Association of Colonoscopy and Colorectal Cancer Mortality in the VA Population

White River Junction Va Medical Center, White River Junction VT

Investigators

Abstract

DESCRIPTION (provided by applicant): Background: Colorectal cancer (CRC) is the second most common cause of cancer death in the United States. Screening colonoscopy aims at the detection and removal of the premalignant precursor, the adenomatous polyp. Recent population-based studies from Canada have shown that exposure to colonoscopy reduces CRC mortality by 29-31%. However, this reduction was seen for left-sided CRC (47- 67%), but not for right-sided CRC. These findings raise questions about the effectiveness of colonoscopy. In the absence of randomized trials, estimates have to be derived from observational studies. Observational studies using an individual level analysis may be limited by confounding. For instance individuals who undergo CRC screening may be different from those who do not. Such individual level confounding should be reduced when examining the exposure to a CRC screening test at a hospital level for all individuals who receive their care at that hospital. While we do not have any effectiveness data of CRC screening on CRC mortality within the VA population, preliminary results show a wide variation in the use of colonoscopy across VA hospitals. We found a concerning multiple-fold difference in the polypectomy rate between high and low polypectomy hospitals. A high variation in the intensity of screening, especially with respect to polyp removal, may translate into different outcomes for Veterans exposed to different hospitals. Assessment of the variation in colonoscopy intensity across VA hospitals should be completed, validated, and understood, and its effect on CRC mortality examined. Aims: The first aim of the study is to examine the association between individual exposure to CRC screening (colonoscopy) and CRC mortality in the VA system. The effect of colonoscopy on CRC mortality will be calculated for CRC overall, and by anatomic location of the cancer (VA Indianapolis group). The individual level analysis will be complemented by a system's level analysis. The second aim will therefore assess the variation of colonoscopy use across VA hospitals and across time and examine whether a varying hospital intensity in the use of colonoscopy procedures (all colonoscopies, screening colonoscopies and colonoscopies with polypectomy) affects CRC mortality (VA WRJ/ Dartmouth group). Method: To accomplish both aims we will apply a case control design. The VA central cancer registry and administrative data files will be used to ascertain cases and controls, examine exposure to relevant CRC screening tests (colonoscopy), and determine outcome (CRC mortality). Cases and controls will be identical for both aims. Cases are Veterans who were diagnosed with CRC between fiscal year (FY) 2003 and FY 2008 and died of CRC prior to FY 2010. Each case will be matched to four controls for each study aim by the time of CRC diagnosis, by age, and by gender. CPT and ICD coding will be used to identify exposure to colonoscopy starting in the fiscal 1997 to an index date six months before the diagnosis of CRC in cases. To examine colonoscopy intensity we will compute a rate of colonoscopy use per VA hospitals. Hospitals will be ranked into quintiles of test intensity. These quintiles will serve as a measure of colonoscopy intensity, which will be correlated with CRC mortality to determine the strength of the relationship.