Does a patient's race affect the odds of hospital transfer?

A study found differences by race, region, and diagnosis.


As the U.S. reckons with its long history of racial inequality, researchers have recently investigated the possibility of racial and ethnic disparities in a specific part of hospitalist practice: interhospital transfers. Using data from more than 5.7 million admissions in the 2016 National Inpatient Sample, a research team estimated adjusted odds of transfer by race/ethnicity, controlling for patient demographics, clinical variables, and hospital characteristics.

The results were, in a word, nuanced, according to lead author Evan Michael Shannon, MD, MPH, a chief medical resident for the internal medicine residency program at Brigham and Women's Hospital in Boston.

While Black and Hispanic patients had lower odds of interhospital transfer than white patients, this was largely explained by a higher likelihood of being hospitalized at urban teaching hospitals, according to results published online in July by the Journal of General Internal Medicine. However, disparities in transfer varied by region and medical diagnosis. In addition, secondary analyses showed that among patients hospitalized at community hospitals, Hispanic patients had lower adjusted odds of transfer than white patients.

Dr Shannon
Dr. Shannon

Dr. Shannon recently spoke with ACP Hospitalist about the findings, offering potential explanations and takeaway points for hospitalists.

Q: What led you to study this issue?

A: I was first interested in this topic when I was a resident at Brigham and Women's Hospital from 2015 to 2018. At our hospital, we have a pretty high volume of patients who are transferred from other hospitals because we're a tertiary/quaternary care center and we have very advanced specialized care, specifically in cardiology and oncology. So what I was noticing was that anytime we received a transfer patient from another hospital, they tended to be white/nonminority. And similarly, most of the patients on the cardiology and oncology teams were white and not really reflective of the Boston demographic, whereas on our general medicine team, we tend to see patients who are of all different socioeconomic, racial, and ethnic backgrounds, as well as in our outpatient ambulatory clinics. So I was thinking to myself, “Why are we seeing these disparities in what the demographic makeup is of our transfer population and these specialty care services?”

Q: What were your most interesting findings?

A: The first thing I was surprised to see was what the demographic breakdown was for the different variables I was examining. Compared to white patients, the Black and Latinx patients were so much more likely to have Medicaid, to be in the lowest income quartile. And then also just seeing how much more likely that Black and Latinx patients were admitted to large urban centers, which tend to be more teaching hospitals. I think you can extrapolate that those are hospitals that tend to have more specialty services available, although that's an assumption.

I think the most interesting findings to me were looking at this subset of patients who are transferred from community hospitals. . . . Latinx patients overall were less likely to be transferred, which I thought was very interesting because this really hadn't been looked at much in the literature at all. . . . And I was surprised that for Black patients in that first subgroup analysis, there wasn't a significant difference in transfer, but then was not surprised that if you looked at a certain subset of diseases, you could see some differences in transfer rates.

Q: What are some potential reasons behind the disparities in transfers of Hispanic patients from community hospitals?

A: I think that language barrier probably had something to do with it. There's emerging data that when there's language concordance between a provider and the patient, the patient is more likely to have a better outcome for certain diseases. So one hypothesis is that if a patient has limited English proficiency and their provider speaks English or another language, then there may be some disconnect as far as communication, which could then lead them to potentially not understand, “Maybe I should go to another facility.” Or, in some cases, what I would be most concerned about is that there is actual bias on the standpoint of the provider—either explicit bias, which is likely rare, but some implicit bias that would mean, for whatever reason, a provider is discounting their symptoms or can't really communicate so doesn't have the same sort of conversation or make the same sort of recommendations they would if there's language concordance. I think the other hypothesis, that there's been some data behind, is patients are admitted to hospitals where they feel more comfortable. So there's certain hospitals that I think Latinx and Black patients see as a hospital where they feel like they're heard, they feel like they're not subject to bias and discrimination.

Q: Do you have any potential explanation for the differences in transfers by diagnosis among Black patients?

A: I think it may have to do with the lack of strict criteria around when transfer is indicated. A lot of the work that my research mentor, Stephanie Mueller, MD, MPH, and others have done found that for some diseases, it's a little bit of a gestalt when a provider in a community hospital might decide to transfer them to a tertiary hospital. . . . For certain diseases (e.g., [myocardial infarction] and stroke), it's pretty clear-cut when a patient needs to go to a higher level of care. But then there's a lot of other diseases that people get transferred for where it's not very clear when exactly to say, “I don't think the hospital where I'm operating has enough of the resources necessary to take the best care of this patient.” So what I would be concerned about with these findings . . . is that for some of these diseases where there's a little bit more discretion on the part of the provider, that that can [introduce] bias, which may explain some of the differences that we observed.

Q: What would be your main takeaway points from the study for hospitalists?

A: I think it's important for hospitalists and any physician taking care of patients just to be aware that implicit bias is real and that it can affect your decisions, myself included. I always try to check myself whenever I make certain assumptions about patients before meeting them in some cases, or even after meeting them. One thing I would hope that could emerge from this, especially given a time where there's finally, importantly, much more emphasis on racial justice and breaking down some of those barriers of systemic racism, is to start to implement programs where there's implicit bias training, antiracism training at these hospitals. I think that could help make providers more aware that these do exist and, by being more thoughtful about it, could reduce some of the biases in practice that we see—not only in this study, but in various studies that have shown disparities in our care for white versus Black and Latinx patients.

Another finding of this study was that patients on Medicaid, for example, were substantially less likely to be to be transferred as well, as were women. So I think this points to potential bias not only based on race/ethnicity, but also based on gender or self-described gender. . . . My fear would be that a patient on Medicaid would be seen as an unattractive patient for a hospital to admit because they'd be getting less reimbursement. So that might be a reason why actually a tertiary facility might block a patient's transfer from another hospital. . . . Suffice to say that there's lots of biases in how we practice, and I think another thing that we as hospitalists can do is ensuring that patients are insured and being thoughtful about how to advocate for universal coverage, or a universal Medicare or Medicaid option, as tools to reduce some of these disparities by improving and increasing access.