The influence of body weight on response to ovulation induction with gonadotrophins in 335 women with World Health Organization group II anovulatory infertility
Dr AH Balen, Department of Obstetrics and Gynaecology, Leeds General Infirmary, Leeds LS2 9NS, UK. Email firstname.lastname@example.org
Objective To assess the influence of body weight on the outcome of ovulation induction in women with World Health Organization (WHO) group II anovulatory infertility.
Design The combined results of two studies in which either a highly purified urinary follicle-stimulating hormone or highly purified urinary menotrophin were compared with recombinant follicle-stimulating hormone.
Setting Thirty-six fertility clinics.
Population A total of 335 women with WHO group II anovulatory infertility failing to ovulate or conceive on clomifene citrate.
Methods Ovarian stimulation using a low-dose step-up protocol.
Main outcome measures The effects of body weight on ovarian response, ovulation rate and pregnancy rate after one treatment cycle.
Results With increasing body mass index (BMI), a higher threshold dose of gonadotrophins was required and there were more days of stimulation; yet, despite a greater concentration of antral follicles, there were fewer intermediate and large follicles. There was no difference in the rates of ovulation and clinical pregnancy in relation to body weight.
Conclusions Body weight affects gonadotrophin requirements but not overall outcome of ovulation induction in women with anovulatory polycystic ovary syndrome and a BMI of less than 35 kg/m2.
The majority of women with World Health Organization (WHO) group II anovulatory infertility have polycystic ovary syndrome (PCOS), of whom at least 40–50% are overweight.1,2 Obesity has a significant effect both on menstrual irregularity, spontaneous ovulation and the likelihood of responding to ovulation induction therapy.3 Standard algorithms use clomifene citrate initially. If clomifene therapy is unsuccessful, ovarian stimulation with follicle-stimulating hormone (FSH) preparations in low-dose step-up protocols has been shown to be successful in inducing ovulation in about 85% but producing monofollicular development in only one-half of women.4,5 A multifollicular response increases the risks of multiple pregnancy and ovarian hyperstimulation syndrome (OHSS). It can be difficult to achieve the threshold for unifollicular response rather than over-response as the polycystic ovary can be extremely sensitive to stimulation.
We have reported the results of two large, multicentre ovulation induction studies which were designed to compare different FSH preparations in low-dose step-up protocols in a total of 335 women with anovulatory PCOS (A. H. Balen et al., unpubl obs.).6 We first compared highly purified urinary FSH (HP-FSH, Bravelle®; Ferring Pharmaceuticals A/S, Copenhagen, Denmark) (A. H. Balen et al., unpubl obs.) and recombinant FSH (follitropin alfa, GONAL-F®; Serono, Geneva, Switzerland) and second, compared highly purified urinary menotrophin (HP-hMG, Menopur®; Ferring Pharmaceuticals A/S) with recombinant FSH (follitropin alfa, GONAL-F®; Serono).6 Both studies demonstrated similar ovulation rates (85–91%), continuing pregnancy rates (14–21%) and singleton live birth rates (14–15%) with the use of different preparations in clomifene-citrate-resistant women.
The total number of women recruited into the two studies was 335—one of the largest ovulation induction data sets. We therefore pooled the data from these two studies in order to assess the influence of obesity on the outcome of ovulation induction. We examined the data set to determine any effect that body weight might have on outcome, with respect to rate of ovulation, number of follicles, total dose of gonadotrophin and threshold dose.
Materials and methods
Anovulatory WHO group II women who had failed to ovulate or conceive on clomifene citrate were recruited at a total of 36 fertility centres (13 in Belgium, 9 in Denmark, 5 in Sweden and 9 in the UK). The inclusion criteria were: (i) women with good physical and mental health, aged between 18 and 39 years, who failed to ovulate with clomifene citrate doses of at least 100 mg/day for at least 5 days or failed to conceive after three cycles of ovulation induction with clomifene citrate; (ii) WHO group II infertility with chronic anovulation (amenorrhoea or oligomenorrhoea, or based on progesterone levels in women with cycles of duration of 21–35 days); (iii) infertility for ≥1 year before randomisation; (iv) body mass index (BMI) of 19–35 kg/m2 at the time of randomisation; (v) at least one patent tube documented within 3 years prior to screening; (vi) a normal pelvis documented by a transvaginal ultrasound scan with respect to uterus, fallopian tubes and ovaries within 3 months prior to screening; (vii) early follicular serum FSH levels between 1 and 12 iu/l, levels of prolactin and total testosterone not suggestive of hyperprolactinaemia or androgen-secreting tumours; (viii) a male partner with a normal semen analysis or semen from a donor and (ix) signed informed consent form prior to screening. The population consisted of a heterogeneous mix of clomifene-resistant women and women who failed to conceive after three cycles of clomifene therapy. This heterogeneity will have been taken care of by the randomisation process.
The exclusion criteria included: (i) a history of ≥12 unsuccessful ovulation induction cycles; (ii) persistent ovarian cysts (≥15 mm in size) for more than one cycle or ovarian endometrioma on ultrasound scan; (iii) any significant systemic disease, endocrine or metabolic abnormalities (pituitary, thyroid, adrenal, pancreas, liver or kidney); (iv) use of any nonregistered investigational drug during the 3 months before screening or previous participation in the study and any concomitant medication that would interfere with the evaluation of the study medication (nonstudy hormonal therapy, except thyroid medication, antipsychotics, anxiolytics, hypnotics, sedatives and need for continuous use of prostaglandin inhibitors); (v) treatment with clomifene citrate, metformin, gonadotrophins or gonadotrophin-releasing hormone analogues within 1 month prior to randomisation; (vi) pregnancy, lactation or contraindication to pregnancy; (vii) current or past (last 12 months) abuse of alcohol or drugs; (viii) a history of chemotherapy (except for gestational conditions) or radiotherapy (ix) undiagnosed vaginal bleeding; (x) tumours of the ovary, breast, adrenal gland, pituitary or hypothalamus; malformation of sexual organs incompatible with pregnancy and (xi) hypersensitivity to any trial product.
Details of the study design may be found in the respective publications (A. H. Balen et al., unpubl obs.).6 With respect to the ovulation induction protocol, essentially stimulation was started 2–5 days after a spontaneous or progesterone-induced menstrual bleed. The starting dose of gonadotrophin was 75 iu daily, which was maintained for 7 days. After the first 7 days, the dose was either maintained or increased by 37.5-iu increments according to individual response. All subjects were maintained on their specific dose level for at least 7 days. The maximum allowed daily dose was 225 iu and subjects were treated with the gonadotrophin for a maximum of 6 weeks.
Gonadotrophin stimulation was maintained until at least one of the following criteria for human chorionic gonadotrophin (hCG) administration were met: one follicle with a diameter of ≥17 mm or 2–3 follicles with a diameter of ≥15 mm. Subjects were not given hCG in either of the following situations: no follicular response after 6 weeks of gonadotrophin treatment or four or more follicles with a diameter of ≥15 mm. Subjects who reached the hCG criteria received a single subcutaneous or intramuscular injection of hCG (Profasi; Serono, Switzerland) at a dose of 5000 iu to trigger ovulation. Subjects given hCG were recommended sexual intercourse or were planned for intrauterine insemination according to the standards at the investigational site; luteal support was prohibited. At least one blood sample was taken during the midluteal phase (6–9 days after hCG administration) and analysed for progesterone by a central laboratory. A quantitative pregnancy test (serum β-hCG) was taken 12–16 days after hCG administration. In case of pregnancy, a transvaginal ultrasound scan was performed in weeks 7 and 12 to confirm clinical and continuing pregnancy, respectively. All pregnancies were followed up to delivery.
The primary outcome was the rate of ovulation. Ovulation was defined as a midluteal serum progesterone concentration of ≥25 nmol/l (≥7.9 ng/ml), and the presence of a clinical pregnancy was considered to have been a successful ovulation, regardless of the progesterone level. Measurement of midluteal progesterone was performed by a central laboratory using a competitive immunoassay using direct chemiluminometric technology with a sensitivity of 0.48 nmol/l (Quest Diagnostics Limited, Heston, UK).
Other clinical parameters evaluated were positive β-hCG rate at 12–16 days after hCG administration, clinical pregnancy rate (transvaginal ultrasound scan showing at least one intrauterine gestation sac with fetal heart beat at 7 ± 2 weeks after hCG administration), continuing pregnancy rate (transvaginal ultrasound scan showing at least one viable fetus at 12 ± 2 weeks after hCG administration), live birth rate, singleton live birth rate, total number of follicles, number of subjects with monofollicular (one follicle ≥17 mm and no follicles of 15 or 16 mm in size) and bifollicular/multifollicular (≥2 follicles ≥15 mm in size) development, number of follicles of ≥12, ≥15 and ≥18 mm in size, endometrial thickness at the time of hCG administration and efficiency in terms of total gonadotrophin dose administered and duration of gonadotrophin treatment. The major safety endpoints were the incidence of OHSS (categorised as mild, moderate or severe according to Golan’s classification), multiple gestations and the number of cancellations due to risk of over-response.
BMI was considered in categories (<25, 25.1–30, >30.1 kg/m2) and as a continuous variable if appropriate.
The influence of BMI (kg/m2) was investigated unadjusted and adjusted for potential confounding factors. Adjustment for study, age, baseline total number of follicles (antral follicles) and serum FSH concentration were planned a priori. In addition, as an extra analysis, menstrual history (amenorrhoea, oligoamenorrhoea, cycle length 21–35 days) was added as it was seen to influence many of the outcome measurements. Furthermore, the investigation of endometrial thickness (mm) at the end of stimulation was adjusted for baseline endometrial thickness (mm) in addition to other factors. Interaction between BMI and age was investigated. There was no statistically significant interaction (all P values > 0.07) observed.
For binary outcome (i.e. yes/no to response), logistic regression models were applied. The influence of BMI is expressed as odds ratios with 95% confidence intervals and overall statistical significance tests. For the categorised analysis, ‘<25 kg/m2’ is selected as reference.
For continuous outcome (i.e. number of follicles, endometrial thickness, treatment days and total gonadotrophin dose), analysis of variance models (i.e. linear regression models) were applied. For the categorised analysis, ‘<25 kg/m2’ is selected as reference.
The analysis of threshold dose was based on a proportional odds polytomous logistic regression model.
The age of the women in the three BMI groups was similar (Table 1). Waist circumference increased significantly with increasing BMI (P < 0.001), but waist:hip ratio was not significantly different. The more obese women were relatively less likely to be amenorrhoeic (P= 0.011) than the normal weight women. There was also a correlation between antral follicle count and total follicle count with increasing body mass (P= 0.042 and P= 0.013, respectively), although ovarian volume was not different (P= 0.983).
Table 1. Demographic data compared by BMI (mean ± SD)
|Age (years)||29.1 ± 3.6||29.4 ± 4.2||28.3 ± 4.1||0.175||29.0 ± 3.9|
|BMI (kg/m2)||21.6 ± 2.0||27.6 ± 1.5||32.3 ± 1.9||<0.001||25.3 ± 4.7|
|Hip circumference (cm)||94.8 ± 7.4||105.6 ±7.9||117.4 ± 8.6||<0.001||102.2 ± 11.7|
|Waist circumference (cm)||75.8 ± 7.7||90.1 ± 10.4||98.6 ± 11.1||<0.001||84.2 ± 13.1|
|Waist:hip ratio||0.80 ± 0.1||0.86 ± 0.1||0.84 ± 0.1||<0.001||0.83 ± 0.1|
|Cycles 21–35 days||28.3%||28.0%||27.7%|| ||28.1%|
|Baseline ovarian morphology|
|Antral follicles >2 mm||18.7 ± 15.8||19.2 ± 14.0||24.3 ± 16.6||0.042||20.0 ±15.6|
|Total follicles||18.3 ± 15.7||17.6 ± 14.5||24.4 ±16.6||0.013||19.2 ±15.8|
|Mean ovarian volume (cm3)||7.7 ± 4.1||7.8 ± 4.2||7.9 ± 3.6||0.963||7.8 ± 4.0|
|Endometrial thickness (mm)||3.8 ± 2.0||4.0 ± 1.9||4.4 ± 1.9||0.120||4.0 ± 2.0|
There was no difference in baseline concentrations of FSH, luteinizing hormone or estradiol between the groups, but, as expected, with increasing obesity, there was an increase in fasting insulin concentrations (P < 0.001), insulin resistance, fall in sex-hormone-binding globulin (P < 0.001) and a concomitant increase in free androgen index (P < 0.001) (Table 2).
Table 2. Endocrine data (mean ± SD)
|FSH (iu/l)||5.6 ± 2.4||5.2 ± 2.4||4.8 ± 1.2||0.200||5.3 ± 2.2|
|Testosterone (nmol/l)||1.68 ± 0.57||1.82 ± 0.64||1.91 ± 0.66||0.025||1.76 ± 0.62|
|Sex-hormone-binding globulin||76 ± 42||46 ± 33||34 ± 20||<0.001||59 ± 40|
|Free androgen index||3.00 ± 2.23||6.03 ± 4.92||7.46 ± 5.04||<0.001||4.73 ± 4.22|
|Fasting glucose (mmol/l)||5.1 ± 0.6||5.0 ± 0.5||5.2 ± 0.7||0.729||5.1 ± 0.6|
|Fasting insulin (pmol/l)||62 ± 51||92 ± 112||104 ± 61||<0.001||77 ± 73|
|Insulin:glucose ratio||2.14 ± 1.52||3.63 ± 3.67||4.12 ± 3.48||<0.001||2.94 ± 2.82|
There was no significant influence of BMI on the rate of ovulation (P= 0.363) or pregnancy (as assessed by positive hCG, P= 0.596; clinical pregnancy rate, P= 0.781 and continuing pregnancy rate, P= 0.828), even after adjustment for study, age, baseline antral follicle count and serum FSH concentration (Table 3).
Table 3. Outcomes (percentage or mean ± SD)
|Days of stimulation||12.0 ± 4.4||13.9 ± 6.8||15.9 ± 8.1||13.3 ± 6.2|
|Total dose of gonadotrophin (iu)||1025 ± 512||1302 ± 899||1633 ± 1279||1220 ± 854|
|Threshold dose (iu)|
|Follicles of size (mm)|
|<12||12.8 ± 13.1||12.5 ± 12.2||18.8 ± 15.5||13.9 ± 13.5|
|12–16||1.6 ± 2.9||1.2 ± 2.4||0.6 ± 1.4||1.3 ± 2.6|
|17||1.3 ± 0.9||1.0 ± 0.6||1.1 ± 0.6||1.2 ± 0.8|
|Total follicles||15.7 ± 13.5||14.8 ± 12.3||21.0 ± 15.5||16.5 ± 13.7|
|Follicular development (%)|
|Endometrial thickness||8.7 ± 2.1||9.4 ± 2.4||10.0 ± 2.0||9.1 ± 2.2|
|Ovulation rate (%)||87||83||80||84|
|Positive hCG (%)||20||24||26||22|
|Clinical pregnancy rate (%)||17||20||18||18|
|Continuing pregnancy rate (%)||16||19||17||17|
The group with a BMI of greater than 30 kg/m2 produced more number of small follicles (P= 0.005) and fewer intermediate follicles (P= 0.036) than the less overweight and normal weight women, despite a higher antral follicle count (Table 4). After adjusting for study, age, baseline antral follicle count and serum FSH concentration, the relationship only remains with intermediate-sized follicles and even then the mean difference between the lightest and heaviest groups is only about one intermediate-sized follicle and the effect of a 1 kg/m2 BMI increase is a decrease of 0.09 intermediate-sized follicles. The influence of BMI on the number of follicles of 17 mm in size or more is not linear; however, there tends to be more number of larger follicles in the group with BMI < 25 kg/m2. There is also a tendency towards an increase in endometrial thickness with increasing BMI after adjusting for other factors (including baseline endometrial thickness). Adjustment for menstrual history, in addition to the other factors, did not affect the conclusions (data not shown).
Table 4. Ovarian response to stimulation related to BMI
|Follicles of size <12 mm at end of stimulation|
| <25 (reference)||0||0|
| 25–30||−0.29 (−3.66 to 3.08)||−0.09 (−2.54 to 2.36)|
| 30+||5.97 (2.16–9.79)||1.72 (−1.09 to 4.53)|
| ||P= 0.005||P= 0.441|
|Follicles of size 12–16 mm at end of stimulation|
| <25 (reference)||0||0|
| 25–30||−0.35 (−0.99 to 0.29)||−0.37 (−1.02 to 0.27)|
| 30+||−0.95 (−1.67 to −0.22)||−1.19 (−1.93 to −0.46)|
| ||P= 0.036||P= 0.007|
|Effect of 1 kg/m2 BMI increase||−0.06 (−0.12 to −0.01)||−0.09 (−0.15 to −0.03)|
| ||P= 0.029||P= 0.005|
|Follicles of 17 mm or more at end of stimulation|
| <25 (reference)||0||0|
| 25–30||−0.25 (−0.44 to −0.05)||−0.26 (−0.46 to −0.06)|
| 30+||−0.15 (−0.37 to 0.07)||−0.15 (−0.38 to 0.08)|
| ||P= 0.040||P= 0.031|
|Total number of follicles at end of stimulation|
| <25 (reference)||0||0|
| 25–30||−0.88 (−4.30 to 2.54)||−0.72 (−3.18 to 1.75)|
| 30+||5.26 (1.38–9.13)||0.77 (−2.06 to 3.60)|
| ||P= 0.012||P= 0.648|
|Endometrial thickness (mm)** at end of stimulation|
| <25 (reference)||0||0|
| 25–30||0.66 (0.11–1.21)||0.66 (0.11–1.21)|
| 30+||1.26 (0.64–1.88)||1.05 (0.41–1.68)|
| ||P < 0.001||P= 0.002|
|Effect of 1 kg/m2 BMI increase||0.11 (0.06–0.16)||0.10 (0.05–0.15)|
| ||P < 0.001||P < 0.001|
As can be seen in Tables 3 and 5, an increasing BMI is associated with more treatment days, a higher total dose and a higher threshold dose of gonadotrophins.
Table 5. Treatment days and dose of gonadotrophins related to BMI
| <25 (reference)||0||0|
| 25–30||1.85 (0.34–3.36)||1.80 (0.32–3.27)|
| 30+||3.84 (2.14–5.56)||2.91 (1.21–4.60)|
| ||P < 0.001||P= 0.001|
|Effect of 1 kg/m2 BMI increase||0.37 (0.24–0.51)||0.31 (0.17–0.44)|
| ||P < 0.001||P < 0.001|
|Total dose (iu)|
| <25 (reference)||0||0|
| 25–30||278 (70–485)||268 (65–471)|
| 30+||608 (373–843)||480 (247–714)|
| ||P < 0.001||P= 0.001|
|Effect of 1 kg/m2 BMI increase||57 (39–76)||49 (30–67)|
| ||P < 0.001||P < 0.001|
| 25–30||1.62 (0.96–2.76)||1.47 (0.84–2.55)|
| 30+||2.41 (1.35–4.32)||2.15 (1.17–3.94)|
| ||P= 0.009||P= 0.041|
We have shown that in women with anovulatory infertility, as expected, an increasing body mass is associated with worsening insulin resistance, an increased free androgen index and ovaries with more immature follicles. Obesity is associated with a significantly higher threshold dose for stimulation, a greater total dose of gonadotrophins required and a longer duration of stimulation. While this is also associated with the development of more number of small follicles, there were fewer large follicles and no overall difference in ovulation or pregnancy rates. These might result in a lower rate of multiple pregnancy, with an increased risk of OHSS, although this was not observed in our study.
The data set reported is one of the largest series of women undergoing ovulation induction and is strengthened by the uniformity of the protocol used, which was strictly monitored. Furthermore, the entry criteria only allowed women with a BMI of <35 kg/m2 so that those with extreme obesity were excluded. Indeed, the mean BMI was 25.3 ± 4.7 kg/m2.
While the findings reported are in keeping with what might be expected, it is interesting to note the differential response with respect to the size of the follicles. The use of gonadotrophin therapy for anovulatory infertility requires careful adjustment of dose in order to avoid over-response. Women who are overweight are both harder to monitor accurately by transvaginal ultrasound scan, and we have shown that they are at greater potential risk of over-response. There was a greater number of antral follicles in those who were overweight; yet, this translated to a smaller number of intermediate-sized and large-sized follicles. This is a further reflection of the differential response to ovarian stimulation.
At least 40–50% of women with PCOS are overweight,1,2 and those who are overweight are more likely to have menstrual cycle dysfunction and anovulatory infertility.2,7 Even moderate obesity, BMI > 27 kg/m2, is associated with a reduced chance of ovulation,8 and a visceral body fat distribution leading to an increased waist:hip ratio appears to have a more important effect than body weight alone.9,10 Obese women (BMI > 30 kg/m2) should be encouraged to lose weight in order to improve ovarian function.11–13 A study by Clark et al.11 looked at the effect of a weight loss and exercise programme on women with anovulatory infertility, clomifene resistance and a BMI > 30 kg/m2 and confirmed that weight loss had a significant effect on endocrine function, ovulation and subsequent pregnancy. An extension of this study, in women with a variety of diagnoses, demonstrated that in 60 out of 67 subjects, weight loss resulted in spontaneous ovulation with lower than anticipated rates of miscarriage and a significant saving in the cost of treatment.12
A reduction in body weight of 5–10% will cause a 30% reduction in visceral fat, which is often sufficient to restore ovulation and reduce markers for metabolic disease.14 Weight loss should be encouraged prior to ovulation induction treatments, as they appear to be less effective when the BMI is greater than 28–30 kg/m2.3 Others have also reported that more gonadotrophins are required to achieve ovulation in insulin-resistant women.15 Obese women being treated with low-dose therapy have inferior pregnancy and miscarriage rates.3 Both obese16 and insulin-resistant15 women with PCOS, even on low-dose FSH stimulation, have a much greater tendency to a multifollicular response and thus a relatively high cycle cancellation rate in order to avoid hyperstimulation.
National guidelines in the UK for the management of overweight women with PCOS advise weight loss, preferably to a BMI of <30 kg/m2 prior to commencing drugs for ovarian stimulation.17 Pregnancy carries significant risks for those who are obese with increased rates of congenital anomalies (neural tube and cardiac defects), miscarriage, gestational diabetes, hypertension and problems during delivery.18,19 Furthermore, pregnancy exacerbates underlying insulin resistance so women with PCOS have an increased risk of developing gestational diabetes.20 The potential risks of fertility treatment for women with anovulatory PCOS have recently been highlighted.21
Gonadotrophin preparations appear to be absorbed equally well by the subcutaneous and intramuscular route, irrespective of body mass,22 and so the difference in response is likely to be a true reflection of metabolic and endocrine differences. A meta-analysis of 13 studies confirmed a positive association between degree of obesity and amount of gonadotrophin required, with a weighted mean difference of 771 iu more needed (95% CI: 700–842) and also a higher rate of cycle cancellation in the obese women (pooled OR 1.86, 95% CI: 1.13–3.06).23 There was also a reduction in ovulation rate associated with obesity compared with nonobese women (OR 0.44, 95% CI: 0.31–0.61). While there was no difference in pregnancy rates associated with obesity, there was a negative association with insulin resistance (pooled OR 0.29, 95% CI: 0.10–0.80). Thus, the combination of obesity and insulin resistance appear to be the most significant determinants for the outcome of ovulation induction therapy, with degree of insulin resistance being more important.
This study confirms the effect of increasing body weight on ovarian response to gonadotrophin stimulation but also indicates that carefully conducted ovulation induction therapy can achieve satisfactory rates of ovulation and pregnancy in women with a BMI up to 35 kg/m2.
The study was sponsored by Ferring Pharmaceuticals A/S, Copenhagen, Denmark. Drs J.-C.A. and L.H. coordinated the design of the two randomised controlled studies and assisted with data collection, and Dr P.S. assisted with the statistical analyses.
We would like to thank all the participating centres:
Belgium: AZ-VUB, Brussels; Virga Jesse Ziekenhuis, Hasselt; AZ Groeninge, Kortrijk; CHR Citadelle, Liège; Hôpital Erasme, Brussels; Hôpital Saint Vincent, Rocourt; ZOL Campus St Jan, Genk; Centre Hospitalier Notre Dame, Charleroi; AZ St Lucas, Gent; Private Practice, Aalter; AZ Jan Portaels Campus Zuid, Vilvoorde; UZ Gasthuisberg, Leuven and Universitair Ziekenhuis, Gent.
Denmark: Copenhagen University Hospital; Brædstrup Hospital; Randers Hospital; Skive Hospital; Holbæk Hospital; Herlev Hospital; Hvidovre Hospital; Odense University Hospital and Skejby Hospital.
Sweden: Uppsala University Hospital; Sahlgrenska University Hospital, Gothenburg; Lund University Hospital; Karlstad Hospital and Helsingborg Hospital.
UK: Leeds General Infirmary, Leeds; Birmingham Women’s Hospital; St Michael’s Hospital, Bristol; Ninewells Hospital, Dundee; Glasgow Royal Infirmary; The Jessop Wing, Sheffield; Liverpool Women’s Hospital; Princess Anne Hospital, Southampton and Guy’s Hospital, London.