OP83 Defining depression in health datasets: a comparison of depression definitions using data from the UK Biobank

Background Epidemiological studies use a wide range of depression definitions. The aim of this study was to compare some of the most common definitions of depression in a large prospective cohort study with a wide range of information collected at baseline and during follow-up. Methods We used data from the UK Biobank, a cohort study of approximately half a million middle-aged adults, to compare different definitions of depression. We defined prevalent depression based on (1) primary care records (using Read codes); (2) hospital records (using ICD-10 codes); (3) participant self-report of seeing a general practitioner (GP) for nerves, anxiety, tension or depression; (4) participant self-report of seeing a psychiatrist for nerves, anxiety, tension or depression and (5) a published definition of ‘probable major depression’ that combined participant responses to several questions. We defined incident depression based on (1) primary care records; (2) hospital records; and (3) the Composite International Diagnostic Interview – Short Form (CIDI-SF) definition of lifetime major depression. We compared prevalence of depression using different definitions of prevalent depression and incidence of depression using different definitions of incident depression among participants with data on all relevant prevalent and incident depression measures. Results Among the 61,666 participants with data on all prevalent depression measures, lifetime prevalence of depression at baseline was 19% based on primary care records, 1% based on hospital records, 34% based on participant self-report of seeing a GP, 11% based on participant self-report of seeing a psychiatrist and 19% based on the probable major depression definition. Among participants with a primary care record of depression at baseline, 3% also had a hospital record of depression, 85% self-reported seeing a GP, 34% self-reported seeing a psychiatrist and 50% had probable major depression. Among 45,708 participants with data on all incident depression measures (mean follow-up of 7.4 years), incidence of depression was 3% based on primary care records, <1% based on hospital records and 9% based on the CIDI-SF. Among participants with incident depression based on primary care records, 7% also had incident depression based on hospital records and 30% also had incident depression based on the CIDI-SF. Conclusion Different definitions of depression gave very different estimates of its prevalence and incidence in a cohort of UK adults. Researchers must be aware of these differences when designing, conducting and critically evaluating studies of depression.