Two weeks ago, we commented on findings from two recent papers analyzing outcomes disparities for Covid-19 by race based on claims and electronic health record data. Hospital quality appears to be one of the drivers of that disparity: Black patients were more likely to be treated at hospitals with worse outcomes for Covid-19 care.
We talked to Natalie Sheils, one of the authors, to learn more about what's driving that difference and what it means for health system leaders. Note that we've edited the conversation for clarity and for concision.
Question: What explains the difference in outcomes by facility?
Natalie Sheils: We're currently doing some research to look into that. I can't talk about it in depth right now, but to give you a quick preview: It looks like the difference came down mostly to resources. Hospitals serving more Black patients were less well-resourced on staff and on bed mix.
Whether the hospital was facing a surge also mattered. Hospitals experiencing Covid-19 volume surges usually had worse outcomes than those that weren't. And hospitals experiencing surges were more likely to treat Black patients.
Question: What's the implication for health systems?
Sheils: I think it's important to look at what we didn't find: We didn't find differences within hospitals for mortality and discharge to hospice for Covid-19 patients by race. So we don't think this is about bias within the hospital or in the care provided—at least in the context of those outcomes for Covid-19.
The much bigger driver was the difference between hospitals. And that outcome differences between hospitals seem to stem from resource gaps.
Which means that the problem is more systematic—and the solution needs to be, too. If the problem is that Black patients are more likely to be treated at hospitals with fewer resources, then the solutions need to address the funding or investment incentives that create that disparity.
Question: Did the gap close as the pandemic continued?
Sheils: In another paper, we found that hospitals improved outcomes for Covid-19 over the course of the pandemic. That makes sense, given advances in treatment methods and provider experience with a new disease. But we also saw that hospitals improved at different rates. And hospitals treating more Black patients saw less improvement than hospitals treating more White patients.
Hospitals treating more Black patients not only had worse outcomes. They also improved less over time, exacerbating the outcomes disparity.
Question: This analysis is limited to Medicare patients. How do you think these findings might change if it were possible to look across the full population?
Sheils: We chose to limit our sample to Medicare patients only because that is where we have reliable race data. We wish that data were more comprehensive, for example by including ethnicity separate from race, and were available in a broader population. But we worked with what we had. We know that this will skew our sample older than what may be truly representative.
However, there are also some methodological advantages. For example, we're not as worried about things like continuous enrollment because Medicare beneficiaries are not subject to the changes in insurance that may come with a job change or job loss. It might not be possible to control for patient comorbidity in other datasets as effectively if you have beneficiaries dropping in and out of the dataset based on those insurance changes.
It's impossible to say how the results would have changed in a different population, but since Covid hospitalizations are concentrated in the 65+ age group, we have confidence in the direction of our results.
Question: The two papers we wrote about in the previous post had slightly different findings. Where your paper explained outcomes disparities by differences between facilities, the other paper explained disparities as a result of a higher rate of hospitalization for patients of color. How do you explain those divergent findings?
Sheils: We used different data with access to different kinds of clinical variables and the studies were asking slightly different questions, so it's not surprising that we reached different conclusions. That said, both papers found outcomes disparities. While we attributed those in part to differences between hospitals, the other paper hypothesized that they might stem in part from disparities in access to care, with patients of color less likely to seek care for asymptomatic or mild infection. The result would be that confirmed cases among patients of color would show disproportionately severe symptoms and require more frequent hospitalization.
I don't think the findings are contradictory. It could be that hospitals serving more Black patients were more likely to see Covid-19 patients who hadn't yet received care elsewhere. So hospitals with fewer resources might also be seeing patients without primary care providers, or who present to the hospital with more advanced symptoms.