Objective: There is an increasing interest in trying to identify patients with chronic physical illness who would benefit from interventions to decrease psychological distress. The Hospital Anxiety and Depression Scale (HADS) is one measure that can be effectively used to identify patients struggling with anxiety and/or depression in addition to comorbid medical illness such as cardiac disease. The aim of this study is to determine if the HADS correctly identified patients with depression according to the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition [determined using the Primary Care Evaluation of Mental Disorders (PRIME-MD), a gold-standard diagnostic tool], as depressed and to determine which items on the HADS most strongly predicted the PRIME-MD major depression diagnosis among cardiac patients. Method: Patient data were obtained from a randomized controlled trial of treatment for psychological distress in patients surviving recent myocardial infarction or life-threatening ischemic heart disease. The HADS was used to evaluate depression and/or anxiety among patients hospitalized for heart disease, and the PRIME-MD was used to diagnose major depressive disorder (MDD) in these patients. We used receiver-operating characteristic (ROC) curves and logistic regression to analyze patient responses to both of these measures. Results: Among the 79 study patients, the area under the ROC curve (AUC) for the HADS depression subscale was 0.81 with an S.E. of 0.05, and the AUC was 0.70 for the HADS anxiety subscale with an S.E. of 0.06. Using a cutoff of 7 on each HADS subscale to determine PRIME-MD diagnosis of MDD, the sensitivity and specificity were 81% and 54%, respectively, for the depression subscale and 81% and 40%, respectively, for the anxiety subscale. In addition, a few individual HADS items predicted PRIME-MD diagnosis of MDD, namely, concentration and positive anticipation (depression items) and relaxed and worry (anxiety items), each of which were strongly correlated with PRIME-MD diagnosis of MDD. Finally, items positive anticipation and relaxed provided reasonable approximations to the results found from the HADS depression subscale measure, which performed better than the HADS anxiety subscale in predicting PRIME-MD diagnosis of depression. Conclusions: Using the HADS to detect mental disorders in primary care populations can save valuable physician time, and perhaps ensure that patients are treated for their mental distress as well as medical disease. (c) 2005 Elsevier Inc. All rights reserved.