RESUMO
PURPOSE: Ultrasound is the primary method for prenatal identification of myelomeningocele and is critical to prognostication and treatment planning. No study has considered the degree of inaccuracy of prenatal US lesion level estimates and anatomic lesion level on postnatal imaging using the weighted kappa coefficient (κw), nor the impact of maternal BMI on agreement. We examined the accuracy of prenatal ultrasound lesion level estimation in a cohort of patients with myelomeningocele using κw and determined whether BMI influenced accuracy. METHODS: The study is a retrospective review including patients born 2011-2019 who had prenatal imaging and primary myelomeningocele closure at a single institution. Lesion levels from prenatal ultrasound and postnatal imaging studies were analyzed for agreement at exact level, within 1 level, and within 2 levels using κw. Maternal BMI was examined for correlation with accuracy. RESULTS: Fifty-seven patients met inclusion criteria. Mean BMI was 31.2. There was no association between maternal BMI and agreement at any level. Lesion level on prenatal ultrasound agreed with postnatal imaging to the exact level in 13 (22.8%) cases, within 1 level in 38 (66.7%) cases, and within 2 levels in 50 (87.7%) cases. Weighted kappa showed moderate agreement at exact level (κw = 0.494) and substantial agreement within 1 (κw = 0.761) and 2 levels (κw = 0.902). CONCLUSION: Weighted kappa adds confidence for clinical decision making by accounting for accuracy. Prenatal ultrasound is a reliable and accurate method of determining lesion level with near-perfect agreement to postnatal imaging within 2 spinal levels. Maternal BMI may not influence lesion level determination after initial diagnosis.
