Nomogram predicts hydrocephalus risk in children with bacterial meningitis

A recent study has designed a predictive model that can be used by clinicians to determine the risk of hydrocephalus in children with bacterial meningitis.

A team of investigators conducted a single-centre retrospective study using data of children with bacterial meningitis admitted to a hospital between January 2016 and December 2020. They evaluated risk factors using univariate analysis and developed the predictive model or nomogram using binary logistic analysis.

The predictive performance of the model was assessed via a nomogram calibration plot, Hosmer‒Lemeshow test, and receiver operating characteristic (ROC) curve. Internal validation was processed by ordinary bootstrapping.

A total of 283 patients were enrolled in the study, of whom 41 (14.49 percent) had confirmed bacterial meningitis-associated hydrocephalus (BMAH). The incidence of sequelae in patients with BMAH was 22 out of 27 (88.9 percent), and this was markedly higher than in those without BMAH.

Fourteen clinical indicators were found to be related to BMAH in univariate analysis. Multivariate analysis identified four variables as independent risk factors, namely repeated seizures, loss of consciousness, procalcitonin ≥7.5 ng/dL, and mechanical ventilation. Using these variables, a graphical nomogram was designed.

The area under the ROC curve of the predictive model was 0.910 (Hosmer‒Lemeshow test: p=0.610). Its mean absolute error in the calibration plot was 0.02. Of note, internal validation confirmed the agreement of the testing set with the original set.