 females) were transplanted. Of these patients, 90 had died by December 1994. The results of this ongoing longitudinal study indicated that all dimensions of quality of life were significantly improved by liver transplantation when compared to end-stage liver cirrhosis before transplantation. This improvement, however, was usually not evident until one year after the operation, and in some patients, even longer. Further, it should be noted that the transplanted patients perceived that their overall quality of life was even better than healthy controls, although they rated some liver-specific symptoms as still being bothersome. Since self-perceived quality of life depends on such factors as the underlying diagnosis, the pre-operative status, the degree of support from family and friends as well as psychotherapeutic support, these factors must be taken into account when considering a cause-effect relationship. Comparing a health status instrument with conventional utility instruments Across all utility measures, cardiac patients rated their own state of health poorer than healthy individuals. The mean utilities decreased with greater impairment in functional status as assessed by the SAS. Across all respondent groups, the SG and TTO yielded higher scores than the RS. The RS, SG and TTO were moderately correlated (Spearman Y = 0.31-0.53) with each other. For all SF-36 subscales, cardiac patients, on average, had lower scores (i.e. poorer health) than healthy individuals. The greatest difference was observed in the Role Functioning-physical subscale (mean = 94 for healthy group vs. mean = 54 for cardiac patients). Mean differences for all other subscales varied from 7-22 points. All SF-36 scales were moderately correlated with RS and TTO (Spearman Y = 0.30-0.70) with the exception of the Role Functioning-Emotional scale which was poorly correlated with RS (Y = 0.24) and TTO (Y = 0.27). The Vitality and the Physical and Social Functioning scores were moderately correlated with SG (Y = 0.30-0.39). All the other scales were only poorly correlated with SG (Y = 0.11-0.28). The correlations between each SF-36 subscale score and the utility score was highest with the RS (with the exception of the Role Functioning-Emotional) and lowest with the SC {with the exception of the Physical and Social Functioning). Patients with cardiac conditions rated their own state of health poorer than healthy individuals on the SF-36 health status instrument and the RS, TTO and SG. However, the modest correlations we observed between the SF-36 subscales and the utilities, particularly those elicited by the SG, suggest that determinants other than health status may influence utilities. has been gathered on the potentially unique caveats of translating child self-report instruments relative to parent-completed forms. Thus, we launched a two-step effort to test the feasibility of translating parallel child self-report and parent versions of the Child Health Questionnaire (CHQ) into Swedish using adaptations of published methodologies. The psychometric properties of the resulting forms will be tested in children of low stature and their parents. The prototype questionnaires consists of multi-item scales that measure a child's physical and psychosocial well-being: Physical Functioning, General Health, Pain, Role/Social Limitations physical, emotional/behavioural, Self-esteem, Behaviour, Mental Health and Parental Impact-Emotional and Time, Family Impact and Family Cohesion. The panel was comprised of two professional translators, a bilingual lay translator, two bilingual study investigators, and a principal architect of the CHQ The investigators have extensive experience in design and administration of child instruments. The lay translator ensured that the translation would be easily understood by the general public. Three forward translations of both questionnaires were prepared by the two professional translators and the bilingual lay representative resulting in six complementary forms. The three versions of the self-report form were reviewed simultaneously. Differences in items, phrasing and response options were discussed. Endorsement of a reconciled form was achieved by evaluating which iteration came closest to the intent of the English version, but did not violate Swedish culture and expression. The same process was completed for the parent form. Whenever possible, complementary wording was used between the self-report and the parent form. Related issues such as reading and comprehension level and the age for self-report were also discussed in considerable detail. Backtranslation of the Swedish versions into English were evaluated qualitatively by the instrument developer and the bilingual study investigators. A focus group with children will further inform the translation process. The reconciliation meeting highlighted several important differences in the translation of the two parallel Swedish forms and underscored the need for representation by a primary developer of the original instrument. English was viewed by colleagues as a more formal language than Swedish, requiring several words to express a concept and resulted in a translation that was more direct in tone than the English version. Some English words were identified Item scaling analysis will provide empirical evidence of the conceptual framework of the Swedish CHO. It is feasible to translate the CHQ child and parent report questionnaires in parallel and achieve conceptually complimentary results in meaning. Of note was our use of three translators, a stringent translation procedure, and the incorporation of a well-grounded child development perspective. Qualitative findings will be augmented with empirical evidence using multi-trait item scaling analysis as pilot data becomes available. Standardization across countries of parallel child self-report and parent forms is essential to the international measurement of health in pediatric populations. Using this criterion, three English speakers were eliminated leaving 51 for analysis. Scale scores were calculated such that high scores indicate better function. Cronbach's CL was used to estimate internal consistency of each scale. Floor and ceiling effects were examined. In the English and French versions respectively, 73% and 89% of all item-scale correlations reached acceptable levels (2 0.4). Tests of item discriminant validity, the extent to which an item correlates more highly with its hypothesized scale than with other scales, achieved success rates of 90% and 88% in both samples. These tests were very stringent given small sample sizes and resulting large standard errors (0.14 and 0.25 in English and French speakers, respectively). Tests of equal item variance were conducted by examining the extent to which items correlated within f0.02 SD of each other within their hypothesized scales. Seventy-seven per cent and 76% of items in English and French respondents met this criterion. With the exception of a single scale among English speakers, all cx coefficients exceeded the acceptable standard for group-level comparison (2 0.7). Median alphas were 0.84 and 0.87 among English and French speakers, respectively. The only floor effects observed were in Role Limitations scales in the English sample (2%-4%) and a single Role scale in the French sample (6%). Generally, more ceiling effects were observed in the English than in the French sample. Analyses based on the combined sample (n = 67) did not materially change overall findings. These preliminary results offer strong support for the separate health-related quality of life dimensions included in the CHQ-PF, based on both item and scale analyses, among both English-and French-speaking Canadian samples. Our findings are comparable to other psychometric results obtained with the CHQ-PF in both a USA population-based sample and chronic condition groups. Stringent translation procedures were used to develop the Canadian French version. If subsequent analyses from larger respondent groups confirm these preliminary conclusions, we will recommend that the Canadian French version tested here be the basis for the standard one for use among Canadian The measurement of the quality of life of cancer patients can be used to assess the impact of medical intervention. The objective of this work has been the development of an instrument for the measurement of the quality of life of cancer patients without social security and mostly from rural places. As a first step we interviewed 42 cancer patients to identify the problems that bothered them most frequently and that could decrease their quality of life. With these problems we developed a questionnaire with 87 questions divided on three subscales: physical symptoms, psychological symptoms and social support. Forty-nine questions had a second part to ask how much distress the patient had with the symptom. With this questionnaire we interviewed 74 cancer patients from the Oncology Service of the Hospital Universitario de la Ciudad de Puebla, Mexico: 61 women with an average age of 48 years and 13 men with an average age of 52 years. They had been to school five years; we read aloud the questionnaire. All of them had had at least one antineoplasic treatment. Besides our questionnaire, they were evaluated with the Quality of Life Index (Spitzer) and with the Karnofsky Scale. The Kamofsky average value was 77 (min 50, max 100). The correlation coefficient (Pearson) between our questionnaire and the Quality of Life Index was 0.63. With the Karnofsky Scale was 0.42. The correlation between Spitzer and Kamofsky was 0.60. The subscale of physical symptoms had 31 questions and its average was 0.44 (zero means no symptom and no distress, three means 'a lot of') with an average total of 13.77. The psychological subscale had 35 questions, and an average of 0.56 for each question and an average total of 19.75. The social support subscale had 21 questions and an average of 0.56 for each question and an average total of 11.81. The Quality of Life Index average was 7.8 (min 2, max 10) Fifty-one patients were out-patients, 12 were in-patients and 11 were in radiotherapy. The patients with the worst quality of life were the m-patients as measured with our questionnaire, the Index of Spitzer and the Karnofsky Scale. We detected some differences between women and men. Women had worst quality of life. We think this questionnaire could be used in clinical work. It should be shortened and we have to do a consistency survey. We used a French validated translation of the FLIC (Functional Living Index Cancer) in a prospective study to assess the quality of life (QOL) in patients receiving adluvant chemotherapy for breast cancer. QOL is supposed to be impaired due to side effects of the treatment Seventy-five female breast cancer patients, included after surgery had to fill up the FLIC questionnaire, given by a research nurse, before the first course (baseline), the third, the fifth and the sixth. Simultaneously, nausea/vomiting and alopecia were assessed. Sixty-seven agreed to enter the study. Total score at baseline was 107 (47-151), at the third 99.5 (46-151) and at the sixth 100.5 (59-154), no difference was observed. No difference was observed between median scores studied at baseline 5 (l-7), the third 5 (l-7) and the sixth course 5 (l-7) for all items, as well as for principal components (somatic, psychological and social). Differences (p values) between the scores observed after the third course and baseline were significant for two items (Q5 0.0004 and Q17 0.0009). Differences between baseline and sixth course score were significant for five items (Q5 0.001, Q7 0.006, Q8 0.004, Q16 0.01 and Q17 0.012). Q5 and Q 17 are relative to nausea; 47, QS and Q16 are somatic and social items. A highly significant correlation was found between FLIC score and the intensity of nausea at the third course (Q5 0.0006, 417 0.0011) and also with somatic and social items (Q17 0.013, QS 0.0019, 416 0.0003). At the sixth course, a highly significant correlation was found also between FLIC scores and the intensity of nausea (Q5 0.0001, Q17 0.0032). These results do not confirm the deleterious effect of adjuvant chemotherapy on the QOL of breast cancer patients as assessed by the FLIC questionnaire. The most important factor which determines the QOL of patients is nausea. A considerable amount of work has been done in recent years on the cross-cultural adaptation of health status and quality of life measures by a number of different teams with a view to obtaining exploitable data at international level. A great number of problems arise given the subjective and cultural nature of what is being measured, and given the variety of contexts in which human-beings live and are sick. This paper compares extensively the development and validation methodologies followed by three international teams: the European Group for Health and Quality of Life Measurement (Euro-NHP), the International Quality Of Life Assessment (IQOLA) project team and the World Health Organization Quality Of Life (WHOQOL) group. The authors examine the following methodological issues: translation (conceptual and a practical issues, role of backtranslation, quality control), cross-cultural adaptation (role of experts and lay people) and psychometric validation. The advantages and inconveniences of the different options undertaken by each team are discussed in the view of the particular objectives of each project (sequential adaptation of existing instruments; simultaneous or parallel development of an instrument in different languages). They also discuss how to compare crossculturally developed instruments. Finally, some methodological recommendations are suggested, in the light of the authors' experience in the development, translation, cross-cultural adaptation and psychometric validation of several HRQOL measures such as the SF-36, the NHP, the WHOQOL and other disease specific instruments in French. The Health Institute, New England Medical Centre, USA Valid, reliable and practical self-report questionnaires measuring the impact of ongoing health problems on people's working lives would represent valuable additions to the current family of health status assessment tools. This presentation will describe a conceptual framework for assessing ability to function in work roles and work contexts. This framework will then be used to evaluate current methods for measuring ability to work. Finally, an inter-disciplinary approach, which has been used to develop questionnaires, will be described. It combines advances in health status assessment with social science methods for classifying work content and context. Both the Institute of Medicine's concept of disability and the World Health Organization's definition of handicap suggest that work role functioning is the result of a relationship between an individual's health resources and the expectations and structural conditions that operate within social settings such as the workplace. These and other perspectives indicate that instrument development efforts must seek to: (1) employ relational items that capture the correspondence between health problems and work role functioning; (2) use domain-specific indicators of functioning that reflect the range of physical, social, psychological and environmental work demands; and (3) incorporate scaling techniques that capture the varying degrees of limitation which occur inconjunction with chronic illnessand impairment. Three dominant methodologies are now in use: the impairment-based type used by the Social Security Administration, work evaluations common to physical and occupational therapy and functional assessment questionnaires used in outcomes research. None meets all of the necessary conceptual criteria. In the absence of improved measurement methods, a core component of health-related quality of life is being overlooked. A national survey of limitations in ability to work using a new work role functioning questionnaire D. J. Lerner, S. Malspeis, B. Rogers and B. C. Amick Because most of the available methods for assessing work role functioning either are not specific to work or focus only on the performance of physical work demands, a new 15 item self-report questionnaire was developed. Its purpose was to measure the frequency distribution and sources of work limitations within a national sample of 972 employed adults. This mail questionnaire elicited reports of the degree to which individuals had difficulty during the past four weeks meeting physical, mental, social and environmental work demands. Responses were scaled from 0 ('no difficulty') to 6 ( so much difficulty 'not done at all') and 'does not apply to my job'. Subjects were drawn from a 1994 national follow-up household survey of functional health status. In this employed subsample, 74% reported having one or more chronic or impairing conditions. The instruments met standard psychometric criteria for discriminant and convergent validity, internal homogeneity, item to total correlation and two-week test-retest reliability. The 15 items were classified into three relatively homogenous, mutually exclusive dimensions based on the strength of each item's relationship to SF-36 physical and mental health summary index scores. One, consisting of physically demanding tasks, was related more strongly to physical health. Another, comprised of psycho-social demands, was related principally to mental health. The last included mainly environmental demands and it was related to both health indexes. Subjects were classified as 'limited' on each dimension if they reported some or more difficulty on at least one of the relevant items and as 'not limited' if none or slight difficulty was reported for every item. Work limitation rates by the eight largest conditioning groups portray each as impacting on the subjects' ability to meet all three types of work demands. The purposes of this study are: (1) to develop an instrument with satisfactory validity-reliability and sensitivity which can be applied to assess QOL for general population; (2) to investigate QOL and its related factors on general population both in urban and rural areas of Hunan Province in China; (3) to understand the interaction between the objective life status and subjective life satisfaction. The subjects were 8,550 urban and rural area residents from six major geographic regions of Hunan Province and were investigated by stratified-random-cluster sampling. Of them, 4,374 were men (51.2%) and 4176 women (48.8%); the average age was 41.94 years (sd = 13.05), ranging from 21-89 years. The instrument used in this study was a selfadministered QOL Inventory (OOLI) developed by authors which was pilot-studied on 430 subjects in a community and was modified. The QOLI included 112 items grouped into four dimensions: 'physical function', 'psvchologicai function', 'social function' and ' living condition'. Each dimension includes some items of objective life status and some items of subjective satisfaction for these relevant objective life status. Items use a f-point scale. Independent scores on both objective and subjective items can be obtained for each dimension respectively. 3) Data were collected by trained investigators who visited the sampled subjects one by one and asked if the subject had finished the self-administered QOLI. The 64 items in QOLI were found to be highly representative of QOL by means of item-scale correlation, cluster analysis and factor analysis. Test-retest coefficients ranged from 0.66-0.69. and item-total correlations were found to be adequate. Factor analysis identified 11 factors accounting for 64.4% of the total variance, which could be grouped into four dimensions as above. Meanwhile, QOLI was sensitive in differentiating different sub-populations. The scores on objective life status in each dimension from younger, urban residents and employees are significantly higher than those of elderly people, rural residents and unemployed; however, the scores on subjective life satisfaction in each dimension, especially in the social function and living condition dimension, is tending towards normal distribution in different sub-populations. The divergence between individual objective life status and subjective life satisfaction is associated with his/her own aspirations for life, external reference standard of life and personal ability, resources, etc. The multiple regression results show the living condition dimension (income, housing and environment, etc.) is still the highest value indicator of life and has the greatest influence on evaluation of overall QOL. The developed QOLI has potential as a valid programme evaluation tool for general population in community. The hypothesis of this study is that the institutional costs of care [reflected by direct costs and contribution margin (net revenue-direct costs)] correlate with perception of quality of life measured by specific domains in the SF-36 health survey It has been argued that the objective of medical care for most patients is the achievement of a more effective life (McDermott, 1981) and the preservation of function and well-being (American College of Physicians, 1988; Cluff, 1981; Ellwood, 1988; Schroeder, 1987; Tarlov, 1983) . In an era of managed care, analysts will seek the best value for each health care dollar, carefully weighing perceived quality of life and the costs of medical care required to achieve that status. Instruments which identify patients with greater requirements of programmatic resources may be useful in containing costs through patient triage to lower cost treatment approaches which address their needs with equal effectiveness. Approximately eighty consecutive patients at each of five sites within the Department of General Internal Medicine at the Cleveland Clinic Foundation (Diagnostic Medicine, Primary Care, Emergency Access, Headache Centre and a community satellite office) were asked to complete a SF-36 health survey on the day of an appointment prior to seeing their physician. The number of office visits in a nine month period (November 15,1994 to January 15,1995 and the direct costs and contribution margin for these visits were identified and compared between groups. Multiple regression models were developed to determine which of the domains within the SF-36 best predicted institutional costs of care and contribution margins accumulated during the study period. Cost data for 260 of 414 (63%) patients were available (no data were accessible for patients seen at the community satellite). Patients treated in the Headache Centre were seen significantly less often (p < O.OOOl), but with similar direct costs than other areas within the department. Although direct costs of care were lowest in the Primary Care Clinic where nurse practitioners are used to augment physician staffing, significantly higher contribution margins could not be demonstrated (except in a comparison between Diagnostic Medicine Section and Primary Care). The best independent variables predicting lower contribution margins using multivariate techniques included patient age group by ten year increments (p < O.OOOl), and the patient's perception of both their general health (p = 0.005) and role physical (p = 0.07) on the SF-36 health survey. Patient perception of their mental health (p = 0.04), the amount of bodily pain (p = 0.01) and their level of social functioning (p = 0.04) were the best independent predictors for total direct cost in a second multivariate regression model. The SF-36 health survey may serve as a valuable tool in triaging patients with higher programmatic needs at medical centres. Because resource utilization translates to higher costs, early identification of patients who might benefit from equally effective, lower cost evaluation and treatment will assist the institution in containing costs in a managed care environment. and debilitating and treatment is palliative. This paper reports on annual rates of change for eight HRQOL dimensions (physical, social and mental functioning, pain, energy, cognition, health concern and general health) for HlV-infected males with varying disease progression, and the clinical correlates associated with these changes. ATHOS (AIDS Time-Orlented Health Outcome Study) is an observational data base of HIV-positive and at-risk patients seen by community-based physicians. Clinical data are abstracted at the time of a patient visit. All participants also complete quarterly questionnaires covering HRQOL, and resource utilization. All HIV-positive males enrolled in ATHOS were screened for length of follow-up (minimum of 90 days) and classified by initial and final disease stage. (Females were excluded because there were not sufficient numbers of patients with varying follow-up for separate analysis by disease stage.) Patients were initially categorized by HIV-status at entry: asymptomatic, symptomatic, patients with CABG surgery, 57 patients with lung cancer and 47 controls participated in this study. A self-administered questionnaire including 60 questions divided into 16 categories was used. We used the SAS analytical method for the statistical analysis. Compared with lung cancer, the patients with CABG surgery were significantly excellent in the sphere of well-being (p < 0.05), physical symptoms (p < 0.05), social participation (p < 0.05), mental problems (p < 0.05) and dietary problems (p < 0.05). In comparison with controls, the patients with CABG surgery were disturbed in the spheres of sexual life (p < 0.05) and physical symptoms (p < 0.05). In another comparison before and after CABG surgery, the significant improvements were noted in the spheres of pain (p < O.Ol), well-being (p < O.Ol), physical symptoms (p < O.Ol), physical function (p < O.Ol), medication (p < O.Ol), sleep (p < 0.05), life satisfaction (p < 0.05) and physical activity (p < 0.05). These findings indicate that CABG surgery is very effective for the improvement of quality of life in patients with coronary artery disease. or AIDS (1993 clinical criteria). Opportunistic infections, disease symptoms, and CD4+ lymphocyte count were reported at the time of the patient's entry into the study, and at each subsequent visit. An annual rate of change was computed for each HRQOL measure for each patient based on the respondent's completed questionnaires (which varied from 2-16); mean rates of change were then computed for each disease group using a multivariate regression analysis which also controlled for baseline age, education and health status. For each of the eight HRQOL dimensions, differences between disease stage were evaluated at baseline, and for the annual rate of change. Of 943 HIV-positive patients, 425 were asymptomatic (HIV+) at baseline, 272 remained HIV+ after one year (Group l), 83 became symptomatic (Group Z), and 70 progressed to AIDS (Group 3). Of 257 symptomatic patients at baseline (SYM), 122 remained SYM (Group 4) and 135 progressed to AIDS (Group 5); 261 patients were AIDS at baseline through follow-up (Group 6). Mean follow-up for all patients was 1.6 years. Scores on all HRQOL measures at baseline were worse for AIDS patients, followed by SYM patients, and HIV+ patients. Rate of change in HRQOL over one year was greatest for HIV+ patients who progressed to AIDS, however, for all domains but cognition and pain. All six groups reported significantly worse general health over a year (-4.4, -8.7, -20.5, -8.1, -14.9, -8.9 for Groups l-6 respectively measured on a O-100 scale.) In multivariate stepwise regressions which evaluated the relationship between HRQOL and disease symptoms, opportunistic infections, and CD4+ count, controlling for baseline disease stage and demographic characteristics, only the frequency and type of disease symptoms were associated with decreased HRQOL. Symptoms most frequently associated with decreased HRQOL were: fatigue, loss of sexual desire, muscle weakness, diarrhea, fever, headache, depression and weight loss. These results indicate that overall HRQOL decreases over time for all HIV infected male patients, no matter what the initial disease stage. However, the greatest decreases occur for those patients with greater symptomatology, controlling for disease stage. Efforts to detect and treat disease symptoms may have important effects on HRQOL, no matter what the patient's disease stage. Reflux (GOR). We therefore developed a self-administered QOL questionnaire to assess the impact of GOR and its treatment. This one year project was divided into five parts: (1) Individual qualitative interviews. Six GPs, six pharmacists and 35 patients were interviewed to determine the patients' main complaints and the specific ways they described their disease. tionnaire. The item reduction was performed using descriptive statistics, principal component analysis, multi-trait analysis and stepwise discriminant analysis according to symptom severity. Items were deleted according to pre-establish selection criteria. (5) Psychometric validation. In the follow-up of 349 patients, the remaining items were evaluated for scaling assumptions, construct validity, reliability (reproducibility/internal consistency) and responsiveness over time (effect size). The qualitative interviews followed by the item generation and the content validity resulted in the elaboration of a 104 item-questionnaire. Item reduction analysis deleted 67 items. The final questionnaire is made up of 37 items evaluating seven domains determined by their content and the factorial structure and confirmed by the multi-trait analysis: Daily activities (six items), Relationships (two items), Quality of Life (five items), Mental Health (seven items), Worries (five items), Sleep (four items), and Food (four items). Reliability coefficients were satisfying (a Cronbath and Intraclass Coefficient > 0.70). All scales discriminated groups according to categories of symptom severity. The responsiveness over time has been shown in the patients improved by their treatment (effect size from 0.38-1.05). This 37-item questionnaire, developed according to recommended guidelines, has shown the necessary properties to allow its use in clinical trials. To assess Quality of Life (QOL) in Intermittent Claudication (IC) a specific questionnaire was compiled from the literature using a generic scale supplemented by specific modules. It has been validated in a cross-sectional study This strategy offered the advantage of being comprehensive, but the length of the final instrument limited its use in large clinical practice. We therefore developed a short form to provide clinicians with QOL data for the management of their patients. The original questionnaire was compiled after interviewing 30 patients and six clinicians. It was based on the SF-36 supplemented by six modules: health distress, sexuality, impact of pain, daily activities, satisfaction and future. The final questionnaire was a profile of 15 scales. A crosssectional study on 123 patients was carried out for item selection. To ensure comprehensiveness, we decided that the short form should include items evaluating at least five major health concepts: physical, mental, social, perceived health and satisfaction. The selection was performed per dimension using stepwise regression, stepwise discrimi-nant analysis according to symptoms severity, and relative precision ratio (selected items F statistic value divided by complete scale F statistic value). Internal consistency reliability and clinical validity of the short form were then assessed in the sample. The responsiveness over time was evaluated using the effect size in a cohort of 3,871 IC patients followed during six months. In the stepwise regression the items which explained best the total score of the scale were selected (R2 > 0.90). In parallel, a second set of items were selected by the stepwise discriminant analysis which discriminated two groups of patients (walking distance > 500 meters and 5 500 meters; n = respectively 71 and 52) with a F statistic p value of 0.15. The final selection of the pre-selected items combinations was done using the relative precision ratio. Selected items compared to complete scales performed better in discriminating patients' severity groups (relative precision ratio > 1). The short form resulted in 12 items evaluating physical functioning (2), well-being (3), daily activities (4), perceived health (1) Women who suffer from urinary urge incontinency (UUI) find their Quality of Life (QOL) affected. No diseasespecific questionnaire was found in medical and specialized data banks, to assess QOL in that particular condition. We decided to develop a disease-specific selfadministered questionnaire for UUI in women using a pre-existing pelvian handicap evaluation scale (MHU) as a point of departure. A multi-disciplinary team of specialists consisting of tional cross-sectional study. The QOL questionnaire should be filled in four times during the one-year follow-up by 10,000 patients. The monitoring of the questionnaires is performed by a European centre in charge of their reception, data entry, analysis and which provides logistical and relational support to stomacare nurses throughout Europe. Monthly data entry allows quality control and monitoring of the study The results of the 903 first patients already enrolled are presented here. In August 1995, a second analysis will be done on the 2,500 patients expected. Results will be integrated in the presentation. The aim of this analysis was to describe the patients' characteristics, acceptability and psychometric properties of the questionnaire including reliability, responsiveness over time and cross cultural differences. QOL scores were also studied according to answers to selected items of the satisfaction scale ('satisfaction with medical care'). and selfefficacy scale ('confidence in changing appliance'). Patients were 63 years old (f13), 54% were men, 68% inactive and 67% had colostomy. The average duration of stoma was 24 days. The number of missing data in the questionnaires were low and more than 75% of the questionnaires were sent back fully completed. The internal consistency reliability was satisfying (a > 0.70). Country differences in QOL were reported, from 55 for Holland to 70 for Sweden. No QOL differences were shown according to sex, age, employment status and type of stoma. The QOL scores improved by + 13 (+16) 3 months after discharge from hospital (effect size = 0.82). The more the patients are satisfied with medical care, and the more confident they are about changing their appliance, the better the QOL. The first results enlightened the importance of the stomacare nurses in training patients to manage with their appliance in order to improve their level of QOL and satisfaction. These results should be confirmed by the long term follow-up. three urologists, two gynecologists and two functional physiotherapists generated the instrument using words and expressions of their patients, broadening the concepts measured in the source scale, formulating the items, grouping the items in dimensions, determining a period of reference and choosing response options. The content validity and the acceptability of the first questionnaire was tested on 20 women with UUI during qualitative face-to-face interviews. A cross-sectional study including 98 patients was carried out for item reduction and psychometric validation. The item reduction was performed using descriptive statistics, principal component analysis, multi-trait analysis and stepwise discriminant analysis according to symptom severity. The psychometric validation consisted in scaling assumptions, internal consistency reliability and clinical validity. A reference general health questionnaire (SF-36) was added to establish concurrent validity The first instrument was composed of 57 items scattered over 11 dimensions. Item reduction analysis deleted 33 items. The final questionnaire resulted in 24 items aggregated in five scales: activities (8 items), emotional impact (5 items), self-image (5 items), sleep (3 items), well-being (3 items). The response options are expressed on a fivepoint Likert scale. The scaling assumption was supported by the principal component analysis and a multi-trait analysis. The internal consistency reliability ranged from 0.69-0.84. The questionnaire was able to discriminate group according to symptom severity (urge seventy, number of urine lost, frequency of urination). Patients' QOL scores deteriorate as the symptoms become more important. The correlations between the dimensions of the specific questionnaire and the SF-36 which measure closely related concepts are higher than in dimensions which explore opposing concepts. Our work enabled us to develop and check the main psychometric properties of the instrument. The responsiveness over time and the reproducibility within the framework of a longitudinal study is currently being carried out. The instrument will then be able for use in clinical trials to measure or compare evolutions in QOL of UUI of women. International quality of life study in patients with ostomy persons with anxiety and explored how these data are correlated and how they can be used to predict the selection of treatment. Five hundred thirty seven persons with anxiety completed a baseline and follow-up psychological distress (SCL-90-R), health status (SF-36) and health state preference assessment (using a rating scale and a standard gamble technique). An anxiety-specific health state classification system, containing five attributes with four levels of functioning, was developed specifically for use in this study. Patients were categorized into four anxiety subgroups based on baseline SCL-90-R scores and DSM-III-R criteria. Sixteen per cent of persons with anxiety did not provide us with usable data on the standard gamble assessment (were confused by interview, provided current anxiety state values worse than death or greater than perfect health). Because of this, most of our analyses are based on our findings from the rating scale technique. At baseline there were statistically significant differences in current anxiety states between those persons with anxiety symptoms only, anxiety disorders only, mixed anxiety and depression symptoms only and mixed anxiety and depression disorders (86 vs. 82 vs. 78 vs. 69; p ~0.001). The odds of receiving a mental health referral were not statistically different for those persons with a low (compared to medium) or high (compared to medium) current anxiety state (1.48, 95% CI (0.71, 3.12), 0.52, 95% CI (0.17, 1.58)). The odds of receiving a psychotropic medication for those persons with a low (compared to medium) current anxiety state was 1 .47,95% CI (0.76,2.86), but was statistically greater for those persons with a high (compared to medium) current anxiety state (2.31,95% CI (1.24.4.31)). Correlations between health status and health state preference measures were moderate at most (0.12-0.46, Pearson correlation coefficient), with both improving at follow-up regardless of the receipt of treatment. In addition to being health state preference values in the more typical descriptive approach, our study explored an innovative use, as a predictor for the selection of a mental health treatment. This study was supported by the Upjohn Company. There are a few studies evaluating the quality of life implications of new drug treatments for acute stroke. Given the potential for these treatments to improve the functioning and well-being of stroke patients, quality of life should be assessed. One of the malor challenges with assessing quality of life is the high likelihood that following a stroke a patient will not be able to complete such an assessment. One practical solution is to request a family caregiver to complete the assessment on behalf of the patient. The purpose of this study was to have stroke patients and/or caregivers complete the Health Utilities Index (HUI) to provide key information about possible systematic differences that may exist between patients and caregivers A total of 74 patients who suffered an ischemic stroke and 41 family caregivers from seven centres in the USA took part in this study. Data from 37 pairs of patients and caregivers were available. Both patients and caregivers completed an interviewer-administered combined Mark II/III version of the HUI and a self-administered demographic questionnaire. Patients also provided information on their clinical history. One of the key aspects of the HUI is that in addition to providing descriptive information on each of the attributes addressed in the questionnaire (nine for the Mark II and eight for Mark III), it is designed to produce a single summary measure of quality of life, the multi-attribute utility, determined from an algorithm based on weights derived from a community-based Canadian sample. The utility scoring function is only available for the Mark II health status classification; the Mark III utility scoring function is expected to be available in late 1995 from McMaster University. Approximately one-half of the caregivers were married to the stroke patient. The mean initial post-stroke hospitalization lasted 8.7 days and the majority of stroke patients were residing in their own homes (78%). The mean time since stroke was 57.3 days (range: l-339 days). Intraclass correlation coefficients (ICC), a measure of inter-rater reliability, were high for the HUI (2 0.65 for 15 of the 17 attributes; Mark II hearing was 0.52 and Mark II pain was 0.64). The mean calculated utility score and its standard deviation for the Mark II (ranging from -0.03 to 1.0) was the same for both patients and caregivers (0.6tiO.19) with an ICC of 0.72. These high correlations suggest the use of caregivers to complete the HUI when patients are unable to do so. This study was supported by Janssen Pharmaceutics. Correlates of health status among persons seeking treatment for anxiety The objectives of this study are to describe (1) the characteristics of persons with anxiety disorders who seek treatment from mental health specialists and (2) the outcomes of such treatment. In this paper we report on the correlates of health status in this sample, based on data collected at entry to treatment. The study is a multi-centre longitudinal design at seven mental health clinics located in four urban areas in the province of Quebec. Persons 460 Quality of Life Research Vol 4 1995 A cross-sectional study of 50 community dwelling individuals 12 months post-stroke is being conducted to examine the relationships between quality of life (SF-36) and the following performance-based items: activities of daily living (OARS IADL/ADL Scale), ambulation (TImed Up and Go), perception (Albert's Test), state of motor recovery of limbs (Chedoke McMaster Stroke Assessment Scale, Impairment Inventory), and gross manual dexterity (Box and Block Test). Complete data is available on 25 subjects. These preliminary results indicate that the mean SF-36 scores for six of the eight health dimensions are not different from population values. However, the mean score for Physical Role Limitation was 56.0, and for Physical Functioning was 65.2; both values are markedly lower than the population normative values of 81.2 and 84.5 respectively. Multiple linear regression analysis determined that a model including ADL, unaffected hand manual dexterity, and perception explains 40% of the variance in the Physical Health Summary Score of the SF-36. A model including age, sex, ADL, perception and stage motor recovery explains 75% of the variance in the Mental Health Summary Score. Preliminary findings suggest that some performance measures may be useful proxies for both the physical and mental components of health-related quality of life in those individuals who are unable to complete questionnaires or interviews. (Funded in part by the Royal Canadian Legion Fellowship in Gerontology, and le Fonds de la recherche en Sante du Quebec: Bourse du R&eau provincial de recherche en adaptation-readaptation.) eligible for the study are those who are visiting the clinic for the first time; have a diagnosis of an anxiety disorder; and speak and read French or English. Following informed consent, participants self-complete a questionnaire that includes the Short-Form Health Survey (SF-36), the Beck Anxiety Inventory (BAI), and questions on pathways to care (e.g., self-referral vs. referral from a general physician or specialist), health service utilization, medications and sociodemographic variables. Statistical methods include analysis of variance and covariance, and multiple linear regression. Over 50 patients have been enrolled to date and data from approximately 100 patients will be presented. Data were analyzed from the first 20 patients enrolled in the study at two clinics, of whom nine had panic disorder with or without agarophobia, and 11 had other anxiety disorders. In comparison with published population norms, these patients reported marked role disability due to both physical problems (mean 32.50, SD 40.64) and emotional problems (mean 23.33, SD 30.78), as well as poor mental health (mean 40.87, SD 19.65). In contrast, levels of physical functioning were high (mean 83.00, SD 15.34), while levels of general health, pain, vitality, social functioning and mental health were intermediate. Patients who were selfreferred in response to media announcements had generally poorer scores on all SF-36 subscales than did those referred by general physicians or mental health specialists, particularly with regard to social functioning, mental health and role disfunction due to emotional problems. Different pathways to specialized mental health care for anxiety disorders are associated with different health status profiles and these differences are likely to influence the outcomes of treatment. Both referral source and baseline The objective of this study is to determine the relationships between performance-based measures of physical function and health-related quality of life as reported by communitydwelling persons one year post-stroke. While health-related quality of life is certainly diminished in stroke patients as compared to healthy individuals of the same age, between 11% and 27% of the stroke population have never been studied because communication and cognitive deficits render them unable to complete QOL questionnaires or interviews. Performance-based measures have been suggested as a proxy to infer quality of life in these individuals, but the relationships have not been evaluated. A project is underway to produce ten conceptually equivalent versions of the Quality of Life in Depression Scale (QLDS) for use in Europe, North America,.Australia and North Africa. The measure is intended to be incorporated in multi-national clinical trials of antidepressant therapy for depression. The original instrument was developed in parallel in the UK and the Netherlands and employed the needs-based model of quality of life. All items in the instrument were derived from interviews with depressed patients. Both original versions had high test-retest reliability and internal consistency ( > 0.85). The Dutch version has been used in studies of depression in general practice and elderly populations. The instrument was well received by both patients and clinicians and had good responsiveness (effect size-2). A standard method of adaptation has been employed in the international study. This involves the use of professional and lay panels to produce conceptually equivalent items in the target languages. The new versions are then field-Qualify of Life Research Vol 4 2995 tested with depressed patients to ensure that they are understood and that their content and format are acceptable. Finally, test-retest reliability, internal consistency and validity are tested in postal surveys. Minor changes to the standard methodology have been made where necessary to allow for cultural differences. Few problems were found in producing conceptually equivalent items in the target languages. This is an advantage of the needs-based approach, as needs are largely culture-free. By avoiding an emphasis on function, individuals with differing lifestyles find the form and content of the original items relevant to them. The translations produced have been shown to be relevant to and understandable by depressed patients in each country, indicating good face and content validity. Completed adaptations indicate that the new versions have adequate test-retest reliability (USA, I = 0.82; Italy, T = 0.95) and internal consistency (Cronbach's cr: USA = 0.94; Italy = 0.96). Correlations with related measures suggest good convergent and discriminant validity. In Italy the QLDS correlated moderately highly with the Nottingham Health Profile sections; energy level (0.71), emotional reactions (0.84) and social isolation (0.88). Low correlations were found with pain (0.28) and physical mobility (0.38). In the United States statistically significant differences were found between the QLDS scores of patients scoring high or low on the depression scale of the Hospital Anxiety and Depression Scale (p < 0.005) and patients scoring high or low on the Hamilton Depression Rating Scale (p < 0.0001). It is concluded that the adapted versions of the QLDS are suitable for use in multinational clinical trials of pharmaceutical treatments for depression. The aim of the studies was to evaluate the psychometric properties and construct validity of the Diabetes Health Profile (DHP-l), which is a recently developed patient selfcompletion disease specific multidimensional screening questionnaire for psychological dysfunctioning of insulin dependent and insulin requiring patients in an ambulatory care setting. Relevant areas and content for the DHP-1 was derived following a series of m-depth interviews with 25 insulin dependent and insulin requiring patients aged 20-65 years, a review of the literature and discussions with health care professionals. The DHP-1 comprising 43 items was completed by 2,239 insulin dependent and insulin requiring patients recruited from 54 hospital outpatient sites throughout England and Wales. Mean age of sample was 39.8 (SD IO) years, range 16-84 years. Fifty-one per cent (n = 1144) were men. According to Cattell's scree test criterion, a forced three factor Principal Factoring Analysis (PAF) with varimax rotation was carried out on the 43 items. Eleven items were excluded from further analysis with item factor cross loadings > 0.30 or item factor loadings < 0.30. A further PAF analysis of the 32 items resulted in three factors accounting for 33% of the total explained variance. The three factors were interpreted as Psychological distress, Barriers to activity and Disinhibited eating. In order to examine the reliability of the DHP-1 factor structure across samples, separate PAF analyses were performed on two subsamples (n = 1108 and n = 1124) and for men (n = 1136) and women (n = 1085). Congruence between factor structures was assessed using Harman's coefficient of congruence where 1.0 represents total agreement. Coefficients of congruence between subsamples were: Psychological distress (0.93). Barriers to activity (0.93) and Disinhibited eating (0.99). Coefficients of congruence between men and women were (1) Cultural Adaptation through Lay Panels. Two separate Lay Panels were organized, one for each linguistic group (French and English). The objective of the Lay Panel was to review in a group format (five to six participants) each of the 34 statements appearing in the Quality of Life in Depression Scale and to elicit participants' opinions and suggestions on the wording and language. Fifty per cent of the participants in the Lay Panels had experienced some form of depression in the past. (2) Content and Face Validity through Interviews. In order to ascertain whether this instrument is relevant and easily understood by the targeted population, a total of thirty people currently suffering from depression were interviewed individually, 15 French speaking and 15 English speaking people. (3) Reliability and Validity through a Postal Survey. In order to test for reliability and validity a postal survey was administered using a test-retest method. Two separate groups were involved; 40 French speaking and 40 English speaking participants all suffering from depression. Overall findings demonstrate that there are no major problems associated with the future usage of the Quality of Life in Depression Scale in Canada, both in its Canadian English and French Canadian versions. The majority of linguistic modifications emanated from the Lay Panels. Although participants in both the English and French Panels were able to reach consensus on the items in the questionnaire they made suggestions on the specific wording. These modifications resulted in the instrument being more relevant to each of the different cultural groups. Results from the individual interviews reveal that some respondents had difficulty with the response format (true/false, yes/no, how you are feeling at the moment). Others identified specific words (i.e. optimist, reluctant, dread) which they had difficulty understanding. Preliminary findings from the Postal Survey show that respondents have a certain amount of difficultly in following directions (giving one response only) and in using the response format (true/false, yes/no, how you are feeling at the moment). The test-retest results are forthcoming. Findings from this study demonstrate the relevance and usefulness of this type of validation. Although issues relating to translation of quality of life instruments have previously been discussed and different methodologies proposed, findings from this research point to significant differences existing within the same linguistic group, in this case English speaking Canadians and British as well as French Canadians with those in France. In order to reflect the linguistic and cultural colloquialism of the particular respondents under study, this type of cultural approach to validation should be considered in the development of future quality of life measures. Prospective assessment of neuropsychological function and quality of life (QOL) in a clinical trial for malignant brain tumors glioma is a generally incurable intracerebral neoplasm that causes progressive neurological, cognitive, and psychological impairment and disability Patients also usually receive therapies that are neurotoxic, including surgery, radiation, chemotherapy and immunotherapy, often in combination. Small gains in survival have been achieved because of the increasing aggressiveness of therapy. However, minor gains in survival are not warranted if patients experience significant declines in their mental abilities and functioning beyond what is caused by the tumor. The usual assessment of patient performance status, the Karnofsky Performance Status KPS) does not measure cognitive functioning or QOL. In addition, it is not possible to tell if a change in KPS is due to tumor recurrence or the effects of treatment, a distinction of utmost importance for decisions regarding further treatment and the evaluation of treatment risk. We propose a short, repeatable battery of tests assessing cognitive function, functional status and QOL that will be integrated into a treatment protocol for recurrent malignant glioma. A similar battery has been successfully used in other non-brain tumor Phase II chemotherapy trials (Gelke et al., Proc AACR, 1995) . These tests will assess neurological outcome along with standard oncology outcomes based on neuroimaging (response rates and time to tumor progression). Glioma patients who recur after primary therapy will be randomized to treatment with the anticancer drug carboplatin, plus either RMP-7 or placebo in a double-blind fashion. RMP-7 is a novel nonapeptide bradykinin agonist that affects transport of small water-soluble molecules across the blood-brain and blood-tumor barriers. Patients will be monitored for time to tumor progression and changes in the following tests: (1) Changes in the test scores over time will be calculated by a reliable change index, and differences between the two groups (RMP-7 vs. placebo) will be analyzed. In this way it can be determined if RMP-7 has an impact on time to tumor progression and any differential effect on neurobehavioural functioning compared to placebo. Developing a disease-specific quality of life assessment protocol for persons with multiple sclerosis (MS) poses a unique set challenges because of the relapsing-remitting nature of this chronic disease, the variety of physical and cognitive symptoms it produces, the lack of definitive therapeutic interventions to manage it and the sociodemographic characteristics of those most commonly affected. The Multiple Sclerosis Quality of Life Inventory (MSQLI) represents the efforts of a multi-disciplinary, international research group to develop an inventory of generic and disease-specific measures that is appropriate for both crosssectional and longitudinal assessment. It is expected that the final version of this measure will be available in a variety of formats to allow self-administration and will be sensitive to between-group comparisons as well as changes in disease status. The development approach used to generate the MSQLI has included a variety of techniques so as to assure its utility for researchers and clinicians, the inclusiveness of assessment domains based on patient perception and assurance of psychometric integrity. A theoretical approach was initially adopted to assure comprehensiveness of domains. A nominal group process was used to: (1) delineate the anticipated applications and modes of administration of the instrument; (2) specify the relevant domains of assessment to be included; (3) operationalize those key constructs utilizing generic measures and disease-specific measures which were identified in aggregate as 'candidate measures' (CM). These were reviewed by three expert panels composed of: (1) persons with MS and their family caregivers; (2) allied health professionals and (3) neurologists who were directed to identify any significant domains that were under-assessed or unaddressed and CM was modified in response to that feed-back. This process generated the Multiple Sclerosis Quality of Life Inventory (MSQLI) that includes two generic measures, the Medical Outcomes Study Short Form 36 (SF-36) and the Sickness Impact Profile (SIP) as well as disease-specific measures. An empirical process was subsequently initiated and is underway. The MSQLI is now being field tested in four MS research centres with 300 clinic attending patients. Reliability analyses of component measures, factor analyses to identify independent dimensions of quality of life for MS patients and multivariate analyses of the domains of quality of life to clarify whether the addition of MS-specific measures improve the ability of generic scales discriminative ability are planned. Through this process we expect to reduce the MSQLI to a practical length and construct a set of test modules that can be selected according to the investigation requirements The details of this two stage process and the benefits of it to assuring instrument quality and investigator implementation will be discussed. Perceptions of patients, physicians and nurses regarding the quality of life of individuals with end stage renal disease A. F. Molzahn The purpose of this study was to explain the perceptions of physicians, nurses and patients regarding the quality of life (QOL) of patients with end stage renal disease (ESRD). The sample consisted of 215 patients (96 with renal transplants, 52 receiving in-centre hemodialysis, 37 receiving home hemodialysis and 30 receiving continuous ambulatory peritoneal dialysis) and their nurses and physicians. Three instruments were used to measure QOL: the Self-Anchoring Striving Scale (Cantril, 1965) , the Index of Well-being (Campbell, Converse and Rodgers, 1976) . and the Health State Utility Score-Time Trade Off Technique (Churchill, Morgan and Torrance, 1984) . As well, data was collected pertaining to the care providers' demographic characteristics, the patients' socio-demographic and medical characteristics and the patients' health status, functional status support and outlook. The reliability and validity of the instruments were examined. There was some evidence for the discriminant and concurrent validity of the independent measures. It was found using Hotelling's t-square that there were some statistically significant differences among the mean scores of the groups. The nurses, on average, rated the patients' QOL lower than did either the patients or the physicians. The correlations of nurses' and patients' scores ranged between 0.19-0.47. The correlations of physicians' and patients' scores ranged from 0.26-0.45. The correlations between nurses' and physicians' scores ranged from 0.35-0.49 on the various measures. A series of multiple regression analyses were performed to explain the patients', nurses' and physicians' perceptions of the patients' QOL. Patients' perceptions of their quality of life could be predicted from their health status, outlook on life and treatment modality (i.e., dialysis or transplant). Nurses' perceptions of their patients' quality of life could be predicted by the patients' educational background, the nurses' educational background and the nurses' perceptions of the patients' functional status and outlook. Physicians' perceptions were more difficult to predict; physicians' perceptions of the patients' support were the only consistent predictors of their perceptions. A series of models of patients', nurses' and physicians' perceptions researching QOL of the patients with ESRD were developed and tested using LISREL. The patients' model, with some revisions, was shown to be consistent with the data collected for the study. The models of the Responsiveness to change (from baseline to third assessment) was assessed by comparing SF-36 scale scores in three groups of patients: those whose KPS had remained stable, had increased and had decreased over time. A repeated measures ANOVA revealed significant group differences over time for the physical, role-physical, social, vitality and general health scales. As expected, patients whose KPS improved over time also had improved SF-36 scores; those whose KPS deteriorated exhibited a decrease in SF-36 scores. The MOS SF-36 health survey is a reliable and valid instrument for use in a cancer population, It can detect differences between clinically relevant subgroups of patients and is responsive to change in performance status over time. The MOS SF-36 health survey is a generic health status instrument designed for use with both the general population and chronic disease populations. To date, the psychometrics of the SF-36 have not been tested among cancer patients. In this study the internal consistency, testretest reliability and clinical validity of the SF-36 is examined when used with a heterogeneous sample of cancer patients. The SF-36 was administered at three points in time: before start of treatment (baseline), four weeks after baseline and three months later. For purposes of examining test-retest reliability, the questionnaire was administered to a subsample of patients one week after the last assessment. The study sample consisted of 485 cancer patients with a mean age of 57 years (SD = 12.1); 58% was female. Primary diagnoses included breast cancer (35%), colorectal cancer (24%), lung cancer (31%) and other (10%). Disease stage at baseline was 19% local, 42% loco-regional and 39% distant metastases. Fifty per cent of the sample was beginning chemotherapy and 50% radiotherapy. The mean score on the Karnofsky Performance Status Scale (KPS) was 78 (SD = 13.6, range 30-100). The Cronbach's a coefficients for the SF-36 at baseline was moderate to high, ranging from 0.66 (for social functioning) to 0.90 (for physical functioning). Alpha coefficients at subsequent assessment points were similar. Test-retest reliability, measured by means of intraclass correlations, ranged from 0.57-0.87. Known groups comparisons were used for testing clinical validity. All differences were in the expected directions. At baseline, patients with local/locoregional disease had significantly (p < 0.01) higher scores (higher scores representing a higher level of functioning) on all scales except for the role-physical and role-emotional scales than patients with metastatic disease. Patients with KPS scores of 80 or higher had significantly (p < 0 ,001) higher scores on all scales than patients with a KPS of 70 or lower. At the second assessment, patients receiving chemotherapy had a significantly higher score on the bodily pain scale than patients receiving radiotherapy. Patients receiving radiotherapy had a significantly higher score on the general health scale than patients receiving chemotherapy. No other differences were observed between these subgroups in SF-36 scale scores. The objective of this study was to develop subscales for the BCQ reflecting different domains of quality of life. The NCIC CTG has collected quality of life data using a selfadministered version of the BCQ on 710 pre-menopausal women with node positive breast cancer enroled in a randomized trial comparing two different adjuvant chemotherapy regimens. The form was administered before treatment, monthly for the six months of treatment and quarterly for 1.5 years of follow-up. The BCQ is a diseasespecific instrument for measuring quality of life in patients undergoing chemotherapy for breast cancer. It has 30 questions answered on a seven point Likert scale and uses the mean answer as a summary score for overall quality of life. The questionnaire was designed to measure the impact of seven different domains of quality of life. The domains include, the consequences of hair loss, emotional dysfunction, physical symptoms, trouble and inconvenience associated with treatment, fatigue, nausea and positive well-being. Each domain was assigned at least four items in the scale. We evaluated the completeness of the data with tables of received vs. expected forms and the number of questions not answered. We then performed three exploratory factor analyses in order to determine if the questions were being interpreted as they were intended and to try and develop subscales for use in future analyses. The data used for the three analyses were baseline forms, on-treatment forms and off-treatment forms. We used a principal components procedure with a varimax rotation selecting the number of factors ranging from 5-15. The factors change in number and composition from baseline to on-study to off-treatment. The on-study factors most closely resemble the intended question groupings while the off-treatment factors are the most different. The factors identified in the baseline data were quite similar to the factors identified in the on-study data. Most of the items associated with physical symptoms, trouble and inconvenience associated with treatment, nausea or consequences of hair loss combined into one factor in the off-treatment data. At the same time four factors emerged that seemed to be indicative of more general quality of life components. These included a domain concerned with fatigue, another combining parts of the emotional dysfunction, consequences of hair loss and positive well-being scales and two individual items asking about the support received from family and friends. There was little variation in answers to these last two items. The original seven factors are apparent while the patient is on treatment although the items do not appear to always load in the expected domain. After the patient completes her treatment there are fewer important components of quality of life being measured. The questionnaire asks questions which are difficult to answer before the start of treatment, although the domain patterns are similar to the expected pattern. The on-study subscales developed here need further testing for validity and reliability but otherwise they could be used in future analyses of the BCQ. The Medical Outcomes Study Short-Form 12 (SF-L?), a shorter version of the validated Medical Outcomes Study Short-Form 36 (SF-36), has recently been developed to provide an alternative, validated measure of health status that requires less effort for the patient to complete. Both instruments can provide summary scores representing overall physical and mental constructs. The relationship between these two instruments requires ongoing study in various patient populations. The purpose of this investigation was to assess the relationship between the Mental and Physical Component Summary scores (MCS and PCS scores) from the 12-item and 36-item versions of the instrument. The SF-36 and SF-12 were administered in the context of a larger study to assess the impact of influenza-like illness on an employer-based sample. It was expected that, maintaining discriminant validity of the measures, the PCS scores and MCS scores would be uncorrelated with each other in either instrument. However, as a demonstration of criterion validity, each component score from the SF-12 was expected to correlate positively and significantly with its correspond-ing component score in the SF-36. Employees in a large company who reported to their occupational health clinic within the first 72 hours of signs and symptoms of influenza-like illness and provided informed consent to participate in the study were enrolled. Sublects were administered the SF-36 (Acute) and an 'acute' version of the SF-12 in a randomized, counter-balanced order on two consecutive days. Data from each instrument were scored according to the authors' recommendations to yield the PCS score and the MCS score. Summary scores were subjected to correlation between and within instruments, and f-test was used to assess whether scores (within-instruments) differed by day of administration. Evaluable questionnaire data were obtained from 43 subjects. As expected, the PCS scores did not correlate significantly with the MCS in either the SF-12 or the SF-36 (I = -0.04-0.16, p's 2 0.31). PCS scores correlated moderately (r = 0.43, p < 0.0037) between the SF-36 and the SF-12. MCS scores correlated highly (r = 0.77, p < 0.0001) between the two instruments. Scores on the SF-36 PCS were higher for subjects completing the SF-36 on day one (mean = 39.7, SD = 8.0 on day one vs. mean = 32.4, SD = 11.4 on day two; t = -2.44, p < 0.0189); otherwise, no effects of day of administration were observed. The SF-12 maintains the discrimination between physical and mental functioning observed in its longer predecessor, even in a relatively small sample of subjects with an acute, short-term, variable illness. The correlation of the PCS scores between the two instruments was moderate and in the hypothesized direction. The correlation of the MCS scores was high and also in the predicted direction. The findings support the criterion validity of the SF-12 and the discriminant validity of the SF-12 and the SF-36 in this population. Differences in PCS scores on the SF-36 by day of administration could be related to changes in symptom severity or to repeated testing. The SF-36, by virtue of its larger number of items, may be a better measure of health status when time and subject fatigue are not an issue. However, the SF-12 appears to be an adequate substitute, especially in situations in which time is short or subject patience is minimal. Is clinical cross-cultural research feasible in Africa? The purpose of this communication is to show that crosscultural research is feasible in an African setting. Cross-cultural research embraces several possibilities: (1) a scientist from one country (usually developed) going to Although clinical experience indicates that health-related quality of life (HRQOL) is adversely affected by deterioration in neurological function in adult patients with brain cancer, precise documentation with data derived from reliable and valid self-report HRQOL measures is lacking. We studied 105 adult patients who had either recently-diagnosed (n = 41) or recurrent (n = 64) high-grade malignant glioma. Patients completed the European Organization for Research and Treatment (EORTC) core Quality of Life Questionnaire (QLQ-CSO), and a psychometrically validated, 24-item brain cancer module (BCM24) at baseline and several weeks later, at follow-up. The attending physicians carried out a standardized neurological examination and completed a physician-rated Mental Function Questionnaire (MFQ) at both time points. The standardized neurological examination included a grading of dysphasia, mental confusion and motor deficit along with other parameters. It also provided for an overall five-category assessment of neurological status ranging from 'definitely better', through 'unchanged', to 'definitely worse'. As expected, patients with either dysphasia, mental confusion or motor deficit (n = 58) reported more cognitive dysfunction Between baseline and follow-up, the neurological status was stable or improved in 75 patients but had deteriorated in 16 patients. In general, HRQOL scores did not change between baseline and follow-up for patients with stable/improved neurological status, but there was a marked decline in several HRQOL scores for patients with a deterioration in neurological status. By means of repeated measures ANOVA statistically significant between-group differences over time were observed not only for cognitive functioning (p < 0.051, visual disorder (p < 0.05) and motor dysfunction (p < O.OOl), but also for physical functioning (p < O.OOl), role functioning (p < O.Ol), emotional functioning (p < O.Ol), social functioning (p < O.oOl), global quality of life (p < O.OOl), fatigue (p < O.OOl), weakness of both legs (p < 0.05)' trouble controlling the bladder (p < 0.05), emotional distress (p < 0.001) and future uncertainty (p < 0.001). Thus deteriorating neurological status is accompanied by profound deterioration in HRQOL in patients with brain cancer. (Supported by Integrated Therapeutics, Inc-a subsidiary of Schering-Plough.) another country (usually developing) to undertake medical or psychosocial research; (2) a scientist undertaking research on recent immigrants; (3) the use of a research methodology developed in another cultural environment; (4) comparative study of different clinical conditions in two different cultural settings and applying similar instruments after careful translation and backtranslation to ensure conceptual equivalence. Funding agencies have often contended that crosscultural research is at best, difficult and at worst, not feasible. They argue that accrual of subjects would be difficult, follow-up would be impossible and, because of cultural differences, study instruments developed in one setting would simply not work in a new cultural milieu. We have two on-going randomized clmical studies in collaboration with the Department of Medicine, University of Zimbabwe, designed to determine the most appropriate way to treat patients with epidemic Kaposi's sarcoma (EKS) and hepatocellular carcinoma (HCC). Since the treatment offered is palliative for both diseases, we chose quality of life (QOL) as the principal measure of success. The study instruments developed in Canada, were the Functional Living Index-Cancer (FLIC) as the core instrument, and disease-specific modules. These were carefully translated and backtranslated into Shona, the main local language in Harare, the capital of Zimbabwe. We recently undertook an on-the-spot audit of 20% randomly selected case files for each of the two studies. The following observations were made: (1) accrual is on target for EKS, but slow for HCC. The slow HCC accrual is related to physician nihilistic attitude and not due to lack of patients; (2) follow-up has been excellent, with 0% loss to follow-up in HCC and 1.8% loss to follow-up in EKS; (3) all case files had informed consent appropriately signed (one had thumbprint); (4) all case files had diagnostic histology forms; (5) clinical documentation was detailed and above-board; (6) QOL data appeared clean with little or no missing data; (7) preliminary analysis of the data revealed the FLIC functions well in the Shona culture and appeared to exhibit the same factor structure as was observed in North America. We conclude that clinical research in Africa is feasible and should be encouraged in other areas. It has long been claimed that failure to control nausea and vomiting (NV) after chemotherapy adversely affects QOL, but there is little if any empirical evidence demonstrating that superior anti-emetic control is reflected in better QOL as measured by a validated instrument. In a study comparing dolasetron (dol) to ondansetron (ond) + dexamethasone (dex) 696 patients were randomized in a factorial design to receive one of four regimens (IV do1 f IV dex or IV ond f IV dex) immediately prior to MEC. NV were measured by patient diaries. QOL was assessed with the EORTC QLQ C30+3 administered prior to chemotherapy and on day 4 or 8 after chemotherapy. Overall the administration of dex markedly improved NV both acutely and over the seven days post MEC while the differences between do1 and ond were moderate acutely and disappeared over the post 24 hour period. These differences were reflected in QOL assessments. When changes in QOL scores from baseline were compared in patients receiving or not receiving dex, significant differences in favour of dex were found for: global QOL (p = O.OOOl), physical (p = 0.005) and social (p = 0.001) function and m the symptoms fatigue (p = O.OOOl), pain (p = 0.004), NV (p = O.OOOl), appetite loss (p = O.OOOl), constipation (p = 0.015) and diarrhea (p = 0.007). In the do1 to ond comparison the only significant QOL difference was in the symptom constipation (p = 0.004). Anti-emetic therapy which results in substantial improvement in NV control also produces detectable benefits with respect to broader aspects of QOL. Inst. 1993; 85: 365-76) . The results indicated that reliability coefficients (Cronbach's IX) for multi-item scales ranged from 0.52-0.89. The lowest coefficient was for the two-item role function scale, which was 0.54 before treatment and 0.52 after treatment. On the basis of this result and a re-examination of the wording of the two items, the wording of the items was changed from 'Are you limited in any way in doing either your work or doing household jobs?' and 'Are you completely unable to work at a job or do household jobs?' to 'Were you limited in doing either your work or other daily activities?' and 'Were you limited in pursuing your hobbies or other leisure time activities?' The previous 'No' and 'Yes' response options were changed to a 4-category response format, in keeping with the response format of 'Not at all', 'A little', 'Quite a bit' and 'Very much' used for most of the other items in the QLQ-C30. In addition, a new item 'How would you rate your overall health during the past week?' was added to the global quality of life scale with the intent of replacmg an original item asking for a rating of 'overall physical condition' but retaining an original item asking about 'overall quality o_f life'. This modified questionnaire, termed the QLQ-C30(+3), including the original and new items, was tested in 696 patients with heterogeneous cancer diagnoses in multicentre studies in Canada and 485 patients at the Netherlands Cancer Institute. In the Canadian studies, Cronbach's alphas for the role function scale varied from 0.59-0.67 over three time points for the original questions with a dichotomous response format and from 0.81 -0.88 for the revised questions with a four-category response format. In Amsterdam, Cronbath's alphas were 0.26-0.35 over three times points for the original version and 0 78-0.86 for the revised version and test-retest reliability over five days was 0.59 and 0.82 respectively. Alphas for the global quality of life scale changed modestly, from 0.83-0.90 for the two-item scale to 0.90-0.95 for the three-item scale. However, the new twoitem scale had alphas ranging from 0.79-0.87 as compared to 0.81-0.86 for the original two items in the Amsterdam sample. Test-retest reliability was 0.83 and 0.82 respectively. Based on these results, the two new role function items using the four-category response format will replace the previous two items with the dichotomous response format. In the global quality of life domain, since the original intent was to replace the 'overall physxal conditzorz' item with the new 'overall Iz~nlfh' item, the new item will be retained and used with the previous item about 'overall qua&y of life'; the previous item asking about 'overall physical condition' will be discarded. In addition, in the NCIC studies, Cronbath's alphas for the physical function scale using the dichotomous response format ('No' and 'Yes') and the fourcategory response format as indicated above) were compared. Alphas for the four-category response format were 0.81-0.84 compared to 0.70-O 76 for the dichotomous response format. The four-category response format is undergoing further field testing. Pending the results of these trials, the current version of the QLQ-C30 (2.0) now incorporates the new role functlon and global quality of life scales Significance of changes in health-related quality of life (QOL) scores in women receiving chemotherapy for recurrent or metastatic breast cancer 468 Qua&y of Life Research Vol 4 1995 indicating 'moderate' or 'very much' change had corresponding mean change scores ranging from 4.1-30.0 on the QLQ-C30. Thus, despite a lack of a high correlation between all of the domains in the SSQ and the corresponding QLQ-C30 domains, the changes in the average QLQ-C30 scores appeared to be meaningful in the expected direction and magnitude. D. Osoba, G. Rodrigues, J. Myles and J. Pater In studies of QOL, it would be desirable to know whether statistically significant numerical changes observed over time or between patients are perceived as being significant to the subjects themselves. However, determining the 'subjective' significance of differences in scores derived from multidimensional QOL instruments may be complicated by the perceived meaning (to the subject) of the wording in the scales. In this study, we assessed whether the perception of change over time in certain QOL domains was accompanied by a corresponding change in the direction and magnitude of the responses in the analogous domains in the EORTC QLQ-C30. The study sample consisted of 154 women with recurrent or metastatic breast cancer who were receiving chemotherapy with either doxorubicin alone or doxorubicin and vinorelbine every 3 weeks in a phase III trial. All women completed the QLQ-C30 before the first (baseline completion) and third (follow-up completion) cycles of chemotherapy. At follow-up they also completed an additional four-item questionnaire, the subjective significance questionnaire (SSQ), which asked 'Since the last time I filled out the questionnaire, my physzcnl condition/my emotional state/my ability to enloy social life/the overall quality of my life, is: very much worse/moderately worse/a little worse/about the same/a little better/moderately better/very much better'. Non-parametric Spearman rank correlations for the changes between baseline and follow-up in each of the domains were calculated in order to determine whether patients' responses about changes in physical condition/emotional state/social life/overall QOL in the SSQ were related to the responses in the physical, emotional and social function and global QOL domains in the QLQ-C30. The highest correlations for like domains were for social life and social function (0.46) and for overall QOL and global QOL (0.43). However, the correlation between emotional state and emotional function was 0.32 with a higher correlation between emotional state and global QOL (0.42). Similarly, the correlation between physical condition and physical function was 0.21, while the correlation between physical condition and global QOL was higher (0.48). Thus, the subjects appeared to interpret social life and overall QOL on the SSQ as being similar to social function and global QOL of the QLQ-C30, but interpreted emotional state and physical condition to be similar to global QOL rather than the corresponding domains of the QLQ-C30. When the direction and magnitude of change m the scores between baseline and follow-up were compared between the corresponding domains on the SSQ and the QLQ-C30, there was reasonably good correspondence between the scores. Patients reporting 'no change' on the SSQ had little difference in QLQ-C30 scores (range f3.3-7.2), while those reporting change on the SSQ usually indicated change in the appropriate direction in the QLQ-C30 domains. Those well-being (11 items), physical-functional well-being (eight items), symptom distress (eight items), and attitude of worry (three items). Internal consistency (Cronbach's a) ranged from 0.52-0.91. The construct validity analysis indicated more than 50% significant correlations between subscale scores in the OMFAQ, the KPS and the QOL-CA. The highest significant correlations were found between patient ratings on the OMFAQ and similar QOL-CA subscale scores, while lower correlations were found between unlike concepts in the three instruments. QOL-CA psychological-existential well-being correlated with OM- The purpose of this study was an initial attempt to develop a valid, reliable instrument capable of measuring a construct of interest to the disciple of nursing, i.e. 'quality of life'. The building of the instrument relies on Newman's nursing framework of health. Newman considers health as the evolving pattern of the whole of life. Her theory of health involves the concepts of movement, space and time all correlated to consciousness. Two of her theoretical propositions are important in the building of the Auto-Patterning Index (API): (1) time (24 hours) is a function of movement (rhythmic behaviours); and (2) movement is a means by which space and time become a reality. Movement is a fundamental unit of analysis in Newman's theory and implies at the macro level, 11 daily rhythmic behaviours for the present study. Within the Newman perspective human patterning expressed in rhythmic behaviours can be viewed as rhythmic expressions of QOL. The rhythmic behaviours are operationalized by duration, rate and amount. The API is a self-assessment tool which involves 11 daily rhythmic (occurrences of similar behaviours within similar intervals) behaviours (each one measured for frequency and duration). A pilot study involving 11 behaviours (sleeping, eating, drinking, eliminating, working, hygiene, leisure, physical activities, learning activities, personal communications and others) was conducted. The 11 items, consistently appeared in the literature and were chosen based on face validity. A revision of the items by three nurses familiar with the concept of patterning and operating independently, insured the interrater reliability. The convenience sample of 23 women (who agreed to participate) between -16.5 pounds (underweight) to +22 pounds (overweight) according to the revised Metropolitan Insurance Chart, were asked to complete the index and 22 did so without difficulty. The relationship between scores obtained on the API and the women's weights (used as a criterion) was analyzed by the stepwise multiple regression procedure (SPSS). The results showed that with 11 variables and ten DF, the R2 = 0.97 and the adj R2 = 0.88. The results obtained were considered promising enough to pursue the validation of the API. A three measurement period study followed with two groups of women (clinic and non-clinic) involving 50 women per group at time one, 31 women per group at time two and time three. These women who signed a written consent were 12-45 pounds overweight at time one and between the ages of 25 and 45. At three points in time, multivariate analyses of variance with an a set at 0.05 were used to identify and measure the relationships between the scores obtained from the API and the women's weights. Results showed that the two groups differed on duration and amount of their daily behaviours at time two and time three of the study. However, the frequency of their behaviours was not statistically significant. The daily behaviours that were dissimilar between the groups were at time two: hygiene, sleep, learning, elimination and eating which were increased for the clinic women. At time three, the significant behaviours were: hygiene, elimination, drinking, sleeping and eating which remained increased for the clinic women who maintained their weight loss. Furthermore, the women who lost weight at time two and maintained their weight loss at time three, showed an increase in their combined behaviours and the summation of the duration of their behaviours in comparison to the non-clinic women. Communicating, working, passive activities, physical activities and 'other behaviours' were non-significant at time two and time three between the groups. Further research is recommended to continue the development and evaluation of the API based on Newman's framework. Nausea and vomiting (NV) after MEC is usually most severe in the first few days after treatment and resolves by the end of one week. In a previous study (J C/in Oncoll994; 12(5): 1050-1057) we found modest correlation between control of NV and QOL measured at day 8. We postulated that, despite the 7 day TF of the questionnaire, patient assessments reflected the most recent days' experience. We therefore developed this study to explore whether administering a QOL questionnaire at different days after MEC and/or with different TF would affect the results reported. Six hundred and ninety-six patients participating in a clinical trial of anti-emetic regimens for MEC were randomized by centre to one of four different QOL schedules. Each patient completed the EORTC QLQ C30+3 prior to chemotherapy and on day 4 or day 8 after initiation of chemotherapy, using one of two different versions of the questionnaire. One version framed questions with reference to the previous three days, while the other version framed questions with reference to the previous seven days. For each patient, results were expressed in terms of change in QOL scores from baseline (CS) for each component of the QLQ C30+3. A general linear model was used to analyze the effect of DA of the questionnaire (day 4 vs. day 8) and the TF of the questions (3 days vs. 7 days), by controlling for the effect of antiemetic regimens in the study. A test of interaction between DA and TF was performed within the general linear model for each of the components of the QLQ C30+3. Initial analyses indicated that there was an interaction between the effects of DA and TF on social function (p = 0.068), global QOL (p = O.OOl), and NV (p = 0.004). Results are therefore presented separately by DA and TF. Patients completing the QLQ C30+ 3 at day 4 had significantly worse social function than those at day 8 (CS: -11.4 p = 0.282; CS for NV: 14.5 vs. 18.4; p = 0.163) . Both DA and TF affect patient reports of QOL after MEC. The first is not surprising, given the time course of NV after MEC. However, the fact that patients distinguish between 3 day and 7 day TF was unexpected and, we think, noteworthy. Besides morbidity and mortality, Quality of Life (QOL) is the important outcome parameter following OLT. The purpose of this study was to develop a QL-Index and provide QOL data in liver transplant patients. Between January 1992 and March 1993 a disease specific QL-Index (LSQL) was developed and validated in 347 comparable adult patients with chronic liver disease as well as before and after OLT. Initially, 82 possible relevant items could be reduced to 28 items by analyzing three different sets of questionnaires in 147 patients. Item reduction was achieved according to their frequency and ability to discriminate. The questionnaire was validated by testing for internal consistency, sensitivity and reliability and by comparison with a known general health measure . Internal consistency of the variables as assessed by Cronbath's a was found to be > 0.8 in all dimensions of QOL. A strong correlation with the number of associated diseases was found and 14-28 day test-retest reliability in stable patients was r = 0.7. Discrimination among different stages of disease was higher when compared with the SF-36. is on a 7-point scale ranging from 1 (completely dissatisfied) to 7 (completely satisfied). A total score is then computed from the simple sum of the 12 items. Its construction followed that of the survey questions designed by Campbell, Converse, and Rodgers (1976) . The use of an aggregate sum score to represent overall quality of life follows a linear additive model of unitweighted satisfaction judgements which, historically, has been found to be superior than any other model in predicting overall well-being. Recently, some domain-based life satisfaction measures have adopted a multiplicativeadditive model which allows the rater to subjectively evaluate the importance of each domain to their lives. In this model, an overall score would then be the sum of multiplicative composites ('importance x satisfaction' ratings). A quantitative approach to address this issue was explored with a version of the PQOL which included importance and satisfaction ratings. Preliminary results suggested that the multiplicative model did not add appreciable variance to the prediction of well-being/quality of life outcomes. Therefore, the PQOL maintains a simple, additive model. The theoretical perspective implied in the measure has been commonly labelled as 'bottom-up', indicating the direction of causality from domain to global well-being. Evaluation of the psychometric properties of the PQOL has been based, to date, on three community samples (total n = 625) and five undergraduate student samples (total n = 1081) in the context of various research programmes. Internal consistency for the entire scale was found to be good across all samples (average a = 0.75 for student samples; average a = 0.80 for the community samples). In addition, principal components analysis consistently reveals that the scale is best described by a uni-factorial solution. These results reflect that, overall, the items of the I'QOL are assessing a common latent variable (factor) which can be labelled life domain satisfaction. Test-retest reliability data were also collected from a small subset of one of the student samples (n = 53). Over a time interval of approximately 1 month, I'QOL total scores demonstrated adequate temporal stability (Y = 0.75, p c 0.001). Convergent validity was obtained with other measures of subjective well-being (Positive and Negative Affect Schedule-Watson, Clark and Tellegen, 1988; Satisfaction with Life Scale-Pavot and Diener, 1993 ) and a behavioural domain-based measure of quality of hfe (Quality of Life Questionnaire-Evans and Cope, 1989) . Across the community samples, the overall I'QOL score showed fairly strong positive correlations with global life satisfaction (r's range from 0.47-0.62, p < 0.001) and positive affect (r's range from 0.42-0.52, p c 0.001). Correlations between the I'QOL and negative affect were moderate and in the expected direction (r's range from -0.31--0.57, p < 0.001). In addition, strong correlations between the total PQOL score and the aggregate behavioural quality of life score were obtained (r's range from 0.46-0.68, p < 0.001). Similar patterns emerged from the student samples. Additional correlational data with measures of psychological well-being, personality measures (neuroticism, extraversion, self-esteem, optimism, coping, social support and control), and social desirability have provided further evidence for the validity of the PQOL as a measure of well-being, quality of life or psychological adjustment. The PQOL can be viewed as being a complement to any subjective well-being or quality of life assessment battery. It was developed with particular attention focused on convergent validity with other measures of well-being and quality of life as well as empirical attention given to the validity of the underlying additive model. It provides a sound assessment of the cognitive (as opposed to affective) component of well-being or quality of life. Currently, its use is being explored in theoretical studies regarding the role of satisfaction judgements in the assessment of wellbeing and mental health as well as part of a quality of life outcome battery in several programme evaluation studies. With sound psychometric properties established among normal populations, its use can now be explored among clinical samples within the context of intervention planning or programme evaluation. The purpose of the present study was to examine the validity of the Swedish SF-36 Health Survey. First, the internal structure of the eight subscales was tested according to their expected relations with the two major dimensions of health-physical and mental health, and second, the hypothesized theoretical relationships between subscales and external criteria were examined. Similar methods were used as in the corresponding tests of the original USA version. The study was performed pursuant to the IQOLA Prolect. The subscales measuring physical functioning, role limitations due to physical health problems and bodily pain were hypothesized to: (a) be most highly correlated with a physlcal health dimension; (b) be most valid in distinguishing groups differing in medical conditions; (c) show little association with a mental health dimension; and (d) perform less well in distinguishing groups differing m severity of mental distress. Likewise, the subscales measuring mental health, role limitations due to emotional problems and social functioning were expected to: (a) be most highly correlated with a mental health dimension; (b) be most valid in distinguishing groups differing in mental distress; (c) show little association with a physical health dimension; and (d) perform less well in distinguishing subjects differing in medical conditions. The vitality and the general health perception scales were expected to be moderately correlated with both the physical and the mental dimensions and to distinguish groups differing in both physical and mental health status. Data came from the well-known population studies of health and ageing-H70 and the Population Study of Women. Cohorts were here studied representing the middle-and old ages; 38, 50, 62, 70, 74, 78 and 84 years (n = 1295; 83% of them were females; the mean age was 66 years and the participation rate was around 80%). From this subject pool four mutually exclusive groups of subjects were selected according to external criteria: (a) no medical conditions and no mental distress (n = 165); (b) several medical conditions but no mental distress (n = 141); (c) no medical conditions but mental distress (n = 42) and (d) both several medical conditions and mental distress (n = 105). The relative validity (RV) of the eight SF-36 scales was assessed, both as measures of the physical and mental health dimensions, and as estimates of how efficiently they discriminated relative to each other between the selected groups differing in physical and mental health. The hypothesized dimensionality was tested by rotated principal components analysis of the total sample. Both the principal components analysis and clinical group contrasts agreed largely with the hypotheses and the results from the corresponding USA study, despite differences in subject selection. In both studies the general health perception scale was more associated with physical health. and less with mental health, than expected. This confirms the cross-cultural stability of SF-36. The vitality scale was, however more associated with the mental dimension and the social functioning scale less associated with the physical dimension compared with the USA studies. These differ- (EPO) was also carried out. Of the sample 54% were females and 46% were males. The mean age of the sample was 38.9f7.2 years. The duration of kidney disease was 14.9k4.6 years, the mean number of months on dialysis 45.2k17.6. The dialysis treatment was conducted three hours per week on the monitors of the firms 'Gambro' and 'Althin' with bicarbonate on the dialysate and system of water purification. The patients were in satisfactory somatic state. However the tendency to anaemia was observed (hemoglobin 90.4s1.7, hematocrit 27.9f9.8). Assessment psychological techniques were used including the MMPI, Rosenzweig Picture Frustration study, the Karnofsky scale, the self-report scale of selffeeling, activity and mood and the questionnaire of attitude to illness. Psychiatric observation of patients was also carried out. The data analysis included paired Student's and Wilcoxon's criteria, Kendall's and Friedman's criteria, and linear regression analysis. Chronic hemodialysis patients showed nearly-normal index of social adjustment (GCR = 63. 6k17.8, norm -64.3). Forty per cent of patients were employed full-time or students. The level of employment depended positively on the education, negatively on the anxious type of attitude to illness and on the duration of illness and dialysis treatment. Patients' level of employment was not connected with their somatic state and intelligence. Frequency of psychic disorders among the unemployed patients was three times higher than among employed. As a whole, 80% of patients had asthenic syndrome, 46%had sleep disorders and 38% had depression. The sphere of personal contacts was a source of psychological conflict only for 14% patients. Seventy-six per cent of patients gave positive appraisal of such spheres of life as social activities, family, sex and work The subjective quality of life diminishes mainly because of the insufficient physical activity of patients. The objective quality of life of patients was also satisfactory. During the treatment with EPO the Karnofsky scale index increased from 64.2-68.5 and symptoms of depression decreased. Improvement of mood reached the pathologic level (euphoria) in 26% cases. Improvement of self-feeling and physical Despite uncertainty surrounding the use of proxies, health professionals have depended on them to assess health status of elderly individuals unable to provide this information. This study set out to determine the type of proxy likely to provide reliable information on perceived health status of elderly patients according to their current living situation. We wanted to estimate agreement between elderly subjects and their proxies on their perceptions of health status, when patients were in-patients in a rehabilitation setting or out-patients in a day-hospital and to determine if there was consistently lower or higher ratings by type of proxy, health professional caregiver or significant other. Eighty-three consenting elderly patients who were receiving rehabilitative treatment in the two different health care settings were entered into the study. Each completed a generic health-status questionnaire, the SF-36, and named a significant other and a health care provider who knew them well. These individuals completed the same questionnaire on behalf of the elderly patients. Questionnaires were completed in either English or French. Agreement between patient and proxy was assessed pair-wise (patient/health professional and patient/significant other) using Intraclass Correlation Coeffficients. The possibility of a biased rating by type of proxy was determined via a 2-way ANOVA. The study contained 203 individuals grouped into 79 patient/ health professional pairs and 41 patient/significant other pairs. Analysis revealed that the English and French patient groups were similar. When concordance within respondent pairs was examined, results indicated only poor to moderate agreement regardless of setting and type of proxy for all subscales except physical functioning where the agreement was substantial. );however54%ofthepatients withoutantiretroviraltherapywereatstageC.Scoreswerenot related to age, sex or leve1 of CD4 (except for energy: r = -0.24, p = 0.04).Afterameanfollow-upof6months,20HIV-infected patients had died. Univariate analysis revealed that mean scoresofphysicalmobility,painandenergyweresignificantly higher in this group of patients. Also, mean Karnofsky score andCD4levelwereassociatedwithadverseoutcome,However, multivariate analysis indicated that only Karnofsky and emotional reactions scores were independently associated with adverseoutcome. Despite the low number of patients, these results demonstrate the feasibility and usefulness of assessing the QOL of patients with a generic questionnaire. In this study, the NHP discriminates between patients infected with HIV and patients with malignancy. In addition, such a profile enables researchers to identify the dimensions which are mostly affected in subgroups of patients (i.e. social isolation and sleep in IV-drug users). Further studies should assess the prognostic value of NHP scores. Analysis of research published over the past ten years revealed six patterns in quality of life measurements among populations that included cognitively impaired adults. The patterns were: limited conceptualization of quality of life; reliance on proxies; exclusion of all cognitively impaired subjects; inclusion of data contributed by cognitively impaired sublects without verifying comprehension; inability to determine clinically significant change; and limited psychometric support for potentially useful measures. Preliminary observations are reported from two current studies involving adults who have some cognitive impairment associated with brain injury, multiple sclerosis, schizophrenia or severe obstructive sleep apnea, and cognitively normal proxies. Measures being used include the Sickness Impact Profile and SF-36. Four patterns are emerging: low tolerance for ambiguity; literal interpretation; importance of validity; and differences in how life is experienced. Recognition of these patterns raises questions about attention and comprehension of cognitively normal subjects who appear to ignore ambiguity, make interpretations at varying levels of abstractness and refrain from commenting on what they consider irrelevant. It is argued also that the importance ascribed to how life is experienced is the barometer by which quality of life data in populations including cognitive impaired adults should be interpreted. chosocial and physical well-being of patients, the subjective perception of the treatment and treatment-related stressors, the consequences of hemodialysis on family, social and vocational areas and the perception of life-satisfaction and quality of life of the patients. The semi-standardized clinical questionnaire pertained to medical data, the current psychological well-being, impairments caused by treatment setting, social rehabilitation, vocational or scholastic rehabilitation, anamnestic data and sociodemographic data. One hundred twelve dialysis patients from three different centres were included in the study, 92 adults (23-84 years of age) and 20 children (13-18 years of age). Statistical analysis used the SPSS system and was oriented towards correlation of indicators of well-being and clinical data, as well as comparisons between subgroups of the study sample. The results of the study show that hemodialysis is subjectively perceived as satisfactory treatment of modus which, however, is associated with massive psychological and somatic impairments which are perceived as stressful by the patients. Quality of life is reported to be reduced through hemodialysis treatment. There is social isolation in terms of increasing distance from friends and acquaintances, accompanied by a retreat in family and private life. The majority of the patients are to be categorized as insufficiently rehabilitated in terms of vocational or scholastic integration. The treatment related impairments of the dialysis are poorly tolerated by a high percentage of patients, especially younger patients have great difficulties with the treatment which are connected to compliance problems. The study shows that there is an acute and massive need to improve the psychosocial care of chronically ill dialysis patients especially in young age, as well as suggesting a deficit in adequate instruments to assess quality of life in this patient population, specifically as concerns chronically ill children. The disparity between subjective satisfaction with the treatment and its perceived effects on well-being can be explained by the lack of perceived treatment alternatives, as well as the wish of the patients to lead as normal life as possible. The cognitive evaluation of the treatment can be seen as an adaptation and coping strategy through which toleration of the impairing effects of dialysis is possible. measures in bipolar patients. Asample of 70 bipolar patients will be recruited from two clinical centres: 30 with major depression, 30 who are clinically stable, and 10 with acute mania. All patients will have confirmed DSM-IV diagnosis of bipolar disorder and clinicians will complete the Young Mania Rating Scale and the Hamilton Depression Rating Scale at study entry and at 8 weeks. The MOS SF-36, cognitive function scale, Mental Health Index-17, Endicott Work Productivity Scale and the Quality of Life in Depression Scale will be administered at baseline and after 8 weeks. Data collection is currently underway and will be completed by August 1995. Internal consistency reliability and intraclass correlation coefficients will be calculated for all clinical and health status scales. The relationship between the different clinical and health status measures will be assessed using correlations. Regression analysis and paired t-tests will be used to evaluate changes in health status in the depressed group. The findings of this study will assist in the planning of health status outcome components of clinical trials of medical treatments for bipolar disorder. (2) to evaluate the usefulness of the EORTC QLQ-C30 in this population. Subjects were 212 community based patients eligible for palliative care services. 69% were diagnosed as having cancer, with multiple sclerosis forming the only other major group. The average age of patients was 65; 60% were female. All patients were asked to complete the EORTC QLQ-C30 questionnaire, with help if required. The baseline questionnaire was completed within 3 days of diagnosis, the first follow-up within 7 days of baseline, the second and third follow-up at 28 day intervals and the fourth follow-up at a further interval of 3 months. In general, patients complied well with completing the questionnaire. The performance of the questionnaire was consistent with known data on its use with other patients with advanced cancer. Observing trends in the data showed that patients who go on to complete the next follow-up have slightly better than average scores, whilst patients who either die or leave the setting at the next follow-up have worse scores. Although these differences are small they are consistent for the functioning, symptom and global quality of life scores. Significant improvements in symptoms scores between baseline and first follow-up are observed. Patients without carers report significantly better functioning and global quality of life scores compared to those with carers, although this is not true of symptom scores. The use of self reporting quality of life questionnaires as a means of assessing palliative care patients appears to be a vaiid approach. The results presented here suggest that such measures can provide interesting insights into the progression of such patients over time with the possibility of yielding prognostic information (HYE) alternative to the QALY has debated whether the HYE and time-tradeoff (TTO) methods of preference assessment are essentially the same. Defenders of the HYE as different allege that those who argue the equivalence do not appreciate the important differences in measurement of preferences under conditions of certainty and uncertainty. The HYE is alleged to measure preferences under conditions of uncertainty whereas the TTO does not. The resulting ambiguity in the literature is resolved in this work where the equivalence of these methods is proven. The HYE in its two-stage process does in fact inject and then subsequently remove the aspect of uncertainty from the assessment process, making it, at best, a cumbersome equivalent to the TTO. Development of a scale to assess quality of life in persons undergoing cochlear implantation those who responded to the questionnaire, the reliability of the scales was high with few missing values. Significant differences between groups supports the validity of these scales which are currently being used in a longitudinal observational study of cochlear implants. G. de Lissovoy); Johns Hopkins University, Baltimore, MD, USA (J. R. Wyatt, J. K. Niparko) The purpose of this study is to describe the development of an instrument designed to assess the quality of life (QOL) of persons undergoing cochlear implant implantation. Cochlear implants are devices used to provide the sensation of sound to those who are profoundly deaf, but unable to benefit from conventional hearing aids. While many believe that these devices represent a tremendous benefit to the recipient, only one large trial has specifically examined quality of life after receipt of a cochlear implant. As part of the planning for a study of the QOL and economic outcomes associated with cochlear implantation, we developed a disease specific scale designed to detect the improvement in QOL after cochlear implantation. We followed established procedures for the development of a disease specific measure of QOL. We first reviewed the limited information on QOL for this population and the scales currently available; however, none covered the relevant dimensions of QOL in sufficient detail to detect expected differences due to cochlear implantation. Two scales were developed: the Hearing Quality of Life (H-QOL) scale and the Hearing Work (HW) scale. Items for each of the scales were developed through review of literature on QOL in hearing impaired persons, review of existing scales, interviews with persons with profound hearing loss who had received cochlear implants, and focus groups with persons with severe to profound hearing loss who had not received an implant. Cognitive interviews were conducted with profoundly hearing impaired persons to detect difficulties in understanding items and instructions. Revisions were made based on these interviews. A pilot study was conducted in which a questionnaire including the new scales as well as other QOL scales selected for the trial were mailed to 350 persons who had received a cochlear implant and 60 persons who were eligible for, but who had not received an implant (86%and 52% of implantees and non implantees respectively, returned usable questionnaires). The questionnaire included the SF-36, Psychological General Well-being (PGWB) scale, Health Utilities Index (HUI), items related to hearing impairment and demographic information. Internal consistency of the H-QOL and HW were 0.89 and 0.88 respectively. Factor analysis yielded three factors for both the H-QOL and HW scale. Sixty-one per cent of the variance was accounted for in the H-QOL scale and 74% in the HW scale. Both the H-QOL and the HW differentiated between those who had received an implant and those who had not (p 5 0.05). Questions related to work satisfaction showed a trend in the expected direction; this difference was not significant. Significant differences between the groups were also found on the HUI and SF-36 (social function). Differences on the PGWB approached significance (p 5 0.06). The relatively high rate of response, especially among the implant group, and the many positive comments written on the questionnaire suggest that a mailed survey using the described battery of instruments is feasible. Among Hope and quality of life in cancer patients-important issues in nursing practice and research T. Rustsen Oslo College, Department of Nursing Education, Oslo, Norway During the last years more and more people are diagnosed with cancer. Cancer might be a threat against life itself and the individual"s perception of quality of life (QOL). Hope is frequently referred to as an important strategy in coping with a disease such as cancer, and as an important aspect of nursing care. The objectives of this paper are to (1) give a review of the importance of hope and QOL for cancer patients and (2) show some results about the correlation between hope and QOL. Hope enables people to cope with difficult situations and suffering. Hope might be defined as a six dimensional dynamic attribute including: active involvement; comes from within; is possible; relates to others; and relates to meaningful outcomes to the individual. Increasing patient's perceptions of hope may result in increased QOL. QOL is being increasingly recognized as an important outcome in nursing research and practice. The concept of QOL is concerned with the experience of the individual's life situation either in general or in relation to different life domains including physical status, functional abilities, psychological status and well-being and social interactions. Traditionally QOL studies have been descriptive and/or comparative. (3) number of days that normal activities could not be completed; (4) amount of work missed; (5) number of medications taken; (6) level of effectiveness at work; and (7) interference with leisure or recreational activities. All data used were collected using a 5-day diary and a followup questionnaire that was mailed two weeks later. For the 43 evaluable employees, the PCS score from the SF-36 was significantly correlated with the symptom assessment ( As expected m a short-term physical illness, the MCS from the SF-36 and SF-12 did not significantly predict any of the work or home-related outcomes. The role of health status should be considered when assessing acute, as well as chronic, illnesses. The PCS of the SF-36 and SF-12 can be effectively used in prospective studies to document the impact of health status on subsequent work productivity and home activities, although the PCS from the SF-36 was a slightly more powerful predictor in this study. The results are significant despite the fact that influenza-like illness is self-limiting and of varying severity. ticipants were followed for 6 months, the 6 month data have been used for the primary analysis in this report. Patients treated with 20mg of lovastatin had a 17% reduction in total cholesterol and a 24% reduction in LDL-cholesterol, while those treated with the 40mg lovastatin dose achieved reductions of 20% for total cholesterol and 28% for LDL-cholesterol. Patients treated with placebo had no reductions in either total or LDL-cholesterol. Complaints of possible adverse events were remarkably similar in the two active treatment groups and the placebo group. At 6 months of follow-up there were no statistically significant differences found in mean change scores from baseline between treatment groups on the cognitive functioning, depression, health perceptions, sleep behaviour, physical functioning or social support scales. Nor were any statistically significant differences found with the global change questions. The CRISP trial demonstrates that lovastatin was extremely well tolerated in an older cohort, both with regards to symptoms and health-related quality of life. Standardized measures of health status predict the impact of influenza-like illness on work and other activities W. B. Saunders, K. Nielsen and J. A. Scott-Lennox Glaxo Wellcome Inc. Self-reported health status data can be a useful predictor of direct and indirect costs of illness. However, limited data exists to document the relationship between standardized measures of health status and other aspects of functioning, such as the ability to work effectively and to perform usual activities at home. The Medical Outcomes Study Short-Form 36-Acute (SF-36) is a validated measure of health status. The Short-Form 12 (SF-12), a shorter version of the SF-36, has recently been developed to provide an alternative validated measure that requires less effort for the patient to complete. Each instrument includes a physical and mental component score. The relationship between these measures of health status and separate measures of functioning at work and at home was examined in an employer-based study designed to assess the impact of influenza-like illness on workers and the employer. An observational study completed in the winter of 1994-1995 enrolled employees who presented to one of four occupational health clinics fever and at least two of four respiratory symptoms (cough, headache, body aches or sore throat). During the clinic visit, employees who qualified for the study completed a baseline questionnaire and a health status questionnaire and a pharyngeal swab was performed for viral isolation. Employees were randomly assigned an SF-36 or SF-12 health status questionnaire to be completed during the clinic visit, with the other to be completed one day later. Data from each health status at the Delhi centre. The pilot version of the WHOQOL was derived from extensive discussions, item generation and item ratings by an international expert group followed by focus group work in all the participating countries. The pilot questionnaire included 41 questions on the importance of the various facets/ subfacets. These questions asked about 'how important various aspects of your life are to you' and 'how much these affect your quality of life'. Each question was rated on a five point rating scale, derived from earlier pilot work. The 305 respondents included 254 adult patients undergoing diagnosis and treatment in a large tertiary care general hospital at Delhi and 51 'well' family members. About half (50.8%) were males and mean age was 40.69 (sd 14.27) years. Ratings were scored between 1 (not important) to 5 (extremely important) and mean score for each item was calculated. The results revealed that the mean scores varied between 3.30-4.52, indicating that all the items were rated as moderately to extremely important. There was no significant difference for any item between the ill and the well subjects. The items rated the highest on the importance scale were: to hear well, to be able to move around, to be able to take care of daily living activities, to see well and to be free of any pain. These items pertained to the physical and the level of independence domains. The items rated the lowest were: support from others, environment, relaxation or leisure, chances to learn new skills and caring and providing for others. These items pertained to the domains of social relationships and environment. The results provide partial validation for the facet structure of the WHOQOL pilot version and also of its stability across ill and well populations. However, limitations of this approach have been discussed. The results also suggest that facets related to physical and level of independence domains are perceived as highly important for quality of life in a developing country like India. There is growing mterest in using the approach in areas of the world where both language and culture are different. The motivation comes from three directions: locally common diseases which merit study; diseases which are rare in one setting, but common elsewhere, such that experience gained in another culture may be transferable; and large scale trans-cultural studies. Two important questions arise. Is there an efficient reliable approach to language translation? Are the underlying constructs in quality of life measures cross-culturally valid? To begin to answer these questions the WHO Collaborating Centre for Quality of Life in Cancer Care has undertaken two studies, one in Singapore (a mixed Chinese/Malay/ English society), and one in Zimbabwe (English/Shona). In both studies the Functional Living Index for Cancer (FLIC) was translated by a three stage process including translation/backtranslation; focus group pilot testing; and formal testing on a suitable cancer population. In Singapore, 246 patients in two major hospitals and a home-based palliative care programme participated in a free-standing validation study. They represented the spectrum of demographics and disease seen in the community. Two changes were made because of cultural differences. The first removed the word 'cancer' from all items where it appeared. The second changed the description of 'household chores', since most Singaporean males, and middle to upper income women do not do such activities. Questionnaire compliance was excellent, and the factor analysis was the same as the original FLIC, with the exception that the ordering of the family and social interaction factors was reversed. In Zimbabwe, 93 Shona speaking patients participated in a similar study, this time in conjunction with a randomized trial of treatment for epidemic Kaposi's Sarcoma. A principal factor analysis was performed on the answers to the 22 items on the FLIC. A six-factor solution seemed the best data fit and explained 56% of the variance. Forcing a five-factor solution provided a result almost identical to that of Morrow et al. (Qual. Life Res., 1992, 2: 287-296) . In both Singapore and Zimbabwe the factor analyses obtained were very similar to those obtained during development of the FLIC in English. This suggests that the quality of life construct may be cross-culturally robust. However, there are important differences in both language and cultural nuance which must be recognized and accommodated. The objective of this study was to evaluate the sensitivity of patients' self-reported health-related quality of life (HRQOL) scores to adverse events (AEs) reported during clinical trials. To be effective measures for research and clinical practice, QOL self-assessment questionnaires must be sensitive to the impact of health problems, such as drug toxicities, symptoms and disease progression on patient functioning and well-being. While other factors beyond clinical AEs are important, the recent occurrence of AEs relevant to a particular QOL dimension should affect its score. Four hundred and fifty-nine HIV patients self-administered the 34-item version of the Medical Outcomes Health Status Questionnaire for HIV (MOS-HIV) at baseline and one subsequent visit during clinic visits for multicentre, randomized, double-blinded clinical trials of a new antiretroviral therapy Based on physician assigned codes and descriptions, AEs occurring within 28 days of the last observed patient questionnaire were assigned to one or more of the following categories: moderate or severe AE, serious (grade 3-4) toxicity, painful, fatigue-related, headache, nausea/vomiting and mood-related. Change in MOS-HIV scores from baseline to last observed questionnaire were compared for patients with and without each of these types of recent health problems within 28 days of their last completed questionnaire. Of these 459 patients, 214 (46.6%) reported no AE in the 28 days preceding the last completed questionnaire. The remaining 245 (53.4%) reported moderate or severe AEs (12.8%), serious (grade 34) toxicities (6.7%), depression (4%), fatigue-related conditions (8.1%), painful conditions (22%), nausea/vomiting (2%) or headache (1.5%). In general, patients who reported a recent AE had lower change scores (less improvement or greater decline) than those without AEs. MOS-HIV scores declined most in patients who had mood disturbances, nausea/vomiting or whose AEs were coded as functional-limiting toxicities. Physical functioning, role functioning, health distress and social functioning varied most in response to AEs, while cognitive functioning and general health perception scores did not vary greatly regardless of the presence, severity or type of AE experienced. Patients experiencing AEs tended to have lower QOL change scores than patients without AEs. Both nature and severity of AEs can help document the sensitivity of QOL instruments to clinically relevant changes in patients' health and functioning. The lack of perfect association between QOL scores and AEs underscores the importance of QOL as a distinct outcome of clinical research. Comorbid anxiety disorder and the functioningand well-being of seriously ill patients of general medical providers From RAND, Santa Monica, CA, USA In recent years, the comorbidity of psychiatric disorders with chronic health conditions has emerged as a topic of considerable clinical and policy interest. There is evidence that anxiety disorders are associated with high use of health services and increased disability. However, the implications for health-related quality of life resulting from anxiety disorders which are comorbid to serious medical or psychiatric illness are not well-understood. Using longitudinal data from the Medical Outcomes Study, we tested the effects of comorbid anxiety disorder on the functioning and wellbeing of general medical patients with serious chronic medical conditions (hypertension, diabetes, heart disease) or depression (current depressive disorder or subthreshold depression). HRQOL was assessed using the SF-36, a multidimensional specification of health status. We estimated the unique effect of any comorbid anxiety disorder on functioning and well-being as well as the differential impact at baseline, year 2 and change over time, of any comorbid anxiety disorder for patients with chronic medical conditions or depression using multiple linear regression methods. We found that general medical patients with comorbid anxiety had lower levels of functioning and well-being than those without comorbid anxiety. These unique effects were most pronounced in mental health HRQOL measures. Comorbid anxiety disorder had an additional negative impact on HRQOL for medically ill patients, especially those with hypertension or diabetes. Hypertensive and diabetic patients with comorbid anxiety were as debilitated as patients with depression or serious heart disease. Over time the emotional well-being and social functioning of patients with heart disease who had comorbid anxiety disorder declined significantly. The finding of substantial differences in quality of life between chronically medically ill patients with and without comorbid anxiety disorder highlights the importance of identifying comorbid anxiety in primary care patients. to NHP and SF-20 in the responsiveness to change. An additive Z-stage or 3-stage valuation model, based on the multi-attribute utility theory, has been used in a postal survey to elicit valuations from the general public for generating a single index score (on a O-l scale) for each possible profile as defined by the 15D health state descriptive system. In the light of response and completion rates the valuation approach was feasible. It also produces valuations with a high reliability as suggested by oneway ANOVA with Tukey multiple comparisons tests, correlation and regression analyses. A rough reflective equilibriumbased validity test and a mapping procedure suggest that the 15D scores exhibit a more plausible trade-off between length and quality of life than those of some other single index score measures (EuroQol, QWB, McMaster). As a profile measure with comparable dimensions the 15D performs equally well as NHP and SF-20, in some respects even better, and much better than EuroQol. Moreover, in comparison with them, the remaining 9-10 dimensions of 15D provide a great extra reserve of discriminatory power and responsiveness to change. The valuation approach is feasible and produces valuations with a high reliability. The resulting scores for health states on a O-l scale exhibit a more plausible trade-off between length and quality of life than those of some other single index score measures and are thus more valid for QALY calculations in cost-utility analysis. thinking about your physical health, which includes physical illness and injury, for how many days during the past 30 days was your physical health not good? (3) Now thinking about your mental health, which includes stress, depression and problems with emotions, for how many days during the past 30 days was your mental health not good? and (4) During the past 30 days, for about how many days did poor physical or mental health keep you from doing your usual activities, such as self-care, work or recreation? Between 67% and 81% of respondents reported that during all of the past thirty days their physical health or mental health was good or better and that poor physical or mental health did not keep them from their usual activities. Selfreported health status was more strongly associated with days physical health not good than with days mental health not good. Respondents with the following demographic characteristics and medical conditions were less likely to report very good or excellent health: older age, blacks, unmarried women, lower education, unemployment, disability, lack of health care coverage, obesity, teetotalling, smoking, hypertension, elevated cholesterol and diabetes mellitus. The 15D is a generic, 15-dimensional, standardized, selfadministered measure of health-related quality of life. It can be used both as a profile and a single index score measure. The purpose of this paper is to examine the properties of 15D in both capacities in terms of acceptability/ feasibility, reliability, validity, discriminatory power and responsiveness to change, and introduce briefly some applications. As a profile measure its properties are compared with the Nottingham Health Profile (NHP), SF-20 and EuroQol by using several data sets and methods. The acceptability as indicated by response and completion rates is comparable to NHP, SF-20 and EuroQol. The reliability measured by repeatability coefficient compared equally, even favourable with that of the NHP, being quite high for both measures. Multitrait-multimethod matrix of correlations between comparable dimensions of 15D, NHP, SF-20 and EuroQol produced solid evidence of convergent and discriminant evidence for the construct validity of 15D. On roughly comparable dimensions and as a whole measure the 15D is superior to NHP, about equivalent to SF-20 and much better than EuroQol in discriminatory power as indicated by percentages of respondents at the 'ceiling' and 'floor'. The effect size and standardized mean response suggest that on comparable dimensions the 15D is roughly The five aims of the study were first, to find out how people in contemporary Britain perceive and represent different facets of their quality of life. This enabled achievement of the second aim which was to find out whether a proposed internationally agreed list of facets was sufficiently comprehensive for use in a cross-cultural measure of quality of life (the WHOQOL) designated for the assessment of health and health care. The third aim was to use a qualitative focus group methodology to encourage patients, health professionals and 'well' people from the community to generate items in the vernacular, for use in a UK version of the WHOQOL. The fourth aim was to determine whether there were broad areas of QOL which were important and specific to QOL locally, which might be appended to an international core of items for use in the UK. Lastly the study aimed to weight and rank for national importance the list of specific areas or facets for their relevance to quality of life in health. Seven 45 minute focus groups were convened with 45 participants split into equal numbers of patients, health professionals and people from the community, including informal caregivers. Local and national statistics were used to derive a representative cross-section of participants with reference to age, sex, socio-economic status, ethnic group, diagnosis, severity and specialty/experience. Participants were asked to free-list those things which could affect QOL positively or negatively. Then they discussed 33 definitions of QOL provided in a manual and suggested ways of asking about these aspects of QOL. Finally they provided 5-point ratings to show the importance of each facet to QOL. Transcriptions and questions generated were used by a four person writing panel to turn them into items and this process will be described. The results from this qualitative data show: (1) Thirtyone out of 33 facets were endorsed as important in Britain and it was possible to derive items for all of these facets for use with the WHOQOL. (2) The validity and comprehensiveness of the facet list is largely confirmed. Omissions included the need for intimacy of loving relationships, job security and fears of unemployment, fears of bereavement or loss of loved ones and concerns about the environment. These were either integrated into the international design or into the UK version. (3) The integrity of five domains was spontaneously confirmed. The happiness facet was seen as so important that is needs to be accorded domain status in the UK. The strengths and weaknesses of the method and design are discussed in conclusion. The results confirm that the proposed facet list and domain structure of the WHOQOL are relevant, important and largely comprehensive to QOL as conceptualized in Britain today. The study is set within the context of an international l4-centre collaborative project. Distinguishing between quality of life and health status in quality of life Quality of life research continues to be hindered by a lack of conceptual clarity regarding the construct of quality of life (QOL). Instruments used for QOL research typically measure various aspects of health without specifying how these domains determine a patient's QOL. The distinction between health status and QOL is further obscured by references to 'health-related quality of life'. We investigated the manner in which patients distinguish health status and QOL in their subjective assessments of these two constructs. We performed a meta-analysis of 11 chronic disease studies that asked patients to make global ratings of QOL (using visual analogue scales or life satisfaction measures) and health status (ranging from poor to excellent health). Instruments used in these studies included the RAND-36, MOS SF-30, EORTC QLQ-30, MILQ and MQOL-HIV. For each study, we estimated regression models using the global rating as the criterion and domain scores as predictors. The effects of individual domains across studies was summarized for each construct by computing standardized regression coefficients weighted by study variance. The results of this analysis indicate that patients rely on two major domains-psychological functioning and physical functioning -to assess both constructs, but that the importance of these domains differs. When rating QOL, patients give greater emphasis to psychological functioning (beta = 0.37) than to physical functioning (beta = 0.20). This pattern is reversed for appraisals of health status in which physical functioning (beta = 0.36) is more important that psychological functioning (beta = 0.12). These results suggests that quality of life and health status mean different things to patients, and that an overall QOL index can be derived by appropriately weighting the relevant domain scores. The use of significant others as proxy raters of the quality of life of patients with brain cancer The aim of the present study was to examine the feasibility and appropriateness of employing significant others as proxy raters of the health-related quality of life of patients with brain cancer. Toward that purpose, we evaluated the level of agreement between patient and proxy responses to the questions of the EORTC QLQ-C30 supplemented by a brain cancer-specific module (the QLQ-BCM). We also examined whether patient-proxy agreement was associated with the patients' level of physical and neurological impairment and with changes in the degree of impairment over time. Finally, we examined the influence of several background characteristics of the significant others on the degree of patient-proxy agreement. The study sample consisted of 103 patients with either recently diagnosed or recurrent brain cancer, and their significant others (75% spouses, 25% relatives or friends). Patients and proxies completed independently the EORTC QLQ-C30 and the QLQ-BCM at three points in time. The physicians evaluated the patients' physical and neurological status using the Karnofsky Performance Status rating, (NL/SP/USA). Differences in psychometric performance between the three countries could be attributed, in part, to differences in sample composition. The EORTC breast cancer module exhibits satisfactory psychometric properties and is currently being subjected to cross-cultural field-testing in 18 countries. In employing self-report quality of life (QOL) measures, it is assumed that the individuals under study have an internalized standard for judging their current level of QOL, and that this internalized standard will remain stable over time. However there is evidence to suggest that cancer patients' internal standards for assessing their QOL may not be as stable as has been assumed. Such shifting of internal standards (termed 'response shift') may jeopardize the validity of results of studies employing longitudinal designs. The use of a retrospective pretest as an extension of the conventional pretest-posttest design has been proposed as a means of avoiding such response shift bias. At the retrospective pretest patients are asked to provide a renewed judgement about their pretest level of QOL. The use of such a retrospective pretest was found to be feasible in an earlier study with 26 cancer patients. The primary aims of the current study are: (1) to determine the extent to which cancer patients experience a response shift as a result of their oncologic treatment; and (2) to determine whether posttest minus retrospective pretest difference scores yield more valid results than posttest minus conventional pretest difference scores. Ninety-nine newly diagnosed cancer patients undergoing radiotherapy were administered a QOL questionnaire (the EORTC QLQ-C30) and a questionnaire on fatigue (the MFI-20) at two points in time: (1) prior to the radiotherapy (pre-treatment) and (2) after completion of the radiotherapy (post-treatment). At the post-treatment session, patients were asked to complete the questionnaires twice (i.e. as a posttest and as a retrospective pretest). Patients were then interviewed about their responses. Criterion measures included: (1) an interview to establish the patients' level of performance status, which was held at both assessments; and (2) a questionnaire on reported change with respect to a range of QOL domains since the start of radiotherapy, administered at the post-treatment session. Contrary to expectation, no evidence for response-shift effects was found for those patients whose health status deteriorated over time. However, response shift effects were found in the area of fatigue, as assessed by the MFI-20, for those patients who reported a stable or improved health status over time. For these patients, the mean fatigue level of the retrospective pretest was significantly higher than that of the conventional pretest, with higher scores indicating a higher level of fatigue (repeated measures ANOVA; p < 0.01). For the entire sample, posttest minus retrospective pretest difference scores were more valid than posttest minus conventional pretest difference scores in that they generally yielded higher correlation coefficients with criterion measures of change. The methodological implications of these fndings for assessing QOL over time will be discussed. Although no significant differences were observed between healthy controls and burned individuals, the correlation analysis revealed that burn severity (%TBSA) affected specific quality of life domains. Indeed, the more severe the injury, the lower the score for physical functioning (r = -0.37, p < 0.03), role physical (r = -0.30, p = 0.06), vitality (T = -0.34 p < 0.05) and role emotional (r = -0.43, p < 0.01). Furthermore, individuals burned more than 30% TBSA (n = 15) reported lower physical functioning than individuals with less severe bums (79.3f26.2 and 94.1f6.6, respectively). Reported levels of general health (p = 0.08) and vitality (p = 0.07) also tended to be lower in individuals with major burns (~30% TBSA). Muscle weakness in the knee extensors was observed (15%-20%. depending on the tested velocity) in subjects that had been burned more than 30% TBSA. Significant correlations were found in the burned subjects between specific domains of the SF-36 and muscle strength of the knee: measure of general health (I = 0.50, p = 0.005), vitality (I = 0.37, p = 0.03), and social functioning (r = 0.33, p = 0.05). This study is the first to demonstrate that a valid and standardized measure of health-related quality of life reflects the severity of bum impact, even years after the initial trauma. (This study was supported by FRSQ-FPQGB.) The influence of the disposable income on health contentment in Germany On the basis of the German Cardiovascular Prevention Study (GCP) and its national surveys it is possible to inquire after aspects of quality of life. We were interested in correlation between the disposable income of the household and the health contentment-one of the components of quality of life (QOL). Using SPSS for Windows 6.0, we tried to discover whether or not there is a coherence between aforesaid elements of the national survey. Question No. 79 asked for the disposable income of the household. As an element of question No. 62 the state of health contentment has to be given. The presentation of the results based on the national survey tl (1987) (1988) , obtainable as a public use file by the Robert Koch Institut (RKI) in Berlin. The question about health contentment required a decision between seven alternative answers (from 1 = very dissatisfied to 7 = very satisfied). The disposable income of the household is placed in 11 groups from 1 = lower than DM 1000 to 11 = over DM 6000. We found two groups: those workers who presented to seven Hamilton, Ontario physiotherapy clinics with soft tissue disorders of the neck and upper limb. Study subjects were interviewed twice during the first week of treatment, once in the last week of treatment (usually 6 weeks later) and two weeks following treatment. A second examiner, blinded to the NULI responses, administered the Sickness Impact Profile, a global question of the subject's perception of overall functional status, a global question of the treating physiotherapist's perception of the subject's overall functional status and carried out physical examinations including hand grip strength, shoulder abduction by goniometer and the 'scratch test'. The third and fourth interviews included a brief questionnaire on work status and re-administration of the NULI items as well as the SIP global questions and physical examinations. The following criteria were used to identify items for exclusion and inclusion: (1) exclusion of items reported not applicable by more than 30% of the respondents (applicability); (2) exclusion of items found to be unreliable (reliability); (3) exclusion of items found to correlate poorly with the overall domain score when correlating each item score with the score of its domain without the item (internal consistency); (4) exclusion of redundant items based on item to item correlations where correlations are > 0.9 (redundancy); (5) inclusion of items demonstrating responsiveness to change measured by correlating the residual change scores of each item from pre-treatment to post-treatment with the residual change scores of the patient and the treating physiotherapist's global assessment measures (responsiveness to change); (6) inclusion of items demonstrating discriminative ability measured by (a) review of histograms for distribution of the responses on each of the points of the seven-point scale for each item; and (b) correlating each item score with the mean of the patient's and the treating physiotherapist's global assessment measures and the score on the global question for that domain (discrimination). The item reduced instrument will be presented as well as a more detailed description of the methods and analyses. with income under DM 3500, and those with income higher than DM 3500. For the group with income less than DM 3500, the state of health contentment increased continuously as income increased. The mean of contentment increased as follows: 4.5473 (income DM lOOO,-); 4.7927 (1000 < x < 1500); 4.8763 (1500 36) in ARDS patients was 23%, 11.9% in control group one and < 0.5% in control groups two and three (&-square test, p = 0.0000). Patients with PTSD scored signifi cantly lower in all health dimensions of the SF-36 except Social Functioning (Mann-Whitney U-test, p = 0.024).Despite impairments in HRQOL, the majority of patients from this sample of extremely sick ICU patients are able to lead an independent life and show good physical and social functioning.Patients after prolonged exposure to ICU environment and procedures show a high incidence of persistent psychological alterations. Whether this is due to traumatic experiences and awareness during ICU treatment or to factors leading to the disease process itself remains to be determined. The study hypothesis was supported and a significant association demonstrated linking the occurrence of PTSD symptoms to loss of HRQOL. Our findings give evidence of the need to identify and provide therapeutic support for patients with PTSD after prolonged ICU treatment. The primary purpose of this presentation is to overview the public health issues that are emerging for survivors of traumatic brain injury and their families. Particular emphasis will be given to the compromised quality of life issues that often accompany post-injury adjustment. A secondary purpose of this presentation is to discuss an innovative model of rehabilitation for survivors-the 'Clubhouse.' The purpose of a 'Clubhouse' is to provide a centre for socialization which includes educational, recreational and work activities. The philosophy of the 'Clubhouse' is to enhance personal empowerment and self-esteem through individual and community involvement. Discussion will centre on how 'Clubhouses' differ from traditional day treatment and rehabilitation models. Lastly, this presentation will describe four steps necessary to establish a free-standing 'Clubhouse' Programme; these are: community needs assessment, focus groups, pre-programme committee work and fund raising. Limitations with this approach will also be discussed. Research indicates that 9.75 million people in the United States sustain some form of head injury each year; one quarter of these injuries result in long lasting and often permanent alterations in cognitive, emotional and behavioural functioning.Survival rates have increased, in part, due to the recent advances in medical and trauma care thus creating a rehabilitation need unprecedented in public health care systems. Additionally, the psychosocial support needs of people with traumatic brain injury are as great as those for people with chronic mental illness, yet few programmes exist that provide services (e.g., peer support services, recreational activities and open-ended socialization opportunities) for this dual need population. Common problems reported by survivors and their family members are impaired social relations, feelings of estrangement and isolation, and difficulty accepting physical health and lifestyle changes. This underscores the need for adults with head injury to have available appropriate resources that will contribute to their social, emotional and physical quality of life. 'Clubhouse' programmes are one such resource. The purpose of this presentation is to discuss how clinicians can use qualitative and quantitative date to understand what constitutes good quality of life and satisfaction with services from the client/consumer perspective. Crossnational survey data from four countries (e.g., Canada, Cuba, Mexico and the United States) will illustrate these techniques. The benefits and limitations of this evaluation method will be discussed.Increasingly, health and mental health providers are attempting to assess clients/consumers quality of life and satisfaction with services. In doing so, there exists an opportunity for clinicians and clients to work in partnership to address social and contextual issues that goes beyond research-driven indices. As an indices of whether these service areas have been met, clinicians are turning to standardized measurement tools to evaluate their treatment outcomes, namely Quality of Life and Service Satisfaction surveys. Recipients of these services (clients/consumers and their families), however, are seldom consulted about what they want or what is working in their lives. As a result, survey results may provide statistically useful information but may not represent the contextual reality of the clients' lives.In order to illustrate how clinicians and clients can collaborate on assessing the efficacy of the mental health care system, the authors will present date results from a cross-national survey conducted in four different countries between 1992-95. Participants (n = 121) were all outpatients at community based (mental) health care clinics; each had a diagnosis of chronic schizophrenia and had been actively involved in a local health care system. Significant variables which contributed to perceived satisfaction included living situation and finances (p < 0.05) and leisure activities (p < 0.10). Quantitative and qualitative methods were used for data collection.The benefits and obstacles to this sort of assessment procedure, particularly from a cross-national perspective, will be discussed. Recommendations will be offered regarding how to maximize client/consumer participation in data gathering procedures and application of the results. affective responses, to create five &item HRQOL scales (item scores: 1 = no health problem; 2 = health problem, but no negative affective response to it; 3 = health problem with a negative affective response). Psychometric performance was evaluated in three ways. First, to estimate homogeneity, Cronbach's a for the five &item HRQOL-scales were calculated. In addition, Cronbath's a were calculated for the two overall emotional functioning scales. Second, for each item, corrected itemscale correlations were calculated and compared to correlations of each item with all other scales. Finally, a confirmative factor analysis was performed. Cronbach's a for the seven scales were: basic physical functioning (0.70), gross motor functioning (0.80), role functioning (0.85), cognitive functioning (0.88), social functioning (0.76), positive moods (0.88) and negative moods (0.77). The correlation of any item with its own scale was always higher than with any other scale in the questionnaire. In a factor analysis, the fit of a five-factor solution to the 40 HRQOL items was very good. The definition of HRQOL as health status combined with affective responses to problems in health status led to promising results. The psychometric performance of the Dutch HRQOL measure for children was quite satisfactory. As a further step in the development of the measure, norms for the measure, as well as data on convergent validity involving a translated version of the German KINDL (Bullinger et al., 1994) , will be collected in a large sample study (n = 1200) in the general child population. In lack of curative therapy for the HIV-infection in the near future, the development and improvement of palliative care gets all the more important. In this context not only clinical parameters (like T cell counts) are relevant, but especially the patients subjective well-being must be considered. For this purpose the recently emerged concept of multidimensional Quality of Life (QOL) assessment, which has been successfully applied in a number of chronic diseases (e.g. hypertonia, epilepsy etc.) offers the appropriate tools for evaluating the effectiveness of drug therapy. As cognitive and neurological impairment is frequently seen already in the earlier stages of HIV-infection and appears even to be a predictor for disease aggravation, the inclusion of tests for mental competence in the concept of QOL assessment in a broader sense seems necessary for this condition. Early diagnosis of these kind of impairments also offers better perspectives for rehabilitation or disease management. Therefore, a neuropsychological test battery measuring complex cognitive as well as elementary brain functions has been included in the study while health-related QOL in a constricted sense is assessed with generic and diseasespecific questionnaires. The generic instruments used in this study were: MLDL (Munich QOL Dimension List), PGWB (Psychological General Well-Being Index), POMS (Profile of Mood States), VAS-General Health (visual analogue scale for General Health), VAS-QOL (visual analogue scale for QOL); the FKV (Freiburger Fragebogen zur Krankheits verarbeitung) was applied as a predictor for strategies of coping with illness and medication. As a disease-specific questionnaire the recently translated and validated German version of the MOS-HIV (Medical Outcome Study for HIV Infection) had been selected including the HIV-specific symptom list. For the assessment of complex cognitive functioning the following tests were administered: a multiple choice lexical intelligence test, the auditory verbal learning test (AVLT), digit span and digit symbol tests (from the HAWIE-R), the trail making test B, the grooved pegboard and the Stroop test (colour-word interference). Two basic psychophysical paradigms addressing elementary brain functions were used to investigate the hypothesis of underlying declaration of central temporal processing. They contained the measurement of auditory and visual simple and choice reaction times as well as three finger tapping tests measuring maximal tempo, personal tempo and sensorimotor synchronization.One hundred and ninety-two patients with symptomatic HIV-infection and T4 lymphocyte counts in the range of 0-9OO/ul took part in the study and underwent initial assessment. No more than 2 weeks later (to) each of them was assigned randomly to one of two equally sized treatment groups receiving either Peptide-T or placebo during the following 8 weeks (t8) of treatment phase. At t0 and t8 patients' QOL and neuropsychological functions were assessed with the above described means. After final testing patients were given the opportunity to continue or begin Peptide-T treatment (according to clinical demand) if they wished to do so. development of a quality of life measure for men with erection difficulties The objective of this study was to develop a cross-cultural instrument measuring quality of life specific to male erection difficulties, the QOL-MED.We developed the QOL-MED from interviewing forty men with erection difficulties in Seattle and Boston. To determine if the QOL-MED was comprehensive and intelligible, it was administered to twelve men in the USA and twenty men in England.For psychometric testing, the QOL-MED was administered twice over two weeks. In both the USA (n = 40) and the UK (n = 29), participants completed PGWB; USA participants also completed the SF-36. By removing items with decreasingly troublesome psychometric characteristics, three possible measures were identified: an l&item, 14item, and IO-item version. These three versions were compared with respect to internal consistency reliability, validity and reproducibility.All three met acceptable criteria, however, we were unable to identify the most suitable version for clinical trials due to important differences in the USA and UK samples. Our objective was to develop a self-report quality of life measure specific to urinary incontinence (I-QOL) that could be used as an outcome measure in clinical trials and in patient care centres. The I-QOL was developed from interviews of individuals with urinary incontinence (n = 90). Refining the questionnaire was accomplished by structured interviews of individuals with urinary incontinence (n = 17). The rigorous development process ensured that the measure was complete and understandable.Testing the I-QOL's psychometric properties involved two administrations (n = 62) along with measures of psychological well-being and functional status. The I-QOL proved to be internally consistent (a 0.95) and highly reproducible (r = 0.93, 18 days, SD 4). For discriminant validity, severity of incontinence (p < 0.0001) and number of medical appointments in the past year to treat incontinence (p < 0.0001) significantly predicted I-QOL scores using one-way analysis of variance. Convergent validity analyses confirmed our predictions that the I-QOL scores were more closely related to overall well-being than bodily pain The I-QOL proved to be valid and reproducible as a self-administered measure for assessing quality of life of people with urinary incontinence. A. Wahl Department of Public Health and Primary Health Care, University of Bergen Psoriasis is one of the most common chronic skin diseases characterized by distinctive lesions being a silvery gray scaling papule or plaque. Few studies describe quality of life and coping in patients with psoriasis. Quality of life and coping are important concepts in nursing science and nursing practice. Therefore the aim of this study is to analyze the perception of quality of life and coping in patients suffering from psoriasis. So far, 175 patients are included in the study. Patients are recruited from hospitals and outpatient clinics. Quality of life is measured using generic and disease specific instruments: Quality of life scale, SF-36 and Psoriasis Disability Index. Coping is based on the definition of Lazarus and Folkman (1984) and is measured using the Jalowiec Coping Scale. Results from the study will be presented at the conference. The purpose of this study was to determine if quality of life, as operationalized by the Sickness Impact Profile (SIP) would improve in patients treated with three doses of tolcapone when compared to patients treated with placebo. Health-related quality of life (HRQOL) is emerging as a critical outcome measure in health care and more recently in neurology. Traditional outcomes in Parkinson's Disease (I'D) have focused on motoric function. I'D also has an enormous impact on psychosocial well being, role fulfillment, sexuality and self esteem. Psychosocial dysfunction may render a patient as disabled as changes in motor function. This experimental study was done with a subset of patients diagnosed with idiopathic Parkinson's disease (n = 46) who were participating in a double-blind placebo controlled dose finding study of 50mg, 200mg and 400mg of tolcapone. Nineteen females and 27 males were randomly assigned to three doses of tolcapone or placebo. Duration of PD averaged 10 years (range 2-20), mean motor function as determined by the Unified Parkinson's Disease Rating Scale (UPDRS) was 37.5 prior to treatment.Improvement in overall quality of life (repeated measures ANOVA) was significantly improved in patients treated with all doses of tolcapone (p = 0.003). Overall illness impact per cent change improvement in the 400 mg group was 43% compared to the 200mg group (23%) and the 50mg group (20%). Greater per cent change improvement was found in HRQOL outcome measures than in the traditional motor function measure. Quality of life per cent change ranged from 3%-43% compared to motor function change range of 2%-S% in patients treated with tolcapone. In contrast patients treated with placebo experienced a 2% worsening in motor function.The findings of this evaluative study suggest that HRQOL was improved for patients taking tolcapone. HRQOL may be a very promising outcome measure for determining both the impact of Parkinson's disease as well as to determine the efficacy of new treatments over time. Apart from combining into an overall score, specific information on aspects of well-being in terms of anxiety, depression, vitality, self-control, health and general well-being can be derived. The index is psychometrically well-documented, and has been used extensively in clinical trials in hypertension in order to ensure that potential adverse events do not affect the wellbeing. Similarly, it has been used in symptomatic conditions to study the impact of symptom relief on well-being. It has also been advocated as the instrument of choice for the assessment of well-being in clinical trials. (Bulpitt CJ, Fletcher AE. PharmacoEconomics 1994; 6: 523-535. Fletcher A. Br J Clin Phmmac 1995; 39: 217-222. Beth P. PharmacoEconomics 1995; 7: 141-151.) Translations into a number of languages have been performed in accordance with proposed guidelines (Guillemin F, Bombardier C, Beaton D. J Clin Epidemiol 1993; 46: 1417-1432.). Using the baseline values of multinational, multicentre clinical trials in hypertension and gastrointestinal disease, the internal consistency reliability (Cronbach's a coefficient) was calculated for each of the participating countries. Similarly, the baseline values in the two conditions were compared, and finally the well-being scores, distributed by gender were computed. The internal-consistency reliability was generally high for the total score as well as for the different dimensions irrespective of whether a hypertension or a GI population was used to compile the data. For example, using a hypertension study, the internal consistency reliability for the total score of the PGWB was for Australia (n = 115) a = 0.94, Canada (n = 250) a = 0.94, Denmark (n = 210) a = 0.95, Sweden (n = 159) a = 0.95, and Italy (n = 191) a = 0.95. Similarly, a high internal consistency reliability a = 0.93 was found in gastrointestinal patients in the UK (n = 954), in Denmark a = 0.94 (n = 117), and in Sweden a = 0.95 (n = 354) for the total score of the PGWB index. As expected, the mean baseline values of hypertensive patients were compatible with those of a normal population, i.e. around 103-105, whereas gastrointestinal patients scored much lower, from 85-95 depending on whether they suffered from ulcer disease or symptomatic GERD. In terms of gender differences, women scored consistently somewhat and significantly lower than men.The suitability of the PGWB index for use in international multicentre clinical trials was supported bv the fact that despite translations into a number of different languages, the homogeneity and consistency of the scale were supported. Similar to other observations, gender differences were observed. The validity was supported by its ability to discriminate between conditions of varying severity. ESRD-specific domains.In logistic regression analyses assessing the independent association of dialysis modality on the product of frequency and bother, the domains of time, social function and access problems were more frequent and bothersome for HD than I'D, controlling for age, gender, race, comorbidity, education, months on dialysis and living situation.We used a structured literature review, content analysis of patient and provider concerns and patient rankings of the importance of these problems to determine domains to include in a questionnaire for patients on dialysis. The questionnaire will be subjected to a formal pre-test to assess reliability, construct and criterion validity.The final questionnaire, entitled the CHOICE Health Experience Questionnaire (CHEQ), will be used in a large cohort study of incident dialysis patients.living. In 1992, Elizabeth Condie suggested that therapists 'are good at dealing with impairment; are generally less competent at coping with disability; and are really poor at helping individuals to minimize or manage their handicap'. Could we as therapists have a greater impact on some of the disability and handicap that is seen in this clientele? If so, is it our role to reduce handicap or is it an unrealistic expectation in today's restrictive health economy?Development of a quality of life measure for renal dialysis patients The objective of this study was to develop a measure of health-related quality of life for the End-Stage Renal Disease (ESRD) Patient Outcomes Research Team (PORT) that would complement the SF-36 and be sensitive to differences in dialysis modality (hemodialysis, HD, and peritoneal dialysis, PD) and dose (intensity and duration). Selection of domains was based on four sources of data:(1) a structured literature review evaluating 65 articles that described 83 instruments; (2) five focus groups conducted with either HD patients, I'D patients, nephrologists or other dialysis providers (transcripts subjected to content analysis); (3) a survey of 110 dialysis asked about patient concerns and relevant features of different modalities;and (4) an open-ended, semi-structured survey of 25 ESRD patients asked about the effects of dialysis on functioning and quality of life to gain additional breadth of responses. Two additional steps helped to prioritize domains and items. First, a representative sample of 152 dialysis patients rated each item for frequency and bother using Likert-type rating scales. Domains were selected for the pilot questionnaire based upon item ranking of the product of frequency and bother and discordance between HD and I'D patients. Second, we obtained advice from a panel of nephrologists about saliency of items to modality or dose differences. Items and scales representing the domains were selected, with a preference for the use of existing measures which were reliable and valid in previous studies of ESRD patients.These first four steps yielded 21 candidate domains composed of 100 total items: eight domains in the SF-36 (general health perceptions, physical functioning, pain, mental health, energy, physical role functioning, emotional role functioning and social functioning); seven additional generic domains (cognitive functioning, sexual functioning, sleep, work, recreation, finances and quality of life); and six ESRD specific domains including dialysis access (i.e., catheters or vascular access), diet, freedom, time, body image and symptoms. New items were developed or adapted to assess Health status in a double-blind randomized comparison of three oral regimens for pneumocystis carinii pneumonia in AIDS A. W. Wu, S. Gray, R. Brookmeyer and S. Safrin Johns Hopkins University, Baltimore, MD, USA; University of California, San Francisco, CA, USA The objective of this study was to examine patient-reported health status in a randomized, double-blind, multicentre trial of oral trimethoprim-sulfamethoxazole (TS), dapsonetrimethoprim (DT), and clindamycin-primaquine (CP) for treatment of mild-to-moderate Pneumocystis curinii pneumonia (PCP) in patients with AIDS (AIDS Clinical Trials Group Protocol 108). Subjects were enrolled from May 1991-June 1993. Eligible patients had HIV-infection, a suspected diagnosis of PCP and an alveolar-arterial gradient of < 45 mmHg.Patients were also eligible if PCP was not confirmed.Study therapy was administered for 21 days +24 hours. Study endpoints included death, therapeutic failure, toxicity and patient-reported functional status and symptoms, assessed using a self-administered battery of scales. Physical functioning was assessed using the Duke Activity Status Index (DASI), a 12-item index weighted on the basis of known metabolic costs of each activity. Energy, pain and general health perceptions were measured using scales from the MOS-HIV. Disability was measured using two items for days spent in bed and decreased usual activities. Symptom severity was assessed for pulmonary symptoms (cough, dyspnea and chest tightness) and for other symptoms (fever, pain, nausea, rash and dizziness). All scales are scored so that a higher score indicates better health. The questionnaire was available in English and Spanish, and took about 5 minutes to complete. Changes in health status scores were analyzed using Kruskal-Wallis and Wilcoxon rank-sum tests at day 7 and day 21. To provide a summary statistic, a random effects model was