Will hospital peer grouping by patient socioeconomic status fix the Medicare HRRP or create new problems?

March 21, 2018 / By Richard Fuller, MS

In this blog, we return to how socioeconomic status is accounted for when measuring patient outcomes. We have actively participated in the debate on how to identify potentially preventable readmissions and how to factor in socioeconomic status.  We have previously commented on Medicare’s Hospital Readmission Reduction Program (HRRP) , highlighting the impact of small performance variation in generating large penalties and, along  with many others^1–5, noting that the policy disproportionately impacts hospitals serving large number of low socioeconomic status (SES) patients.

Let’s be clear, it is a very complicated task to define what is meant by socioeconomic status and to gather standardized variables that capture that concept once it is defined. We can argue that the concept is wrapped up in some mixture of dimensions that includes income (flow of money), capital (store of money), social capital (having those around us that will aid and care for us when needed) and education/literacy (the ability to understand and succeed in our environment). Factors such as housing insecurity/homelessness are correlated with all of these factors and can be considered as factors in their own right or a byproduct of poverty and social isolation. It is also understood that measurement of these dimensions are context dependent. For example, an individual without their own housing, lower than high school education, without income or capital may be of low SES or may be a child. Context is key and providing defining principles and structure a complicated task. With this as background we turn to our recent article in the Journal of the Joint Commission.

The article has three main findings. First, the use of SES peer groups ignores distortion that results from other hospital characteristics influencing readmission rates that may be unevenly distributed across the peer groups. Second, that peer groups have been established too broadly (in order to promote stability) such that hospital differences within peer groups are larger than differences between them. And third, that there remains an observable (but unexplained) phenomenon by which smaller hospitals tend to have lower readmission rates for medical admissions than large hospitals. (As expected however, larger hospitals have lower readmission rates than smaller hospitals for surgical admissions),

While we recognize the complexity of the task at hand, and maintain the belief that for the Medicare HRRP an adjustment for SES is necessary, both the approach and implementation by CMS is inadequate and possibly harmful. Faced with the complexity of defining precisely what SES means (and how to measure it) the approach enshrined within the 21^st Century Cures Act was to develop “peer groups” that would indirectly capture its effects. Thus the hard, time consuming labor of developing an SES measure to account for its effects upon the readmission risk of an individual was avoided and replaced by a measure of how many lower income people were on average seen by a hospital. Having ranked hospitals by this variable and constructing peer groups from the result, all that was left was to compare  relative performance and to give each hospital the target from their peer group. But notice that SES no longer resembles a complex variable but rather it is reduced to the average poverty of patients and the effects are no longer tied to those being treated for a particular condition but the hospital’s global average. Most importantly , as we expound in the article, indirect measurement via peer grouping picks up not only variation in the intended variable (SES) but other factors that may influence outcomes – such as the expected hospital volume response for larger hospitals (“practice makes perfect”) with surgical admissions and the aforementioned unexplained performance of smaller hospitals with medical admissions.

But even accepting that limitations of approach were forced upon CMS through the regulations, the implementation set out to minimize the measured effect of SES on outcomes. It became clear that hospitals with larger shares of patients identified as being of low SES had different outcomes than those with which they share an assigned peer group. It is highly likely that those hospitals that lobbied long and hard for an SES adjustment, in particular safety net hospitals, will be left wondering why they bothered.

In closing, we state our hope that this is not the end but the beginning. We need to address the issue of disparities, how to measure them and when to incorporate them when evaluating provider performance (and when not). Specifically we believe that CMS should consider the following research and policy steps:

1. Build a working model of how to capture SES differences from the patient up—ideally one that can be used with consistency outside of the HRRP and by programs other than Medicare.
2. Work to ensure that the data required to power such a model is clearly defined and captured/reported consistently.
3. Track the outcomes of this patient population both as part of the overarching performance measure and in its own right.

Having an operational SES definition as we’ve outlined is a necessary first step. What we fear most is that the SES adjustment developed for the HRRP will be used to demonstrate that such an adjustment is of little consequence and not worthy of our attention.

This blog was authored by Richard Fuller, MS, an economist with 3M Clinical and Economic Research, Norbert Goldfield, MD, medical director for 3M Clinical and Economic Research and John Hughes, MD, professor, department of medicine, Yale School of Medicine.

References

1. Thompson MP, Kaplan CM, Cao Y, Bazzoli GJ, Waters TM. Reliability of 30-Day Readmission Measures Used in the Hospital Readmission Reduction Program. Health Serv Res. 2016;51(6):2095-2114. doi:10.1111/1475-6773.12587.
2. Joynt KE, Figueroa JE, Oray J, Jha AK. Opinions on the Hospital Readmission Reduction Program: results of a national survey of hospital leaders. Am J Manag Care. 2016;22(8):e287-94. http://www.ncbi.nlm.nih.gov/pubmed/27556831. Accessed September 19, 2016.
3. Gu Q, Koenig L, Faerberg J, Steinberg CR, Vaz C, Wheatley MP. The Medicare Hospital Readmissions Reduction Program: Potential Unintended Consequences for Hospitals Serving Vulnerable Populations. Health Serv Res. January 2014. doi:10.1111/1475-6773.12150.
4. Hu J, Gonsahn MD, Nerenz DR. Socioeconomic status and readmissions: evidence from an urban teaching hospital. Health Aff (Millwood). 2014;33(5):778-785. doi:10.1377/hlthaff.2013.0816.
5. Atkinson G, Giovanis T. Conceptual errors in the CMS refusal to make socioeconomic adjustments in readmission and other quality measures. J Ambul Care Manage. 37(3):269-272. doi:10.1097/JAC.0000000000000042.