Development and validation of a novel scoring system based on a nomogram for predicting inadequate bowel preparation

Abstract

Background and aims

Adequate bowel preparation (BP) is crucial for the diagnosis of colorectal diseases. Identifying patients at risk of inadequate BP allows for targeted interventions and improved outcomes. We aimed to develop a model for predicting inadequate BP based on preparation-related factors.

Methods

Adult outpatients scheduled for colonoscopy between May 2022 and October 2022 were enrolled. One set (N = 913) was used to develop and internally validate the predictive model. The primary predictive model was displayed as a nomogram and then modified into a novel scoring system, which was externally validated in an independent set (N = 177). Inadequate BP was defined as a Boston Bowel Preparedness Scale (BBPS) score of less than 2 for any colonic segment. The model was evaluated by the receiver operating characteristic (ROC) curve, calibration plots, and decision curve analysis (DCA).

Results

Independent factors included in the prediction model were stool frequency ≤ 5 (15 points), preparation-to-colonoscopy interval ≥ 5 h (15 points), incomplete dosage (100 points), non-split dose (90 points), unrestricted diet (88 points), no additional water intake (15 points), and last stool appearance as an opaque liquid (0–80 points). The training set exhibited the following performance metrics for identifying BP failure: area under the curve (AUC) of 0.818, accuracy (ACC) of 0.818, positive likelihood ratio (PLR) of 2.397, negative likelihood ratio (NLR) of 0.162, positive predictive value (PPV) of 0.850, and negative predictive value (NPV) of 0.723. In the internal validation set, these metrics were 0.747, 0.776, 2.099, 0.278, 0.866, and 0.538, respectively. The external validation set showed values of 0.728, 0.757, 2.10, 0.247, 0.782, and 0.704, respectively, indicating strong discriminative ability. Calibration curves demonstrated close agreement, and DCA indicated superior clinical benefits at a threshold probability of 0.73 in the training cohort and 0.75 in the validation cohort for this model.

Conclusions

This novel scoring system was developed from a prospective study and externally validated in an independent set based on 7 easily accessible variables, demonstrating robust performance in predicting inadequate BP.