Development of an anthropometric-based algorithm to estimate feeding tube insertion depth in neonates

To develop an anthropometric-based algorithm for estimating feeding tube (FT) insertion depth in neonates using retrospectively collected radiographic data.

This observational study reviewed records of infants admitted to a Danish level IV Neonatal Intensive Care Unit from January 2023 to December 2024. Infants with FT placement, using the conventional Nose-Ear-Xiphoid (NEX) method, and radiographic verification within 24 h were included. Radiographically verified FT position and anthropometric data were used to develop new linear and multivariate quadratic models to optimize placement.

Among 198 infants were included (median gestational age [GA] of 35 + 0 weeks [interquartile range, IQR:28 + 6-38 + 1]), 301 radiographic images showed, 144 FTs located at lumbar vertebra 1 (L1), 139 within one vertebra of L1, and 18 outside the stomach. Radiologists recommended adjustment in 107 images. Linear models (weight- or length-dependent) showed strong predictive power (R² > 0.84), while the multivariate quadratic model reduced bias and achieved R² = 0.88, confirmed by cross-validation. Prediction accuracy within the predefined target range (L1 ± 1.5 cm) was 77% n, compared with 64% of clinically acceptable NEX placed tubes (p < 0.001). The optimal algorithm was: FT depth = -3.46 + 0.00229 × weight (g) - 9.48 × 10^-8 × weight (g)² + 0.5992 × length (cm) -0.005918 × length(cm)² + 0.07808 × correted GA(w).

The quadratic prediction model showed good internal performance and reduced bias at anthropometric extremes, offering a standardized approach to optimize FT placement and enhance safety in neonatal care. Prospective external validation is required before clinical implementation.