Introduction

In September 2011, the Society for Maternal-Fetal Medicine issued an opinion on expectant management of severe preeclampsia in specially-selected women who are at less than 34 weeks’ gestation. The Society noted that, “with careful in-hospital maternal and fetal surveillance,” expectant management might improve neonatal outcomes. If the status of the mother or fetus deteriorates, delivery is likely warranted. This recommendation was based on studies that were primarily conducted in developed countries, so it is reasonable to ask whether it is generalizable to nations in the developing world. This month, Journal Club members discussed an article by Vigil-De Gracia et al that addresses this question.

Ebony Boyce Carter, MD and George A. Macones, MD, MSCE, Associate Editor

Introduction

In September 2011, the Society for Maternal-Fetal Medicine issued an opinion on expectant management of severe preeclampsia in specially-selected women who are at less than 34 weeks’ gestation. The Society noted that, “with careful in-hospital maternal and fetal surveillance,” expectant management might improve neonatal outcomes. If the status of the mother or fetus deteriorates, delivery is likely warranted. This recommendation was based on studies that were primarily conducted in developed countries, so it is reasonable to ask whether it is generalizable to nations in the developing world. This month, Journal Club members discussed an article by Vigil-De Gracia et al that addresses this question.

Ebony Boyce Carter, MD and George A. Macones, MD, MSCE, Associate Editor

Background

Macones: Is this an important research question, and if so, why?

Fjeldstad: The authors of the article posed the question of whether expectant management of women with severe hypertensive disorders occurring at 28-33 weeks’ gestation improves perinatal outcomes, as compared to prompt delivery after steroid administration. This is an important question because, as the authors point out in their discussion, preeclampsia/eclampsia is the primary cause of maternal mortality in Latin American countries. But delivery is the only known treatment, and early preterm delivery is associated with poor neonatal outcomes. In order to make recommendations for optimal management in this patient population, it is necessary to determine whether prolonging time to delivery (ie, treatment) with expectant management improves perinatal outcomes or whether management increases the risk for maternal morbidity and mortality.

As cited in the article, 2 prior randomized studies also investigated maternal and neonatal complications after conservative management of preeclampsia remote from term, and both demonstrated a reduction in neonatal complications with conservative management. However, these studies were limited by small sample sizes; therefore it is essential to gather more data on this subject, as this study aimed to do, in order to determine guidelines for management in this setting.

Study Design

Macones: What was the overall study design? What types of sites were included?

Fjeldstad: This was a randomized, multicenter, parallel, open-label clinical trial that was conducted between August 2010 and August 2012. The study was open label due to the clearly identifiable nature of the interventions. Eight tertiary teaching hospitals in Latin America were included, all with experience in managing severe preeclampsia. These sites were selected based on their ability to immediately respond to obstetric emergencies and their capacity for 24-hour neonatal intensive care. Each hospital also had experience with conducting clinical trials.

Macones: What were the inclusion and exclusion criteria?

Theilen: Pregnant women with singleton or twin gestations and severe hypertensive disorders at 28-33 weeks of gestation were included. Severe hypertensive disorders included severe preeclampsia, severe gestational hypertension, and chronic hypertension with superimposed preeclampsia. Grouping these 3 entities together allowed a larger number of women with significant hypertensive disease during pregnancy to be included. However, there may be distinctions between these groups in terms of underlying pathophysiology, and this might affect pregnancy outcomes.

Exclusion criteria consisted of gestational age less than 28 weeks, eclampsia, HELLP syndrome (hemolysis, elevated liver enzyme levels and a low platelet count), preeclampsia with renal failure or pulmonary edema, active vaginal bleeding, ruptured membranes, placenta previa, pregestational or gestational diabetes, preexisting renal disease, autoimmune disease, major fetal anomalies, fetal growth restriction (defined as estimated fetal weight on ultrasound less than the tenth percentile for gestational age), oligohydramnios, and reverse umbilical artery Doppler flow. These criteria excluded patients with contraindications to expectant management and in whom poor outcomes could be attributable to other comorbidities.

Macones: Discuss the interventions and primary and secondary outcomes.

Theilen: Patients in both groups received glucocorticoid therapy, were placed on bed rest, initially received intravenous magnesium sulfate for 24-48 hours, and were given bolus doses of antihypertensives to control blood pressures greater than 160/110 mmHg. In addition, they underwent laboratory evaluation on admission and daily for 48-72 hours, had ultrasounds to assess fetal growth and amniotic fluid volume on admission, and had daily nonstress tests to evaluate fetal wellbeing.

In the prompt delivery group, patients were delivered 24-72 hours after administration of glucocorticoids, and magnesium sulfate was continued until 24 hours postpartum. In the expectant management group, patients were delivered at 34 weeks of gestation or when specific maternal or fetal indications were present, including uncontrollable hypertension, nonreassuring fetal heart tracing, abnormal fetal testing, fetal growth restriction, placental abruption, decline in renal function, HELLP syndrome, and persistent headache or vision changes or epigastric pain. Magnesium sulfate was restarted when delivery was indicated and was continued for 24 hours postpartum. Laboratory testing was repeated twice weekly, and serial ultrasounds were performed every 1-2 weeks. Oral antihypertensive medications, including methyldopa, nifedipine, and hydralazine, were only used as maintenance therapy in some participating hospitals at the discretion of the treating physician.

The primary outcome was perinatal (fetal and neonatal) mortality. The secondary outcomes were composite neonatal morbidities, neonatal data, and maternal morbidities and mortality. Neonatal morbidities included respiratory distress syndrome, intraventricular hemorrhage, necrotizing enterocolitis, and neonatal sepsis. Neonatal data included birthweight, small size for gestational age, Apgar scores, neonatal intensive care unit (NICU) admissions, and NICU length of stay. Maternal morbidities included placental abruption, pulmonary edema, HELLP syndrome, renal insufficiency, eclampsia, and disseminated intravascular coagulation.

Macones: Can you describe the process of randomization and the block sizes?

Carter: Block randomization was used to divide the patients into block sizes of 4 and 6. Double-blinding was not possible since patients were randomized to prompt delivery or expectant management, but the authors tried to prevent selection bias to the extent possible through central assignment of blocks, concealment of allocation to study staff, and use of variable block sizes.

Macones: What analytic approach and sample-size estimate were used?

Carter: When investigators estimated the sample size, they assumed a 15% perinatal mortality rate, based on a 1990 South African study by Odendaal et al. The authors stated that the high perinatal mortality rate in the study was reflective of Latin American countries. However, the perinatal mortality in the prompt delivery group in the cited study was 5 out of 20 patients (25%). We would have liked further insight from the authors regarding how they arrived at this estimate. Nonetheless, based on the author’s assumptions, the study was adequately powered.

Macones: Can you walk us through the CONSORT flow diagram? Why is this important?

Fjeldstad: The CONSORT flow diagram depicts the study participants’ passage through the 4 trial stages; their enrollment, allocation, follow-up, and analysis. Of 425 eligible women, 267 agreed to participate—133 and 134 women were randomly selected to the prompt delivery and expectant management groups, respectively. In the expectant management group, 3 women did not complete the study—2 because they refused to remain in the hospital and 1 because a physician’s error resulted in incorrect management—and these data were excluded from the analysis. This omission is important to note because it could confound some of the findings for that group in the paper, including the secondary outcomes of increased rates of placental abruption and small size for gestational age. Even with this omission, the sample size in both groups is still large enough to preserve statistical power.