Perinatal morbidity prediction utilizing ultrasonographic central adiposity measures compared to BMI in patients with BMI≥40

Abstract

We observed that a 3-variable model of late 3rd trimester ultrasonographic (US) measures of central adiposity is superior to BMI alone in predicting cesarean delivery (CD) in patients with BMI≥40 kg/m2. As these patients are more likely to experience maternal and neonatal morbidity, we evaluated the predictive ability of these measurements as well as a new primary model for adverse perinatal outcomes and compared these to BMI alone. Planned secondary analysis of a single center prospective cohort study of patients with BMI≥40 at >36 weeks’ GA and viable singletons. Patients underwent a single study visit >36 weeks’ (blinded to clinicians) at which BMI and 19 US/anthropometric central adiposity measures were obtained by trained personnel. Exclusions were contraindications to vaginal birth or prior CD. For this analysis, the primary outcome was composite perinatal morbidity (maternal puerperal infection, umbilical cord arterial pH< 7.1, NICU admission and other neonatal morbidities, Table). Receiver operator characteristics (ROC) curves with area under the curve (AUC) assessed the original 3-variable model’s predictive ability for the composite; AUCs for BMI, the 3-variable model from the primary analysis, and a new best-fit forward selection (α =0.05) model for the composite were compared. From 2017-2020, 149 women (mean BMI 45.4±5.2) were enrolled. The rate of perinatal morbidity was 18%. Multivariable logistic regression retained only one measurement, minimum pre-peritoneal subxiphoid fat depth, that was significantly associated with the perinatal composite (Model 1), but was poorly predictive (AUC 0.64 95%CI 0.52-0.76). BMI had no predictive ability (AUC 0.49 [0.37-0.61]), while our original 3-variable model (Model 2) had poor prediction, similar to Model 1 (Table). In patients with BMI≥40, maternal US fat depth measurements are no more predictive of adverse perinatal outcome than BMI alone, itself a poor predictor. As perinatal outcome may reflect other maternal and labor characteristics, incorporating these into models may improve prediction.