Abstract 10: Socioeconomically Disadvantaged Medicare Beneficiaries Had Worse One Year Clinical Outcomes But Lower Total Annual Costs of Care

Background: Recent data suggest that cost prediction models in the private insurance setting may systematically underpredict costs in high-risk subgroups due to uneven access to health care. This may lead to disparities in allocation of preventive services and other interventions. However, whether these patterns persist in an older population with higher rates of comorbidity and chronic disease is uncertain.

Objective: We set out to determine whether socially disadvantaged Medicare beneficiaries had disproportionately low costs compared to their clinical outcomes, which might serve as a marker for inadequate access to care.

Methods: We calculated one year all cause and CV hospitalization, death, and total costs for Current Medicare Beneficiary Survey participants from 2016-2017 (N=3,614). We used logistic and linear regression, as appropriate, to examine the relationship between key social determinants of health and clinical outcomes, as well as annual costs of care.

Results: For age and clinical risk, there was directional agreement between adverse events and annual cost. For example, those in the highest tertile of comorbidity index had the highest rates of adverse outcomes as well as higher costs compared to those in mid and low tertiles (Incidence Rate Ratio (IRR) for all-cause and cardiovascular hospitalization =3.29 and 4.37 respectively, OR for death=3.35; cost differential $8641, p<0.001 for all). However among the poor, minorities, and those with low education, adverse clinical outcomes were inversely related to cost. After multivariable adjustment, blacks/Hispanics had higher rates of CV hospitalization (IRR 1.78, p=0.012), but similar annual costs ($-336, p=0.77) compared to whites. This pattern of similar annual costs, despite higher rates of adverse outcomes, was also observed for those whose income was <25,000/year, those who had less than a college education, and those reporting living at an income poverty ratio of <100% (difference in costs $-373, $-712, & $-638, respectively, p>0.05).

Conclusions: A paradoxical relationship was observed between adverse outcomes and cost among racial/ethnic minorities, and those from low socioeconomic groups. Because of systematic differences in access to care, cost may not be an appropriate surrogate for predicting clinical risk among the poor or other vulnerable populations, and algorithms that are based on cost alone may lead to inadequate risk prediction.