Developing an economic model for LEAN: a text-messaging intervention for schizophrenia in rural China

Abstract

Background: Stakeholders in schizophrenia care in China are unaware of the economic value and overall cost savings associated with a text-messaging intervention for schizophrenia, known as LEAN, an intervention shown to improve adherence to oral psychotropic medications and reduce relapses and re-hospitalizations in rural China. Objective: To develop an economic model and utilize this model to perform a preliminary analysis of the cost-effectiveness of the LEAN intervention for schizophrenia in rural China. Methods: Data from the LEAN trial and other published sources were utilized to create the model. An iterative process involving repeated cycles of looking at this data, constructing models, and discussing with a team was used to actually develop the economic model. We decided to use a cohort simulation to test the model and perform a preliminary cost-effectiveness analysis. Results: Model Development: A decision-analysis model consisting of two linked models, a decision tree (Figure 1) and a Markov model (Figure 2) took shape, and we applied the final parameters (time horizon, cycle length, transition probabilities, utilities, costs, etc) to align with each aspect of the model. A cost-effectiveness analysis was performed using decision and Markov modeling of schizophrenia patients divided into well (stable) and severe (severely relapsed) states of the illness from the perspective of the Chinese healthcare system. The main outcome measures were Quality-adjusted life years (QALYs) gained over 2 years and the incremental cost-effectiveness ratio (ICER) measured as a cost per QALY gained over 2 years. In running a base-case analysis using cohort simulation with our model to assess the cost-effectiveness of LEAN, we found the 2-year discounted QALYs were higher for the LEAN plus 686 Program [intervention] arm (1.45) than for the 686 Program alone [control] arm (1.42). In the incremental analysis, the LEAN plus 686 Program [intervention] arm was dominant, i.e., less costly (by 196 USD or 1,362 RMB, over 2 years) and more effective by 0.03 QALYs (over 2 years). Discussion: We were able to construct an economic model and in our preliminary analysis, the addition of LEAN to the 686 Program is cost-effective. However, better data are needed to both further develop the model and better parameterize it. While we’d like to be able to say that our base-case analysis results mean that LEAN is a good candidate for scaling up and modifying for sustainability in order to improve the provision of oral psychotropic medication for people with schizophrenia in rural China, with potential adaptation to other contexts as well, we cannot say this at this time. Future adaptations of this model with better data and included sensitivity analyses will have to provide this.