Readmission rate improvements often attributed to an Affordable Care Act policy fining hospitals with higher-than-expected readmissions may actually be the result of changes in how visits were being classified, according to a recent retrospective study.
Implementation of the Medicare Hospital Readmissions Reduction Program (HRRP) coincided with other Medicare policies that “resulted in many inpatient admissions being reclassified as observation stays,” researchers wrote in JAMA Network Open.
These reclassified observation days were often “clinically indistinguishable from short inpatient admissions,” the researchers wrote, and ultimately represented less than 5% of total hospitalizations for conditions specifically targeted by the program.
Nonetheless, by reviewing and analyzing nearly 9 million hospitalizations among Medicare fee-for-service beneficiaries between 2009 and 2015, the researchers found that more than half of the decline in readmissions for target conditions could be attributed to these observation stay reclassifications.
“Our findings suggest that the program may be underperforming relative to the penalties levied on 93% of hospitals since the inception of the program,” researchers wrote in the journal.
The study’s sample represented 20% of both Medicare inpatient admissions and observation stays during a period before and after the announcement and implementation of HRRP. The researchers looked for changes in 30-day readmissions for HRRP target conditions (heart failure, acute myocardial infarction and pneumonia) and nontarget conditions, both accounting for and excluding observation stays.
The analysis found a 93.4% relative increase in observation stays among initial hospitalizations for the target conditions as well as a 51.1% relative increase in observation stays among the initial hospitalizations for nontarget conditions.
As suggested in prior studies, target condition readmission rates among the sample decreased shortly following HRRP’s announcement and then returned to baseline by the time penalties began to be implemented, the researchers wrote.
When looking at inpatient hospitalizations alone, researchers found a -1.48 percentage point absolute reduction and a -1.13 percentage point absolute reduction in readmission rates for target and nontarget conditions, respectively.
Including observation stays, however, “more than halved” the readmission rate absolute decline for target conditions to -0.66 percentage points while nontarget readmissions’ absolute decline landed at -0.76 percentage points, according to the study.
“Ignoring the growth of observation stays results in a measurement problem for estimating the potential outcomes associated with HRRP,” researchers wrote. “Readmissions associated with these index events—nearly one in five hospitalizations in the Medicare population—have fallen out of the calculation of readmission rates over time in a nonrandom way, introducing bias in longitudinal assessments of the HRRP to date, as well as misclassifying the true performance of hospitals.
“Our results suggest that an increasingly larger share of hospital care will be invisible to quality metrics if shifts in observation stay practices are not accounted for in readmissions algorithms,” they wrote.
In an accompanying invited commentary, Jose Figueroa, M.D., of the Harvard T.H. Chan School of Public Health, and Rishi Wadhera, M.D., of Beth Israel Deaconess Medical Center, described the study as the latest addition to a growing body of evidence that HRRP “has not been an effective policy.”
They pointed to arguments from front-line clinicians that variation in readmission rates can often be explained by socioeconomic circumstances in the surrounding communities that aren’t captured by the CMS’ quality program. Poverty and similar social determinants can drive more frequent readmissions, they wrote, making HRRP “incredibly regressive [by] disproportionately penalizing safety-net hospitals that care for low-income, minoritized and marginalized populations.”
Figueroa and Wadhera also highlighted clinician complaints that HRRP has introduced incentives to avoid readmissions and more often rely on observation visits. Though data on this point are "mixed," they stressed that clinicians are in the best position to determine what is best for their patients and that “pressure from administrators” should be given greater consideration when evaluating HRRP’s impacts.
“At best, the evidence to date suggests that the HRRP has had no meaningful effect on the rate at which patients return to the hospital within 30 days of discharge,” Figueroa and Wadhera wrote in the journal. “At worst, the HRRP has unfairly penalized hospitals caring for the most vulnerable populations in our country and potentially resulted in patient harm. How much more evidence will it take for policymakers to officially end this program?”