Objective:
There has been an alarming increase of lung cancer, 3.4-fold at age 65-69 rate and 13-fold in total number in 35 years, among nonsmoking women in Taiwan, The worried well, concerned with their risk, demanded LDCT screening. However, current recommendation was limited to heavy smokers with 30 pack years. Screening low-risk individuals had the potential of causing substantial harms.

Methods:
A cohort of 439,119 adults, who had complete medical work up during health surveillance program, between 1994 and 2008, had their IDs matched for lung cancer with National Cancer Registry. Higher lung cancer risk nonsmokers could be identified with prediction models.

Results:
Significant risks were found among nonsmokers as follows: Personal history of any cancer or family history of lung cancer; body weight, Reduced spirometry such as FEV1; biomarkers, including elevated Alfa-feto-protean(AFP), Carcino-Embronic Antigen(CEA), and C-reactive Protean (CRP), low bilirubin, second hand smoking, abnormal chest X-ray and geo-coded PM2.5 for air pollution. With 1,307 incidence lung cancer, including nonsmokers (607), light smokers (274) and heavy smokers (426), a prediction model was developed, with ROC 0.806-0.847. A simple 5-year risk score was established for nonsmokers and light smokers to check their risk before LDCT.

Conclusions:
With risk-data available, a small portion of nonsmokers (3%) and one fifth of light smokers (20%) could be found qualified for LDCT screening, as they reached 5-year lung cancer risk of heavy smokers, 1.5% in 5 years. This pre-screening model could save unnecessary LDCT procedures and avoid harms and cost associated with screening low risk individuals.