Development of Screeners for Depressive Disorders and Substance Disorder History

TLDR

If screeners are to be widely incorporated in clinical care to increase detection of common psychiatric disorders, they must be brief, easy to score, and sensitive across diverse patient populations; few existing screeners meet these criteria, and if further validation supports these sensitivity and specificity estimates, these new instruments may also be valuable as initial brief screeners in a two-stage screening process to improve clinicians' recognition of common mental health problems that complicate case management and impair patient functioning. The study addresses this problem by identifying a subset of Diagnostic Interview Schedule questions that predict simultaneously obtained DIS diagnoses. The authors tested a two‑item depression/dysthymia screener and three‑item drug and alcohol disorder screeners in community, medical, and mental health patients. The depression screener achieved 83–94% sensitivity, the drug screener 91–94% (excluding one low‑prevalence site), and the alcohol screener 87–92% sensitivity, with all three screeners exceeding 90% specificity in community and medical samples and somewhat lower specificity among mental health patients, indicating that these brief screeners may be useful in epidemiologic studies and as initial brief screeners in a two‑stage process.

Abstract

If screeners are going to be widely incorporated in clinical care to increase the detection of common psychiatric disorders, they need to be brief, easy to score, and sensitive across diverse patient populations. Few screeners exist that meet these criteria. This problem is addressed in this study by identifying a subset of questions from the Diagnostic Interview Schedule (DIS), which predicted simultaneously obtained DIS diagnoses. A two-item screener to detect depression or dysthymia within the last year and three-item screeners for lifetime drug disorders and alcohol disorders were tested in community residents, medical, and mental health patients. The sensitivity of the depression screener ranged between 83% and 94%. The sensitivity of the drug screener ranged between 91% and 94%, excluding one site with an extremely low prevalence of drug problems. The sensitivity of the alcohol screener ranged between 87% and 92%. Specificity for all three screeners exceeded 90% in community and medical samples, while being somewhat lower among mental health patients. These findings indicate that these brief screeners may be useful in a variety of epidemiologic studies to provide estimates of common psychiatric disorders when complete diagnostic interviews are not feasible. If further validation studies support these sensitivity and specificity estimates, these new instruments may also be valuable as initial brief screeners in a two-stage screening process to improve clinicians' recognition of common mental health problems that complicate case management and impair patient functioning.