Welcome to the 51st Clinical Care Congress of SCCM. Over the course of the next hour or so, I will be speaking to you about how to address disparities and biases in critical care medicine. I'm Dr. Michelle Ngong, Chief of the Division of Critical Care Medicine and Chief of the Division of Pulmonary Medicine at Montefiore Medical Center and Albert Einstein College of Medicine. So just in terms of disclosure, I do have grant support from the AHRQ, NIH, as well as CDC. And I do have some fees for serving on the DSMB board and the scientific advisory panels and committees of various entities. None of them, however, are related to the talk. So over the course of the next hour or so, I'm going to talk a bit about racial disparities in adult critical illness, and then explore a bit about the mechanisms underlying disparities in critical illness with a special attention towards biases, both implicit and the effective biases in clinical decision making, and then end with a section on strategies to try to decrease bias as a way of decreasing disparities. So first, it may be helpful to start with a discussion about the differences between disparities versus biases. Health disparities can be defined as preventable differences in the burden of disease, injury, violence, or in opportunities to achieve optimal health. Disparities is often experienced by socially disadvantaged racial, ethnic, gender, or other population groups. And bias, however, refers to the negative evaluation of one group and its members relative to another. They are not the same, but they are correlated, and one feeds into the other. Disparities can result in biases. For example, disparities in terms of obesity in certain racial groups may lead to some biases about that racial group tending to be more obese and susceptible to overeating, for example. Whereas biases can also lead to health disparities. Assuming that women, for example, may be more likely to report shortness of breath or pain from anxiety rather from a physiological entity may lead to actually disparity in terms of treatment for pain. Racial disparities has been described in a number of different critical illnesses as illustrated by this word diagram. The existence of these racial disparities in critical illness is proven and pervasive, and it's not so important anymore to keep documenting these disparities. Rather, we need to actually shift towards effective interventions to decrease disparities. But to be able to devise effective interventions means that we need to have an understanding of the mechanisms of disparities, because that's actually how you're going to end up devising the interventions to target. So, for example, let's talk a little bit about racial disparities in severe sepsis, which has been well documented. We know, for example, that black patients were more likely to die of severe sepsis than whites. In fact, there's anywhere between a 1.5 to 1.8-fold higher mortality. But what is actually the mechanisms that may be driving that increased mortality in severe sepsis? Because when you look at actually different studies on development of severe sepsis and outcome in severe sepsis, you realize that there actually are different mechanisms at play. For example, in this study, you can see actually that black patients as compared to whites tend to be younger on presentation. They tend to have different types of infections. For example, black patients are more likely to have respiratory infections than white patients. And whereas, actually, when it comes to similarly general urinary infections are more common in black patients than white patients. And black patients also tend to have more comorbidities, like diabetes, chronic kidney disease, on presentation. But even after they have developed an infection, there's a difference in how black patients may present compared to white patients. Among patients who are hospitalized with infection, if you look over here in terms of age and in terms of those who have severe sepsis, you see that across all age groups, black patients are more likely to have severe sepsis with infection compared to actually non-black or white patients with infections. In addition, if you look at the organ dysfunction as stratified by race among patients who are hospitalized with severe sepsis, you notice actually that there's statistically significantly more organ failures occurring among black patients with severe sepsis than white patients with severe sepsis. So not only are they having different types of infections, different comorbidities, but they're also actually presenting with greater severity of illness as well as organ dysfunction. A good example of these different stages of development along continual critical illness can be found in looking at the disparities in patients with COVID-19. We know actually, if you look at these numbers of deaths per 100,000, that there is a strikingly higher mortality from COVID among black patients and Hispanic patients compared to whites. But this is a fascinating study, actually, where they looked at the entire course of developing COVID-19 from susceptibility to having an infection to likelihood of being hospitalized to dying from it. And what you can see here is U.S., whereas you might actually see a higher percentage of patients who are hospitalized who die from COVID-19 among black patients compared to white patients, you don't quite see the difference in Hispanic patients. The biggest disparity is not actually in terms of mortality and from COVID-19. It is actually in terms of likelihood of coming down with COVID-19 to begin with at the infection stage, where black patients and Hispanic patients have two to three times greater percentages of patients coming in with a COVID-19 infection. Similarly, actually, the other large disparity we see is in whether or not these patients get admitted to the hospital. A much higher percentage, more than double the percentages of those patients with COVID-19 who are white gets hospitalized for COVID-19 if you're black or Hispanic. So the largest disparity that we see in COVID-19 is not necessarily from death, but from the development of infection and hospitalization. So when you talk about actually the mechanisms of health disparities, it's also important to think about the mechanism along the continuum of acute critical illnesses, of which actually health disparities have been reported along the entire continuum from actually disparities in terms of immunization rates, susceptibility to infections, or comorbidities that put you at a higher risk for infections. As well as different types of infection. Two, as we talked about actually how they present in terms of severity and illness, organ failure, and which types of hospital certain patients from different racial, ethnic populations may be presenting to. To what happens in the hospital, whether or not they get access to certain treatment, to communication about end of life, to complications, to transfers, to higher acuity long-term care, to the ultimate outcome of whether or not they survive from the critical illness or not. So if we want to be able to reduce disparities in acute critical illness, it's important that we devise interventions that target the right mechanism. And as you can see here, there are a number of different mechanisms that has already been identified that contribute to disparate health outcomes. From societal and institutional factors that lead from like socioeconomic issues, healthcare system, racial prejudice, to individual social contacts and relationship factors related to the community, to being access to healthcare, to employment, to insurance status, to individual factors like, you know, the health status of the patient, like their health behavior, their mental health, and so on. To biological factors, such as genetics, prior illnesses, medications that may actually predispose them to one thing or another. It is important that when we devise interventions, we devise the interventions with regards to the mechanism, because it tells you what interventions to test, who you should be targeting in the intervention, where and when the intervention should be implemented, and what you should be doing. So here's where I'm going to shift a little bit and talk about actually individual-based versus community-based versus hospital-based mechanisms for racial justice. And while yes, there is definitely overlap between these different groups, it's also actually important to understand whether or not we're talking about disparities that exist in the community. So we have an understanding of where all potential roles might be and where we could potentially intervene to decrease disparities. For example, under the individual base, there is known genetic polymorphisms that has been linked to increased susceptibility and worse outcome of sepsis in patients who have African-American or African-American parents. That is a very genetics or individual-based mechanism for racial disparity and sepsis, but one that not necessarily will be able to change, except for maybe having a better understanding of what's going on in the community. And then there are community-based mechanisms for racial disparity, by which, in order to address, we might have to look at, for example, how we might be able to address racial disparities in patients who are African-American or African-American parents. And then there are community-based mechanisms for racial disparity, by which, in order to address, we may need to actually address some of these other factors, such as preventive care, environmental factors, hospital resources. One example of that is this relationship that has been found between medically underserved areas and increasing incidence, as well as increasing mortality in sepsis, where you see, actually, that there is a significant differences in terms of incidence and mortality in sepsis. among those patients who develop sepsis in what's called medically underserved area, which is defined by low access to primary care physicians, high infant mortality, high poverty level, and a high percentage of births. And this is true regardless of race or gender. And that reflects, actually, a community issue with regards to resources, such as, for example, health insurance. We know the lack of insurance is a big indicator of disparities, even in critical illness. When you look at, actually, predictors with regards to what disparities are, you can see that there are a lot of disparities in health insurance. And this is true regardless of race or gender. We know the lack of insurance is a big indicator of disparities, even in critical illness. When you look at, actually, predictors with regards to relationship to lack of insurance, we find that the lack of insurance is related to increased mortality in severe sepsis, greater likelihood that the patient will present in certain types of hospitals, such as actually presenting in a government hospital. They also tend to present with greater severity of illness, such as more organ failure, and also get certain interventions like mechanical ventilation more often, but less likely to get other interventions, like transfusions or central line placement or new dialysis or TPN. And the lack of insurance means the patients are less likely to be discharged to a skilled nursing facilities or with home health care, and are much more often to be sent home after critical illness without any support. And we know, actually, that this also translates to increasing mortality, even adjusting for other factors, such as age, gender, and race, for example. So compared to having no insurance, any kind of insurance, private, Medicare, Medicaid, any other type of insurance, is associated with lower mortality, even after adjusting for multiple things like age, gender, race, zip code, how sick they are, comorbidities, what kind of procedure, and even clustering within the hospitals themselves. And that is a community-based or societal-based issue that contributes hugely to health disparities that we see in critical illness. But over the rest of the talk, actually, what I want to do now is shift from some of these community-based mechanism, of which many of us may not necessarily have within our power to change, and focus more on the intersection between the individual base and the hospital base that we may have greater ability to intervene on and reduce disparity. And one of the ones I want to focus on is the role of bias. Bias in critical care medicine most commonly exists on this intersection between the individual and hospital-based level, where the patient interacts with the healthcare professional. Remember, bias is defined as the negative evaluation of one group and its members relative to another. But there actually are two kinds of biases. There is the explicit bias, where a person is aware that their evaluation of the group differs from other groups, but believes that it is correct in some way and intends to act on it in its current situation. They know that they're treating one group differently than the other, but they believe that that is justified, and they will intend and continue to do so. This kind of explicit bias has significantly declined over the last 50 years in terms of report. But what we're seeing more of is implicit bias. And this is especially true because we now have validated instruments to test for implicit bias. And implicit bias, unlike explicit bias, is unintentional and unconscious. The person is completely unaware of this bias and doesn't realize that they're treating one group differently or see one group differently than another. This bias is often activated quickly, without any thinking, without any deliberation, and unknowingly to the person themselves. And it's usually triggered after some kind of a situation cue, whether it be gender, skin color, dress, language, how they act. And that kind of cue is oftentimes unconscious to the individual. They didn't realize that that had triggered something. Now, this concept of health care providers' susceptibility to bias, especially unconscious bias, is not new to medicine. In fact, we know from decades of literature that bias is common in medical decision making. Just look at the word document and you'll see, actually, all the number of different types of biases that has been found and measured to be prevalent with regards to clinical decision making. And these kind of biases, like anchoring biases, availability bias, confirmation bias, these are oftentimes common lexicon of clinical decision making. These are oftentimes common lexicon that we're all familiar with, with regards to cognitive errors and errors in diagnostic and management of patients. The other thing about bias that's relevant to critical care is that bias is particularly common and easy to trigger in acute time-sensitive situation where there's limited information and you need to be able to make a decision very rapidly. So in critical illness, we are particularly susceptible to these kind of cognitive biases. So by that manner, we should not assume that we are above having any kind of implicit bias and how it might influence our medical decision making and our management towards patients. And this has been studied in medicine. So as I had mentioned, actually, explicit bias is self-reported, the person is conscious. And you can see, actually, in terms of measures of explicit bias, it's generally a low prevalence. But we now have instruments that look at implicit bias that many of you may be already familiar with, that basically forces you to rapidly answer questions and associations with certain words that have both positive and negative connotation with either overweight people or underweight people to look at weight bias or racial groups or gender, male and female. And using these kind of implicit bias tools, so using the IAT, we can measure, actually, the presence of unconscious bias or implicit bias. And we find that that is a much higher percentage found among healthcare providers. And now we also have studies that have shown that these kind of implicit bias can lead to poor care. So this is a study, actually, that look at, actually, scenarios of patients who present with chest pain and acute MI. And looking to see, actually, where clinicians will respond with regards to treatment with thrombolysis. And then also, actually, having these physicians participate in the IAT to determine whether or not that there may be some unconscious implicit bias. What they found was that among patients who are African-Americans, so in this dotted line here, we find, actually, that there is a difference in terms of whether or not the clinician will recommend treatment with thrombolysis based upon, actually, whether they have a low or high implicit bias to heart rate. So clinicians who have a low implicit bias will have a higher likelihood of saying that black patients should be treated with thrombolysis. Whereas, actually, physicians who have a high implicit bias will be more likely to say that black patients should not get treated with thrombolysis. And the opposite is actually true among white patients. White patients are more likely to be said to need thrombolysis among those clinicians that have high implicit bias. So the greater the implicit bias, the higher the likelihood to treat with thrombolysis for acute MI if the patient is white, but lower if the patient is black. And these kind of biases, what does that mean? So with regards to critical illness, this is further compounded by the fact that actually some of our medical devices, like the pulse oximeter, may not have been validated on all types of patients with different skin color. So in this study by Michael Salinger, he has demonstrated that undetected hypoxemia is common, but more common among black patients than among white patients. But that does not necessarily mean that there is disparity that comes from it. Because even though that is true and that there may be some indication, actually, that our devices are not necessarily well validated for all types of patients, we don't yet know if undetected hypoxemia can lead to differences in escalation respiratory support, intubation, ICU admission, or outcomes for acute respiratory failure. Because remember, the pulse oximeter is only one data point in our clinical assessment about whether or not, actually, to move to a higher respiratory support or to intubate a patient. And sometimes, actually, it requires us to see what the patient reported symptoms of shortness of breath or distress or pain is. But that actually is more concerning and more susceptible to bias. Take, for example, pain. This has been found in multiple studies in which they've seen that there's differences in terms of how physicians opt to treat for pain depending on the race of the patient and whether or not the physicians have a high or low implicit bias with regards to race. So if you look at oxycodone, which is on the left here, and you look at the African-American patients, you see, actually, clinicians who have a high implicit bias for race are much less likely to agree to give oxycodone to a Black patient in pain than clinicians who have a low implicit bias. And this is much different, significantly different, than if the scenario involved a White patient. Whereas, actually, if you look at the issue about giving ibuprofen for pain, you see less of a difference between the African-Americans and the White patients and among clinicians who have high or low implicit bias. So with regards to hypoxemia, the other factors that are symptom-based or patient-based may also be susceptible to biases. In addition, implicit bias can also lead to poor communication, and this is considerably more concerning in acute critical illness, where we may need to actually make some difficult decisions with the patients and with the family about goals of care, about treatment, about the balances of risk and benefit. So this was a study in which they looked at, actually, a clinician's implicit attitude about race, and they were looking to see, actually, by observing and taping their interaction, how they interact with patients. And they also actually looked at how the patients regarded that interaction. What they found was that clinicians who have more implicit bias against Blacks were also associated with an increased dominance of the conversation by the clinicians for both black and white patients. So clinicians who have high implicit bias against blacks tend to dominate the conversations more with their patients in regards to place. And this is actually sensed by the patients themselves, even if it's unconscious on the provider level. And more implicit bias in a clinician means their patients have a lower probability of perceiving that there was respect from that clinician. They're less likely to like the clinician, they're less likely to have confidence in the clinician, and they're less likely to recommend the clinician to others. They can also look at actually whether or not clinicians may have some bias against race and medical compliance, for example. Whether they have unconsciously believe that certain patients, whether it may be black patients, for example, are going to be just more non-compliant with medications. What they found is this kind of bias between race and medical compliance is associated with less dialogue that's been centered around the patients, but among only the black patients and not with the white patients. And again, this is sensed by the patients themselves. With black patients rating lower trust and lower confidence in the clinicians as a result. So in the next half of this talk, I wanna transition to talking less about actually that disparities exist in critical illness and that this bias exists in critical illness, and bringing those two things around to talk about actually strategies for reducing bias in critical care medicine and how that we can then take responsibility in trying to reduce some of that disparities by focusing on reducing bias. So just as we had talked about actually, when we look at health disparities and mechanisms of health disparities, you need to look at it along the entire continuum of acute critical illness from when it is in patients who are susceptible to when they develop it, to how they develop it, and ultimately to whether they survive it or not. It is also actually important to think about opportunities to decrease disparities by reducing biases, by focusing on biases among the clinicians, among clinical leaders within the institution and when the institution itself. So let's start a bit with the clinicians. So this is a very interesting study, partly because it was an experimental trial looking to see how we can reduce biases among clinicians in recommendations for best care. So what they did was they took a study of about 840 clinicians and they exposed all these clinicians to a video, like a set video with a scenario of a patient with heart disease. And after the first time they showed the video, they asked what would they recommend for this patient, both in terms of diagnostic assessment and also in terms of management to see whether or not they were adhering to evidence-based best practices in terms of recommendations for management for typical patients with this video, or that they deviated from that evidence-based practice. Then they showed a video two more times for a total of three. But after the second time and the third time, they will ask them if the clinician, if they wanna revise their prior recommendation on heart risk and follow-up care. Then they actually randomized these clinicians to a video with the same scenario, whether it be a white male or black female, so they'll randomize that. And they will also randomize to getting some kind of a peer networking on the decision-making after the first video, meaning half of the clinicians after their first video will be told what their peers had recommended and what the peers had suggested for the management of that patient, and then ask if they wanna revise their prior recommendation. And they did that on the second and the third time, whereas half the clinicians did not get feedback in terms of what their peers were recommending. What they found was, was that when you look at actually the diagnostic assessment and what is compliant, which is the higher number to what's recommended and what's lower, you see actually that given feedback on peer networking from the group significantly improved compliance with evidence-based practices in terms of diagnostic assessment for this patient scenario. And this improvement basically eradicated any disparity that you could see among the black patient and the white patient that you saw under those patient clinicians who did not get that peer review. And the same thing actually, when they looked at management, a follow-up management, when whether or not they are adherent to evidence-based guidelines. In those clinicians that was randomized to control, so no peer networking, you see actually that there were big differences between likelihood of black patients being much less likely to get evidence-based guideline care, concordant care, compared to the white patients. But after actually network peer feedback, you see actually that after the second and the third, and after the peer review before the second showing of the video, you already see improvement in terms of the black patients in terms of more adherence to evidence-based practices and narrowing of the difference between the two groups. And again, if you look at it on the flip side, looking at guideline discordant care, you see something similar. And the network conditions where they give peer networking, you see after the first video, a big drop in terms of a guideline discordant care and obliteration of any disparities between the black and white patients. Indicating actually that understanding how your clinical decisions may be made in the context of evidence-based practices as well as your peers may actually help reduce differences that might be seen from any kind of bias or gaps in knowledge or education. So from a clinician level, we have opportunities to decrease disparities among all of us by reducing bias. First, take responsibility about how we interact with patients. Be aware of the possibility of implicit bias in ourselves. Just as like we should be aware that we may be susceptible to anchoring bias, availability bias, confirmatory bias, any of the other biases that lead to cognitive errors in medical decision-making, be aware that we also may be susceptible to implicit bias. Minimize the focus on race and gender and focus instead on the individual. It's really necessary to identify them in your notes and our presentation as a African-American male when it's maybe not relevant to their presentation and management. Learn to take the patient's perspective. Seeing things from their point of view can completely change our attitudes and taking that practice of learning to see from their point of view can have been shown in other instances outside of healthcare to reduce implicit bias. And based upon that study I told you about, practice evidence-based practices, but also actually be aware of how your practice may be in alignment or differ from your peers, from national guidelines, from what's doing around the country. So let's shift now to opportunities to decrease disparities by reducing bias among clinical leaders. So here I wanna talk a bit about the gender and racial imbalance in critical paramedicine. If you look at actually women representation at the undergraduate level and even at the med school level, you see actually women constitute almost 50, 55% of the medical school class and undergraduate. Where you start to see a decrease is where you start looking at the pipeline and residency to critical care medicine. So for example, internal medicine, surgery, anesthesia, as well as actually critical care fellows gender. Whereas over the last 10 years, 10, 20 years, you do see an increase in terms of female representation in the pipeline to critical care and in the critical care fellows. You see that that's still lags behind the female representation in medical school. And if you look actually at say black applicants for critical care medicine, as well as actually black patients, you'll see that there is a gross under-representation of black students in medical school, as well as actually really low numbers in the pipeline into critical care, as well as actually critical care fellows. And that over the last decade or two, this has not changed very much at all. Representation matters, okay? Because given the health disparity that we see, it's important to reduce the disparities by having the healthcare providers look like and can empathize and understand what their patients are going through. It's also been known actually that implicit bias and cultural competency improves, there's less implicit bias and more cultural competency when you have greater diversity and greater representatives of minorities on the treatment team. So when you look at actually IAT, you find actually that there is a fair prevalence of IAT among different races, but much less so among the African-American students. So one way to decrease bias and health disparity may be actually increasing representation within critical care medicine. But why, for example, are there less representations of women and minorities in critical care medicine? Unfortunately, there are some studies indicate that there may be a disparate evaluations of trainees in the critical care training path. So this one study, which came out of Canada, founds that female trainees as general get a lower score on their evaluation and their ICU rotation, which will invariably translate to less likelihood these candidates will wanna pursue critical care and also with less likelihood they'll match in a competitive program. Now, the good news is that this lower of scores among female residents is not universal. And it was driven mostly actually by the anesthesia residency program and the anesthesia residents was rotating through the ICU. So indicating actually that there may be an opportunity here to focus on the problem area and solve it. But it is not just in the training process in which we see a disparity in terms of representation, which may translate to how we deliver care to outpatients. If you look at actually authorships and publications, as well as an authorship and clinical practice guideline, which is by all count the most influential with regards to dictating clinical care and management of patients. You see that there is no increase in the percentage of female authors leading some of these papers. And if you look at it by specialty, critical care medicine, which is up here, ranks among the bottom six subspecialties in terms of representation of female members, authors on the practice guidelines. With only about 20% of the authors being female. So I wanna actually part with some suggestions for clinical leaders in critical care medicine to try to avoid bias during mentoring and to try to use that as a tool to increase representation, increase recruitment of minorities and women into critical care medicine. One is be aware of your own bias as a mentor. How do you write letters of representation? How do you write letters of recommendation? How do you evaluate your residents? What standard are you holding up to? Make your compensation plan, your fellowship selection process, your promotion metrics, and your recruitment criteria open and transparent so that everybody can see what are the things that you're using to evaluate selection of these patients. Of these candidates, how they're paid. Not just for you, but for them, so they know how they can reach up to that standard. One of the things that I oftentimes see that I try to actively discourage is where feedback about a new candidate for a fellowship, a new candidate for a job, the feedback from attendings that says, well, it may not be a good fit. I don't know what that means. And that kind of vague terminology can be very prone to bias. I don't know why. What is the reason for that? What is it that would not make them function well in this environment? You need to actively offer opportunities to the underrepresentative individuals on your team. They don't get the opportunities unless you offer them opportunities to them. And that sometimes actually, they will turn down your opportunities. So you need to work harder on convincing your female and minority mentees that they are capable of and should consider these new challenges and opportunities. Remember, female and underrepresented minorities are much more susceptible to the imposter syndrome in which they think that they haven't accomplished enough or that they don't have the criterias necessary to get to the next stage or that they don't necessarily deserve or have earned the praise that they've gotten. So you need to work harder on helping them get past that and to recognize that what they, not only what they have done and their achievement, but what their potential is. And then after they have accepted those opportunities for advancement or challenge, you can't just say you're done. You have to actually also offer the tools, the additional training, the mentorship that will help them succeed with those opportunities. And then when they exceed, celebrate the work, the quality, the leadership of your mentees, but don't celebrate that they did this as a black candidate. Celebrate just their achievements and what they have done, not their race or their gender. And you as mentees also has a responsibility. You should ask for and accept new opportunities, even if they're scary. Take some time to think about these new opportunities and challenge. Think about what it is that will help you meet that challenge and make it successful and ask for it. Ask for the training, the mentoring, the tools that will help you succeed with these challenges and opportunities. Seek multiple mentors from multiple places, okay? Because you're going to need guidance and different perspective. And mentors should be understanding that you're going to need different mentors from different places. Educate yourself on salary, promotion metrics. I just said that the mentors should be transparent. The mentees do their homework in understanding what those processes is so they know where they stack up and how they get to the next stage. And then you need to sponsor each other. You need to support each other. Celebrate your underrepresented colleagues for their achievements rather than their representation. So we talked a little bit about how clinical leaders have opportunities to decrease disparity, mostly by increasing diversity, inclusions, but also actually be transparent about it. Collect the data on your diversity and equity. Be transparent, increase the mentoring, actively offer opportunities. What about the institution itself? Institution itself actually could be a big driver in terms of disparities. When you actually look at some of the outcomes with regards to race and say mortality and sepsis, for example, you find actually, and like this particular study is one example of it, that black patients, for example, and Hispanic patients may have a higher mortality rate than say white patients after adjusting for patient characteristics. But it turns out actually that this might be due to differences in terms of the type of hospitals that black patients go to versus actually white patients. So if you were to actually account for both the patient characteristics as well as actually the hospital characteristics and whether or not the patients actually are treated in one type of hospital or another, those differences disappear. And comparison actually within the hospitals themselves by race no longer matters. So here's another example. This is looking at actually the quality of care for patients admitted to the hospital for pneumonia. So compared to whites, black patients were less likely to get timely antibiotics and less likely to get the right antibiotics, but ironically enough, more likely to use mechanical ventilation. However, if you look within the hospitals themselves, because it turned out to be actually great hospital variability in terms of percentage of their patients who get timely antibiotics, correct antibiotics, who use mechanical ventilation. So if you look within the hospital itself, there is no differences among patients by race. And if you adjust for the case mix of the patients, how sick they are, as well as the effect of what hospital they were admitted to, differences in terms of quality of care and pneumonia treatment disappears between the races. So there is opportunities then for improving the quality of care, the delivery of care to patients in a more systemic fashion so as to reduce disparities and biases. And this is true even of research because currently actually the more well-resourced hospital systems and academic centers, which tend to have a population that is less likely to be underserved, they're also more likely to be able to get funding for research and resources to be able to improve quality and implementation of best practices. And to change this, we actually need to have more funding to reduce bias and variability in care practices. That means an increased federal funding to safety net hospitals so that they will have the resources like the technology, the data science, the research, the clinical trials to be able to actually improve and leverage the kind of care they can have and improve the quality of the care that they give. They need to actually give incentives to safety net hospitals to do research. You know, one of the things is about research grants is the whole section that you get graded on in terms of environment and whether the environment is conducive to research and will support the research. Unfortunately, lower resource institutions that tends to serve the underserved and underrepresented also have more minority patients, more minority faculty that has a lower resources so tend to score lower on the environmental component for research grants. And they also have less resources to attract the star faculty and research to come and do the research there and to build the infrastructure. So it becomes this terrible cycle by which safety net hospitals are always going to be disadvantaged in being able to generate the knowledge and the resources available to change care and outcomes in their patients. Institutions have an opportunity to reduce disparity by promoting quality and evidence-based practices, by creating a system that will identify and address bias and discrimination, by recruitment to increase diversity, and by creating an institutional culture of respect and inclusion and equity. And they have to take a stance against bullying because bullying is prevalent in the hospital system and oftentimes women and minorities are suffers from that bullying. So in conclusion, racial disparities in acute critical illness do exist and it is pervasive, unfortunately. Mechanisms underlying the disparities have to be defined in order to find the effective interventions. But one of the mechanisms that has been well described stands at the intersection between the patient, the hospital and the provider is presence of bias, both in terms of our cognitive decision-making, but also actually in terms of implicit bias against race, against gender, against entities like elderly or overweight people, all of which can contribute to disparities in critical care delivery. But there are strategies for reducing bias at the clinician level, at the clinical leader level, at the institutional level and the research level. We can start with ourselves, be conscious of the possibility of bias in ourselves and in our system and its effect on the patients and on their colleagues. Be purposeful in avoiding bias, be purposeful in terms of placing ourselves from the perspective of the patient so that we can better understand where they come from and better understand actually where biases may leak in even as we are unconscious about them. So with that, I wanna thank you for your attention and I hope you enjoy the rest of the conference. Thank you.

disparities

biases

critical care medicine

racial disparities

severe sepsis

COVID-19

clinical decision-making

implicit biases

reducing bias