key: cord-0038963-4g906q7y authors: Waltner-Toews, D.; Martin, S.W.; Meek, A.H. title: Dairy calf management, morbidity and mortality in Ontario Holstein herds. III. Association of management with morbidity date: 2002-11-14 journal: Prev Vet Med DOI: 10.1016/0167-5877(86)90019-x sha: e8de5a4dbf41b74dea9968852364d84316fb2f83 doc_id: 38963 cord_uid: 4g906q7y Associations between heifer calf management and morbidity, particularly scours and pneumonia, were studied on 104 randomly selected Holstein dairy farms in southwestern Ontario between October 1980 and July 1983. At the farm level, data were stratified by season, with two six-month seasons (winter and summer) per year. The odds of farms with particular management strategies having above median morbidity were calculated. At the individual calf level, the odds of a calf being treated, controlling for farm of origin and month of birth, were calculated for different management practices. Farm size, and policies related to anti-scour vaccination, offering free-choice water and minerals to calves, methods of feeding, and the use of medicated feeds significantly altered the odds of a farm experiencing above-median pneumonia rates. Farm policies with regard to anti-scour vaccination, offering free-choice salt to calves, age at teat removal, type of calf housing, and use of preventive antimicrobials significantly altered the odds of a farm experiencing above-median scours rates. Scours and pneumonia were significantly associated with each other at both the farm and the calf level. No significant associations were found between individual calf management practices and the odds of being treated for scours. Sire used, method of first colostrum feeding, navel treatment, use of anti-scour vaccine in the dam, and the administration of preventive antimicrobials significantly altered the odds of a calf being treated for pneumonia. The study of calf morbidity under field conditions raises some serious questions fundamental to the practice of both animal husbandry and veterinary medicine. The most serious and controversial of these questions concerns the nature of disease itself. Apart from those animals which are moribund or obviously debilitated, when and how do we judge an animal to be sick? Are there objective criteria by which one can judge this? Some veterinary researchers have attempted to avoid this shady area by evaluating the impact of calf diseases solely on the basis of microbiological laboratory diagnoses (House, 1978) . As Murphy (1976) has pointed out, however, any individual will ultimately be declared to be diseased if a sufficient number of laboratory diagnostic tests are performed. In addition, there is no clearly defined association between laboratory test results and true morbidity. Other researchers, such as Hancock (1983) , have developed their own scales for measuring the looseness of calf stools in order to give what are essentially qualitative judgements a semblance of objectivity. In many ways this begs the central question: do loose stools reflect disease, and, if so, how, or under what conditions? It is one thing to standardize one's observations; it is another matter entirely to say what the standardized observations mean. The most obvious measure of morbidity --farmer diagnosed and treated disease --appears on the face of it to be an unreliable measure of disease. Nevertheless, it can be argued that calf diseases measured in this way reflect the clinical syndromes of interest to dairy farmers, and deserves study if only for that reason. The ability of a farmer to recognize clinical signs, his willingness to treat, and his record-keeping habits all contribute to whether or not calves are recorded as having been treated on a particular farm. Some, but not all, of these difficulties can be overcome by standardizing and monitoring record-keeping practices, and by using large numbers of farms. Also, since these factors are likely to be consistent within farms, farmlevel management effects can be assessed by looking for consonant patterns at the individual animal level on a within-farm basis. For instance, if farms which usually house calves in hutches have, in general, lower rates of treatments for scours, then calves housed in hutches on farms with more than one housing system should be less likely to be treated for scours than their herd-mates. Consistency between farm-level and individual animal-level findings lends strength to a hypothesis of causality (Susser, 1973) . The objectives of this observational study were to assess the impact of dairy calf management policies on rates of disease treatment by farmers, and to assess the impact of individual calf management factors on individual calf morbidity. Analyses were carried out with all diagnoses combined, and two sub-classes of clinical diagnosis --pneumonia and scours --as endpoints. In this study, the terms "crude morbidity" and "all diagnoses" are used interchangeably to refer to a pooled morbidity category, which includes all calves treated therapeutically for whatever reason. Farm-level calf management and outcome data were collected from an initial random sample of 110 Holstein dairy farms in southwestern Ontario, over a period of about two and one-half years. Individual heifer calf data were collected on a subsample of 35 of these farms. The method of selection of these farms and a description of calf management and outcomes were presented earlier (Waltner-Toews et al., 1986a) . All analyses were carried out based on heifer calf data only. For analytical purposes, the neonatal period was defined as the period from birth up to and including 28 days of age. The pre-weaning period covered the entire time from birth until calves were no longer fed milk or milk substitutes on a regular basis. This latter time period varied from farm to farm. Records were kept on all farms on standardized forms. Record-keeping habits were monitored and scored as poor, fair and good by technicians who regularly visited the farms. Farm calf management data were derived from interview questionnaires and farm visits; details of individual calf management were based on specially designed calf birthday forms. Preventive treatments, at both the farm and the individual calf level, were those given to a dam within six weeks before calving, or to the newborn calf in the absence of recognizable disease. At the farm level, morbidity rates were estimated by treatment days per calf during a given 'season'; winter was the period 16 November--15 May and summer was the period 16 May--15 November. The rates were calculated as (number of calf treatment days) -: (number of live-born calves) during each time period. Losses due to death or sale were small and were not accounted for in these analyses. For analytical purposes, farms were classified as being above or below the median treatment days per calf for a given season. On the subset of farms where individual calf information was kept, measuring tapes (Coburn weight-by-breed dairy cow tape, the Coburn Co., Inc., Whitewater, WI 53190, USA) were provided part-way through the study, and farmers were instructed how to take heart-girth measurements. These measurements were to be taken on calves at birth, two weeks of age, and at weaning. Since farms were classified as being above or below the median value for treatment days per calf, the effects of management on all diagnoses combined, and on pneumonia and scours in particular, were assessed using multiple stepwise logistic regression (Breslow and Day, 1980; Dixon, 1983) . The county in which the farm was located (COUNTY), year (YEAR), record-keeping score (REC) and usual weaning age per farm (WEAN) were included in all models. Some variables, such as those concerned with whether or not hutches were moved between calves, or whether or not indoor calf pens were in an air space separate from the adult cows, only applied to particular subsets of farms. These were analyzed by means of two-way tables, stratified by year, with differences tested by chi-square or Fisher's Exact test. At the individual calf level, whether or not calves were treated (all diagnoses), or treated specifically for pneumonia or for scours were analyzed using multiple stepwise logistic regression. Farm, and time of birth (month and year) were included in all models. Initially, variables were allowed to enter the logistic models if they explained a significant portion of the residual variation, based on an estimated F, and a p-value to enter at ~< 0.10. If a variable, once in the model, was found not to contribute significantly to the model, controlling for other variables in the model, it was removed, based on an estimated F, and a pvalue to remove at ~< 0.15. In order to determine whether or not some factors in the model might significantly alter the effects of other factors, also in the model, (for instance, assisted colostrum feeding in calves that required birthing assistance, versus feeding assistance in naturally-born calves), models including interaction terms were constructed. Only biologically plausible interactions of those variables which had entered the main effects models were considered for entry. In these analyses, the p-values for F-to-enter and F-to-remove were set at ~< 0.05 and ~< 0.10, respectively, with all terms (main effects and interactions) being offered for entry. The adequacy of the logistic model for the data was assessed using Brown's statistic, and the fit of the specific model to the data was evaluated based on Lemeshow and Hosmer's statistic (Lemeshow and Hosmer, 1982; Dixon, 1983) . Variables and associations reported in the tables are those included in the interaction models. Factors which apparently influenced age at first treatment, duration of treatment and weight gains were analyzed using multiple linear least squares regression (Draper and Smith, 1981; SAS, 1982; Dixon, 1983) . In the first step, all possible subsets regressions were run. The 'best' model was selected based on the lowest value for Mallow's Cp statistic. This model was then more fully evaluated by examination of residuals and normal probability plots. A dummy variable for each farm was included in all models. Time of birth (season) was included as a binary variable (WINTER). This was coded as one if the calf was born between 16 November and 15 May, and zero otherwise, and was available for entry, but was not forced in. Other variables, such as EASE, were also recoded into the binary form (CALVE-ASSIST) for the multiple linear regression analysis. Variables with coefficients in the 'best' subset model which were not significant at P ~< 0.05 were then excluded, and the model re-run. At the farm level, 30 different management variables (apart from REC, COUNTY, YEAR and WEAN) could be evaluated using the complete set of farms. These variables covered a wide range of aspects of calving and calf management (Table I ). At the individual calf level, 15 management variables (other than farm effect and month and year of birth) were evaluated in about 1800 calves (Table I) . On different subsets of calves, the effects of estimated birthweight (from heart-girth measurements) (n=614), calf housing and age at removal from dam (n=682) could also be evaluated. As reported earlier (Waltner-Toews et al., 1986a) , 20.5% of 1968 heifer calves were treated for scours, and 15.4% were treated for pneumonia on the study farms. Case fatality rates for pneumonia, scours, a combination thereof, or other diseases, did not differ significantly from each other, and ranged from 5.5% (pneumonia) to 7.1% (diseases other than scours and pneumonia). Of 1280 calves which were not seen to be sick, 2.3% died. This was significantly lower (×2=20.3, P ~< 0.0001)than the case fatality rate in the pooled disease categories. The risk of pneumonia in individual calves was significantly increased by a factor (odds ratio) of three if they also had scours during the same time period (Table II) . Treatment for scours did not always precede treatment for pneumonia. However, since the risk of scours treatment peaked earlier in the life of these calves (at about 10 days) than the risk of pneumonia treatment (at about six weeks), scours is presented as if it were an antecedent variable. The median farm treatment days per calf are displayed in Table III . The median number of treatment days per calf for scours was zero for every season. Hence these analyses differentiated between farms with no treatments for scours and those with any treatment for scours. The median number of treatment days per calf for pneumonia, and for all diagnoses combined, varied from season to season. At the farm level, the association between scours and pneumonia for both summer and winter seasons was consistent with what was seen at the individual calf level. Using the method of Mantel and Haenszel (1959) to summarize across years, the winter odds ratio was 2.2 (×2=9.4; P ~< 0.005) Place where calf was born Ease of calving Calves routinely pulled or not Time of birth. Day = 6 AM to 10 PM Was calf's navel treated at birth Time post-partum first colostrum offered How first colostrum offered to calf Antimicrobials given healthy newborn calf Vitamins given healthy newborn calf Selenium given healthy newborn calf Scour vaccination given healthy newborn calf Prey. vit. ADE to dam < 6 wks prepartum Prey. vit. D to dam < 6 wks prepartum Prev. selenium to dam < 6 wks prepartum Scour vaccination to dam < 6 wks prepartum Sire of calf Type of housing for calves to weaning Age in days when calf removed from dam Estimated from heart-girth measurement If teats removed preweaning, at what age If horns removed preweaning, at what age Liquid calf feed (milk, milk replacer, etc) Volume of liquid feed fed to calves Antimicrobials in starter or replacer Age at which pail feeding started Age salt first offered to calves Age mineral first offered to calves Age fresh water first offered to calves Age hay first offered to calves Age starter first offered to calves Type of calf starter (commercial, home-mix, etc.) Proportion of calvings attended No. of calvings resulting in a lactation/year Age to which bull calves reared Person(s) responsible for calf care aAssessed at individual calf level. bAssessed at farm level as a general policy. and the summer odds ratio was 3.0 (×5=12.2; P < 0.0005). That is, farms that had above the median number of treatment days for scours (or, in other words, which experienced at least one episode of calf scours) had a two to three times increase in risk of having above-median treatment days per calf for pneumonia. Also, farms with above the median treatment days per calf for pneumonia were 2.3 times more likely to have above-median mortality rates than farms with below the median treatment days per calf. For scours treatment, the odds ratio was 1.9 for its association ~vith mortality. For both scours and pneumonia treatment days per calf, the association with mortality was strongest during the summer months. Farm-level calf management policies included in the final logistic regression models as predictors of disease in general (all diagnoses), and for pneumonia and scours are displayed in Tables IV to VI. Among those variables which only applied to some of the farms and which were analyzed by the chi-square or Fisher's Exact tests, none was significantly associated with whether or not farms experienced above-median treatment days. Individual calf management factors associated with all diseases, and with pneumonia, are displayed in Tables VII and VIII. In Tables IV to VIII, only those odds ratios whose 95% confidence interval (CI) did not include 1 are displayed. The logistic regression coefficients and their standard errors can be calculated from the adjusted odds ratios and either one of the confidence limits (CL). For instance, the odds ratio and 95% CL for COW-VACCINATION in Table IV are 5.47 and 2.01, 14.85. The logistic regression coefficient is the natural log (in) of 5.47, that is, 1.699. The standard error, calculated from the lower confidence limit, is (1.699-1n(2.01)) -: 1.96 = 0.51. For variables, or levels of variables, which were not significant at P ~< 0.05, but were included in the final model, only the direction of the effect is shown. A positive sign indicates an odds ratio of > 1, and a negative sign indicates an odds ratio of < 1. Lemeshow and Hosmer's statistics for the final models indicated an adequate fit to the data for all the logistic models except that for neonatal pneumonia (Table IX) . No individual calf management factors were significantly associated with whether or not a calf was treated for scours neonatally (~< 28 days) or during the pre-weaning period. Only COW-VACCINATION significantly aOdds ratio of having above-median treatment days (see Table III ), controlling for county, year, record-keeping score and weaning age. Figures are only given for odds ratios significant at P < 0.05. For other variables, or levels of variables, a '+' indicates a mean odds ratio of > 1, and a '-' indicates an odds ratio of < 1. bSour or preserved colostrum. increased the odds of being treated for pneumonia during the neonatal period (odds ratio 2.98; 95% CL 1.11, 3.29). The effect of SIRE for each of the 16 selected bulls was calculated with all other sires as the base-line category. In this study, the SIRE effect was significant only for pre-weaning pneumonia. The overall effect was primarily driven by a few bulls whose calves experienced higher treatment rates for pneumonia than the calves from other bulls. When only those calves were considered for which REMOVAL-AGE and CALF-HOUSING details were available (n=682), CALF-HOUSING appeared as the first significant factor in the logistic regression models for all diagnoses, for pneumonia, and for scours. In each case, this was because calves raised in outdoor hutches were much less likely to be treated than calves raised in inside individual pens, the latter being the baseline category. Mean odds ratios varied from 0.04 for hutches versus individual pens in the pneumonia model to 0.12 for hutches versus individual pens in the scour model. In other words, calves raised in hutches were 25 times less likely to be treated for pneumonia, and 8 times less likely to be treated for scours, than calves raised inside, in individual pens. Calves raised in other forms of housing, bCalves in the suckled category received no assistance or supplementation for first colostrum feeding. including group pens, did not differ significantly in outcome from those raised inside in individual pens. In the crude morbidity (all diagnoses combined) model, calves left with the dam longer tended to be less likely to be treated (P ~< 0.10) than calves removed at one day of age. When only those calves for which a heart-girth measurement at birth was available (n=614) were considered, estimated BIRTHWEIGHT did not significantly affect any of the outcomes. Individual calf management factors included in the best subset model for age of onset and duration of disease are displayed in Tables X and XI. After controlling for the farm effect, only the season of birth added significantly to the model predicting two-week weight gains. Calves born in winter had lower rates of gain (coefficient = 0.285, P = 0.017) than calves born during the summer. Pasture-born calves tended (P ~< 0.08) to have a higher rate of gain than calves born elsewhere. When farm effect was controlled, none of 55 (18,129) aNumbers in parentheses are numerator and denominator degrees of freedom, respectively. All models were significant at P • 0.001. the variables measured added significantly to the weight-gains to weaning model. The models selected were able to explain 20--25% of the variation in age at first treatment, 40-50% in duration of treatment, and 65% in estimated two-week weight gains (Table XII) . On the basis of residual analysis, the multiple regression models were deemed to be adequate for the analyses of these data. The association of above-median treatment days for scours with abovemedian treatment days for pneumonia, at the farm level, suggests that at least some common factors pre 1 for crude morbidity and scours at the farm level, and for scours at the calf level, but negative for crude morbidity and pneumonia at the calf level. None of the effects were close to being statistically significant, so whatever the effect it appears to be minimal. The shorter pneumonia treatments in pasture-born calves is consistent with the better two-week weight gains in the same calves. This, in combination with the calf housing results, indicate that calves born and raised outdoors were generally better off, healthwise, than calves born and raised in confinement. At the farm level, pail feeding of calves subsequent to the colostral period (as opposed to the use of nipple feeders of some sort) was associated with decreased odds of having above-median summer morbidity and winter pneumonia treatment days. It may be relevant that delayed introduction of pail feeding was associated with increased odds of having calves which harboured salmonellae, enteropathogenic E. coli and/or rota-and coronavirus (Waltner-Toews et al., 1986f) . This, in turn, is hypothesized to relate to improper or inadequate disinfection of nipple feeders. However, under experimental conditions, no significant differences have been demonstrated in the disease rates in pail-and nipple-fed calves (Kesler et al., 1956; Wise and LaMaster, 1968; Morrill and Dayton, 1981) . Although isolation of these organisms from calves on these farms was not significantly associated with morbidity (Waltner-Toews et al., 1986f) , the circumstantial evidence is sufficiently compelling to suggest a rigorous study of the use of nipple feeders in young calves under common dairy farm conditions. Farms where calves were fed waste milk (that is, from treated or mastitic cows) or milk replacer had significantly increased odds of having abovemedian summer morbidity (Table IV) . A recent review of feeding waste milk to calves concluded that, with a few exceptions, no increased risk of disease in calves was associated with this practice (Kesler, 1981) . The health effects of feeding milk replacer have been reported to be related to the quality of the replacer used, and the methods of preparation (Roy, 1980a) . The negative effects seen in this study were significant only during the summer. This suggests that perhaps preparation practices of milk replacer and sanitation related to calf feeding were more problematic during the warm summer months. Beneficial health effects of offering salt (winter scours) or fresh water (winter pneumonia) to calves have not been previously reported, but are biologically plausible. For instance, calves in the early stages of a disease would be able to take in salt and/or rehydrate themselves without having to wait for someone to take notice of them at the next feeding period. Recently, it was reported that ad libitum water intake in neonatal calves was associated with significantly increased calf starter intake and weight gains (Kertz et al., 1984) . Although no differences in rates of scours were observed in that study between the two experimental groups of calves (one with ad libitum water available, and one with no water available), the effects on weight gains would certainly be consistent with a beneficial effect on the health of the calves. The detrimental effect of early teat removal on winter rates of calf scours treatments and of offering mineral supplements on pneumonia treatments are not open to any obvious biological interpretation and may represent chance associations, or may be surrogate measures for other factors not measured in this study. The relationship between serum immune globulin deficiency and disease in calves: a farm survey Some observations on serum gammaglobulin concentrations in suckled beef calves Statistical Methods in Cancer Research Outdoor individual portable pens compared with conventional housing for raising dairy calves Applied Regression Analysis Open shed versus conventional housing for dairy calves The effect of different methods of feeding colostrum on calf blood serum immunoglobulin levels Studies on the epidemiology of mortality and diarrheal morbidity in heifer calves in northeastern Ohio dairy herds Teat dip protects calves from navel ill. Hoard's Dairyman Bovine immunoglobulins following induced parturition Economic impact of rotavirus and other neonatal disease agents of animals Comparative studies of the gammaglobulin level in sera of market calves in relation to their health Incidence of disease in colostrum deprived calves Indoor versus outdoor calf rearing at three weaning ages Ad libitum water intake by neonatal calves and its relationship to calf starter intake, weight gain, feces score, and season Feeding mastitic milk to calves: a review Nipple versus pail feeding of milk to Holstein calves A review of goodness of fit statistics for use in the development of logistic regression models Statistical aspects of the analysis of data from retrospective studies of disease Performance of newborn dairy calves in hutch housing Method of feeding and aqcess to fiber source for young calves The effect of management and housing upon growth and health of dairy calves Open shed and portable pens versus conventional housing for young dairy calves The Logic of Medicine Factors affecting susceptibility of calves to disease The Calf SAS User's Guide: Basics, 1982 edn (94 pp.) and SAS User's Guide: Statistics, 1982 edn The absorption of colostrum globulins by newborn calves The serum immune globulin concentrations of newborn dairy heifer calves: a farm survey Studies on dairy calves allowed to suckle their dams at fixed times post partum An epidemiologic study of calf health and performance in Norwegian dairy herds Perinatal ill-health in calves. Commission of the European Communities Colostral immunoglobulin transfer in calves Causal thinking in the health sciences A field trial to evaluate the efficacy of a combined rotavirus-coronavirus/E, coli vaccine in dairy cattle Dairy calf management, morbidity and mortality in Ontario Holstein herds. I. The data. Prey Dairy calf management, morbidity and mortality in Ontario Holstein herds. II. Age and seasonal patterns Dairy calf management, morbidity and mortality in Ontario Holstein herds. IV. Association of management with mortality. Prey Calf-related drug use on Holstein dairy farms in southwestern Ontario Selenium content in the hair of newborn dairy heifer calves and its association with pre-weaning morbidity and mortality An epidemiological study of selected calf pathogens on Holstein dairy farms in southwestern Ontario A retrospective study of respiratory disease in a cohort of bacon pigs Evaluation of three housing systems for raising dairy calves Responses of calves to open-pail and nipple-pail systems of milk feeding This research was supported by the Ontario Ministry of Agriculture and Food through its Provincial Lottery Awards Program. Also, the senior author was a Fellow of the Medical Research Council of Canada while this research was undertaken, and gratefully acknowledges their support.