Comparing medical versus surgical termination of pregnancy at 13–20 weeks of gestation: a randomised controlled trial

Authors


T Kelly, Research Midwives Office, Level 6 Leazes Wing, Royal Victoria Infirmary, Queen Victoria Road, NE1 4LP, UK. Email teresa.kelly@ncl.ac.uk

Abstract

Please cite this paper as: Kelly T, Suddes J, Howel D, Hewison J, Robson S. Comparing medical versus surgical termination of pregnancy at 13–20 weeks of gestation: a randomised controlled trial. BJOG 2010;117:1512–1520.

Objective  To compare the psychological impact, acceptability and clinical effectiveness of medical versus surgical termination of pregnancy (TOP) at 13–20 weeks of gestation.

Design  Randomised trial.

Setting  Large UK tertiary centre.

Sample  Women accepted for TOP at 13–20 weeks of gestation.

Methods  Medical TOP (MTOP) using mifepristone and misoprostol or surgical TOP (STOP) by vacuum aspiration at <15 weeks of gestation, and by dilatation and evacuation at 15 or more weeks of gestation.

Main outcome measures  Distress 2 weeks after TOP, measured by the impact of events scale (IES), and acceptability, measured by the proportion of women who would opt for the same procedure again.

Results  One hundred and twenty two women were randomised: 60 to the MTOP group and 62 to the STOP group. Twelve women opted to continue their pregnancy. Follow-up rates were low (n = 66/110; 60%). At 2 weeks post-procedure there was no difference in total IES score between groups. However, compared with women undergoing STOP, women undergoing MTOP had a higher score on the IES intrusion subscale (mean difference 6.6; 95% CI 1.4–11.8), and a higher score on the general health questionnaire (GHQ) (= 0.033). Women found STOP more acceptable: compared with MTOP, more women would opt for the same procedure again (100% versus 53%, ≤ 0.001), and fewer women found the experience to be worse than expected (0% versus 53%, = 0.001). Women who had MTOP experienced more bleeding (= 0.003), more pain on the day of the procedure (= 0.008), and more days of pain (= 0.020). Of the 107 women who declined to participate, 58 (67%) preferred a STOP.

Conclusions  Randomised trials of women requesting midtrimester TOP are challenging. Women found STOP less painful and more acceptable than MTOP.

Introduction

Termination of pregnancy (TOP) is the most commonly performed gynaecological procedure in England and Wales: in 2008, a total of 195 296 procedures were performed, of which 18 990 (9%) were performed after 12 weeks of gestation, and with the majority taking place between 13 and 19 weeks of gestation.1

Up to 15 weeks of gestation surgical termination of pregnancy (STOP) can usually be achieved by vacuum aspiration (VA). After this, fetal size precludes complete aspiration, and dilatation and evacuation (D&E) is necessary.2 Currently D&E accounts for 5% of all terminations in England and Wales.1 For medical termination (MTOP) after 12 weeks of pregnancy, the Royal College of Obstetricians and Gynaecologists (RCOG) recommends mifepristone followed 36–48 hours later by either misoprostol or gemeprost.2 Both surgical and medical methods appear safe and effective, although the RCOG only recommends D&E when ‘undertaken by specialist practitioners with access to the necessary instruments and who have a sufficiently large caseload to maintain their skills’.2

Eight studies have compared the safety and efficacy of medical and surgical abortion.3,4 Most of these were in women below 13 weeks of gestation. In 1980 Grimes et al.5 undertook a randomised control trial (RCT) of midtrimester abortion by intra-amniotic injection of prostaglandin F2a versus D&E; however, installation techniques are rarely used today. Grimes et al.6 subsequently undertook a pilot RCT comparing MTOP using mifepristone and misoprostol with D&E in women between 13 + 6 and 19 + 6 weeks of gestation. The trial was abandoned after enrolling only 18 women in 12 months because most women were unwilling to accept randomisation to MTOP.

Thus, the aim of this RCT was to compare psychological outcomes and acceptability of MTOP and STOP in women undergoing abortion between 13 and 20 weeks of gestation. Based on one prior non-randomised comparison of D&E and MTOP,7 we hypothesised that compared with STOP, MTOP would be associated with greater psychological distress at 2 weeks after the procedure, as measured by the impact of events scale (IES). A comparison of effectiveness and acceptability was also undertaken.

Methods

An RCT of MTOP versus STOP between 13 + 0 and 19 + 6 weeks of gestation was conducted in a tertiary teaching hospital (Royal Victoria Infirmary, Newcastle upon Tyne, UK) between May 2000 and February 2004 (46 months). The trial was approved by the Newcastle and North Tyneside Local Research Ethics Committee: ref. no. 2000/63.

Women accepted for TOP under clause C of the human Fertilisation and Embryology Act (1990) amendment of the Abortion Act (1967), with pregnancies between 13 + 0 and 19 + 6 weeks of gestation at the time of abortion were eligible for inclusion. Women under 16 years of age were eligible for inclusion if deemed Fraser competent by the clinical practitioner,8 and where a parent or guardian was present and also willing to give written consent. Women with a history of previous caesarean section were not excluded. Gestational age was assessed by ultrasound measurement of the fetal biparietal diameter (BPD) at the time of the outpatient assessment (usually <7 days before the TOP). Exclusion criteria were fetal congenital abnormality and a medical disease precluding MTOP. Women who were unable to speak English were also excluded (<5% of women presenting for TOP).

After a decision had been made to offer TOP by the responsible health professional, eligible women were given a patient information sheet. Those expressing an interest in participating in the study were seen by the research nurse, and any questions they had at that time were answered. Women giving informed written consent were randomised to MTOP or STOP by choosing the next sequentially numbered, sealed and opaque envelope. Envelopes were prepared based on a computer-generated randomisation table by a medical statistician who was not involved in the clinical care of the women. Women who declined involvement were asked to state their reasons for not participating. Because of funding constraints the research nurse was unable to attend all TOP assessment clinics. Baseline characteristics including age, previous TOP and gestation were recorded. Following randomisation, participants were asked to complete the general health questionnaire (GHQ-12)9 and the Hospital Anxiety and Depression Scale (HADS).10

Women randomised to STOP were admitted to the gynaecological day unit on the morning of the procedure. All nulliparous women and multiparous women at 17 or more weeks of gestation were primed with Gemeprost 1 mg vaginally 3 and 6 hours prior to the anticipated time of the STOP. Multiparous women between 13 + 0 and 16 + 6 weeks were primed with Gemeprost 1 mg vaginally 3 hours prior to the anticipated time of STOP.2 All STOPs were performed under general anaesthesia by one experienced surgeon (SCR). Gestational age was confirmed on the operating table prior to cervical dilatation by ultrasound measurement of the BPD. Cases between 13 + 0 and 13 + 6 weeks of gestation were progressively dilated to 13 mm using Hegar graded cervical dilators. Vacuum aspiration was performed using a 12-mm aspiration curette. Cases between 14 + 0 and 14 + 6 weeks of gestation were dilated to 15 mm and vacuum aspiration was performed using a 14-mm aspiration curette. Complete evacuation of products of conception was confirmed by ultrasound and, where necessary, residual products were removed with sponge forceps under ultrasound guidance. Cases of 15 + 0 or more weeks of gestation were progressively dilated to a cervical diameter in mm equivalent to the gestational age in weeks using Hegar graded cervical dilators. Products of conception were removed using Sopher’s forceps under ultrasound guidance. Routine perioperative uterotonic agents were not used. Intravenous oxytocin (5 units) was administered in cases with persistent post-evacuation bleeding (n = 2).

Women randomised to MTOP were given mifepristone 200 mg orally. They were asked to attend the gynaecological day unit at 0800 hours 36–48 hours later. Misoprostol 800 μg was administered vaginally followed by 400 μg vaginally or orally (depending on level of vaginal bleeding) every 3 hours up to a maximum of four doses.2 If abortion had not occurred by midnight a further dose of mifepristone 200 mg orally was administered, followed by gemeprost 1 mg vaginally 3-hourly from 0800 hours up to a maximum of five doses. If abortion had still not occurred by 0800 hours the following morning the MTOP was deemed to have failed and STOP was arranged. In all cases expelled products were examined to confirm that they were complete. Where this was uncertain, an ultrasound scan was performed to confirm the uterus was empty. Cases where the placenta was retained after expulsion of the fetus (confirmed by speculum examination) without significant bleeding were managed with a further dose of prostaglandin. Failure to pass the placenta after a further 4 hours was managed with surgical evacuation of retained products. All women received periabortion antibiotic prophylaxis with doxycycline 100 mg orally twice daily, commencing on the day prior to abortion. Women undergoing STOP also received metronidazole 1 g rectally at the time of abortion.

All women in the trial were invited to return to the hospital 14 days after the TOP for follow-up. At this visit women were questioned about bleeding and pain. If vaginal bleeding had not stopped, or women complained of persistent pain, an ultrasound scan was performed.

The primary outcome measure was the IES at 2 weeks after the procedure.11 This 15-item scale, which has seven intrusion (IES-I) and eight avoidance (IES-A) items, has been widely used as a measure of stress reactions after traumatic events, and has convergent validity with observer-diagnosed post-traumatic stress disorder.12 Items reflect symptom occurrence in the past week and are rated 0 (not at all), 1 (seldom), 3 (sometimes) and 5 (often), resulting in subscale scores ranging from 0 to 30 (IES-I) and from 0 to 35 (IES-A). Participants were instructed to relate the items to the TOP.

Secondary outcomes included:

  • • General health questionnaire with 12 items (GHQ-12) score: the GHQ-12 has been widely used as a screening instrument for psychological distress in the general population.9 All items have a four-point scoring system that ranges from a ‘better/healthier than normal’ option, through a ‘same as usual’ and ‘worse/more than usual’ to a ‘much worse/more than usual’ option, depending on the particular nature of the item. Each item was scored using the GHQ method (0-0-1-1), yielding a score ranging from 0 to 12.
  • • HADS score: the HADS has been widely used to detect symptoms of anxiety and depression in a variety of patients.10 It contains 14 questions (seven related to anxiety and seven related to depression), each rated 0–3, yielding scores for anxiety (HAD-A) and depression (HAD-D) ranging from 0 to 21.
  • • Physical symptoms during and after the procedure: information was collected on the duration of vaginal bleeding, heaviest bleeding after discharge (as compared with a menstrual period), duration of pain and maximum pain during TOP or after discharge (as assessed by a 10-cm visual analogue scale, VAS).
  • • Complications (including haemorrhage, defined as an estimated blood loss of >500 ml),13 failure, uterine evacuation for retained products of conception and pelvic infection (defined by pain and/or bleeding with a temperature of >37.5°C).
  • • Acceptability (as assessed by the proportion of women who would have the same procedure again if they ever had to have another abortion, and the proportion of women who rated the procedure better or worse than expected).
  • • Satisfaction with care (as assessed using the satisfaction with care scale, SCS).14 Women were asked to rate their satisfaction with three components of abortion care (information and counselling prior to the TOP, care during the TOP and counselling and support after the TOP) by indicating whether care was: poor, fair, good, very good or excellent.

Outcomes were recorded using a questionnaire administered at the time of the 2-week review. All complications were confirmed by an examination of the hospital notes and/or by telephone discussion with the general practitioner. Women who failed to attend the follow-up clinic and gave consent to postal contact were sent the questionnaire by post; 11 women who failed to attend follow-up did not consent to be contacted after the TOP.

The trial was designed to demonstrate a 5-point difference in the IES, assuming a SD of 8.15 With a power of 90%, and a two-sided type-1 error of 0.05, the sample size was estimated to be 54 per group. Allowing for a 20% loss to follow-up, we aimed to recruit a total of 130 women.

An intention-to-treat analysis was performed. Statistical analysis was undertaken using stata v9. Comparisons of mean and proportions were made across intervention groups with 95% CI for these differences. For GHQ, HAD-D and HAD-A, analysis of covariance was used to estimate the difference between mean scores at 2 weeks post-TOP while adjusting for the same scales at baseline. As some of the distributions appeared to be non-normal, these comparisons were made using bootstrap procedures.

Results

During the study period 720 women were eligible to participate based on ultrasound measurement of BPD in the assessment clinic (Figure 1). Despite assigning women likely to be more than 12 weeks pregnant (based on their last menstrual period) to a specific clinic, only 229 (32%) of the eligible women were approached because of the limited availability of the research nurse. One hundred and seven women (47%) declined randomisation and 122 women were randomised in the trial: 60 to the MTOP group and 62 to the STOP group.

Figure 1.

 CONSORT flowchart.

Of those who declined to be randomised 87 cited the principle reason as a strong preference for one procedure; 57 (67%) wanted a STOP. The reasons cited were: wanting to be asleep (n = 37); perception that a STOP was psychologically less traumatic (n = 20); perception that a STOP was less painful (n = 2); positive previous experience of STOP (n = 2); and shorter time to wait (n = 2). Twenty-nine women (33%) had a strong preference for MTOP. The reasons cited were: not wanting a general anaesthetic (n = 14); a shorter time to wait (n = 12); and a positive previous experience of MTOP (n = 3).

Twelve women subsequently decided to continue their pregnancies and were therefore only included in the baseline data comparisons: this left 52 women in the MTOP group and 58 in the STOP group. Only 66 women either attended the 2-week follow-up clinic or returned the questionnaire by post (30 in the MTOP group and 36 in the STOP group). Compared with subjects who did (n = 65) provide data on the primary outcome, those that did not (n = 47) were less likely to be in their first pregnancy (32% versus 54%, = 0.013) but otherwise were similar in terms of age, history of previous TOP, gestation at TOP, baseline GHQ and HAD scores (P > 0.55 for all comparisons). Data on complications and unexpected overnight stays were available from hospital notes for all 110 women who underwent TOP.

The baseline characteristics of participants are shown in Table 1. The distributions appeared to be similar in the MTOP and STOP groups. Psychological and preference outcomes 2 weeks after the procedures are shown in Table 2. The mean IES-I score was higher in the MTOP group (difference = 6.6, 95% CI 1.4–11.8), indicating that women had more intrusive thoughts about the procedure. Mean IES-total and IES-A scores were higher in the MTOP group, but these differences were not statistically significant, and the wide 95% CI showed that the data was consistent with quite large differences in either direction. Mean GHQ-12 scores post procedure were also higher after MTOP (adjusted difference 2.6, 95% CI 0.7–4.4); however, the small differences in mean HAD-A and HAD-D scores were not statistically significant. Compared with the MTOP group, far more women in the STOP group would opt to have the same procedure again (100 versus 53%, 95% CI for difference 29–65%), and far fewer women in the STOP group commented that the procedure was worse than expected (0 versus 53%, 95% CI for difference 35–71%).

Table 1.   Baseline characteristics
VariableMedical TOP
n = 60*
Surgical TOP
n = 62**
  1. *GHQ and HAD data missing for 12 women.

  2. **GHQ and HAD missing for five women.

Mean age (years)
(SD)
23.9
(6.3)
23.5
(5.8)
Mean gestation (weeks)
(SD)
14.7
(1.6)
15.1
(1.9)
Mean GHQ-12
(SD)
4.5
(3.9)
5.3
(3.6)
Mean HAD-A
(SD)
8.3
(4.3)
9.6
(3.9)
Mean HAD-D
(SD)
5.9
(4.2)
7.0
(3.7)
Primapara (n)2429
Previous TOP (n)1421
Previous CS (n)31
Table 2.   Psychological and preference outcomes at 2 weeks after treatment
VariableMedical TOPSurgical TOPDifference95% CI
  1. *Results at 2 weeks after treatment adjusted for those at baseline.

Mean IES-I18.6 (n = 28)12.0 (n = 36)6.61.4 to 11.8
Mean IES-A18.2 (n = 29)17.3 (n = 36)0.94.0 to 5.8
Mean IES-T36.8 (n = 28)30.1 (n = 35)6.7−2.5 to 15.9
Mean GHQ-124.4 (n = 29)2.1 (n = 35)2.6*0.7 to 4.4
Mean HAD-A6.3 (n = 29)6.5 (n = 36)0.0*−1.8 to 1.8
Mean HAD-D5.0 (n = 29)3.9 (n = 35)0.9*−1.1 to 2.9
Would choose same method again (%)53% (n = 30)100% (n = 26)−47%−29% to −65%
Experience of TOP worse than expected (%)53% (n = 30)0% (n = 26)53%35% to 71%

Complication rates and reported symptoms are shown in Table 3. Overall complication rates were similar in the two groups (six in the MTOP group and seven in the STOP group). The types of complications differed between groups. Five women in the MTOP group underwent evacuation for retained products of conception (ERPC) and one woman had a haemorrhage that required a blood transfusion. Five women in the STOP group had an estimated blood loss >500 ml, but none required blood transfusion, one further woman in the STOP group had a cervical laceration that required suturing and one required ERPC. A total of 16 women in the MTOP group were unexpectedly required to stay overnight compared with none in the STOP group (difference 31%, 95% CI 18–43%): of those requiring an overnight stay, nine women passed products late in the evening or early the following morning, five needed surgical ERPC, two women required further observation, one for heavy bleeding the other after a fainting episode.

Table 3.   Complications and symptoms
VariablesTrial armWould opt for same procedure again
MedicalSurgicalDifference95% CIYes (n = 42)No (n = 14)Difference95% CI
Any complications (%)11.5 (n = 52)12.0 (n = 58)−0.5−13 to 121477−13 to 27
ERPC (%)10 (n = 52)2 (n = 58)8−1 to 1714212−7 to 31
Unexpected overnight stay (%)31 (n = 52)0 (n = 58)3118 to 43437369 to 63
Bleeding severity > menstrual period (%)37 (n = 30)4 (n = 26)3314 to 5250123810 to 66
Pain severity moderate or severe (%)43 (n = 30)23 (n = 26)20−4 to 44502921−8 to 51
Mean pain duration44 (n = 30)1.9 (n = 26)2.50.4 to 4.52.74.82.0−0.7 to 4.8
Mean VAS score6.8 (n = 30)4.6 (n = 26)2.20.7 to 3.85.08.02.91.7 to 4.1

Although a similar proportion of women were still bleeding after 2 weeks in both groups (MTOP n = 14; STOP n = 11), after MTOP more women described their bleeding as heavier than a menstrual period (37 versus 4%, 95% CI for difference 14–52%). The duration of heavy bleeding was also longer in the MTOP group (difference in means 2.2 days, 95% CI 0.4–4.5). Pain after TOP was rated as moderate or severe more often in the MTOP group, but this difference was not statistically significant (43 versus 23%, 95% CI for difference −4 to 44%). When asked to rate the worst pain on the day of procedure VAS scores were higher in the MTOP group (difference between means 2.2, 95% CI 0.7–3.8).

Compared with women who would opt to have the same procedure again, women who would not had higher mean pain scores on the day of procedure (difference 2.9, 95% CI for difference 1.7–4.1), more unexpected overnight stays (difference 36%, 95% CI for difference 9–63%) and more women stated bleeding was more than experienced during their menstrual period (difference 38%, 95% CI for difference 10–66%). There were no differences in complication rates or duration of pain.

Satisfaction ratings are shown in Table 4. There were no differences in satisfaction ratings for information and counselling pre-procedure, care during procedure, and counselling and support post-procedure between the two groups; between 98 and 100% of women were satisfied with their care in all three categories.

Table 4.   Satisfaction ratings
VariablesMedical
n = 30
n (%)
Surgical
n = 36
n (%)
Info and counselling pre-TOP
Satisfied29 (100)35 (97.2)
Not satisfied0 (0)1 (2.8)
Care during TOP
Satisfied29 (100)35 (97.2)
Not satisfied0 (0)1 (2.8)
Counsel and support post-TOP
Satisfied28 (96.6)35 (100)
Not satisfied1 (3.4)0 (0)

Discussion

This is the second RCT comparing MTOP and STOP after 13 weeks of gestation using a currently recommended MTOP regime.2 The first trial performed by Grimes et al.6 was abandoned after 12 months because women were unwilling to be randomised. We were unsuccessful in recruiting and following up the intended sample size, underscoring the challenges of studying this group of women. The conclusions that can be drawn from our results are therefore limited, and the possible improvement in psychological outcome associated with STOP needs confirming in an adequately powered study. Women did seem to prefer STOP. This may be because of the heavier bleeding and greater pain and/or the increased frequency of an unplanned overnight stay with MTOP. However, the majority of women were satisfied with their care. These findings suggest that a patient-centred abortion service should offer the choice of MTOP or STOP after 13 weeks of gestation.

The study has a number of limitations. The proposed sample size of 130 women was not achieved. Funding constraints meant that recruitment could not be extended beyond 46 months. During this time the research nurse was only able to approach 32% of suitable women, and, of this group, 38% declined randomisation because of a preference for one method: a problem also experienced by Grimes et al.6 in their abandoned RCT of midtrimester termination methods. Women’s reluctance to have their method of TOP randomly assigned is not confined to late abortion: Creinin only managed to enrol fewer than 10% of suitable women to an RCT of MTOP versus STOP before 49 days of gestation. More recent RCTs of first-trimester abortion methods have incorporated preference arms: although two studies from Aberdeen reported randomisation rates of 54 and 82%,16,17 only 19% of women recruited to our recently completed randomised preference trial of medical versus surgical TOP before 14 weeks of gestation agreed to randomisation.18 Furthermore, primary outcome data was obtained in only 60% of our subjects, a figure lower than we predicted and less than reported in prior trials of earlier TOP: in the two trials conducted in Aberdeen, 2-week outcome data was obtained in 56 and 96% of women, whereas Rørbye et al. achieved an 81% response rate in Denmark.14,16,19 More women might have contributed outcome data if additional methods of data collection had been employed, e.g. telephone/text contact and/or web-based questionnaires: incorporating these methods resulted in primary outcome data being available from 74% of 1877 recruits to our recent trial of TOP methods before 14 weeks of gestation.18

We hypothesised that MTOP would be associated with greater psychological distress, and therefore chose the IES as the main outcome measure. The IES was designed to study the psychological impact of a specific ‘trauma’. Total IES scores were similar in the two groups, although mean scores on the intrusion subscale were higher after MTOP. Consistent with the hypothesis that midtrimester STOP may be associated with better short-term psychological outcomes, mean GHQ-12 scores were higher in the MTOP group. Where the differences between groups were not significant, the relatively small sample size meant that the data were consistent with quite large differences in both directions. There have been three RCTs that have compared psychological outcomes after MTOP and STOP.20–22 All were undertaken in the first trimester and none reported any differences in psychological outcome, whether assessed during the first 4 weeks after the procedure,20,21 or at 2 years.22 Only one of these studies included the IES, and the authors reported that, within the MTOP group, women who saw the fetus were more likely to experience intrusive events (nightmares, unwanted thoughts and images) than those who did not. As the fetus is more likely to be seen and recognised at later gestations, this may account for the higher IES intrusion subscale scores in the current study. However, we cannot test this hypothesis as whether women saw the fetus or not was not recorded.22 The greater pain experienced during and after the MTOP procedure may also have contributed to women’s distress. Mean IES scores in the present study were higher than those reported by Broen et al.23 10 days after STOP performed prior to 13 weeks of gestation (IES-I 11.9, IES-A 11.1). Interestingly, in their study IES fell progressively, reaching mean values of 3.6 (IES-I) and 8.3 (IES-A) at 5 years after the procedure.23

Approximately one in ten women experienced a complication. This rate is higher than encountered in prior RCTs of early gestation methods,17,24,25 underscoring that TOP is safer the earlier in pregnancy it is performed.1,6 The most common complications encountered in this study were RPC after MTOP requiring evacuation (9.6%) and haemorrhage at the time of STOP (6.9%). Ashok et al.26 reviewed 1002 consecutive cases of MTOP at 13–21 weeks of gestation using a similar regime: rates of ERPC (7.4%) and transfusion (0.7%) were comparable with those reported here. Grimes et al.27 reviewed their experience with D&E in 6213 terminations between 13 and 20 weeks of gestation, and reported rates of haemorrhage, RPC, cervical injury and uterine perforation of 0.7, 0.9, 1.2 and 0.3%, respectively. In their 10–13 weeks of gestation RCT, Ashok et al. reported that only one woman (0.5%) required ERPC after MTOP, whereas five (2.1%) women required ERPC in the STOP group. In this study the consultant carried out 66% of the STOP procedures. Expertise performing STOP can impact on efficacy.28 All surgical procedures in the current study were performed by one consultant operator, with experience of more than 1000 STOPs after 14 weeks of pregnancy, and ultrasound was used to ensure complete uterine evacuation. These two factors are likely to have contributed to the low rate of RPC in the surgical group. Overnight stay was required in 30% of the MTOP group, a similar rate to that reported by Ashok et al.17 Although women undergoing an MTOP were warned about this possibility, many women commented that this was a significant problem. Although some primiparous women in the STOP group required an overnight stay, these were all planned admissions for cervical priming prior to morning surgery. These can be completely avoided with an afternoon operating list.

Pain experienced during and after the procedure was greater in the MTOP group, as was heavy bleeding after discharge. It was not possible to assess the duration of bleeding as a substantial number of women were still bleeding at the time of follow-up, albeit their loss was declining. These results are consistent with previous studies of earlier abortion,17,25 although Creinin24 found no differences before 7 weeks of gestation. However, there is evidence that pain related to MTOP increases as gestation increases.25,26 There are, however, no studies comparing bleeding or pain after discharge in midtrimester abortion.

The use of a simple question (i.e. if you ever had to have another termination would you have the same method again?) to explore preference for abortion method has been used in several previous studies.6,16,24,29 It has the advantage that responses can easily be determined through a variety of data collection methods (including sms text messaging. Based on this measure, more women preferred the STOP. Consistent with this finding, no woman in the STOP group rated the procedure as ‘worse than expected’, compared with nearly one-half of the MTOP group. Specific aspects of the MTOP procedure that contributed to the reduced preference for MTOP were unexpected overnight stay, heavy bleeding and more pain on the day of procedure than STOP. Prior studies have indicated that seeing the fetus,30 pain, prolonged bleeding and possible psychological consequences16,29 contribute to dissatisfaction with MTOP and influence future preference. It is also noteworthy that more women who expressed a preference prior to the procedure preferred STOP. Although the commonest reason cited was a desire to be asleep (which probably relates to a desire to avoid pain and/or seeing the fetus) a significant number of women commented that they felt MTOP was psychologically more traumatic.

Women who attend for midtrimester abortion have a range of complex factors that contribute to their delay in accessing services, e.g. ambivalence about their pregnancy and the decision to terminate,31 and relative delays at all stages of the referral pathway.32 The proportion of women from vulnerable groups, e.g. teenage pregnancies, also increases as gestational age increases.1 Despite the available evidence that STOP in experienced hands is at least as safe as MTOP,27 and the evidence presented here that many women prefer STOP, access to midtrimester surgical abortion within the NHS remains limited.33 Although most (71%) NHS hospitals provide facilities for abortion after 13 weeks of gestation, the majority (79%) only provide MTOP.33 In contrast, whereas only 11% of abortion providers are from the private sector, 88% offer STOP.33 The number of midtrimester abortions undertaken in NHS hospitals has been falling since 1992. In 2008, although 94% of abortions between 13 and 20 weeks of gestation were NHS funded, 64% were carried out in the independent sector (under NHS contract), and most of these were performed surgically (76%).1 An increasing number of gynaecologists are opting out of providing STOP, especially after 12 weeks of gestation,34 thereby severely compromising training opportunities within the NHS: the RCOG Advanced Training Skills Module (ATSM) in Abortion Care has a compulsory element covering MTOP at early and late gestations, but only up to 14 weeks of gestation for STOP.35 If women are to be offered midtrimester surgical abortion in the future, there is an urgent need to introduce novel training strategies, including making optimal use of the skills available within the independent sector.

Conclusion

We found MTOP to be associated with more pain and bleeding, and also more unexpected overnight stays. It may also be associated with more distress 2 weeks after the procedure, but larger trials are needed to confirm this observation. More women would prefer STOP if requiring another termination in the future; however, some women would choose MTOP to avoid general anaesthesia and surgery. This study suggests that women should be offered a choice of method to allow for personal preferences.

Disclosure of interests

There are no conflicting interests to disclose.

Contribution to authorship

TK completed the first draft and further revisions, undertook initial analysis and agreed the final version. JS was responsible for the clinical facilitation of the project and agreed the final version. DH was involved in the initial design, undertook final data analysis and agreed the final version. JH was involved in the initial design, contributed to the psychological interpretations and agreed the final version. SR was responsible for the conception and design, interpretation of data, revisions and agreed the final version.

Details of ethics approval

Ethical approval was obtained from the Joint Ethics Committee of the Newcastle and North Tyneside Health Authority on 26 April 2000 (ref. no. 2000/63). Trial registration number: ISRCTN17262711.

Funding

This trial was undertaken with funding from the University of Newcastle upon Tyne.

Acknowledgements

The authors would like to thank all the women who participated in the study and the nursing staff from the gynaecology ward and day-case theatre who supported the study at the Royal Victoria Infirmary, Newcastle upon Tyne.

Ancillary