[Evaluation of population-based lung cancer screening in Niigata and analysis of interval cases based on comparison lung cancer registry with screening records].

Survival analysis has been used as one way to examine the efficacy of cancer screening. However, this type of study is susceptible to many sources of bias, especially to a length bias. In the city of Sukumo in Japan, a survival analysis considering screening history for the purpose of eliminating the length bias was conducted. The efficacy of lung cancer screening was assessed by comparing those who participated in the screening in the year preceding diagnosis, with those who did not. The 5-year survival rate was 44% in the former and 16% in the latter. We tried to evaluate population-based lung-cancer screening in Niigata using the method as that of Sukumo by comparing lung-cancer registry data with screening records, those who died of lung cancer from 1991 to 1994 in the study area and had the opportunity to be screened were categorized according to the screening history for the year preceding diagnosis. The 5-year survival rate was 41% in those who were screened and 19% in those not screened. In addition, the median survival time of the former was 37 months, significantly longer than the 12 months of the latter. Replication of study results in different populations may suggest the efficacy of lung-cancer screening on an annual basis. We also analyzed the clinical background of 47 interval cases in the search for a future direction for the improvement of the screening method.