WHO multinational study of three misoprostol regimens after mifepristone for early medical abortion.

II: Side effects and women's perceptions

Authors


*Dr H. von Hertzen, UNDP/UNFPA/WHO/World Bank Special Programme of Research, Development and Research Training in Human Reproduction, Department of Reproductive Health and Research, World Health Organization, Geneva, Switzerland.

Abstract

Objectives  To compare the side effect profiles of regimens of oral and vaginal administration of misoprostol after a single oral dose of 200 mg of mifepristone and to investigate patients' perceptions of medical abortion.

Design  Double-blind, randomised controlled trial.

Setting  Fifteen gynaecological clinics in 11 countries.

Population  A total of 2219 healthy pregnant women requesting medical abortion with ≤63 days of amenorrhoea. Two thousand women were asked about their perceptions of the method.

Methods  Mifepristone 200 mg orally on day one, followed by 0.8 mg misoprostol either orally or vaginally on day three. The oral group (O/O group) and one of the vaginal groups (V/O group) continued with 0.4 mg of oral misoprostol, and the vaginal-only group (V-only group) with oral placebo, twice daily for seven days. Side effects were recorded daily by women and reported at each visit. After misoprostol administration at the clinic, side effects were recorded at 1-hour interval up to 3 hours. Patients' perceptions were asked at the second follow up visit, six weeks after treatment.

Main outcome measures  The outcome measures were the following: pregnancy-related symptoms (nausea, vomiting, breast tenderness, fatigue, dizziness, headache), drug-related side effects (diarrhoea, fever, rash and blood pressure change), side effects related to the abortion process (lower abdominal pain) and women's perceptions of the method.

Results  The pregnancy-related symptoms decreased in all groups after misoprostol, and breast tenderness decreased already after mifepristone. Oral administration of misoprostol was associated with a higher frequency of nausea and vomiting than vaginal administration at 1 hour after administration. With oral misoprostol, diarrhoea was more frequent at 1, 2 and at 3 hours after administration than with vaginal administration. Misoprostol induced fever during at least 3 hours after administration in up to 6% of the women, this peak being slightly higher and taking place later with the vaginal route. Lower abdominal pain peaked at 1 and 2 hours after oral misoprostol, while it did so at 2 and 3 hours after vaginal misoprostol. In the two groups that continued misoprostol, 27% of women had diarrhoea between the misoprostol visit and the two-week follow up visit, compared with 9% in the placebo group. Among the women studied, 84% would choose medical abortion again, 9% would choose surgical abortion and 7% did not know. Twenty-three percent of the women would choose to have a possible future abortion at home, 70% at a health facility and 7% did not know.

Conclusions  The pregnancy-related symptoms decrease significantly with time during medical abortion. Nausea, vomiting and diarrhoea were more frequent after oral administration of misoprostol. Pain related to the abortion process occurs earlier after oral misoprostol. Should a need arise, a majority of women would choose medical abortion again and would prefer to have it at a health facility rather than at home.

INTRODUCTION

A combination of mifepristone followed two days later by a prostaglandin analogue misoprostol has proved an effective method for termination of early pregnancy. Misoprostol is designed for oral use, for prevention and treatment of gastric ulcers. However, as in termination of early pregnancy, after pretreatment with mifepristone, vaginal administration of misoprostol results in better efficacy.1,2 We wanted to study whether the frequency of side effects in vaginal use differs from that in oral use.

The side effects most commonly related to misoprostol use in medical abortion are lower abdominal pain, nausea, vomiting and diarrhoea.3 Nausea and vomiting are also common complaints in early pregnancy, in addition to being possible side effects of misoprostol. A previous study has shown that the incidence of diarrhoea and vomiting is significantly lower when misoprostol is used vaginally rather than orally.1

As some studies suggest that the majority of women prefer the oral route of misoprostol because they feel it is more convenient and private,4 we wanted to assess whether the efficacy of oral misoprostol could be improved by continuing its administration for one week. Continued misoprostol might increase the abortion rate and shorten the duration of bleeding,5–8 therefore we wanted to assess whether the continued regimen would be feasible regarding the possible side effects.

Administration of misoprostol at home has been proposed especially for countries where there is a shortage of service capacity for the provision of abortion.9,10 We asked the women after undergoing medical abortion in a clinic whether they would prefer to have a possible future medical abortion at home or at a health facility.

The purpose of this trial was to compare efficacy, duration of bleeding, side effect profiles and women's perceptions of three mifepristone–misoprostol regimens for medical abortion differing in the route and duration of administration of misoprostol. In a previous paper, we reported comparisons of the abortifacient efficacy, timing of expulsion and the duration of bleeding between the three regimens.2 In this paper, we compare the detailed side effect profiles and women's perceptions of the treatments.

METHODS

We conducted a multicentre, double-blind, randomised controlled trial to compare three mifepristone–misoprostol regimens for medical abortion differing in the route of the initial administration of misoprostol and in the continuation or not of oral misoprostol for additional seven days. Centres in 15 cities participated in the trial: Beijing, Hong Kong and Shanghai in China; Chandigarh, Mumbai and New Delhi in India; Helsinki in Finland; Ho Chi Minh City in Viet nam; Ljubljana in Slovenia; Oslo in Norway; Singapore in Singapore; Stockholm in Sweden; Szeged in Hungary; Targu Mures in Romania; Ulaanbaatar in Mongolia.

The trial included pregnant women requesting legal termination of pregnancy who were in good general health, with duration of pregnancy not more advanced than nine weeks or ≤63 days of amenorrhoea since the onset of the last menstrual period, confirmed by ultrasonography, and who were older than the age of legal consent. Other inclusion and exclusion criteria, procedures for the ethical approval of the trial, for randomisation, allocation concealment and blinding are described in detail in the previous paper.2

The dose of mifepristone in the three arms was a single oral dose of 200 mg on day one of the study (Table 1 in the previous paper).2 On day three of the study (36–48 hours after the administration of mifepristone), the oral/oral group (O/O) received 0.8 mg of misoprostol orally and placebo tablets vaginally, the vaginal/oral (V/O) and the vaginal-only (V-only) groups received 0.8 mg of misoprostol vaginally and placebo tablets orally. After administration of misoprostol, women were observed for 3 hours and side effects were recorded at 1-hour intervals during this time. At admission, women were requested to maintain a diary card to record days of bleeding and the occurrence of side effects for the duration of the study. Those in groups O/O and V/O continued with 0.4 mg of oral misoprostol twice daily for seven days, starting on day four of the study, while those in group V-only took placebo tablets. Participants returned for follow up evaluations on days 15 and 43 after beginning the treatment.Side effects and medical treatment used, if any, were recorded daily by women and reported at each visit.

We collected data on signs and symptoms, including blood pressure, pulse and temperature, and on medication used, recorded before treatment (symptoms during this pregnancy), two days after administration of mifepristone (recording period from mifepristone to arrival at the misoprostol visit), immediately after misoprostol administration and at 1-hour intervals during the 3-hour observation period. At the two-week follow up visit, women were interviewed and the diary cards were checked for side effects experienced since the previous visit. All had haemoglobin determinations at every visit after admission. At the end of the visit that took place six weeks after administration of mifepristone, women were asked three questions on their perceptions of the method (acceptability). The interviews started after about 200 women had completed the follow up, therefore these participants did not provide information on acceptability.

The outcome measures of the trial that we consider in this paper are symptoms and side effects that we have classified as follows: ‘pregnancy-related symptoms’ (nausea, vomiting, breast tenderness, fatigue, dizziness, headache and fainting), ‘drug-related side effects’ (diarrhoea, fever defined as temperature >38°C, rash, blood pressure and pulse change)and ‘side effects related to the abortion process’ (lower abdominal pain). Women's perceptions of the method (acceptability) were assessed through three questions regarding future choice: preference for surgical or medical abortion, willingness to take medicines without supervision by medical professionals and preference for having medical abortion at home or at a medical facility. Duration of bleeding, an abortion-related variable, was considered in the previous paper.2

The main outcome of the trial was efficacy to achieve complete abortion and therefore the number of subjects to be recruited into the trial was calculated based on this criterion. The number required, 700 per group, or about 2100 total, provided at least 96% power in a two-sided 5% significance level test to detect a difference of at least 10% between regimens in side effects with prevalences ranging from 20% to 60%. For rare side effects, ranging from 2% to 10%, the power provided by this sample size to detect a difference of at least 4% between regimens was at least 78%.

The data were analysed centrally in Geneva. All subjects with side effects were included in the safety analysis as randomised. All subjects interviewed for acceptability were included in the acceptability analysis. We compared baseline demographic characteristics of women in the three groups, as well as gynaecological and obstetric history and characteristics of the present pregnancy, for these populations.

Percentages of women with each side effect were calculated and simple χ2 tests were used to explore differences between groups in side effects. A repeated-measures analysis using an extension of logistic regression to account for intrasubject correlation provided information on the effects of treatment, time and interaction of treatment by time. The same technique was used to investigate the effect of other factors, like parity, on side effects. Tests of significance for continuous variables, like blood pressure, were done using repeated-measures analysis of variance. All analyses were adjusted for centre and for the baseline value of the symptom analysed (except fever and rash, for which we did not record admission values).

The association of baseline characteristics with preference for surgical or medical abortion, willingness to take medicines without supervision by medical professionals and preference for having medical abortion at home or at a medical facility, were explored through univariate χ2 tests first and then through multivariate logistic regression. Thus, the effect of group, age, gravidity, previous induced abortions, parity, length of amenorrhoea and ethnic group on the acceptability outcomes was adjusted for the other variables (all variables were included in the model, except that gravidity, previous induced abortions and parity were included one at a time because of their high correlation). In the study of the association of the abortion outcome, the duration of bleeding and abdominal pain with the acceptability outcomes, the adjustment was made for group, age, parity, length of amenorrhoea and ethnic group.

RESULTS

A total of 2219 participants were enrolled in 15 centres, 740 women were randomised to oral misoprostol and continued doses of oral misoprostol (O/O), 741 to vaginal misoprostol and continued doses of oral misoprostol (V/O) and 738 to vaginal-only misoprostol (V-only). All of them provided information on side effects at admission and during the misoprostol visit on day three. At the follow up visit on day 15, 2126 participants provided information on side effects (4.2% lost to follow up for safety outcomes). Details on lost to follow up, eligibility and compliance were given in the Efficacy paper.2

Acceptability interviews on the second follow up visit were conducted in 2000 women, 665 in the O/O group, 668 in the V/O group and 667 in the V-only group.

The participants' baseline characteristics by group were reported in the previous paper.2

Baseline characteristics of participants with acceptability information were similar in all important respects, between groups and also compared with those of the recruited population, used in the efficacy analysis, and are not shown.

Baseline (day one) ‘pregnancy-related symptoms’ (nausea, vomiting, breast tenderness, fatigue, dizziness, headache and fainting), ‘drug-related side effects’ (diarrhoea, fever and rash) and ‘side effects related to the abortion process’ (lower abdominal pain) are shown by group in Tables 1, 2 and 3, respectively.

Table 1.  Number and percentages of women with the pregnancy-related symptoms, nausea, vomiting, breast tenderness. fatigue, dizziness, headache and fainting, by treatment group and time periods (1 = at admission, 3 = between mifepristone and misoprostol, 3–0 = immediately after misoprostol administration, 3–1 = 1 hour after misoprostol, 3–2 = 2 hours after misoprostol, 3–3 = 3 hours after misoprostol, 15 = between misoprostol visit and day 15).
Timing (day–hour)GroupP(2)
O/OV/OV-only
No. of cases (n= 740)(1)%No. of cases (n= 741)(1)%No. of cases (n= 738)(1)%
  • (1)

    Day 15: n= 705 for O/O, n= 715 for V/O and n= 706 for V-only.

  • (2)

    With Bonferroni correction for simultaneous inferences for 11 side effects and time points, results are significant at α= 0.05 if P < 0.0007.

Nausea
145261.144359.845761.90.70
342557.444860.545261.20.29
3–018625.118625.118825.50.98
3–123031.115821.316622.5<0.0001
3–213418.112516.912617.10.80
3–39012.210714.49512.90.41
1512317.411616.210815.30.58
 
Vomiting
119125.819926.918324.80.66
319626.521829.421429.00.40
3–0395.3364.9364.90.92
3–113217.8496.6516.9<0.0001
3–2567.6516.9527.00.87
3–3283.8354.7324.30.67
15395.5344.8233.30.12
 
Breast tenderness
133745.534045.934847.20.81
318424.916021.616922.90.32
3–012617.010814.610914.80.35
3–19512.88912.08211.10.59
3–28811.98811.97910.70.71
3–38912.08711.77610.30.53
1513218.713018.213118.60.99
 
Fatigue
130741.527036.428438.50.13
326235.424833.526335.60.63
3–011415.411115.010614.40.85
3–114018.913418.112817.30.73
3–213418.113317.913818.70.93
3–311715.813818.612617.10.36
1519327.417724.816923.90.34
 
Dizziness
118625.117223.217523.70.67
316121.814219.215120.50.40
3–0486.5486.5456.10.94
3–1679.1618.2527.00.36
3–2527.0567.6709.50.19
3–3527.0608.1628.40.59
1510014.28511.99213.00.49
 
Headache
112617.013418.112316.70.76
313618.413718.514119.10.93
3–0486.5385.1446.00.53
3–1385.1375.0364.90.97
3–2334.5344.6324.30.97
3–3253.4263.5283.80.91
1512918.314019.615421.80.24
 
Fainting
1223.0131.8131.80.18
3253.4233.1334.50.33
3–081.150.740.50.46
3–1121.691.250.70.24
3–2152.070.9101.40.21
3–3162.281.1111.50.24
15253.5192.7213.00.64
Table 2.  Number and percentages of women with the drug-related side effects, diarrhoea, fever and rash, by treatment group and time periods (1 = at admission, 3 = between mifepristone and misoprostol, 3–0 = immediately after misoprostol administration, 3–1 = 1 hour after misoprostol, 3–2 = 2 hours after misoprostol, 3–3 = 3 hours after misoprostol, 15 = between misoprostol visit and day 15).
Timing (day–hour)GroupP(2)
O/OV/OV-only
No. of cases (n = 740)(1)%No. of cases (n = 741)(1)%No. of cases (n = 738)(1)%
  • (1)

    Day 15: n= 705 for O/O, n= 715 for V/O and n= 706 for V-only.

  • (2)

    With Bonferroni correction for simultaneous inferences for 11 side effects and time points, results are significant at α= 0.05 if P < 0.0007.

Diarrhoea
1415.5476.3527.00.49
3435.8405.4506.80.52
3–091.291.2101.40.96
3–1506.8131.881.1<0.0001
3–2527.0304.0273.70.01
3–3425.7233.1202.70.01
1518926.818826.3638.9<0.0001
 
Fever
3–081.140.570.90.50
3–1263.570.9131.80.0019
3–2354.7344.6324.30.93
3–3273.6456.1425.70.08
15182.6212.9162.30.72
 
Rash
3–020.310.110.10.78
3–160.840.520.30.37
3–291.240.550.70.31
3–360.860.840.50.78
1530.4101.420.30.02
Table 3.  Number and percentages of women with lower abdominal pain by treatment group and time periods (1 = at admission, 3 = between mifepristone and misoprostol, 3–0 = immediately after misoprostol administration, 3–1 = 1 hour after misoprostol, 3–2 = 2 hours after misoprostol, 3–3 = 3 hours after misoprostol, 15 = between misoprostol visit and day 15).
Timing (day–hour)GroupP(2)
O/OV/OV-only
No. of cases (n= 740)(1)%No. of cases(n= 741)(1)%No. of cases (n= 738)(1)%
  • (1)

    Day 15: n= 705 for O/O, n= 715 for V/O and n= 706 for V-only.

  • (2)

    With Bonferroni correction for simultaneous inferences for 11 side effects and time points, results are significant at α= 0.05 if P < 0.0007.

Lower abdominal pain
120928.221729.321128.60.90
332243.531242.131743.00.86
3–024332.818124.417323.4<0.0001
3–151369.348966.046162.50.02
3–253271.955574.954173.30.42
3–349166.454573.554073.20.0027
1539355.739455.136852.10.35

Upon admission, on average, about 60% of the women reported nausea, 26% vomiting, 46% breast tenderness, 39% fatigue, 24% dizziness, 17% headache and 2% fainting, similarly for the three groups. About 6.3% of the women reported diarrhoea. Fever and rash were not assessed at admission. The mean systolic blood pressure was 111 mmHg (SD between 11 and 12) and the mean diastolic blood pressure was 70 mmHg (SD between 8 and 9) in the three groups. The mean pulse rate was 78 beats/minute (SD between 8 and 9) in the three groups. About 29% of the women reported lower abdominal pain.

The percentages of women with pregnancy-related symptoms are shown in Table 1, by treatment group and time periods (admission, between mifepristone and misoprostol, immediately after misoprostol administration and after 1, 2, 3 hours, and between misoprostol and day 15).

There was a consistent decrease in the three groups in the occurrence of all these symptoms after misoprostol, when the abortion process had started. The time effect was highly significant for all the symptoms (P < 0.0001), except for headache (P= 0.6065). In the case of breast tenderness, most of the decrease seems to take place after mifepristone and before misoprostol. These time changes were similar between treatment groups (all the interactions Group by Time were non-significant), except for vomiting, for which the occurrence was higher 1 hour after misoprostol in the O/O group (interaction Group by Time was almost significant, P= 0.0512). The number of events of fainting was too small to draw any conclusions.

For all pregnancy-related symptoms except vomiting, there was no difference among the groups on average over the time periods (all P > 0.30). From mifepristone until misoprostol administration, on average 60% (1325/2219) of the women had nausea and 28% (628/2219) had vomiting, and the frequencies of the other pregnancy-related symptoms were also maintained with respect to admission levels, except for breast tenderness, for which the frequency was reduced by half.

At 1 hour after administration of misoprostol, the occurrence of nausea was 31% (230/740) in the O/O group, 21% (158/741) in the V/O group and 23% (166/738) in the V-only group. There was a threefold increase in the incidence of vomiting in the O/O group compared with the other two groups: 18% (132/740) of the women had vomiting in the O/O group, 7% (49/741) in the V/O group and 7% (51/738) in the V-only group. On the other hand, the occurrence of women with breast tenderness, fatigue, dizziness, headache and fainting was very similar in the three groups.

At 2 and 3 hours after misoprostol, the frequency of all pregnancy-related symptoms was similar between the groups.

The percentages of women with drug-related side effects are shown in Table 2, by treatment group and time periods.

The occurrence of the drug-related side effects varied significantly at the different time points (P < 0.0001 for the time effect for diarrhoea and fever, P= 0.0061 for rash), but presented a different pattern.

No effect of mifepristone was observed on the incidence of diarrhoea, which was maintained at admission levels, similarly in the three groups. Only 1% of women in the three groups reported diarrhoea immediately after misoprostol. One hour after administration of misoprostol, the incidence of diarrhoea increased fivefold in the oral group, and was maintained 2 and 3 hours later, while in the vaginal groups the increase was slighter, and happened only later, 2 hours after misoprostol. At the follow up visit on day 15, 27% (189/705) of the women in the O/O group and 26% (188/715) in the V/O group reported having had diarrhoea since day three, while 9% (63/706) did in the V-only group (the interaction Group by Time was highly significant, P < 0.0001).

There was a rise in the incidence of fever after administration of misoprostol, the rise being higher in the vaginal groups (up to 6% at 3 hours vs 4.7% in the O/O group at 2 hours) (The interaction Group by Time was significant, P < 0.0001).

The percentage of women with rash was very low, less than 1.2% in all groups at all times except in the V/O group in day 15, when it reached 1.4% (P= 0.3493 for the difference between groups, P= 0.1662 for the interaction Group by Time).

The treatment groups did not differ in the frequency of blood pressure changes on day three. An increase in systolic blood pressure >10 mmHg or in diastolic blood pressure >5 mmHg occurred in 15.5% of women for the three groups combined at 1 hour after misoprostol, in 17.6% at 2 hours and in 18.6% at 3 hours after misoprostol. 25.5% of women had a decrease in systolic blood pressure of more than 10 mmHg or in diastolic blood pressure of more than 5 mmHg at 1 hour, 26.8% at 2 hours and 27.4% at 3 hours.

There were two cases with periorbital oedema reported soon after administration of misoprostol on day three, one in the O/O group and the other in the V/O group.

The percentages of women with lower abdominal pain at different time points are shown in Table 3 and Fig. 1, by treatment group.

Figure 1.

Percentages of women with lower abdominal pain, by treatment group and time periods (1: at admission, 3: between mifepristone and misoprostol, 3-0: immediately after misoprostol administration, 3-1: 1 hour after misopostrol, 3-2: 2 hours after misoprostol, 3-3: 3 hours after misoprostol, 15: between misoprostol visit and day 15).

The incidence of lower abdominal pain increased after mifepristone to over 40%, similarly in the three groups. There was also an increase after misoprostol to values around 70% (P < 0.0001 for the time effect). The O/O group seemed to peak at 1 and 2 hours after misoprostol, reaching 69%, while the two vaginal groups did so at 2 and 3 hours after misoprostol, reaching 73% to 74% (the interaction Group by Time was significant, P= 0.0237). By day 15, the incidence of women reporting pain since the misoprostol visit was 54%.

There was no evidence of any effect of length of amenorrhoea on lower abdominal pain at any point in time (in all cases P > 0.30 from χ2 tests, results not shown), the incidence of pain being similar in the three length of amenorrhoea groups of less than 49 days, 50–56 days and 57 or more days.

In contrast, parity was significantly associated with a decrease in the frequency of lower abdominal pain at all recording points: the percentage of nulliparous and parous women reporting this symptom was respectively 35% (402/1134) and 22% (235/1085) at admission, 48% (539/1122) and 38% (412/1079) on day three, before misoprostol, 31% (352/1122) and 23% (245/1079) immediately after misoprostol, 77% (863/1122) and 56% (600/1079) 1 hour after misoprostol, 81% (912/1120) and 66% (716/1079) 2 hours after, 77% (856/1118) and 67% (720/1078) 3 hours after, 66% (709/1082) and 43% (446/1042) on day 15 (all P < 0.0001).

The most common treatment given for side effects was treatment for pain. There were major differences between the centres in the use of pain-killing medication. In some centres, like in Stockholm, Helsinki and Oslo, pain killers were given almost routinely, whereas in some centres women received hardly any pain-killing medication. During the first hour after administration of misoprostol, 19.4% (431/2219) of the women received pain-killing medication, mainly paracetamol, and during the second and third hours the percentages were 9.2% (204/2219) and 9.8% (217/2219), respectively. Between day three and day 15 visits, 6.9% (147/2126) needed medication for lower abdominal pain.

There were 206 women who came for unscheduled visits, some of them more than once. Among 222 unscheduled visits, vaginal bleeding was mentioned as one of the main reason(s) in 138 cases and lower abdominal pain in 60 cases. At unscheduled visits a total of 31 (15.0%) women were hospitalised. Treatment for heavy bleeding was the main reason for hospitalisation in 26 out of 31 women, and two of these women received blood transfusion. Initiation of treatment for a gynaecological infection was started in five women in the ward. During the study, 13 women received antibiotic therapy for a suspected upper gynaecological infection. There was one case of suspected deep venous thromboembolism treated six weeks after abortion.

Among the 1998 women who answered the question, 1676 (84%) would choose medical abortion in the future should the need arise, 174 (9%) would choose surgical abortion and 148 (7%) did not know.

Eighty-seven percent (865/1000) of parous women would choose medical abortion in the future compared with 81% (811/998) of non-parous women (P= 0.0062), the association being maintained when adjusted for all other baseline variables (P= 0.0201).

The ethnic groups differed significantly in the choice for future method (P < 0.0001): 88% of other Asian women (there were no African centres in this study, the few black women in different centres were included in this category) would choose medical abortion, 85% of Caucasians and 78% of Chinese. However, this association was not present when adjusting for the other baseline factors (P= 0.2796).

There was no association of treatment group, age, history of previous pregnancies, history of previous induced abortions and duration of pregnancy with preference for medical abortion, neither crude nor adjusted (in all cases P > 0.05).

Preference was significantly associated with the outcome of medical abortion: 85% of women with complete abortion would choose medical abortion again, and only 37% of those with failures (P < 0.0001). Adjusting for baseline variables did not change this result. The duration of bleeding among women with complete abortion was associated with preference, although the percentages of women preferring medical abortion were very high in both groups: higher, 93.6% (1052/1124), among women with duration of bleeding less than 15 days, than among women with duration of bleeding 15 days or more, 90.5% (571/631) (P= 0.0180). This association just failed to be significant when adjusted (P= 0.0764).

Preference was not associated with the occurrence of lower abdominal pain or fever on day three, neither crude nor adjusted.

Of the total of 1998 women, 645 (32%) women were of the opinion that the given medicines could be taken by the woman herself, without supervision by medical professionals, 1148 (58%) disagreed and 205 (10%) did not know.

The reasons most often cited for taking the medicines by oneself were that it is easy, and it does not cause any problems. The reasons most often cited for not taking the medicines by oneself were concern about safety, fear of possible problems and side effects.

There was an association of ethnic group with the opinion on taking medicines without supervision: 41% (297/719) of Caucasians thought that the medicines could be taken without medical supervision, 34% (207/618) of other Asian women and 21% (137/647) of Chinese (P < 0.0001). This result was maintained when adjusted.

Women older than 30 years were more frequently positive towards taking medicines by themselves (38%, 184/485) than younger women (between 33% and 29%: 82/249, 213/684, 166/580 in the age groups 20 or less, 21–25 and 26–30, respectively) (P= 0.0016). This association was maintained when adjusted (P= 0.0348).

There was no association of treatment group, history of previous pregnancies, history of previous induced abortions, parity and duration of pregnancy with the opinion on taking medicines without supervision, neither crude nor adjusted (in all cases P > 0.05).

The outcome of the present medical abortion was associated with the opinion on taking medicines without supervision: 33% (636/1936) of women with complete abortion and 15% (9/62) of women with failures were of the opinion that the medicines could be taken by themselves (P= 0.0066 crude, P= 0.0055 adjusted).

The opinion on taking medicines without supervision was not associated with duration of bleeding, the occurrence of lower abdominal pain or fever on day three, neither crude nor adjusted.

Of the total of 1995 women, 451 (23%) would choose to have a possible future abortion at home, 1400 (70%) at a health facility and 144 (7%) did not know.

The reasons most often cited for preferring to have a medical abortion at home were the easiness of the method, privacy and convenience. The main reasons for preferring to have a medical abortion at a health facility were safety, need of support and fear of complications.

Nineteen percent (192/994) of non-parous women and 26% (259/1001) of parous women would prefer abortion at home (P < 0.0002). This association was not present when adjusted.

Medical abortion at home is preferred by 25% of both Caucasians (177/717) and of other Asian women (156/618) and 18% (116/646) of Chinese women (P < 0.0001 crude, P= 0.0523 adjusted).

Women in different age groups differed about the choice of where to have medical abortion: those older than 30 years were more frequently inclined to home abortion (31%, 148/485) than younger women (only 15% in the age groups 20 or less, 21% in the age group 21–25 and 22% in the age group 26–30) (P < 0.0001 crude, P= 0.0018 adjusted).

Twenty-six percent (187/728) of women with previous induced abortions and 21% (264/1267) of women with no previous induced abortions would choose home abortion (P= 0.0178). Nineteen percent (128/687) of women with no previous pregnancies and 25% (323/1308) of women who had had previous pregnancie(s) would choose home abortion (P= 0.0005).

The outcome of the present medical abortion, duration of pregnancy, occurence of abdominal pain and fever did not have an effect on the future choice on where to have medical abortion (both crude and adjusted).

DISCUSSION

In this study, we compared the side effect profiles of oral and vaginal administration of misoprostol, and investigated the effect of continuing the initial administration with oral misoprostol administration. Our results show that the pregnancy-related symptoms decrease when the abortion process has started, after misoprostol administration (nausea, vomiting, fatigue, dizziness) or even after mifepristone (breast tenderness). At 1 hour after administration of misoprostol, the occurrence of nausea and vomiting was higher in the oral group than in the vaginal groups. This is explained by nausea and vomiting being also misoprostol-related side effects.11

The drug-related side effects that we found to occur were diarrhoea, fever and rash. At 1, 2 and 3 hours after administration of misoprostol, diarrhoea was more frequent in the oral group than in the vaginal groups. At the two-week follow up visit, 26% of women who continued with oral misoprostol twice daily for one week reported having suffered from diarrhoea, compared with only 9% of women in the group which took placebo for a week. This finding is expected, as diarrhoea is a known side effect of misoprostol used for prevention of gastric ulcer.11

There was an increase in the occurrence of fever after misoprostol administration on day three, peaking earlier in the oral group compared with the two vaginal groups. The incidence of fever up to 6.1% after misoprostol in our study is within the lower range compared with findings from previous studies where repeated misoprostol has been used to treat first trimester missed abortion and in second trimester medical abortion, where a rise in temperature over 38.0°C has been reported in 5–32% of women.12,13

There was also an increase in rash, but this was at very low levels and the power to draw conclusions is low. In a registry of adverse reactions to drugs, several cases of rash and allergic reactions and some cases of anaphylaxis have been reported with misoprostol use.14 The registry covers both long-term and once-only use and for different indications. In our study of >2000 women, we found a low incidence of rash, ≤1.2% within 3 hours after misoprostol use.

An increase in systolic blood pressure >10 mmHg or in diastolic blood pressure >5 mmHg during the 3-hour time period after misoprostol occurred in less than one-fifth of the women, and a decrease of the same amount in approximately one-fourth of the women, which is in line with prostaglandins having both vasodilator as well as vasoconstrictor effects. Changes in blood pressure can also be related to the abortion process itself; pain may cause an increase, and a vasovagal reaction—caused by passage of the products of conception through the cervical canal—a decrease in blood pressure. In previous studies, no significant changes in blood pressure and pulse after administration of misoprostol have been noted.1,15

The occurrence of lower abdominal pain immediately after and 1 hour after misoprostol was higher in the oral than in the vaginal groups, whereas at 3 hours it became more frequent in the vaginal groups. However, the oral administration of misoprostol/placebo took place first, whereafter misoprostol/placebo was administered vaginally, after which the side effects were recorded. The delay between oral administration and the ‘immediately after misoprostol’ recording could thus be as long as 10 minutes, which may count for the higher frequency of lower abdominal pain already at this point with oral administration of misoprostol.

After oral administration of misoprostol, the peak plasma concentration is reached already in 14–20 minutes,11 and the absorption is faster and the peak plasma concentrations are higher than after vaginal administration.16 Uterine contractility starts earlier after oral administration, but it is more regular and lasts longer after vaginal administration.16 These differences in the pharmacokinetics and uterine contractility well explain the different side effect profiles after oral and vaginal administration of misoprostol.

The use of pain-killing medication varied a lot among the centres. After administration of misoprostol, some centres used pain-killing medication almost routinely, while in some other centres no women requested any. Bleeding and pain were the most common reasons for unscheduled visits—10% had an unscheduled visit, and 1% of all women were taken into ward for treatment of bleeding or infection.

We recorded the symptoms, signs and side effects at several time points. However, the recording periods were of different length: Day one recording covered the ongoing pregnancy until mifepristone administration, whereas day three first recording, at arrival, covered the time since mifepristone administration. The recording periods immediately after misoprostol and hourly until 3 hours later were shorter. The likelihood for a side effect or symptom to occur is greater during a longer period than during 1 hour, thus the comparisons between frequencies at different recording points are subject to this limitation. Also, the recordings after misoprostol were made only until 3 hours after administration, therefore we do not have complete information thereafter, but only the report at day 15 visit about the whole period elapsed since the misoprostol visit.

At the six-week follow up visit, we evaluated women's perceptions of the method. Most women, 84%, would choose medical abortion again if there was a future need for abortion. Our finding is in line with other studies showing that the majority of women who have undergone medical abortion would choose it again should the need arise.17,18 Parous women were more favourable towards medical abortion than non-parous women.

Success of the current abortion process was strongly associated with the future choice of method: 85% of women with complete abortion would choose medical abortion again, whereas only 37% of those who had a failure would do so. This coincides with findings from other studies.17

Contrary to previous studies,19 duration of pregnancy and woman's age did not seem to play a role in the future choice of method in the present study. The occurrence of pain on day three was not associated with the future choice.

As home abortion has lately been proposed, we asked these women after having their experience of medical abortion in the clinic, which place they would prefer to have a possible future medical abortion, at home or in the clinic. Contrary to USA studies reporting satisfaction with home use of misoprostol,9,20 our study showed that the majority of women, 70%, would prefer to be in a clinic for a possible future medical abortion. The main reasons for this were safety, support needed and fear of possible complications.

Older and parous women, in line with findings from other studies,9 and women with previous induced abortion(s) were more positive towards home abortion.

Duration of pregnancy did not have an effect on the future choice of method, on whether to take the medicines by oneself or in the place where abortion is to be conducted.

CONCLUSIONS

Most women reporting the pregnancy-related symptoms nausea, vomiting, breast tenderness, fatigue, dizziness and headache have these symptoms alleviated after the abortion process is started. Oral administration of misoprostol is associated with higher frequency of nausea and vomiting at 1 hour after administration, and with higher frequency of diarrhoea at every hour during the 3 hours after drug intake, than vaginal administration. Frequency of fever after misoprostol increases up to 5% of women at 2 hours with oral administration and up to 6% at 3 hours with vaginal administration. The occurrence of lower abdominal pain increases to similar levels with the three regimens, peaking at 72% at 2 hours with oral administration and at 73% to 75% with vaginal administration. Continuation of misoprostol for one week increases the incidence of diarrhoea by more than threefold, from 9% to 26–27%. Should a need arise, most women would choose medical abortion again, and would prefer clinic to home abortion.

Acknowledgements

The study was funded by the UNDP/UNFPA/WHO/World Bank Special Programme of Research, Development and Research Training in Human Reproduction, WHO, Geneva, Switzerland.

Statement of responsibility and of conflict of interest

Overall responsibility for this paper lies with Dr H. von Hertzen of the UNDP/UNFPA/WHO/World Bank Special Programme of Research, Development and Research Training in Human Reproduction, World Health Organization. Neither she nor the Special Programme or any of its co-sponsors have a conflict of interest.

World Health Organization Research Group on Post-Ovulatory Methods for Fertility Regulation

Manuscript prepared by: H. Honkanen, G. Piaggio, H. von Hertzen.

Study design and coordination: H. von Hertzen, H. Honkanen, G. Piaggio.

Data coordination and statistical analysis: G. Piaggio, A. Peregoudov, M. Vucurevic.

Principal investigators and centres

S. C. Wu (National Research Institute for Family Planning, Beijing, China); S. Gopalan, S. Lyall, K. Gulati (Postgraduate Institute of Medical Education and Research, Chandigarh, India); H. Honkanen, O. Heikinheimo (Department of Obstetrics and Gynaecology, Helsinki University Central Hospital, Helsinki, Finland); N. T. N. Ngoc, N. T. B. Nga (Hungvuong Hospital, Ho Chi Minh City, Viet nam); O. S. Tang, P. C. Ho (Queen Mary Hospital, Hong Kong, China); A. Pretnar-Darovec, N. Vojnovic (University Department of Obstetrics and Gynaecology, Ljubljana, Slovenia); R. S. Shah, S. Kalgutkar, P. Anjaria, V. Walvekar (Institute for Research in Reproduction, I.C.M.R., Mumbai, India); S. Mittal, S. Kumar, S. Agarwal, A. Batra (All India Institute of Medical Sciences, New Delhi, India); F. Jerve, B. Nesheim, R. Baldursdottir (Department of Obstetrics and Gynecology, Ulleval University Hospital, Oslo, Norway); S. Song, M. L. Qian, F. Y. Wei (Shanghai Institute of Planned Parenthood Research, Shanghai, China); R. N. V. Prasad (National University Hospital, Singapore, Singapore); K. Gemzell-Danielsson, L. Marions, L. Helström, M. Bygdeman (Division for Obstetrics and Gynaecology, Karolinska Hospital, Stockholm, Sweden); G. Bártfai, A. Keresztúri (University of Szeged, Faculty of General Medicine, Albert Szent-Györgyi Medical and Pharmaceutical Centre, Department of Obstetrics and Gynecology, Szeged, Hungary); M. Horga, C. Enciulescu, S. Bela, F. Prundaru, I. Cozos (Centre of Public Health, Targu Mures, Romania); R. Erdenetungalag, B. Enkhbileg, M. Bayalag, J. Radnaabazar (Department of Human Genetics and Human Reproduction, Maternal and Child Health Research Centre, Ulaanbaatar, Mongolia).

Accepted 12 February 2004

Ancillary