Background. The next generation of studies on antidepressant drug prescriptions in general practice needs to assess both the patterns of prescription and its appropriateness. This study aimed to assess the performance of the Personal Health Questionnaire (PHQ), a new questionnaire for detecting individuals with ICD-10 depressive disorders, to be used in association with companion instruments for assessing the ‘quality’ of antidepressant prescriptions in primary care settings.

Methods. The PHQ was completed by 1413 primary care attenders (100 were re-tested after 7–14 days) and 139 were selected and interviewed using the SCAN-2 and the 17-item HDRS. All data were analysed using appropriately weighted procedures to control for two-phase sampling design and non-response bias. Individual weights were estimated by logistic regression analysis and trimming strategy.

Results. PHQ internal consistency and test–retest on both Likert score and number of symptoms were high. The PHQ discriminated well between individuals with and without depressive disorders. A Likert score [ges ] 9 provided a good trade-off between sensitivity (0·78) and specificity (0·83). The screening accuracy of the PHQ in detecting subjects likely to benefit from antidepressant drug treatment (SCAN cases with a HDRS total score of 13 or higher) was satisfactory (ROC area 0·87; sensitivity 0·84; specificity 0·78).

Conclusions. The PHQ can be strongly suggested as an accurate and economic screener to identify primary care attenders at high risk of being clinically depressed. However, in order to identify patients requiring antidepressant drug treatment, a second-phase assessment of PHQ high scorers (total score of [ges ] 10), using the HDRS, is needed.