The aim of this study is to investigate whether prophylactic administration of 17OHPC will lower the incidence of neonatal morbidity by reducing the number of preterm births.
We will use a randomised, placebo-controlled double blinded trial (the AMPHIA-trial: 17-Alpha hydroxyprogesterone in Multiple pregnancies to Prevent Handicapped InfAnts) to assess the effects of 17OHPC injections on neonatal outcome. The study is set in the Dutch Obstetric Research Consortium, a collaboration of obstetric practices in the Netherlands. Approximately 50 clinics, including academic hospitals, non-academic teaching hospitals and non-teaching hospitals will participate in this trial.
All women with a multiple pregnancy, both mono- and multichorionic, and a gestational age ≤ 19 weeks are eligible for participation in the AMPHIA-trial. Before inclusion, chorionicity must be accurately determined by means of ultrasound.
Women with a previous spontaneous preterm birth < 34 weeks, serious congenital defects or death of one or more foetuses, early signs of twin-to-twin transfusion syndrome or primary cerclage are excluded from the study.
Procedures, recruitment, randomisation and collection of baseline data
All women with a multiple pregnancy who present at one of the participating clinics will be referred to an obstetrician or a specifically appointed research nurse/midwife for counselling. Eligible women will receive an information sheet and, where possible, are given two weeks time to reflect on participation. Once women have given consent for the trial, they are randomised through a website, according to a computer-generated randomisation sequence. Stratification will be applied for chorionicity (mono- versus multichorionic), number of multiples (twin versus higher order multiple) and previous vaginal birth (nullipara versus multipara). Randomisation will be 1:1 for progesterone and placebo.
When a patient is randomised, the company that supplies the study medication will receive an automatic notification by e-mail. A medication package with the corresponding randomisation number, containing 20 ampoules of 1 cc of either 250 milligrams 17OHPC or placebo, is then sent to the pharmacist in the clinic where the patient has been randomised. After a maximum of 7 days, the medication package will be available to the obstetrician or research nurse/midwife. Content of the package is blinded to the practitioner, the patient and the local pharmacist. In case of severe side-effects, allocation can be disclosed by coded envelopes that are available to the principal investigator.
Baseline demographic, past obstetric and medical histories will be recorded for all women. Cervical length will be measured by transvaginal ultrasound at the time of randomisation or at the next visit.
The patient will receive the first intramuscular injection of 250 milligrams of 17OHPC or placebo at a gestational age between 16+0 and 20+0 weeks. For the remaining duration of the pregnancy, the medication package will be stored by the patient herself. Weekly injections will be administered by a nurse at the outpatient clinic, a general practitioner or the patient herself. Dates of injection are noted in a schedule that is kept both by the patient and in her medical record. Non-compliance is defined as an interval of more than ten days between two injections. Injections are continued weekly until a gestational age of 36 weeks or, in case of delivery before 36 weeks, until delivery. Management of the pregnancy is done according to the local protocol.
Follow up of women and infants
The patient will be invited for a final visit 6 weeks after delivery. At this visit, all medication that has not been used must be returned and the patient is asked about the condition of her children.
All details of delivery, maternal assessments and admittance during pregnancy are recorded in the case record form that is accessible through the website. In case of admittance of one or more children to the neonatal intensive care unit, details of this admittance are also recorded.
Long-term follow up of children is desirable, but is depending on future funding.
The primary outcome measure is composite neonatal morbidity. This composite morbidity rate contains severe RDS, BPD, IVH grade II B or worse, NEC, proven sepsis and death before discharge from the hospital [8
Secondary outcome measures are time until delivery, preterm birth before 32 and 37 weeks, length of admission at the neonatal intensive care unit, maternal morbidity, hospitalization of the mother due to (threatened) preterm labour and costs.
Analysis is performed according to the intention-to-treat principle. The effectiveness of progesterone compared with placebo will be assessed by calculating a relative risk and 95% confidence interval. Time to delivery will be assessed with Kaplan-Meier analysis and Cox proportional hazard analysis. We will also assess whether there is an interaction between treatment effect and cervical length at 20 weeks.
The aim of the economic evaluation is to compare the optimality, in terms of costs and health effects, of weekly (prophylactic) administration of 17OHPC progesterone injections (16–20 to 36 weeks) with no intervention. As the clinical study is based on a superiority design (it is hypothesized that progesterone decreases preterm birth considerably), the primary economic evaluation is a cost-effectiveness analysis (CEA): the optimal strategy is the one with the largest health gain at the smallest extra costs.
Costs and outcomes are analysed according to intention-to-treat and described with appropriate statistical measures. The sensitivity of costs and health outcomes for various parameters is tested in sensitivity analysis. Scenario analyses for relevant subgroups (e.g. gestational age at birth, parity, monochorionic/dichorionic multiples) are added.
The process of care is distinguished into three cost stages (antenatal stage, delivery/childbirth, postnatal stage) and three cost categories (direct medical costs [all costs in the health care sector], direct non-medical costs [costs outside the health care sector that are affected by health status or health care] and indirect costs [costs of sick leave]). For each stage and each cost category, costs are measured as the volumes of resources used multiplied with appropriate valuations (cost-per-unit estimates, fees, national reference prices).
Cost volumes in the antenatal stage consist of direct medical costs (e.g. hospital care, outpatient visits, administration of 17OHPC progesterone injections). Direct non-medical and indirect costs in that stage may occur if role patterns or household routines shift.
Costs during childbirth are dominated by the course of childbirth and type of delivery. Cost volumes in the postnatal stage consist of hospital-based maternal care (hospitalization etc.), neonatal care (admission to NICU/medium care or maternity ward, related diagnostic care and treatment, outpatient visits) and primary care. If neonatal health is suboptimal, further direct medical, direct non-medical and indirect costs may occur. Hence, for these infants, resource use of infants and/or parents is measured during 12 months after childbirth.
Volumes of health care resource use are measured prospectively alongside the clinical study in all participating centres as part of the CRF.
Valuations of direct medical resources are estimated as cost per unit estimates comprising "true economic" costs, i.e. including shares of fixed costs and hospital overheads. Costs per units are estimated for at least teaching and one non-teaching hospital. An analysis based on reimbursement fees is added. Direct medical volumes outside the hospital and direct non-medical volumes are valued using national reference prices [17
]. Indirect costs are quantified but remain unvalued. Study-specific costs are excluded from analysis.
The sample size is calculated based on the primary outcome 'bad neonatal outcome'. In the control group, 'bad neonatal outcome' is expected in 7.2% of the children (1.8%*77% + 5.4%*35% + 7.2%*12% + 35.6%*8% + 50%*.5% = 7.2%). In this calculation, the first rate represents the probability that a patient delivers at that gestational age, whereas the second rate represents the probability of 'bad neonatal outcome' at that particular gestational age. In case of treatment, 'bad neonatal outcome' is then expected in 3.9% of the children (0.9%*77% + 2.7%*35% + 3.6%*12% + 17.8%*8% + 75%*.5% = 3.9%).
Because the outcomes in children from multiple pregnancies are to a certain extent non-independent, we adjusted our sample size assuming a correlation of 0.6 for composite neonatal morbidity between two children born from the same pregnancy [18
]. Using a two-sided test with an alpha of 0.05 and a power of 0.80 we need 330 women in the control group and 330 in the intervention group.
Data will initially be analysed according to the intention to treat method. The main outcome variable, 'bad neonatal outcome', will be assessed by calculating rates in the two groups, relative risks and 95% confidence intervals as well as numbers needed to treat. To evaluate the potential of each of the strategies, we will also perform a par protocol analysis, taking into account only those cases that were treated according to protocol.
Time to delivery will be evaluated by Kaplan-Meier estimates, with account for different durations of gestation at entry, and will be tested with the log rank test. The other secondary outcome measures will be approached similarly to the primary outcome measure. The analysis will be stratified for chorionicity, number of multiples and parity. Furthermore, we will take account of non-independence between children born from the same pregnancy using cluster analysis.
An interim analysis will be performed after the inclusion of 300 women. This analysis will be done by an independent data and safety monitoring committee that will not be aware of the allocation of treatment or placebo when they judge data on effectiveness. In case of severe side-effects, the safety monitoring committee can order to disclose the label of patients with such side effects.
This study has been approved by the ethics committee of the Academic Medical Centre Amsterdam (Ref. No. MEC 05/102) and by the boards of management of all participating hospitals.