Growth failure prediction in Chile

Area of Science:

• Pediatrics
• Public Health
• Biostatistics

Background:

• Infant growth failure is a significant public health concern.
• Early identification of at-risk infants is crucial for timely intervention.
• Existing screening methods may lack optimal predictive accuracy.

Purpose of the Study:

• To develop and evaluate screening instruments for predicting infant growth failure.
• To identify biological and social risk factors associated with infants being underweight/age.
• To compare the efficacy of different statistical methodologies for screening.

Main Methods:

• Prospective collection of biological and social variables from infants in Santiago, Chile.
• Application of three distinct statistical methodologies: simple relative risk, logistic regression, and classification and regression trees (CART).
• Evaluation of screening instrument performance based on sensitivity and specificity.

Main Results:

• Classification and regression trees (CART) demonstrated superior performance compared to simple relative risk and logistic regression.
• The CART-developed screening instrument showed improved sensitivity and specificity for predicting infant growth failure.
• Identified key biological and social variables as predictors for infants being underweight/age.

Conclusions:

• Classification and regression trees (CART) provide a highly effective methodology for developing infant growth failure screening instruments.
• The developed CART-based screening tool offers enhanced accuracy for identifying infants at risk of being underweight.
• This approach can aid public health initiatives in early detection and intervention for pediatric malnutrition.