Cancer Incidence Tracing in the U.S.

Retrospective cohort studies have typically relied upon cause-of-death information to identify the occurrences of cancer among cohort members. Because the United States has the National Death Index, a well-established and thorough repository for death information, tracing vital status and obtaining cause-of-death information is relatively straightforward. However, some researchers have expressed concern about the accuracy and completeness of cancer data available through death certificates.

To date, most of the cancer incidence studies conducted on retrospective cohorts have traced through only one state cancer registry or used one registry supplemented with self-identified cases. However, this limited tracing leads to loss of case ascertainment due to out-migration. One way to avoid these losses in retrospective epidemiology studies is to trace through multiple cancer registries rather than only the state in which the population of interest is located.

The COBE investigators published in 2009 a manuscript describing some of the methodological issues encountered during multiple-state cancer incidence tracing (Buchanich JM, Youk AO, Marsh GM, Bornemann Z, Lacey SE, Kennedy KJ, Hancock RH, Esmen NA, Lieberman FS. Methodological Issues in United States Retrospective Cancer Incidence Studies. American Journal of Epidemiology, 170:112-119, 2009).

In brief, while they spent considerable time completing applications and obtaining the cohort matches, they were able to obtain cancer incidence data from 21 state registries (they declined to pursue matches from three registries due to the high cost). Despite some obstacles, cancer incidence studies of retrospective cohorts using multiple cancer registries are feasible.

Background and Experience

The COBE faculty has a working relationship with cancer registry personnel in 22 states, with the North American Association of Central Cancer Registries (NAACCR) and the Central Brain Tumor of the United States (CBTRUS). They have successfully completed the various state applications, obtained registry approval for individual-level data and have worked with registry personnel to ensure accurate and complete matching.

Recent Example

In 2013, COBE investigators completed a very large occupational cohort study of jet engine manufacturing workers. One component of that study was to evaluate the incidence of central nervous system (CNS) cancer among employees. They identified cancer incidence cases occurring between 1976 and 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. The methodological issues were presented in poster form during the 2008 Society of Neuro-Oncology meeting, available for viewing here (2008 SNO poster link).

The investigators published 9 papers detailing the results of the malignant CNS cancer analysis, the non-malignant CNS analysis, a comparison of the mortality and cancer incidence results and a novel methodology for analyzing non-malignant cancer incidence.

The jet engine manufacturing workers study was the first known to have successfully traced through multiple state cancer registries, putting the COBE researchers in a unique position to lend their expertise to the successful completion of the cancer incidence studies using multiple state cancer registries.

Conclusions regarding the feasibility and time requirements of pursuing multiple state cancer incidence tracing were presented during the 2009 Society for Epidemiologic Research annual meeting (2009 SER poster link).

Available Resources

The COBE faculty and staff are available for consultation and implementation of cancer incidence tracing projects for federally, industry or trade group-funded projects being conducted by universities, companies or consulting firms.