Baseline bicarbonate level is associated with apnea of prematurity in preterm neonates with neonatal jaundice: insights from the Medical Information Mart for Intensive Care III (MIMIC-III) database.

BACKGROUND

Apnea of prematurity (AOP) and neonatal jaundice are among the most common diagnoses in the neonatal intensive care unit (NICU). This study aims to evaluate the association between baseline serum bicarbonate levels and the risk of AOP in preterm neonates with neonatal jaundice.

METHODS

Using data from the MIMIC-III database (2001-2012), we identified 2,170 preterm neonates diagnosed with neonatal jaundice. Baseline bicarbonate was defined as the initial serum level measured within 24 hours of admission and categorized into quartiles (Q1-Q4). Multivariate logistic regression, mediation analysis, propensity score matching (PSM), subgroup analyses, and restricted cubic spline (RCS) analysis were performed.

RESULTS

Baseline bicarbonate levels showed a significant inverse association with AOP risk when analyzed as a continuous variable {Model 1: odds ratio (OR) =0.964 [95% confidence interval (CI): 0.937-0.992], P=0.01; Model 2: OR =0.919 (95% CI: 0.895-0.944), P<0.001; Model 3: OR =0.912 (95% CI: 0.888-0.936), P<0.001}. When bicarbonate levels were divided into quartiles, individuals with baseline levels in the Q2 range exhibited the lowest risk of AOP [Model 1: OR =0.545 (95% CI: 0.423-0.702), P<0.001; Model 2: OR =0.383 (95% CI: 0.300-0.488), P<0.001; Model 3: OR =0.363 (95% CI: 0.285-0.461), P<0.001; Model 4: OR =0.758 (95% CI: 0.579-0.992), P=0.044]. Further evidence from mediation analysis and subgroup analysis following PSM consistently supported the mediating role of caffeine treatment (indirect effect: β=-0.059, P<0.001; direct effect: β=0.048, P=0.97). RCS analysis revealed a non-linear relationship between bicarbonate levels and AOP risk (P for nonlinear <0.05).

CONCLUSIONS

This study found that lower baseline bicarbonate levels were linked to higher AOP risk in preterm neonates with jaundice. Caffeine treatment fully mediated this nonlinear association, and those with bicarbonate levels in Q2 (16-19 mEq/L) had the lowest AOP risk.