Effect of written and computerized decision support aids for the U.S. Agency for Health Care Policy and Research depression guidelines on the evaluation of hypothetical clinical scenarios.

OBJECTIVE: The objective of this study was to compare the effects of written and computerized decision support aids (DSAs) based on U.S. Agency for Health Care Policy and Research depression guidelines. METHODS: Fifty-six internal medicine residents were randomized to evaluate clinical scenarios using either a written or a computerized DSA after first assessing scenarios without a DSA. The paired difference between aided and unaided scores was determined for diagnostic accuracy, treatment selection, severity and subtype classification, antipsychotic use, and mental health consultations. RESULTS: Diagnostic accuracy with the written DSA increased from 64% to 73%, and with the computerized DSA decreased from 67% to 64% (P=0.0065). Residents using the computerized DSA (vs. no DSA) requested fewer consultations (65% vs. 52%, P=0.028). In post hoc analysis, the written DSA increased sensitivity (66% to 89%, P<0.001) and the computerized DSA improved specificity (66% to 86%, P=0.0020) but reduced sensitivity (67% to 49%, P = 0.011). CONCLUSIONS: A written DSA improved diagnostic accuracy, whereas a computerized DSA did not. However, the computerized DSA improved specificity and reduced mental health consultations.

Internal medical residents' ability to diagnose and characterize major depression.

The purpose of this study was to assess medical residents' knowledge of symptom criteria and subtypes of major depressive episode and their accuracy in diagnosing major depressive disorders and classifying episode severity and subtype according to criteria of the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition. Thirty-five third-year internal medicine residents completed a self-administered, written instrument containing 2 open-ended questions and 21 hypothetical scenarios. The sensitivity for recognizing major depressive disorder was 64%, and the specificity was 69%. The sensitivity for classifying severity was 86% for mild, 66% for moderate, 71% for severe, and 66% for severe with psychosis. Misclassification of severity was most commonly to a less severe class. For scenarios with a diagnosable subtype of a major depressive disorder, the sensitivity for classification was 34% for atypical, 51% for catatonic, 74% for melancholic, 100% for postpartum, and 94% for seasonal depression. When asked to enumerate the criteria symptoms for depression, 80% or more of the residents listed sad mood, loss of interest, weight change, and sleep disturbances; 14 to 21 (40%-60%) listed thoughts of death and worthlessness; other criteria were listed by 7 to 11 (20%-31%). When asked to list the episode subtypes, none was listed by more than 3 (9%) residents, although 13 (37%) residents volunteered psychotic as a subtype. Residents frequently failed to recognize the presence or absence of major depressive disorder and often misclassified episode severity and subtype on scenarios. Few could spontaneously list the episode subtypes. Methods must be developed to improve the recognition and classification of major depressive episodes to better direct treatment.