Real-world evidence of improved healthcare utilization in patients with schizophrenia or schizoaffective disorder after early treatment of paliperidone palmitate once-monthly treatment in Hong Kong.

Background: Very few data are available to demonstrate the economic benefit of early paliperidone palmitate once-monthly long-acting injectable (PP1M) treatment in patients with schizophrenia or schizoaffective disorder. Methods and materials: This study has retrospectively compared the healthcare utilization and associated costs of pre- and post-PPIM treatment in 413 patients with schizophrenia or schizoaffective disorder recruited from three major public hospitals providing psychiatric services in Hong Kong. Patients were categorized into early treatment (<= 3 years since diagnosis) and chronic (>3 years) groups, and also whether they were receiving polypharmacy (POP). Results: It was found that patients who were started on early therapy with no POP had the most favourable outcomes. Overall results of the entire cohort, including both early and late treatments, indicate that there was a slight increase in annual in-patient days (IP) per patient and outpatient visit (OP) by 3.18 and 1.87, respectively, and a decrease in emergency room visit (ER) of 0.9 (p < 0.05). For non-polypharmacy (NP) patients receiving early PP1M therapy, there was a significant decrease in IP and ER of 21.56 (p < 0.05) and 1.15 (p < 0.05), respectively, but an increase in OP of 1.88 (p < 0.05). For patients with POP, there was an all-across increase in IP and all-across decrease in OP and ER. In monetary terms, a NP patient receiving early therapy may have an overall saving of HKD40,878 (USD5,241, 1USD = 7.8HKD) per year compared to HKD6,224 (USD798) in patients where therapy was given after 3 years. For patients with POP, there was an all-across increase in overall spending despite reductions in OP and ER. Conclusions: From the 413 patients studied, potential annual savings is higher by early administration of PPIM in patients with NP. Analysis using multivariate linear regression based on generalized estimating equations and sensitivity analysis using a linear mixed model supported the findings.