Risk Adjusting Readmissions for Social Factors

January 27, 2016

Author: Mat Reidhead

What did it take to convince one of the world’s most preeminent health policy researchers that risk adjustment for sociodemographic factors is necessary for equitable hospital assessments on readmissions performance? One, a homeless veteran with uncontrolled diabetes and two, a Harvard health policy statistician.

Ashish Jha is a general internist, professor of health policy at Harvard’s T.H. Chan School of Public Health and possibly the most heavily-cited researcher on the policy implications underlying the Hospital Readmissions Reduction Program. Dr. Jha had been a long-time subscriber to the notion that adjusting for sociodemographic status would statistically codify a lower standard of care for low-income patients, creating a “soft bigotry of low expectations” for hospitals who treat these patients, which in turn would perpetuate their health disparities. Then he discussed the readmission case of a 64 year-old diabetic homeless veteran with Alan Zaslavsky, the second aforementioned ingredient. Using data, Dr. Zaslavsky demonstrated how, when applied effectively, risk adjustment for SDS does not excuse poor care for poor patients, but rather avoids penalizing safety-net hospitals simply for caring for more poor patients. This is a very important distinction.

The continued decision by the Centers for Medicare & Medicaid Services to exclude SDS factors in the risk adjustment of outcomes measures that determine “value-based” carrots and sticks carries the risk of perpetuating disparities in its own right. This risk is borne by the potential of misguided policies to penalize our safety net out of business, or equally tragic, by disincentivizing the provision of care for these patients. This would be catastrophic for low-income communities that often depend on this safety net for access to the entire spectrum of care — primary to quaternary.

In the spirit of Zaslavsky, The Missouri Hospital Association is using data to expedite what many policy experts view as CMS’ inevitable reversal of its policy stance on SDS. Using an augmented version of CMS’ own readmissions methodology, we developed measures that account for Medicaid status, poverty and risk factors attributable to patients’ communities. The results are striking. For all conditions modeled, SDS factors significantly reduce the range of measured quality differences between hospitals, suggesting that 43 to 88 percent of this variation can be explained by Medicaid status and the communities in which patients reside. Hospitals with the highest mix of patients with Medicaid, and from low-income communities, saw the most improvement relative to the standard CMS methods.

Armed with these data, MHA has garnered the support of several national and state hospital associations and policy experts from across the country. We are working to coalesce these voices and bring the issue to CMS, their measure developers at Yale and Congress.

Modern Healthcare recently ran an article, “CMS admits to bad dual-eligible math.” In it, Sean Cavanaugh (who leads the CMS Center for Medicare) conceded that dual-eligibility and disability status should be used in risk-adjusted quality measures. “What I want you to take away from this is that the industry brought an issue to us and we took it seriously.” Unfortunately, the industry he referred to was Medicare MCOs and the quality measures in question were the health plans’ star rating system. This concession begs the question — when will CMS take the hospital industry seriously on virtually the exact same issue?