Editor's Pick: Algorithmic bias in lung cancer screening eligibility
Algorithmic bias in lung cancer screening eligibility
Manful A, Mercaldo S, Blume JD, Aldrich MC. Addressing algorithmic bias in lung cancer screening eligibility. J Natl Cancer Inst. 2026;118(2):343-353. doi:10.1093/jnci/djaf29
JNCI: Journal of the National Cancer Institute
February 1, 2026
__
Q&A Video Featuring Anand Narayan, MD, PhD
Watch the full Q&A here.
The teaser video can be found here.
Anand Narayan, MD, PhD
Breast Imaging Radiologist, Professor, and Vice Chair of Health Equity
Department of Radiology, University of Wisconsin
Anand grew up in Baltimore, MD, and completed his dual degree at the Johns Hopkins School of Medicine (MD) and the Johns Hopkins School of Public Health (PhD, Clinical Epidemiology). He completed his radiology residency at Johns Hopkins Hospital and pursued a breast imaging fellowship at Memorial Sloan Kettering Cancer Center from 2016-2017. He then became an assistant professor at Massachusetts General Hospital until 2021, when he served as the Department of Radiology's Diversity and Inclusion Officer. His clinical, research, and administrative interests focus on equity, diversity and inclusion, with a particular focus on reducing breast cancer disparities through early detection and diagnosis. He has published 110 peer-reviewed journal articles and three book chapters, and he serves as a reviewer for several radiology journals. He currently serves on the editorial board of Radiology and the Journal of the American College of Radiology and is an associate editor for health policy and practice for the journal Radiology, and is an assistant editor for the Journal of the American College of Radiology and the American Journal of Roentgenology. He was selected as one of the National Academy of Medicine’s Scholars in Diagnostic Excellence in 2023. He currently serves as chair of the American College of Radiology’s Patient and Family Outreach Committee and president of the Wisconsin Radiological Society.
__
(Note: The responses below are highlights from the Q&A video above from Dr. Narayan.)
What's the point?
Lung cancer screening is a relatively new diagnostic test to identify patients and reduce lung cancer deaths. We’ve known for several years that some of the criteria used to define eligibility for lung cancer screening with low-dose chest CT—specifically age, whether or not patients formerly smoked, and pack-year criteria—have had the effect of reducing eligibility for specific patient populations who have increased risk for morbidity and mortality from lung cancer. So, you've seen in previous papers that Black and African American individuals are less likely to be identified as being eligible for lung cancer screening, despite the fact that they experience increased morbidity and mortality from lung cancer.
This particular paper explores a number of different ways that we can look at risk prediction modeling and define risk eligibility, or eligibility for lung cancer screening, to figure out if there are ways we can reduce some of those racial disparities in terms of lung cancer screening eligibility. The paper uses the Southern Communities Cohort Study, which is a wonderful resource and database that specifically over-represents Black or African American study participants, and it applies to a number of different types of ways to study risk modeling, and to define eligibility for lung cancer screening. And based upon those definitions of eligibility, the existing US Preventive Services Task Force criteria, and several other different methods to define eligibility, [the study’s authors] said, well, if we test out these different methods to define eligibility, what happens then to racial disparities?
One of the most exciting things about this particular paper is that they tested many of these different methods, and some of these different methods that they used to redefine eligibility actually reduced, or in some ways nearly eliminated, racial disparities and lung cancer screening eligibility. What they concluded then was that by redesigning and reconfiguring some of our eligibility criteria, we can either reduce, substantially reduce, or almost eliminate eligibility criteria disparities for lung cancer screening.
The Bottom Line: That's one of the most exciting takeaways from this particular paper—we do have ways to define lung cancer screening eligibility to make sure that we're reducing racial disparities and not creating disparities in terms of whether or not patients are eligible for lung cancer screening.
Why does this matter?
Why does this matter for diagnostic excellence and improving the care we have for our patients? There are several reasons. For one, we know that existing criteria undercount specifically Black or African American individuals in terms of being eligible for lung cancer screening. With the development and testing of these different types of eligibility criteria, we have different ways that we can make sure that our models are more accurate in diagnosing which particular individuals are at risk for developing adverse outcomes from lung cancer. And, with these particular models, we now have some systems in place when we're looking across our health systems and can ask how we're defining eligibility for lung cancer screening, we have some methods now we can put into place, and some of them are more practically implementable versus others. With the data provided in this particular paper, [we can] say which systems can be put in place to ensure high-quality outcomes for all the patients that we serve.
Who does this impact?
One of the things that is both a challenge and an exciting opportunity as we think about designing lung cancer screening systems across the country is that lung cancer screening is a relatively new test.
[For patients,] most of the individuals who are eligible for lung cancer screening have not obtained it. As we develop these new systems to identify patients who are eligible and get them into our systems for lung cancer screening, we have a real opportunity to drive improvements that actually truly benefit all the different patient populations that we serve.
Health system administrators, researchers, and individuals who are involved in the implementation of lung cancer screening programs can think carefully about different patient populations that you serve within your own communities. [For example, asking] “are there formulas that we're potentially thinking about using that may exclude patients who are otherwise high risk for developing adverse outcomes from lung cancer?” And as you think about the different patient populations you serve in your own communities, it's important to connect with their community groups and other individuals as well to try to figure out, amongst the different eligibility systems here, which of these types of systems will actually serve your community best. [You can] then advocate for policy changes within your own institution, and nationally, too, to support those different types of policy and eligibility criteria that would really maximize benefits for as many patients as we possibly can.
__
Share your thoughts and join the conversation on LinkedIn.
About Editor's Picks
Curated by the UCSF CODEX team, each Editor’s Pick features a standout study or article that moves the conversation on diagnostic excellence forward. These pieces offer meaningful, patient-centered insights, use innovative approaches, and speak to the needs of patients, clinicians, researchers, and decision-makers alike. All are selected from respected journals or outlets for their rigor and real-world relevance.
View more Editor's Picks here.