BACKGROUND: Previous studies have attempted to forecast the costs of mental health care, using clinical and individual variables; the inclusion of ecological measures could improve the knowledge of predictors of psychiatric service utilisation and costs to support clinical and strategic decision-making. METHODS: Using a Psychiatric Case Register (PCR), all patients with an ICD-10 psychiatric diagnosis, who had at least one contact with community-based psychiatric services in the Verona Health District, Northern Italy, were included in the study (N = 4558). For each patient, one year's total cost of care was calculated by merging service contact data with unit cost estimates and clinical and socio-demographic variables were collected. A socio-economic status (SES) index was developed, as a proxy of deprivation, using census data. Multilevel multiple regression models, considering socio-demographic and clinical characteristics of patients as well as socioeconomic local characteristics, were estimated to predict costs. RESULTS: The mean annual cost for all patients was 2,606.11 Euros; patients with an ongoing episode of care and with psychosis presented higher mean costs. Previous psychiatric history represented the most significant predictor of cost (36.99% R2 increase) and diagnosis was also a significant predictor but explained only 4.96% of cost variance. Psychiatric costs were uniform throughout the Verona Health District and SES characteristics alone contributed towards less than 1% of the cost variance. CONCLUSIONS: For all patients of community-based psychiatric services, a comprehensive model, including both patients' individual characteristics and socioeconomic local status, was able to predict 43% of variance in costs of care.
BACKGROUND: Previous studies have attempted to forecast the costs of mental health care, using clinical and individual variables; the inclusion of ecological measures could improve the knowledge of predictors of psychiatric service utilisation and costs to support clinical and strategic decision-making. METHODS: Using a Psychiatric Case Register (PCR), all patients with an ICD-10 psychiatric diagnosis, who had at least one contact with community-based psychiatric services in the Verona Health District, Northern Italy, were included in the study (N = 4558). For each patient, one year's total cost of care was calculated by merging service contact data with unit cost estimates and clinical and socio-demographic variables were collected. A socio-economic status (SES) index was developed, as a proxy of deprivation, using census data. Multilevel multiple regression models, considering socio-demographic and clinical characteristics of patients as well as socioeconomic local characteristics, were estimated to predict costs. RESULTS: The mean annual cost for all patients was 2,606.11 Euros; patients with an ongoing episode of care and with psychosis presented higher mean costs. Previous psychiatric history represented the most significant predictor of cost (36.99% R2 increase) and diagnosis was also a significant predictor but explained only 4.96% of cost variance. Psychiatric costs were uniform throughout the Verona Health District and SES characteristics alone contributed towards less than 1% of the cost variance. CONCLUSIONS: For all patients of community-based psychiatric services, a comprehensive model, including both patients' individual characteristics and socioeconomic local status, was able to predict 43% of variance in costs of care.