Unaided general practitioners' clinical diagnosis in evaluation of depressive patients: a pilot study.

BACKGROUND In this study we wished to determine the diagnostic accuracy of unaided general practitioners' (GPs') clinical diagnosis in the evaluation of depression in depressed patients under their care compared with the Beck Depression Inventory II (BDI-II). SUBJECTS AND METHODS From 17,000 patients in 10 GPs' offices as representative sample in the city of Zagreb, 5100 patients from three GPs' offices were selected. The sample consisted of 53 out of 76 depressed patients with a diagnosis of Depressive episode (F32) or Recurrent depressive disorder (F33) classified according to ICD-10 and assessed by review of the GP's standardized medical records. Cross-sectional investigation was performed during February 2008. GPs classified depressed patients as either nondepressed without therapy, nondepressed with therapy or depressed with therapy. Within a two-week period, the unaided GPs' diagnosis was compared with BDI-II performed by psychologists unfamiliar with the GPs' assessment. Based on the GP vs. BDI-II comparison, patients were classified as either positive, false positive, false negative or negative. Sensitivity, specificity, PPV, and NPV associated with physician identification of depression were calculated by standard methods. RESULTS Depressiveness was found by BD-II in the group 'depressed with therapy' (24.39±10.91). ANOVA found a significant difference in BDI-II means between the outcome groups (P<0.001). Scheffe's procedure found a significant difference in BDI-II in patients with therapy (nondepressed vs. depressed) (P<0.001) and nondepressed without therapy vs. depressed with therapy (P<0.001). There were 16 depressed patients, 27 nondepressed, 2 false positive, and 8 false negative. Unaided GPs' clinical diagnosis showed 66% sensitivity, 93%, specificity, 88% PPV, and 77% NPV. CONCLUSION Unaided GPs' clinical diagnosis with 88% PPV outperforms other measures of patient depression and is easier to implement when compared to the psychiatric model of caseness, which is based on screening instruments.