The Many Faces of Bias and Their Impact
on Medical Outcomes
Fredrick Wans submitting for:
“What is bias and how can it affect patient care?”
Common consensus suggests the most apt description of bias to be a tendency
towards prejudice or a variable that impedes removing prejudice.1,2 Generally divided into
cognitive and social types, it is notably present in all humans. In medicine, specific
prejudices act directly, leading to sub-optimal treatment due to flawed decision-making,
discrimination and issues with patient compliance, or indirectly, by hindering research efforts.
Thinking About Thinking
Cognitive biases are non-pathological systematic errors in thinking that prevent
logical/objective thought due to the brain’s tendency to process information based on
preferences and patterns common to it.3 During the COVID-19 pandemic, they played a role
in unnecessary testing, policies and the use of uncertain medications such as
hydroxychloroquine - an example of action bias (the tendency to choose action over
inaction), as its significant side effects were neglected (e.g. heart failure)4 in favour of its low
probability of success. In addition to altering decision-making, they plague research
methodology in the unequal treatment of study groups (therefore altering participant
behaviour and consequently results), selectively reporting only significant outcomes or
unjustifiably omitting some participants. Even greats like William Halsted, whose vehement
1
Delgado‐Rodríguez and Miguel, “Bias.” Journal of Epidemiology and Community Health 58:8 (2004): 635
D. P. Gopal and others, “Implicit Bias in Healthcare: Clinical Practice, Research and Decision Making.”, Future Healthcare
Journal 8, no. 1 (March 2021): 40–48.
3
G. Ellis, “So, What Are Cognitive Biases?,” in Springer eBooks, (2018): 1–10.
4
Alanagreh, Lo’ai, F. Alzoughool., and M. Atoum, “Risk of Using Hydroxychloroquine as a Treatment of COVID-19.” The
International Journal of Risk and Safety in Medicine 31:3 (2020): 111–116.
2
belief in the unsupported ‘surgical radicalism’ concept unnecessarily disfigured many
women, fall to cognitive bias.5
People-Prejudice: “All animals are equal, but some animals are more equal
than others” (George Orwell, Animal Farm)
On the other hand lie ubiquitous implicit (subconscious) or explicit (conscious)
prejudices directed at certain groupings such as ethnicity, race, gender, weight and age.6
Whilst it is illegal for care providers to deliberately mistreat patients, these biases often
operate subconsciously. Drug abusers being denied necessary treatment under the umbrella
justification of “drug-seeking behaviour” is a classic example.7
Bias may also be patient-side, such as refusing treatment from providers of a specific
social group, with studies showing up to a quintupled chance of refusal in some instances.8
This may lead to delayed treatment, which can have a negative prognostic outcome in
emergencies9 or might reduce treatment efficacy by the nocebo effect.10 Finally, insurance
companies have historically used discriminatory standards of "risk" that result in weaker
policies, higher prices and even coverage denial for those deemed "high risk".11 US-based
“Cigna”’s claims that the use of AI in making coverage decisions would ‘improve health
outcomes12 are outright ridiculous. AI training on biassed data can only ever result in biassed
outputs: the “garbage in garbage out” phenomenon.13 Insurance companies claim that their
discriminatory risk calculations accurately reflect the real world, but this is questionable.
5
Mukherjee, Siddhartha The Emperor of All Maladies: A Biography of Cancer, (2010) 78-87
Banaji, R. Mahzarin., C. D. Hardin, and A. J. Rothman. “Implicit Stereotyping in Person Judgment.” Journal of Personality
and Social Psychology 65:2 (1993): 272–281.
7
Carusone and others, “‘Maybe If I Stop the Drugs, Then Maybe They’d Care?’—Hospital Care Experiences of People Who
Use Drugs.” Harm Reduction Journal 16:1 (2019).
8
R. Waldron. and others. “Effect of Gender on Prehospital Refusal of Medical Aid.” The Journal of Emergency Medicine
43:2 (2012): 283–290.
9
Hede, Karyn, “Emergency Medicine: The Need for Speed.” Nature 503:7475 (2013): S14–S15.
10
Tu, Yiheng, et al. "Placebo and Nocebo Effects: From Observation to Harnessing and Clinical Application." Translational
Psychiatry, 12:1, (2022): 1-9, [https://doi.org/10.1038/s41398-022-02293-2/ accessed 4 Jan. 2024]
11
Avraham, Ronen, "Understanding Insurance Anti-Discrimination Laws." S. Cal. L. Rev. 87:2 (2014): 195-274.
12
Cigna Uses Data and AI to Improve Patient Outcomes. Cigna Healthcare Newsroom, n.d.
[https://newsroom.cigna.com/how-cigna-uses-data-and-ai-to-improve-patient-outcomes]
13
The Role of Data in AI, J. Tennison, M. B. Jančič, November 2020
[https://gpai.ai/projects/data-governance/role-of-data-in-ai.pdf]
6
To what extent can we modify bias?
Legally enforced by some American states, “implicit bias training” seeks to quash the
source by recalibrating medical students’ mindsets, but does it serve its intended function?
Although these interventions lower IAT scores (the most common measure used), they do not
deal with the crux of the issue.14 To thwart bias, a system overhaul is required; one entirely
out of the budget of most developing countries, where funds would be better spent on new
medications, enhanced sanitation and public medical education.15 It’s therefore not
unreasonable to ask: Is “How does bias affect patient care” even the right question? People in
power might even say that if it is impractical to prevent the result of a lesser issue, why focus
on it at the moment?
In short, bias negatively impacts patients at all levels, from fouling doctor-patient
interactions to worsening decision-making to pitfalls in research to invisible discrimination.
Affecting some more than others, deciding if remedying it is a priority is time and
location-specific.
14
Green, L. Tiffany, and N. Hagiwara, “The Problem with Implicit Bias Training.” Scientific American, (2020):
[https://www.scientificamerican.com/article/the-problem-with-implicit-bias-training/]
15
[https://www.jica.go.jp/Resource/jica-ri/IFIC_and_JBICI-Studies/english/publications/reports/study/topical/health/pdf/healt
h_02.pdf]
Bibliography
1: Delgado‐Rodríguez and Miguel, “Bias.” Journal of Epidemiology and Community Health
58:8 (2004): 635
2: Gopal, D. P. and others, “Implicit Bias in Healthcare: Clinical Practice, Research and
Decision Making.”, Future Healthcare Journal 8, no. 1 (March 2021): 40–48.
3: Ellis, G., “So, What Are Cognitive Biases?,” in Springer eBooks, (2018): 1–10.
4: Alanagreh, Lo’ai, Alzoughool, F., and Atoum, M., “Risk of Using Hydroxychloroquine as
a Treatment of COVID-19.” The International Journal of Risk and Safety in Medicine 31:3
(2020): 111–116.
5: Mukherjee, Siddhartha The Emperor of All Maladies: A Biography of Cancer, (2010)
78-87
6: Banaji, Mahzarin, R., Hardin, C. D., and Rothman, A. J., “Implicit Stereotyping in Person
Judgment.” Journal of Personality and Social Psychology 65:2 (1993): 272–281.
7: Carusone and others, “‘Maybe If I Stop the Drugs, Then Maybe They’d Care?’—Hospital
Care Experiences of People Who Use Drugs.” Harm Reduction Journal 16:1 (2019).
8: Waldron, R., and others. “Effect of Gender on Prehospital Refusal of Medical Aid.” The
Journal of Emergency Medicine 43:2 (2012): 283–290.
9: Hede, Karyn, “Emergency Medicine: The Need for Speed.” Nature 503:7475 (2013):
S14–S15.
10: Tu, Yiheng, et al. "Placebo and Nocebo Effects: From Observation to Harnessing and
Clinical Application." Translational Psychiatry, 12:1, (2022): 1-9,
[https://doi.org/10.1038/s41398-022-02293-2/ accessed 4 Jan. 2024]
11: Avraham, Ronen, "Understanding Insurance Anti-Discrimination Laws." S. Cal. L. Rev.
87:2 (2014): 195-274.
12: Cigna Uses Data and AI to Improve Patient Outcomes. Cigna Healthcare Newsroom, n.d.
[https://newsroom.cigna.com/how-cigna-uses-data-and-ai-to-improve-patient-outcomes]
13: The Role of Data in AI, J. Tennison, M. B. Jančič, November 2020
[https://gpai.ai/projects/data-governance/role-of-data-in-ai.pdf]
14: Green, L. Tiffany, and N. Hagiwara, “The Problem with Implicit Bias Training.”
Scientific American, (2020):
[https://www.scientificamerican.com/article/the-problem-with-implicit-bias-training/]
15:[https://www.jica.go.jp/Resource/jica-ri/IFIC_and_JBICI-Studies/english/publications/rep
orts/study/topical/health/pdf/health_02.pdf]