screened the commonly used drugs after each additional 2 years of follow-up
for cancer. The first follow-up was through 1976, the most recent through
2002. We routinely included analyses with one- or two-year lag, that is,
ignoring cancers that developed in the first year or first two years after
the drug was first dispensed. The value of this feature will be described
subsequently. Subdividing recipients of a drug by number of prescriptions
dispensed to evaluate possible dose-response was done selectively.