Do primary care physicians have particular difficulty identifying late-life depression? A meta-analysis stratified by age.

BACKGROUND

There is long-standing concern regarding poor recognition of depression in primary care, especially in older people.

METHODS

Studies that examined the unassisted (clinical) ability of general practitioners (GPs; primary care physicians) to identify depression were divided into those of older adults, younger adults and mixed populations. Data were extracted by 3 reviewers independently and pooled using a Bayesian meta-analysis.

RESULTS

We identified 31 valid studies that examined both sensitivity and specificity (or rule-in and rule-out accuracy), involving 52,513 individuals. Twelve studies recruited older individuals, 12 recruited younger adults and 7 recruited both younger and older adults (mixed populations). In the most robust studies the point prevalence of depression in late life was 13.2% (95% CI = 7.9-19.6). GPs were able to correctly identify 47.3% of the late-life depressions and 78.6% of the non-cases (71.0% overall accuracy). In younger adults GPs were able to identify 39.7% of the mid-life depressions and 85.1% of the non-depressed (77.8% overall accuracy). In mixed aged groups GPs we able to correctly identify 46.6% of the depressed individuals and 86.2% of the non-depressed (79.6% overall accuracy). The overall fraction correctly identified was significantly lower in older compared with younger adults. Correcting for differences in prevalence showed a statistically lower rule-in performance for older compared with younger adults. There was no difference in ability to identify non-depressed (healthy) individuals by age.

CONCLUSIONS

In clinical practice GPs appear to be less successful in identifying depression in older people than in younger adults, however there have been few head-to-head studies stratified by age from one centre.