Intended for healthcare professionals

Rapid response to:


When to induce late term pregnancies

BMJ 2019; 367 doi: (Published 20 November 2019) Cite this as: BMJ 2019;367:l6486

Linked Research

Induction of labour at 41 weeks versus expectant management and induction of labour at 42 weeks

Rapid Response:

Re: When to induce late term pregnancies - Misleading analysis and results in editorial

Dear Editor

I read the editorial (1) commenting on the SWEPIS trial with great interest. The trial was stopped early and unfortunately seems to suffer from serious errors owing to conduct, analysis and reporting (2). In the editorial a two-paragraph systematic review with meta-analysis is carried out and it is noted that “The number of events is small and a lot of imprecision exists in the estimate of treatment effect”. Nevertheless it is concluded that “this evidence is likely to be compelling enough [...]” and in the very last sentence a costly change of clinical practice is proposed: “maternity units must now find ways to overcome the challenge of implementation”.

The problem is the following. In the editorial the mortality data from three trials are combined (10 deaths in total) and a Peto odds ratio from a meta-analysis is presented (0.20, 95% confidence interval 0.06 to 0.70) (1). However, results “from apparently conclusive meta-analyses may be false”, particularly when a limited number of events from a few small trials are combined (3). It has been advised that if “a meta-analysis is conducted before a sufficient IS [information size] is reached, it should be evaluated in a manner that accounts for the increased risk that the result might represent a chance finding (i.e. applying trial sequential monitoring boundaries)” (3). This seems not to have been done. The editorial thus seems to add to the errors already present in the analysis of the SWEPIS trial.

Furthermore, in the editorial it is not clearly stated that for a long time it has been known that meta-analyses stopped early for benefit has a too high risk of random error to be ignored (4). This is particularly so for cumulative meta-analyses of sparse data as in the present case (4). “Given that point estimates from meta-analyses will be used as a basis for recommendations in clinical practice guidelines and decision making in individual patient care, grossly erroneous treatment effects can be highly problematic.” (3) “A meta-analysis of sparse evidence will typically be among the first in a series of meta-analytic updates. Thus, enforcing some degree of conservatism at this stage seems appropriate.” (4). This advice also seems to have been ignored.

The trial on which the editorial is based is flawed and a substantial proportion of the weight in the meta-analysis is from that trial. It has been noted that “if a substantial proportion of trials is biased, then any TSA [trial sequential analysis] would also provide misleading results” (5). The authors of the editorial should either demonstrate that their pooling of trial results does not provide misleading results or they should seriously revise the editorial. The meta-analysis should provide a reliable confidence interval and the implications and advice in the editorial should be in accordance with the corrected analysis.

Ole Olsen
Independent researcher
Academic employee at The Research Unit for General Practice in Copenhagen
Department of Public Health
Faculty of Health Sciences
University of Copenhagen
Øster Farimagsgade 5
P.O. Box 2099
DK-1014 Copenhagen


1. Kenyon S, Middleton L, Skrybant M, Johnston T. When to induce late term pregnancies. BMJ 2019;367:l6486.

2. Olsen O, Rapid Response: Statistical analysis in SWEPIS trial is flawed (retrieved 8th of January 2021 from

3. Thorlund K, Devereaux PJ, Wetterslev J, Guyatt G, Ioannidis JP, Thabane L, Gluud LL, Als-Nielsen B, Gluud C. Can trial sequential monitoring boundaries reduce spurious inferences from meta-analyses? Int J Epidemiol. 2009;38:276-86. doi: 10.1093/ije/dyn179.

4. Brok J, Thorlund K, Wetterslev J, Gluud C. Apparently conclusive meta-analyses may be inconclusive--Trial sequential analysis adjustment of random error risk due to repetitive testing of accumulating data in apparently conclusive neonatal meta-analyses. Int J Epidemiol. 2009;38:287-98. doi: 10.1093/ije/dyn188.

5. Brok J, Thorlund K, Gluud C, Wetterslev J. Trial sequential analysis reveals insufficient information size and potentially false positive results in many meta-analyses. J Clin Epidemiol. 2008;61:763-9. doi: 10.1016/j.jclinepi.2007.10.007.

Competing interests: No competing interests

12 January 2021
Ole Olsen
Independent researcher and Academic employee
The Research Unit for General Practice in Copenhagen
Department of Public Health Faculty of Health Sciences University of Copenhagen Øster Farimagsgade 5 P.O. Box 2099 DK-1014 Copenhagen