LAUSR

PURPOSE Screening for potentially inappropriate medications (PIM) is recommended in older patients with cancer receiving chemotherapy, given the concern for adverse drug reactions, drug-drug interactions and non-adherence. Our objective was to determine the impact of PIM on outcomes in patients with breast and colorectal cancers receiving chemotherapy. METHODS We used data from the SEER-Medicare database, including patients >/= 66 years old with a diagnosis of Stage II/III breast and colorectal cancer made between 7/1/2007-12/31/2009. We used modified STOPP criteria to define baseline PIM as a dichotomous variable in the 4 months prior to diagnosis. STOPP criteria was used based on its performance as a robust measure of PIM. Outcomes measures included ER visits, hospitalizations, and death within 3 months from the last chemotherapy, and a composite of the three. We used Chi-square or Fisher's exact test to determine associations of PIM with covariates and outcomes, and Cox proportional hazards (PH) model for the time-to-event analysis. RESULTS Final analysis included 1,595 patients with breast cancer and 1,528 patients with colorectal cancer. Frequency of baseline PIM by STOPP was 31.5% in the breast and 30.9% in the colorectal cohort. In the breast cohort, associations with the composite outcome in the Cox PH model included disease stage, comorbidity, medication number and baseline ER visits/hospitalization. Age, gender, race, comorbidity and baseline ER visits/hospitalization were associated in the colorectal cohort. PIM was not associated with outcomes in either cohort, aside from hospitalization in the breast. CONCLUSIONS We found no consistent association between pre-chemotherapy PIM defined by STOPP and outcomes.