OBJECTIVES : Developing a method to adjust for lead time bias in the quantification of the health benefit of cancer screening

METHODS : We interlinked Taiwan Cancer registry and National Mortality Registry and extrapolated throughout life to estimate LE (life expectancy) of patients of non-small cell lung cancer (NSCLC) and breast cancer diagnosed at stages I and IV. Then, we subtracted the above figures from age-, sex-, and calendar year-matched referents simulated from national vital statistics to obtain the loss-of-LE, or, expected years of life loss (EYLL).

RESULTS : We found that the LE's of non-small cell lung cancer diagnosed at stage I (n=6449) and IV (n=27258) were 11.59 and 1.50 life-years respectively. Apparently, the health benefit of early diagnosis would be 10.05 years. Since the average ages of patients diagnosed at these two stages were 68.9 and 71.7 year-old, respectively, there would be a potential lead time bias about 2.8 years. Similarly, the ΔLE's of breast cancer between stage I (n=32908) versus IV (n=5836) were 22.3 (=29.1-6.8) years with average diagnosed ages of 52.5 and 55.4 year-old, respectively, and the potential lead time bias was 2.9 years. Alternatively, we estimated the saving of EYLL for each subcohort through subtracting the LE from age-, sex-, and calendar year-matched referents simulated from national life tables, or, health benefits that could be saved from treating such a cancer. The difference-in-differences, or, ΔEYLL would be the net health benefit that could be saved from early detection of cancer at stage I compared with stage IV, of which the non-small cell lung cancer would be 7.97 (=12.62-4.65) years and that of breast cancer would be 19.7 (=22.4-2.7) years.

CONCLUSIONS : Estimation of difference in saving of loss-of-LE for each subcohort could adjust for the potential bias resulted from early diagnosis of a specific cancer.