Lung cancer risk prediction model lacks efficacy in diverse populations
Click Here to Manage Email Alerts
A risk prediction model for lung cancer calculated scores that did not align with diagnoses in African American patients, according to results of a cross-sectional study.
African American individuals are less often eligible for lung cancer screening based on United States Preventive Services Task Force criteria, Christine S. Shusted, MPH, a research data analyst at Thomas Jefferson University, and colleagues wrote in a research letter published in JAMA Network Open.
“One approach to mitigate disparities in [lung cancer screening] eligibility is to use lung cancer risk prediction models to identify the patients with the highest risk,” they wrote. “Existing models have been derived from screening trials including 5% or fewer African American individuals and may not apply equitably to real-world screening participants.”
Shusted and colleagues conducted a cross-sectional, single-center study that calculated the risk for lung cancer using the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial modified logistic regression model (PLCOm2012) in 1,276 African American and white patients (mean age, 64.25 years; 42.7% African American; 59.3% women). There were “significant differences” in demographics between African American and white patients, with more African American patients being women (62.9% vs. 56.6%), smokers (63.9% vs. 49%) and Medicaid or dual-eligible beneficiaries (35.2% vs. 18.3%). African American patients were less likely to have private insurance only compared with white patients (25.3% vs. 33.2%).
The researchers categorized patients into risk quartiles based on PLCOm2012 values in a screening cohort and a cancer cohort, which was comprised of patients who had screen-detected lung cancer, 44% of whom were African American.
In the screening cohort, African American patients had a PLCOm2012 median risk score of 5.81%, which researchers wrote was “significantly higher” than median risk score of 4.1% in white patients. Additionally, more African American patients had risk scores in quartile 3 and 4 compared with white patients (59.8% vs. 42.5%), according to the researchers.
However, among the cancer cohort, more white patients had a risk score in quartile 3 or 4 compared with African American patients (61.1% vs. 35.7%).
“We found that lung cancer risk scores were not aligned with lung cancer diagnoses in African American patients,” researchers wrote. “Our findings suggest that we should use caution in applying risk models to diverse populations, given that our current understanding of lung cancer risk is incomplete.”