A simple preoperative index for neurological deterioration in children with cervical and thoracic split cord malformation: development and temporal validation of the P(ND) score

Abstract

Background: Children with split cord malformation (SCM) of the cervical and thoracic spine often present only after a neurological deficit has developed, and no validated tool estimates who will deteriorate before surgery. We developed a simple preoperative index, P(ND), from routine imaging and neurophysiology and tested it in a later cohort. Methods: Single-centre study of 42 children with cervical or thoracic SCM operated on in 2015–2026: a retrospective derivation cohort (n = 22, 2015–2021) and a prospective temporal validation cohort (n = 20, 2021–2026). Candidate predictors were tested against neurological status with Kendall’s τ-b and combined into a weighted index. Discrimination was assessed by ROC analysis. Results: Spur asymmetry (τ-b = −0.72; p < 0.001), abnormal tibial somatosensory evoked potentials (τ-b = 0.60; p = 0.01) and the number of involved segments (τ-b = 0.45; p = 0.02) were retained, giving P(ND) = 0.41·S + 0.34·E + 0.25·g(Seg). Across 42 children (32 with deterioration), the AUC was 0.924 (95% CI 0.843–1.000); at P(ND) ≥ 0.60, sensitivity was 93.8% and specificity 90.0%. Children treated under the P(ND)-guided protocol with ventral autologous duraplasty improved more often than historical controls (80.0% vs 40.9%; p = 0.014). Conclusion: P(ND) is a three-item index computed from standard MRI, CT and SSEP data that discriminated children with preoperative deterioration well in a single centre. External validation is needed before routine use.