Psychometric properties of the GHQ-28 in opiate-dependent patients

To know the psychometric properties of the GHQ-28 when it is applied to a sample of opiate-dependent patients. Participants were 138 opiate-dependent patients. Reliability was estimated by means of the Cronbach's alpha and Spearman-Brown coefficients. Confirmatory Factor Analysis was applied to check the four-factor structure proposed by Goldberg and Hillier. The sensitivity and specificity of the instrument for the detection of possible psychiatric cases were measured using ROC (Receiver Operating Characteristic) analysis, using as criterion the DSMIV diagnosis from the Mini International Neuropsychiatric Interview. Reliability estimation was adequate. Validity results could not confirm the four-factor structure of GHQ-28 proposed by Goldberg and Hillier. The sensitivity and specificity parameters show optimum values of 60.7 % and 73.7 %, respectively, 65.9 % of the participants being correctly classified. GHQ-28 shows adequate reliability and appropriate sensitivity and specificity parameters when it is applied to opiate-dependent patients. However, the original factorial structure could not be confirmed. Given these results and those from other validity studies in which the relationships with Severity of Dependence, Physical Health and Health-Related Quality of Life were shown, the GHQ-28 can be considered a useful instrument for screening in clinical practice.