86 - Gastrostomy Tube Placement in Large-for-Gestational-Age Infants: Frequency, Associated Co-Morbidities, and Outcomes

Publication Number: 3082.86

Samuel Mikhail, Nemours Children's Hospital, Melbourne, FL, United States; Harish Chandra Dega, Nemours Children's Hospital, Jersey City, NJ, United States; Kera M. McNelis, Emory University School of Medicine, Atlanta, GA, United States; Sreekanth Viswanathan, Nemours Children's Hospital, ORLANDO, FL, United States

Background: Gastrostomy tube (GT) placement is an intervention for oral feeding difficulties, yet its frequency and practice variation among large-for-gestational-age (LGA) infants remain unclear.

Objective: To estimate GT placement frequency among LGA infants in US neonatal intensive care units (NICUs), quantify statewide variation in GT use, and identify associated co-morbidities and predictors of placement.

Design/Methods: Retrospective cross-sectional analysis of a de-identified, multi-health-system repository (Epic Cosmos). LGA infants ≥35 weeks’ gestation with NICU-Level conditions (2020–2024) were included. GT placement was ascertained using CPT/ICD-10 procedure codes. We summarized cohort characteristics and fit mixed-effects logistic regression with a state random effect, adjusting a priori for preterm status (35–36 vs 37–40 weeks), birthweight, maternal diabetes/obesity, race/ethnicity, 5-minute Apgar, and major morbidities. We report adjusted odds ratios (aORs) with 95% CIs and median odds ratio (MOR) to quantify between-state heterogeneity.

Results: Among 431,908 LGA infants, 3,670 received a GT (0.85%). Rates were higher in preterm (1.85%) vs term (0.53%; risk ratio 3.50; p< 0.001). State-level GT use varied from 0.10%-1.94% (19.7-fold difference); the state random effect was significant (p < 0.001) with MOR 10.36. Multivariable regression showed co-morbidities that were the strongest independent predictors of GT placement (all p< 0.001): Bronchopulmonary Dysplasia (aOR 39.73), Necrotizing Enterocolitis (aOR 6.56), and Intraventricular Hemorrhage (aOR 4.09). Other significant predictors included Hypoxic-Ischemic Encephalopathy (aOR 2.25), Sepsis (aOR 1.84), and Feeding Difficulty (aOR 1.63). Cardiac surgery was associated with a near-100% GT rate. After adjustment, significant unexplained state-level variation persisted, quantified by a MOR of 10.36 (p < 0.001). This indicates that two identical infants from different states have over 10x different odds of receiving a GT based solely on their location.

Conclusion(s): GT placement in LGA infants is rare (0.85%) and strongly associated with severe neonatal co-morbidities. This study has a limitation in the retrospective database design that relies on accuracy of entered CPT/ICD-10 codes. The finding of extreme, unexplained 19.7-fold statewide variation (MOR 10.36) deserves further study for contributing factors. These results highlight a critical opportunity for quality improvement through the development and implementation of standardized feeding and GT placement protocols for LGA infants in the NICU.