Assessments of service utilization is often based on self-reports. Concerns regarding the accuracy of self-reports are raised especially in mental health care. The purpose of this study was to analyze the accuracy of self-reports and calculated costs of mental health services. In a prospective cohort study in Germany, self-reports regarding psychiatric inpatient and day-care use collected by telephone interviews based on the Client Socio-Demographic and Service Receipt Inventory (CSSRI) as well as calculated costs were compared to computerized hospital records. The sample consisted of patients with mental and behavioral disorders resulting from alcohol (ICD-10 F10, n=84), schizophrenia, schizophrenic and delusional disturbances (F2, n=122) and affective disorders (F3, n=124). Agreement was assessed using the concordance correlation coefficient (CCC), mean difference (95% confidence intervals (CI)) and the 95% limits of agreement. Predictors for disagreement were derived. Overall agreement of mean total costs was excellent (CCC=0.8432). Costs calculated based on self-reports were higher than costs calculated based on hospital records (15 EUR (95% CI -434 to 405)). Overall agreement of total costs for F2 patients was CCC=0.8651, for F3 CCC=0.7850 and for F10 CCC=0.6180. Depending on type of service, measure of service utilization and costs agreement ranged from excellent to poor and varied substantially between individuals. The number of admissions documented in hospital records was significantly associated with disagreement. Telephone interviews can be an accurate data collection method for calculating mean total costs in mental health care. In the future more standardization is needed.
