A report finding no connection between socio-economic status and readmissions rates is being faulted for its lack of granularity by advocates for safety net hospitals.
The ongoing debate about the role of socio-economic status in hospital readmissions was rekindled this month with a report in Health Affairs showing no connection between the two.
"Our paper should not be interpreted to suggest that socioeconomic disparities in health do not exist," Susannah M. Bernheim, MD, MHS, director, Quality Measurement Programs at the Yale-New Haven Hospital Center for Outcomes Research and Evaluation (CORE), said in an email exchange with HealthLeaders Media.
"Socioeconomic status is an important determinant of health status," she says. "What we have shown is that for quality measures, like the readmission measures that already account for differences in patient illness, socioeconomic indicators do not have a big additional effect on hospital results."
Pressure Mounts to Risk-Adjust for Poverty
America's Essential Hospitals, an association representing safety net hospitals, took issue with the findings and said it relies upon ZIP code-level data that doesn't provide enough granularity about specific neighborhoods and streets where some safety net patients reside, or what goes on inside those homes.
"We believe the authors of the article have not truly identified the data that can help tease out the variables in geographic areas to fully understand why some neighborhoods are more likely to see readmissions than others," says Beth Feldpush, senior vice president of policy and advocacy at America's Essential Hospitals.
'Vested Interest' in Existing Measures
"They did take a look at demographic variables. The problem is they looked at ZIP codes where patients reside as well as dual eligibility for Medicare and Medicaid. We believe that ZIP code does not get you down to a finely divided enough area to understand the impacts."
Feldpush also noted that the Yale researchers are also "the developers of the readmissions measures themselves. So, you can understand they already have a vested interest in the measures as they are constructed."
Bernheim says CORE research uses U.S. Census data.
"The neighborhood data that we used comes from what is called the American Community Survey, which is collected by the U.S. Census Bureau… which includes information about things such as poverty, income and education levels, or housing within a neighborhood," she says.
She defended the study findings as "quite robust."
"Since we completed this paper, we have continued to study this topic and we consistently get the same results. Adjusting the readmission measures for patient socioeconomic status does not make a difference to hospitals' results even when we use other indicators of patients' socioeconomic status," Bernheim says.
Disparity-Busting Payment Reform, Validated
"For instance, we recently presented results showing that adjustment for socioeconomic status measured at the census block group-level also does not affect hospitals' readmission results. That research used an indicator linked to a much smaller neighborhood area than the ZIP code areas used for this paper, but it didn't change the findings. Some people think that results would change if we just had the perfect way to assess socioeconomic status, but that doesn't seem to be true."
Bernheim concedes that people may find it "surprising" that hospitals caring for poorer patients generally perform well on Medicare readmissions measures, and she offered "few things (that) may help explain this finding."
"Most importantly, the readmission measures that Medicare reports are adjusted for patients' level of illness. So if poorer patients are sicker when they arrive at the hospital, the readmission measures already account for those differences among the patients," she says.
"Our study used the measures that are reported by Medicare nationally, which take into account differences in how sick patients are," she says.
"We asked whether socioeconomic status played a major role in hospital readmission rates above and beyond the differences in illness that the measures already take into account, and we found it did not."
John Commins is the news editor for HealthLeaders.