Early versus Late Ventricular Intervention Study (ELVIS) in post-hemorrhagic ventricular dilatation: Bayesian reanalysis of brain injury and outcomes.

Abstract

To assess the probability of benefit and harm with early versus late intervention for post-hemorrhagic ventricular dilatation (PHVD). Bayesian logistic regression analysis was conducted using data from 126 preterm infants (64 Early; 62 Late) with progressive PHVD randomized to early (ventricular indices, VI >97th percentile) or late (VI >97th percentile +4 mm) intervention thresholds. Four potential prior distributions were tested: optimistic, skeptical, neutral, and informative. Posterior estimates were generated for secondary brain injury, and death or neurodevelopmental impairment (NDI) at 2 years. Early intervention was associated with smaller ventricles and lower global brain injury scores on term-equivalent age MRI. Across all priors, late intervention was associated with higher odds of adverse outcomes of death or NDI. Posterior odds ratios (OR) for late versus early treatment ranged from 1.56 to 2.42, corresponding to a 91.1-99.7% posterior probability of harm. Under a neutral prior, the estimated OR was 2.09 (95% credible interval 1.01-4.40). The absolute risk difference in the 2-year outcome between groups was 16%, corresponding to a number needed to treat of 7. Early, cUS-based intervention for progressive PHVD confers a high probability of improved neuroimaging and neurodevelopmental outcomes at 2 years compared with later intervention. ISRCTN43171322. Bayesian reanalysis of the randomized ELVIS (Early versus Late Ventricular Intervention Study) trial demonstrates a high probability (>90%) that delaying intervention for progressive post-hemorrhagic ventricular dilatation (PHVD) increases the risk of death or moderate to severe neurodevelopmental impairment at 2 years. Early cranial ultrasound-guided intervention was associated with smaller ventricular size and significantly lower global brain injury scores on term-equivalent age MRI, supporting a reduction in secondary brain injury. The absolute risk reduction of 16% translates to a number needed to treat of 7, providing clinically interpretable evidence that earlier intervention is a modifiable strategy to improve neurodevelopmental outcomes in preterm infants with progressive PHVD.