Could Medicare Readmission Policy Exacerbate Health Care System Inequity? FREE

The Centers for Medicare & Medicaid Services recently started publicly reporting readmission rates.

Options for health care reform would create differentiated hospital payments based on readmission rates.

Hospital performance is one of many factors that determine readmission.

Current models and reform discussions largely omit the contribution of demographic, socioeconomic, and community variables.

Demographic, socioeconomic, and community variables may play a substantial role in determining readmission in hospitals serving vulnerable communities.

Generalizable techniques to manage care cost-effectively are still evolving.

Evaluating hospitals that serve challenged communities on the basis of readmission rates and imposing corresponding financial penalties may amplify inequity in the health care system.

Join the dialogue on health care reform. Comment on the perspectives published in Annals and offer ideas of your own. All thoughtful voices should be heard.

The Centers for Medicare & Medicaid Services (CMS) began publicly reporting hospital readmission rates in July 2009 (1), and proposals to reduce hospital readmission have gained prominence in health care reform discussions. A recent article by Jencks and colleagues (2) showed that 19.6% of Medicare beneficiaries were readmitted within 30 days of hospital discharge. Nearly $17.4 billion could be saved annually by preventing readmissions. If hospitals were evaluated and differentiated payments were generated on the basis of a 30-day period (beginning at discharge), a major step toward reversing financial incentives, which now paradoxically reward hospitals for readmission and discourage care coordination, would be made.

Although the possible salutary effects of readmission-related policies have dominated current discourse, little attention has been given to the potential for unintended consequences on hospitals serving vulnerable communities. In this article, we offer the perspective of a health care provider serving 1 of the poorest counties in the United States. We argue that the potential for proposed hospital readmission policies to create further inequity in the health care system is real. If students in inner-city school districts scored lower than those in affluent suburban communities on standardized tests, it would be implausible to reduce funding to inner-city schools as a result. Yet, if hastily implemented, readmission-related proposals may be paving this path in health care.

Readmission rates are now publicly reported as part of the CMS Reporting Hospital and Quality Data Annual Payment Update program. The 43 hospital quality measures reported to date have encompassed effectiveness of care, patient safety, and patient-centeredness. Inclusion of readmission rates allows the CMS to add measures of hospital efficiency and care coordination as part of its overall value-based purchasing efforts (3). The 30-day readmission rates that the CMS publicly reports include acute myocardial infarction, congestive heart failure, and pneumonia.

Executive and congressional health care reform proposals from 2009 have relied on readmission provisions to catalyze payment system reform, and such proposals have even been supported by prominent media sources (4 - 6). A U.S. Senate Finance Committee policy option called for withholding up to 20% of a hospital's inpatient payments on the basis of comparative readmission rates (4). Original U.S. House of Representatives options called for penalties of up to 5% of hospital payments for facilities with higher-than-expected readmission rates (5). Readmission-based reimbursement would serve as a precursor to “bundled payments,” whereby hospitals, physicians, skilled nursing facilities, home care agencies, and other providers would receive a single payment under the auspices of an accountable care organization (7), and all would be expected to work together to effectively manage care transitions.

Consider 2 patients. Patient A, a man aged 70 years, is a Medicare beneficiary with hypertension, diabetes, and congestive heart failure. He is an affluent retiree with a doctoral degree; he lives with his wife, plays golf, drives to appointments to see his primary care physician (of 10 years), and has ample coverage for prescription drugs. Patient B, a man aged 70 years, is also a Medicare beneficiary with hypertension, diabetes, and congestive heart failure. However, patient B is of modest means; he did not complete high school, lives alone in urban housing, speaks limited English, relies on his working children to make medical appointments at a clinic, and has gaps in his prescription drug coverage. If a hospital discharged both of these patients with congestive heart failure, which one would be more likely to be readmitted within 30 days?

Although patients A and B are medically “identical,” the current CMS readmission measurement model does not account well for the important differences in the 2 patients' situations. Demographic variables, such as race, ethnicity, and preferred language, and socioeconomic variables (for example, education level and income) are not included (8). Although hospital readmission rates are adjusted for coded comorbid conditions, clinical factors, such as degree of left ventricular or kidney function, are excluded. The all-cause method also does not exclude readmissions for unrelated conditions—a patient discharged with pneumonia and readmitted 3 weeks later after a motor vehicle accident would count in the hospital's pneumonia readmission rate.

Although the socioeconomic circumstances for patients A and B contrast anecdotally, the effect of these factors has been shown to predict health care utilization and outcomes. Ross and coworkers (9) completed a systematic review of 117 studies that included models to assess the relationship between patient characteristics and heart failure readmission rates. Among models in which sociodemographic variables were included, several studies were identified in which race, ethnicity, living status, insurance, and income were statistically significantly associated with heart failure readmission.

Rathore and colleagues (10) analyzed a national sample of records of Medicare beneficiaries with heart failure to evaluate the relationship among socioeconomic status, quality of inpatient care, and outcomes. They used a composite socioeconomic status measure derived from community data and concluded that socioeconomic status was independently and significantly associated with higher 1-year readmission and mortality rates. A recent statistical brief from the Agency for Healthcare Research and Quality evaluated the national frequency and costs of preventable hospitalizations (11). Rates of hospitalization for all 12 adult medical conditions evaluated were higher for patients residing in the lowest-income communities than for those living in the highest-income communities.

With respect to education level and its effect on health, a recent report from the Robert Wood Johnson Foundation assessed adult health in the United States. Compared with the most educated adults, the least educated adults in every state were more likely to be in less than very good health. (12). The Institute of Medicine's report on racial and ethnic disparities summarized more than 2 decades of work documenting disparities in health care outcomes among Hispanic/Latino, African-American/black, and American-Indian populations (13). These gaps persist, and the potential for “pay-for-performance programs” to contribute to them has been noted (14 - 15).

Publicly reported quality measures already show the effect of sociodemographic variables. Hospitals report patient satisfaction scores to the CMS by using the Hospital Consumer Assessment of Health Care Providers and Systems Survey. Publicly reported scores are adjusted to account for method of survey administration and patient characteristics. In the March 2009 public report of data, such characteristics as patient-designated education level, health status, and primary language other than English, all required adjustment of satisfaction scores (16).

Regional variation in health care spending and utilization has received much attention recently (17). Analyses of hospital readmission data have revealed variation in readmission rates at state and local levels (2,18). Although such findings can be viewed as indicative of suboptimal hospital performance, they may also indicate that demographic and socioeconomic factors, health care service provider accessibility, and community characteristics produce variability in hospital utilization, in addition to patient medical characteristics and hospital performance. Access to primary care; subacute rehabilitation; skilled nursing; home care; and services unrelated to health care, such as transportation, can determine hospital utilization rates. Comparison of recently released readmission data from hospitals in different communities illustrates this finding (Appendix Table).

Publicly reported quality measures released by the CMS for congestive heart failure care include comprehensive discharge instruction and readmission rates (1). The county our institution serves, Bronx, New York, is a highly diverse urban community and 1 of the poorest counties in the United States. In contrast, some of the most affluent suburban counties in the United States are located within a 30-minute commute. For the provision of heart failure discharge instructions, a quality variable that is fully under a hospital's control, 6 of 7 Bronx county hospitals reporting data perform more favorably than the national average. In contrast, only 3 of 6 facilities in a neighboring affluent county do. Heart failure mortality rates are also similar. However, for corresponding heart failure readmission rates, in which socioeconomic and community factors exert a significant influence, Bronx hospitals do not perform as well as the same neighboring affluent county.

A central tenet of basing future hospital payments on readmission rates is that hospitals and postdischarge providers will successfully work together to reduce hospital readmission. This introduces questions about which interventions they would use to accomplish this. Quality improvement rests on the application of established evidence-based practices. In contrast, the state of the science on effective care management methods is evolving, and the generalizability of approaches in varied patient populations remains to be seen. In such areas as heart failure care, in which effective care management methods have been identified, it is unclear that overall spending is reduced (19).

Peikes and colleagues (20) recently evaluated the effectiveness of the CMS Coordinated Care Demonstration. Fifteen care coordination programs across diverse settings used different techniques to manage chronic conditions, including congestive heart failure, coronary artery disease, and diabetes. Thirteen showed no significant difference in hospitalizations, and none of the 15 yielded net savings. These results from motivated and funded care coordination entities suggest the difficulties that nascent coalitions of hospitals and disparate postdischarge providers will confront in developing “home-grown” programs to prevent readmission.

Preventable hospital readmissions present an opportunity to improve care coordination and achieve savings. However, current approaches omit consideration of important variables that determine readmission. The addition of financial penalties for hospitals with higher readmission rates treating vulnerable populations in challenged communities amplifies these policy shortcomings. Hospitals' disproportionate share funding was intended to remedy these challenges. Eliminating this funding and reducing payments based on readmission rates may create financial “double jeopardy” and paradoxically worsen care coordination and exacerbate health disparities in such communities.

Given this deleterious potential, the CMS should not broadly apply readmission policies until specific policy remedies are incorporated. Readmission measurement models should include additional factors, such as patient sociodemographic and community characteristics. The CMS could leverage disproportionate share funding methods, census and national health survey data, and the data it already uses to adjust federal patient satisfaction surveys to account for these factors. If hospital discharge performance is the primary variable of interest, readmission for unrelated conditions should be excluded from calculation of rates, and a shorter readmission window, such as 7 days, should be used.

Hospitals should be judged on the basis of their own expected performance rather than penalized for having higher readmission rates than those of other facilities. The relationships among patient sociodemographic variables, community characteristics, and readmission rates should be studied to promote fair hospital comparison and further advance the evidence base on care management. Critical evaluation of demonstration projects, such as the recently launched CMS Care Transitions Project, remains essential in determining the effectiveness of care management efforts in varied patient populations.

Informed readmission policy can be a transformational force in promoting coordinated and patient-centered care. However, gains in health care system efficiency should not come at the expense of equity. A hasty one-size-fits-all approach is unlikely to work in formulating details for the most important health care reform in a generation.