Good performance of bioimpedance in early pregnancy to predict preeclampsia.

INTRODUCTION

Preeclampsia (PE) affects 2-8% of pregnancies and is one of the main causes of maternal morbidity and mortality worldwide. Early identification of pregnant women at higher risk for PE would allow the use of interventions to reduce adverse maternal and perinatal outcomes.

OBJECTIVE

To assess the ability of bioelectrical impedance analysis (BIA) in pregnancy to predict the development of PE.

METHODS

This prospective cohort involved healthy nulliparas who underwent BIA at 17-20 weekś gestation and were followed until delivery. We used univariate and multivariate logistic regression to assess the ability of BIA measures to predict the occurrence of PE. We used an adjusted regression model to estimate the probability of developing PE, the Hosmer-Lemeshow test to assess the adequacy of the final model, and ROC curves to assess the sensitivity and specificity of different BIA measures in the prediction of PE.

RESULTS

Twelve (6.1%) of the 196 participants developed PE. In the final multivariate model, the following BIA measures were associated with the occurrence of PE: extracellular water/intracellular water ≤ 0.618, skeletal muscle mass ≥ 25 Kg, and body fat percentage ≥ 44%. The combination of these three measures had a predictive accuracy of 83.7%, a sensitivity of 83.3%, a specificity of 83.7%, and a negative predictive value of 98.7% for PE.

CONCLUSION

BIA done on nulliparous women at 17-20 weekś gestation has a good accuracy and high negative predictive value for the risk of developing PE.