(C) PLOS One
This story was originally published by PLOS One and is unaltered.
. . . . . . . . . .



Determining optimal timing of birth for women with chronic or gestational hypertension at term: The WILL (When to Induce Labour to Limit risk in pregnancy hypertension) randomised trial [1]

['Laura A. Magee', 'Department Of Women', 'Children S Health', 'School Of Life Course', 'Population Sciences', 'Faculty Of Life Sciences', 'Medicine', 'King S College London', 'London', 'United Kingdom']

Date: 2024-12

In this study, we observed that most women with chronic or gestational hypertension required labour induction, and planned birth at 38 +0–3 weeks (versus usual care) resulted in birth an average of 6 days earlier, and no differences in poor maternal outcome or neonatal morbidity. Our findings provide reassurance about planned birth at 38 +0–3 weeks as a clinical option for these women.

This 50-centre, open-label, randomised trial in the United Kingdom included an economic analysis. WILL randomised women with chronic or gestational hypertension at 36 to 37 weeks and a singleton fetus, and who provided documented informed consent to “Planned early term birth at 38 +0–3 weeks” (intervention) or “usual care at term” (control). The coprimary outcomes were “poor maternal outcome” (composite of severe hypertension, maternal death, or maternal morbidity; superiority hypothesis) and “neonatal care unit admission for ≥4 hours” (noninferiority hypothesis). The key secondary was cesarean. Follow-up was to 6 weeks postpartum. The planned sample size was 540/group. Analysis was by intention-to-treat. A total of 403 participants (37.3% of target) were randomised to the intervention (n = 201) or control group (n = 202), from 3 June 2019 to 19 December 2022, when the funder stopped the trial for delayed recruitment. In the intervention (versus control) group, losses to follow-up were 18/201 (9%) versus 15/202 (7%). In each group, maternal age was about 30 years, about one-fifth of women were from ethnic minorities, over half had obesity, approximately half had chronic hypertension, and most were on antihypertensives with normal blood pressure. In the intervention (versus control) group, birth was a median of 0.9 weeks earlier (38.4 [38.3 to 38.6] versus 39.3 [38.7 to 39.9] weeks). There was no evidence of a difference in “poor maternal outcome” (27/201 [13%] versus 24/202 [12%], respectively; adjusted risk ratio [aRR] 1.16, 95% confidence interval [CI] 0.72 to 1.87). For “neonatal care unit admission for ≥4 hours,” the intervention was considered noninferior to the control as the adjusted risk difference (aRD) 95% CI upper bound did not cross the 8% prespecified noninferiority margin (14/201 [7%] versus 14/202 [7%], respectively; aRD 0.003, 95% CI −0.05 to +0.06), although event rates were lower-than-estimated. The intervention (versus control) was associated with no difference in cesarean (58/201 [29%] versus 72/202 [36%], respectively; aRR 0.81, 95% CI 0.61 to 1.08. There were no serious adverse events. Limitations include our smaller-than-planned sample size, and lower-than-anticipated event rates, so the findings may not be generalisable to where hypertension is not treated with antihypertensive therapy.

Chronic or gestational hypertension complicates approximately 7% of pregnancies, half of which reach 37 weeks’ gestation. Early term birth (at 37 to 38 weeks) may reduce maternal complications, cesareans, stillbirths, and costs but may increase neonatal morbidity. In the WILL Trial (When to Induce Labour to Limit risk in pregnancy hypertension), we aimed to establish optimal timing of birth for women with chronic or gestational hypertension who reach term and remain well.

Funding: The trial was funded by the National Institute of Health Research (NIHR) Health Technology Assessment Programme (project number 16/167/123 to LAM, with co-applicants JD, JH, BWM, JS, PH, JGT, PvD, and from the WILL Study Group, PB, LC, MF, TR, and JS [PPIE]). The funder was not involved in study design, data collection or analysis, the decision to publish, or preparation of the manuscript.

Copyright: © 2024 Magee et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Timing of birth recommendations vary, demonstrating clinical equipoise. United Kingdom (UK) guidance advises timing of birth “be agreed between the woman and the senior obstetrician” [ 11 ]. International guidance states timed birth may be offered from 37 +0 weeks (37 weeks and 0 days) for women with gestational hypertension and 38 +0 weeks for those with chronic hypertension (weak recommendations) [ 1 ].

Observational data suggest that early term birth (at 37 to 38 weeks) may reduce maternal complications (e.g., preeclampsia), cesareans, stillbirths [ 3 – 6 ], and costs, related primarily to a shorter duration of maternal-fetal surveillance [ 7 ]; however, early term birth may increase neonatal morbidity [ 8 ]. An individual patient data meta-analysis of randomised trials comparing early birth with expectant management in women with a hypertensive disorder of pregnancy also suggested earlier birth may benefit women without harming babies, including in subgroup analyses limited to participants with chronic and gestational hypertension [ 9 ]; however, these subgroup analyses included pregnancies randomised at preterm gestations, when the balance of harms and benefits associated with earlier birth are likely different. Also, the number of women enrolled in prior trials was small (e.g., 134 women with chronic hypertension), and the research was conducted in settings with differences in antenatal care, such as less frequent use of antihypertensive medication [ 10 ].

The primary economic analysis was a cost-consequence analysis from a NHS perspective, comparing intervention and control management strategies. All resource use was valued with unit cost data (2020 to 2021 prices) obtained from NHS Reference Costs ( Table D in S3 Appendix ) [ 22 ]. Overall mean costs and their variance were calculated for outpatient visits, hospital admissions, tests of maternal-fetal well-being, maternity care, and neonatal care for both groups. Mean differences in costs were calculated using regression analysis, with bootstrapped bias-corrected 95% CIs (1,000 samples) (Health Economics Analysis Plan, S2 Appendix ).

Sensitivity analyses limited to coprimary and key secondary outcomes were to assess the impact of missing data; further adjust for baseline characteristics; exclude women and babies if birth in the intervention arm was before 38 +0–3 weeks, and in the control group before 39 +0 weeks; assess heterogeneity of treatment effect due to the change to usual care (control arm); assess the impact on the neonatal coprimary outcome of stillbirths or neonatal deaths before neonatal unit admission. Unadjusted differences in medians (and corresponding 95% CI) were performed using bootstrapping methods (repetitions = 1,000, seed = 123,456). Complier Average Causal Effect analyses were not performed due to analytical difficulties in applying the standardisation approach.

Preplanned subgroup analyses were limited to coprimary outcomes and undertaken on variables used in the minimisation algorithm, except for recruiting centre; and ethnicity, body mass index, prior severe hypertension (index pregnancy), or any of the following at randomisation: antihypertensive therapy, gestational diabetes mellitus, or smoking.

A statistical analysis plan ( S1 Appendix ) was developed before analyses which were intention-to-treat. Coprimary outcomes were analysed using mixed effects logistic regression, adjusted for hypertension type and prior cesarean as fixed effects (when convergence was possible), and recruiting centre as a random effect. Adjusted risk ratios (aRR) and adjusted risk differences (aRD) with 95% CIs were calculated by marginal standardisation for covariate adjustment [ 21 ]. For the neonatal coprimary outcome, noninferiority was based on the upper limit of the 95% CI in relation to our noninferiority margin of 8%. Binary secondary outcomes were analysed as per primary outcomes. Continuous outcomes were analysed using mixed effects linear regression to generate adjusted mean differences and 95% CIs.

We estimated that a sample size of 1,080 women (540/group) was required to detect a relative risk of 0.68, corresponding to an absolute risk reduction of 8% in poor maternal outcome, from a control group rate of 25% [ 20 ], 90% power, and two-sided alpha = 0.05 (superiority hypothesis). For the neonatal coprimary outcome, we estimated 540/group would have 88% power to detect a noninferiority margin of 8%, assuming a control group rate of 23% (i.e., the upper bound of the 95% confidence interval (CI) is <8%), two-sided alpha = 0.05, and 90% power to detect a 10% decrease in cesarean [ 10 ], assuming a control group rate of 45%. No adjustment was made for loss to follow-up or dropouts.

Other secondary outcomes included potential cointerventions, other pregnancy outcomes, maternal satisfaction, and healthcare resource use; for definitions, see Table C in S3 Appendix [ 12 ]. We included core outcomes in hypertensive pregnancy [ 18 ], except neonatal seizures. Adverse events were captured via predefined outcome measures in this high-risk population. A serious adverse event was one that resulted in death, was life-threatening, required hospitalisation or prolongation of existing hospitalisation, resulted in persistent or significant disability or incapacity, may have jeopardised the pregnancy, or may have required intervention to prevent one of the other outcomes listed above. Responses to the postpartum questionnaires will be reported elsewhere.

The maternal coprimary outcome was a composite of poor maternal outcome until primary hospital discharge home or 28 days after birth (whichever was earlier), defined as severe hypertension, maternal death, or maternal morbidity, modelled on Delphi consensus [ 17 ] and the core outcome set in pregnancy hypertension [ 18 ] (for details, see Table B in S3 Appendix ). This outcome was adjudicated by the local site principal investigator (or delegate), masked to allocated group and uninvolved in the woman’s care, and based on review of primary case notes.

In the intervention group, birth could be initiated by labour induction or elective cesarean, by local protocol. In the control group, care was based on national guidance and local policy (as below) [ 11 ]. On 11 August 2022 (after randomisation of 348 women), the control group was changed from “expectant care until at least 40 +0 weeks” to “usual care at term,” to reflect practice change related to timed birth in other populations [ 13 , 14 ], the COVID-19 pandemic [ 15 ], and draft national labour induction guidance suggesting timed birth at 39 weeks may be appropriate for women at increased risk of term complications [ 16 ].

Women who provided documented informed consent were randomly assigned (1:1 ratio) to “planned early term birth at 38 +0–3 weeks” (intervention) or “usual care at term” (control). Randomisation was by a central computerised service at the Birmingham Clinical Trials Unit and minimised for site, hypertension type, and prior cesarean. A “random element” was included in the minimisation algorithm, so that each woman had a probability of 20%, of being randomised to the opposite intervention that they would have otherwise received.

Participants were recruited from consultant-led UK maternity units. Women were eligible if they were ≥16 years of age, had chronic or gestational hypertension, and a live singleton fetus at 36 +0 to 37 +6 weeks. Hypertension was a systolic blood pressure (BP) ≥140 mm Hg or diastolic BP ≥90 mm Hg. Chronic hypertension was diagnosed before pregnancy or before 20 weeks, and gestational hypertension from 20 weeks [ 1 ]. Women were excluded if they had a contraindication to either trial group (e.g., preeclampsia), severe hypertension (systolic BP ≥160 mm Hg or diastolic BP ≥110 mm Hg) until resolved, a major fetal anomaly anticipated to require neonatal unit admission, or had consented to participate in another timed birth trial.

WILL was a 2-arm, parallel-group, open-label, multicentre, randomised trial in the UK (International Standard Randomised Controlled Trial Number, ISRCTN77258279; https://www.isrctn.com/ISRCTN77258279 ). The trial was approved by the National Health Service (NHS) Health Research Authority London Fulham Research Ethics Committee (reference 18/LO/2033). A 9-month internal pilot (3 June 2019 to 20 March 2020) tested trial processes in 20 centres; the Trial Steering Committee (TSC) and Data Monitoring Committee recommended the trial continue to the main phase. The protocol has been published [ 12 ] ( S1 Protocol ). This study is reported as per the CONsolidated Standards Of Reporting Trials (CONSORT) guideline and Guidelines for Reporting Trial Protocols and Completed Trials Modified Due to the COVID-19 Pandemic and Other Extenuating Circumstances (CONSERVE-CONSORT) ( S1 Consort Checklist ).

The intervention (versus control) was associated with less antihypertensive therapy use ( Table 3 ); most women received 1 agent, usually labetalol. The intervention (versus control) was associated with less monitoring of well-being, with regard to preeclampsia blood or urine testing; outpatient visits by midwives or in the office/clinic, maternity assessment unit, or emergency department; or fetal cardiotocography or ultrasound. As such, over median [IQR] follow-up (to primary discharge home) of 10 days [ 8 – 12 ] in the intervention and 16 [ 11 – 20 ] in the control groups, the intervention (versus control) was associated with lower absolute rates of resource use and costs (mean ± SD): £6,659.57 ± 1,871.63 for the intervention and £7,067.37 ± 2,350.80 for the control groups (mean difference £−407.80, 95% CI −793.47 to +39.55; p = 0.054), with significantly lower costs for outpatient visits and tests of maternal-fetal well-being ( Table 4 ; details, Table K in S3 Appendix ). There were no serious adverse events.

For the woman, there was no association between the intervention (versus control) and preeclampsia (56/201 [28%] versus 76/202 [38%], respectively; aRR 0.74, 95% CI 0.56 to 0.98; p = 0.039; Table 3 ), before and after birth; the absolute reduction translates into a number-needed-to-treat-for-benefit (NNTB) of 10. There was no association between the intervention (versus control) and PPH, sepsis, or intensive therapy unit admission. (For details, see Tables 3 and J in S3 Appendix .)

There was no association between the intervention (versus control) for cesarean (58/201 [29%] versus 72/202 [36%], respectively; aRR 0.81, 95% CI 0.61 to 1.08; p = 0.149) ( Table 2 ). However, the aRD and 95% CI (−0.07, 95% CI −0.16 to 0.02; p = 0.146) included the prespecified minimal clinically important difference of 10%. The trend towards a difference in cesarean was due to cesarean in labour (following spontaneous onset or induction, 18/201 [9%, intervention] versus 28/202 [14%, control]). The indication for cesarean in the intervention group was most often the study protocol (32/39 [82%]), and in the control group, based on maternal (30/72 [42%]) or fetal (30/72 [42%]) indications.

For high-level neonatal care for ≥4 hours, the intervention group was considered noninferior to the control, as the upper bound of the aRD 95% CI did not cross the prespecified noninferiority margin of 8% (14/201 [7%] versus 14/202 [7%], aRD 0.003, 95% CI −0.05 to 0.06; p = 0.912); however, events rates were lower than estimated ( Table 2 ). There were no stillbirths or neonatal deaths. High-level neonatal care was required most commonly for suspected/confirmed infection, respiratory disease, or “poor condition” at birth.

We found no evidence of a difference in the maternal coprimary (“poor maternal”) outcome between intervention and control groups: 27/201 (13%) versus 24/202 (12%), respectively; aRR 1.16, 95% CI 0.72 to 1.87; p = 0.538 ( Table 2 ). The 95% CI of the aRD (−0.05 to +0.09; p = 0.539) did not include the prespecified effect size of −0.08 (which corresponds to a “poor maternal” outcome event rate being 8% lower in the intervention versus control groups in absolute terms). There was evidence to suggest that receipt of transfusion (of any blood product), as a component of the composite outcome, occurred more often in the intervention (9/201 [4.5%]) versus control (2/202 [1.0%]) group, but the 95% CI reflected high levels of uncertainty due to low event rates (aRR 4.68, 95% CI 1.05 to 20.84; p = 0.043). All transfusions were postpartum, but there was no between-group difference evident in postpartum haemorrhage (PPH; see below).

Adherence to the intervention was high ( Table 2 ); nonadherence was most often due to busy hospital induction or theatre schedules. Gestational age at initiation of birth and gestational age at birth were each a median difference of 0.9 (95% CI 0.7 to 1.0; p < 0.001) weeks earlier in the intervention (versus control) group. The interval from initiation of birth to actual birth was a median difference of 0.3 weeks (95% CI 0.26 to 0.31) in the intervention group, and 0.3 weeks (95% CI 0.14 to 0.43) in the control group. A minority of women in the control group went into spontaneous labour; most were induced. (Further details are in Table G in S3 Appendix ).

Both groups were similar at trial entry ( Table 1 ). On average, women were just over 30 years of age, with slightly more than one-fifth from ethnic minority groups and over half with BMI ≥30kg/m 2 . Approximately half of women had chronic hypertension. Among 209 parous women (52%), approximately one-sixth had a prior cesarean. The gestational age at randomisation was just over 37 weeks. Most women were on antihypertensive medication at enrolment—almost always 1 agent, usually labetalol. BP was <140/90 mm Hg for most participants ( Tables 1 and F in S3 Appendix ).

Of 2,822 women screened, 1,030 were eligible, of whom 432 (42%) consented to participate ( Fig 1 ); details of nonparticipation are in Table D in S3 Appendix . A total of 403 women were randomised, 201 to the intervention and 202 to the control group. There were 2 protocol deviations: inclusion in the control group of one woman with planned timed birth, and another who was randomised in error on the training randomisation website; both were analysed in their allocated group. Follow-up was complete for the coprimary outcomes, but 18/201 (9.0%) in the intervention and 15/202 (7.4%) in the control arms were lost to follow-up after hospital discharge, by 6 weeks postpartum.

Among 50 participating sites with median 4,976 births annually (interquartile range: 3,400 to 5,800), 46 sites consented at least 1 woman from 3 June 2019 to 19 December 2022, with a pause after the internal pilot (20 March 2020 to 6 July 2020) due to the pandemic. During this, recruitment was delayed and the funder directed recruitment to stop, without knowledge of the results, as part of “post-pandemic reset.”

Discussion

For women with chronic or gestational hypertension who reach term and remain well, we found that planned early term birth at 38+0–3 weeks, versus usual care at term, resulted in birth an average of 6 days earlier, although almost 70% of women in the usual care at term group still required labour induction. Planned early term birth at 38+0–3 weeks resulted in lower-than-anticipated rates of adverse maternal and fetal/newborn coprimary outcomes, with no evidence of differences compared with usual care at term. The 95% CI around our comparative estimate for the maternal coprimary outcome excludes our target difference of a 32% relative risk reduction and an 8% absolute risk reduction. Similarly, the 95% CI around our comparative estimate for the neonatal coprimary outcome excludes our target noninferiority margin of 8% increase in risk.

Also, we found that planned early term birth (versus usual care at term) was associated with no increase (and potential reduction) in cesarean, with the 95% CI from the comparative estimate that included a 10% reduction in risk set as the minimally-clinically important difference a priori. The intervention (versus control) was associated with a reduction in preeclampsia (defined broadly) for the woman (NNTB = 10), with no evidence of increased health problems for the baby. Also, the intervention (versus control) was associated with lower healthcare utilisation (monitoring of maternal-fetal well-being, and obstetric outpatient visits) and associated costs, with the direction of costs overall favouring planned early term birth.

To the best of our knowledge, WILL is the largest randomised evaluation of timed birth for women with chronic or gestational hypertension at term. Most prior trials enrolled women with preeclampsia and have dominated meta-analyses of timed birth for women with pregnancy hypertension [9,23–25]. While 4 trials have included at least some women who would have met WILL eligibility criteria (at least 340 participants), only 1 trial excluded women with preeclampsia [26]. That trial was small (N = 102), not prospectively registered, and found no differences in a composite of maternal/neonatal mortality/morbidity or cesarean. There is one ongoing trial (250 women) in India of timed birth at 38 (versus 40) weeks for mild gestational hypertension (CTRI/2022/06/043028, recruitment anticipated to end in 2024).

Our findings are consistent with observational data suggesting 38+0 to 39+6 weeks is the optimal timing of birth for women with chronic or gestational hypertension at term [3,4]. Our observed trend towards a reduction in cesarean associated with planned early term birth is consistent with the HYPITAT trial [10] and trials of induction for other indications [27,28]. While we observed an increase in transfusion associated with planned early term birth (versus usual care at term), the 95% CI ranged from 0.1% to 8.0% increased risk, reflecting substantial uncertainty. There was no evidence of an increase in PPH, consistent with systematic reviews of labour induction for either any indication (versus expectant care) at term [29], or for pregnancy hypertension, including chronic or gestational hypertension [9].

The WILL trial demonstrated low rates of maternal and fetal/newborn morbidities at term for women with chronic or gestational hypertension. This may be due to WILL being undertaken in the current era of good BP control [11]. Most women in WILL were taking antihypertensive therapy at enrolment and had BP <140/90 mm Hg. Improved maternal outcomes are consistent with the reduction in severe hypertension and maternal end-organ complications that define preeclampsia (e.g., thrombocytopoenia), as seen with BP control in the Control of Hypertension In Pregnancy Study (CHIPS) and the Chronic Hypertension And Pregnancy (CHAP) trials [2,30]. A recent retrospective cohort study of timed birth in women with chronic hypertension controlled with antihypertensive therapy found similarly low adverse outcome rates [31].

Our healthcare utilisation and economic findings are similar to those of the HYPITAT trial, in which earlier birth was associated with a shorter duration of (and less overall) maternal-fetal surveillance, and lower associated costs [7].

A strength of the trial was its generalisability to real-world care of women with chronic or gestational hypertension, through inclusion of women with comorbidities, contemporary treatment (control) of hypertension with antihypertensive therapy, and comparison of planned early term birth with usual clinical practice. Of note, the trial was reviewed independently and found to exceed expectations for having a diverse study population [32].

Our major limitation is that we reached only 37% of our recruitment target before trial cessation by the funder. Nevertheless, to the best of our knowledge, WILL is the largest randomised evaluation of timed birth for this group of women who reach term gestational age and remain well. It is likely that no more than 122 women in the HYPITAT trial would have been eligible for WILL, given that (i) 65% of participants had gestational hypertension; (ii) they were recruited at the time that they developed that gestational hypertension; and (iii) a minority (188/756) of participants overall were recruited at 37+0–6 weeks, when they were randomised to labour induction within 24 hours or ongoing expectant care (whereas women in WILL were randomised to planned timed birth at 38+0–3 weeks versus usual care). We included women with either chronic or gestational hypertension; while recruiting a population of mixed hypertensive type is common in pregnancy hypertension trials [2] and results for the coprimary outcomes were similar by hypertension type, such subgroup analyses were predictably underpowered. While the label of our control arm was changed to “usual care at term,” this applied only to the final 14% of recruits, and throughout, the control group reflected current practice. The findings may not be generalisable to where hypertension is not treated with antihypertensive therapy, despite international recommendations. The event rates of the 2 coprimary outcomes were lower than anticipated; while the relative risk of 0.68 set a priori for the maternal coprimary outcome was excluded, and the clinically important changes specified in absolute risks were also achieved, those changes in absolute risk for the maternal (8% reduction) and neonatal (8% noninferiority margin) coprimary outcomes were unrealistically large given the lower-than-anticipated event rates. Finally, we did not collect information on the level of neonatal care required.

The WILL trial results indicate that for women with chronic or gestational hypertension whose BP is controlled, who have reached term, and remain well, most (78%) women managed expectantly will require iatrogenic birth for clinical need, prior to the onset of spontaneous labour. While the likelihood is low that planned early birth is harmful for the baby, such a management strategy may be beneficial for women. Planned early term birth is associated with a clinically important, lower risk of progression to preeclampsia, albeit potentially, associated with a smaller, increased risk of transfusion; this stands alone as an intervention that could reduce the risk of progression to preeclampsia at term in women with chronic or gestational hypertension, similar to development of de novo preeclampsia in the ARRIVE trial of timed birth at term for low-risk nulliparous women [13]. The potential reduction in cesarean may appeal to women, and the associated reduction in healthcare utilisation and some health system costs may prompt some units to recommend planned early term birth to these women. Thus, it appears on balance that planned early term birth at 38+0–3 weeks may be the preferred clinical option.

Future work should address whether planned early term birth in women with chronic or gestational hypertension reduces cesarean; an individual participant data meta-analysis is planned (CRD42024498376), as it would be difficult to justify mounting another randomised trial given our low adverse event rates affecting feasibility. Also, observational data have raised concerns that in the general population, across the whole range of gestational at birth, gestational age has a strong, dose-dependent relationship with special educational needs, including mild learning disabilities such as dyslexia and attention deficit hyperactivity disorder [33]. While data from nonrandomised comparisons of labour induction or expectant care at term have been reassuring with regard to neurodevelopmental outcomes [34,35], condition-specific data are needed. WILL participants were asked for consent for collection of routinely collected health data, including those measuring school performance.

Pending definitive data, the WILL trial findings provide reassurance about planned early term birth at 38+0–3 weeks as a clinical option for women with chronic or gestational hypertension who reach this gestational age undelivered.

[END]
---
[1] Url: https://journals.plos.org/plosmedicine/article?id=10.1371/journal.pmed.1004481

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.