Association of measures of socioeconomic deprivation with healthcare utilization in elderly patients enrolled in SWOG cancer clinical trials.

Abstract

6527 Background: Reducing unplanned Emergency Room (ER) visits and hospitalizations is an important strategy for improving value. Patients enrolled in clinical trials are treated per protocol. While demographic factors have been associated with increased healthcare utilization, less is known about measures of socioeconomic deprivation (SD), which may be a proxy for structural barriers to access. We tested whether ER and hospital stays are more common among trials participants who live in areas of SD or have Medicaid insurance (MI). Methods: We examined ER visits and hospitalizations among cancer patients ≥65 years treated on SWOG clinical trials from 1999-2018 using trial data linked to Medicare claims. Neighborhood socioeconomic deprivation (NSD) was measured using patients’ zip code linked to the Area Deprivation Index (ADI), measured on a 0-100 scale and categorized into tertiles (T1-T3). Higher ADI (T3) denotes areas of higher deprivation. Type of insurance was classified as Medicare alone or with Commercial, versus Medicare + MI. Outcomes were ER visits, hospital stays, and costs, all in the first year. Demographic, clinical, and prognostic factors were captured from trial records. Generalized estimating equations (GEE) were used, accounting for clustering by cancer type. Regression models were adjusted for age, race, and a study-specific prognostic risk score, and stratified by study and treatment. ADI and Insurance were considered separately to reflect neighborhood-level and individual-level measures of SD. Results: In total, N = 3,027 participants from 27 trials in bladder, breast, colorectal, lung, myeloma, and prostate cancers were analyzed. Median age was 71 years, 3% had Medicare + Medicaid and 22.3% were in the Highest ADI tertile. In all, 983 (32%) patients experienced hospitalization, and 1094 (36%) visited the ER. In multivariate GEE, patients living in more deprived areas were more likely to experience hospitalization (OR for ADI T3 compared to T1: 1.36, 95% CI, 0.95-1.94, p=.09) and to visit an ER (OR for ADI T3 compared to T1: 1.33, 95% CI, 1.08-1.63, p=.008). Overall, patients from the most deprived areas had a 61% increase in risk of either ER visit or hospitalization (OR for ADI T3 compared to T1: 1.61, 95% CI, 1.23-2.10, p<.001). Patients with MI were 86% more likely to visit an ER in the first year (OR 1.86, 95% CI, 1.49-2.33, p<.001. No increased risk of hospitalization was observed. Conclusions: Despite participation in cancer clinical trials, patients living in areas with higher deprivation or those with Medicare + MI had an increased risk of unplanned ER visits. These findings suggest that neighborhood deprivation and economic disadvantage may increase ER visits for socioeconomically vulnerable older patients with cancer. Efforts are needed to ensure adequate resources to prevent unplanned use of acute care in vulnerable populations.