Depressive disorders are a major health problem in primary care, and at least half of these disorders remain undetected. 1 There are two recommended approaches to diagnosing depression in primary care: one is to perform routine screening, and the other is to evaluate patients only when the clinical presentation triggers the suspicion of depression. Our aim was to compare these two approaches, and to compare three different screening tools in order to evaluate which would be most appropriate for use in primary care. From among the many available screening tools, we selected three brief, self rating instruments: one disorder-specific (the depression module of the brief patient health questionnaire (B-PHQ, 9 items)), 2 one broad based (the general health questionnaire (GHQ-12, 12 items)), 3 and one that is less restricted to both issues (WHO-5 wellbeing index (WHO-5, 5 items)).
4
Methods and resultsEighteen primary care facilities participated in our prospective cohort study. The study protocol was approved by our local ethics committee. On one given day, all patients who presented in one of the practices were asked to complete the three screening questionnaires before seeing a doctor. The doctors who treated the patients remained blind to the questionnaire results until they had completed a brief "physician's encounter form" to indicate their clinical assessment of their patient's current diagnoses.Within a period not exceeding six days after they had completed the questionnaires, the patients were contacted by telephone for a fully structured, standardised psychiatric interview (composite international diagnostic interview (CIDI)) conducted by a trained psychologist blind to the screening results. We chose the composite international diagnostic interview as the reference standard because its reliability and validity have been established. 5 The interviewing psychologists met a high standard of inter-rater reliability.The main outcome measures were, firstly, the family doctors' performance in detecting depression without any tool to help guide diagnosis decisions and, secondly, the test accuracy of the screening questionnaires. We calculated sensitivity, specificity, and predictive values using two-by-two tables. We used two statistical tests to compare differences of characteristics of test accuracy (table). For 431 patients, all screening questionnaires, the composite international diagnostic interview, and the physician's encounter form were completed. Of these patients, 17% suffered from any depressive disorder and 83% did not.
CommentThe sensitivity of the family doctors' unaided clinical diagnoses was 65%. With standard cut-off points, the briefest screening questionnaire (and therefore the most practical to use), the WHO-5, produced significantly greater sensitivity (93%) and a better negative predictive value (98%) than the other questionnaires (see table). However, the brief patient health questionnaire and unaided clinical diagnosis produced better specificity. The brief patient health questionnaire al...