DO FINANCIAL PENALTIES REDUCE READMISSION RATES? A CONTINUOUS TREATMENT EFFECT EVALUATION

Readmission rates have declined for Medicare fee-for-service beneficiaries since the introduction of financial penalties for providers with higher-than-expected rates under the Hospital Readmission Reduction Program (HRRP) in 2012. The HRRP has been controversial with safety-net hospitals attracting larger penalties than other providers, but no controlled study of this effect has been undertaken to date.

Similar policies introduced in the English National Health Service (NHS) may provide some evidence on the impact of the HRRP penalty. English hospitals have been penalized since 2011 for readmissions above an a priori agreed threshold, and unlike in the US, the threshold varies between local health care economies. Each threshold was determined by a review, involving commissioners and providers, taking into account previous hospital performance and patient-case mix. In the 2012 fiscal year, providers were not reimbursed for an average of 22% of readmissions. A particular feature of the NHS policy is that any occurring penalty must be reinvested in services that aim at reducing readmission rates in the future. It has been shown that local health care organizations chose to invest predominately in services addressing the immediate post discharge period, such as acute post discharge services (33%), community care (26%), intermediate care (12%), and home re-ablement (4%).

This is the first study to evaluate the effectiveness of the English penalization policy and uses generalized propensity scores (GPS) to construct dose-response functions. English Hospital Episode Statistics (HES) data was aggregated at provider level and linked to financial accounts data. The GPS was defined as the conditional density of the treatment given observed covariates such as area-level prevalence of chronic conditions, patient case-mix, and historic readmission levels, covering a six-year period prior to the policy implementation. The balancing properties were satisfied allowing an unbiased estimation of the dose-response function under the assumption of weak confoundedness. The outcome, 30-day emergency readmission rate, was risk adjusted using the shrinkage estimator method used by the Centers for Medicare & Medicaid Services (CMS). 

The analysis showed no statistically significant marginal treatment effect of increasing financial penalties on adjusted 30-day emergency readmission rates. Additional sensitivity analysis looked at treatment effects on readmission rates for patients following surgical admissions, as well as readmissions following an index admission of myocardial infarction, pneumonia, and stroke.  Possible reasons for the failure may relate to the coexistence of policies that also aim at improving provider efficiencies by using financial incentives, with the penalty effect neutralized through extra income generated elsewhere. Compared to a penalty size of about 5% in the US, English penalties are significantly larger and could therefore lead to funding gaps that may result in deterioration of quality of care. Based on current findings, US policymakers should be careful when interpreting reductions in readmission rates as a result of penalization policies.