A Novel Model to Predict Inadequate Bowel Preparation Prior to Colonoscopy Incorporating Patients' Reactions to Drinking the Laxative

Background and Aims: Prior studies have identified predictors of inadequate preparation with limited success. We aimed to build a model that could predict the likelihood of inadequate preparation by also including factors related to the patient's reaction to drinking the laxative preparation. Methods: Demographic, clinical, and preparation-related data were prospectively collected on patients undergoing colonoscopy. An inadequate preparation was defined as a Boston Bowel Preparation Scale < 6. Statistical analyses were performed to identify predictors of inadequate preparation and create a predictive model. Results: 324 patients were included (age 67 +/- 14 years, 52% male). 77 (23.7%) had inadequate preparations. Diabetes (p < 0.001), cerebrovascular accident (CVA) (p < 0.001), incomplete prep consumption (p = 0.007), high school level education and above (p < 0.001), use of Bisacodyl (p = 0.005), >10 bowel movements (p = 0.02), and use of Sodium Picosulfate or low-volume polyethylene glycol (PEG) solution (2L) compared to PEG 3L (p < 0.001) were significant variables. In a multivariate analysis, prior CVA increased the risk for inadequate preparation (OR = 4.8, CI 1.6-14.5), whereas high school level education and above (OR = 0.4, CI 0.2-0.8), consumption of Bisacodyl (OR = 0.4, CI 0.2-0.8), >10 bowel movements (OR = 0.5, CI 0.3-0.9), and use of Sodium Picosulfate (OR = 0.5, CI 0.3-0.9) decreased the risk for inadequate prep. Using these, a predictive model for patients likely to have an inadequate colon preparation was created with an area under the curve of 0.74 (35% sensitivity, 90% specificity at a cutoff point of 39%). Conclusion: Given the low sensitivity, this predictive model does not appear ready for clinical use. However, due to its high specificity, it may be helpful in high-risk, sicker populations by preventing inadequately prepped procedures.