Disentangling true inspection time from therapeutic intervals: Validation of effective withdrawal time as a colonoscopy quality metric

Figure 1 Schematic representation of the study workflow and the artificial intelligence architecture for effective withdrawal time quantification. AI: Artificial intelligence; EWT: Effective withdrawal time; SWT: Standard withdrawal time; BBPS: Boston Bowel Preparation Scale.

Figure 2 Adjusted odds ratios for lesion detection according to quintiles of withdrawal time. A: Adjusted odds ratio (aORs) for polyp detection according to quintiles of effective withdrawal time (EWT) (orange) and standard withdrawal time (SWT) (blue); B: The aORs for adenoma detection according to quintiles of EWT (orange) and SWT (blue); C: The aORs for cancer detection according to quintiles of EWT (orange) and SWT (blue). EWT: Effective withdrawal time; SWT: Standard withdrawal time; aOR: Adjusted odds ratio; Q1: Quintile 1; Q2: Quintile 2; Q3: Quintile 3; Q4: Quintile 4; Q5: Quintile 5.

Figure 3 Comparative predictive performance of withdrawal times for lesion detection. A: Receiver operating characteristic (ROC) curve for polyp detection using the continuous model, comparing effective withdrawal time (EWT) (orange) and standard withdrawal time (SWT) (blue); B: ROC curve for polyp detection using the quintile model, comparing EWT (orange) and SWT (blue); C: ROC curve for adenoma detection using the continuity model, comparing EWT (orange) and SWT (blue); D: ROC curve for adenoma detection using the quintile model, comparing EWT (orange) and SWT (blue); E: ROC curve for colorectal cancer detection using the continuity model, comparing EWT (orange) and SWT (blue); F: ROC curve for colorectal cancer detection using the quintile model, comparing EWT (orange) and SWT (blue). EWT: Effective withdrawal time; SWT: Standard withdrawal time; AUC: Area under the curve.

Figure 4 Association between withdrawal times and adjusted detection rates. A: Adjusted adenoma detection rate (ADR) plotted against effective withdrawal time; B: Adjusted ADR plotted against standard withdrawal time; C: Adjusted polyp detection rate (PDR) plotted against effective withdrawal time; D: Adjusted PDR plotted against standard withdrawal time. ADR and PDR were adjusted for age, sex, bowel preparation quality (Boston Bowel Preparation Scale), and endoscopist experience. ADR: Adenoma detection rate; PDR: Polyp detection rate.

Figure 5 Subgroup analyses of adjusted adenoma detection rate across withdrawal times. A: Adjusted adenoma detection rate (ADR) plotted against effective withdrawal time (EWT) stratified by polyp number; B: Adjusted ADR plotted against standard withdrawal time (SWT) stratified by polyp number; C: Adjusted ADR plotted against EWT stratified by physician seniority; D: Adjusted ADR plotted against SWT stratified by physician seniority; E: Adjusted ADR plotted against EWT stratified by bowel preparation quality; F: Adjusted ADR plotted against SWT stratified by bowel preparation quality. The smooth curves represent the adjusted ADR (left Y-axis), while the background bar histograms indicate the number of procedures (right Y-axis). ADR was adjusted for age, sex, Boston Bowel Preparation Scale, and endoscopist experience. ADR: Adenoma detection rate; BBPS: Boston Bowel Preparation Scale.

Figure 6 Subgroup analyses of adjusted polyp detection rate across withdrawal times. A: Adjusted polyp detection rate (PDR) plotted against effective withdrawal time stratified by bowel preparation quality; B: Adjusted PDR plotted against standard withdrawal time stratified by bowel preparation quality; C: Adjusted PDR plotted against effective withdrawal time stratified by physician seniority; D: Adjusted PDR plotted against standard withdrawal time stratified by physician seniority. PDR was adjusted for age, sex, Boston Bowel Preparation Scale, and endoscopist experience. PDR: Polyp detection rate; BBPS: Boston Bowel Preparation Scale.