Person-time was defined as the time from date of
baseline examination to date of diagnosis for patients
and to last date known alive for the cohort. Person-time
for patients ranged from 3 to 13 years, and, for the cohort,
from 1 to 16 years. Stratified analysis and Poisson
regression were used to estimate rate ratios, adjusted for
age, race, and gender, for various levels of aspirin use.
We also used Poisson regression to assess confounding
effects of alcohol consumption (at least monthly versus
not) and cigarette smoking (ever versus never smoked
regularly). Subjects with missing values were excluded.
When there were no patients in the exposure category,
the upper bound of an exact 95% confidence interval
was calculated using the BIN20M software routine. l4
All other analyses were performed using the SAS statistical
package.