Detecting mental disorders in general hospitals by the SCL-8 scale

Objective: The objective of this study is to validate the eight-item dichotomised version of the Symptoms Check List (SCL-8d) as a screening tool for psychiatric disorders. Methods: The study population included 198 consecutive new neurological inpatients and outpatients and 294 consecutive internal medical inpatients, aged 18 or older. All patients received the SCL-8d questionnaire, and a stratified subsample was interviewed using the Schedules for Clinical Assessment in Neuropsychiatry (SCAN) interview. We tested the external SCL-8d validity using the SCAN interview as gold standard. The test was performed based on weighted data to correct for the skewness introduced by stratification. Results: The diagnostic performance of the SCL-8d was excellent in the internal medical setting but not quite as good in the neurological sample. It performed better among the older compared with the younger patients, whereas the scale was not affected by gender. In the combined sample at the cut point 0/1, the sensitivity (SE) of the SCL-8d was 0.73 (confidence interval [CI](95%): 0.60-0.82), the specificity (SP) 0.61 (CI95%: 0.53-0.68) and the positive predictive value (PPV) 0.42 (CI95%: 0.34-0.50), using any International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) psychiatric disorder, excluding somatoform and substance abuse disorders, as gold standard. The risk of a patient having a mental disorder (except phobia, substance abuse or somatoform disorder) was less than 6% in case of a negative screening test. In patients with a current depressive disorder, 87.1% (27/31) were screening positive, and all except 1 (93.0%) of the 14 patients with a modest to severe depression scored 1 or higher on the SCL-8d. All 17 patients with an anxiety disorder, excluding phobias, were screening positive. Conclusion: The study suggests that the SCL-8d is a valid, brief screening tool for use in nonpsychiatric medical settings, especially to detect emotional psychiatric disorders (EPDs). (C) 2004 Elsevier Inc. All rights reserved.