A meta-analysis of primary care studies has concluded that overdiagnosis of depression is more of a problem than missed cases.

 

Forty one studies met the inclusion criteria, giving a total pooled sample of 50,371 patients. Participating GPs did not receive any assistance from severity measures when making a diagnosis. Diagnostic accuracy was assessed by comparison with structured or semi-structured interviews. In all, 22 studies assessed diagnostic sensitivity only and 19 assessed both sensitivity and specificity.

 

GPs correctly identified just under half (47.3%) of the patients with depression. For the 19 studies which also measured ability to rule out depression correctly, the weighted sensitivity was 50.1% and the specificity was 81.3%. The prevalence was 21.9%, giving a positive predictive value (PPV) of 42.0% and a negative predictive value (NPV) of 85.8%. 

 

The 12 studies that evaluated GP performance by retrospective case-note analysis found a lower sensitivity (33.6%) than the 29 prospective studies (52.6%).

 

Studies have found that between 4% and 10% of consecutive primary care patients have major depression, and that a similar percentage have subthreshold depression (ICD-10 mild depressive episode).¹ The recent PREDICT study from six European countries reported that 12.15% of general practice attendees had major depression. The UK figure was 13.1%, but this may be inaccurate because of the low participation rate (44%).²

The present study suggests that, if the prevalence is 20%, out of every 100 consecutive patients, there will be 10 patients with unrecognised depression and 15 patients with an incorrect diagnosis of depression. Overall, 60% of depression diagnoses will be erroneous. If the prevalence is 10%, there will be five patients with unrecognised depression and 16.8 with misdiagnosed depression. Fewer than 1 in 4 patients diagnosed with depression will actually be depressed (PPV 23%). 

 

The ability of GPs to recognise depression is probably better than this study would suggest. In the first place, the most important clinical boundary lies between major and subthreshold depression, and GPs are more likely to recognise depression once this threshold has been crossed.³ Second, cross-sectional studies fail to recognise the longitudinal nature of general practice: a small study found that 22/54 patients not detected at baseline were diagnosed during follow-up.⁴

 

The most important message to draw from this study is that unassisted GP diagnoses of depression are usually wrong. Depression severity measures can play an important role in filtering out some of the false positives, providing the cut-off score is set at the appropriate level.