Low-dose CT screening for lung cancer may be more effective when eligibility is based on an individual's overall risk of developing the disease rather than on age and smoking history alone. That is the conclusion of researchers from the Lung Cancer Cohort Consortium, who compared multiple lung cancer risk prediction models with the 2021 U.S. Preventive Services Task Force (USPSTF) screening criteria.

The researchers analyzed data from 641,830 adults aged 50 to 80 years from 12 U.S. cohort studies. All participants had a history of smoking. The study compared 16 lung cancer risk prediction models with the current USPSTF eligibility guidelines.

Under the existing recommendations, annual low-dose computed tomography (LDCT) screening is advised for adults aged 50–80 years with a smoking history of at least 20 pack-years—for example, smoking one pack of cigarettes per day for 20 years—who currently smoke or quit within the past 15 years.

The analysis showed that, while referring the same number of people for screening, individualized risk models identified more individuals who were likely to benefit than the standard USPSTF criteria. The PLCOm2012 and Life Years Gained From Screening–CT models demonstrated the best overall performance.

However, the researchers also found important limitations. Among non-Hispanic Black participants, 11 of the 16 models underestimated lung cancer risk. Among Asian participants, 13 models were less accurate in distinguishing between individuals who would and would not develop lung cancer. In some population groups, these disparities persisted even when compared with non-Hispanic White participants.

The authors concluded that no single approach simultaneously improved all key measures, including screening eligibility, sensitivity, predicted clinical benefit, and equitable performance across racial and ethnic groups. While individualized risk models appear promising, they are not yet ready for universal implementation without further validation and calibration across diverse populations.