Pre-eclampsia in Hospital: Early Induction or Expectant management (PHOENIX).

Summary

Women with pre-eclampsia are at risk of morbidity or in severe cases, maternal or fetal death. Delivery after 37 weeks' gestation is routinely indicated, and does not significantly increase operative morbidity in the mother. In contrast, the risk/ benefit of delivery between 34+0 and 36+6 weeks gestation is not clear, as reduced risks to the mother and infant associated with pregnancy cessation, and therefore disease progression, must be balanced against risks of immaturity due to premature birth. In the UK, approximately 10,000 women per year will present with mild to moderate disease during this gestational 'window'; equivalent to approximately 33% of all women who develop preeclampsia. This randomised controlled trial will investigate whether delivery or expectant management leads to better outcomes in both mother and baby (including at two year follow up) in pregnancies affected by pre-eclampsia between 34 and 37 weeks' gestation.

Women with confirmed pre-eclampsia between 34+0 and 36+6 weeks' gestation are randomised to delivery (commencement within 48 hours) or to expectant management. A composite maternal morbidity outcome measure has been selected that clinicians recognise as important, and with sufficient prevalence to allow the trial to be delivered in a feasible timeframe, while aiming to detect a meaningful and plausible relative risk reduction. It would not be feasible to power the study on rare permanent morbidity to the mother. The primary outcome measure is a composite of maternal morbidity adapted from the fullPIERS model for prediction of pre-eclampsia adverse events (von Dadelszen et al, Lancet 2011) which includes central nervous system, cardiorespiratory, haematological, hepatic and renal outcomes together with placental abruption, intensive care unit admission as well as confirmed severe systolic hypertension (>=160mmHg). All these outcomes are included in the recently completed Dutch HYPITAT II trial and will allow prospective meta-analysis.

Taking into account a 5% loss to follow-up and a two-sided significance level of 5%, 900 women will be required to determine at 90% power a 25% relative risk reduction in the adverse maternal outcome from 43% in the expectant management group to 32.25% in the planned delivery group. For the neonatal outcome, these numbers will give sufficient power (93%) to demonstrate a non-inferiority difference between the group proportions of no less than 10% based on the composite of perinatal death and neonatal admission rate of 24% in the expectant group. The long term outcome at two years will use the revised Parent Report of Children's Abilities (PARCA-R), and 345 women in each arm (assuming a 20% loss to follow up) will achieve 94% power to detect a non-inferiority difference in mean of no less than 4 points. Other outcome measures will include complications of planned delivery, including sepsis, thromoembolism and operative vaginal delivery. Health service use outcomes will include antenatal and postnatal inpatient days, as well as maternal and fetal intensive care unit admissions. Patient reported outcome measures that have been developed in our unit will be used. Health economic outcome data will be collected until the infants are two years old, and include tests and procedures on both mother and infants. If clinically important differences between randomised groups are observed at two years, further funding will be sought to examine longer term effects on intelligence quotient at 5 years and other neuropsychological outcomes.