The Logic Was Irresistible
Hospital readmissions cost Medicare more than $17 billion per year. Many seemed avoidable. A patient with heart failure gets discharged, deteriorates at home, and bounces back three weeks later. Why not penalize hospitals for letting that happen?
The reasoning was intuitive and bipartisan. In 2010, the Affordable Care Act created the Hospital Readmissions Reduction Program, which required the Centers for Medicare & Medicaid Services to impose financial penalties on hospitals whose 30-day readmission rates for heart failure, acute myocardial infarction, and pneumonia exceeded expected levels. Penalties began in October 2012, capping at 3% of a hospital's total Medicare base DRG payments. By 2018, CMS had collected nearly $2 billion in fines. Two-thirds of American hospitals had been penalized at least once.
Readmission rates fell. The program looked like a success story, cited in policy circles as evidence that financial incentives could discipline healthcare delivery. Then researchers at Harvard checked whether the patients were still alive.
What 8 Million Hospitalizations Revealed
Rishi Wadhera and colleagues at Beth Israel Deaconess Medical Center analyzed 8.3 million Medicare fee-for-service hospitalizations from April 2005 through March 2015 — the decade straddling the HRRP's announcement and implementation. Among the 7.9 million patients discharged alive (mean age 79.6 years, 53.4% women), they tracked what happened in the 30 days after going home: readmission, death, or both.
The study design was careful. Four time periods of equal length allowed comparison of pre-HRRP mortality trends against post-policy changes. Inverse probability weighting controlled for shifting case mix between periods. The team examined three conditions separately: heart failure (3.2 million hospitalizations), acute myocardial infarction (1.8 million), and pneumonia (3.0 million).
The findings split cleanly. For heart failure, post-discharge mortality had already been rising slowly before the HRRP at 0.27 percentage points per period. After implementation, the rise accelerated: an additional 0.25 percentage-point jump beyond the baseline trend (P = .001). For pneumonia, the picture was worse. Mortality had been stable before the policy. After HRRP implementation, it began climbing — a 0.40 percentage-point increase above the pre-existing flat trend (P < .001). For acute myocardial infarction, there was no significant change.
The most troubling detail lay in the subgroup analysis. The increase in deaths was concentrated among patients who were not readmitted. They went home and died there. The proportion of patients who died without being readmitted grew significantly for both heart failure and pneumonia after the HRRP took effect — meaning the policy's core mechanism of discouraging readmissions may have kept dying patients out of the hospitals that could have saved them.
An Independent Team Got the Same Answer
Seven months before Wadhera's JAMA paper, Ankur Gupta, Gregg Fonarow, and nine co-authors published a separate analysis in JAMA Cardiology using the American Heart Association's Get With The Guidelines-Heart Failure registry — a prospectively collected clinical dataset with far richer risk adjustment than the administrative claims data most HRRP studies rely on (DOI: 10.1001/jamacardio.2017.4265).
Across 115,245 Medicare beneficiaries at 416 hospitals from 2006 to 2014, the pattern was stark. Thirty-day readmissions fell from 20.0% to 18.4%. Thirty-day mortality rose from 7.2% to 8.6% — a hazard ratio of 1.18 (95% CI: 1.10–1.27, P < .001). One-year mortality climbed from 31.3% to 36.3%. As Fonarow put it at the American Heart Association Scientific Sessions: "It was really the worst of our fears realized." The declining mortality trend that had characterized heart failure care for over a decade reversed after the HRRP began.
How Good Intentions Create Perverse Incentives
The mechanism is not mysterious. When you fine hospitals for readmitting patients, hospitals find ways to avoid readmissions. Some strategies genuinely improve care: better discharge planning, follow-up phone calls, transitional care nurses. But others involve keeping patients out of the hospital who need to be there.
Emergency departments developed triage protocols to avoid admitting returning patients. Observation stays — which don't count as admissions — surged after the HRRP. A 2019 analysis by Wadhera and colleagues in the BMJ confirmed that total hospital revisits (readmissions plus observation stays) declined less than readmissions alone, suggesting hospitals were reclassifying visits rather than preventing them. Safety-net hospitals and academic medical centers that serve sicker, poorer patient populations were disproportionately penalized, draining resources from the institutions that needed them most.
A 2018 review in the European Journal of Heart Failure cataloged the gaming: observation stays increased, some readmissions were delayed past the 30-day window, and emergency department triage shifted to avoid formal admission. The achieved reduction in heart failure readmissions was roughly 9%, far below the 25% target, and some of that reduction was attributable to administrative upcoding rather than actual clinical improvement (DOI: 10.1002/ejhf.1212).
The Strongest Case Against Causation
The most credible challenge comes from Teryl Nuckols and colleagues at Cedars-Sinai, who compared mortality trends at hospitals that received HRRP penalties with those that did not. Their reasoning was sharp: if the HRRP caused the mortality increase, you would expect it to be larger at penalized hospitals, since those hospitals faced the strongest pressure to change behavior. Instead, Nuckols found that mortality rose at nearly identical rates at penalized and non-penalized institutions.
This is a serious finding. It suggests the mortality increase may reflect a secular trend — perhaps increasing patient acuity, changes in coding practices, or an aging population — rather than a direct consequence of HRRP penalties. If something unrelated to the policy was driving deaths upward everywhere, the HRRP would be guilty by temporal association rather than causation.
The Wadhera team acknowledged this limitation directly. Their JAMA paper noted that "whether this finding is a result of the policy requires further research." And their 45-day post-admission analysis — which included in-hospital deaths and a longer observation window — did not show a significant mortality increase, adding another layer of ambiguity to the causal story.
What We Didn't Prove
This is an observational study, not a randomized trial. The HRRP was implemented nationally, which means there is no true control group — every hospital in the country was simultaneously exposed to the same policy. The study uses temporal trends and statistical modeling to infer an association, but confounding is possible. Patient severity may have increased over time in ways that administrative data cannot fully capture.
The Nuckols counter-evidence genuinely weakens the causal claim. If penalized and non-penalized hospitals showed identical mortality trends, the policy's specific incentive mechanism may not be the primary driver.
The 45-day post-admission analysis yielding no significant increase complicates interpretation. It could mean the HRRP shifted where patients died (at home rather than in hospital) without changing total mortality. Or it could mean the 30-day post-discharge window captures noise rather than signal.
Finally, both Wadhera and Gupta used data ending in 2014–2015. The HRRP has been modified since, and hospitals may have adapted in ways that altered the mortality trajectory. Whether the pattern persists into the 2020s remains an open question complicated by the COVID-19 pandemic's massive disruption to hospital care patterns.
The Bottom Line
The HRRP reduced readmissions. It may also have increased deaths. Across two independent analyses comprising over 8 million hospitalizations and more than 600 hospitals, implementation of the readmission penalty coincided with a significant reversal in a decade-long trend of declining heart failure mortality. The excess deaths were concentrated among patients who stayed home instead of returning to the hospital.
Wadhera, Yeh, and Joynt Maddox called for a "reboot" of the HRRP in a 2019 New England Journal of Medicine perspective, arguing the program needed to incorporate mortality as a balancing measure alongside readmissions. As of 2026, CMS still does not penalize hospitals for post-discharge deaths the way it penalizes them for readmissions.
Here is an original calculation that captures the asymmetry: at current penalty rates, a hospital loses up to 3% of its Medicare DRG payments for high readmission rates. It loses exactly 0% for high post-discharge mortality. The financial incentive is structurally one-sided, and the 8-million-patient dataset suggests patients paid the difference.
What You Can Do
If you or a family member is discharged from the hospital with heart failure or pneumonia, do not interpret "staying out of the hospital" as a goal in itself. Worsening symptoms — increased shortness of breath, sudden weight gain from fluid retention, confusion, persistent fever — warrant emergency care regardless of how recently you were discharged.
Ask about transitional care programs before discharge. Hospitals that reduced readmissions without increasing mortality generally did so through nurse-led follow-up visits, telehealth monitoring, and structured medication reconciliation rather than through triage avoidance.
Know the difference between observation status and admission. If you return to the hospital and are placed in "observation" rather than admitted, this affects your Medicare coverage for subsequent skilled nursing care. Ask your care team explicitly whether you are being admitted or observed, and request admission if you believe your condition warrants it.