Use of Screening versus All Exams to Calculate Mean Adenomas per Colonoscopy: Data From the New Hampshire Colonoscopy Registry

Introduction: Adenomas per colonoscopy (APC) may be a better quality measure than adenoma detection rate since it reflects the ability of an endoscopist to optimize colorectal cancer prevention by clearing the colon of all precursors. A major limitation of all detection rates is that some endoscopists have a lower volume of exams. A proposed solution is to use all exams as opposed to current calculation using only screening colonoscopies. We used data from the New Hampshire Colonoscopy Registry (NHCR) to compare APC calculated with data from screening versus all exams. Methods: Our sample consisted of patients enrolled in the NHCR with at least one follow up event 3 months of later than index exam. Follow up were events were a colonoscopy or CRC diagnosis in the New Hampshire State Cancer Registry which collects data from NH and other states (VT, MA, ME). The exposure variable was APC which was calculated as the total number of adenomas for colonoscopies divided by number of colonoscopies for each endoscopist. Screening APC (APC-S) used data from screening exams and APC-A used all exams, regardless of indication. APC was examined as continuous variables as well by categories, 0.2, 0.4, 0.6 and 0.8. We examined risk for PCCRC defined as any CRC diagnosed 3 months after an index exam. Exclusion criteria were any CRC diagnosed at index or within 3 months, incomplete exams, IBD, and genetic syndromes. Cox regression was used to model the Hazard of PCCRC on APC controlling for age, sex, index exam year, index findings, bowel prep quality, having more than 1 surveillance exam and family history of CRC. Results: Our sample included 27,688 exams performed by 152 endoscopists with 153 CRCs diagnosed after the index exam. APC-A and APC-S had a high correlation (Spearman’s rho=0.90; p< 0.001) but the mean APC-A was higher (0.69) than APC-S (0.43) Both APCs were associated with a reduction of PCCRC as a continuous variable as well as stratified as above (Table). The median percentage of screening exams across endoscopists was 50% (IQR=16). Median difference between APCs was 0.21 (IQR=0.12) Conclusion: Our novel data support the use of APC as calculated for all exams as a quality measure by demonstrating a reduction in PCCRC risk in exams performed by endoscopists with higher APC-As, similar to that for APC-S. In addition, the 2 rates correlated closely. However, varying proportions of screening exams may make it difficult to develop benchmarks without adjusting for endoscopist case mix. Table 1. - Post colonoscopy CRC and APC calculated with screening (APC-S) and all exams (APC-A) APC-S< 0.2 (REF) APC-S0.2-< 0.4 APC-S0.4-< 0.6 APC-S0.6-< 0.8 APC-S0.8+ P value Continuous APC p value APC-S HR 1.0 0.20 0.17 0.15 0.08 0.001 0.10 0.001 95 % CI REF 0.12-0.34 0.10-0.29 0.07-0.32 0.03-0.25 0.03-0.32 AbsoluteRisk 2.6% 0.7% 0.5% 0.3% 0.2% 0.001 --- --- N 688 10175 10424 4042 2359 --- --- --- APC-A< 0.2 (REF) APC-A0.2-< 0.4 APC-A0.4-< 0.6 APC-A0.6-< 0.8 APC-A0.8+ P value ContinuousAPC p value APC-A HR 1.0 0.22 0.21 0.18 0.11 0.001 0.21 0.001 95% CI REF 0.11-0.44 0.12-0.40 0.10-0.34 0.06-0.23 0.10-0.45 AbsoluteRisk 2.5% 0.7% 0.7% 0.5% 0.2% 0.001 --- --- N 522 3540 7889 7686 8051 --- --- --- Cox regression was used to model the Hazard of PCCRC on APC controlling for age, sex, index exam year, index findings, bowel prep quality, having more than 1 surveillance exam and family history of CRC.