Accuracy of case-finding instruments for depression: a meta-analysis

Article type
Authors
Ramirez G, Mulrow CD, Williams JW, Gerety MB, Kerber C
Abstract
Introduction: Although depressive disorders are common among patients of general practitioners, they are poorly recognized. Several case-finding instruments (CFIs) are available to identify patients with depression.

Objective: To evaluate the sensitivity and specificity of CFIs in general practice settings for identifying patients with depression.

Methods: Computerized data-bases of English-language medical literature, bibliographies, and experts were culled to identify studies conducted in general practice settings that compared CFIs to accepted criterion standards for major depression. Two independent reviewers rated quality and abstracted data of selected studies. Data was formatted using two-by-two tables. Kraemer's adjustment method was used if the criterion standard had been given only to a random sample of persons screening negative on CFIs. Weighted average false-positives, false-negatives, sensitivities, and specificities were calculated for specific CFIs and overall CFIs.

Results: Of 906 articles identified from the search, 210 were deemed potentially relevant. Of these, 29 met selection criteria. Most common quality problems were selection biases and lack of independent criterion standard assessment. Nine different CFIs were studied. Sensitivities and specificities for major depression ranged from 80% to 100% and 60% to 90%, respectively. No significant differences between most CFIs were found. Overall sensitivity and specificity was approximately 80%. Sensitivities decreased when identifying dysthymia. Stratified analyses of data by quality and other study characteristics are in progress.

Discussion: Several different CFIs with reasonable operating characteristics are available to help practitioners identify major depression. Choices between CFIs should be dependent on other issues such as feasibility, cultural appropriateness, administration and scoring times, and ability to also use CFIs for monitoring severity and/or therapeutic response.