Prognostic Models for Predicting the Severity and Mortality in People With Acute Pancreatitis - A Systematic Review and Meta-Analysis.
Angus Hann, Anne Nagelhout, Gianluca Rompianesi et al.
✦ AI-curated · Sources linked
In 30 seconds
This systematic review and meta-analysis assessed how accurately four clinical scoring systems predict severe acute pancreatitis (SAP) and mortality in adults. It included 87 studies, comprising 92 cohorts and 65,942 participants. The highest pooled AUCs for predicting SAP ranged from 0.82 to 0.88 across the scores; for mortality, they ranged from 0.86 to 0.89. The authors found no evidence that any one system outperformed the others, and reported that accuracy varied with the timing of assessment.
Key findings
- For SAP, the highest pooled AUC was 0.88 (0.83-0.92) for Glasgow-Imrie within 48 h.
- For mortality, the highest pooled AUC was 0.89 (0.84-0.92) for BISAP within 24 h.
- The review found no evidence that Ranson, Glasgow-Imrie, APACHE II, or BISAP was superior for predicting SAP or mortality.
- Predictive accuracy varied according to when the scoring system was performed.
Why it matters
These findings help frame the role of commonly used scores in early risk assessment: each showed predictive accuracy, but the pooled results do not establish one as best. Because performance varied with assessment timing, interpreting a score requires attention to when it was calculated as well as which system was used.
What to keep in mind
The review included studies with differing assessment times and pooled results across multiple cohorts; the abstract does not report comparative evidence establishing superiority between scores.