Higher versus lower protein intake in formula-fed low birth weight infants.

Systematic review/meta-analysis of formula-fed hospitalized neonates weighing less than 2.5 kg at birth, including preterm and low birth weight infants who were appropriate-for-gestational-age or small-for-gestational-age. Included trials compared different formula protein doses during hospitalization, with several small parallel-group studies and one crossover study.

Formula protein intake was compared across randomized/controlled studies in hospitalized preterm or low birth weight infants during the initial hospital stay. The main comparison was higher protein intake at about 3.0 to <4.0 g/kg/day versus lower intake <3.0 g/kg/day; a few studies also evaluated very high intake ≥4.0 g/kg/day, including up to 6 to 7.2 g/kg/day.

Higher formula protein intake generally improved short-term growth and nitrogen retention. Across individual trials, weight gain was higher with higher protein, including 20.3 vs 18.9 g/kg/day in Bhatia 1991, 17.4 vs 12.9 g/kg/day in Hillman 1994, 18.3 vs 13.9 g/kg/day in Kashyap 1986, 16.45 vs 11.54 g/kg/day in Wauben 1995, and 23.1±7 vs 16.7±6 g/kg/day in the Cooke 2006 crossover study. Higher protein also increased nitrogen accretion in Kashyap 1986 (422.2 vs 270 mg/kg/day) and was associated with higher blood urea nitrogen in some studies, such as 9.85 vs 7.62 mg/dL in Bhatia 1991 and 3.1 vs 1.4 mg/dL in Kashyap 1986. There were no consistent differences in necrotizing enterocolitis, sepsis, or diarrhea, but very high protein intake had limited evidence and included reports of uremia in a subset of infants.

The evidence base was small, with several trials enrolling few infants and using different protein doses, formulas, and outcome definitions. Long-term outcomes such as neurodevelopment were sparsely studied, and safety data for very high protein intake were limited. Some biochemical adverse signals, including higher blood urea nitrogen and occasional uremia, raise uncertainty about optimal upper protein limits.