Bulletin of the World Health Organization

Comparative effectiveness, safety and acceptability of medical abortion at home and in a clinic: a systematic review

Thoai D Ngo a, Min Hae Park b, Haleema Shakur a & Caroline Free a

a. Department of Epidemiology and Population Health, London School of Hygiene and Tropical Medicine, 50 Keppel Street, London, WC1E 7HT, England.
b. Health System Department, Marie Stopes International, London, England.

Correspondence to Thoai D Ngo (e-mail: Thoai.Ngo@lshtm.ac.uk).

(Submitted: 28 November 2010 – Revised version received: 01 February 2011 – Accepted: 10 February 2011 – Published online: 04 March 2011.)

Bulletin of the World Health Organization 2011;89:360-370. doi: 10.2471/BLT.10.084046

Introduction

Medical abortion consists of using drugs to terminate a pregnancy. It is an important alternative to surgical methods.1 Although many different drugs have been used, alone and in combination, to induce abortion, a regimen composed of mifepristone plus misoprostol has been the one most widely used since mifepristone was first approved as an abortifacient in China and France in 1988.2 The World Health Organization (WHO) recommends this drug combination, with an initial dose of mifepristone followed by misoprostol 36 to 48 hours later, for early medical abortion.2 In 2005, an estimated 26 million women worldwide used this drug combination to terminate their pregnancies.3

In home-based medical abortion, a health-care provider administers mifepristone at the clinic and the pregnant woman later takes misoprostol at home. This protocol is intended to simplify the medical abortion regimen. Home-based medical abortion may improve the acceptability of medical abortion by allowing for greater privacy than in-clinic abortion, giving women greater control over the timing of abortion (anytime before the seventh week of pregnancy) and making it possible for family members or friends to be present to provide emotional support.4,5 Home-based medical abortion also reduces the number of clinic visits required, and hence the burden on women and services.5 In studies from France, Sweden, Tunisia and the United States of America, the majority of women opted for home-based medical abortion when offered the choice between home and clinic.3 Self-administration of misoprostol is already common in France and the United States.6

Medical abortion has been practiced to varying degrees across different settings. Despite this, whether home-based methods are as effective as clinic-based methods remains unclear. Studies evaluating regimens consisting of mifepristone and misoprostol in various combinations suggest that home-based medical abortion is effective and safe. Clinical trials from Canada,7 Turkey8,9 and the United States10,11 report rates of complete abortion ranging from 91% to 98% for pregnancies up to 9 weeks when misoprostol is administered at home (Appendix A, available at: http://www.mariestopes.org/documents/Home-based%20Medical-Abortion-Systematic-Review-Appendix.pdf). Observational studies1220 have also shown that home-based medical abortion is well accepted and effective, with 86–98% of women reporting satisfaction with the method and complete abortion achieved in 87% to 98% of cases. However, none of these studies has compared home-based medical abortion with clinic-based protocols. To fill this research gap, in this paper we review the evidence on the comparative effectiveness, safety and acceptability of medical abortion practised at home and in clinic.

Methods

Study selection

In this review we searched for published studies on home-based medical abortion that tested different drugs, routes of administration and doses or regimens. Randomized controlled trials (RCTs) and prospective cohort studies were eligible for inclusion; service evaluations, case series and review articles were excluded. The inclusion criteria were: (i) a comparison between home-based and clinic-based medical abortion;(ii) a prospective assessment of outcomes; and (iii) reporting of the primary outcome of interest.

Participants

Participants of interest were women of reproductive age (15–49 years) in resource-rich or resource-limited settings who were seeking an abortion.

Outcomes

The primary outcome was the proportion of successful abortions (i.e. complete evacuation of the uterine contents without surgery). Other outcomes of interest were acceptability and the development of side-effects, which included bleeding, vomiting, diarrhoea, fever, pain and infection. Mortality was expected to be low.2 There were three common measures of acceptability: satisfaction with the method, likelihood of choosing it again and likelihood of recommending it to a friend.21

Search strategy

We developed a search strategy based on search terms and filters used by the Cochrane Fertility Regulation Group22 (Table 1). Ovid MEDLINE (1950–December 2009), EMBASE (1980–2010), Popline (2004–2010) and the Cochrane Central Register of Controlled Trials (1990–2010) were searched electronically. The following web sites were hand searched for relevant publications: Marie Stopes International, Ipas, Gynuity, Population Council, the International Consortium for Medical Abortion and Google Scholar. Searching was limited to publications from 1990 or later. No limits were placed on language.

Validity assessment

Open-label trials and prospective cohort studies were eligible for inclusion. Studies were excluded if outcome data were collected retrospectively.

Study quality

Studies were assessed for quality based on a scale adapted from the Newcastle−Ottawa Scale. They were awarded points based on:Selection bias: “A” if same inclusion criteria for both study arms and “I” if different eligibility criteria for each study arm or if criteria unclear; Control for confounders: “A” if study controlled for gestational age in design/analysis (gestational age is an important confounder; previous reviews have indicated lower efficacy in more advanced pregnancies);23 “I” if no adjustment for confounders or this unclear. Assessment of gestational age: “A” if gestational age determined according to standard protocol (i.e. interview, bimanual exam and/or ultrasound); “I” if not assessed or inadequately assessed; and Adequacy of follow-up: “A” if all study participants accounted for or if ≤ 10% lost to follow up; “I” if no description of those lost to follow-up and if drop-out rate > 10%.To be categorized as high quality, studies had to score positively on selection bias, assessment of gestational age and adequacy of follow-up, as we felt that these three categories could have the most direct influence on the outcomes and study design.

  • Selection bias: “A” if same inclusion criteria for both study arms and “I” if different eligibility criteria for each study arm or if criteria unclear;
  • Control for confounders: “A” if study controlled for gestational age in design/analysis (gestational age is an important confounder; previous reviews have indicated lower efficacy in more advanced pregnancies);23 “I” if no adjustment for confounders or this unclear.
  • Assessment of gestational age: “A” if gestational age determined according to standard protocol (i.e. interview, bimanual exam and/or ultrasound); “I” if not assessed or inadequately assessed; and
  • Adequacy of follow-up: “A” if all study participants accounted for or if ≤ 10% lost to follow up; “I” if no description of those lost to follow-up and if drop-out rate > 10%.

Data abstraction

Two independent reviewers screened and extracted the data using a pre-designed form. A researcher who fluently spoke French and English translated the French-language papers. We made three attempts within one month to contact the authors of studies whose eligibility for inclusion depended on unpublished information.

We recorded the number of women recruited to each intervention group and the number of complete abortions. The drugs used, dose and route of administration were noted, along with each study’s inclusion and exclusion criteria. We used study participant characteristics at baseline and inclusion/exclusion criteria to qualitatively assess clinical heterogeneity.

Data synthesis

The principal measure of effect was the odds of having a successful medical abortion at home relative to the odds of having a successful medical abortion in clinic and the 95% confidence interval (CI) of this odds ratio (OR). We calculated the odds of having a successful abortion using the number of women recruited for each study and an intention-to-treat approach. We synthesized effectiveness in a meta-analysis, specifying a random-effects model to produce a pooled OR and CI. This model was selected a priori to incorporate the effect of trial heterogeneity among prospective studies from different settings. Statistical heterogeneity was assessed using χ2 tests and I2 statistics. The following sensitivity analyses were carried out: (i) separate analysis of high-quality studies to explore the effect of biases on study heterogeneity; (ii) analyses by maximum gestational age, and (iii) analyses by resource-rich versus resource-limited study setting.

We present a forest plot showing relative risks and 95% CIs for the primary outcome. Owing to the small number of studies included in the data synthesis, we did not assess publication bias. Analyses were carried out using Stata version 11 (StataCorp. LP, College Station, USA).

Results

Description of included studies

Nine studies were included in this review (Fig. 1) (one study was conducted in both Tunisia and Viet Nam and we present the findings separately for each setting). All were prospective cohort studies and included a total of 4522 participants (3478 home users, 1044 clinic users) (Table 2). The studies were carried out in Albania (n = 409), France (n = 714 women), India (n = 599), Nepal (n = 400), Tunisia (n = 518), Turkey (n = 208) and Viet Nam (n = 1674), between 199724–2008.25,31

Fig. 1. Summary of study selection for systematic review of studies comparing home-based and clinic-based medical abortion
Fig. 1. Summary of study selection for systematic review of studies comparing home-based and clinic-based medical abortion
a These studies involved medical abortion at home and in clinic but did not compare outcomes by home versus clinic status; data for this comparison were unavailable.

Study participants

The mean age of study participants ranged from 24.7 to 32.2 years. The maximum gestational age was 56 days in seven studies and 49 days in the two French studies. Gestational age was assessed by last menstrual period (LMP) and confirmed by clinical examination.

One Vietnamese study30 reported that home users were more educated and had a lower gestational age and higher gravidity on average, and that fewer of them were primigravidas when compared with clinic users. An Indian study reported that home users were 1.6 years older on average than clinic users (P = 0.008).25 The Tunisian study indicated that women using the home-based abortion method were more educated than those who opted for a clinic-based method.28 One study did not compare participant characteristics at baseline.27

Interventions

In all studies, oral mifepristone and misoprostol were used in combination to produce a medical abortion. In seven studies. 200 mg of mifepristone were used, while in the two French studies 600 mg were used.26,31 The time between mifepristone and misoprostol administration was 48 hours in seven studies, 36–48 hours in one French study31 and unspecified in another study.26 All studies used misoprostol 400 µg, given as two tablets of 200 µg where specified.9,24,26,28,30 In all studies, women were given mifepristone at the clinic but could choose between taking misoprostol at home or returning to take it at the clinic. All protocols detailed the use of painkillers (paracetamol, codeine or ibuprofen), which women were advised to take as needed. In seven studies women were followed up 2 weeks after mifepristone administration, while in one study they were followed up at 10–20 days.31

Six studies required that participants live or work within 1 hour of the study site.24,26,27,2931 In the French studies, women who lived farther than 1 hour away from the referral hospital were ineligible for the home-based protocol but were included in the study.

Study quality

The quality assessment based on four criteria described previously resulted in seven studies being categorized as high quality and two as low quality. Two studies scored 4/4,24,30 five studies scored 3/49,25,2729 and the two French studies scored 1/4.26,31 (Appendix B, available at: http://www.mariestopes.org/documents/Home-based%20Medical-Abortion-Systematic-Review-Appendix.pdf).

Complete abortion

Among the 3478 women who took misoprostol at home, the proportion who succeeded in having a complete abortion ranged from 86% in India25 to 97% in Albania.24 The average success rate in this group was 89.7% (95% CI: 88.7–90.7%) (Table 2). Among the 1044 women who took misoprostol in clinic, the success rate ranged from 80% in Turkey9 to 99% in France.26 The average success rate in this group was 93.1% (95% CI: 91.4–94.5%). The ORs for complete abortion at home versus in a clinic showed no difference in effectiveness in five studies.9,24,25,28,29 In three studies (two French,26,31 one Vietnamese)30 medical abortion in clinic settings was found to be more effective, while in the study conducted in both Tunisia and Viet Nam abortion at home proved more effective (OR: 2.9; 95% CI: 1.1–8.1).27 Pooled data from all nine studies showed no evidence of a difference in complete abortion rates (OR: 0.8; 95% CI: 0.5–1.5). However, study heterogeneity was high (I2: = 69.4%).

Subgroup analyses by study quality, maximum gestational age and study setting were equivalent, as the two studies from France also had low quality assessment scores and a maximum gestational age of 49 days (versus 56 days in the other studies). Pooled analysis of the seven high-quality studies showed no difference in complete abortion rates between women who took misoprostol at home (n = 3215) and those who took it in clinic (n = 593) (OR: 1.1; 95% CI: 0.7–1.9; I2: 62.0%) (Fig. 2). Pooled analysis of the findings of the two French studies indicated a higher rate of successful abortion among women who took misoprostol in clinic (n = 263 and 451) (OR: 0.2; 95% CI: 0.1–0.6).

Fig. 2. Forest plota comparing rates of complete abortion in women who underwent home-based and clinic-based medical abortion
Fig. 2. Forest plot<sup>a</sup> comparing rates of complete abortion in women who underwent home-based and clinic-based medical abortion
CI, confidence interval; OR, odds ratio.a Pooled data from all nine studies showed no evidence of a difference in complete abortion (odds ratio: 0.8; 95% confidence interval: 0.5–1.5).

Side-effects and complications

Side-effects were inconsistently reported across studies (Table 3). Pain and cramping were experienced by > 90% of women and lasted from 0.5 to 3 days.24 Pooled analysis of the mean number of days of pain from three studies indicated that pain lasted 0.3 days longer in women who took misoprostol at home (n = 1761) than in those who took misoprostol in clinic (n = 297) (weighted mean difference, WMD: 0.3 days; 95% CI: 0.1–0.5).9,24,30 Vomiting was reported by 12–34% of women24,29 and lasted 0.3 days longer in women who took misoprostol at home (n = 1761) than in women who took misoprostol in clinic (n = 297) (WMD: 0.32; 95% CI: 0.1–0.5). Fever also lasted longer in the former (n = 2058) than in the latter (WMD: 0.3 days; 95% CI: 0–0.6). Pooled analysis (n = 2058) showed no difference between home-based and clinic-based abortion in the reported duration of nausea (WMD: 0.3 days; 95% CI: −0.2 to 0.9), or the duration of heavy bleeding (WMD: 0.1 day; 95% CI: −0.1 to 0.4). In one study,30 women who took misoprostol at home were more likely to contact health services by phone or make unscheduled clinic visits (Table 2). In three other studies, women who took misoprostol in clinic were more likely to call clinic hotlines and make unscheduled visits than women who took misoprostol at home.9,27,29

Serious complications were rare. Four women had severe bleeding that required transfusion. Two of these women had taken misoprostol at home;29 where the other two women took the misoprosotol was not reported.25 Suspected infection requiring hospitalization occurred in one case,25 but where the woman took the misoprostol was not reported.

Acceptability

In reporting acceptability, most studies used the following criteria: satisfaction with the method (8 studies); the likelihood of choosing the method again (9 studies); and the likelihood of recommending medical abortion to a friend (4 studies) (Table 4). Of the women who underwent a home-based medical abortion (n = 3138), 84–99% were satisfied. Their average satisfaction rate was 88.4% (95% CI: 86.9–89.1). Among women who took misoprostol in clinic (n = 867), 72–97% were satisfied. Their average satisfaction rate was 85.6% (95% CI: 82.6–87.4). Pooled analysis showed no difference in satisfaction rates between women taking misoprostol at home or in clinic (OR: 1.46; 95% CI: 0.59–3.60; I2: 82.2%).

About 78–97% of home users (mean 94.4%; n = 3239) and 40–93% of clinic users (mean 61.6%; n = 963) stated that they would choose medical abortion again. Women who took misoprostol at home were seven times more likely to choose medical abortion again than women who took misoprostol in clinic (pooled OR: 7.1; 95% CI: 2.7–18.6), although heterogeneity was high (I2: 94%). Four studies reported the number of women who would recommend medical abortion to a friend (n = 1194).26,28,29,31 The pooled OR was 2.8 (95% CI: 0.5–17.3). Thus, no difference was found in this respect between women who took misoprostol at home and in clinic.

Discussion

Medical abortion is an important alternative to surgical methods for the termination of pregnancy. Our review is the first to systematically compare home-based to clinic-based medical abortion. Other reviews have compared medical abortion methods by regimen22,23 and gestational age.23 We have shown that the rate of complete abortion among women using home-based medical abortion across diverse study settings is high (~90%), and there is no evidence of a difference in effectiveness when compared to clinic-based protocols. The rate of complete abortion reported in our review is similar to the rates reported in other reviews. For example, the Cochrane Review of medical methods for first trimester abortion noted success rates of > 90% in all studies.22 Loss to follow-up was low in the included studies (4% in home-based groups, and 6% in clinic-based groups).

This review has limitations. Meta-analysis of the findings of non-randomized studies increases the possibility of biases,32 particularly self-selection bias stemming from the fact that women could choose between home-based and clinic-based medical abortion. As an illustration, women wanting to keep the procedure confidential and concealed from family members would be more likely than others to choose clinic-based abortion. Only one study reported the findings with adjustment for potential confounders.30 Our reported effect sizes are unadjusted and residual confounding and other biases are likely to have affected the estimates and study heterogeneity. Thus, the pooled estimates should be interpreted with caution.

Data from this review are limited to pregnancies no longer than 56 days. Eight of the nine included studies assessed gestational age from the date of the last menstrual period, which, according to a recent trial (n = 4484) that compared this method with pelvic bimanual and ultrasound, is effective in determining gestational age for early medical abortion.33 In all studies, participants were required to live or work within 1 hour of the study site and to be in good health, and in several studies they were required to have access to a telephone. However, there was no stipulation regarding the means of transportation for getting to the study site or the presence of another person at home in case of emergency. Women participating in studies were also screened for allergies to mifepristone and misoprostol. The effectiveness of home-based medical abortion in non-research settings without the precautionary measures and support systems that were most likely in place in these and other studies, essential for compliance with ethical norms, may be less satisfactory. From the data provided in this paper it is not possible to determine just how safe medical abortion practised at home would be if back up safety measures (e.g. easy access to a health facility, consultation by phone) were absent or more relaxed. Home-based medical abortion does not preclude prior screening for ectopic pregnancy, which is a standard of care even in resource-limited settings.34 Because of the inclusion of the two French studies, both categorized as low quality, this review was unable to provide robust estimates of the effectiveness of home-based medical abortion in developed settings.

Safety data from the included studies showed that of 3478 women who underwent abortion at home, two experienced heavy bleeding requiring transfusion. Thus, complications arising from use of misoprostol at home were rare. Two cases of heavy bleeding and one case of suspected infection were reported in the Indian study.25 However, the paper did not specify where the misoprostol was administered. If we assume that it was always administered at home the proportions of women affected (0.03% with infection, 0.1% with heavy bleeding) are comparable to the proportions reported in other reviews.3,35 Most women experienced pain and cramping after misoprostol administration. Women in the home-based groups reported experiencing pain and vomiting for slightly longer (0.3 days) than those in clinic-based groups, but they did not have more contact with health services. In the included studies, women who took misoprostol at home used self-report study cards to record any side-effects, while those who aborted in clinic were observed in the facility. Therefore, the safety data may be subject to reporting bias. Data on side-effects were also inconsistently reported across studies and this limits their generalizability.

Women who practised home-based medical abortion appeared satisfied and likely to choose the method again. Acceptability is subject to the influence of costs and convenience, data that were unavailable for the included studies. Furthermore, the included studies did not report on other factors, such as tolerance for bleeding and pain, that could have affected acceptability.

Our findings only apply to pregnancies up to 56 days and to the oral use of mifepristone–misoprostol. Data from our review cannot be generalized to settings where mifepristone is unavailable or where misoprostol is used in higher doses to induce abortion. It is also important to emphasize that the mifepristone-misoprostol regimen is not an alternative contraceptive method.

Implications

There is no evidence that home-based medical abortion is less effective, safe or acceptable than clinic-based medical abortion. Simplified protocols could give greater access to medical abortion to women living in restrictive and/or resource-limited settings where mortality related to unsafe abortion remains high.36 Adequate safety measures and support systems should be in place before home-based medical abortion can be offered. To further clarify the comparative effectiveness, safety and acceptability of home-based medical abortion, further studies should be conducted to explore different regimens, routes of administration and its use for gestational ages, as well as on the use of misoprostol only for home-based medical abortion, given the high cost of mifepristone and the fact that its use is restricted in many settings.


Acknowledgements

The authors thank Nathalie Kapp from the Department of Reproductive Health and Research of the World Health Organization for her critical review of the manuscript and Kristen Hopkins from Marie Stopes International (MSI) for translating the French-language articles. They also thank the authors of the original studies who provided additional information for this review.

Thoai D Ngo is also based within the Research and Metrics Team at MSI, and Min Hae Park is also affiliated with the Department of Epidemiology and Population Health of the London School of Hygiene and Tropical Medicine.

Funding:

The Research and Metrics Team at Marie Stopes International provided funding for this review.

Competing interests:

TDN and MHP work for Marie Stopes International, an organization that provides medical abortion procedures in the United Kingdom and globally. All authors have completed the Unified Competing Interest form at www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declare: (i) no financial support for the submitted work from anyone other than their employer; (ii) no financial relationships with commercial entities that might have an interest in this review; (iii) no spouses, partners or children with relationships with commercial entities that might have an interest in this review; (iv) no non-financial interests that may be relevant to this review.

References

Share