Predicting Harm: A Systematic Review of Pediatric Pressure Injury Models and Their Clinical Shortcomings

Area of Science:

• Pediatric healthcare
• Clinical prediction modeling
• Patient safety

Background:

• Pediatric pressure injuries (PIs) are a significant clinical concern.
• Existing risk prediction models for pediatric PIs are underdeveloped.
• A systematic review is needed to evaluate current models.

Purpose of the Study:

• To critically evaluate existing pediatric PI risk prediction models.
• To assess their methodological rigor, predictive performance, and clinical applicability.
• To identify gaps and inform future model development.

Main Methods:

• Systematic review following PRISMA 2020 guidelines.
• Searched nine databases for PI prediction models in patients <18 years.
• Assessed risk of bias and applicability using PROBAST and PROBAST + AI tools.

Main Results:

• Twelve models were included; all showed acceptable discrimination (AUC 0.612-0.978).
• Calibration reporting was inconsistent; only two models assessed clinical utility.
• All models had high risk of bias, with major concerns regarding applicability and generalizability.

Conclusions:

• Current pediatric PI prediction models are methodologically flawed and lack clinical applicability.
• Future models need to adhere to reporting standards (PROBAST, PROBAST + AI), use multicenter data, and stratify by age/setting.
• Improvements are crucial for early PI identification and prevention in pediatric care.