Editors' Note: Risk of Hospitalization and Death Associated With Pimavanserin Use in Older Adults With Parkinson Disease

Abstract

Dr. Hwang et al. examined the risk of hospitalization and death associated with pimavanserin use in a retrospective cohort study of 2,186 pimavanserin users with Parkinson disease (PD) compared with 18,212 nonusers with PD, all aged 65 years or older and residing in Medicare-certified long-term care facilities. They found that pimavanserin use was associated with a higher risk of hospitalization within 1 month of initiation and higher mortality up to 1 year after initiation. In response, Dr. Subbiah et al. argued that the study design and cohort selection do not support the authors' conclusions, citing the following key reasons: 1, pimavanserin is the only approved medication for patients with PD psychosis, itself associated with more advanced disease and mortality, whereas PD psychosis was not confirmed in nonusers in the cohort; and 2, pimavanserin users had higher rates of use of other antipsychotics, which are themselves associated with higher mortality in PD. They caution that such major baseline imbalances between groups cannot be adequately adjusted with the authors' propensity-weighted statistical analyses. They also point to pimavanserin's safety profile in the literature, particularly the reduction in all-cause mortality with pimavanserin compared with other atypical antipsychotics in PD in a previous study. Responding to these comments, the authors counter that they included several patient characteristics of psychosis in their analysis and that they limited their sample to long-term care residents with PD (presumably having more advanced PD). They point to the lack of dedicated International Classification of Diseases codes for Parkinson disease psychosis. They also note that after inverse probability of treatment weighting in their analyses, pimavanserin users and nonusers were well balanced for key patient characteristics including use of other antipsychotics. They argued that comparing mortality risk with pimavanserin vs other antipsychotics would not entail an inert comparison (although the editors believe that the reader response was in fact arguing for an active comparison). They cite previous studies that have raised concerns about mortality associated with pimavanserin. This exchange highlights the limitations of an administrative database in definitively capturing PD psychosis and the challenges of residual confounding in such studies of antipsychotic safety in these patients. Although pimavanserin did have an apparent association with hospitalization and mortality in this large retrospective cohort, the relative risks of hospitalization and mortality with pimavanserin compared with other antipsychotics remain to be elucidated. Dr. Hwang et al. examined the risk of hospitalization and death associated with pimavanserin use in a retrospective cohort study of 2,186 pimavanserin users with Parkinson disease (PD) compared with 18,212 nonusers with PD, all aged 65 years or older and residing in Medicare-certified long-term care facilities. They found that pimavanserin use was associated with a higher risk of hospitalization within 1 month of initiation and higher mortality up to 1 year after initiation. In response, Dr. Subbiah et al. argued that the study design and cohort selection do not support the authors' conclusions, citing the following key reasons: 1, pimavanserin is the only approved medication for patients with PD psychosis, itself associated with more advanced disease and mortality, whereas PD psychosis was not confirmed in nonusers in the cohort; and 2, pimavanserin users had higher rates of use of other antipsychotics, which are themselves associated with higher mortality in PD. They caution that such major baseline imbalances between groups cannot be adequately adjusted with the authors' propensity-weighted statistical analyses. They also point to pimavanserin's safety profile in the literature, particularly the reduction in all-cause mortality with pimavanserin compared with other atypical antipsychotics in PD in a previous study. Responding to these comments, the authors counter that they included several patient characteristics of psychosis in their analysis and that they limited their sample to long-term care residents with PD (presumably having more advanced PD). They point to the lack of dedicated International Classification of Diseases codes for Parkinson disease psychosis. They also note that after inverse probability of treatment weighting in their analyses, pimavanserin users and nonusers were well balanced for key patient characteristics including use of other antipsychotics. They argued that comparing mortality risk with pimavanserin vs other antipsychotics would not entail an inert comparison (although the editors believe that the reader response was in fact arguing for an active comparison). They cite previous studies that have raised concerns about mortality associated with pimavanserin. This exchange highlights the limitations of an administrative database in definitively capturing PD psychosis and the challenges of residual confounding in such studies of antipsychotic safety in these patients. Although pimavanserin did have an apparent association with hospitalization and mortality in this large retrospective cohort, the relative risks of hospitalization and mortality with pimavanserin compared with other antipsychotics remain to be elucidated.