Prediction Models and Risk Factors for Steroid Resistance in Children with Nephrotic Syndrome: A Systematic Review...

Area of Science:

• Pediatric Nephrology
• Clinical Epidemiology

Background:

• Steroid resistance in pediatric nephrotic syndrome (SRNS) is linked to poor prognosis.
• Predictive models and risk factors for SRNS are not well-established.

Purpose of the Study:

• To systematically review and meta-analyze studies developing SRNS prediction models.
• To identify risk factors associated with SRNS.

Main Methods:

• Comprehensive literature search across multiple databases (PubMed, Embase, etc.) up to March 2025.
• Random-effects meta-analysis of odds ratios and AUC for prediction models.
• Risk of bias assessment using PROBAST and Newcastle-Ottawa Scale.

Main Results:

• 23 studies were included from 2264 initially identified.
• Logistic regression was the primary method for model development (94.1%).
• Reported AUC for models ranged from 0.75 to 0.88, with limited external validation.
• 22 independent risk factors were identified, with five novel factors not previously included in models.
• Significant risk of bias was noted in 76% of model studies and 26% of risk factor analyses.

Conclusions:

• Current SRNS prediction models exhibit good discrimination but are limited by high risk of bias and insufficient external validation.
• Several important risk factors are not yet incorporated into existing models.
• Future research must focus on robust model development and multi-center external validation for clinical utility.