Larynx cancer rates have generally increased over time, largely due to increased smoking. Furthermore, larynx cancer rates, like most cancer rates, increase rapidly with age. Hence, age and calendar time were associated with disease and could act as potential confounders if exposed and non exposed groups differed for these factors. To control for this possible confounding, data on larynx cancer incidence were to be stratified by age and calendar time. This would require agespecific and calendar-time-specific rates in both the exposed and nonexposed populations. Indirect standardization was used to create a summary rate ratio across age and calendar-time strata (see Question and Answer 9).
While age-specific rates were available both for the exposed cohort and the various possible comparison populations (Connecticut, New York, and the United States), the comparison populations did not consistently have rates available during the entire calendar period of interest (1940s through the 1980s). Some procedure would have to be adopted to confront this problem.