While physicians intend to treat all their patients with equal respect and compassion, studies show that favoritism and other implicit attitudes can emerge, especially in times of stress, affecting medical decisions and care quality. Pulmonary and critical care specialist Aaron Baugh, MD, discusses the evidence on how and why bias surfaces in everyday practice and offers simple solutions that benefit not only patients but the providers themselves. Bonus: Get a better understanding of bias self-tests.
Thank you so much. My name is Aaron Baugh. I am an Assistant Professor of Medicine at the University of California San Francisco. I work at our Parnassus campus uh on the uh Pulmonary Consult Service and in the Pulmonary Function lab. And uh I actually spend a lot of time in research um in the airway Clinical Research Center. We do uh work on asthma uh COPD and also just in pulmonary function in general where especially of my work has been trying to understand uh health disparities and how that contribute to um uh patient outcomes along uh socio-economic and racial ethnic lines. Uh I'm really happy uh to be joining you all to uh spend uh a little bit of time talking about unconscious bias inequity uh and um in the health care system, right? Um this is going to be a uh brief quality improvement talk and like everything under that umbrella, we have a couple tasks. First of all, we wanna, you know, define the uh type of error that we're talking about and we wanna see if we can understand really like how is it that this emerges in spite of everyone's, you know, best intentions to do better. Uh We wanna look at uh what evidence we have to understand what is its clinical impact and how does that contribute to poor outcomes in our patients? And the last thing I wanna do is see if we can understand from both the systemic and an individual standpoint, what we can do that uh will allow us to hopefully uh stop or limit this from happening in the future. So, um couple key principles we're gonna look at is a bias and we can just define that broadly or, or simply you see the formal definition up there. But I just want you to think about favoritism. And I, I'd like to start here because oftentimes uh when we're talking about topics like these of uh of a bias or bigotry or inequity, uh people think in, in very broad and vivid colors, right? That Hitler hated the Jew or, you know, the Ku Klux Klan or something like this, right? Where there's one person that's very strongly uh in oppo opposition or antipathy to another group of people. Um But I wanna suggest to you that um oftentimes just as important of a mechanism is not so much that you are mistreating one person, but that um how is it that you treat, you know, selectively, uh uh favoritism towards another group, right? And this can be somewhat tricky for us because I think we all it's a common thing that we uh find points of similarity with our patients and their interests or background or things. And these can help us remember the cases and help us connect with them and help motivate us. But uh we then have to be careful about how we then treat people that don't share those points of similarity or, or uh likeness with us. Um One really important principle that helps us and we'll see it emerge a couple times in this talk is cultural humility and that's trying to use self reflection and uh you know, really mindfulness to think about. How can we identify um when we're confronted with someone that's not really what we um you know, not like us or not the background or the situation we're familiar with. How are there ways that we can catch ourselves to identify when bias is happening and stop that? And then finally, I just want to come to this point of intent versus impact. Um And uh a lot of times again when uh we start to have these discussions, uh a natural thing people wanna ask again is like, uh you know, oh, well, like, what did you mean to do? Were you trying to offend someone or were you trying to mistreat someone? And I think, you know, that is a natural question, but in some ways, I think it's a um it's a, it's not a helpful framework, right? Like, uh again, if we think about our broader quality improvement umbrella. No one is trying to cause a, you know, a side infection or no one is trying to leave in a fully, too long and uh inappropriately. Right? Um And so, uh, it's not that we want to question each other's intent. Um, but we wanna say accepting that we're trying to be all good physicians and trying to deliver uh high quality care to every person. Um, what we should really be looking at is in terms of the outcomes. Uh why might that not have happened or why might not that happen? And then again, are there ways that we can either modify our behavior or modify the system that we're practicing in that makes it a little easier for us to uh you know, do the right thing or do the thing that's gonna deliver a more equitable and high quality care. OK. So let's get into this. There is multiple levels of bias that can happen on the um on a uh uh that, that might be relevant to us, especially when we look from a social determinants of health perspective, uh which teaches us, of course, that the, the major uh influence of how people's overall outcomes come are not necessarily what happens in our clinics or in the hospital, but in the broader context of our lives. So let's look at some levels of bias that could impact uh impact patient outcomes at our broader societal level at the level of the health care system and in uh individual patient provider interactions and will go down that whole scale. So, first of all societal biases, uh let's look at these inequities that exist in um American society right now. And so, uh if you look at, for instance, uh Latin X communities, native Americans, African Americans, right? Um these uh different not bins of nonwhite groups. And you say that, you know, what, how do they compare in terms of their outcomes? Well, um even after you control for things like uh highest level of educational attainment and household income, um you find that they tend to be in neighborhoods that have greater exposure to pollution. Uh And part of this connects to a story that we'll get into just a moment about redlining where uh the US really uh developed its housing market along racialized lines uh by federal policy. Um And so uh systematically uh kind of crowded into neighborhoods that were uh a poor quality, you know, next to factories uh less desirable to live in, right? And that's gonna influence maybe their pulmonary health, right? Um We see that if you look outside of medicine, uh and you look into like, you know, criminal proceedings, even when you look at criminals who have the same uh or people who committed crimes that have the same prior records, the same severity of the current offense and like those sorts of factors that might adjust for things. Uh You again, see that, uh, racial, ethnic minorities tend to be more harshly sentenced for, um, the same offense, which is a little bit odd. Right. Um, this, uh, next point, um, is, uh, a meta analysis actually that was done over about 20 to 30 years. And this, uh, this finding has been, as long as it's been sustained in America that if you take two identical CV S or resumes and so they have the same prior work experience, same letters of recommendation, uh the same uh educational background, right? Um But you just change the name on the top and you ask that it is to uh you know, suggest a name that might be more stereotypically uh like African American or black sounding and one that might be more uh non Hispanic white sounding, right? It's the white name that's about 1.5 times more likely to be called back for an interview or for hiring, even though everything else was the same. And this is true even whether you have like an exotically uh say African American name or uh you know, very ethnic size name or uh it's just one that people could clearly say this is a little more associated, right? So this is uh you know, the difference between uh you know, Reginald Washington or something like that and Kamal Kambo or Barack Obama, uh we can tell the difference in those names, but it doesn't make any difference for the hiring, right. Um, they're both, you know, more African American names and neither of them would get called for that interview or less likely. Um And then, of course, we know, and uh I'll just mention uh for this one, they've also done this study uh similarly with uh just gender, right? Changing. So you, everything is the same, but the person's female, right? And again, much less likely to be called for that hiring or that interview. And that makes the important point that um we're gonna be talking a lot about uh racial ethnic disparities and evidence in this talk. But I don't want you to think of it as only segregated to that area. It's uh where a lot of evidence and research has been done because, you know, kind of, of our national history and there's interest there. But all these potential um places where there's power dynamics by gender, by race, uh by religion, by language. Um You know, we should be concerned about potential bias and we need to be thinking critically um about keeping all these lessons in mind. Um The last we just say, uh again, uh with all that in mind, African Americans have about, you know, less than 1/10 of wealth of uh white families even after controlling for current income. Uh So, a suggestion of the historic impact of the discrimination inequity that these communities have faced over time. So certainly these things could all you imagine might uh influence a patients, you know, availability to come see you in clinic or uh you know, how much time they have to look into their conditions and be uh their own health conditions and be proactive or even what exposures they might have that put them at risk for different diseases. So we have to think about these. On the other hand, even if we say, well, those things are equal, uh we also have a health system that is not uh entirely well equipped to deal with them. And so we have a couple of things. So uh there's clinical trial under representation and this is uh Doctor mcgary study was in CF where uh you know, there is almost no uh you know, nonwhite patients uh that are take part in cystic fibro trials, even though very clearly, there are uh nonwhite populations that have cystic fibrosis and meaningful numbers. And so then you get situations where uh because their mutations haven't been tested in these drugs that the CFTR modifiers don't apply as immediately to them and they get um or don't are not, you know, FDA authorized for them and then, or not uncovered by insurance. So then you get uh you know, some outcome differences based on that. Uh there's physician distribution inequities uh right? Not only between communities of color and not, but, you know, rich and poor communities. We're all in urban communities, right? So there's not an equal uh access to physicians and to specialist care and that sort of thing that you might also imagine makes a difference within a uh given uh uh clinic or something. Uh We of course saw this work uh doctor Shouldering and this uh Tom Valley is his um kind of senior author there. Um looking at pulse oximetry and people with darker skin, it gives falsely reassuring uh high numbers. Uh and so mass uh clinically significant hypoxemia in those with dark skin. And you can imagine this might have influenced like in COVID or things when we were trying to decide who to send home and who to uh keep in our hospitals or how much to escalate care, right? Um uh And it's not so much the mean was different but the, the uh 95% confidence interval that spread went so much lower on uh our African American patients when we used like, you know, 89 or 90%. Um Now, then the last one, this is some of my own work, I'll share with you. And what we looked at was the use of race specific versus uh a race neutral or multi rial uh pulmonary function reference equation to understand the relationship between symptoms and uh lung function in COPD. Um Because this idea, we used to use race specific equations thinking that oh, it's going to like, you know, uh there's differences in baseline lung function. So we should account for those by race and use race specific equations. Um But what we found is that it created this very weird situation where for the same level of lung function, say 75%. It meant one thing for a white patient in a very different thing, a much higher level of symptoms for a black patient. And so that's kind of weird. And so if you're thinking like, oh about 70% should correlate to this level of severity. Um You know, you were really missing the vote again, just like this pulse oximetry, uh getting a falsely reassuring impression of how sick your patient was. Um but this effect really goes away or significantly reduced when we use a multi racial or race neutral equation. And so thankfully, just this spring, we were able as the American Thoracic Society and pulmonologist to recommend everyone use now these race neutral equations rather than race specific. But so this is something that has been fixed. But, you know, all these other things are uh really potential problems that we haven't addressed yet to say, well, how uh can you then as a provider uh even do this even if you wanna treat everyone, right? Um You know, there's these differences in the quality of the data we have available to us from the diagnostic tools to the medical literature down with doctor mcgary that inhibits our ability to do so. And now the last one though those are all important. But um it's fair to say that while we can advocate within our professional societies and in our voting and our activism in the community. Um you know, those are not things that we can do so much about as uh clinicians from day to day. So, uh what we want to understand is something that we have the most control over, which is how do we as individual providers, how can bias manifest in us and how can we take a better control of that? So at least we're not contributing to the problem anymore. Ok. Well, so first of all, we understand, why does uh this bias persist, how does it keep going uh for any of these? Really? And so one of them is a thing is we want to say that um especially after, you know, things like World War Two and the Holocaust Hitler and um you know, more generally, um their response to the civil rights movement and things, uh people have thankfully gotten a sensitive that this is not, you know, something to be proud of or we're not a society that wants to discriminate. And I don't think there's really uh anyone among us that would say like, oh, I want to be biased or I want, you know, to have skewed judgment. Um But then because it's socially desirable, sometimes this emerges as an unwillingness to look and say, well, is this a problem or not or just say like, oh, I don't think it's a problem. I'm not gonna address that. Uh which makes it hard to then suss out where there's still problem areas. Uh Number two is the issue of perspective. So I'll tell you that when I was in training, I had a uh female senior resident and we had an incident where a patient was just repeatedly kind of disrespecting her and referring to her as like a nurse and things. And I was really horrified and I was like, oh my goodness, I've never seen anything like this. Like this must be like the first time, right? And her response is kind of like, no, I'm actually pretty used to it. Like it is terrible, but it's not my first time seeing this because I as a woman and in the position to see this all the time because people are doing it to me where, you know, I as a man, uh I'm not necessarily seeing that because so it looks rare to me from my perspective where I am, but from the person that may be suffering from the bias, uh you know, they're seeing it quite a bit. Um And that can, it can certainly happen. You can imagine in these other settings, right? If you're one of a few racial minorities in a community, uh you know, the times that other people see you get discriminated against might be rare, but the times that it is happening to you might be uh frequent from week to week and then the last one we want to get into is this ordinariness and this is where uh you know, something might be happening. Um that is biased. But um we have an alternative explanation for or kind of a cover of it. Um And this is not necessarily malicious, but it is something of like, uh you know, we, we uh uh II I, it's cognitive dissonance, right? We don't like to say there's uh that or think bad things about ourselves. And so if there's some other explanation, we can kind of latch on to it even if it's not true. So, uh here's an example of this, we talked about redlining earlier or I referenced it. This was a federal policy where when uh the government was really helping people get home loans because no one really had private homes at that time and they were expanding homeownership and uh after the Great Depression and during, um, they said, well, let's go out and let's rate all these neighborhoods. What is the, uh credit worthiness? What is the creditworthiness of, um, of these different areas of these different neighborhoods for getting a homework? Uh But the problem is the way they did it was if you were not white, they specifically gave you the lowest rating no matter what. And now, um, the reality of how this happened, uh, let's take a look at this study that was done in the desoto car area of Saint Louis. Um and in this neighborhood, um, what happened is that Saint Louis, uh, had a segregation and so, uh, blacks were all confined to just one kind of new neighborhood. Um And so because they couldn't live anywhere else, they had to live in the city, it became a lot of overcrowding and then the landlords didn't have to do anything to keep it up because there was nowhere else they could stay anyway. Right. So you have very poor conditions, lots of overcrowding. Of course, that's a perfect place for TV to bloom. But now, rather than talking about like, hey, maybe our problem is we have a segregated city and we're only allowing these people to live in this one block, like what they said was, well, look, you know, we have all these TV cases in this one area. And so if we're gonna like, mark that, this is a poor risk, it's like, yes, we're using a racial rule, but we're not trying to be, you know, racially discriminatory. It's about, you know, public health and that, oh, it's all this TV, here. And this became a big justification for this redlining policy which was really racial. And so it's giving this sense of ordinariness of like, oh, I'm not actually trying to make a racially biased decision. I'm making a public health decision. This alternate explanation for something that at its core was about um these housing covenants and uh restricting where people could live and why? And when uh and so that's just some examples from Barter Society. So let's see how these all apply in clinical medicine, which is what we were coming to, right? OK. So, first of all, where can we see this? So this is a study that was done uh about uh half a decade ago now or more. Um And what they did is they use standardized patients. Um and they looked at um these pairs where they had either, you know, a white or black physician and then whiter black patient that uh had this uh this scenario was like metastatic uh cancer of some kind. And then they're gonna be talking about end of life and goals of care and how does that conversation go? And they taped them and then they scored them for both verbal and nonverbal communication. Now, in terms of what the physicians said, the information they communicated the options they offered. It was exactly the same, which is again, speaks to this, that we as a physi as a profession, I think, do value equity, do value trying to treat all our patients the same. Um But yet, when they looked at not what was said, but how this was presented, did they touch the patient? Were they, you know, smiling or open? Um you know, did they, um you know, was their posture with their arms open or closed, like all these sorts of non verbal cues that might be happening unintentionally that you might not think so consciously about. Um, there was less positivity and kind of more guardedness or hostility, uh when it was this opposite pair of like a white uh physician with a black patient. And you know, why did this happen again? We can't say we can't speculate into their mind. Maybe, um, they didn't like it or maybe they were just nervous of what the patient would think of them. But uh regardless these cues came out and uh what we noticed that the patients were responding to that, like when the, the physician was less open with them in that interaction, it changed the way the interaction went. And indeed, if we look at kind of meta analysis or big studies, uh what we see is that when we pair, uh it, it's not so much been seen in uh uh the black physicians, but when we look at um uh white physicians pair it with uh minority patients, um those discussions tend to have less patient centers, um there tend to be less of those positive nonverbal cues we were talking about. And then at the end of the visit, the patients tend to rate those or whether it's standardized patients or real study with real patients, right? Um And so again, I wanna make this point, uh we can see this, especially with like nationality or language and also that these uh ratings are associated with implicit bias. And so because there's a lot of the, you know, uh this, this kind of uh uh standardized patients and discussion is good for trying to figure out whether there's bias in, in these kind of talks where it's all just a patient discussion, information sharing decision making, right? But a lot of things we do in as clinicians are not like that. So how do we understand whether there's bias in those or not? And so this is uh where we come to a tool that has been pretty widely used. And maybe you've heard of it before, this is called the implicit association test. So the implicit association test is a speed sorting task. So first you're presented with uh you know, uh pictures or words and you're asked to sort on to the right or left hand is something uh you know, is this like a black person or white person? Is this a man or a woman? Is this whatever thing? Right? And then another time you're going to be tested on some idea like, is this strong, is this weak, is this um you know, uh you know, young or old? Right? And so then you get these uh combos together. So like uh you know, woman, strong or man, weak or black, good and white, bad or white, good and black bad, right? And you do all these different combinations. And so the idea is that uh do you have a differential like speed and complexity and sorting um the thought being that if you've been uh by making it a speed test, you're again trying to stop the things that you think about uh uh consciously and kind of really force those automatic reflexes, the unconscious things that we do. Um our heuristics and our kind of default modes of thought or action into the forefront. And then that if something is really, you know, makes sense to you, you'll be able to hit that really fast and reliably. Whereas if something doesn't, is, is uncomfortable to you, or it is unusual to you because of the way you've grown up or the way things have been presented in you in the past, you might be slower, make more errors, right? So if you've been in a very uh society with very traditional gender roles and you've had, that's all the, you know, information you see, then if you're as a sort of task and it's asking you to pair things that are like, you know, woman, strong woman, independent, like that might be like, oh, that's, you know, not usually that doesn't fit with my uh gender roles that I've usually seen. So I might go a little slower or score differently on that. And so that's kind of how it works. Now, there's a couple of things that people say about. Well, is this really a good tool? Um What do we know about this? Um So first thing they bring up is like, you know, sometimes people will take this and it will say it biased against their own self. How can that be true? How can you know, a, a black person or a Latinx person be biased against Latinx? Like that can't be right? Or how can a woman have an anti female bias? What's going on? Um And I'll just point out this is not new, like in 1950 when they made school integration and thing, they cited an even earlier psychological study from 1950 by Clark and Clark. This is a, a really wonderful uh two of the first uh black uh psychologists to graduate from uh the program they were working at. Um and they did this famous doll study where they asked black Children whether they preferred a black doll or a white doll, right? Um And there was this overwhelming preference again uh for white dolls. And when you ask the Children, they would start explaining uh really negative attitudes about their own race, not because, you know, they were necessarily hateful Children or whatever. So that's not the way to interpret this but that, you know, the environment that they were in had a lot of negative messages and even, you know, without thinking about it, they had absorbed some of these and could reflect it back in their behavior. Um Not so that I don't think invalidates anything. Number two people talk about the reliability that sometimes they take the test and they would get like, oh, I get um you know, high bias, but then the next time I got medium bias and I took them right after each other. So, is that really reliable? Um And so the uh test retest reliability does have some uh flow or error to it. And the reason for that is that we're trying to approximate and get at this underlying uh unconscious association. And it certainly, I don't think anyone would claim it does so perfectly. Uh But when you look at the internal reliability in terms of its psychometrics, um it's very strong and so that doesn't necessarily mean it's a bad tool. It means that it's testing a real idea underlying imperfect. Now, the last thing that you guys might be saying is, well, what does this have to do with anything with, with, does it have any connection to behaviors or attitudes or clinical management right now that I've introduced this tool? Why am I spending so much time on it? Well, I wanna spend time on it for a couple reasons. Uh First of all, these higher biases, um mean this is uh more of a pro white bias, uh these higher numbers. Um And uh what you can see is that relative to all groups and even relative to other professionals like lawyers or phd, doctors have some of the highest uh uh racial bias, pro racial bias scores as a group. And so that's the difference even though in individuals, you can have test retest uh very, when you look at the in aggregate and over averages, these are very strong and reliable. So we're seeing that uh physicians as a group, there is more of us on average with this pro white bias even compared to other groups. So that's a concern. That's, and so now does this now influence our clinical care as our next question? Well, let's take this in a couple parts. First of all, um uh this is just, um, you see, and, and I do apologies, I don't mean to offend or anything. But, um, uh I want you to see really what's going on. So when you use Google searches, so anonymize, how do people talk, how do they behave? Right? Um There's this very strong correlation where uh in the state level average, um when you look at uh whites from that state, they're much more likely to use these pejoratives and things. Uh then blacks. And so it seems to be, uh you know, influencing something, right? But you might say, well, this is just, you know, that's just another way of expressing beliefs or ideas, that's not really a behavior or an action so fair. Ok. So this is a study of schools across the nation and they're looking at county level. Uh So they have these, uh these hundreds thousands of teachers take this implicit Biot test and these higher numbers again is more pro white bias and then they looked at in school and out of school suspensions for, uh, black versus white students. Um, and this little dotted line is where you start to get at least slight pro white bias and you see it really kind of takes off that, um, the more pro white bias they had, I'm sorry, more pro white bias they had, the more likely they were to be doing this discipline of sending these, uh, students with a much bigger gap for black and white students. This is true even after control, for class size, after control for socioeconomics in the county. Um, and so then you ask, well, ok, but is it possible that maybe the black students were just not as well behaved? I guess that's one possibility. But another thing we have to take in account is that in this same study, they found that, uh, teachers that operated in the areas with the most black students actually had the least anti, uh, the least pro white bias and least anti-black bias. So that was, um, you know, not be consistent with just, oh, this one group was really bad. Right. So something is changing about maybe the response to how they're dealing with discipline based on the level of racial bias they have on this IAT, um, that doesn't really correspond. It's not over bias of, I hate, you know, this group. But, uh, when you do this test to try to look at their automatic processing and their modes of thought, their heuristics. Um This is what emerges now, now, we gotta say, OK, so that's behavior among some professional, but it's not a clinical medicine, right? So let's go over to some clinical medicine and we'll start again at pain management, right? So, um this is again, oncologic cancer management, uh oncological pain management. And again, we have increasing uh these positive values are more um uh uh pro white bias. Um And so what we see is that uh when you have a black standardized patient, as they uh get more as the the provider reports on their test, not overtly, uh there was no correlation, but when they looked at the implicit association test, the more pro white bias that they have, they then had uh this is an justice scale. So the, the wax, this is not the same across these studies, but the more pro white bias they had um the less likely they were to give an appropriate prescription for pain medication also not shown here, much less likely to um uh communicate and ask a lot about the pain. And uh one thing you'll know if we just look at what we were talking about earlier is not, this line is so steeply down that you're so much decreasing it for blacks. But like, you know, as pro white bias increase, you're much more likely to give it to a white patient. So that, that again, that idea of it's not so much you need to be against someone but sometimes, uh trying to do, you know, above and beyond for someone when it's only to one group can be inequitable. Now, pain management is a little fuzzy, right? And so what about if we do a heart thing who needs a cardiac cath, who needs angiography? So, this study looked at 503 cardiologists, they gave them scenarios, um, clinical vignettes about uh different people having chest pain. And they compared um uh they just changed the gender of these patients by changing the pronouns, he or she, right. Um And then they also had them take this, it about associating uh gender and risk taking right now, what they found is that as you showed more uh uh male bias is saying like, well, males are more risk takers, you're going to then get where you have a thinking that uh even this is after controlling for the actual, you know, indication of C ad in the vignette that they're much more likely to say, well, all this smell, he really needs a calf and like, oh, I'm not sure this woman doesn't. So um is it true that uh you know, females can have chest pain? Um and is not a representative of underlying CD? Yes, absolutely. But is it very odd that it correlates well, with um the, the level of implicit bias that these people have by gender? Yes, that's the unusual part. And now all these studies you're seeing are interaction effects because they're taking a different uh you're seeing that the different groups uh have a different effect depending on or your, your level of implicit bias. Let me say, has a different effect depending on which group the patient is a part of male or woman, black or white, right? So that's our general format for these studies. Now, interestingly, they did this also not with just clinical vignettes, but just a stress test. And when it was a stress test, there was no effect for IAT which suggests that where we have firm clinical guidelines or a clearer evidence, then people are not gonna let this come into play because now they're saying like I know what the guidelines say. You have a positive stress test, you need a cap, right? But where there's this more fuzziness, more clinical uncertainty, now we're opening up a door for us to fall back on our uh background thinking. Um and maybe let bias in the door. Now, you might lastly say, well, um well, why do I need to care about this? I don't, you know, I don't treat my patients badly. I don't do this and maybe you don't. But um even if you manage to treat all your patients um perfectly well, um what this suggests is that uh they use in um over 2000 residents had them do again, the racial implicit bias test uh and what they found is that as these residents reported more implicit bias anti-black, they were also more likely to experience depersonalization and burnout symptoms. So, some correlation with their own quality of life. Um And so even for your own sake, um it's worth looking into this problem right now, what kind of lessons can we learn from this? So, first of all, we talked about um in that cardiac cast study that um when do we need to be concerned when there is issues that maybe underlying clinical uncertainty, when it's not 100% clear what to do? That's when we should be more careful about. Are we falling back into um, you know, a habit or routine way of thinking a like a kind of rule of thumb that may not be evidence based and that uh you know, may be biased uh bringing other background impressions and advice. Another thing we need to think about is stressors that we might have. So this Johnson study found that when E er s were overcrowded or when people had higher uh patient loads, then post that shift, they had by the end of that shift would tend to have higher its than before the shift, right on average. So, um in these situations where we're really overworked and working hard again, that's when we really have to watch ourselves that maybe this bias can emerge more and affect how we're treating patients. And then think about like uh what happens when, you know, maybe if you've had a bad encounter with a person of a different gender or different uh sexual orientation, different race. And does that gonna influence the way you deal with them? Next, we always have that um you know, prior case bias going on. And so, um uh think about that. So we've talked about really um some, some different ways that we can really see there is uh evidence for bias that can emerge and how we manage patients by no mean comprehensive. But I hope I've convinced you that both we have some bias in our society and even in the level of the individual patient provider interactions, uh it can influence how we manage our patients. So what can we do about this? Um Well, one thing that's very, and I've tried to stress here is that um we are not looking to, you know, sometimes people want vindication and like I told you so, uh but really, we're not looking to uh make anyone feel bad or say that you're a terrible person. Um These uh just like everyone that commits a medical error or um you know, these kind of things is not a bad doctor. Uh Everyone that has some evidence of bias, even that influences their practice is certainly not a bad doctor. Um This is just means it's something that we can work on and continue to improve because we know you are dedicated to equity, right? Um, but that if you say, well, I don't believe in any of this, like, then you don't have the opportunity to work on it, right? So we don't want to, to, to, uh, you know, convict people. Um, we're not over judging anyone, but we are trying to help everyone, help each other improve as clinicians. Um, now this is a solution, um, perspective taking. Uh, we're running short on time. So I'll just say, um, this is looking at asking people to rate who looks trustworthy and who doesn't. Um, perspective taking is where you put yourself in the shoes of another person and imagine what it's like. They had them actually write an essay. What is it like to, you know, live as like if they're English doctors, what is it like to live as like a Asian person? Imagine you're a Chinese person and then take this test and then rate how uh um trustworthy people look in general before you do that. If you don't do perspective, taking English people tended to English people, other English people as very trustworthy Asian looking people as very not trustworthy. But in when you perspective take you really improve that effect. So that's one thing we can do is try to put yourself in the shoes of that other group that you're not like and see how it can change. Because the idea is you're disrupting those automatic thoughts and those heuristics changing the way you think um and trying to get better, we won't go further into the details of that just for time. Um And then the other solutions we can do are building on that. So, mindfulness thinking about, you know, how am I approaching this problem? What am I thinking? Is there a bias? Like am I reacting maybe negatively or more guarded to this patient because I'm uncomfortable or I've not have a lot of friends in this group, right? Um of this type of person or background, right? Giving yourself internal feedback where you say, well, I know that, you know, in past situations, gonna love it comfortable or I know before I think I, you know, made this call, uh you know, I didn't quite offer all the option to this patient because I had this um you know, preconception about how much, how aggressive a treatment this group wants or doesn't want. Am I doing it again or not? And then reflecting after you get done with a week of service, maybe or after you come home from clinic, how did that day go? Are there places that I think bias emerged and not? And so, um you know, those are things that we can do. So in summary, I'll just show you that uh distinct from overt bias, which I think is not as much of a problem in medicine. Uh There can be implicit bias that arises when we're in um automatic kind of thinking modes or press because of stress because of time, because of business, because of uncertainty, we fall back into these uh kind of patterns of thought that can be good but can also be biased sometimes that this biased decision making can lead to worse quality of care. Um And so what are we gonna do about it? Most of all, we got to disrupt those automatic thoughts, uh do some kind of uh mindfulness reflection perspective, taking this humility, thinking critically about trying to identify your biases um and assess them. And most of all in this work is not just saying, oh, I looked at it one time. I sat in on one talk at one medical about health inequity and disparity and I'm good for the next 10 years, right? But the more you try to engage longitudinally exposing yourself to people who are unlike you trying to understand different perspectives, trying to bring this humility, the better you're gonna get at it. And the more you're gonna be able to limit um uh any bias that might observe uh emerge in your clinical care. And so that way we're all doing our part to minimize these kind of errors while the rest of the system and our society does theirs. And thank you so much for listening to me. Sorry, I ran a little bit longer than playing but um opening up the questions now.
