Development of a Simple Clinical Risk Score for Mortality Prediction in Neonatal Gram-Negative Sepsis.

OBJECTIVES

To identify clinical and laboratory predictors of mortality and develop a simple bedside risk score for neonates with Gram-negative sepsis.

METHODS

A total of 50 neonates satisfying as per the inclusion criteria during the reference period were included in the study. Independent predictors of mortality were identified using univariable and multivariable binary logistic regression and Least Absolute Shrinkage and Selection Operator (LASSO) regression, and were used to develop a Neonatal Risk Scoring System (NRSS).

RESULTS

Mortality occurred in 38% (19/50). On multivariable analysis, necrotizing enterocolitis (NEC) ≥ Stage II (OR 45.24, p = 0.003), refractory shock (OR 6.54, p = 0.047), and leukopenia (OR 20.01, p = 0.018) independently predicted mortality. The NRSS (weights 8, 4, and 6, respectively) showed excellent discrimination (AUC > 0.9). Mean NRSS was higher in non-survivors (9.47 ± 5.37) vs. survivors (1.74 ± 2.35, p < 0.001). A cut-off ≥ 7 predicted mortality with 73.7% sensitivity and 96.8% specificity.

CONCLUSIONS

NEC ≥ Stage II, refractory shock, and leukopenia independently predict mortality. The NRSS provides a simple, reliable tool for early risk stratification.