Hospital Length of Stay Can Be a Predictor of Bowel Preparation Quality in Hospitalized Patients: Results of a Pilot Study in a Tertiary Care Teaching Hospital

Abstract

Introduction: Aim: To identify the predictive factors for poor bowel preparation quality in hospitalized patients at a tertiary care hospital. Methods: Nursing staff and endoscopists assessed bowel preparation quality using the Boston Bowel Preparation Score after inpatient colonoscopies. We looked for potential predictors of poor bowel preparation scores. We reviewed patients' charts and examined their age, gender, h/o prior colonoscopy, length of hospitalization, h/o opiate use, BMI, h/o dementia or psychiatric illness, h/o constipation, h/o prior abdominal surgery, procedure type (i.e. colonoscopy vs colonoscopy with EGD), hospital unit, and number of medications. Data was analyzed utilizing SPSS software. Results: Twenty-eight forms were filled out during a pilot period evaluating 24 patients. For the 4 patients who had 2 forms completed, inter-rater reliability was 100%, so no averaging of scores was needed. BBPS scores ranged from 3 to 9. Nearly a third of the patients (29%, n=7) had a score of 3 while only 8% (n=2) had a perfect score of 9. Seven patients (29%) had a BBPS score of 7 or higher. When analyzing the subscores of the R-sided, transverse and L-sided colon, BBPS scores became lower as the colon segment became more proximal. The number of full subscores (i.e. excellently prepped) awarded decreased from 33% (n=8) in the left colon to 17% (n=4) in the transverse colon to 8% (n=2) in the right colon. For the analysis of predictive variables, 23 patients were analyzed as one patient had an invalid medical record number recorded. After running a 1-tailed Pearson correlation, these were gender (r = .355, p = 0.048, with females prepping better), length of stay (r = -0.364, p = 0.044), and number of medications (r = -.432, p = 0.020) found to be significant. Younger age trended toward better bowel preps, but this did not reach significance (r = -0.313, p = 0.073). A forced-entry multiple regression was conducted on these four factors; the factors explained a significant amount of variance in bowel prep quality (R2 = 0.427, F(4,18) = 3.359, p = 0.032). In terms of individual factors, only length of stay significantly predicted bowel prep quality (β = -0.14, p = 0.037). Conclusion: Our study suggests that hospital length of stay could be utilized as a predictor of patients that may need additional time or bowel prep for optimal inpatient colonoscopy. This may be useful in avoiding repeat colonoscopies due to poor preparations which can lengthen hospital stays.