Estimates of lead time and length bias in a breast cancer screening program.

Assume that the benefit of screening for breast cancer with a combination of mammography and a clinical examination is measured by comparing the five-year survival rate of women detected by screening to the five-year survival rate of women surface clinically with breast cancer, e.g., to the survival rate of women reported in End Results In Cancer. In this paper a mathematical model is used to estimate the percentage of the observed difference in five-year survival rates between women in these two groups that is due to lead time and length bias. For women detected at an initial screen, the best estimate is that about 50% of the observed difference is due to bias and 50% to earlier detection. However, the percent due to bias may be as low as 20% and as high as 72%. These estimates are relatively insensitive to the age of the women screened. For women detected at a second screen given one year later, the estimated percentage of the observed difference in survival rates between women detected at the screen and a control group that is due to bias is between 33 and 42%.