Background: Central nervous system medication (CNS med) use among older adults is associated with high morbidity and mortality. This study described patterns of CNS medication use, and identified people most likely to have CNS medications initiated or deprescribed around hospitalization.

Methods: Retrospective cohort study using electronic health record data from 3-hospitals. Patients aged 65 years or older, hospitalized on general medicine between 2018-2020, and prescribed an outpatient CNS med (antianxiety, anticholinergics, anticonvulsants, antidepressants antipsychotic, hypnotics, muscle relaxants, and opioids) between the 90 days prior- to 90 days after hospitalization were included (N=4666). Latent transitions analysis (LTA) was used to identify profiles of CNS med class use, examine transitions between profiles across four time-points (90 days before hospitalization, admission, discharge, 90 days after hospitalization), and determine predictors of transitions between profiles.

Results: Mean age was 74.3±9.3 years; 63% were female, 70% were White. The most commonly prescribed CNS meds were antidepressants (54%) and opioids (49%). Overall, 74% (n=3446) of patients were persistent users of a CNS med (any) across all four time points; 7% (n=388) had a CNS med deprescribed during hospitalization, but 64% (216/388) of these patients had a CNS med started or restarted within 90 days following hospitalization. LTA identified three profile groups: 1) overall low CNS med users, 54-60% of patients; 2) mental health med users (predominantly antianxiety, antidepressants), 30-36%; and 3) acute/chronic pain med users (predominantly muscle relaxants, opioids), 9-10%. Probability of staying in same group across the four time-points was high (0.88-1.00). Of those who transitioned to the low CNS med use group (indicating deprescribing), the probability was highest from admission to discharge (0.09 for chronic pain med users, 0.05 for mental health med users). Being female increased (OR 2.4, 95% CI 1.3-4.3) and having chronic kidney disease lowered (OR 0.5, 0.2-0.9) the odds of deprescribing.

Conclusions: Around an episode of hospitalization the majority of patients were persistent users of CNS meds and there was little deprescribing. CNS meds could represent important targets for hospital deprescribing, but additional work on patient outcomes is needed to understand potential benefits and harms.