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