ANTICIPATED IMPACTS ON VETERAN?S HEALTHCARE: Knowledge generated by this proposal will inform VHA about: 1) the prevalence (and incidence) of post-colonoscopy colorectal cancer (PCCRC) in Veterans; 2) determine whether and the extent to which patient outcomes are affected relative to detected colorectal cancer (DCRC); 3) identify patient-, endoscopist-, and facility- / system-specific factors associated with PCCRC, in patients who do and who do not have one or more index polyps identified and removed. Identifying remediable factors associated with PCCRC will lead to interventions to improve colonoscopy performance and adherence to appropriate surveillance intervals, aligning with recent VHA directives for high-quality colonoscopy. Deployment of these interventions will help ensure that Veterans receive colonoscopy of the highest quality. BACKGROUND: Colorectal cancer (CRC) that occurs after a colonoscopy showing no CRC but prior to the recommended interval for follow-up colonoscopy is referred to as ?post-colonoscopy CRC? (PCCRC). PCCRC results from missed colorectal lesions, incompletely resected lesions, or from de novo, fast-growing lesions. A robust, but heterogeneous literature shows that 3-9% of all CRCs are ICCs. More limited studies show an inconsistent effect of PCCRC on patient outcomes as compared to DCRC, and attribute PCCRC to specific colonoscopy-related factors and to polyp characteristics. As the prevalence of PCCRC, its associated factors, and effect on patient outcomes have not been well-studied within VHA, we propose the following specific aims: PROJECT AIMS: 1) Quantify the a) prevalence and incidence, and b) outcomes of PCCRC in Veterans, as compared with DCRC; 2) Assess the role of colonoscopy-related factors, polyp characteristics, patient factors, and facility factors for the risk for CRC after colonoscopy a) with polypectomy, and; b) without polypectomy METHODS: Using VA electronic databases (VA Central Cancer Registry, Corporate Data Warehouse, VA- CMS data repository, VA Informatics and Computer Infrastructure, VA Vital Status File, and others), we will perform a retrospective cross-sectional study (for prevalence), a retrospective cohort study (for incidence and outcomes) and nested case-control studies (to identify risk factors). The retrospective cross-sectional study will quantify prevalence of PCCRC, using definitions consistent with the published literature and experience from other large healthcare systems in order to facilitate comparison of PCCRC prevalence with those other systems for the interval 1/1/06-12/31/2011. From all patients undergoing colonoscopy during this interval, we will calculate PCCRC incidence for Veterans with non-advanced neoplasia and no neoplasia for whom a 5- year and 10-year surveillance / rescreening interval, respectively, is recommended. Incidence and prevalence estimates will be adjusted for diagnostic-error rates, which will be based on manual medical record review. We will conduct a retrospective cohort study to compare Veterans aged 50-85 years diagnosed with PCCRC to those diagnosed with DCRC between 1/1/2006 and 12/31/2011, examining the primary outcome of 5-year overall survival and secondary outcomes of urgent hospitalization, disease stage, surgery, and 30-day post- operative mortality. Multivariate analysis will include adjustment for covariates including age, sex, rurality, comorbidity, and cancer site. For all CRC diagnosed between 2004 and 2011, we will use a case-control study (CCS) design to identify risk factors for PCCRC among Veterans ages 50-85 years who did or did not have polypectomy. Cases will be Veterans with PCCRC either following polypectomy (CCS-1) or not (CCS-2). For both CCSs, controls will be Veterans who do not have PCCRC during the same timeframe as that of the cases. Exposure variables will be procedure-related (extent of exam, preparation quality, others), endoscopist-related (specialty, level of training, others), and institution-related (volume, mechanisms for ensuring follow-up, complexity, others). Odds ratios and attributable (etiologic) fractions will be derived using multiple logistic regression and Greenland?s method for logistic regression, respectively.
Ensuring that colonoscopy is performed well is necessary for optimizing prevention of colorectal cancer (CRC). One of the most important ways of measuring the quality of colonoscopy is to see how often CRC occurs following an initial colonoscopy that does not show CRC, an entity known as ?post-colonoscopy CRC? (PCCRC). We plan to: 1) measure how often PCCRCs occur in Veterans; 2) compare Veterans who have PCCRC with those who have an initially detected CRC for cancer stage, surgical complications, and survival; 3) identify factors associated with PCCRC, some of which may lend themselves to interventions that improve how colonoscopy is performed and how well follow-up occurs, measures that will make sure Veterans are receiving colonoscopy and follow-up of the highest quality.
