Implicit bias in diagnosis: gender, race and age
Table of contents
Two people arrive at the clinic with the same complaint, described in almost the same words. One is a forty-year-old man; the other, a seventy-year-old Black woman. The student who sees them swears he would treat both the same way, and he believes it. Even so, the list of hypotheses, the questions he asks and how seriously he takes the complaint can change without him noticing.
NEXO is a daily medical diagnosis game: one clinical case a day, six progressive clues, and ICD-10 proximity feedback across five levels.
This is what implicit bias is about: attitudes and stereotypes that operate outside awareness and can diverge from what a person explicitly believes. This guide explains how it is measured, what reviews show about gender, race and age in health care, and what the student and the preceptor can do to notice their own bias.
Key points
- Implicit bias is an automatic association, outside awareness, between a group of people and an attitude or a stereotype.
- Systematic reviews found implicit bias in health professionals at levels similar to those of the general population.
- In some studies, higher levels of implicit bias were associated with worse communication and different treatment decisions, but the results are mixed.
- In a study with students and residents, half endorsed false beliefs about biological differences between Black and white people, and this was associated with pain assessment.
- Training the skill with practice and feedback reduced bias in pain assessment in an experiment, more than raising awareness alone.
What implicit bias is
Explicit attitude and automatic association
Explicit attitude is what a person says and believes. Implicit association is what the mind links automatically, quickly and without intention. The two can diverge. In a study with 287 internal medicine and emergency residents, participants reported no explicit preference between white and Black patients, but association tests showed an implicit preference for white people (Green et al., 2007).
Different from heuristics
Availability, anchoring and representativeness are shortcuts about the information in the case, described in the guide on examples of heuristics in clinical reasoning. Implicit bias starts somewhere else: from who the patient is. It can change the probability the physician assigns to each hypothesis, as if it changed the base rate with no new data.
Close neighbors
The emotion of the moment, affective bias, also shifts attention, but it has another origin. The diagnostic pause is a procedure that can include a question about implicit bias. And judging an error after the outcome is the subject of hindsight bias. Each topic has its own guide in this series.
How implicit bias is measured
The implicit association test
The most used instrument is the implicit association test, or IAT, which measures reaction time when associating faces or words from different groups with positive or negative concepts. A review of 37 studies on racial bias in health professionals using the IAT found evidence of pro-white or anti-Black, Hispanic, Indigenous or dark-skinned bias in 31 of them (Maina et al., 2018).
Vignettes with the patient swapped
The second method uses identical vignettes in which only one characteristic of the patient changes, such as race, gender or age. A systematic review included this method because psychologists in the field recognize it as a way to detect the possible presence of implicit bias, although on its own it isolates implicit attitude poorly (FitzGerald and Hurst, 2017).
Limits of the measure
The IAT measures association, not behavior. A high score is a risk signal, not proof that the person will discriminate against someone. That is why the most useful studies are those that link the measure to decisions or interactions.
What the reviews show
Levels similar to those of the population
A systematic review of 42 articles found evidence of implicit bias in health professionals in 35 of them, and all the studies that tested correlation found a relationship between the level of implicit bias and lower quality of care (FitzGerald and Hurst, 2017). Of the 42 articles, 27 dealt with racial or ethnic bias, and ten with other biases, such as gender, age and weight (FitzGerald and Hurst, 2017).
Another review, with 15 studies, found low to moderate levels of implicit racial or ethnic bias in almost all of them, similar to those of the general population (Hall et al., 2015). The bias was significantly associated with patient and professional interaction, treatment decisions, adherence and outcomes, although some of the associations tested did not reach significance (Hall et al., 2015).
Mixed results on care
In the 2018 review, among 14 studies with vignettes or simulated patients, eight found no significant association between implicit bias and care and six found an association with differences in treatment recommendations, pain management and empathy (Maina et al., 2018). The seven studies that observed real interactions found worse communication among professionals with stronger bias (Maina et al., 2018).
The picture is this: implicit bias exists in professionals, appears consistently in communication and appears less consistently in decisions. The authors call for more research in real care (Maina et al., 2018).
.jpg){kind=link}
Gender, race and age
Race and decision
In the 2007 study, residents read the vignette of a patient with an acute coronary condition in the emergency department, randomized as white or black (Green et al., 2007). As pro-white implicit bias increased, so did the likelihood of indicating treatment to the white patient and of failing to indicate it to the black patient (Green et al., 2007).
Race, false beliefs and pain
One study examined false beliefs about biological differences between black and white people, such as the belief that black people's skin is thicker (Hoffman et al., 2016). Half of a sample of white medical students and residents endorsed these beliefs, and those who endorsed them rated the black patient's pain as lower and made less accurate treatment recommendations (Hoffman et al., 2016).
The result is useful for teaching because it shows a concrete mechanism: a wrong idea learned at some point, and never revisited, that changes the assessment of a symptom.
Gender and race together
In three studies with 532 participants, using videos of real people in pain, observers disproportionately underestimated the pain of black women compared with black men, white women and white men (Ruben and Stosic, 2024). The effect of the sum of characteristics, and not only of each one, is a reminder that bias rarely acts on a single axis.
Age
A recent essay describes ageism as a silent threat to quality of care: symptoms attributed to "normal aging", missed diagnoses and insufficient treatment (Cheung, 2026). The author points to anchoring on age, instead of individual assessment, as the central mechanism (Cheung, 2026).
What to do with your own bias
Accept that you have it too
Studies show bias in professionals at levels similar to those of the population (Hall et al., 2015). Starting from this finding, and not from certainty of impartiality, is the first step. Metacognition in medicine helps you look at your own reasoning without defensiveness.
Swap the patient in your head
When facing a decision, ask: "would I do the same if the patient were of another gender, another color or another age?". The exercise mimics the design of vignettes with the patient swapped and is quick enough to fit into the routine.
Review learned beliefs
The 2016 study shows that false beliefs about biological differences persist in training (Hoffman et al., 2016). When a rule of thumb links a group to a symptom or to a response to treatment, it is worth looking into where it came from.
Train the skill, not only awareness
In the 2024 experiment, an intervention based on practice and immediate feedback reduced biases in pain assessment more than training that only raised awareness of social or personal biases (Ruben and Stosic, 2024). For preceptors, the lesson is that discussing bias helps, but practicing assessment with feedback helps more. The guide on feedback in clinical cases covers how to give that feedback.
Use the full history
A careful history taking reduces the room for stereotype, because it gives individual data in place of the group's imagined average. The guide on signs, symptoms and history shows the weight of history in diagnosis, and the one on differential diagnosis helps keep the list open.
Reasoning from the data in NEXO
In NEXO, the case of the day arrives in up to six clues, and the player forms a hypothesis from the data that appear, one at a time. Practicing with varied cases helps separate the clinical data from the image one forms of the patient. With each guess, the proximity in ICD-10 across five levels shows how far the hypothesis landed.
The cases are fictional with a real clinical presentation, and the guide what NEXO is explains the whole game. How the modes are unlocked is on the NEXO economy page.
If you coordinate a medical school, a student league or a residency and want to discuss biases with the class, see NEXO Closer. Researchers in medical education can propose studies through NEXO Researcher. Physicians who review cases take part through NEXO Curator, and those who create health content on social media can take part through NEXO Creators.
Frequently asked questions
What is implicit bias?
It is an automatic association, outside awareness, between a group of people and an attitude or a stereotype, which may diverge from what the person explicitly believes.
Do physicians have implicit bias?
Reviews found implicit bias in health professionals at levels similar to those of the general population (Hall et al., 2015).
How is implicit bias measured?
Mainly through the implicit association test and through identical vignettes in which only one characteristic of the patient changes (FitzGerald and Hurst, 2017).
Does implicit bias change diagnosis?
The results are mixed: it appears consistently in communication and less consistently in decisions (Maina et al., 2018).
Can implicit bias be reduced?
One experiment showed that practice with immediate feedback reduced bias in pain assessment more than raising awareness alone (Ruben and Stosic, 2024).
What is ageism in medicine?
It is treating the patient differently because of age, for example attributing symptoms to aging without individual assessment (Cheung, 2026).
Implicit bias checklist
- Start from the premise that you also have bias.
- Swap the patient in your mind before deciding.
- Look for the origin of the practical rules about groups.
- Take the full history, with individual data.
- Keep the list of hypotheses open.
- Practice with feedback, beyond discussing the topic.
- Notice communication, where bias appears first.
Want to train reasoning from the data? Play the case of the day in NEXO.
References
- FitzGerald C, Hurst S. Implicit bias in healthcare professionals: a systematic review. BMC Med Ethics. 2017. https://pubmed.ncbi.nlm.nih.gov/28249596/
- Hall WJ, Chapman MV, Lee KM, Merino YM, Thomas TW, Payne BK, et al. Implicit Racial/Ethnic Bias Among Health Care Professionals and Its Influence on Health Care Outcomes: A Systematic Review. Am J Public Health. 2015. https://pubmed.ncbi.nlm.nih.gov/26469668/
- Maina IW, Belton TD, Ginzberg S, Singh A, Johnson TJ. A decade of studying implicit racial/ethnic bias in healthcare providers using the implicit association test. Soc Sci Med. 2018. https://pubmed.ncbi.nlm.nih.gov/28532892/
- Green AR, Carney DR, Pallin DJ, Ngo LH, Raymond KL, Iezzoni LI, et al. Implicit bias among physicians and its prediction of thrombolysis decisions for black and white patients. J Gen Intern Med. 2007. https://pubmed.ncbi.nlm.nih.gov/17594129/
- Hoffman KM, Trawalter S, Axt JR, Oliver MN. Racial bias in pain assessment and treatment recommendations, and false beliefs about biological differences between blacks and whites. Proc Natl Acad Sci U S A. 2016. https://pubmed.ncbi.nlm.nih.gov/27044069/
- Ruben MA, Stosic MD. Documenting Race and Gender Biases in Pain Assessment and a Novel Intervention Designed to Reduce Biases. J Pain. 2024. https://pubmed.ncbi.nlm.nih.gov/38692397/
- Cheung K. The Silent Threat to Clinical Excellence: Ageism in Medicine. J Gen Intern Med. 2026. https://pubmed.ncbi.nlm.nih.gov/41239116/