Prediction models for primary graft dysfunction after lung transplantation: Systematic review.

Primary graft dysfunction (PGD) is the leading cause of early mortality following lung transplantation (LTx). Accurate prediction of PGD is critical to implementing effective preventive measures at the earliest opportunity. A systematic literature search about PGD prediction models was conducted, and results were synthesized narratively and descriptively. This systematic review adhered the Transparent Reporting of Multi-variable Prediction Models for Individual Prognosis or Diagnosis: Checklist for Systematic Reviews and Meta-analyses (TRIPOD -SRMA). The risk of bias was assessed using the Prediction Model Risk of Bias Assessment Tool (PROBAST). Nine original studies comprising 12 predictive models were included. Most models were rated as high risk of bias, despite claims of TRIPOD adherence. Common methodological limitations included inadequate reporting of performance measures, lack of blinding to predictors and outcomes, retrospective design, and insufficient sample size. Only four studies conducted both model development and external validation. Discriminative performance, as measured by the area under the curve (AUC), ranged from 0.63 to 0.94 in derivation cohorts and 0.66 to 0.82 in validation cohorts. Commonly identified predictors encompassed recipient-, donor-, and operation-specific factors, with primary diagnosis emerging as the most frequently reported. There is an urgent need for stricter adherence to TRIPOD guidelines and use PROBAST as a methodological reference. Although several prognostic models for PGD have been theoretically developed, none have been translated into clinical practice or evaluated for their impact on clinical outcomes. Future studies should prioritize rigorous methodological design and external validation as well as facilitating the integration of predictive models into clinical decision-making.