A randomized double-blind comparison of two single doses of mifepristone for emergency contraception*

B.L. Xiao1,4, H. von Hertzen3, H. Zhao2 and G. Piaggio3

1 Department of Reproductive Endocrinology and 2 Computer Centre, National Research Institute for Family Planning, No.12 Da Hui Si, Beijing 100081, People's Republic of China and 3 Department of Reproductive Health and Research World Health Organization, 1211 Geneva 27, Switzerland


    Abstract
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
BACKGROUND: Previous trials have shown the potential of 10 mg of mifepristone in emergency contraception. The aim of this trial was to investigate whether 10 mg of mifepristone has the same efficacy as 25 mg. METHODS: This double-blind, randomized trial was carried out in 10 family planning institutes and hospitals in China. Women who met recruitment criteria and requested emergency contraception within 120 h of a single act of unprotected coitus were randomized using a computer-generated list to either 10 or 25 mg of mifepristone within each centre. RESULTS: Among 3052 women enrolled, the outcome was known for 3030 women, 1516 in the 10 mg group and 1514 in the 25 mg group. Seventeen pregnancies occurred in each group, giving a pregnancy rate of 1.1%. The relative risk of pregnancy for women treated with 10 mg mifepristone compared with those treated with 25 mg was 1.0 (95% CI: 0.51–1.95) and equivalence was demonstrated within a two-fold margin. Both doses prevented 85–86% of pregnancies expected to have occurred if no treatment had been given. The pregnancy rate nearly doubled if women had further acts of intercourse. Efficacy decreased with treatment delay. Side-effects were uncommon and mild. CONCLUSIONS: A dose of 10 mg of mifepristone is sufficient for emergency contraception. Earlier treatment is preferable, although the method can be used effectively for up to 5 days after intercourse.

Key words: dose/efficacy/emergency contraception/mifepristone/RCT


    Introduction
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Five randomized trials (Xie et al., 1998Go; Zhang et al., 1998Go; Cheng et al., 1999Go; Sang et al., 1999Go; Task Force on Post-ovulatory Methods of Fertility Regulation, 1999Go) have compared the efficacy and side-effects of different doses of mifepristone in emergency contraception. These trials have been included in a systematic review on interventions for emergency contraception (Cheng et al., 2001Go). The largest of these trials was a multicentre trial undertaken by the World Health Organization (WHO), in which the efficacy of three doses of mifepristone (600, 50 and 10 mg) (Roussel-Uclaf, France) for emergency contraception were compared, when administered up to 120 h after intercourse. The three doses did not differ significantly in proportions of pregnancies (1.3, 1.1 and 1.2% respectively), and the authors concluded that lowering the dose of mifepristone 60-fold did not seem to decrease its effectiveness as an emergency contraceptive. However, due to the size of the study (559, 560, and 565 women in the three treatment groups respectively) the investigators could not exclude up to almost 3-fold differences in efficacy: the relative risks (RR) of pregnancy in the 50 and 10 mg groups compared with the 600 mg group were 0.85 [95% confidence interval (CI): 0.29–2.51] and 0.99 (0.35–2.80) respectively. No major side-effects occurred in that study. However, the delay in the onset of next menstruation was significantly related to the mifepristone dose (P < 0.01). The other four trials were conducted in China using Chinese mifepristone (Xie et al., 1998Go; Zheng et al., 1998; Cheng et al., 1999Go; Sang et al., 1999Go). The combined results of the five trials comparing doses of mifepristone showed no evidence of a dose-related efficacy (Cheng et al., 2001Go). The side-effect profile was better with lower doses, since higher doses were shown to be associated with a delay in menses (Task Force on Post-ovulatory Methods of Fertility Regulation, 1999Go). A double-blind randomized study based in China, which compared a 10 mg dose of mifepristone and a dose of levonorgestrel, found a lower pregnancy rate after mifepristone (Wu et al., 1999Go). To date, the results from studies suggest that mifepristone could have advantages compared with existing emergency contraceptives, since it can be administered in one low dose with minimal side-effects.

As both 10 and 25 mg tablets of mifepristone have been used in China, we carried out this randomized, double-blind study to investigate the effectiveness and side-effects of the two doses of the drug, including timing of subsequent menstrual period, when the treatment was given within 120 h (5 days) of unprotected coitus. Our a priori hypothesis was that the two doses would have similar efficacy in preventing pregnancy. The main outcome measures included pregnancy rates, proportions of pregnancies prevented and side-effects. We planned to explore the influence of treatment delay as well as of further acts of intercourse on the efficacy of the treatment, although we acknowledged that the power might not be sufficient to detect relevant differences induced by these variables.


    Materials and methods
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Protocol
This study was conducted in 10 family planning institutes or hospitals in Beijing, Guangdong, Jiangsu, Liaoning, Shangdong, Shanghai, Sichuan and Tianjin. Approval for the study was obtained from the institutional review board of each of the participating centres, the Ethics Committee of the State Family Planning Committee of China as well as from the Scientific and Ethical Review Group of the WHO Department for Reproductive Health and Research.

Participants eligible for the study were healthy women with regular menstrual cycles (24–42 days duration) requesting emergency contraception within 120 h of a single act of unprotected coitus in the present menstrual cycle; who were willing to abstain from further acts of unprotected intercourse during that cycle; and who were available for follow-up over the next 6 weeks. Women who had recently discontinued hormonal contraception or had been pregnant were included only if they had had at least one normal menstrual cycle before the current cycle. Pregnancy was excluded by means of a sensitive urinary pregnancy test. Exclusion criteria were: current pregnancy or breast-feeding; use of hormonal contraception in the current cycle, and contra-indications to mifepristone use. In addition the centres did not enrol women likely to continue a pregnancy should the method fail. Women satisfying these criteria were admitted into the study after they had given written informed consent.

At admission, relevant medical, gynaecological and obstetric histories were recorded. Each participant received two tablets orally, one active and one placebo, at the assigned dose of mifepristone, at the clinic. They were advised not to have further unprotected sex but use condoms, which were provided. Women were asked to keep a diary of side-effects for the 7 days following the treatment and to record any bleeding and acts of intercourse until the follow-up visit about a week after the expected onset of next menstruation.

If the woman had a normal menstruation, it was considered that she had completed the study. In cases where menstruation was not normal or had not started by the time of the follow-up visit, hCG was measured in blood or urine. If negative, a further follow-up visit was arranged and if positive, ultrasound examination was carried out to estimate the duration of gestation. If menses had not resumed by the time of the second follow-up visit and the pregnancy test was negative, the treatment was regarded as successful. The primary outcome measure was unintended confirmed pregnancy. Secondary outcome measures included side-effects and delay in the onset of the next menstruation.

For the sample size calculation we assumed equal efficacy between the two doses at the level of 1% in the pregnancy rate for the 10 mg dose of mifepristone. To demonstrate one-sided equivalence (non-inferiority) within a margin of equivalence of 1% on the absolute scale with 80% power and using a 95% CI, 1224 subjects per group are required. Allowing for a 10% of loss to follow-up and further increasing the sample size by 10% to allow for a mis-specification of the pregnancy rate, the proposed sample size for the trial was about 1550 women per group (3100 in total). The margin of 1% on the absolute scale from a pregnancy rate of 1% implies a margin of equivalence of two on the relative scale.

Analysis
Analysis was done by intention to treat, except for the exclusion of women lost to follow-up (who had unknown outcome). The efficacy was measured in two ways, calculating crude (unadjusted) and centre-adjusted pregnancy rates (using the Mantel-Haenszel procedure) as well as the estimated reduction in expected pregnancies or prevented fraction (one minus the ratio of observed pregnancies to expected pregnancies). The expected number of pregnancies in both groups was estimated by multiplying the number of women having unprotected coitus on each day of the cycle by the conception probability on that cycle day. The date of ovulation was estimated by subtracting 14 days from the expected date of the next menstrual period. The probabilities of conception by cycle day were obtained from the recognizable conceptions pooled from two data sets (Trussell et al., 1998Go), which only include clinical pregnancies and not those diagnosed only by biochemistry.

To compare the efficacy of the treatments, RR were calculated by standard methods, and their 95% CI using the Taylor series. The Mantel-Haenszel stratified RR was also calculated to adjust for centres, and the Breslow and Day {chi}2 statistic was used to test the homogeneity between centres. Odds ratios and their 95% CI were also computed, using the statistical software SAS (version 6.12) to test interactions (delay by regimen, further acts by regimen). Stratified analyses were done to compare the efficacy between regimens among women who had no further acts of intercourse between the treatment and the follow-up visit(s) with those who had intercourse with or without contraception.

We also examined the effect of delay in treatment administration on the efficacy and calculated the crude pregnancy rates for each 24 h interval of delay. Then we used logistic regression with group and linear term for delay to estimate the odds ratio for delay and its 95% CI, to test the linear trend and lack of fit of the model, and to adjust for possible confounders.

Numbers and percentages of women with side-effects were compared between groups, using the Mantel-Haenszel method to combine centres. Survival analysis with log-rank tests was used to estimate percentiles and to compare the delay of menses between the two groups.

Prior to commencing the study it was decided that the study would be stopped if the lower 95% confidence limit on the failure rate in a group was >3.5%. We conducted one interim analysis for this purpose and the trial was not stopped. We did not use stopping rules based on the comparison of the two treatments.

Assignment
The individual participant was the unit of randomization. The randomization sequence was generated by the data management group in the National Research Institute for Family Planning (NRIFP), Beijing, using a programme to assign participants to treatment groups within centres. Each centre received assignments by randomly permuted blocks with a fixed block size of 10.

The manufacturer (Hualian Pharmaceuticals Co., Ltd, Shanghai, China) supplied the NRIFP with two types of sealed packs containing: (i) one small tablet of 10 mg of mifepristone and one larger placebo tablet of 25 mg; and (ii) one small placebo tablet of 10 mg and one larger tablet of 25 mg of mifepristone. The active tablet and the corresponding placebo tablet were identical in size and shape. The packs were labelled at the NRIFP sequentially with the number of the centre, participant number and the expiry date. Random samples of labelled packages were taken for quality control purposes, both prior to and during the trial.


    Results
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Between May, 1998 and April, 1999 a total of 3052 women were enrolled for the study in 10 centres (Table IGo), 1527 of them for the 10 mg group and 1525 for the 25 mg group. The outcome was unknown for 22 (0.7%) women, who were lost to follow-up. There were four women with cycle length outside the range 24–42 days, that were protocol violations. All women received the assigned treatment. Seven (0.5%) women in the 10 mg group and eight (0.5%) in the 25 mg group vomited within 1 h of treatment. The women lost to follow-up were excluded from the analysis and those with protocol violations or vomiting within 1 h after treatment were included (Figure 1Go).


View this table:
[in this window]
[in a new window]
 
Table I. Numbers of women enrolled and pregnancies by treatment group
 


View larger version (21K):
[in this window]
[in a new window]
 
Figure 1. Flow-chart of trial participants.

 
The two randomized treatment groups were similar in baseline characteristics (Table IIGo). The mean age of the women was 30 years, ~67% of them had had previous normal deliveries and 80% had had at least one induced abortion. Most (70%) women had experience of the use of barrier methods, mainly condom, and 21% had used emergency contraception in the past, while 9% of the women had never used a method of contraception. As a reason for requesting emergency contraception, >58% of the women reported that they had not used any contraception at coitus, while 38% had a condom failure or suspected such failure. In all, 1645 (54%) women requested emergency contraception within the first 24 h following intercourse and only 243 (8%) requested it after 72 h.


View this table:
[in this window]
[in a new window]
 
Table II. Baseline characteristics of participants
 
The 22 women lost to follow-up were younger (mean age 26 years) than the other women, had longer mean time from coitus to treatment (34.7 h), fewer had previous normal deliveries (32%), induced abortions (59%) or had used barrier methods before (14%), and more had used natural methods (73%).

Of the 3030 women who completed the follow-up 34 were found to be pregnant, 17 of 1516 (1.1%) in the 10 mg group and 17 of 1514 (1.1%) in the 25 mg group (95% CI: 0.7–1.8 in each group) (Table IIIGo). The RR of pregnancy for women treated with 10 mg mifepristone compared with those treated with 25 mg was 1.0 (95% CI: 0.51–1.95). All pregnancies were intrauterine and all pregnant women opted to have induced abortion. If no treatment had been provided, we estimated that there would have been a total of 115 pregnancies in the 10 mg group and 126 pregnancies in the 25 mg group. Thus, among these women the 10 mg dose prevented 85% of pregnancies and the 25 mg dose prevented 86% (Figure 2Go). Adjustment for centre using the Mantel-Haenszel procedure produced almost identical results (RR = 1.00, 95% CI: 0.50–2.00). There was no statistical heterogeneity between centres (P = 0.58 from a Breslow and Day test of homogeneity of odds ratios).


View this table:
[in this window]
[in a new window]
 
Table III. Crude pregnancy rates and prevented fractions by regimen
 


View larger version (17K):
[in this window]
[in a new window]
 
Figure 2. Observed ({blacklozenge}) and expected ({blacksquare}) pregnancies by day of the cycle in which unprotected intercourse took place, for the two groups. Day 0 is defined as the day of ovulation, calculated as the expected date of next menstruation minus 14 days.

 
We observed a decline in efficacy when considering a delay in treatment in each of 5 successive days within intercourse (Table IVGo). This trend was similar in both dose groups (P = 0.44 for the regimen by delay group interaction) and the linear trend for both groups combined was significant (P = 0.02 from the {chi}2 for trends, P = 0.03 from the logistic regression). The relative increase in the odds of pregnancy was 35% for a delay of 24 h in treatment administration (OR = 1.35, 95% CI: 1.02–1.78; P = 0.03), adjusting for age, weight, gravidity and day of the cycle in which coitus took place (P = 0.03). When comparing pregnancy rates for women treated within 48 h of coitus with that for women treated later, we obtained 21/2381 (0.9%) and 13/649 (2.0%) respectively (RR = 2.30, 95% CI: 1.17–4.53).


View this table:
[in this window]
[in a new window]
 
Table IV. Treatment delay and crude pregnancy rates
 
A total of 1506 women reported not having had intercourse between treatment and menstruation, or between treatment and last visit, if pregnant, while 1524 women reported at least one. There were 12 (0.8%) and 22 (1.4%) pregnancies among these groups respectively. Thus, the pregnancy risk was nearly twice as high among women with further acts of intercourse (RR = 1.81, 95% CI: 0.91–3.61). Further, seven of the 82 women (8.5%) who had intercourse after treatment for failure of the method or with no contraception became pregnant while 15 of the 1442 (1.0%) who had used contraception successfully became pregnant (RR = 8.21, 95% CI: 3.84–17.30). There was no interaction of regimen by this variable (P = 0.99).

It is unlikely that unreported pregnancies among women lost to follow-up affect the conclusions regarding the detection of a significant difference. Even if all 11 women lost to follow-up in the 10 mg group had been pregnant and none of the 11 women lost from the 25 mg group, the RR of pregnancy in the 10 mg group compared with that of the 25 mg group would have been non-significant (RR = 1.64, 95% CI: 0.90–2.99). However, if the number of unreported pregnancies were not balanced between regimens among women lost to follow-up, equivalence of the two treatments would not have been demonstrated within a margin of equivalence of two on the relative scale.

Data on side-effects were available for 3033 women, 1517 in the 10 mg group and 1516 in the 25 mg group. A total of 10–11% of the women reported nausea or vomiting during the week following treatment. The prevalence of other side-effects was also low and similar in both treatment groups (Table VGo). Most women, 964 from 1499 (64%) in the 10 mg group and 919 from 1497 (61%) in the 25 mg group, had their menses about the expected time (within 3 days) and 149 from 1499 (9.9%) and 137 from 1497 (9.1%) respectively had a delay of >1 week. The timing of menses did not differ between the two regimens (P = 0.33, Figure 3Go).


View this table:
[in this window]
[in a new window]
 
Table V. Side-effects within the first week after treatment
 


View larger version (15K):
[in this window]
[in a new window]
 
Figure 3. Timing of menses after treatment for the two groups ({blacksquare} = 10 mg; {square} = 25 mg) expressed as the day of start of menses in relation to the expected day (Day 0). Minus are days of menses started earlier than 0, and plus are days delayed.

 

    Discussion
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
The results of this study show that the doses of 10 and 25 mg of mifepristone have an equivalent efficacy in preventing pregnancy within a margin of equivalence of two on the relative scale, when used for emergency contraception. The incidence of side-effects after treatment is low and very similar in both dose groups. The study also confirms the finding of previous studies that the 10 mg dose of mifepristone is effective for emergency contraception and that the use of a higher dose of mifepristone has no advantage.

The WHO trial included women from different populations while this study only enrolled women from Chinese centres. Nevertheless, the efficacy of this method of emergency contraception was similar in the two trials: pregnancy rates for all regimens combined were 1.1% (95% CI: 0.8–1.6) in this trial, and 1.2% (95% CI: 0.7–1.8) in the WHO trial. Also, the pregnancy rates for the 10 mg dose were similar between the two trials (1.1%, 95% CI: 0.7–1.8 and 1.2%, 95% CI: 0.5–2.5 respectively). When the number of observed pregnancies was compared with the number of those expected to have occurred without treatment, the 10 mg dose prevented 85% of pregnancies in both trials. Other trials using Chinese mifepristone reported higher pregnancy rates (Cheng et al., 1999Go; Xie et al., 1998Go; Sang et al., 1999Go), but one of these studies (Sang et al., 1999Go) recruited women between 25 and 96 h of intercourse. Two studies, on the other hand, recruited women up to 72 h of intercourse only (Xie et al., 1998Go; Zhang et al., 1998Go). Apart from the WHO trial, the other trials comparing mifepristone doses were small and pregnancy rates could have been due to chance.

The two earlier trials that included women who had had intercourse up to 120 h previously (Cheng et al., 1999Go; Task Force on Post-ovulatory Methods of Fertility Regulation, 1999Go), similar to the present study, had sample sizes which did not provide sufficient power to enable conclusions to be drawn from their results as to whether treatment delay had an influence on efficacy. In this study the earlier treatment appeared to be more effective. The risk of pregnancy increased by 35% when the treatment was administered on any given day after the first one, compared with 24 h earlier. The pregnancy rate was 0.9% if treatment administration was within 24 h, while the rate increased to 3.0% if administration was during the fifth day. It should be noted, however, that even when administered 72 h after intercourse the treatment still prevented 67% (95% CI: 24–90) of pregnancies. A declining trend in efficacy with time has also been demonstrated for other hormonal methods of emergency contraception (Piaggio et al., 1999Go).

The occurrence of the most notable side-effect of the treatment, namely delay in the next menses, was reported to be related to the dose. The proportion of women in the 10 mg group who had a delay of >7 days was 18% (97/553) in the previous WHO trial, and 10% (149/1499) in this trial. The difference may be due to the fact that in the WHO trial any bleeding that occurred within 5 days of treatment was regarded as treatment-related and not as menses, and thus in that study menses delay may be somewhat over-reported.

The rate of other side-effects was lower in the present trial compared with the previous multinational trial. This could be due to under-reporting, or a different perception of side-effects among Chinese women. However, in both trials the rates are low, which suggests that mifepristone is very well tolerated.

The delay of ovulation caused by the drug means that women may continue to be at risk for pregnancy after treatment should they have further acts of unprotected intercourse. This was also demonstrated in this trial. The pregnancy rate among women with no further acts was 0.8% while it was 1.0% among those women who had further acts of intercourse during which a condom had been used correctly, and as high as 8.5% among those who continued to have unprotected intercourse. It is important to emphasize to women during counselling that there is a high risk of pregnancy if acts of unprotected intercourse take place after emergency contraception with mifepristone.

One trial compared a lower 5 mg dose with doses of 10 and 25 mg (Xie et al., 1998Go), but the sample size was not sufficient to obtain evidence about the relative efficacy of the doses. To demonstrate equivalent efficacy for doses <10 mg for emergency contraception would require larger trials. Doses of <10 mg, however, even if found effective, are unlikely to provide any practical advantage. The 10 mg dose is already so low that it would not make a relevant difference in the price of the treatment should lower doses prove to be effective. The side-effect profile is also unlikely to differ.

We believe that this study has internal validity. The randomization produced treatment groups similar in all the important respects. The allocation was concealed from the centres participating in the trial. All pill packs were identical, containing two tablets—one active and one placebo—but it was impossible to distinguish the active one. The analysis of samples of pills at the beginning and during the study showed that the packing and drug content were correct. The sample size was large enough to show equivalence of the two doses within a two-fold margin.

This study confirms the potential of 10 mg of mifepristone for emergency contraception. It can be administered in one single dose and it causes very few side-effects. Although its efficacy appears to decline with time it still prevents a high percentage of pregnancies even up to 5 days following coitus.


    Acknowledgements
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
This study was funded by the Rockefeller Foundation and was part of the `Collaborative research and development on mifepristone to reduce unwanted pregnancies and recourse to abortion' initiative. Hua Lian Pharmaceuticals Co., Ltd, Shanghai, China, provided the drugs used in the study. We also like to acknowledge the contribution of the State Family Planning Commission, Department of Reproductive Health and Research of the World Health Organization, Concept Foundation and PATH who have made this national and international collaboration possible and successful.


    Notes
 
* *On behalf of the Clinical Research Team of Collaborative Research and Development on Mifepristone to reduce unwanted pregnancies and recourse to Abortion. Back

4 To whom correspondence should be addressed. E-mail: xiaobl{at}public.bta.net.cn Back


    References
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Cheng L., Tong, C. and Xiao, Z. (1999) Low doses of mifepristone for emergency contraception. Chin. J. Obstet. Gynecol., 34, 335–338.

Cheng, L., Gülmezoglu, A.M., Ezcurra, E. and Van Look, P.F.A. (2001) Interventions for emergency contraception (Cochrane Review). The Cochrane Library, Issue 1. Oxford Update Software.

Piaggio, G., von Hertzen, H., Grimes, D.A. and Van Look, P.F.A. (1999) Timing of emergency contraception with levonorgestrel and the Yuzpe regimen. Lancet, 353, 721.[ISI][Medline]

Sang, G.W., Shao, Q.X., Zhang, L.J., Zhang, M.Y., Chen, S.F., Song, S., Du, M.K., Wu, X.Z., Ding, J.H. and Weng, L.J. (1999) Mifepristone in combination with anordrin for emergency contraception: A randomized multicentre study. Chin. J. Obstet. Gynecol., 34, 331–334.

Task Force on Post-ovulatory Methods of Fertility Regulation (1999) Comparison of three single doses of mifepristone as emergency contraception: a randomized trial. Lancet, 353, 697–702.[ISI][Medline]

Trussell, J., Rodriquez, G. and Ellertson, C. (1998) New estimates of the effectiveness of the Yuzpe regimen of emergency contraception.Contraception, 57, 363–369.[ISI][Medline]

Wu, S.C., Wang, C.P., Wang, Y.Q., Cheng, W.Y., Zuo, S.H., Li, H., Xu, X., Wang, R.F. and Dong, J. (1999) A randomized double-blind, multicentre study on comparing levonorgestrel and mifepristone for emergency contraception. Chin. J. Obstet. Gynecol., 34, 327–330.

Xie, X., Liu, Y. and Lin, X. (1998) A clinical study on 600 cases of mifepristone for emergency contraception. Reprod. Contraception, 18, 224–226.

Zhang, Y., Qiao, G., Zhu, P., Zhang, S., Zhang, J. and Zhu, N. (1998) Clinical observation of three lower doses of mifepristone for emergency contraception. Chin. J. Fam. Plan., 8, 343–345.

Submitted on May 21, 2002; accepted on September 2, 2002.