Sevil Bali, Burcu Cebeci
Background/Objectives: Diazoxide is used for persistent or recurrent hypoglycemia when the clinical and biochemical pattern is compatible with hyperinsulinism, but neonatal treatment populations are heterogeneous. We evaluated the clinical and hormonal characteristics and short-term outcomes associated with diazoxide use in hypoglycemic neonates. Methods: This single-center retrospective cohort included 124 hypoglycemic neonates treated between January 2020 and January 2026; 36 received diazoxide and 88 did not. Critical hormonal samples were obtained concurrently with glucose < 47 mg/dL; in treated infants, sampling preceded the first diazoxide dose. Hormonal effects were estimated with parsimonious multivariable models. Diazoxide-selection factors were examined with a reduced logistic model and Firth sensitivity analysis. Results: Treated infants were more frequently preterm (80.6% vs. 52.3%; p = 0.003) and had greater pre-treatment sepsis and ventilation burden. Severe hypoglycemia < 25 mg/dL did not differ significantly (27.8% vs. 17.0%; p = 0.176). Insulin was higher with diazoxide (median 6.05 vs. 2.77 mU/L; adjusted GMR 2.23, 95% CI 1.38-3.61; p = 0.001), as was C-peptide (adjusted GMR 1.59, 95% CI 1.03-2.47; p = 0.038). Among treated infants, 35/36 had insulin > 1.25 mU/L; 17/29 with available maximum glucose-infusion-rate data required >8 mg/kg/min. Diazoxide began at a median postnatal age of 10.5 days; 30/36 infants remained on therapy at discharge. Fluid retention requiring diuretics occurred in 11/36 (30.6%) versus 11/88 (12.5%; p = 0.017). Pulmonary hypertension was not documented on routine day 5-7 echocardiography (0/36; exact 95% CI 0.0-9.7%). Conclusions: Diazoxide use identified a clinically vulnerable, treatment-selected subgroup with evidence of inadequate insulin suppression during hypoglycemia. The data do not establish uniformly confirmed hyperinsulinism. Fluid retention was the main documented treatment-period adverse event; rare or later cardiopulmonary complications cannot be excluded by this retrospective cohort.