Case-control studies evaluating a screening test's efficacy in reducing cancer mortality require accurate classification of test indication to obtain a valid result. However, for analogous studies of cancer incidence, determination of test indication is not as critical because, to define exposure, we need consider only tests that can identify precursor lesions whose treatment might prevent cancer, not tests leading to cancer diagnosis. This study utilizes US Surveillance, Epidemiology, and End Results (SEER)-Medicare data, which do not include information about colonoscopy indication, to evaluate the efficacy of colonoscopy in preventing colorectal cancer (CRC) incidence. Cases were Medicare enrollees diagnosed with CRC between 1996 and 2013; up to 3 controls were matched to each case. Colonoscopy receipt prior to presumed onset of occult cancer was associated with an approximately 60% reduction in CRC incidence (odds ratio = 0.41, 95% confidence interval: 0.40, 0.42). The association was robust to differing exposure windows and estimates of occult cancer duration and is similar to those from CRC incidence studies in which exam indication was available. Our results suggest that, when it is impractical/impossible to determine whether tests were conducted for screening, the efficacy of a test in preventing cancer incidence can still be estimated using a case-control study design.
Keywords: cancer screening; case-control studies; colorectal cancer; incidence.
Published by Oxford University Press on behalf of the Johns Hopkins Bloomberg School of Public Health 2019.