Serial amplitude-integrated electroencephalography trajectories and brain-injury-associated functional immaturity in extremely preterm infants

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Objective To characterize postmenstrual-age (PMA)-specific Burdjalov maturation trajectories in extremely preterm infants and to evaluate how imaging-defined major brain injury was associated with these trajectories and with apparent within-cohort discrimination between imaging-defined groups across PMA windows. Methods We conducted a single-center retrospective cohort study of 156 infants born before 28 weeks’ gestation, comprising 109 infants without major brain injury (reference group) and 47 with brain injury, defined as grade III–IV intraventricular hemorrhage or white-matter injury. These infants contributed 723 analyzable serial amplitude-integrated electroencephalography (aEEG) recordings across PMA 26 to 36 weeks. Burdjalov scores were analyzed using linear mixed-effects models, categorical-PMA contrasts, a patient-clustered ordinal model, complete time-window sensitivity analyses, and joint patient-level bootstrap comparisons of PMA-specific areas under the curve. Results The Burdjalov total score increased by 0.92 points per week of PMA. Infants with brain injury scored on average 1.45 points lower at PMA 32 weeks (95% confidence interval −1.65 to −1.25). Although the continuous-scale model produced a positive PMA-by-group interaction, its magnitude and statistical significance varied across analysis windows and it was not reproduced by the ordinal model ( p = 0.979). Categorical-PMA estimates showed a non-monotonic pattern across the full observation period: the imprecise difference at PMA 26 weeks was followed by the largest observed difference at PMA 28 weeks and progressively smaller differences thereafter. At PMA 36 weeks, 87.4% of reference-group and 27.3% of brain-injury recordings reached the maximum score. The largest observed area under the curve was at PMA 32 weeks (0.886, 95% confidence interval 0.824 to 0.940), but joint patient-level bootstrap comparisons did not distinguish PMA 32 from PMA 28, 30, or 36 weeks. Conclusions Imaging-defined major brain injury was associated with persistently lower Burdjalov scores. The unstable linear interaction on this bounded scale should not be interpreted as biological recovery or catch-up maturation. The retrospective within-cohort discrimination estimates do not establish predictive, diagnostic, screening, or clinical decision utility.