Cervical Pessary Treatment for Prevention of s PTB in Twin Pregnancies on Children's Long-Term Outcome

Impetus is a prospective, multicentre, multinational, open-label, randomised, controlled clinical trail in parallel groups.

For sample size calculation, a the stratified design is accounted for and three equally large gestation groups assumed. For the pessary group a combined event rate of at least 8% for the primary outcome is assumed and for the comparison of the pessary group with the control group an odds ratio of 2.29 is assumed. This odds ratio correspond to the lower bound of a one-sided confidence interval for the event rate given in van´t Hooft (ProTwin Trial). To reach a power of at least 80%, at least 500 patients will be evaluated, 250 in the pessary group and 250 in the control group. To account for a drop out rate of 25%, overall n=672 pregnant women will be recruited.

The primary statistical aim is to compare the primary combined outcome "long-term survival without neuro-developmental disability at 3 years follow up" with a two-sided Cochran-Mantel-Haenszel-Test and a significance level of alpha=0.05. The primary outcome refers to a combined event in any of the twin and will be analysed for all pregnancies with available primary endpoint. The stratified study design is accounted by this stratified test according to the gestation groups.

The main statistical evaluation will be performed at two time points. (1) The complete data set for the secondary endpoints will be available after the last women enrolled in this study has delivered her twins, so the analysis of these outcome parameter will be done right after this event. (2) The primary outcome will be evaluated 3 years after the last woman enrolled in this study has delivered her twins. A descriptive analysis by preterm birth will be carried out calculating means and medians for quantitative variables and proportions with 95% confidence intervals for categorical variables. In general, statistical comparisons with the pessary arm and the control arms or other group comparisons for primary and secondary outcomes will be performed with stratified tests as well as comparisons in the gestation subgroups. Events will be analysed for each twin and for single children assuming appropriate random effect regression models. Further subgroup analyses regarding the cervical length will be performed (e.g. Cervical Length (CL) 15 to 25mm and below 15mm). All tests, see also examples in the synopsis, will be two-sided using a significance level of alpha=0.05.

For the primary endpoint a drop out rate of up to 25% is expected due to the long follow-up time (3 years) of the study; but no lost data for the secondary endpoints are expected because for these parameters the study has a short follow-up time till time to birth only.

An interim analysis shall be conducted on key safety parameters after birth of 300 twins: the following safety endpoints will be assessed by a one-sided test with alpha=1%

• on level of the neonates: rate of preterm birth, time to birth, birth weight, death, neonatal morbidity, harm of intervention
• and on the maternal level: rate of hospitalisation for threatened preterm labour < 32 weeks, rate of premature rupture of membranes (PRoM) <32 weeks, rate of infection / inflammation, rate of physical or psychological intolerance to pessary, rate of SAR/SAE, death.

The trial will be terminated as negative if a disadvantage for the pessary-treatment can be found in one of these tests. To guarantee a high safety level the significance level is chosen more conservatively than in a Bonferroni correction. All analysis will be carried out with SPSS® version 19.0 or later (IBM Company SPSS Inc. Headquarters, Chicago, Illinois. USA) and R version 3.2.3 or later (R Foundation for Statistical Computing, Vienna, Austria).

Methods against bias:

All women will be randomly allocated to the cervical pessary group or the control-group in a 1:1 ratio. The randomisation sequence is computer generated with variable block sizes using a web-based e-CRF (Online-Software Castor is a fully GCP compliant system) stratified for gestation groups and centers. The allocation code will be disclosed after the patient´s initials will be confirmed. The investigators or the trial coordinator will not have access to the randomization sequence.

Exclusion criteria were chosen to ensure an equal risk distribution for pregnancy complications and fetal morbidity / mortality rate for both study groups.

The study is open label since masking the intervention is not possible. All investigators should be trained in pessary application and cerclage placement. Quality protocols should be submitted according to the Clara-Angela Foundation requirements for pessary placement. Outcome assessors will be blinded to the interventions. Group allocations will base on an intention to treat basis with a per protocol allocation as sensitivity analysis.

The study will be registered and the study protocol is available. Outcome measures meet the core-outcome set for the evaluation of interventions to prevent PTB published by the crown-initiative in 2016.