Identifying potentially inappropriate prescribing in entire populations: coding the STOPP-START and Beers criteria for use with large, routinely collected population health datasets

Background and objectives — Adverse medication events significantly contribute to emergency department visits, unplanned hospitalizations, and in-hospital morbidity and mortality, particularly in older adults. Identifying potentially inappropriate prescribing is essential to improve safety, care quality, and medication management. The objective of this study was to identify and codify the Screening Tool of Older People’s Prescriptions (STOPP)-Screening Tool to Alert to Right Treatment (START) and Beers criteria for large, population-level health databases to detect potentially inappropriate prescribing at the population level.

Methods — A subset of the 2014 STOPP/START and 2015 Beers criteria applicable to health administrative data was codified using diagnostic (ICD; International Classification of Diseases) and medication (DIN; Drug Identification Number) codes using provincial health administrative databases in Ontario, Canada, which comprise individual-level linked information on medication dispensation, physician services use, emergency room visits, hospitalizations, mortality, and sociodemographic data.

Results — Overall, 103 of 177 (58.2%) criteria were codable (76.5% of 81 STOPP, 23.5% of 34 START, and 53.2% of 62 Beers). Some criteria could not be coded because the population health data used were missing information necessary to the operationalization of these criteria. This included laboratory values (renal function) and diagnostic or clinical information (such as blood pressure, body mass index) that could not be indirectly derived from the population health data.

Conclusions — Applying a large subset of codified and well-established clinical criteria to health administrative data offers a promising and potentially cost-effective approach to detect potentially inappropriate prescribing at the population level. The present study contributes the required coding which constitutes the core of this approach.