Background: Metabolic acidosis is a frequent and serious complication in critically ill neonates, particularly preterm infants, and is associated with an increased risk of mortality, intraventricular hemorrhage, and long-term neurodevelopmental impairment. Despite limited evidence, sodium bicarbonate (SB) is widely administered in neonatal intensive care units (NICUs) to correct acidosis, largely extrapolated from adult and pediatric practice. However, concerns have been raised about its potential adverse effects, including paradoxical intracellular acidosis, impaired cerebral autoregulation, and increased risk of neurological injury. Given the uncertainty regarding both its efficacy and safety, we conducted a systematic review and meta-analysis to evaluate the role of SB administration in the neonatal population. Methods: MEDLINE, Scopus, and the Cochrane Library were searched using specific medical subject headings and terms. We included all study published up to July 2025 that involved newborns treated with SB. The primary outcome was positive response to treatment, while secondary outcomes included mortality, morbidity, and long-term impairment. Results: We analyzed 10 studies (9 randomized and 1 unrandomized study, including 660 neonates). Pooled results from the randomized controlled studies showed no efficacy of SB in newborns. Data from one unrandomized study showed an increased risk for mortality (OR 13.1 p = 0.02), clinical seizures (OR 2.8, p = 0.01), and a combined outcome of death or neurological damage (OR 3.1 p < 0.01) for neonates treated with SB. Conclusions: Current evidence is insufficient to support the routine administration of SB in NICUs. Neonatologists have the responsibility to administer only drugs of proven efficacy, personalizing therapy on the basis of a pathology’s etiology, in order to reduce risk and optimize benefits. In the absence of robust, statistically significant data, the indiscriminate use of SB should be discouraged in current clinical practice. PROSPERO registration number: CRD420251132502.
Boscarino et al. (Mon,) studied this question.