Re: Induction of labour at 41 weeks versus expectant management and induction of labour at 42 weeks (SWEdish Post-term Induction Study, SWEPIS): multicentre, open label, randomised, superiority trial
We thank Dr Simon Timka and Dr Johanna Larsson for comments on our paper, the SWEPIS trial, recently published in the BMJ1. We agree with the authors that the relevant question is if the risk of expectant management and induction of labour (IOL) at 42 gestational weeks is out weighted by a potential increase in complications by IOL at 41 gestational weeks.
It is doubtful to use a statistical model with time-at-risk as proposed in the letter, when the cumulative number of outcomes per group is the outcome of clinical interest. Although some studies indicate that the prospective risk of stillbirth in term pregnancies (from 37 to 42 gestational weeks) increases with gestational age2, another Swedish registry study could not confirm a higher risk at 42 gestational weeks compared to the risk at 41 gestational weeks3. Further, the exact risks for low risk singleton pregnancies are not known, i.e. those included in the SWEPIS trial. The hypothesis in the SWEPIS trial was that, by including only women with low risk pregnancies, we should get a somewhat higher risk in the composite adverse perinatal outcome in the expectant management group but that the risk of perinatal death would not be substantially higher. Something else would, for sure, not be accepted by the patients either. Therefore, we think the design in the present trial, as stated in the publication, not using a time-at-risk design, is correct in view of the reasons given above.
We further want to emphasize that our trial was not stopped of solely statistical reasons but more of ethical concerns. The primary mandate for a Data Safety Monitoring Board (DSMB) is to protect the patients in a trial. A secondary aim is to discuss the quality of the trial to improve scientific issues.
It is obvious that ethical considerations are most important, which were established already in the Helsinki declaration4 and Belmont report5. A serious unbalance in a study should initiate termination of the trial.
The DSMB in the SWEPIS trial has performed similar statistical analyses, also based on Fisher’s exact test, as the analyses performed by Drs Timka/Larsson. However, such statistics must be considered very carefully and reviewed in a context where ethical deliberations must play an important inalienable part. While including such ethical deliberations the DSMB did not find that it was acceptable to continue this study when six deaths were seen in the 42 weeks group and zero in the 41 weeks group. In fact, we are not aware of any study with such a drastic difference in mortality during follow up.
The DSMB wants to emphasize that the recommendation is underpinned by the fact that special considerations are necessary given the very nature of the present study. Pregnant women are randomized, and the sensitive traits of the observed mortality must be judged carefully before a final recommendation by the DSMB. The decision to issue this recommendation from the DSMB was unanimous. Moreover, the members of the DSMB are experienced scientists with different scientific backgrounds.
1. Wennerholm UB, Saltvedt S, Wessberg A, Alkmark M, Bergh C, Brismar Wendel S, et al. Swepis. BMJ 2019;367:6131.
2. Muglu J, Rather H, Arroyo-Manzano D, Bhattacharya S, Balchin I, Khalil A, Thilaganathan B, Khan KS, Zamora J, Thangaratinam S. Risks of stillbirth and neonatal death with advancing gestation at term: A systematic review and meta-analysis of cohort studies of 15 million pregnancies. PLoS Med. 2019 Jul 2;16(7):e1002838.
Competing interests: No competing interests