Risk-based models may improve screening efficiency compared to current guidelines, but accuracy issues remain for specific racial and ethnic populations.
RT’s Three Key Takeaways:
- Underestimated Risk: Many lung cancer prediction models were found to underestimate the risk for non-Hispanic Black individuals and showed less precision in identifying future cases for Asian participants.
- Screening Efficiency: Risk-based models generally improved screening efficiency and reduced variation across racial and ethnic groups when compared to the 2021 US Preventive Services Task Force criteria.
- Population Refinement: Researchers indicated a need for further model refinement to achieve optimal accuracy and fairness, noting that race and ethnicity serve as indicators for social factors not explicitly captured in current databases.
A cohort consortium study found that many lung cancer prediction models underestimated risk for certain ethnic groups, according to findings published in Annals of Internal Medicine.
The research found prediction models underestimated risk for non-Hispanic Black individuals and were less able to identify future lung cancers for Asian participants. Still, risk-based approaches overall may improve screening efficiency and make it more equitable compared with current guidelines, according to researchers.
Scientists from the International Agency for Research on Cancer and colleagues assessed 16 lung cancer risk models in 641,830 Asian, Hispanic, non-Hispanic Black, and non-Hispanic White adults aged 50 to 80 years with a history of smoking. They examined how well the models estimated lung cancer cases or deaths and how effectively they distinguished between individuals who would and would not develop disease.
The team also evaluated how each model might guide screening decisions, including eligibility, cases detected, and overall efficiency.
Although risk-based models generally improved screening efficiency and helped reduce some disparities, they tended to underestimate risk in non-Hispanic Black participants and were less precise for identifying future cases in Asian individuals.
All models showed better estimated screening efficiency than the expanded 2021 US Preventive Services Task Force criteria and reduced variation across racial and ethnic groups, but none simultaneously achieved optimal accuracy, fairness, and efficiency, indicating a need for further refinement to better serve diverse populations.
It is important to recognize that race and ethnicity are not biological factors but are population level indicators of social factors not explicitly captured in the database used for the study.
