Induction of labour: when it's recommended and what the evidence says

The question

When is inducing labour actually recommended — and when the choice is genuinely yours (for example, at 39 weeks with no medical problem, or when pregnancy runs past its due date), what do the benefits and risks look like compared with waiting for labour to start on its own?

Short answer

NICE recommends offering induction for specific reasons — most commonly when pregnancy goes past 41 weeks, when waters break without labour starting, or when there are concerns about mother or baby (such as pre-eclampsia or the baby not growing well) — and for uncomplicated pregnancies the routine offer comes at 41+0 weeks; requests before then are handled by discussion of the benefits and risks, not a standing offer [4]. The biggest trial, ARRIVE (6,106 first-time mothers in the US), found that elective induction at 39 weeks led to fewer caesareans than waiting (186 vs 222 per 1,000) with no increase in serious newborn problems [1]. For pregnancies reaching 41 weeks, two European trials found that inducing at 41 weeks rather than waiting until 42 slightly improved newborn outcomes without increasing caesareans [2][3].

What the strongest evidence says

Elective induction at 39 weeks: the ARRIVE trial

The ARRIVE trial (Grobman and colleagues, New England Journal of Medicine, 2018) randomised 6,106 healthy first-time mothers with straightforward singleton pregnancies in 41 US hospitals to either elective induction between 39 weeks 0 days and 39 weeks 4 days, or expectant management (waiting, with induction only if medically indicated) [1]:

• Caesarean section: 18.6% with induction vs 22.2% with waiting — 186 vs 222 per 1,000, i.e. 36 fewer caesareans per 1,000 women induced (relative risk 0.84). About one caesarean avoided for every 28 inductions [1].
• Serious newborn problems (the trial's primary outcome: perinatal death or severe neonatal complications): 4.3% vs 5.4% — 43 vs 54 per 1,000. The confidence interval just reached 1.00, so this is not a clear superiority result: fair to say "no clear difference, with a hint of benefit" [1].
• High blood pressure disorders of pregnancy: 9.1% with induction vs 14.1% with waiting (277 of 3,059 vs 427 of 3,037) — about 91 vs 141 per 1,000, i.e. roughly 50 fewer per 1,000 women induced (relative risk 0.64, 95% CI 0.56–0.74, P<0.001; figures verified against the opened primary paper) [1].
• Newborn breathing support: the need for respiratory support within 72 hours of birth was 3.0% with induction vs 4.2% with waiting (91 of 3,059 vs 127 of 3,037) — about 30 vs 42 per 1,000 (relative risk 0.71, 95% CI 0.55–0.93); babies in the induction group also had a shorter duration of respiratory support [1].
• Women's experience: those induced reported slightly less pain and a slightly greater sense of control on a standard scale — but the differences were very small and probably not meaningful in real life [1].

Important context for ARRIVE [1]:

• Only 6,106 of 22,533 eligible women agreed to be randomised — the quarter who said yes may differ from those who declined (in particular, they were open to induction) [1].
• Care was 94% physician-led in US hospitals with a high background caesarean rate (22% in the waiting group — higher than UK rates), so the caesarean reduction may not translate directly to UK midwifery-led settings [1].
• The trial was unblinded (inevitable), and the two groups gave birth about five days apart on average (39.3 vs 40.0 weeks) — so it compares "birth at ~39 weeks" with "birth at ~40 weeks", not induction magic per se [1].
• It studied only healthy first-time mothers with single babies, head-down, at term. It says nothing about induction for medical reasons, or about second or later births.

Inducing at 41 weeks vs waiting until 42: SWEPIS and INDEX

Once pregnancy passes 41 weeks, two large European trials compared inducing at 41 weeks with waiting (and inducing at 42 if labour hadn't started) [2][3]:

• SWEPIS (Wennerholm and colleagues, BMJ 2019; 2,760 women in 14 Swedish hospitals): the composite outcome of stillbirth, newborn death, and serious newborn illness was 2.4% with induction vs 2.2% with waiting — no significant difference. But there were 6 perinatal deaths in the waiting group and none in the induction group (0 vs about 4 per 1,000), and the trial was stopped early because of this [2]. Stopping early can exaggerate effects, but deaths are deaths: the signal was taken seriously enough to halt the trial.
• INDEX (Keulen and colleagues, BMJ 2019; 1,801 women in the Netherlands): adverse perinatal outcomes were 1.7% with induction vs 3.1% with waiting — 17 vs 31 per 1,000, i.e. 14 fewer per 1,000. Low Apgar scores (<7 at 5 minutes) were 1.2% vs 2.6% (12 vs 26 per 1,000). There was 1 fetal death with induction vs 2 with waiting, and no newborn deaths in either group [3].
• Caesareans: neither trial found a difference — SWEPIS reported no difference in caesarean or immediate maternal outcomes [2]; INDEX had 97 caesareans in each group (about 108 per 1,000 both sides) [3].
• Women's experience: a SWEPIS sub-study on childbirth experience found no significant differences between the induction and waiting groups [5].

What NICE recommends (NG207, November 2021)

NICE's guideline on inducing labour sets out when induction should be offered or discussed [4]:

• Prolonged pregnancy: discuss membrane sweeping (a vaginal examination to separate the membranes — can cause pain, discomfort, and vaginal bleeding) at antenatal appointments after 39+0 weeks; offer induction at 41+0 weeks, or as soon as possible after [4].
• Waters breaking at term without labour (prelabour rupture of membranes): offer a choice — wait up to about 24 hours or induce as soon as possible; after 24 hours without labour, offer induction. If the current pregnancy has had a positive group B streptococcus test, offer immediate induction or caesarean [4].
• Previous caesarean: offer induction, planned caesarean, or expectant management on an individual basis — and explain that induction raises the chance of caesarean and of uterine rupture compared with a planned repeat caesarean [4].
• Suspected large baby: offer a choice after fully discussing the risks — not a routine recommendation to induce [4].
• Baby has died in the womb: offer induction [4].
• Maternal request: discuss the benefits and risks — a woman's informed preference counts [4].
• Avoid routine induction: for breech babies (unless caesarean is declined and external cephalic version has failed, been declined, or is contraindicated); before 34 weeks (unless there are other obstetric reasons); where growth scans show the baby is compromised; or solely to avoid an unattended birth after a previous very fast labour [4].

How induction is done depends on how ready the cervix is (the Bishop score) [4]: membrane sweeping first; then vaginal prostaglandins (dinoprostone or misoprostol — both can overstimulate the uterus, so mother and baby are monitored); a balloon catheter as a mechanical alternative; and with a favourable cervix, breaking the waters (amniotomy) plus an oxytocin drip. Low-risk women with a favourable cervix may be offered outpatient induction (going home to wait for labour after the process is started) [4].

What it means for the parents

Induction is a process, not a moment. It often means admission to hospital, repeated vaginal examinations, hours (sometimes a day or more) of waiting for things to get going, and — once an oxytocin drip is running — continuous monitoring of the baby, which limits moving around [4]. Partners should expect a lot of hanging around in a medical setting; the evidence on partners' experiences is essentially non-existent, which is itself worth saying.

Pain and sense of control: the best data (ARRIVE) suggests induction doesn't meaningfully change how painful labour feels or how in control women feel — slightly less pain and slightly more control on paper, but too small to matter in real life [1]. At 41 weeks, the SWEPIS experience sub-study found no difference at all [5].

Planning your life: the one genuine practical upside of elective induction is predictability — you know roughly when it's happening, which can help with childcare, work cover, and getting a birth partner there. That's a real consideration, not a medical one.

The 41-week decision is the one most UK parents will actually face. The absolute risks of waiting are small — SWEPIS's stopped-early result was 0 vs ~4 perinatal deaths per 1,000 — but they are the reason NICE draws the line at offering induction at 41+0 weeks [2][4]. There's no risk-free option: induction brings its own interventions, waiting carries a small but real increase in rare bad outcomes.

Mental health, sleep, work, finances: no trial measured these as outcomes of induction vs waiting. Any differences would run through the mode of birth and the length of the hospital stay rather than the induction itself.

What remains uncertain

• Long-term outcomes for children after induction: unstudied.
• Whether ARRIVE's caesarean reduction applies in the UK: the trial's US hospitals had a 22% background caesarean rate; UK rates and models of care differ [1].
• Optimal timing for subgroups (e.g. by age, BMI, or cervical readiness): NICE lists this as a research recommendation [4].
• Women's and partners' lived experience: thin evidence; one SWEPIS sub-study found no meaningful difference (one CEQ2 subscale slightly favoured induction with a small effect size) [5].

Benefits and risks in absolute terms

Practical considerations

• A membrane sweep is the gentlest first step and can be done at a routine antenatal appointment after 39 weeks — it doesn't always work, and it can be uncomfortable [4].
• If you're induced, pack for a long stay: induction is often a slow process, and you may be in hospital overnight or longer before labour is established.
• Discuss pain relief before it starts, especially if you'll have an oxytocin drip — NICE specifically advises covering this (NG207 Box 1) [4].
• You can say no. If you're offered induction at 41 weeks, the alternative is monitoring while you wait — NICE expects your informed decision to be respected, and you can change your mind either way [4].
• Ask about outpatient induction if you're low-risk with a favourable cervix — some units let you go home after the process is started [4].

When to talk to your doctor, midwife, or pediatrician

• Reduced fetal movements at any point — contact your maternity unit immediately, don't wait for an appointment.
• Waters break: note the time and colour; contact your midwife or maternity unit the same day to discuss timing [4].
• Approaching 41 weeks: expect a conversation about induction vs monitoring — ask what your unit's monitoring offer looks like if you choose to wait [4].
• Signs of pre-eclampsia: severe headache, visual disturbances, sudden swelling of face/hands, or pain just below the ribs — seek urgent assessment.
• Considering requesting induction without a medical reason: ask for a full discussion of the benefits and risks first [4].

References

1. Grobman WA et al. Labor induction versus expectant management in low-risk nulliparous women (ARRIVE trial). New England Journal of Medicine. 2018;379:513–523. — [A] (primary full text opened and figures verified 2026-09-09) https://www.nejm.org/doi/full/10.1056/NEJMoa1800566
2. Wennerholm U-B et al. Induction of labour at 41 weeks versus expectant management until 42 weeks (SWEPIS). BMJ. 2019;367:l6131. — [A] https://www.bmj.com/content/367/bmj.l6131
3. Keulen JKJ et al. Induction of labour at 41 weeks versus expectant management until 42 weeks (INDEX). BMJ. 2019;364:l344. — [A] https://www.bmj.com/content/364/bmj.l344
4. National Institute for Health and Care Excellence. Inducing labour. NICE guideline NG207. November 2021. — [guideline] https://www.nice.org.uk/guidance/ng207/chapter/recommendations
5. Nilvér H et al. Women's childbirth experiences in the Swedish Post-term Induction Study (SWEPIS): a multicentre, randomised, controlled trial. BMJ Open. 2021;11(4):e042340. — [C] (CEQ2 total score 3.3 vs 3.2, p=0.07; VAS 8.0 vs 8.1, p=0.22; one subscale — participation — slightly higher in the induction group, 3.6 vs 3.4, p=0.02, small effect size)

Changelog

• 2026-09-09: Primary-source compliance pass. Opened the ARRIVE primary full text (NEJM, Grobman et al. 2018) — verified: caesarean 18.6% vs 22.2% (569/3,059 vs 674/3,037; RR 0.84); hypertensive disorders 9.1% vs 14.1% (277 vs 427; RR 0.64, 0.56–0.74, P<0.001); primary composite 4.3% vs 5.4% (132 vs 164; RR 0.80, 0.64–1.00); need for neonatal respiratory support 3.0% vs 4.2% (91 vs 127; RR 0.71, 0.55–0.93) with shorter duration. Removed the earlier "corroborated across multiple reports" rationale and the "exact figures unrecoverable" note; respiratory-support claim now quantitative. ARRIVE rating A now rests on the opened primary.
• 2026-09-09: Author review (v1 model, delegated to agent 2026-09-09): verified against primary/official sources opened 2026-09-09 (NICE NG207 recommendations page; SWEPIS PMC full text; SWEPIS BMJ Open sub-study abstract). Fixes: completed reference [5] (Nilvér H et al., BMJ Open 2021) with the participation-subscale nuance; softened "NICE advises against routine induction before 41 weeks" to the guideline's actual wording (routine offer at 41+0; requests handled by discussion); verified the NG207 indication list (41+0 offer, PROM ≤24h choice, GBS immediate induction, previous-CS individualized, maternal request by discussion, breech/growth-restriction restrictions, woman's right to decline/stop). Status → approved.
• 2026-09-08: Topic created (draft). Evidence gathered 2026-09-08. (Old review model superseded 2026-09-09.)