Potential Unintended Consequences of Antipsychotic Reduction in Ontario Nursing Homes

Abstract

Objective: Antipsychotic reduction in nursing homes has been a focus of research and policy attention for several decades; however, there is evidence that these initiatives may have had unintended consequences, such as medication substitution and changes in diagnosis coding. Our objectives were to describe temporal changes in the use of antipsychotics, potential substitution medications, and diagnoses and symptoms used to establish the appropriateness of antipsychotic prescribing.

Design: Repeated cross-sectional study design.

Setting and participants: Individuals, 66 to 105 years of age, living in nursing homes in Ontario, Canada between April 1, 2010 and December 31, 2019.

Methods: Linked health administrative and clinical data were used to estimate the quarterly prevalence of dispensed antipsychotics, antidepressants, anticonvulsants, and benzodiazepines, as well as diagnoses and symptoms (eg, schizophrenia and delusions) applied to establish the appropriateness of antipsychotic use. Generalized linear models with generalized estimating equations, binomial distribution, and identity link function estimated the absolute changes over time among the population overall, by dementia diagnosis, and by severity of aggressive behaviors and cognitive impairment.

Results: Among over 70,000 nursing home residents each quarter, we observed general declines in antipsychotic (-0.70% per year [95% confidence limit (CL) -0.74%, -0.66%]) and benzodiazepine use (-1.17% per year [95% CL -1.20%, -1.14%]), and increases in antidepressant (0.89% per year [95% CL 0.84%, 0.94%]) and anticonvulsant use (1.06% per year [95% CL 1.03%, 1.09%]). Although initially stable, the coding of delusions increased from 3.5% to 10.2% (1.18% per year [95% CL 1.12%, 1.24%]) between 2014 and 2019, whereas schizophrenia remained stable. Increases in antidepressant use and recorded delusions were more pronounced among residents with dementia and aggressive behaviors.

Conclusions and implications: Medication substitution and changes in diagnosis/symptom coding may be potential unintended consequences of initiatives to reduce inappropriate antipsychotic use. How this corresponds to resident-level prescribing and clinical outcomes warrants further investigation.

Keywords: Medication substitution; antipsychotic reduction; diagnosis coding; nursing homes.