Methodological issues in a retrospective cancer incidence study

The authors traced incidence of central nervous system cancer in a large occupational cohort of jet engine manufacturing workers from 1976 to 2004 in the 24 US states that comprised 95% of the cohort deaths. The cohort of approximately 224,000 employees was matched with cancer registry data; all central nervous system cancer matches were requested with their diagnostic data. This paper highlights the obstacles encountered while conducting this retrospective cancer incidence study. The authors spent approximately 700 hours completing applications and obtaining the cohort matches. Approximately 70% of the cases were identified in the state in which the facility of interest is located. In addition to the large amount of time involved, identified issues include complicated approval processes, high costs, temporal differences among the registries, and registry agency difficulty in performing the matching. Several states do not allow individual-level data to be used for research purposes. Researchers can gain important cancer incidence information by matching retrospective cohorts to multiple state cancer registries. However, they should carefully weigh the time and costs required and plan accordingly. Despite some serious obstacles, many of which are potentially resolvable, cancer incidence studies of retrospective cohorts using multiple cancer registries are feasible.