The aim of this study is to systematically compare early initiation of delivery and expectant management in case of preterm prelabour rupture of membranes in terms of neonatal sepsis and RDS, maternal health, health-related quality-of-life and costs. As it is impossible to blind the healthcare workers and patients involved for the strategy of allocation, we will use a multicentre randomised controlled open label trial to assess the effects of initiation of delivery or expectant management on neonatal outcome. This study is set in the Dutch Obstetric Consortium, a collaboration of obstetric clinics in the Netherlands. Approximately 40 clinics, including academic hospitals, non-academic teaching hospitals and non-teaching hospitals will participate in this trial.
Women presenting with preterm prelabour rupture of the foetal membranes between 34+0 and 37+0 weeks' gestation who have not delivered within 24 hours after rupture of the foetal membranes are eligible for participation in the PPROMEXIL-trial. Even so women presenting with preterm prelabour rupture of foetal membranes after 26+0 weeks gestation who have not delivered at 34+0 weeks of gestation are eligible for participation. Both women with singleton and women with multiple gestations can be included. Women with a child in breech presentation can also be included, and both an elective caesarean section and a vaginal delivery are allowed for these women.
Women with monochorionic multiple pregnancies, abnormal (non-reassuring) cardiotocogram (CTG), meconium stained amniotic fluid, signs of intrauterine infection, major foetal anomalies, labour, HELLP syndrome or severe pre-eclampsia, will not be included in the study.
Procedures, recruitment, randomisation and collection of baseline data
The research nurse and/or the staff of participating hospitals will identify eligible women. After the subject has given informed consent for participation in the study, she will be randomised to either a policy that aims at termination of the pregnancy (intervention group) or a policy that aims at expectant management for spontaneous delivery (expectant group). Once the subject data have been entered anonymously in a web-based database, either a staff member or the research nurse will randomise using an internet based procedure. Randomisation will be 1:1 for intervention or expectant management and stratified for centre and previous delivery.
At study entry baseline demographics, past obstetric and medical history will be recorded. Maternal temperature is measured and baseline blood samples are taken. From all participating women a vaginal swab is collected either at admission to the hospital with premature ruptured membranes or at study entry. Antibiotics are given in accordance to local policy. All women will undergo an ultrasound examination at study entry. If a previous ultrasound examination was done within a fortnight before study entry in the same hospital, which revealed no abnormal results, these outcomes may be recorded at study entry. Women fill out a baseline quality of life questionnaire containing EuroQoL, 5D3L, HADS, SF-36 and background questions. They are also asked on their prior intervention preference.
At local centres data-collection will be the responsibility of the local research coordinator and the regional research nurses. The data for this study must be collected, coded and processed with adequate precautions to ensure patient confidentially.
Women randomised for intervention will be planned for induction of labour or elective caesarean section as soon as reasonable after randomisation. Because of the fact that randomisation will take place preferable 24 hours after rupture of the foetal membranes it is imaginable that due to logistic reasons induction of labour is postponed until the following morning, but within 12 hours after randomisation. Induction of labour will take place according to local policy. In case of breech presentation, a primary caesarean section is allowed based on the preference of both the patient and the gynaecologist. Similarly in case of a dichorionic twin an elective caesarean section may be performed because of the position of the presentation of the first twin.
Women randomised for expectant management will be treated according to local policy. This might be either in an outpatient or inpatient setting. If a patient in the expectant group reaches 37+0 weeks of gestation age, induction of labour is performed according to local policy. Whenever a patient with an indication for elective caesarean section will be allocated to expectant management the caesarean section will be performed as soon as labour commences.
Follow up of women and infants
Information is obtained on the condition, weight, length and neonatal morbidity and mortality from the infant and maternal records as well as maternal complications and length of stay. If applicable, neonatal antibiotic treatment, blood samples and cultures are recorded. In case of admittance of the baby to the neonatal intensive care, high care or medium care unit, details of this admittance are also documented. The placenta will be sent for histological examination on chorioamnionitis and funisitis.
Six weeks and six months post partum neonatal length, weight, neurological disabilities, physical disabilities and maternal disabilities will be recorded. Patients will be asked to fill out follow-up quality of life questionnaires including EuroQoL, 5D3L, HADS, SF-36, SCL-90 also concerning the development of the child. Women are also asked to (re)state their intervention preference. Long-term follow up of the children is not yet planned, and depends on future funding.
The primary outcome measure is neonatal sepsis. Neonatal sepsis is defined as a positive blood culture, biochemical infection parameters (C-reactive protein (CRP) above 20 mg/l) or clinical signs of infection (apnoea, fever, intolerance for feeding, respiratory distress and/or haemodynamic instability) with positive surface cultures. An independent panel of paediatricians will define between proven or probable sepsis without knowledge of the outcome data.
Secondary infant outcome measures are respiratory distress syndrome (RDS) (defined according to the Organ dysfunction criteria), transient tachypnoea of the newborn asphyxia (defined according to Sarnat), pneumothorax/pneumomediastinum, late onset sepsis, hypoglycaemia, meconium aspiration syndrome, necrotizing enterocolitis (NEC) (defined according to Bell staging), hyperbilirubinemia, in, traventricular haemorrhage, periventricular leucomalacia, convulsions, other neurological abnormalities and congenital abnormalities [12–14].
Secondary maternal outcome measures are ante partum haemorrhage, umbilical cord prolapse, signs of chorioamnionitis (defined as fever before or during labour as a temperature greater than 37,5°C on two occasion more than one hour apart or a temperature > 38,0°C with either uterine tenderness (or contractions), leucocytosis, maternal or foetal tachycardia, or a foul-smelling vaginal discharge in absence of any other cause of hyperpyrexia), maternal sepsis (defined as temperature > 38.5°C and positive blood culture or circulatory instability requiring intensive care monitoring), thrombo-embolic complications, urinary tract infection treated with antibiotics, signs of endometritis (defined as temperature of > 38,0 °C on 2 occasions at least one our apart after the first 24 h post-partum with associated uterine tenderness ), pneumonia, anaphylactic shock, HELLP-syndrome and death, incidence of instrumental deliveries, maternal quality of life, maternal intervention preference and costs. 
Other outcomes are direct medical and non-medical costs generated by maternal and neonatal resource utilisation during admission and post-discharge follow-up until 6 weeks after randomisation. The economic evaluation will integrate the primary clinical outcome and costs in a cost-effectiveness analysis.
This trial is designed to demonstrate a 66% reduction of neonatal sepsis from 7.5% in women in the expectant group to 2.5% in women in the induction group. This requires a total sample size of 520 patients with 80 % power and a significance level of p = 0.05.
Data analysis will be performed on an intention to treat basis and study baseline characteristics will be compared. The relative risks and 95% confidence intervals will be calculated for the relevant outcome measures. The analysis will be stratified for centre and parity. Moreover, we will evaluate whether the relative benefits of induction of labour will be stronger in multiparous women and in women with a ripe cervix at baseline.
In case of continuous data in the secondary outcome group, t-tests will be used whereas chi-square tests will be used if data are categorical. In case of equivalence between outcomes, the analysis will be repeated on a par protocol basis. Quality of life as well as pain scores will be analysed using repeated measures analysis of variance. 
Serious Adverse Events will be reported to an independent data safety monitoring committee. A formal interim analysis is not planned.
This study has been approved by the ethic committee of the University Hospital Maastricht (Ref. no. MEC 05–240). A total of about 40 academic and non-academic hospitals have signed an intention form to participate in the trial, most of them have started the medical ethics committee approval procedure. Before a collaboration clinic may start with randomisation of patients the local ethical committee must have given their approval. The trial is registered in the controlled clinical trial register under number: ISRCTN29313500 
Confidentially and data security
Initials of participants as well as a local patient number are recorded in the electronic database. Linking names with patient's numbers can only be done in the local clinics. Data will be collected using Oracle Clinical Remote Data Capture (RDC), which is a new generation of application system that enables collection and cleanup of clinical trial data using the Internet. For detailed information on Oracle RDC, please visit the page of Oracle RDC products. Each participating clinic receives a login name and password to get access to the web-secured database. The access is restricted to the database of the clinic to which the password and login name belongs. Full access to the entire database is restricted to some members of the research staff.