# Influential factors related to feeding disorders in preterm infants and the construction of predictive models

## TL;DR

This study identifies risk factors for feeding disorders in preterm infants and builds a predictive model to help clinicians anticipate these issues.

## Contribution

The study introduces a new predictive model for feeding disorders in preterm infants using clinical and laboratory data.

## Key findings

- Lower gestational age, birth weight, and blood calcium levels are significant risk factors for feeding disorders.
- The predictive model achieved high accuracy with an AUC of 0.866 and a prediction accuracy of 91.4%.
- The model combines seven clinical indicators to effectively predict feeding disorders in preterm infants.

## Abstract

To investigate the influencing factors associated with feeding disorders in preterm infants and to construct a prediction model.

314 cases of preterm infants admitted to our hospital from January 2019 to December 2022 were retrospectively analyzed and divided into feeding disorder group and non-feeding disorder group according to the presence of feeding disorder at 37 weeks of corrected gestational age. Statistical analysis of children's general information, hospitalization measures, laboratory tests, feeding time, etc. Multifactorial Logistic regression analysis of the occurrence of feeding disorders related to the influence of factors, and the use of subjects to make a work characteristic curve to analyze the predictive value of the relevant factors on feeding disorders.

Multifactorial logistic regression analysis suggested that lower birth gestational age, birth weight, white blood cell count, absolute value of monocytes, blood calcium value, Apgar score at 1 min after birth, and longer duration of noninvasive ventilation were risk factors for feeding disorders in preterm infants. ROC curve analysis suggested that the area under the curve of the feeding disorders was predicted by the combination of the above seven indexes to construct the feeding disorders prediction model The AUC was 0.866 (P < 0.001, 95% CI 0.801–0.932), and it had a maximum Yoden index of 0.699, an optimal cutoff value of 0.169, a sensitivity of 85.4%, a specificity of 84.5%, and a prediction accuracy of 91.4%.

Lower birth gestational age, birth weight, white blood cell count, absolute monocyte value, blood calcium value, low Apgar score at 1 min after birth, and prolonged noninvasive ventilation are risk factors for feeding disorders in preterm infants, and the present prediction model is a good predictor of the occurrence of feeding disorders in preterm infants.