Do GPs' medical records demonstrate a good recognition of depression? A new perspective on case extraction.

BACKGROUND Previous estimates of depression recognition in primary care are low and inconsistent. This may be due to registration artifacts and limited extraction efforts. This study investigated a) whether GPs' medical records demonstrate an accurate recognition of depression and b) which combinations of indications within the record most accurately reflect a diagnosis of depression. METHODS GPs' registrations were compared with a reference standard, the Composite International Diagnostic Interview (CIDI), according to DSM-IV criteria. Six definitions of GPs' recognition of depression were tested using diagnostic codes, medication data, referral data and free text in the medical records. The Youden-index was used to select the optimal definition of recognition. Data were derived from the Netherlands Study of Depression and Anxiety. 816 primary care patients from 33 general practitioners were included in the vicinities of Amsterdam and Leiden, The Netherlands. RESULTS Registration of antidepressant prescriptions was the best single indicator of GPs' recognition of CIDI depression with a recognition rate of 0.43. The best combination of indicators increased the recognition rate to 0.69. All indications except the specific diagnostic codes for 'depressive disorder' and 'depressive feelings' were included in this definition. LIMITATIONS Potential bias due to the selection of participating GPs might have influenced our recognition rates. CONCLUSION GPs are aware of mental health problems in most depressed patients, but labeling with specific diagnostic codes is weak. Researchers should consider that diagnostic coding alone is not an accurate measure of the diagnostic ability of depression and strongly underestimates the accuracy of the GP.