Risk prediction models for blood transfusion in patients undergoing total hip and knee arthroplasty: a systematic review and meta-analysis.
Xiaoya Liu, Mengyao Liu, Yuxia Yang et al.
✦ AI-curated · Sources linked
In 30 seconds
This systematic review and meta-analysis evaluated published models predicting perioperative blood transfusion in patients undergoing total hip or knee arthroplasty. Fourteen studies contributed 36 models; transfusion incidence ranged from 3.2% to 30.8%. Across 13 validated models, pooled discrimination was an AUC of 0.87 (95% CI: 0.85-0.90). However, PROBAST assessment rated every model as having a high risk of bias, tempering confidence in their apparent predictive performance.
Key findings
- The pooled AUC for 13 validated models was 0.87 (95% CI: 0.85-0.90).
- Across included studies, blood transfusion incidence ranged from 3.2% to 30.8%.
- All 36 models were assessed as having a high risk of bias; applicability was unclear for four studies.
- Preoperative hemoglobin, tranexamic acid use, operative duration, intraoperative blood loss, and age were the most frequently included predictors.
Why it matters
Reliable transfusion risk estimates could help inform perioperative planning for hip and knee arthroplasty. Although the pooled discrimination appears promising, high risk of bias across all models means reported performance may not translate reliably to other settings. The review highlights the need to distinguish apparent predictive accuracy from evidence that a model is robust and applicable in clinical populations.
What to keep in mind
All included models were rated as having a high risk of bias, and applicability was unclear in four studies. The abstract does not report external validation results beyond the pooled analysis of validated models.