154 Inter-Center Variation in the Clinical Outcomes of Pediatric Crohn's Disease

Abstract

G A A b st ra ct s start time & polyp yield. Methods: We performed a controlled before & after study of patients receiving screening, surveillance, or diagnostic colonoscopies at a University-based VA medical Center. We compared the relationship between polyp yield and start time in our historical cohort (pre period) vs a 3-month intervention (post period) in which 3'x2' informational posters were placed in endoscopy rooms within view of operators & nurses. The poster depicted a bar graph of the previously documented hour-by-hour drops in polyp yield coupled with prominent text: “What time is it now?” The primary outcome was number of polyps reported by the endoscopist. We first measured the unadjusted relationship between colonoscopy start time and polyp yield for each period. We then performed logistic regression to measure the effect of start time on polyp yield while adjusting for period and other covariates. Finally, we created a time*polyp interaction term, and performed regression to evaluate whether period predicted the time-polyp relationship. Results: There were 494 & 108 patients in the pre & post periods, respectively. More polyps were found in the pre vs. post periods (2.1 vs 1.6 polyps per patient; p<0.01). In bivariate analysis, there was a significant inverse relationship between start time and polyp yield for both the pre and post periods (p=0.01 for both). Start time remained a significant predictor of polyp yield independent of period (p=0.001). Moreover, start time remained predictive of polyp yield after adjusting for prep quality, indication, cecal intubation, and period (p=0.01). Period did not predict the time*polyp interaction in multivariate regression. Conclusion: Provision of an auditing and feedback informational poster did not alter the inverse relationship between colonoscopy start time and polyp yield. These data strengthen our previous finding that start time matters, and suggest that passive use of social influence theory is inadequate to modify this robust effect; active auditing and feedback may be necessary.