Prevalence and predictors of potentially inappropriate prescribing using codified STOPP-START and Beers criteria: a retrospective cohort study in Ontario’s older population

Objectives — To quantify potentially inappropriate prescribing (PIP) and assess the association between patient characteristics and PIP using previously coded STOPP-START and Beers criteria in Ontario’s older population.

Study design and setting — An established subset of the 2014 STOPP-START and 2015 Beers criteria applicable to health administrative data were used to identify instances of PIP in health administrative data. Associations between the patient characteristics and PIP were examined using multivariable logistic regression. Using Ontario’s large health administrative databases, which comprise individual-level, linked information on medication dispensation, physician services use, emergency room visits, hospitalizations, mortality, and sociodemographic data, a cohort including all patients ≥65 years who were issued at least 1 prescription between April 2003 and March 2017 (N = 2,937,927) was formed.

Results — From a total of 2,937,927 patients, 2,220,641 (75.6%) patients were identified with at least 1 PIP using the STOPP-START criteria. Using the Beers criteria, 1,505,243 (51.2%) patients were identified. The most common PIP identified by the STOPP-START criteria was the lack of pneumococcal vaccine to be given at least once after age 65 years according to national guidelines (75.9% of patients). Patient characteristics that were found to be strongly associated with PIP identified by both STOPP-START and Beers criteria were age, female sex, long-term care resident, lack of MedsCheck prior to index date, and frailty, among others.

Conclusion — Applying coding for identifying PIP in health administrative data is a promising approach to screen for PIP at the population level in an impactful and cost-effective manner. This approach will allow investigators to identify areas for intervention in terms of PIP in a population.