(299) An Active Safety Surveillance Using Real-World Evidence (ASSURE) Approach to Pharmacovigilance Signal Evaluation: The case of paliperidone, risperidone, and low-energy fracture

Background: While antipsychotics, including paliperidone and risperidone, are known to increase serum prolactin levels, the issue of attendant reductions in bone mineral density or increases in low-energy fracture (LEF) risk remains disputed. Routine pharmacovigilance identified several epidemiological studies in support of these associations, though mostly in comparison to non-user groups or in the context of case-control studies.

Objectives: To inform signal evaluation and causality assessment, we conducted a rapid-cycle analysis of the association between two antipsychotics, paliperidone and risperidone, and LEF using an active comparator new user cohort design in multiple real-world databases.

Methods: We identified new users of all forms of paliperidone and risperidone, as well as of comparator atypical antipsychotics (CAAs, associated with no or minimal increases in prolactin) in ten real-world databases. Patients were followed for the occurrence of diagnosed hyperprolactinemia, osteoporosis, and LEF during the period beginning six months after treatment initiation. Paliperidone and risperidone new users were matched to CAA new users on the basis of a large-scale propensity score in a 1:1 manner. Hazard ratios were estimated separately within each database and summarized using Bayesian random effects meta-analysis. Empirical calibration was performed using 93 negative control outcomes with no known association to any antipsychotic. Objective diagnostics based on covariate balance, equipoise, statistical power, attrition, and systematic error identified by negative controls were used to determine if database-specific results could be unblinded and included in meta-analysis.

Results: New users of paliperidone experienced a roughly four-fold higher incidence of hyperprolactinemia than did new users of CAAs (meta-analytic hazard ratio (HR) 4.31, 95% confidence interval (CI): 2.86 to 6.50), with similar results for risperidone (HR 4.50, 95% CI: 2.51 to 8.07). Similar incidences were seen across groups for both osteoporosis (HR: 1.07, 95% CI: 0.81 to 1.42 for paliperidone vs. CAAs, HR: 1.04, 95% CI: 0.86 to 1.26 for risperidone vs. CAAs) and for LEF (HR: 0.91, 95% CI: 0.70 to 1.17 for paliperidone vs. CAAs, HR: 1.01, 95% CI: 0.87 to 1.17 for risperidone vs. CAAs). These results were not sensitive to alterations in outcome definitions, nor did they differ substantially according to sex or database.

Conclusions: While both paliperidone and risperidone are associated with increased risks of hyperprolactinemia, this evaluation did not find them to be associated with increased risks of osteoporosis or LEF during the first five years of use.