PP053. Development and validation of a model for prepregnant prediction of recurrence of early-onset preeclampsia

S M Van Kuijk; D H Delahaije; L L Peeters; C D Dirksen; H C Scheepers; L Smits; PreCare study group

doi:10.1016/j.preghy.2012.04.164

PP053. Development and validation of a model for prepregnant prediction of recurrence of early-onset preeclampsia

Pregnancy Hypertens. 2012 Jul;2(3):270-1. doi: 10.1016/j.preghy.2012.04.164. Epub 2012 Jun 13.

Authors

S M Van Kuijk¹, D H Delahaije², L L Peeters³, C D Dirksen², H C Scheepers³, L Smits⁴; PreCare study group

Affiliations

¹ Epidemiology, Maastricht University, Maastricht, The Netherlands; Obstetrics & Gynecology, Maastricht, The Netherlands.
² Department of Clinical Epidemiology and Medical Technology Assessment, Maastricht University Medical Center, Maastricht, The Netherlands.
³ Obstetrics & Gynecology, Maastricht, The Netherlands.
⁴ Epidemiology, Maastricht University, Maastricht, The Netherlands.

PMID: 26105376
DOI: 10.1016/j.preghy.2012.04.164

Abstract

Introduction: Women with a history of early-onset preeclampsia, requiring delivery before 34 weeks of gestation, often receive intensive surveillance with far more visits than routine antenatal care, additional testing such as the serial measurement of various fetal arterial Doppler blood flow velocity profiles, repetitive assessment of the fetal biometry, the amount of amniotic fluid, the fetal biophysical profile, and repetitive blood testing. Yet, recurrence risk is generally low.

Objectives: To develop and validate a prepregnant prediction model to identify women at very low risk of recurrence of early-onset preeclampsia. These women may be reassured and offered routine antenatal care.

Methods: For the derivation of the model, we enrolled 407 pregnant women from 5 Dutch hospitals who had experienced early-onset preeclampsia in their previous pregnancy. Based on previous published evidence, we selected five predictor variables (gestational age at the time of previous birth, prior small-for-gestational-age (SGA) newborn, fasting blood glucose, body mass index (BMI) and the presence or absence of chronic hypertension) to be entered in a logistic regression model. Discrimination and calibration measures were evaluated after an internal validation step using standard bootstrapping techniques. After the model was built, we enrolled another 200 women to externally validate the model. For the external validation study, 6 more hospitals provided patients.

Results: The individual risk of recurrence of early-onset preeclampsia using our formula can be calculated as follows: P(recurrence)=1/(1+e(-(linear predictor))), with linear predictor=0.29-0.42*fasting blood glucose (mmol/L) + 0.59* hypertension (yes/no) - 0.01*gestational age at the time of previous birth (days) - 0.41*prior SGA (yes/no)+0.01*BMI (kg/m(2)). After internal validation, the area under the receiver operating characteristic (ROC) curve of the model was 0.65 (95% CI: 0.56-0.74) in the development sample, and was higher in the external validation sample (AUC=0.76, 95% CI=0.58-0.96), indicating that the model discriminates well between women who will develop a recurrence and those who will not. Using a predicted risk threshold of, for example, 4.6%, about one-fourth of the population would be regarded low-risk with a negative predictive value of 100%. Calibration was satisfactory in both samples.

Conclusion: Our model is helpful in the identification of women at very low risk of recurrent early-onset preeclampsia, and may be used to stratify women into normal care and intensified care groups. At present, we are conducting the PreCare study, in which we assess the effects and costs of introducing our prediction model into routine clinical practice.