Introduction Colonoscopy is the standard of care for diagnosing colorectal cancer (CRC). However, 3.4%–7.9% of subjects with CRC are reportedly diagnosed within 3yrs of a colonoscopy that did not detect the cancer (post-colonoscopy colorectal cancer, PCCRC). We have investigated risk factors for these events in a national data set in England.

Methods Hospital Episode Statistics (HES) collates information on all NHS hospital attendances in England. Subjects undergoing colonoscopy without a CRC diagnosis 6–36 months before subsequent CRC diagnosis were identified as PCCRC cases (definitely missed – colonoscopy without CRC diagnosis 6–12 months before CRC diagnosis; probably missed – colonoscopy without CRC diagnosis 12–36 months before CRC diagnosis) and those with no colonoscopy 6–36 months before diagnosis served as controls. The influence of personal and institutional variables on missed PCCRC were examined by multivariate logistic regression.

Results HES records from 2001–12 were analysed including 2874641 colonoscopies in 2263905 subjects. 136237 subjects were diagnosed with CRC. 4219 (3.1%) definitely missed PCCRC cases and 8266 (6.1%) probably missed PCCRC cases occurred. Colonic polyps were the most common coded finding in PCCRC subjects (1553 subjects (12.4%)). Emergency colonoscopies were less likely to fail to diagnose CRC than elective procedures (OR 0.58 (95% CI: 0.5–0.6), p < 0.001). Subjects over age 70 (1.16 (1.1–1.2), p < 0.001), female gender (1.05 (1.0–1.1), p = 0.018) and comorbidities (liver disease (2.18 (1.4–3.5), p = 0.002), peptic ulcer (1.29 (1.1–1.6), p = 0.01), myocardial infarction (1.14 (1.0–1.3), p = 0.046), pulmonary disease (1.11 (1.0–1.2), p = 0.025)) were associated with PCCRC. Ethnicity was not associated with PCCRC. Right sided CRC was more likely to be missed (1.30 (1.25–1.37), p = 0.015). Subjects with PCCRC were less likely to undergo surgery (0.27 (0.26–0.28), p= <0.001) or chemotherapy (0.62 (0.59–0.65), p= <0.001). Overall survival was worse in PCCRC subjects than controls. There was a fourfold variation in PCCRC rates between units. Unit volume was inversely related to PCCRC rate (lowest tertile volume versus highest tertile 1.72 (1.6–1.8), p= <0.001). The annual rate of PCCRC has improved over the study period with a fall in PCCRC rate from 15.9 to 5.1% per annum.

Conclusion The rate of PCCRC up to 3 yrs prior to CRC diagnosis was 9.1% in England between 2001–12. PCCRC was associated with increasing age, female gender, comorbidities, site in right colon and colonoscopy unit volume. PCCRC subjects were less likely to have surgery or chemotherapy and had worse survival rates. Encouragingly, annual rates of PCCRC have fallen over the study period.

Disclosure of Interest None Declared.