Socioeconomic factors may make Medicare's hospital readmissions data more useful

A new study shows that if socioeconomic factors related to patients' income and education are taken into account, differences in readmission rates among hospitals may not be as great as Medicare data indicate.

The study, published today in the May issue of the journal Health Affairs, was conducted by researchers at Washington University School of Medicine in St. Louis and BJC HealthCare. BJC operates Barnes-Jewish and St. Louis Children's hospitals, which are staffed by Washington University physicians.

"We still need to better understand which social factors contribute to patients being readmitted," said senior author William C. Dunagan, MD, professor of medicine and vice president of quality for BJC. "But we think models that do not include social factors deprive hospitals of valuable data that will help them determine how to use limited resources to help the most vulnerable patients."

The formula Medicare uses to calculate readmission rates and levy fines against hospitals does not include socioeconomic factors, which can affect patients' health after they leave the hospital. Poorer patients may not be able to afford prescription medicines or have transportation to doctors' offices for follow-up care. Patients with a limited education may have a hard time understanding hospitals' instructions for care at home.

Hospitals are paying a lot of attention to their readmission rates, which are being used by the federal government as a measure of hospital performance. In October 2012, Medicare began reducing payments to hospitals that have a higher than average share of patients who return within a month of being treated and discharged.

A concern is that not adjusting readmissions data for poverty or other socioeconomic factors could mislead the public into thinking that hospitals with a large share of disadvantaged patients provide lower-quality care than hospitals with more affluent patients.

For the current study, first author Elna Nagasako, MD, PhD, and her colleagues looked at hospital readmissions for nearly 60,000 patients treated for heart attacks, heart failure or pneumonia at acute care hospitals in Missouri from 2009 to 2012. They compared two different models for calculating hospital readmission rates within 30 days of discharge. One is the same model used by the Centers for Medicare and Medicaid Services. The other is a similar model that adds socioeconomic information drawn from census tract data. That model linked a patient's most recent address to poverty rate, level of educational attainment and housing vacancy rate in the census data. The housing vacancy rate can be a measure of neighborhood stability.

The model that incorporated socioeconomic data substantially narrowed the differences in readmission rates among the hospitals. For example, for patients treated for heart attacks, the model used by Medicare shows that readmission rates for Missouri hospitals ranged from 14 percent to nearly 21 percent.

The model that incorporates socioeconomic data showed a much narrower range of readmissions, 15.3 percent to 17.1 percent.

"A narrower range suggests that socioeconomic factors could explain a substantial portion of the observed differences in hospital readmission rates," said Nagasako, an instructor of medicine.

The researchers found a similar narrower range of readmissions for patients treated for heart failure or pneumonia.

For heart failure, the Medicare model showed readmissions ranged from 14.5 percent to 28.5 percent, while the other model showed a range of 17.6 percent to 25 percent.

And for pneumonia, the Medicare model pinned hospital readmissions at 11.2 percent to 18.6 percent, compared with 13.4 percent to 17.1 percent for the model that incorporated socioeconomic data.

The study's results also support the conclusions of an expert panel commissioned by the Obama administration that recommended a closer look at the effects of socioeconomic factors on performance measures. The panel also pointed out that financial penalties levied against hospitals may have the unintended consequence of transferring money away from hospitals that treat large numbers of disadvantaged patients.

Because the study used socioeconomic information from census tract data, which is easily obtained, rather than patients' actual data on income, education and housing, the study can't make specific recommendations about which patients are most likely to be readmitted. And, because the study looked only at Missouri hospitals, it is not known whether these findings extend to other areas of the country.

Rather, given the debate over whether Medicare should include socioeconomic factors in its formula to calculate hospital readmissions, the study raises questions about whether Medicare's readmission rates reflect social factors related to the hospital's patient mix as well as hospital performance and quality.

"We want to make sure that all patients, regardless of their social circumstances, receive the support they need after they are discharged," Nagasako said. "For some patients, this may prevent returning to the hospital, but for others, the best support may mean being readmitted to the hospital for care. So, in addition to tracking the readmission rates themselves, we also need to better understand why patients return."