Reducing Stigmatizing Language in the Electronic Medical Record for More Equitable Care With Alex Boucher, MD

In this interview from the 2023 American Society of Hematology (ASH) Annual Meeting, Oncology Data Advisor speaks with Dr. Alex Boucher, Director of the Pediatric and Adult Sickle Cell Programs at the University of Minnesota Medical School, about his presentation on racial discrepancies in the use of stigmatizing language in hematology/oncology electronic medical records. Dr. Boucher explains the impact that stigmatizing language can have on patient care and ways to raise awareness for mitigating its use.  

Oncology Data Advisor: Welcome to Oncology Data Advisor. Today, we’re here live at the ASH Annual Meeting, and I’m joined by Dr. Alex Boucher. Thanks so much for coming by today.

Alex Boucher, MD: Thank you for inviting me.

Oncology Data Advisor: To start off, would you like to introduce yourself and share what your work focuses on?

Dr. Boucher: Sure, I’m a Pediatric and Adult Hematologist at the University of Minnesota, having been there for a few years. I lead our sickle cell program, and a lot of my research is around health equity, not just for sickle cell disease but even more broadly within the medical community.

Oncology Data Advisor: Awesome. Today, we’re talking about your research here about racial discrepancies in the use of stigmatizing language in hematology/oncology electronic medical records (EMR). For background, would you like to tell us a little bit about what stigmatizing language is and the role that it can play in perpetuating biases?

Dr. Boucher: Part of this idea certainly generated from my patients with sickle cell disease, who are predominantly underrepresented minorities. They often have lower socioeconomic status. When we took this on, it was certainly driven by my personal experience of seeing words in the medical record. These can include “belligerent” or “hostile” or even “noncompliant.” We recognize that those mean something over and above medical connotations; they’re not objectively medical, but they do potentially trigger a reaction in a reader of that document or perhaps a listener if you’re interacting vocally. It starts to paint a picture and perhaps trigger some implicit biases.

With that, and having been triggered myself in many ways with the medical documentation—which unfortunately is how most communication happens in the medical world—we undertook this project. We looked at the adult hematology/oncology field, but actually it’s part of a much larger project across our health care system. We aimed to find out, regardless of age and inpatient versus emergency department setting, if there are certain words that are personal characterizations of patients but have generally agreed-upon positive or negative biases. To figure out what our current baseline is, before we could even intervene, we needed to know where we were starting from.

Oncology Data Advisor: What are some of the ways that this stigmatizing language could affect patient care?

Dr. Boucher: If somebody says a patient is noncompliant, that means a lot of different things to a lot of different people. Oftentimes that will mean that somebody, either that person who says it or potentially someone else down the line, may not offer the same kinds of treatments. Another one may be if somebody says a patient is hostile, and you’re reading that before you go into the room, you’re going to very likely be a little bit more defensive when you walk into that room. You’re probably not going to hear the whole story. You’re potentially not going to offer all the things you might versus somebody who’s polite or respectful.

Those words can mean different things to different people, and they change the nonverbal interactions with people. Whether it’s hematology or oncology, thinking about the word “noncompliant,” you may not offer that person a bone marrow transplant if that person has “not been compliant” with their medications, even though there’s a very likely a good reason that they haven’t been able to follow through. Maybe it’s because of transportation or financial difficulties, but with the term “noncompliance,” the default tends to be, “You didn’t follow the rules that I set out.”

Oncology Data Advisor: Why did you decide to focus on the electronic medical record to investigate stigmatizing language?

Dr. Boucher: To be honest, because that’s unfortunately our most common source of communication. The reality is that we get too busy to pick up the phone and call somebody, or we’re communicating messages across to another system. We may get a referral from a different patient or from a different center. Or maybe somebody is coming to us as a second opinion and we don’t know who the other doctor is and what their time commitments are. So, we send a note and we fax the note, or in the case of our study, we’re in the inpatient setting, and we type in this note because we’ve got to write it and because there’s got to be some communication, but we don’t know who the audience is going to be.

The reality is that these implicit biases, that we may have ourselves, get put into text. It essentially gets put in digital cement. In some ways, it’s like social media. It doesn’t disappear. You can even sometimes try to delete it, and it flags it as deleted, but it’s actually still there at least on the provider side of the medical record. The risk is getting perpetuated, particularly in relation to individual characterizations of patients.

Oncology Data Advisor: How did you and your team go about designing the study?

Dr. Boucher: The impetus was a patient experience that I had where I felt like the text that was written about one of my patients from the emergency department (ED) was very stigmatizing. The likelihood was that the next ED provider would read that note and not my own or somebody else’s, because they were presumably and objectively trying to figure out whether it matched the last admission. When the last few keywords are, say, “problem patient” or “behavioral problem,” that’s going to flag them as something different.

I wanted to figure out, because my patient population is heavily underrepresented minorities, was this just my experience or was this a broader issue? We pulled a big data set. We took two and a half years of data from five different hospitals within our health care system—a couple suburban, one pediatric, and two urban centers, including our core academic center. We basically looked at notes across the board. We looked at notes from any person who wrote in the electronic medical record, whether they were a spiritual health counselor or a pharmacy student or a nurse or a physician. We picked out keywords that we felt we had seen regularly enough that they mentally triggered a picture of who that person was before we even walked in the room. Essentially, they prejudiced us one way or the other. We did make a conscious effort to look at positive terms as well.

Oncology Data Advisor: That’s great. What were the results that you found in the study?

Dr. Boucher: For this abstract, we had about 1,000 patients, and we used natural language processing software, essentially artificial intelligence (AI), to find these words. Then we spent a year or more trying to confirm those words were use in the right way. For example, for “compliant,” we wanted to make sure that it wasn’t “complaint” misspelled and that “not compliant” didn’t register as “compliant” because of the space”. That’s that sort of background. We can’t just assume that what the AI is telling us is accurate. We found those things as we went.

This is a part of a much larger data set that hopefully will be published soon of about 12,000 encounters in 9,000 patients. In our analyses, we looked at the use of these terms within racial groups—primarily White and Black, although we also looked out to other race ethnicities and multiracial groups. When it came down to it, we mainly analyze White and Black, and there were a couple of those negative terms that really stuck out. When they did stick out, the odds ratio showed it was more likely that the negative term was in someone who self-reports as Black in our demographic. On the flip side, things like “polite” and positive terms in general were more likely to be in someone who self-reports as White. Those racial tendencies and stereotypes that are out in the community are being perpetuated, in some ways not surprisingly if we think about the fact that this is the way we communicate in our digital record.

Oncology Data Advisor: Did any of these results stand out to you as particularly striking?

Dr. Boucher Not striking, but alarming. In other words, I don’t think we were surprised by any of these data. I’ve struggled with this in my clinical care. I’m the first to admit, my own documents were in this EMR too, so this is not me standing on the outside of the house throwing stones. This is somebody who lives this and recognizes that I’ve been part of the problem, but I want to address it. While we didn’t see it in this abstract, we did see some instances in the larger document of curse words showing up. There are actually a lot of curse words in there that were being quoted by patients, and they weren’t threats to anybody—we excluded those—but these were ones where you could have said, “Patient cursed at me,” but instead somebody decided to put in the actual text for whatever reason. Again, I think it triggers you in a different way when you read that text.

The fact that it was so skewed that every negative term that had a significant odds ratio was in a Black individual, and every positive was in a White individual, I think is the biggest takeaway versus any one word. Again, there are other words that could certainly be used here. Sometimes “not adherent” becomes the new “noncompliant,” and I don’t like that term either. But the reality is this is just such a heavy skew, and it’s so pervasive. We’re seeing it in our hematology/oncology patients, but that’s just a microcosm of our larger medical community. We also say this is not just a Minnesota issue. Now, maybe we’re one of the first to look at it, but I can pretty much guarantee you that we are not the only one who has had this problem. We may have been one of the few to look at it.

Oncology Data Advisor: Definitely. So, in light of these results that you found, what changes do you recommend for improving awareness of stigmatizing language in documentation and starting to mitigate its use a little more?

Dr. Boucher: One of the things that we’re hoping to do is get people more aware, although I’m also very cognizant that it’s not enough. In other words, making big statements about these things is fine in a very short term, but it’s unlikely to stick. Even as we look at our larger data set, George Floyd was murdered kind of midway through analysis of our data set, and that was just a few miles from our hospitals. Nothing, to be honest, really changed over time in that post–George Floyd versus pre–George Floyd event. This tells me that we have a mission, and it’s an admirable mission to improve our health equity, but it wasn’t being passed down, at least in this manifestation, to the way we communicate.

What I’d like to figure out are ways that we can nudge people to reduce using these terms. It may not be to totally wash away all these terms; maybe there’s an important use for them, but we can be more thoughtful about when to use them. If you’re going to use the terms “noncompliant” or “compliant,” then explain what you mean by that if you write that, because then you take away the reader’s implicit bias of what that means. You’ve put out what this means to you, and then I, as the reader, now can interpret that and decide whether I agree or disagree. Now I’ve gotten more clarity. If the patient is noncompliant because they didn’t have transportation to get to the pharmacy, I wouldn’t call that noncompliant actually. I would say, now we have a social issue we need to address and we can get this patient back on track. That’s one way to do it, and I think it’s a piece of a larger picture of health equity.

This communication modality is not going anywhere, and also, our patients can see it because it is an era of open medical records. They can’t respond to it as directly as others, but they can see it. I think that awareness and vigilance are key. We need to get back to the role of the medical record as core medical information, and that’s it. That’s what it was back in the paper era when nobody had time to type everything and we didn’t have copy forward. You just wrote the core things. That had issues too, but I would argue that in some ways, it didn’t perpetuate these characterization stereotypes because nobody had the time to write them out.

Oncology Data Advisor: That makes sense. Do you have any future research plan in this area?

Dr. Boucher: We do. We’re now trying to figure out, now that we’ve done our full data set and submitted it, how to nudge people in a different direction. In other words, there’s a small but increasing number of journal articles out there saying that this is an issue in various ways, shapes, and forms, but there’s very little, if any, data on how we can use that information to change outcomes. Even in ours, we make it very clear that we can’t even address outcomes here because it’s just too pervasive. There’s not one measurement that we could do other than the use of the words.

I would love to figure out ways that we can, again, nudge. I like the behavioral economics theories of getting people to think differently and recognize their own biases but still giving them the freedom to write what they feel is appropriate. If you’re going to use the word “compliant,” either you have to define it or get rid of it. Either way is probably better than just putting “patient is not compliant,” period. That’s one way, but we’re really trying to figure out what those potentially meaningful clinical outcomes area, whether it’s in the heme/onc world, which is certainly where I spend my time, but also just in the general hospital sphere.

Oncology Data Advisor: Great. Do you have any additional messages or parting words regarding reducing the use of stigmatizing language in the hopes of providing more equitable care?

Dr. Boucher: I think it’s easy for it to be thought about in my world of sickle cell disease. That’s kind of the low-hanging fruit. People say it’s obvious, although we still see it. I also think people are nowadays probably nowadays, I hope, a little more apt to check themselves and not use those languages. But in our broader oncology world, I don’t know if that’s true. What I encourage folks to keep in mind when we’re talking about health equity is that this is everywhere. This is out in high socioeconomic communities and in low socioeconomic communities. Our outside culture tends to seep into medical culture. We don’t live differently or totally separately within one or the other. We leave the hospital, and we go into our communities. We read the news, which is not medical.

I’ve of the mind that if we can find ways to change this, and this just being one little piece of a big puzzle, we can be iterative. We can build upon that and go to the next step. There’s no golden ticket here. There’s no easy fix. But the more we can actually take those second thoughts and think a little more deeply about what we’re writing, not just chalking it up to an issue in sickle cell disease, I think we as a heme/onc community can become just one step better in terms of health equity.

Oncology Data Advisor: Absolutely. This is such an important issue. Thank you so much, not only for stopping by to talk about this today, but for embarking in this work and helping to improve patient care.

Dr. Boucher: I really appreciate it. We’re trying to spread the word, and I’m glad that you were able to find the abstract and reach out.

Oncology Data Advisor: Definitely. Thank you very much again. It was wonderful to talk with you today.

About Dr. Boucher

Alex Boucher, MD, is an Assistant Professor in the Division of Pediatric Hematology/Oncology and Director of the Pediatric and Adult Sickle Cell Programs at the University of Minnesota Medical School. He specializes in the treatment of sickle cell diseases, and his interests include addressing diversity, equity, and inclusion challenges to provide more equitable patient care.

For More Information

Ivy ZK, Hwee S, Kimball BC, et al (2023). Racial discrepancies in the use of stigmatizing language is evidence in hematology/oncology electronic medical records. Presented at: American Society of Hematology 2023 Annual Meeting. Abstract 3758. Available at:

Transcript edited for clarity. Any views expressed above are the speaker’s own and do not necessarily reflect those of Oncology Data Advisor. 

Related Articles


Your email address will not be published. Required fields are marked *