Regimens of misoprostol with mifepristone for early medical abortion: a randomised trial

Authors


Ms Caitlin Shannon, Gynuity Health Projects, 15 East 26th Street, Suite 1617, New York, NY 10010, USA. Email caittespencer@yahoo.com

Abstract

Objective  To compare the efficacy, adverse effects and acceptability of the three most common misoprostol regimens used with mifepristone for medical abortion.

Design  Randomised nonblinded trial.

Setting  Three clinics associated with major research universities in Canada; two in major urban areas and one in a periurban area.

Population  Women of reproductive age.

Methods  Consenting women presenting for abortion services with gestations less than 56 days and who met inclusion criteria were given 200 mg mifepristone orally and then randomised into three misoprostol study groups: (group I) 400 micrograms of oral misoprostol, (group II) 600 micrograms of oral misoprostol, and (group III) 800 micrograms of vaginal misoprostol. Misoprostol was self-administered at home 24–48 hours following mifepristone, and participants were instructed to take a second similar misoprostol dose at 24 hours after the initial dose if bleeding was less than a normal menstrual period.

Main outcome measures  Successful abortion without surgery was 94.1%, with no significant differences across the three study groups (94.7% in group I, 93.4% in group II, and 94.3% in group III; P= 0.975).

Results  Efficacy and adverse effects did not differ significantly across the three study groups. Pain increased significantly across the study and the gestational age groups and was associated with lower acceptability.

Conclusions  There appears to be a range of safe and effective options for early medical abortion with mifepristone including a choice between oral and vaginal administration of misoprostol.

Introduction

The use of mifepristone for early pregnancy termination has increased greatly since its initial approval in France in 1988.1 While in the USA and 30 other countries, the approved dose of mifepristone is 600 mg followed by 400 micrograms of oral misoprostol in 48 hours, there is also a substantial evidence that a 200-mg dose may be as effective.2,3 This lower dose has become the standard of practice in the USA and UK.4–6 In contrast, evidence on the best route and dose of misoprostol for use in combination with 200 mg of mifepristone is more equivocal; thus, practice is not standardised.

In the USA and UK, a regimen of 200 mg mifepristone followed by 800 micrograms of vaginal misoprostol in 24–48 hours has been recommended.4–6 Many studies have reported success of this regimen in 96–98% of cases.7–9 Several studies also suggest that regimens combining 200 mg mifepristone and 400 or 600 micrograms of oral misoprostol administered 24–48 hours following mifepristone are highly effective, with success rates from 92 to 96%.2,3,10–14 Provider practice, including the use of additional misoprostol doses, may account for some of the variance in efficacy reported across studies.15

Thus, without multiple, well-designed comparative trials, it is difficult to determine the relative efficacy of various regimens. While a few trials comparing 400 or 800 micrograms of oral misoprostol and 800 micrograms of vaginal misoprostol have been published, they do not provide convincing evidence that one regimen is preferable, in terms of efficacy, adverse effect profile, and patient acceptability.15–19 In the USA, several published guidelines suggest that a regimen of 200 mg mifepristone and 800 micrograms of vaginal misoprostol is superior, but a review of the literature is inconclusive, in part, because the differences in efficacy reported may not be clinically relevant even if statistically significant.

Other factors such as adverse effects, pain, and acceptability are also important guides, especially when each different regimen is highly effective. Evidence suggests that adverse effects associated with early pregnancy terminations using mifepristone and misoprostol—such as nausea, gastrointestinal complaints, and pain from cramping—may be related to the dose of misoprostol, its route of administration, and the gestational age of the pregnancy.20 Further, although adverse effects are generally mild and short-lived, they may impact patient acceptability. Therefore, while acceptability of medical abortion is known to be high generally, regardless of the regimen used, a better understanding of the impact of both misoprostol dose and route of administration on the various components of acceptability may help improve service delivery.21–23

This study reports on a randomised trial comparing misoprostol regimens after low-dose mifepristone for early pregnancy termination. The objective was to compare efficacy, adverse effects, and acceptability of these three regimens.

Methods

We conducted a multicentre randomised trial to compare three regimens of mifepristone–misoprostol medical abortion. Between February 2001 and September 2001, women at three clinics in Canada who requested a medical abortion for termination of an early pregnancy were recruited for the study. The three centres were the University of British Columbia, Vancouver; Centre Hospitalier Universitaire de Sherbrooke; and Centre Hospitalier Universitaire de Québec, Québec City. The ethics committees of all the centres and of the Population Council approved the study. All 971 women who were enrolled gave a written informed consent.

Participants included in the study were women aged 16 years or older, seeking elective abortion of pregnancies less than 57 days as measured since the onset of their last menstrual period (LMP). Participants were required to provide written informed consent prior to participation in the study, to have evidence of an intrauterine pregnancy of 56 days LMP or less on vaginal ultrasound, to be willing to undergo surgical aspiration in case of failed medical abortion, and to have access to a telephone.

Women were excluded if haemoglobin was less than 9.5 g/dl or if they had active hepatic or renal disease, type I diabetes mellitus, adrenal insufficiency, glaucoma, sickle cell anaemia, coagulopathy, uncontrolled seizure disorder, severe cardiovascular disease, allergy or intolerance to mifepristone or misoprostol, used chronic oral steroid medications or anticoagulants, or had any other medical conditions that, in the opinion of the investigator, would compromise their health during the procedure.

On day 1, participants were given one tablet of 200 mg mifepristone (Mifeprex®; Danco Laboratories, New York, NY, USA) to be taken orally and then were randomised into three study groups: group I, 400 micrograms of oral misoprostol; group II, 600 micrograms of oral misoprostol; and group III, 800 micrograms of vaginal misoprostol. Each participant was given two packages containing misoprostol (Cytotec®; Pfizer, New York, NY, USA) tablets to be administered at home. According to a participant’s study group randomisation, each package contained either 400 or 600 micrograms of misoprostol to be taken orally or 800 micrograms of misoprostol to be taken vaginally. Women were instructed to take the first package at home 24–48 hours after mifepristone administration and the second package 24 hours later, only if the bleeding was less than a usual menstrual period at that time. Participants were given nausea and pain medication, to be taken as needed. Participants were also asked to keep a written record of bleeding, pain, other symptoms, and administration of other medications.

Seven days following administration of mifepristone, vaginal ultrasound was performed to determine the status of the pregnancy. Participants who had a persistent gestational sac received 800 micrograms of misoprostol vaginally and were asked to return to the clinic in 1 week. On day 15, participants with continuing pregnancies were booked for surgical aspiration. Participants with incomplete abortions (e.g. retained products of conception but no embryonic cardiac activity) were asked to return for additional follow up on day 36. At that visit, if the abortion was still incomplete, participants were offered a surgical aspiration or were followed in the clinic until the abortion was complete.

A participant with an incomplete abortion could request a surgical aspiration at any time. Providers used their discretion in determining whether surgical intervention was necessary based on clinical findings, such as haemorrhage and incomplete abortion. Treatment of bleeding complications was also according to provider discretion and included the use of methergine, intravenous fluids, surgical intervention, transfusion, and hospitalisation as necessary.

Participants were considered lost to follow up if there was no successful contact with them by the end of the study; thus, a treatment outcome could not be determined.

Pain, bleeding, adverse effects, and acceptability

At each follow-up visit, women were asked about pain, bleeding, adverse effects, and any medications used. Pain was measured by asking women to estimate their worst pain on a validated scale from 0 (no pain) to 10 (worst pain possible).24,25 The number of days on which bleeding occurred was also recorded together with complications, such as infection and haemorrhage. Infection was defined as any treatment with antibiotics for fever and/or uterine pain or tenderness. Haemorrhage was defined as bleeding that required treatment with methergine, surgery, intravenous fluids or that resulted in a drop of haemoglobin more than 2.0 g/dl or to less than 9.0 g/dl. At the final visit (or phone call), participants were asked whether the bleeding, pain, waiting/time involved, adverse effects, and the overall procedure were acceptable and whether they would prefer to administer misoprostol orally or vaginally if given the choice.

Randomisation

Misoprostol pills were packed in sealed, opaque envelopes according to study group assignment, allocated according to a computer-generated random number of blocks of 15. Randomisation was concealed from providers and participants until after mifepristone administration, when the study packages were opened and distributed to participants for administration of misoprostol at home. This study was not blinded, and participants and providers were aware of the dose and route of administration of misoprostol.

Statistical analysis

We determined the required sample size to be 500 subjects per study group, in order to have greater than 90% power in a two-sided 5% significance level test to detect at least a 5% difference between any two study groups.* This calculation was made based on the assumption of a 92% success rate in the 400 micrograms of oral misoprostol group and a 97% success rate in the 800 micrograms of vaginal misoprostol study group.

Complete medical abortion without surgical intervention was classified as treatment success. All outcomes that resulted in a surgical intervention were classified as treatment failures. Reasons for surgery were coded as (1) continuing pregnancy (i.e. embryonic cardiac activity on ultrasound), (2) medical indications (i.e. for incomplete abortion or for bleeding and/or pain that warranted surgery), or (3) patient request.26 Other outcome measures were adverse effects, pain, and acceptability. A chi-square test was used for pair-wise comparison of outcome measures between study groups. A one-way analysis of variance (ANOVA) test was used to compare mean pain scores across gestational age and study groups. A Cochran’s test for trend was used to compare outcome measures across study groups. For chi-square tests, a Yate’s correction was used where cell sizes were less than 5. P values less than 0.051 were considered statistically significant.

Results

A total of 971 women were enrolled. Fifteen women were excluded from the analysis—14 (1.4%) who had a gestational age greater than 56 days LMP and one (0.1%) who was younger than 16 years. A final treatment outcome was recorded for each enrolled patient. The three study groups (n= 956) were evenly matched for maternal and gestational age, reproductive history, and ethnicity (Table 1).

Table 1.  Participant characteristics and treatment outcomes
 Group I (400 micrograms orally), n= 319Group II (600 micrograms orally), n= 319Group III (800 micrograms vaginally), n= 318
  • *

    Measured in days since LMP.

  • **

    Measured by chi-square test for trend (df= 2).

Participant characteristics
Mean age of woman (years ± SD)27.6 ± 6.627.4 ± 6.527.8 ± 6.6
Mean gestational age (days*± SD)43.0 ± 5.843.7 ± 6.144.0 ± 6.1
Parity > 0, % (n)41.2 (131)42.5 (135)41.8 (133)
Previous induced abortion, % (n)48.7 (155)46.2 (147)48.4 (154)
Previous spontaneous abortion, % (n)10.4 (33)9.7 (31)9.7 (31)
Treatment outcomes
Success, % (95% CI) (n)94.7 (92.4–97.1) (302)93.4 (91.1–95.8) (296)94.3 (92.0–96.7) (299)
Surgery for continuing pregnancy, % (n)0.0 (0)0.3 (1)0.0 (0)
Surgery due to medical indication, % (n)5.0 (16)6.0 (19)5.0 (16)
Surgery due to patient request, % (n)0.3 (1)0.3 (1)0.7 (2)
Lost to follow up, % (n)0.0 (0)0.6 (2)0.3 (1)
Treatment success by gestational age
<42 days, % (n)97.1 (135)97.6 (121)99.1 (112)
42–49 days, % (n)93.0 (120)92.5 (124)92.1 (128)
50–56 days, % (n)92.2 (47)86.4 (51)90.8 (59)
P value**0.1050.0040.010

Efficacy was high in all three groups, with no significant differences in treatment outcomes across the three study groups (Table 1). Overall, 897 participants (94.1%) had complete abortions after completion of the treatment regimen. There was only 1 surgery due to an continuing pregnancy (0.1%), 51 due to medical indication (haemorrhage and incomplete abortion) (5.4%), and 4 due to patient request (0.4%). There were no significant differences in treatment outcomes by maternal age, weight, parity, gravidity, or ethnicity (data not shown). Gestational age did not have an effect on treatment outcomes (Table 1). Across groups II and III but not group I, the overall efficacy significantly decreased with increasing gestational age (group I: χ2= 2.63, P= 0.105; group II: χ2= 8.25, P= 0.004; and group III: χ2= 6.62, P= 0.010).

Overall, 27.9% (260) of women took the second dose of misoprostol provided to them at home (Table 2). The likelihood of taking this second dose decreased with increasing initial misoprostol dose (χ2= 22.53, P < 0.0001). In group I, 36.6% of women took a second dose compared with 28.0% of those in group II (χ2= 4.97, P= 0.026) and 19.2% of those in group III (χ2= 22.55, P < 0.0001). The average total dose of misoprostol was 545 micrograms in group I compared with 768 micrograms in group II and 954 micrograms in group III, or an average of 155 micrograms of additional misoprostol with each increase in initial dose, despite the decreased likelihood of using a second dose.

Table 2.  Doses of misoprostol administered in week 1 by the study group*
DosesGroup I (400 micrograms orally)Group II (600 micrograms orally)Group III (800 micrograms vaginally)
%n%n%n
  • *

    Four women who did not take misoprostol and one who took four doses of misoprostol were excluded from the analysis.

163.720072.022480.8248
236.311428.08719.259
Average total dose (micrograms)545 768 954 

Complications

There were no significant differences among the three groups in terms of complications. There was one hospitalisation of a woman in group III, who died of a systemic Clostridium sordellii infection.27 One woman in group II received a blood transfusion. There were ten emergency room visits (four in group I, four in group II, and two in group III), primarily for bleeding or pain.

Adverse effects

With the exception of chills, there were no differences in adverse effects among the three study groups following administration of mifepristone (Table 3). The frequency of fever, headaches, and chills increased significantly across the three study groups, suggesting a possible dose-correlated increase. No other adverse effects showed a similar trend. However, comparing reports of adverse effects following misoprostol in the two oral groups (groups I and II), the frequency of diarrhoea increased significantly with a higher initial dose of misoprostol (17.9% in group I versus 26.2% in group II, P= 0.025), whereas there was no difference between the vaginal (group III) and oral groups (groups I and II). There were no significant differences or trends for vomiting and nausea.

Table 3.  Adverse effects and acceptability by study group
 Group I (400 micrograms orally), % (n)Group II (600 micrograms orally), % (n)Group III (800 micrograms vaginally), % (n)P value*
  • *

    Chi-square test for trend (df= 2) reported where significant.

  • **

    P= 0.025 for comparison between 400 and 800 micrograms groups.

  • ***

    P= 0.012 for comparison between 400 and 600 micrograms groups.

  • ****

    P < 0.001 for comparison between 400 and 800 micrograms groups; P= 0.006 between 600 and 800 micrograms groups.

  • *****

    P < 0.0001 for comparison between 400 and 800 micrograms groups and between 600 and 800 micrograms groups.

  • ******

    P= 0.011 for comparison between 400 and 800 micrograms groups.

Adverse effects after mifepristone
Any37.3 (117)38.1 (117)36.5 (113) 
Nausea22.9 (72)28.8 (88)24.8 (77) 
Vomiting8.6 (27)6.9 (21)7.1 (22) 
Diarrhoea4.2 (13)3.6 (11)4.5 (14) 
Fever1.9 (6)3.3 (10)2.6 (8) 
Chills**3.8 (12)6.2 (19)8.1 (25)0.033
Headache8.9 (28)11.4 (35)12.9 (40) 
Fatigue20.4 (64)18.6 (57)21.0 (65) 
Adverse effects after misoprostol
Any65.5 (205)71.6 (219)72.4 (223) 
Nausea39.6 (124)46.6 (142)41.6 (128) 
Vomiting19.8 (62)24.6 (75)24.4 (75) 
Diarrhoea***17.9 (56)26.2 (80)19.5 (60) 
Fever****15.0 (47)17.7 (54)26.9 (83)<0.0001
Chills*****25.9 (81)29.2 (89)43.5 (134)<0.0001
Headache******20.4 (64)23.6 (72)29.2 (90)0.011
Fatigue38.7 (121)45.6 (139)42.2 (130) 
Acceptability
Bleeding92.9 (275)93.1 (270)96.2 (275) 
Time94.9 (280)94.8 (272)95.4 (270) 
Adverse effects93.3 (270)91.7 (264)91.9 (262) 
Pain84.6 (252)84.8 (246)75.1 (214)0.003
Overall satisfaction92.4 (281)91.2 (268)88.1 (260) 

Pain

Increased misoprostol dose was associated with greater reported pain (Table 4). Mean pain score was lowest in group I and greatest in group III (5.8 versus 6.7). The greatest difference in mean pain score occurred among women with pregnancies less than 42 days LMP, with a mean pain score of 5.0 in group I and 6.3 in group III (P= 0.02).

Table 4.  Mean pain by gestational age and study group
 Group I (400 micrograms orally)Group II (600 micrograms orally)Group III (800 micrograms vaginally)P value*
  • *

    One-way ANOVA comparing mean pain scores across study groups.

  • **

    One-way ANOVA comparing mean pain scores across gestational age groups.

<42 days5.05.46.30.013
42–49 days6.46.57.00.191
50–56 days6.36.76.90.639
Overall5.86.16.70.001
P value**0.0070.0130.209 

Acceptability

Overall, women were satisfied with the procedure (90.6%, n= 802), and acceptability was not statistically different across the study groups (group I, 92.4%; group II, 91.2%; and group III, 88.1%; P= 0.072). In all the groups, most women indicated that bleeding, time required for treatment, and adverse effects were acceptable (Table 3). The proportion of women indicating that pain was acceptable varied significantly across groups and was significantly lower in the vaginal group (group I, 84.6%; group II, 84.8%; and group III, 74.1%; P= 0.003).

Discussion

We found that regimens of low-dose mifepristone followed by 400 micrograms of oral misoprostol, 600 micrograms of oral misoprostol, or 800 micrograms of vaginal misoprostol can be equally successful in terminating pregnancies up to 8 weeks LMP. These results are both similar and different from what has been previously reported in the literature. Aubeny and Chatellier tested 400-microgram doses of oral and vaginal misoprostol, 48 hours after 600 mg mifepristone, repeated once at 3 hours if the pregnancy had not been expelled. In women with gestations less than 7 weeks LMP, they found no differences in success after 4 hours following the initial dose of misoprostol (76.6% in the oral group versus 77.0% in the vaginal group, P= 0.92).15 Similarly, the World Health Organization (WHO) found no differences in efficacy between regimens with initial doses of 400 micrograms of oral misoprostol and 800 micrograms of vaginal misoprostol following low-dose mifepristone in women up to 8 weeks LMP (93.5% in the oral group versus 93.8% in the vaginal group, P= 0.914).19 In contrast, both Schaff et al.16,17 and el-Refaey et al.18 reported significantly higher efficacy in regimens with vaginal versus oral misoprostol administered 24–48 hours following low-dose mifepristone.

  • image(1)

[ Regimens of misoprostol with mifepristone for early medical abortion. ]

For regimens with oral misoprostol, several published studies have confirmed that there is a significant decrease in success rate among women between 8 and 9 weeks LMP compared with that among those less than 8 weeks LMP. Spitz et al.,28 testing a regimen of 600 mg mifepristone, followed by 400 micrograms of oral misoprostol, reported a 10.6% decrease in efficacy among women between 8 and 9 weeks LMP compared with that among those less than 8 weeks LMP. Hedley et al.,29 in a meta-analysis of data aggregated from four studies testing a 600 mg mifepristone and 400 micrograms of oral misoprostol regimen, found that compared with women between 7 and 8 weeks LMP, the odds of experiencing a successful medical abortion were 25% lower for those between 8 and 9 weeks of gestation. In those studies, women did not receive additional doses of misoprostol. The WHO, however, testing a multidose oral regimen following low-dose mifepristone reported neither a significant decrease in efficacy among women greater than 8 weeks LMP compared with those less than 8 weeks LMP nor a significant trend suggesting a decrease associated with gestational age.19

For regimens with vaginal misoprostol, published data suggest that there may be no significant decrease in efficacy among women between 8 and 9 weeks LMP compared with that among those less than 8 weeks LMP. Testing a multiple dose vaginal misoprostol regimen, the WHO reported neither a significant decrease in efficacy among women between 8 and 9 weeks LMP compared with those less than 8 weeks nor a trend of decreasing efficacy associated with gestational age.19 Data reported by Schaff et al.,16 however, demonstrated a significant trend of decreasing efficacy associated with gestational age following regimens of low-dose mifepristone and one dose of 800 micrograms vaginal misoprostol (P= 0.045). This trend remained significant after a second similar dose of misoprostol (P= 0.050), although the decrease in efficacy among women between 8 and 9 weeks of gestation compared with that among those less than 8 weeks of gestation was not significant.16

Most published studies of mifepristone–misoprostol use for early pregnancy termination, however, are not designed to measure significant effects by gestational age group, and sample sizes among women with later gestational ages are small. Indeed, data from three comparative studies suggest that a regimen with vaginal misoprostol administration among women who are up to 9 weeks LMP is superior to one with oral administration. This putative superiority, however, may be confounded by significant decreases in efficacy following oral misoprostol use among women between 8 and 9 weeks LMP. This trend may also explain why we found no significant efficacy differences between oral and vaginal regimens, as we tested these regimens in women less than 8 weeks LMP.

Data from our study and other published studies also suggest that gestational age, route of misoprostol administration, and amount and doses of misoprostol may have subtle yet important impacts on the overall success of mifepristone–misoprostol regimens for pregnancy termination among women less than 9 weeks LMP. Among women with gestations less than 8 weeks LMP, in particular, the existing literature does not explain sufficiently the influence of these variables, in part, because the initial dose of the most frequently tested vaginal misoprostol regimen (800 micrograms) is twice the most usual oral dose (400 micrograms). Testing equivalent 400-microgram doses of oral and vaginal misoprostol, Aubeny and Chatellier found no differences in efficacy.15 Unlike in that study, we did not compare oral and vaginal administration for each dose tested. Nonetheless, our results suggest that for women up to 8 weeks LMP, differences in efficacy are more likely attributable to gestational age and not to the initial misoprostol dose or route of administration, as only gestational age was significantly associated with success.

We also found few significant differences in adverse effects reported among the study groups. Similar to previous findings,17 our results suggest that the frequency of chills, fever, and headache may increase with increasing doses of misoprostol. In contrast to previous findings, we found few significant differences in gastrointestinal adverse effects, such as nausea, vomiting, or diarrhoea, between oral and vaginal administration. Our findings suggest, however, that the frequency of diarrhoea may increase with increased doses of misoprostol and may be more likely to occur with oral administration. Both Schaff et al.17 and el-Refaey et al.18 found a greater frequency of these adverse effects reported by women administering misoprostol orally, as compared with those administering it vaginally. However, both of those studies tested oral regimens of 800 micrograms of misoprostol, whereas we tested lower oral doses of 400 and 600 micrograms of misoprostol. These lower doses may be better tolerated with respect to gastrointestinal adverse effects.

Despite minor differences in adverse effects, the overall acceptability of the procedure was high in all the study groups. Later gestational age and an initial dose of 800 micrograms of vaginal misoprostol, however, were each predictors that a woman would find the procedure more painful. Additionally, women in the vaginal group reported that pain was less acceptable compared with those in the two oral groups (75.1% acceptability in the vaginal group versus 84.7% in the oral groups, P < 0.001). This result is consistent with previous reports. Schaff et al.17 found that both pain, together with the duration of the procedure and bleeding (amount and duration), were major predictors of rating the procedure as less acceptable.

A possible limitation to this study is that it was not blinded. We agree, however, with Crane et al.30 that controlling all routes of administration with placebo may mask women’s preferences. Nonblinded administration mimics ‘real-world’ medical abortion practice, where women know how they are receiving medication. For this reason, we believe that data from this study provide a more accurate assessment of women’s attitudes towards routes of administration. Further, blinding may also obscure differences in efficacy, if they are associated with women’s reactions to different routes of administration.

In summary, offering women a choice of regimens may improve the acceptability of the procedure. Guidelines for mifepristone medical abortion in the USA and UK have widely recommended vaginal misoprostol with the view that it is more effective and more acceptable. The literature and results from this study indicate that this may not be the case generally, although it may be so for women with gestations greater than 8 weeks LMP. For women with gestations less than 8 weeks LMP, all three regimens tested appear equally effective, easily managed, and well tolerated. Women may also prefer oral misoprostol (Table 5), perhaps because the commonly used vaginal misoprostol dose appears to be associated with increased pain, at least as compared with the two oral regimens studied here.

Table 5.  Treatment preferences by route of administration*
 Oral groups, % (n)Vaginal group, % (n)OR (95% CI)
  • *

    Only the last 287 participants were asked this question.

Prefer oral94.5 (190)55.8 (48)13.67 (6.51–28.72)
Prefer vaginal2.0 (4)38.4 (33)0.03 (0.01–0.10)
Not sure3.5 (7)5.8 (5)0.58 (0.18–1.90)

Acknowledgements

The authors would like to thank all the women who participated in this study and each of the investigators who provided their care. We are indebted to our colleagues Dr Lisa Lugtig, for her work during the initial phases of the study, and Dr Inna Tcherenchova, who was invaluable in managing the data. We also thank the Packard Foundation for its generous support of this project.

Footnotes

  • *

    A death in the study (see below) resulted in its prolonged suspension while sponsors and regulatory agencies determined that the death was not related to the study medication or protocol. As a result, we elected to terminate enrolment so that the data could be analysed. With 318–319 women enrolled per study group, there was greater than 80% statistical power in a two-sided statistical 5% test significant level test to detect the same 5% difference between any two study groups.

Ancillary