Development of a tool to describe case difficulty and educational value of an individual GI bleeding case to aid setting Certification Standards for GI bleed competence.

Tool development was in 3 phases. Phase 1 -EMS Reporting tool (Medtronic) identified 57 patients referred for upper GI bleed [codes ‘malaena’, ‘haematemesis’]. An iterative approach developed a standardised case difficulty scoring system based on domains of case context, endoscopic findings, treatment and outcomes graded against global assessment of case difficulty (easy, moderate, difficult) and educational value (0=no added value above diagnostic case, 1=assessment and management advice, no therapy, 2=assessment, treatment and management). Phase 2 - Putative score applied to a second cohort of 63 patients with adjustment to score weighting across scoring domains to improve alignment with global assessments. Phase 3 - S2T2 tool applied to cases in UK GI bleed database [1] and examined associations with difficulty, EV, clinical outcomes, and existing severity scores.

The final S2T2 tool included four domains – SET (Endoscopy outside of Unit); SITE (Bleeding site not visible or difficult to access); TYPE (variceal or non-variceal bleed, with sub-types defined); TREAT (endoscopic treatment required/not required) - matched to global assessment of case difficulty (easy/moderate/difficult) and weighting of educational value for an endoscopic trainee. Total S2T2 score showed a strong correlation between a global rating of difficulty (r=0.9) and educational level (r=0.8). 25% of GI bleed cases provided Level 2 Educational Value. In phase 3, the S2T2 tool was applied to 3,979 cases from the UK GI bleed database: this defined 64% as technically easy, 24% of moderate difficulty, and 12% as difficult. Increasing S2T2 difficulty was associated with higher rates of endoscopic therapy, rebleeding, need for interventional radiology or surgery, longer length of stay and bleed-related mortality. 

EV increased progressively with higher S2T2 scores: median (IQR) 0 (0–1) for Level 0, 1 (1–2) for Level 1, and 6 (5–8) for Level 2 cases (Kruskal–Wallis p < 0.001). In univariable analyses, S2T2 (OR 4.81), GBS (OR 1.08), Pre-Rockall (OR 1.15), and Full Rockall (OR 1.28) were associated with EV. However, in multivariable modelling, only S2T2 remained independently predictive (OR 4.78, 95% CI 4.37–5.23), while GBS and Pre-Rockall lost significance (Table). Correlations between S2T2 and clinical severity scores were weak (p ≤ 0.35), indicating S2T2 captures a distinct dimension of procedural complexity.

The S2T2 tool is simple, rapid to apply, and correlates strongly with both case difficulty and EV. It captures aspects of procedural complexity not reflected by existing clinical severity scores and may be valuable for defining training requirements and certification criteria in GI bleeding.