Do safety net hospitals categorically under perform the national average in terms of managing readmissions? Or is something else triggering higher rates of readmissions in these facilities? These questions are essential for policymakers to answer as pay-for-performance (P4P) penalties are having a disparate impact on hospitals that serve low-income areas.
Medicare’s Hospital Readmission Reduction Program (HRRP), for example, links risk-adjusted hospital readmission rates to financial penalties. Hospitals with risk-adjusted readmission rates above the national average are penalized by having their annual Medicare payments reduced by up to 2 percent. In 2015, hospital payments are scheduled to be reduced by up to 3 percent.
But the program’s current system for measuring readmission rates may be flawed. Numerous analyses have found that safety net hospitals, which care for low-income patients, are more than twice as likely to be penalized than hospitals caring for higher-income patients.
HRRP penalties can be substantial and can exacerbate the financial challenges already facing safety net hospitals and lead to unintended consequences, such as hospital closures in areas where few providers operate today. These trends, in turn, could worsen health disparities rather than alleviate them among Medicare beneficiaries who live in low-income areas.
Given the potential for unintended consequences, pay-for-performance programs must be carefully designed to ensure at a minimum that health disparities are not exacerbated. Programs should accurately measure the performance of providers for whom payments are put at risk while controlling for factors that are outside a provider’s sphere of influence. In the case of hospital readmission rates, hospitals and other stakeholders argue that the scope of the quality measure (rate of readmissions 30 days after hospital discharges) is too broad to isolate the effect of a hospital’s role because many significant social and demographic factors come into play after discharge.
Risk adjustment is one way to statistically control for factors that lie beyond a provider’s scope of care. HRRP currently applies a risk adjustment formula that includes factors such as a patient’s gender, age and health status, but does not include what are referred to as social determinants of health that are captured in a patient’s socioeconomic status (SES) with factors such as income and education.
National Quality Forum Deliberations
In November 2013, the National Quality Forum (NQF) convened an expert panel at the request of the Centers for Medicare and Medicaid Services (CMS) to evaluate whether risk adjusting performance measures used in pay-for-performance programs, such as the HRRP, should control for factors related to patient SES in addition to patient health status.
The NQF expert panel’s report, Risk Adjustment for Socioeconomic Status and Other Sociodemographic Factors, recommended that NQF overturn its long-held policy of excluding SES factors and require measure developers to consider patient SES or related factors as warranted by research when designing risk adjustment of quality measures. The panel’s recommendation would apply only to measures used in pay-for-performance programs because they have a financial impact.
However, the NQF Board of Directors did not adopt the panel’s recommendation in whole. The Board voted to further study the implications of including SES in risk adjustment using a trial period. The press release noted: “During the trial period, clinically and sociodemographically adjusted measures may be endorsed by NQF. NQF will develop further details regarding the trial period, including its duration, measure submission requirements, and the evaluation objectives necessary to inform future NQF policy.” The NQF Board also voted to create a standing Disparities Committee.
If the NQF Board had adopted the expert panel recommendation, then each hospital’s readmissions rates used in the HRRP would have to have been adjusted for the SES of their readmitted patients. This change would have had important implications for safety net providers. Including SES factors in the hospital readmission risk adjustment formula would have narrowed the observed differences between safety net hospitals and non-safety hospitals.3 And as a result, it may have reduced the number of safety net hospitals facing a financial penalty and the size of the penalty.
Arguments for Including SES in Risk Adjustment
As noted in a recent Health Affairs Blog post by Billy Wynne, the NQF expert panel’s recommendation to adjust performance measures by SES is receiving attention from Congress on a bipartisan basis. Senators Joe Manchin (D-WV), Bill Nelson (D-FL), Mark Kirk (R-IL), and Roger Wicker (R-MS) introduced S.2501, “Hospital Readmissions Program Accuracy and Accountability Act.”
The bill calls on the Department of Health and Human Services (HHS) to include SES factors such as poverty and education in the HRRP risk adjustment formula starting in 2017. It also allows the Secretary to consider alternative proposals such as peer comparisons. In the House, Representative Jim Renacci (R-OH) introduced the “Establishing Beneficiary Equity in the Hospital Readmission Program Act” (H.R. 4188), which calls for readmission rates to be adjusted by a hospital’s proportion of patients eligible for both Medicare and Medicaid, which could be considered a proxy for SES.
There are compelling reasons to include SES factors in the hospital risk adjustment model. Studies have found that hospital readmissions are largely explained by patient-level and community-level factors that are out of the hospital’s control. For example, a recent study found that Medicare beneficiaries enrolled in both Medicare and Medicaid are more likely to be readmitted than non-duals. Another study found that Medicare beneficiaries living alone or with limited education are also more likely to be readmitted.
It is important to acknowledge that a growing evidence base also finds that some hospital interventions are effective at reducing readmission rates. However, these interventions all require resources and technical assistance – both of which safety net hospitals will need to succeed in a value-based payment environment. Policymakers are beginning to worry that pay-for-performance penalties alone will exacerbate financial strains on safety net providers and as a result will be unlikely to move the needle on quality improvement.
Arguments Against Including SES in Risk Adjustment
There are long-held views for not including SES in risk adjustment models applied to quality measures. The two main arguments are that adjusting for SES in outcome measures like readmissions rates would hold hospitals to different standards and adjusting for SES could “obscure differences in quality among providers that are important to identify if we want to reduce disparities – and also diminish the incentives to improve such disparities,” according to the Frequently Asked Questions posted by CMS on its health care quality data reporting website QualityNet. We address these points in more detail below.
First, the idea of holding hospitals to different standards is certainly not desirable. For some measures, such as hospital acquired infections or the use of a pre-surgical checklist, there is no need to consider risk adjustment. Either the hospital took appropriate steps to prevent an infection (like hand washing) or not; either a surgeon used a pre-surgical checklist or not. In both of these cases, the outcome is clearly under the control of the provider – patient characteristics are not relevant.
However, other measures, such as hospital readmission rates are influenced not only by what the hospital does, but also the patient’s home, family availability, use of outpatient care, use of community resources, and health literacy. The standard of care delivered may be the same or comparable while the outcome on readmissions differs in safety net hospitals because of the challenges created by the social circumstances of their patients once they leave the facility.
Second, identifying quality differences among providers is essential if we are going to address disparities in care. In the context of HRRP, one Health Affairs study found that including SES factors in the hospital readmission risk adjustment formula may narrow the observed differences between safety net hospitals and non-safety hospitals. While some may see this as obscuring differences between providers, what it actually means is that there is less variability in provider quality when providers are compared on the same hypothetical mix of patients. Providers do not all have the same set of patients. Therefore, they should be measured and rewarded (or penalized) on what they do, not because of who they serve.
Congressional staff and other policy advisors working on this issue will discover that the debate over what factors to control for in risk adjusting quality measures is not new. The reason for the renewed attention is that as value-based payment programs take effect, their real-world consequences are coming into view. For now, NQF and CMS have approached the debate over adjustment for SES factors in P4P programs as largely a technical matter. The decision to further study the impact of SES adjustment before making a significant change to methods of constructing quality measures reflects an abundance of caution and the difficulty of changing long-held methodology overnight. However, while the trial period begins, the impact of HRRP on safety net hospitals will continue and could widen the gap between providers.
It is clear that Congress must give CMS some policy direction if SES adjustments are to be made in time to help safety net providers. At its core, the issue of risk adjustment of quality measures used in pay-for-performance should be one of public policy and not merely a technicality of measure development to be worked out by CMS or NQF. Specifically, Congress should consider two questions: Are P4P programs that do not adjust for SES equitable and are P4P programs achieving the policy results of improving quality for all populations and reducing costs for the federal government as Congress intended?
While legislative action on the issue is unlikely to happen soon, the debate will give Congress time to understand these key policy questions and determine if it is necessary to act to provide guidance to CMS on the matter. Luckily, there is bipartisan interest to build on. As the NQF trial period begins, Congress has several options:
- Understand the impact of HRRP in states and districts. Before deciding to act (or not), assessing the impact of HRRP in states or Congressional districts may be helpful. Some hospitals appear to be disproportionately affected by the exclusion of SES in risk adjustment.
- Pass legislation to adjust for SES factors. If Congress sees the issue as largely a policy matter, the most direct solution is to instruct CMS via law. Congress could include legislation, such as the Manchin bill, in health legislation expected to be considered over the next year, such as the physician payment fix and the post-acute care reform bills.
- Work with CMS through formal and informal communications. Laws are hard to pass with deep divisions between the House and Senate. As this issue has attracted bipartisan interest, members and their staff can weigh in both formally and informally with CMS and NQF to convey their point of view. The committees of jurisdiction could also hold a hearing to better understand the impact of pay-for-performance programs on hospitals and Medicare Advantage plans and to consider other policy solutions, such as requiring CMS to compare hospitals to their historical performance or creating a mechanism to share responsibility and accountability for readmission rates between outpatient and inpatient providers.
In the meantime, CMS could also proactively take steps to help improve outcomes at safety net hospitals. One option would be for the Center for Medicare and Medicaid Innovation (CMMI) to design, test, and help evaluate initiatives that aim to reduce readmissions and improve other outcomes of care for patients who are served by safety net hospitals. CMMI has the tools and resources to test the effect of targeted health interventions in a range of areas (e.g., rural vs. urban) and move the ball forward to find solutions to improve readmissions rates in communities in need. Similarly, as suggested by MedPAC, Quality Improvement Organizations (QIOs) could provide similar technical support services to safety net hospitals.