Using health production functions to evaluate treatment effectiveness: an application to a community mental health service.

The randomized controlled trial (RCT) is the recommended means of evaluating health care effectiveness and cost-effectiveness. Whilst representing a 'gold-standard' in health services research, RCT evidence on the clinical and economic desirability of services and treatments is often absent. Where RCT evidence is lacking, or where it is infeasible to implement randomized controlled comparisons, longitudinal observational and naturalistic data sources when analysed appropriately can yield useful insights regarding the clinical effectiveness and economic efficiency of treatments. In this paper we demonstrate the utility of applying panel estimation methods to data from an Italian psychiatric case register as a means of modelling the mental health outcomes of patients referred to a community-based mental health service. Emphasis is placed on quantifying the clinical effectiveness of consultations with different mental health professionals (including in-patient days) and whether service outcomes are affected by psychiatric diagnosis. The impact of service consultations and their interaction with different types of psychiatric diagnosis on a measure of patient mental health are found to be statistically significant, although the size of these effects are not substantial from a clinical perspective.