Validation of the 13-Item Beck Depression Inventory in alcohol-dependent people

Abstract

Aims. To validate a self-administered questionnaire (the 13-item Beck Depression Inventory) as an instrument for assessing depressive symptoms in alcohol-dependent people. Design, setting and participants. One hundred and eight treatment-seeking alcohol-dependent patients were recruited from a teaching hospital substance misuse facility. Measures. The Beck Depression Inventory, Montgomery Asberg Depression Rating Scale and Hamilton Rating Scale for Depression were administered. Scores from each instrument were compared using correlation coefficients. Findings. Correlation between the Beck Depression Inventory and Montgomery Asberg Depression Rating Scale was 0.763. The Beck Depression Inventory achieved a high degree of consistency/reliability (Cronbach's α=0.8847). Receiver operated curve analysis gave an optimal cut-off on the 13-item Beck Depression Inventory of 18/19 out of 39 as a screening tool to identify cases with moderate or severe depression. This cut-off gave a sensitivity of 79% and a specificity of 79% compared to the Montgomery Asberg Depression Rating Scale. The diagnostic efficient was 82% for moderate to severe depression compared to the Montgomery Asberg Depression Rating and 85% compared to Hamilton Depression Rating Scale. Conclusions. The self-administered Beck Depression Inventory is a feasible, valid and reliable alternative to the interview for detecting change in depressive symptoms in alcohol-dependent people. This would be particularly useful in services with very limited staffing time such as primary care.

• Alcohol dependence
• depression
• self-completion questionnaire
• 13-item Beck Depression Inventory

All patients were under the care of Dr M. Farrell, Dr S. Welch or Dr E. Finch, Consultant Psychiatrists at Marina House, Maudsley Hospital, London. Dr M. Farrell was the principal applicant on the ethical committee application. Statistical analysis was carried out by Dr V. Nikolaou, Consulting Statistician, Department of Biostatistics, Institute of Psychiatry, Maudsley Hospital. All funding for the project was provided by the principal author (JL).