Hospital Readmissions Reduction Program: Safety-Net Hospitals Show Improvement, Modifications To Penalty Formula Still Needed

Under the Hospital Readmissions Reduction Program (HRRP) of the Affordable Care Act (ACA), the Centers for Medicare and Medicaid Services (CMS) penalizes hospitals for thirty-day readmission rates that exceed national averages. Each year CMS compares each hospital’s performance on thirty-day readmissions over the most recent three years, adjusting for its case-mix (using a methodology that accounts for patients’ age, sex, and clinical factors such as comorbidities and frailty). Hospitals whose readmission rates exceed those of the average hospital with a comparable case-mix are penalized in proportion to their excess rate of readmissions.

Implemented in 2013, the HRRP targeted thirty-day readmission rates for heart attack, heart failure, and pneumonia. For excess readmissions, the program withheld up to 1 percent of a hospital’s total Medicare reimbursements in the first year, with a graduated increase to 3 percent after two years. The ACA specified the penalty formula and ceilings, the timetable for implementation, and the initial conditions for which readmissions penalties would apply. The ACA also granted CMS authority to expand the program as it deemed appropriate. CMS recently added chronic obstructive pulmonary disease and elective knee and hip replacement surgery and is considering including more—possibly all—conditions.

Since implementation of the HRRP, the rate of hospital readmissions has fallen. ¹ While it is difficult to isolate the effects of the program from other trends, there is some evidence that the HRRP has been successful in reducing thirty-day readmissions, at least at the beginning of the program. ² However, despite the program’s seeming success, a number of observers have expressed serious concerns about the penalty rules, and some have recommended changes to the program. ^3,4

A frequently voiced concern is related to the program’s impact on safety-net hospitals, which serve a relatively high proportion of low-income patients—who have a relatively high probability of readmission. ⁵ A disproportionate number of patients discharged from safety-net hospitals lack access to resources needed at that point, such as social support and primary care. ^6,7 The concern is that this places safety-net hospitals at risk for penalties because of risk factors that are largely outside hospital control. ⁸ As a remedy, some policy advisers have argued that the HRRP case-mix adjustment should include patients’ socioeconomic status. ⁵

To date, most assessments of the HRRP have examined overall changes in readmissions ^1,2 or have identified hospital characteristics associated with penalties. County characteristics—particularly access to care—explained nearly half of the national variation in readmission rates, ⁹ and large hospitals, teaching hospitals, and safety-net hospitals received larger penalties compared to other hospitals. ¹⁰ No studies have examined individual hospitals’ improvement of readmission rates or explored the characteristics of hospitals that were most successful in reducing readmissions. This study examined the success of safety-net hospitals in reducing readmissions under the HRRP, compared to other hospitals in the program.

Study Data And Methods

We obtained thirty-day readmission rates for each of the three initial target conditions from Medicare’s Hospital Compare website for fiscal years 2013 and 2016. ¹¹ These rates are adjusted for each hospital’s case-mix; data are not reported for hospitals with fewer than twenty-five discharges for a particular condition. Under the HRRP, only hospitals reimbursed under the inpatient prospective payment system are subject to penalties. We therefore excluded critical access hospitals, Veterans Affairs hospitals, children’s hospitals, and other specialty hospitals that are not reimbursed under that system.

We measured hospitals’ improvement for each condition as the percentage-point change in the readmission rate between the first and fourth years of the program, for each combination of hospital and condition having a reported readmission rate measured in both years. ¹² We defined safety-net hospitals as the highest quartile of hospitals according to their percentage of patients eligible for Supplemental Security Income, ^5,8 and we used Medicare’s Healthcare Cost Report Information System (HCRIS) to identify those hospitals. ¹³

We used two comparisons to examine improvement in safety-net hospitals relative to that in other hospitals in the inpatient prospective payment system. First, we compared safety-net hospitals with all other hospitals subject to the HRRP. Second, we assumed that safety-net hospitals had relatively high readmission rates compared to other hospitals at the outset of the program and therefore had above-average room for improvement. Hence, in an auxiliary analysis we restricted the observations of non-safety-net hospitals by drawing a subsample of these hospitals from the HCRIS for each condition. The subsamples were of similar size to the sample of safety-net hospitals and were matched to that sample by their readmission rates in the first year of the program (for details about the matching strategy, see the online Appendix). ¹⁴

We conducted $t$ -tests to compare safety-net hospitals and other hospitals, and we compared the two groups’ percentage-point changes in readmission rates for each condition. Subsequently, we used regression analysis to compare percentage-point reductions in readmission rates in safety-net hospitals with reductions in the matched other hospitals. The key independent variable was a binary indicator for safety-net status.

The regression analysis incorporated several adjustments for hospital characteristics that might account for variation in improvement. These characteristics were the following four variables: size (fewer than 200 beds, 200–399 beds, or 400 or more beds), resident-to-bed ratio (highest quintile versus four lowest quintiles), percentage of patients with Medicare (less than 50 percent, or 50 percent or more), and occupancy rate (highest quintile versus four lowest quintiles). We obtained data about these variables from the CMS historical impact file for fiscal year 2013. ¹⁵ Finally, using data from HCRIS for fiscal year 2013, we entered binary variables that measured hospitals’ positive financial margin and for-profit status. We estimated the models with SAS, version 9.4, using PROC GENMOD (for further details on the estimation, see the Appendix). ¹⁴

Study Results

The risk-adjusted readmission rates for all three conditions were higher in safety-net hospitals than in other hospitals in the overall sample in both fiscal years 2013 and 2016 ( Exhibit 1 ). The differences, which ranged from 0.370 percentage points for pneumonia in 2016 to 1.017 percentage points for heart failure in 2013, were all significant. The differences in readmission rates between safety-net and other hospitals declined over the first three years of the HRRP. For example, for heart attack, the percentage-point difference dropped from 0.621 to 0.427.

Improvements in unadjusted readmission rates between fiscal year 2013 and fiscal year 2016 were greater in safety-net hospitals than in other hospitals in the overall sample for each condition ( Exhibit 2 ). For example, readmissions for heart attack fell 2.86 percentage points at safety-net hospitals, compared to 2.64 percentage points at other hospitals.

SOURCE Authors’ analysis based on data for fiscal year 2013 from the Medicare Cost Report and for fiscal years 2013 and 2016 from Medicare Hospital Compare. NOTES The overall sample contains all hospitals in the inpatient prospective payment system. The matched sample contains safety-net hospitals and other hospitals matched to them in terms of readmission rate in fiscal year 2013 (the first year of the program). Significance is based on $t$ -tests of difference between means. ** $p<0.05$ *** $p<0.01$ **** $p<0.001$

But as noted above, compared to other hospitals, safety-net hospitals had higher readmission rates to begin with and so had more room for improvement. Accordingly, we also compared changes in readmission rates in safety-net hospitals with changes in other hospitals in the matched sample (whose initial readmission rates were the same as those of safety-net hospitals). In these comparisons, safety-net hospitals did not perform as well as other hospitals. For example, improvement in readmission rates for heart attack at safety-net hospitals was 2.86 percentage points, compared to 3.20 percentage points at other hospitals ( Exhibit 2 ).

Exhibit 3 presents the results of the regression analysis comparing percentage-point differences in readmission rates between safety-net and other hospitals in the matched sample. When we controlled for hospital characteristics, we found that the differences in improvements between the two hospital types were similar to the unadjusted results in Exhibit 2 : Safety-net hospitals had smaller improvements than other hospitals—for example, an improvement for heart attack that was 0.38 percentage points lower. The associations between the variables we controlled for and differences in improvement did not reveal any consistent patterns.

Discussion

A growing literature suggests that characteristics of a hospital’s patients and of the community in which it is located are key factors in explaining variation in thirty-day readmission rates. ^7,8 Some studies have accounted for patient variables such as socioeconomic status in calculating readmission rates and have found considerable disparities between those rates and CMS’s published readmission rates. ^16,17 One study of 4,073 hospitals found that 58 percent of the national variation in readmission rates was explained by characteristics of the county in which the hospital was located. ⁹

The relevance of a hospital’s patient population to its readmission rates is also apparent in the distribution of HRRP penalties. It is well documented that safety-net hospitals have had larger penalties than other hospitals. ^5,10,18–20 This study contributes to knowledge about the performance of safety-net hospitals under the HRRP by identifying how much their readmission rates have improved. Our results indicate that overall, safety-net hospitals achieved greater reductions in readmission rates between fiscal years 2013 and 2016 than other hospitals did.

Although the differences in improvement were not large, the finding that disparities between safety-net and other hospitals in readmission rates and the ensuing penalties have declined is a positive policy outcome. However, when we compared safety-net hospitals with other hospitals that also had high readmission rates to begin with—and, hence, equally great room for improvement—we found that safety-net hospitals had smaller reductions than other hospitals in their readmission rates. This result may reflect the difficulties safety-net hospitals have in dealing with factors that influence readmission rates but are beyond the hospitals’ control—such as patient homelessness or lack of family support.

Concern over the impact of the HRRP is widespread, and policy makers have recommended various approaches to leveling the playing field for hospitals in the program. One option is formally adjusting the penalty algorithm for patients’ socioeconomic status. CMS adjusts for readmission risk using a methodology that has been endorsed by the National Quality Forum, but the forum is reassessing its long-held policy of excluding socioeconomic status and other demographic factors in risk adjustment of quality measures used for pay-for-performance. ²¹ However, CMS argues that incorporating socioeconomic status into its risk-adjustment methodology would have the effect of lowering the standard of performance for hospitals serving populations with low status.

In light of this concern, the Medicare Payment Advisory Commission (MedPAC) has proposed a refinement that would divide hospitals into deciles based on their share of low-income patients and would assess penalties based on the comparative performance of hospitals within the same decile. ⁵ This modification would avoid evaluating safety-net hospitals on the same basis as hospitals that serve populations with higher socioeconomic status, without altering the risk-adjustment methodology.

In 2014 a bill was introduced in the Senate calling for easing penalties for safety-net hospitals, leaving CMS to decide exactly how to do so. ²² More recently, the House of Representatives passed the Helping Hospitals Improve Patient Care Act of 2016, ²³ which instructs CMS to use a measure similar to what MedPAC recommended to account for socioeconomic status in the near term. Under the provisions of this bill, in applying HRRP adjustments, CMS would compare the performances of hospitals with similar proportions of dual-eligible patients (those eligible for both Medicare and Medicaid). ²⁴ In the future, when data required under the Improving Medicare Post-Acute Care Transformation Act (IMPACT) of 2014 become available, CMS would adjust the program according to updated research findings.

In refining the HRRP, policy makers should bear in mind that a penalty program may not provide the best lever for incentivizing performance improvement in safety-net hospitals. Hospitals face mixed incentives, and it is reasonable to assume that for some hospitals, the HRRP’s financial penalties may not provide sufficient motivation for reducing readmission rates. ⁴ While the HRRP penalizes hospitals whose readmission rates exceed national averages, under the inpatient prospective payment system, hospitals are reimbursed for all readmitted patients.

In addition, implementing programs to reduce readmissions have associated costs, which may be particularly burdensome for safety-net hospitals. These include devoting staff time to analyzing data and processes and implementing changes, and perhaps the acquisition and enabling of new technologies. Hospitals’ budgets are limited, and administrators recognize that allocating additional resources to reducing readmission rates requires forgoing alternative efforts to improve quality or other uses of those resources that they may view as more valuable.

Some administrators may conclude that avoiding the possible penalties of the HRRP is not worthwhile. ²⁵ They also are aware that while they incur 100 percent of the penalties, they bear only partial responsibility for readmissions; other providers—particularly physicians and nursing homes—share that responsibility.

Conclusion

This study addressed the question of how safety-net hospitals are performing under the HRRP, relative to other hospitals subject to the program. The main findings are that safety-net hospitals have considerably improved their readmission rates and that the disparity in rates between hospitals serving high shares of poor patients and those serving low shares of such patients has decreased. Still, reducing readmissions may be more challenging for safety-net hospitals than for other hospitals: Safety-net institutions have not improved as much as other hospitals that had initial high readmission rates. Our results support the approach recommended by MedPAC as a way of modifying the amount of penalties imposed on safety-net hospitals: evaluating them against other safety-net hospitals. This would maintain incentives for improvement but would reduce financial pressure on safety-net hospitals, especially for those with patient populations of the lowest socioeconomic status.

This study was an early investigation of the issue, based on readmission rates in the initial years of the HRRP, and the improvements we observed may reflect the program’s success in picking low-hanging fruit. It will be important to continue to monitor the performance of safety-net hospitals under the HRRP. If these hospitals fail to respond to HRRP incentives in the future, CMS might consider using different approaches to reducing the hospitals’ readmission rates—such as assessing the rates against the hospitals’ own historical record or exempting the hospitals from the HRRP altogether and focusing on quality improvement initiatives for them instead.

Reducing hospital readmissions provides a broad-scale opportunity for improving value in the delivery of expensive inpatient care. Yet CMS faces considerable challenges in designing an incentive program that will accomplish that goal without triggering adverse effects. It would be advisable for CMS to pay attention to characteristics of hospitals that succeed in reducing readmissions as it modifies and expands the HRRP.

ACKNOWLEDGMENTS

This work was funded in part by the Agency for Healthcare Research and Quality (Grant No. R03 HS024853-01; Kathleen Carey, principal investigator). [ Published online September 21, 2016. ]

NOTES

• 1 Medicare Payment Advisory Commission . Report to the Congress: Medicare payment policy [Internet]. Washington (DC) : MedPAC ; 2015 Mar [cited 2016 Aug 31 ]. Available from: http://medpac.gov/docs/default-source/reports/mar2015_entirereport_revised.pdf Google Scholar
• 2 Carey K , Lin M-Y . Readmissions to New York hospitals fell for three target conditions from 2008 to 2012, consistent with Medicare goals . Health Aff (Millwood) . 2015 ; 34 ( 6 ): 978 – 85 . Go to the article, Google Scholar
• 3 Glass D , Lisk C , Stensland J . Refining the Hospital Readmissions Reduction Program [Internet]. Washington (DC) : Medicare Payment Advisory Commission ; 2012 Sep 7 [cited 2016 Sep 6 ]. Available from: http://medpac.gov/docs/default-source/meeting-materials/september-2012-meeting-presentation-refining-the-hospital-readmissions-reduction-program.pdf?sfvrsn=0 Google Scholar
• 4 Berenson RA , Paulus RA , Kalman NA . Medicare’s readmissions-reduction program—a positive alternative . N Engl J Med . 2012 ; 366 ( 15 ): 1364 – 6 . Crossref, Medline, Google Scholar
• 5 Medicare Payment Advisory Commission . Report to the Congress: Medicare and the health care delivery system [Internet]. Washington (DC) : MedPAC ; 2013 Jun [cited 2016 Aug 31 ]. Available from: http://medpac.gov/docs/default-source/reports/jun13_entirereport.pdf Google Scholar
• 6 Hu J , Gonsahn MD , Nerenz DR . Socioeconomic status and readmissions: evidence from an urban teaching hospital . Health Aff (Millwood) . 2014 ; 33 ( 5 ): 778 – 85 . Go to the article, Google Scholar
• 7 Singh S , Lin YL , Kuo YF , Nattinger AB , Goodwin JS . Variation in the risk of readmission among hospitals: the relative contribution of patient, hospital, and inpatient provider characteristics . J Gen Intern Med . 2014 ; 29 ( 4 ): 572 – 8 . Crossref, Medline, Google Scholar
• 8 Joynt KE , Jha AK . A path forward on Medicare readmissions . N Engl J Med . 2013 ; 368 ( 13 ): 1175 – 7 . Crossref, Medline, Google Scholar
• 9 Herrin J , St. Andre J , Howard K , Joshi MS , Audet AJ , Hines SC . Community factors and hospital readmission rates . Health Serv Res . 2015 ; 50 ( 1 ): 20 – 39 . Crossref, Medline, Google Scholar
• 10 Joynt KE , Jha AK . Characteristics of hospitals receiving penalties under the Hospital Readmissions Reduction Program . JAMA . 2013 ; 309 ( 4 ): 342 – 3 . Crossref, Medline, Google Scholar
• 11 Data.Medicare.gov . Hospital Compare datasets [Internet]. Baltimore (MD) : Centers for Medicare and Medicaid Services ; [cited 2016 Aug 31 ]. Available from: https://data.medicare.gov/data/hospital-compare Google Scholar
• 12 Both Hospital Compare and the HRRP are based on moving averages from the same three-year reporting period: The first year of the HRRP (fiscal year 2013) is based on the reporting period July 1, 2008–June 30, 2011, and the fourth year (fiscal year 2016) is based on the reporting period July 1, 2011–June 30, 2014.
• 13 CMS.gov . Cost reports [Internet]. Baltimore (MD) : Centers for Medicare and Medicaid Services ; [last modified 2016 Jul 27; cited 2016 Aug 31 ]. Available from: https://www.cms.gov/Research-Statistics-Data-and-Systems/Downloadable-Public-Use-Files/Cost-Reports/ Google Scholar
• 14 To access the Appendix, click on the Appendix link in the box to the right of the article online.
• 15 CMS.gov . Historical impact files for FY 1994 through present [Internet]. Baltimore (MD) : Centers for Medicare and Medicaid Services ; [last modified 2012 Aug 1; cited 2016 Aug 31 ]. Available from: https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/AcuteInpatientPPS/Historical-Impact-Files-for-FY-1994-through-Present.html Google Scholar
• 16 Nagasako EM , Reidhead M , Waterman B , Dunagan WC . Adding socioeconomic data to hospital readmissions calculations may produce more useful results . Health Aff (Millwood) . 2014 ; 33 ( 5 ): 786 – 91 . Go to the article, Google Scholar
• 17 Barnett ML , Hsu J , McWilliams JM . Patient characteristics and differences in hospital readmission rates . JAMA Intern Med . 2015 ; 175 ( 11 ): 1803 – 12 . Crossref, Medline, Google Scholar
• 18 Gilman M , Adams EK , Hockenberry J , Wilson IN , Milstein AS , Becker ER . California safety-net hospitals likely to be penalized by ACA value, readmission, and meaningful-use programs . Health Aff (Millwood) . 2014 ; 33 ( 8 ): 1314 – 22 . Go to the article, Google Scholar
• 19 American Hospital Association . Rethinking the Hospital Readmissions Reduction Program [Internet]. Washington (DC) : AHA ; 2015 Mar [cited 2016 Aug 31 ]. (TrendWatch). Available from: http://www.aha.org/research/reports/tw/15mar-tw-readmissions.pdf Google Scholar
• 20 Gilman M , Hockenberry JM , Adams EK , Milstein AS , Wilson IB , Becker ER . The financial effect of value-based purchasing and the Hospital Readmissions Reduction Program on safety-net hospitals in 2014: a cohort study . Ann Intern Med . 2015 ; 163 ( 6 ): 427 – 36 . Crossref, Medline, Google Scholar
• 21 National Quality Forum . Finding the right adjustment [Internet]. Washington (DC) : NQF ; c 2016 [cited 2016 Aug 31 ]. Available from: http://www.qualityforum.org/Finding_The_Right_Adjustment.aspx Google Scholar
• 22 Congress.gov . S.2501—Hospital Readmissions Program Accuracy and Accountability Act of 2014 [Internet]. Washington (DC) : Congress ; [cited 2016 Sep 6 ]. Available from: https://www.congress.gov/bill/113th-congress/senate-bill/2501/text Google Scholar
• 23 Congress.gov . H.R.5273—Helping Hospitals Improve Patient Care Act of 2016 [Internet]. Washington (DC) : Congress ; [cited 2016 Aug 31 ]. Available from: https://www.congress.gov/bill/114th-congress/house-bill/5273 Google Scholar
• 24 House Ways and Means Committee . The Helping Hospitals Improve Patient Care Act of 2016 (H.R. 5273): summary [Internet]. Washington (DC) : House of Representatives ; [cited 2016 Aug 31 ]. Available from: http://tiberi.house.gov/uploadedfiles/hipc_summary_final.pdf Google Scholar
• 25 Boutwell A . Time to get serious about hospital readmissions . Health Affairs Blog [blog on the Internet]. 2012 Oct 10 [cited 2016 Aug 31 ]. Available from: http://healthaffairs.org/blog/2012/10/10/time-to-get-serious-about-hospital-readmissions/ Google Scholar