Induction of labour at 41 weeks or expectant management until 42 weeks: A systematic review and an individual participant data meta-analysis of randomised trials

Mårten Alkmark*, Judit K.J. Keulen, Joep C. Kortekaas, Christina Bergh, Jeroen Van Dillen, Ruben G. Duijnhoven, Henrik Hagberg, Ben Willem Mol, Mattias Molin, J. A.M. Van Der Post, Sissel Saltvedt, Anna Karin Wikstrom, Ulla Britt Wennerholm, Esteriek De Miranda

*Corresponding author for this work

Research output: Contribution to journalReview articlepeer-review

22 Citations (Scopus)
10 Downloads (Pure)

Abstract

Background The risk of perinatal death and severe neonatal morbidity increases gradually after 41 weeks of pregnancy. Several randomised controlled trials (RCTs) have assessed if induction of labour (IOL) in uncomplicated pregnancies at 41 weeks will improve perinatal outcomes. We performed an individual participant data meta-analysis (IPD-MA) on this subject. Methods and findings We searched PubMed, Excerpta Medica dataBASE (Embase), The Cochrane Library, Cumulative Index of Nursing and Allied Health Literature (CINAHL), and PsycINFO on February 21, 2020 for RCTs comparing IOL at 41 weeks with expectant management until 42 weeks in women with uncomplicated pregnancies. Individual participant data (IPD) were sought from eligible RCTs. Primary outcome was a composite of severe adverse perinatal outcomes: mortality and severe neonatal morbidity. Additional outcomes included neonatal admission, mode of delivery, perineal lacerations, and postpartum haemorrhage. Prespecified subgroup analyses were conducted for parity (nulliparous/multiparous), maternal age (<35/≥35 years), and body mass index (BMI) (<30/≥30). Aggregate data meta-analysis (MA) was performed to include data from RCTs for which IPD was not available. From 89 full-text articles, we identified three eligible RCTs (n = 5,161), and two contributed with IPD (n = 4,561). Baseline characteristics were similar between the groups regarding age, parity, BMI, and higher level of education. IOL resulted overall in a decrease of severe adverse perinatal outcome (0.4% [10/2,281] versus 1.0% [23/2,280]; relative risk [RR] 0.43 [95% confidence interval [CI] 0.21 to 0.91], p-value 0.027, risk difference [RD] -57/10,000 [95% CI -106/10,000 to -8/10,000], I2 0%). The number needed to treat (NNT) was 175 (95% CI 94 to 1,267). Perinatal deaths occurred in one (<0.1%) versus eight (0.4%) pregnancies (Peto odds ratio [OR] 0.21 [95% CI 0.06 to 0.78], p-value 0.019, RD -31/10,000, [95% CI -56/10,000 to -5/10,000], I2 0%, NNT 326, [95% CI 177 to 2,014]) and admission to a neonatal care unit ?4 days occurred in 1.1% (24/2,280) versus 1.9% (46/2,273), (RR 0.52 [95% CI 0.32 to 0.85], p-value 0.009, RD -97/10,000 [95% CI -169/10,000 to -26/10,000], I2 0%, NNT 103 [95% CI 59 to 385]). There was no difference in the rate of cesarean delivery (10.5% versus 10.7%; RR 0.98, [95% CI 0.83 to 1.16], p-value 0.81) nor in other important perinatal, delivery, and maternal outcomes. MA on aggregate data showed similar results. Prespecified subgroup analyses for the primary outcome showed a significant difference in the treatment effect (p = 0.01 for interaction) for parity, but not for maternal age or BMI. The risk of severe adverse perinatal outcome was decreased for nulliparous women in the IOL group (0.3% [4/1,219] versus 1.6% [20/1,264]; RR 0.20 [95% CI 0.07 to 0.60], p-value 0.004, RD -127/10,000, [95% CI -204/10,000 to -50/10,000], I2 0%, NNT 79 [95% CI 49 to 201]) but not for multiparous women (0.6% [6/1,219] versus 0.3% [3/1,264]; RR 1.59 [95% CI 0.15 to 17.30], p-value 0.35, RD 27/10,000, [95% CI -29/10,000 to 84/10,000], I2 55%). A limitation of this IPD-MA was the risk of overestimation of the effect on perinatal mortality due to early stopping of the largest included trial for safety reasons after the advice of the Data and Safety Monitoring Board. Furthermore, only two RCTs were eligible for the IPDMA; thus, the possibility to assess severe adverse neonatal outcomes with few events was limited. Conclusions In this study, we found that, overall, IOL at 41 weeks improved perinatal outcome compared with expectant management until 42 weeks without increasing the cesarean delivery rate. This benefit is shown only in nulliparous women, whereas for multiparous women, the incidence of mortality and morbidity was too low to demonstrate any effect. The magnitude of risk reduction of perinatal mortality remains uncertain. Women with pregnancies approaching 41 weeks should be informed on the risk differences according to parity so that they are able to make an informed choice for IOL at 41 weeks or expectant management until 42 weeks. Study Registration: PROSPERO CRD42020163174.

Original languageEnglish
Article numbere1003436
Number of pages25
JournalPLoS Medicine
Volume17
Issue number12
DOIs
Publication statusPublished - 8 Dec 2020

Bibliographical note

Funding: UBW was supported by the Swedish state under the agreement between the Swedish government and the county councils, the ALFagreement (ALFGBG-718721)(https://www.vr.se/english/mandates/clinical-research/clinicalresearch-within-the-alf-agreement.html). HH was supported by the Foundation of the Health and Medical care committee of the Region of Vastra Gotaland, Sweden (VGFOUREG854081) and UBW was supported by Hjalmar Svensson Foundation
(https://www.stiftelsemedel.se/stiftelsenhandlanden-hjalmar-svenssons-forskningsfond/). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Data cannot be shared publicly directly because use of the data from the individual trials was only approved by the ethical committee for use in this IPDA-MA. Approval for use for other purposes would be needed from the individual trials. Contact information for each study is given in supplement S6 Table.

Fingerprint

Dive into the research topics of 'Induction of labour at 41 weeks or expectant management until 42 weeks: A systematic review and an individual participant data meta-analysis of randomised trials'. Together they form a unique fingerprint.

Cite this