Can prediction models reduce racial/ethnic disparities in lung screening eligibility criteria?

Health inequalities are a growing concern worldwide. In this piece, Rebecca Landy and Corey Young from the Division of Cancer Epidemiology and Genetics in the US National Cancer Institute, part of the National Institutes of Health, discuss whether the recent draft lung cancer screening guidelines in the United States have succeeded in reducing racial/ethnic disparities.

Following the publication of the National Lung Screening Trial (NLST) in 2011 showing that low dose CT screening can prevent lung cancer mortality, the US Preventive Services Task Force produced their first lung cancer screening guidelines in 2013. They recommended annual screening for individuals aged 55-80 years who had a history of smoking at least 30 pack-years (equivalent to at least one pack of cigarettes per day for 30 years or more) and who currently smoke or have quit within the last 15 years. These eligibility criteria were largely based on evidence from the NLST, which included individuals aged 55-74 with at least 30 pack-years who currently smoked or had quit within the last 15 years, and microsimulation modelling.

However the recommendations did not take into account that African Americans have a higher risk of lung cancer, despite smoking less than whites, and develop cancer at younger ages. Therefore the guidelines were criticized for creating racial/ethnic disparities; for example, evidence suggested that 12% more whites who developed lung cancer were eligible for screening compared to African Americans.

In response, draft USPSTF guidelines released in July 2020 expanded the age range and pack-year criteria to try to reduce racial disparities in screening eligibility. Our study examined whether the new screening guidelines succeeded in reducing racial/ethnic disparities. As previously discussed on this blog, there are many ways of defining disparities; we defined disparities as the absolute difference in the percentage of years of life which would be gained from attended screening among people eligible for screening between whites and each minority group separately (African-Americans, Hispanic-Americans and Asian-Americans). We found that racial/ethnic disparities in lung cancer screening eligibility between whites and each minority group actually increased with the draft 2020 guidelines. Although 2020 guidelines increased eligibility for all races/ethnicities, the percent increase for minorities eligible for screening (97%) was only slightly higher than the increase for whites (78%), which was unlikely to sufficiently decrease the racial/ethnic disparities.

To see whether using prediction models to determine screening eligibility could reduce racial/ethnic disparities, we evaluated another approach – benefit was estimated from the Life-Years From Screening-CT (LYFS-CT) prediction model, which calculates the increase in life-expectancy for an individual if that person undergoes screening. We considered people to be eligible for screening by either the 2020 guidelines or if they were predicted to have a high benefit from screening. With this approach, the white/African-American disparity was eliminated. However, this method of identifying people for screening did not reduce disparities for Hispanic-Americans or Asian-Americans, though it did make screening more efficient (lowering the number of people needed to be screened per life-year gained).

These findings illustrate that it is not possible to eliminate racial/ethnic differences in the proportion of people diagnosed with lung cancer who would be eligible for screening using only the variables in the draft USPSTF guidelines, namely age, pack-years and quit-years, as there are larger differences between races/ethnicities in the risk distributions of the individuals who develop cancer. For example, 30% of lung cancer deaths among Hispanic-Americans and Asian-Americans aged 50-80 years who ever-smoked occurred in individuals with <10 pack-years, compared with 19% for African-Americans and 9% for Whites.

The draft 2020 USPSTF guidelines declined to recommend using risk models to define eligibility criteria due to a lack of evidence on their implementation in clinical practice, though other organizations (the National Comprehensive Cancer Network (NCCN) and American College of Chest Physician) do support use of models to augment USPSTF screening eligibility. There is also concern about the implementation of model-based screening guidelines since information on a range of variables is required to calculate the predicted lung cancer risk or benefit from screening, which was identified as a key reason risk models were not included in the draft 2020 USPSTF recommendations.

Even if screening recommendations perfectly identified who should be eligible for screening, disparities would remain in lung cancer mortality, since there are disparities at every stage “from prevention to screening to treatment to survivorship”. More effort will be needed to overcome the remaining disparities that result from the different risk distributions between races and ethnicities. Studies are currently underway comparing the performance of USPSTF eligibility criteria to model-based criteria in the UK.

The views expressed are those of the author. Posting of the blog does not signify that the Cancer Prevention Group endorse those views or opinions.

Share this Page

Subscribe to our mailing list

Be the first to comment

Leave a Reply

Your email address will not be published.