key: cord-0765330-n6pdfepr
authors: Jones, Rupert; Davis, Andrew; Stanley, Brooklyn; Julious, Steven; Ryan, Dermot; Jackson, David J; Halpin, David M G; Hickman, Katherine; Pinnock, Hilary; Quint, Jennifer K; Khunti, Kamlesh; Heaney, Liam G; Oliver, Phillip; Siddiqui, Salman; Pavord, Ian; Jones, David H M; Hyland, Michael; Ritchie, Lewis; Young, Pam; Megaw, Tony; Davis, Steve; Walker, Samantha; Holgate, Stephen; Beecroft, Sue; Kemppinen, Anu; Appiagyei, Francis; Roberts, Emma-Jane; Preston, Megan; Hardjojo, Antony; Carter, Victoria; van Melle, Marije; Price, David
title: Risk Predictors and Symptom Features of Long COVID Within a Broad Primary Care Patient Population Including Both Tested and Untested Patients
date: 2021-08-11
journal: Pragmat Obs Res
DOI: 10.2147/por.s316186
sha: 372d164be782b09d90f979e045649a9fd48ab2de
doc_id: 765330
cord_uid: n6pdfepr

INTRODUCTION: Symptoms may persist after the initial phases of COVID-19 infection, a phenomenon termed long COVID. Current knowledge on long COVID has been mostly derived from test-confirmed and hospitalized COVID-19 patients. Data are required on the burden and predictors of long COVID in a broader patient group, which includes both tested and untested COVID-19 patients in primary care. METHODS: This is an observational study using data from Platform C19, a quality improvement program-derived research database linking primary care electronic health record data (EHR) with patient-reported questionnaire information. Participating general practices invited consenting patients aged 18–85 to complete an online questionnaire since 7th August 2020. COVID-19 self-diagnosis, clinician-diagnosis, testing, and the presence and duration of symptoms were assessed via the questionnaire. Patients were considered present with long COVID if they reported symptoms lasting ≥4 weeks. EHR and questionnaire data up till 22nd January 2021 were extracted for analysis. Multivariable regression analyses were conducted comparing demographics, clinical characteristics, and presence of symptoms between patients with long COVID and patients with shorter symptom duration. RESULTS: Long COVID was present in 310/3151 (9.8%) patients with self-diagnosed, clinician-diagnosed, or test-confirmed COVID-19. Only 106/310 (34.2%) long COVID patients had test-confirmed COVID-19. Risk predictors of long COVID were age ≥40 years (adjusted Odds Ratio [AdjOR]=1.49 [1.05–2.17]), female sex (adjOR=1.37 [1.02–1.85]), frailty (adjOR=2.39 [1.29–4.27]), visit to A&E (adjOR=4.28 [2.31–7.78]), and hospital admission for COVID-19 symptoms (adjOR=3.22 [1.77–5.79]). Aches and pain (adjOR=1.70 [1.21–2.39]), appetite loss (adjOR=3.15 [1.78–5.92]), confusion and disorientation (adjOR=2.17 [1.57–2.99]), diarrhea (adjOR=1.4 [1.03–1.89]), and persistent dry cough (adjOR=2.77 [1.94–3.98]) were symptom features statistically more common in long COVID. CONCLUSION: This study reports the factors and symptom features predicting long COVID in a broad primary care population, including both test-confirmed and the previously missed group of COVID-19 patients.

There have been multiple reports of symptoms of COVID-19 infection persisting long after the initial phases of the infection, a phenomenon termed long COVID or postacute COVID-19. 1,2 There is no single agreed definition for long COVID. The UK National Institute of Health and Care Excellence (NICE) 2,3 and the Wellcome Trust foundation 4 defined long COVID as symptoms lasting 4 weeks or more since onset, while other definitions include a duration cut-off of 12 weeks. 5 Long COVID appears to be common, persistent, and debilitating, causing a wide range of physiological and cognitive disabilities irrespective of patient age and initial disease severity. 1, [6] [7] [8] [9] The impact on social well-being 10 highlights the importance of addressing the persisting symptoms of COVID-19 infection. The common presentation of persisting symptoms includes fatigue, breathlessness, chest pain, and body aches. 9, 11, 12 While comparison is difficult due to the lack of a uniform definition, especially in the early days of the pandemic, a review by Nalbandian et al suggested up to 87.4% of COVID-19 patients may remain symptomatic months since the initial symptom presentation. 1 Symptom of COVID-19 infection has also been reported to persist one year post-hospital discharge. 13 Reports of prevalence from a more general patient population suggest a much lower prevalence. The UK COVID-19 Infection Survey produced experimental results showing a national estimate of one in five patients who were tested positive for COVID-19 were symptomatic 5 weeks into the infection, with a further one in ten patients' symptoms lasted beyond 12 weeks as of 16th December 2020. 14 In the US, telephone interviews of RT-PCR-confirmed symptomatic outpatient adults reported 35% having not recovered after 2-3 weeks, and only 1 in 5 young healthy adults. 11 A better understanding of the symptom characteristics and the risk predictors for long COVID is needed to identify patients at risk of developing the condition. However, current data on persisting symptoms of COVID-19 has been mostly derived from cases identified via positive test results. [7] [8] [9] [10] [11] [14] [15] [16] While the exact numbers are not known, estimates suggested that the majority of COVID-19 cases went undetected, 17, 18 likely due to limited testing capacity in the early phase of the pandemic. Furthermore, many studies on the long-term symptoms of COVID-19 have focused primarily on hospitalized patients. [7] [8] [9] 12, [19] [20] [21] [22] [23] The full burden of long COVID and factors which predict the risk of developing long COVID have not been clearly defined in the primary care population including the untested patients who may have a milder initial disease but may still be at risk of developing long COVID. 24 Supplementing electronic health data (EHR) data with patient-reported information can provide additional insight on these previously missed patients with symptoms of COVID-19 managed in primary care.

In response to the pandemic, Optimum Patient Care (OPC) UK established the COVID-19 Quality Improvement (QI) program to help practices identify and manage patients affected by COVID-19 and related problems including preexisting chronic diseases. (https://optimumpatientcare.org/ covid-qi/). As part of the program, participating general practitioners invite their patients to complete an online questionnaire covering questions on demographics, co-morbid conditions, COVID-19 status, symptoms, and testing. Platform C19 systematically assimilates extensive EHR data with patient-reported outcomes (PROs) through questionnaires in the primary care population, creating a unique platform that enables new analyses and insights on the pandemic.

This study aims to investigate the prevalence, symptom features, and demographical and clinical factors which predict the development of long COVID within a broader primary care population within Platform C19 which 

This is an observational study using EHR data and patientreported information data stored in Platform C19. EHR data is derived from the OPCRD, a de-identified primary care database holding records for more than 12 million patients from over 800 GP practices across the UK. The OPCRD integrates with all UK clinical systems (EMIS, TPP SystmOne, InPS Vision, Microtest Evolution) and holds a long retrospective period of data extending from the conception of summary diagnostic data collection (median time in the database of 13 years). The OPCRD data is further enriched with routine EHR data linked with patient-reported information from GP practices participating in OPC QI programs. The questionnaire was designed via a consensus of experts from the Platform C19 steering committee.

Patients aged 18-85 years at the start of the pandemic (1st March 2020) from practices participating in the COVID-19 QI program were invited by text message from the GP to complete an online COVID-19 questionnaire as part of the QI starting from 7th August 2020. Patients that have opted out from data-sharing for research 25 or from receiving text messages were excluded.

De-identified EHR and questionnaire data collected from 7th August 2020 up till 22nd January 2021 from all patients who responded were included. To prepare data for analysis, questionnaire responses and relevant electronic medical records were collated, cleaned, and summarized. If patients had multiple questionnaires on record only the most recent or most complete one was used. Data is stored in a secure, enterprise database running SQL-Server 

COVID-19 self-diagnosis, clinical diagnosis, and testing were ascertained via the questionnaire. Patients were asked whether they believed they have had a COVID-19 infection, whether they had been diagnosed with COVID-19 by a healthcare professional, and whether they had been tested for COVID-19 and the result of the test. Patients with COVID-19 status were split into three categories: self-diagnosed but not diagnosed by a clinician or test-confirmed, diagnosed by a clinician but not test-confirmed, and test-confirmed positive.

The demographic variables extracted from the EHR include age, sex, and smoking status. Data on ethnicity and BMI were supplemented by the questionnaire. For analysis, age was split into an 18-40 year group and an age 40+ group. BMI was split into below 30 kg/m 2 and greater than or equal to 30 kg/m 2 ; the threshold for obesity.

Smokers were classified as non-smokers, ex-smokers, current smokers, or having an unknown smoking status.

The presence of underlying chronic co-morbid diseases was extracted from both EHR and questionnaire. Diseases investigated were asthma, COPD, diabetes, heart disease or heart failure, kidney disease, and any combination of these diseases. Frailty was defined as either recorded frailty in the EHR or responding to be frail ("you have medical problems that limit how active you are, and you need help with daily activities and personal care") in the question assessing the level of fitness. Patients were also asked if they visited the accident and emergency department, if they were admitted to a hospital, and if they were admitted to intensive or critical care for their COVID-19 infection or symptoms. Presence, start date, and end date of symptoms typical of COVID-19 since January 2020 were asked in the questionnaire. The list of questions within the questionnaire used for analysis and the list of COVID-19 symptoms are available in the Supplementary Material.

The primary outcome of this study is the presence of long COVID based on COVID-19 status and symptom duration based on symptom start and end date as reported by the patients. Patients were considered to have had long COVID if they were self-diagnosed, clinician-diagnosed, or test-confirmed for COVID-19 and have symptoms of COVID which lasted for more than 4 weeks based on the NICE guideline definition for long COVID. 3

All analyses were conducted using Rv4.0.3 (R Core team, 2020). Univariable logistic regression was conducted to summarize each demographic and clinical variable individually. Multivariable logistic regression models were created using all demographic variables, hospital visits for COVID-19, frailty, chronic co-morbid conditions, and COVID-19 status as predictors and long COVID as the response variable in comparison with COVID-19 with <4 week symptom duration. Two separate regression models were created, one with each underlying co-morbid chronic disease entered as separate variables (split condition model) and another with the presence of any underlying co-morbid chronic disease combined as a single variable (combined condition model).

Univariable and multivariable logistic regression analyses to compare the symptom features of long COVID and COVID-19 with <4 week symptom duration were also conducted. The multivariable regression model for symptom comparison included every symptom adjusted for factors that significantly predict long COVID from the analysis above. Odds ratios were estimated along with 95% confidence intervals to evaluate variable influence. There is no allowance for multiplicity. Sensitivity analyses were also conducted on long COVID using 12 week symptom duration compared to COVID-19 with symptoms lasting <12 weeks.

The patient flowchart is presented in Figure 1 

The presence of symptoms in patients with long COVID and COVID-19 with shorter symptom duration is presented in Figure 2 . Almost every (95.2%) patients with long COVID reported a loss of appetite. Other commonly reported symptoms among long COVID patients were persistent dry cough (80.3%), chest pain (79.7%), and fatigue or tiredness (72.9%). The symptom features of long COVID using the 12 week cut-off definition showed a similar pattern (Supplementary Figure 1) . (Table 3) . Surprisingly, shortness of breath was statistically less common (adjusted OR = 0.68 [0.50-0.91]).

The results using the 12 weeks cut-off were similar except "fatigue and tiredness" was also statistically less common in long COVID patients (adjusted OR = 0.66 [0.46-0.96]) (Table 3 ).

In this study, we identified long COVID, defined as COVID-19 with symptoms lasting at least 4 weeks, in nearly one-tenth of COVID-19 cases. This is much lower than the rates reported in other studies which have been reported to be more than half the patient population. 13, 22 However, this could be due to the difference in the method of symptom assessment and patient population. The prevalence observed in this study is closer, albeit still lower, compared to more similar general populations from the UK COVID-19 infection survey and the proportion reported in an observational study based on users of the COVID Symptom Study mobile application in the UK (13.3%). 14, 16 As the patients in the current study were not limited to test confirmed COVID-19 cases, our results may be reflective of the prevalence of long COVID in a more general COVID-19 population. Indeed, only slightly above one-third of long COVID cases identified in this study were test-confirmed, highlighting the magnitude of the important yet previously unquantified population of long COVID patients.

Patients who were diagnosed by a clinician but never testconfirmed were more likely to have long COVID than those who were test-confirmed positive. We speculate that this may be due to patients with milder symptoms only received examination and testing after their symptoms have persisted. Patients who did not receive test confirmation may also make up the group of patients from the initial phases of the pandemic when testing and awareness of long COVID were limited. In contrast, significantly fewer self-diagnosed patients who believed themselves to be infected, without a clinician diagnosis or test confirmation, had long COVID. This may be as patients whose COVID-19 symptoms resolved rapidly did not feel the need to be checked by a clinician or tested. Alternatively, a proportion of these patients may not have had COVID-19 infection and this is difficult to assess. Older age and female sex were independently associated with long COVID. Sexual differences in COVID-19 related symptomatology and mortality had been widely recognized. While male sex had been reported to be associated with a higher risk of COVID-19 hospitalization and mortality, 26 studies have reported female sex to be at a higher risk for prolonged COVID-19 symptoms. 7, 8, 16, 22, 23 Biological differences could contribute to the difference in symptom persisting between sex, though gender differences in symptom reporting behavior may have also played a part. 27 Frailty significantly predicted long COVID even after adjustment with age and chronic diseases. Frailty was previously reported in a multicenter cohort of hospitalized elderly patients to be a strong predictor for poorer outcomes of COVID-19 infection. 28 The National Institute for Health and Care Excellence (NICE) of the UK advised frailty to be included as part of the assessment for patients admitted into intensive care during the pandemic. 29 Awareness of frailty as a risk factor for long COVID may be useful for clinicians when patients present with new-onset prolonged diverse symptoms as reported in the present study.

Patients who visited an emergency department and who were admitted to the hospital for COVID-19 symptoms or infections were statistically more likely to have long COVID. Similar to our observation on COVID-19 status, this may be due to the increasing likeliness of patients visiting the hospital as their symptoms persisted or due to the higher severity of their illness. Admission to intensive or critical care was no longer significant following multivariable adjustment, and this may have been due to the small numbers.

Multiple symptoms across different organ systems were common among our patients, consistent with known information on COVID-19 symptomatology. 1 Loss of appetite, persistent chest pain, persistent dry cough, and fatigue or tiredness were the most commonly reported symptoms. The range of symptoms observed in this study is similar to that observed by previous studies which commonly list symptoms of fatigue, body aches, breathlessness, and cough. 9, 11, 12 However, this study further compares patients with long COVID to patients with shorter symptom duration to investigate the symptoms that can potentially predict if a patient will develop long COVID.

Individually, all symptoms were significantly more commonly reported by long COVID patients than patients with shorter symptom duration. This is similar to the observation in the study using the COVID Symptom Study application. 16 Following multivariable adjustment of the symptoms with each other, several symptoms were no longer independently associated with long COVID. Additionally, shortness of breath and fatigue or tiredness were found to have an inverse association with long COVID using the 12 weeks cutoff definition. These results may indicate that these symptoms were pervasive enough in the acute phase of the infection that they were not strong indicators of COVID-19 infection which will persist into long COVID. Symptoms of aches and pain, appetite loss, confusion and disorientation, diarrhea, and persistent dry cough were the symptom features that predicted long COVID within our patient population. Diarrhea, but not fever or dyspnea, has also been previously reported to be associated with persistent symptoms among hospitalized patients in India. 30 

To our knowledge, this study is the first to identify and report on the scale of long COVID as well as the predicting factors and symptom features of long COVID in the primary care population and also including those who had not received a diagnosis or test confirmation for their symptoms. Due to the limited testing during the initial wave of the pandemic, most COVID-19 cases would have gone undetected and unrecorded in EHRs. This study, therefore, provides a more comprehensive picture of long COVID, which includes the previously undetected group of long COVID patients. A previous investigation on persisting symptoms of COVID-19, which included untested patients, was conducted by Goertz et al. 24 However, in contrast to our study, participants in the study were recruited from online groups of patients with persisting symptoms. The current study further supplements our knowledge by including a broader population of primary care patients utilizing Platform C19's unique linkage to a quality improvement program and access to EHR data supplemented with patient-provided information. This provides the unique capability to answer research questions that are not possible with EHR data or questionnaires alone. The utility of patient-reported outcome measures had been previously utilized to demonstrate the social and physical impact of persisting COVID-19 symptoms. 10

Similar to other questionnaire studies, this study is prone to non-response bias. Patients who respond may not be fully representative of the general population. Due to the electronic nature of the questionnaire, it is possible that the elderly and frail were less likely to respond. The Black, Asian, and ethnic minorities are also currently underrepresented among our respondents.

The retrospective nature of this study may have introduced a recall bias. Patients whose symptoms lasted longer may perceive themselves as having more symptoms. Similarly, patients with more symptoms may have reported their symptoms to have lasted longer. This might have inflated the differences in symptom features between long COVID and COVID-19 of shorter duration. Lastly, symptom duration depends on patient recall as it is impossible to retrospectively confirm the length of COVID-19 related symptoms. We are also unable to ascertain whether the reported symptoms were present only during the acute phases of the infection or if they persisted as long-term symptoms.

This study provides novel insight on the burden and predictors of long COVID in a broad primary care population, which includes the population of previously undetected COVID-19 patients who were not clinically diagnosed or test-confirmed. These insights will be valuable for the identification of care pathways and interventions to support this patient group.

Long COVID was observed in almost 10% of COVID-19 cases and was mostly from the group of patients who did not receive test-confirmed COVID-19. Compared to test-confirmed COVID-19 patients, patients who were diagnosed by a clinician but not test-confirmed were significantly more likely to have long COVID. In contrast, patients who were self-diagnosed without clinician diagnosis nor test confirmation were significantly less likely. Older age, female sex, presence of frailty, visit to an A&E department, and hospital admission for COVID-19 symptoms or infection significantly predicted long COVID. Symptoms of aches and pain, appetite loss, confusion and disorientation, diarrhea, and persistent dry cough were indicative of cases that persisted into long COVID.

Further analyses using data stored within Platform C19 will look at the correlation between the risk of developing long COVID with the number of symptoms as a metric for disease severity, patient behavior, and psychological comorbidities. The multi-organ system symptomatology of COVID-19 is similar to that observed in functional disorders such as irritable bowel syndrome, fibromyalgia, and chronic fatigue syndrome. 31 The potential co-morbid link between long COVID and functional disorders is a potential area for future analysis using Platform C19.

Data collection under the OPC COVID-19 QI program is ongoing. Future studies will investigate the long-term clinical outcome of long COVID and its effect on the well-being of patients. With the initiation of the vaccination program in the UK, questions on vaccination status will be included in future versions of the questionnaire. Future research using data stored in Platform C19 will analyze the protective effects of vaccinations on the development of long COVID.

Access to the de-identified patient data used in this study may be requested via the OPCRD website (https://opcrd. co.uk/platform-c19-new-2/) or via the inquiries email info@ opcrd.co.uk.

The OPCRD has received NHS Research Ethics Committee (REC) approval to provide anonymized data for scientific and medical research since 2010, with its most recent approval in 2015 (NHS HRA REC ref: 15 

prior project grant funding from MedImmune, Novartis UK, Roche/ Genentech, and GlaxoSmithKline; has taken part in Advisory Boards/Lectures supported by Novartis, Roche/Evelo Biosciences, Genentech, GlaxoSmithKline, Teva, Theravance and Vectura; has travel funding support to international respiratory meetings (AstraZeneca, Chiesi, Novartis, Boehringer Ingelheim, Teva, and GlaxoSmithKline) and has taken part in asthma clinical trials (GlaxoSmithKline, Schering Plough, Synairgen, Novartis, and Roche/Genentech) for which his institution was remunerated. Salman Siddiqui reports advisory board/advisory services and speaker fees from AstraZeneca, GlaxoSmithKline, Chiesi, Boehringer Ingelheim, Novartis, Mundipharma, ERT medical

Ian Pavord reports grants from NIHR and personal fees from

Marije van Melle and Victoria Carter are employees of Optimum Patient Care who funded this study. David Price has board membership with AstraZeneca, Boehringer Ingelheim, Chiesi, Mylan, Novartis, Regeneron Pharmaceuticals, Sanofi Genzyme, Thermofisher; consultancy agreements with Airway Vista Secretariat

grants and unrestricted funding for investigator-initiated studies (conducted through Observational and Pragmatic Research Institute Pte Ltd) from AstraZeneca

National institute for health and care excellence: clinical guidelines. In: COVID-19 Rapid Guideline: Managing the Long-Term Effects of COVID-19. London: National Institute for Health and Care Excellence

Why the patient-made term 'long covid' is needed

Management of post-acute covid-19 in primary care

Characterizing long COVID in an international cohort: 7 months of symptoms and their impact

Post-COVID-19 symptom burden: what is long-COVID and how should we manage it?

Persistent fatigue following SARS-CoV-2 infection is common and independent of severity of initial infection

Gemelli against C-P-ACSG. Persistent symptoms in patients after acute COVID-19

PROMIS scales for assessment of persistent post-COVID Symptoms: a cross sectional study

Symptom duration and risk factors for delayed return to usual health among outpatients with COVID-19 in a multistate health care systems network -United States

Postdischarge symptoms and rehabilitation needs in survivors of COVID-19 infection: a cross-sectional evaluation

Long-term outcomes of patients with coronavirus disease 2019 at one year after hospital discharge

The prevalence of long COVID symptoms and COVID-19 complications [Website]. Office for national statistics

Patients with uncomplicated COVID-19 have long-term persistent symptoms and functional impairment similar to patients with severe COVID-19: a cautionary tale during a global pandemic

Attributes and predictors of long COVID

Estimating the undetected infections in the Covid-19 outbreak by harnessing capture-recapture methods

Estimating the size of undetected cases of the COVID-19 outbreak in Europe: an upper bound estimator

Patient outcomes after hospitalisation with COVID-19 and implications for follow-up; results from a prospective UK cohort

Prevalence and correlates of chronic fatigue syndrome and post-traumatic stress disorder after the outbreak of the COVID-19

Sixty-day outcomes among patients hospitalized with COVID-19

6-month consequences of COVID-19 in patients discharged from hospital: a cohort study

Physical, cognitive and mental health impacts of COVID-19 following hospitalisationa multi-centre prospective cohort study

Persistent symptoms 3 months after a SARS-CoV-2 infection: the post-COVID-19 syndrome?

Male sex identified by global COVID-19 meta-analysis as a risk factor for death and ITU admission

Gender differences of symptom reporting and medical health care utilization in the German population

The effect of frailty on survival in patients with COVID-19 (COPE): a multicentre, European, observational cohort study

COVID-19 rapid guideline: critical care in adults. National institute for clinical excellence

Clinical predictors of long COVID-19 and phenotypes of mild COVID-19 at a tertiary care centre in India

Symptom frequency and development of a generic functional disorder symptom scale suitable for use in studies of patients with irritable bowel syndrome, fibromyalgia syndrome or chronic fatigue syndrome

Pragmatic and Observational Research Dovepress Publish your work in this journal Pragmatic and Observational Research is an international, peerreviewed, open access journal that publishes data from studies designed to reflect more closely medical interventions in realworld clinical practice compared with classical randomized controlled trials (RCTs). The manuscript management system is completely online and includes a very quick and fair peer-review system

Other members of the Platform C19 steering committee consist of Sir Prof Lewis Ritchie, OBE from the University of Aberdeen, Steve Davis 

All authors contributed to data analysis, drafting or revising the article, gave final approval of the version to be published, agreed to the submitted journal, and agree to be accountable for all aspects of the work. The overall conduct of this study was supervised by David Price.

The design, conduct, and writing of this study are funded by Optimum Patient Care UK and the Observational and Pragmatic Research Institute Singapore.

Rupert Jones reports grants, personal fees, and nonfinancial support from AstraZeneca and OPRI, personal fees and non-financial support from Boehringer Ingelheim, grants, personal fees, and non-financial support from GSK, grants and non-financial support from Novartis, nonfinancial support from Nutricia, and personal fees from Pfizer outside the submitted work. Dermot Ryan has (in the last 3 years) lectured on behalf of, received sponsorship from, or acted as a paid advisor to Mylan, AZ, Chiesi, Novartis, GSK, Boehringer Ingelheim and Regeneron.David J Jackson has received advisory board and speaker fees from AstraZeneca, GSK, BI, Teva, Napp, Chiesi, Novartis and research grant funding from AstraZeneca.David MG Halpin has received sponsorship to attend international meetings, and honoraria for lecturing, attending advisory boards and preparing educational materials from AstraZeneca, Boehringer Ingelheim, Chiesi, GSK, Novartis and Pfizer.Jennifer K Quint reports grants from MRC, grants from The Health Foundation, grants and personal fees from AZ, grants from Bayer, grants and personal fees from Chiesi, grants and personal fees from GSK, grants and personal fees from BI, outside the submitted work. Michael Hyland reports personal fees from GSK, outside the submitted work.Pam Young and Tony Megaw are employees of Wellbeing Software.The remaining authors report no conflicts of interest.