Factors predicting mortality in newborn ventilation

Objective: Prediction of mortality among newborns on mechanical ventilation is difficult. Our aim was to develop a scoring system for predicting mortality among such neonates.

Methods: This multi centre prospective study was performed to develop and validate a scoring system among two equal cohorts of ventilated newborns in India. Mechanical ventilator was used in pressure-limited time-cycled mode. Arterial blood gas, initial pulmonary pressures, septicemia screen along with other basic parameters were recorded in a pre-structured proforma. Blood samples were analyzed for malondialdehyde to determine the possible role and predictive validity of free radical injury. Multiple logistic regression analysis was done to find out independent predictors of mortality for the variables those were significantly associated with outcome after univariate analysis.

Results: On univariate analysis, birth-weight, oxygenation-index, septicaemia, malondialdehyde level, and inotropic support were significantly associated with mortality. However, after multiple regression analysis gestational-age, pH and FiO₂ lost their significance as predictors. According to cut-off values of ROC-curve, a scoring system ranging from 0 to 20 and four mortality risk groups were developed. Area under ROC-curve was 0.94, compared to 0.90 for both APACHE-III and CRIB-scores; and 0.92 for PRISM-score.

Conclusions: Birth-weight, oxygenation-index, malondialdehyde level, inotropic support, and septicemia are independent mortality predictors of neonatal ventilation. Increase in malondialdehyde level is associated with higher mortality rate, indicating possible role of free radical injury. Pediatr Pulmonol. 2015; 50:271-275. © 2014 Wiley Periodicals, Inc.