Potentially Inappropriate Medications, Frailty, and Outcomes in Patients With Cancer Managed in a National Health Care System

Abstract

Background: Previous studies have operationalized the NCCN list of high-risk medications in older adults into a measurable tool known as the Geriatric Oncology Potentially Inappropriate Medications (GO-PIMs) scale. The current study aims to evaluate the ability of GO-PIMs to identify high-risk medications and their impact on patients with both solid and liquid tumors managed in a large national health care system.

Methods: We performed a retrospective cohort study using data from the national Veterans Affairs (VA) Cancer Registry and electronic health records, including all veterans newly diagnosed with a solid or liquid malignancy from 2000 to 2022. The number of GO-PIMs for each patient was determined from outpatient pharmacy prescriptions filled in the 90 days preceding the initial cancer diagnosis (the index date). We assessed the association of PIMs with baseline frailty-measured using the electronic Veterans Affairs Frailty Index (VA-FI) and categorized as nonfrail (≤0.2), mildly frail (>0.2-0.3), or moderate-to-severely frail (>0.3)-and with unplanned hospitalization and mortality during follow-up, using multivariable models adjusted for age, gender, cancer type and stage, Charlson comorbidity index score, and socioeconomic factors.

Results: Among 388,113 patients with newly diagnosed cancer (median age, 69.3 years [IQR, 62.8-76.7]; most common cancer types: prostate [21.5%], lung [23.7%], and gastrointestinal [20.5%]), GO-PIMs were prevalent, with 38% patients receiving ≥1 GO-PIM. Each additional GO-PIM was associated with a 66% increase in the odds of being mildly or moderate-to-severely frail at diagnosis, after adjusting for all covariates (ordinal regression adjusted odds ratio, 1.66; 95% CI, 1.65-1.67). Each additional GO-PIM was also associated with a higher hazard of unplanned hospitalization (Cox regression adjusted hazard ratio [aHR], 1.08; 95% CI, 1.07-1.08) and death (Cox regression aHR, 1.07; 95% CI, 1.06-1.07), after adjusting for frailty and all covariates.

Conclusions: An increasing number of PIMs, as identified by the GO-PIMs scale, was independently associated with greater risk of frailty at diagnosis, unplanned hospitalization during follow-up, and mortality among patients treated within a large national health care system.