Validity of the 12-item version of the General Health Questionnaire in detecting depression in the general population

The 12-item version of the General Health Questionnaire (GHQ-12) is frequently used to measure common mental disorder in public health surveys, but few population-based validations have been made. We validated the GHQ-12 against structured psychiatric interviews of depression using a population-based cohort in Stockholm, Sweden.

We used a population-based cohort of 484 individuals in Stockholm, Sweden (participation rate 62%). All completed the GHQ-12 and a semi-structured psychiatric interview. Last month DSM-III-R symptoms were used to classify major and minor depression. Three scoring methods for GHQ-12 were assessed, the Standard, Likert and Corrected method. Discriminatory ability was assessed with area under the receiver operating characteristic (ROC) curve.

A total of 9.5% had a major or minor depression. The area under the ROC curve was for the Standard method 0.73 (0.65-0.82), the Likert method 0.80 (0.72-0.87) and the Corrected method 0.80 (0.73-0.87) when using major or minor depression as standard criterion. Adequate sensitivity and specificity for separating those with or without a depressive disorder was reached at ≥12 Likert scored points (80.4 and 69.6%) or ≥6 Corrected GHQ points (78.3 and 73.7%). Sensitivity and specificity was at ≥2 Standard scored points 67.4% and 74.2%.

When scored using the Likert and Corrected methods, the GHQ-12 performed excellently. When scored using the Standard method, performance was acceptable in detecting depressive disorder in the general population. The GHQ-12 appears to be a good proxy for depressive disorder when used in public health surveys.