S245 Risk Factors for 5-Year Post-Colonoscopy Colorectal Cancers (PCCRCs)

Abstract

Introduction: 3-10% of CRCs are PCCRCs, which are CRCs identified after a colonoscopy (CY) that does not find CRC. Risk factors for PCCRCs have been little studied. The study aim was to identify risk factors for PCCRCs at 3 years (PCCRC-3y) based on polypectomy of ≥ 1 neoplastic polyps at index CY. Methods: We assembled a cohort of 50 to 85 year-old Veterans with newly diagnosed CRC from 1/1/2003 to 12/31/2013, examining prior exposure to CY. Those whose CY occurred ≤ 6 months prior to CRC diagnosis with no other CY within the previous 36 months were categorized as having detected CRC (DCRC). Those whose CY occurred 6-36 months prior to CRC diagnosis were categorized as PCCRC-3y. We conducted 2 nested case-control studies (CCS) based on whether polypectomy of neoplastic polyps was performed during index CY, and compared demographics, clinical features, and CY-specific factors (e.g., prep quality, endoscopist training, VA- vs non-VA CY, and recommended surveillance interval) between PCCRC cases and DCRC controls who were matched for age and facility in a 1:2 ratio. Univariable and multivariable logistic regression identified factors independently associated with PCCRC, reported as odds ratios (OR) and 95% CIs. Results: There were 29,877 patients with CRC, with 1785 (6.0%) classified as PCCRC. From this dataset, we identified 402 cases and 804 matched controls w/o polypectomy and 404 cases and 808 controls with polypectomy, with cases and controls comparable demographically. Factors independently associated with PCCRC in the no polypectomy CCS were: Charlson score (OR=1.10; CI, 1.00-1.21); BMI ≥ 30 kg/M2 (OR=1.66; CI, 1.16-2.39); and recommended surveillance interval either < 5 years (OR=4.67; CI, 3.30-6.66) or missing (OR=3.00; CI, 2.17-4.14), c-statistic = 0.70. When the two surveillance variables were removed from the model, good or excellent prep quality was protective (OR=0.64; CI, 0.46-0.78), c-statistic=0.61. In the polypectomy CCS, the factors were: non-VA CY (OR=2.97; CI, 1.11-8.21); non-GI endoscopist (OR=1.64; CI, 1.16-2.33); advanced adenoma (OR=2.02; CI, 1.54-2.65); ≥ 2 proximal polyps (OR=1.43; CI, 1.09-1.87) and recommended follow-up < 1 year (OR=3.38; CI, 2.40-4.80), c-statistic of 0.71. Conclusion: Several factors are associated with PCCRC-3yr, some of which are modifiable, and the most important of which is recommended surveillance interval. These factors may be useful for tailoring surveillance and as targets for quality improvement.