Prioritizing medication review for older individuals: A real-world data study using administrative databases

Abstract

Aim: Optimizing drug treatments in older individuals is essential for improving health outcomes and reducing drug-related issues. However, targeting older adults for interventions is challenging and tools to identify and prioritize individuals with potentially inappropriate medications (PIMs) are lacking. This study aims to develop a prioritization algorithm for medication review, with a proof-of-concept established using Italian administrative data, and to assess the association between PIMs and all-cause hospitalization.
Methods: Eight indicators were selected:
1) medications that should be avoid in elderly,
2) drugs linked to fall risk or orthostatic hypotension,
3) drug-drug interactions,
4) Anticholinergic Cognitive Burden,
5) Sedative Load,
6) therapeutic duplicates,
7) polytherapy,
8) drugs with higher risk of adverse drug reactions.
This study focused on the first indicator. Administrative healthcare data from Local Health Units (LHUs) in Lombardy were used to identify over 65 individuals who redeemed a PIM between 2015 and 2018, with index date defined as the first PIM redemption. Risk-set matching was used to select controls, adjusted for high-dimensional propensity scores (HDPS) logistic regression models were used to assess the odds of all-cause hospitalization within 90 days.
Results: A total of 499,511 over 65 adults across the LHUs were evaluated. Between 27.4% and 37.7% individuals were exposed to at least one PIM with higher prevalence in adults aged 65-74 years and women. After matching, 128,063 pairs were analyzed. Hospitalization rates were higher among exposed individuals (8.3-10.2%) compared to controls (5.1-6.0%). Multivariate regression showed a 55% increased risk of hospitalization for those exposed to PIMs (OR 1.55, 95% CI 1.48-1.62).
Conclusions: This proof-of-concept study made it possible to develop an analytical model, which will be implemented for the other indicators. The strength of the association between each indicator and the risk of hospitalisation will be used as a weight in the construction of the prioritisation algorithm.