Variation and correlation of potential unintended consequences of antipsychotic reduction in Ontario nursing homes over time

Background — Potentially inappropriate antipsychotic use has declined in nursing homes over the past decade; however, increases in the documentation of relevant clinical indications (eg, delusions) and the use of other psychotropic medications have raised concerns about diagnosis upcoding and medication substitution. Few studies have examined how these trends over time vary across and within nursing homes, information that may help to support antipsychotic reduction efforts.

Objective — To jointly model facility-level time trends in potentially inappropriate antipsychotic use, antidepressant use, and the indications used to define appropriate antipsychotic use.

Research Design — We conducted a repeated cross-sectional study of all nursing homes in Ontario, Canada between April 1, 2010 and December 31, 2019 using linked health administrative data (N=649). Each nursing home’s quarterly prevalence of potentially inappropriate antipsychotic use, antidepressant use, and relevant indications were measured as outcome variables. With time as the independent variable, multivariate random effects models jointly estimated time trends for each outcome across nursing homes and the correlations between time trends within nursing homes.

Results — We observed notable variations in the time trends for each outcome across nursing homes, especially for the relevant indications. Within facilities, we found no correlation between time trends for potentially inappropriate antipsychotic and antidepressant use (r=−0.0160), but a strong negative correlation between time trends for potentially inappropriate antipsychotic use and relevant indications (r=−0.5036).

Conclusions — Nursing homes with greater reductions in potentially inappropriate antipsychotics tended to show greater increases in the indications used to define appropriate antipsychotic use—possibly leading to unmonitored use of antipsychotics.