Evaluation of the reliability of current neonatal malnutrition indicators in diagnosis: An analysis of a retrospective cohort

BACKGROUND: Proposed diagnostic criteria for neonatal malnutrition have not been validated. The objective of this study was to evaluate the reliability of the 2018 neonatal malnutrition criteria for identifying malnutrition in preterm infants. METHODS: A retrospective study of premature infants admitted to a level III neonatal intensive care unit from 2013 to 2022 was performed. Malnutrition diagnoses were assigned using each indicator. Weight and linear growth velocities were defined using the preferred approach (maintaining z-score on Fenton growth chart) and generalized growth goals (weight: 15-20 g/kg/day for infants <2000 g, 20-30 g/day for infants ≥2000 g, length: 1 cm/week). Kappa (k) values assessed the inter-indicator reliability of the full criteria and the inter-indicator and intra-indicator reliability of individual malnutrition criteria. Acceptable reliability was defined as k > 0.8. RESULTS: This study included 2305 infants. The 2018 neonatal malnutrition criteria had poor inter-indicator reliability for diagnosis in the first 2 weeks of life (k = 0.054) and after the first 2 weeks (k = 0.087). Inter-indicator reliability varied but was better among growth-based indicators. Intra-indicator reliability was strong for weight gain velocity definitions (k values ranging from 0.86 to 0.94), but poor for linear growth definitions (k = 0.12). CONCLUSION: Proposed neonatal malnutrition criteria have poor reliability when all indicators are used and limited reliability at the individual indicator level. Utilization of generalized weight gain goals may be appropriate for malnutrition diagnosis, but the generalized linear growth goal of 1 cm/week is not. Differing patterns in malnutrition diagnoses should be investigated for clinical significance. Redevelopment of current indicators is likely warranted.