Dr A. Magos, Minimally Invasive Therapy Unit and Endoscopy Training Centre, University Department of Obstetrics and Gynaecology, Royal Free Hospital, Pond Street, Hampstead, London, NW3 2QG UK.
Objectives To evaluate triple tourniquets in controlled conditions and for the first time to investigate the hypothesis that leaving a semi-permanent tourniquet around the uterine artery reduces post-operative bleeding from the uterine incisions.
Design A randomised controlled trial.
Setting Two University teaching hospitals.
Population Twenty-eight patients with symptomatic fibroids and uterine sizes ranging from 14 to 24 weeks of gestation undergoing open myomectomy.
Methods A number 1 polyglactin suture was tied around the cervix to occlude the uterine arteries, and polythene tourniquets were tied around the infundibulopelvic ligament to obstruct the ovarian vessels. At the end of the procedure, the ovarian ties were released but the uterine artery suture remained in situ.
Main outcome measures Intra-operative blood loss, post-operative blood loss, blood transfusion rates, operative morbidity, uterine blood flow and ovarian function.
Results There was significantly less blood lost in the tourniquet group than in the control group (difference between means 1870 mL, 95% CI 1159–2580 mL, P < 0.0001; transfusion rates of 7% and 79%, P= 0.0003). The volume in the pelvic drain 20 min post-operatively and after 48 hours failed to reach statistical significance between the two groups (P= 0.10 and P= 0.165). There were no differences in uterine artery Doppler resistance indices at five days (P= 0.54), six weeks (P= 0.47), three months (P= 0.49) and at six months (P= 0.18). Day two serum FSH concentrations after surgery were unchanged (P= 0.45), compared with baseline values.
Conclusions Triple tourniquets are effective in reducing bleeding and transfusion rates. There appears no obvious adverse effect on uterine perfusion or ovarian function.
Improvements in diagnostic imaging reveal that uterine fibroids are more prevalent than originally described.1 Recent ultrasound data suggest a cumulative incidence of at least 70% for a woman aged 50 years.2 When symptomatic, fibroids are associated with reproductive dysfunction, abnormal bleeding and pelvic pressure symptoms. There is also evidence to suggest that fibroids are linked with dyspareunia and non-cyclic pelvic pain.3
Despite the introduction of new non-surgical techniques,4 where fertility is to be preserved, open myomectomy remains the most likely treatment option. In the UK in 2001, for instance, 1414 open myomectomies were performed, accounting for 7597 bed days.5 Where future fertility is not an issue, hysterectomy remains a common intervention. For all indications, 44,461 hysterectomies were performed in the same period. Recent data for the USA show that 37,000 myomectomies are performed annually.6 Open myomectomy is a major surgical procedure and is associated with considerable morbidity, in particular, operative. One recent study, for instance, reported that 23% patients lost over 1000 mL of blood.7 Other series reported transfusion rates between 18% and 24%.8–10
Haemostatic tourniquets to reduce intra-operative bleeding have long been available,11 but data on their efficacy from controlled trials are lacking. We set out to evaluate the use of triple tourniquets in controlled conditions and for the first time to investigate the hypothesis that leaving a semi-permanent tourniquet around the uterine artery reduces post-operative bleeding from the uterine incisions, a well-recognised sequel to surgery.12–14
From January 2002 to July 2003, patients seen in the Fibroid Clinic and Gynaecology Clinic at the Royal Free Hospital, and in the Gynaecology Clinic at the Whittington Hospital were assessed for entry into this randomised controlled trial. Entry criteria were symptomatic fibroids, a uterine size equivalent ≥14 weeks of gestation and requesting myomectomy. Exclusion criteria were a history of a bleeding disorder, concurrent anticoagulant therapy, a haemoglobin less than 10.5 g/dL at the time of surgery and premalignant endometrial histology.
Diagnosis of uterine fibroids was based on clinical examination and ultrasound scan. Women complaining of abnormal uterine bleeding (e.g. menorrhagia, intermenstrual bleeding) underwent outpatient diagnostic hysteroscopy and endometrial biopsy.
Pre-operative gonadotrophin releasing hormone agonists (GnRHa) were only prescribed if the patient was anaemic (Hb < 10.5 g/dL), to ensure a pre-operative haemoglobin of >10.5 g/dL, or if uterine size was >20 weeks of gestation size (to facilitate the use of a low transverse abdominal wall incision).
The study protocol was approved by the Local Ethics Committee (Application No. 5855).
Patients were allocated to the control or tourniquet by means of sealed, sequentially numbered opaque envelopes containing computer-generated random numbers. Randomisation was done immediately before surgery, and patients were not informed of their assignment.
All patients underwent a routine open myomectomy performed by AT, AM or MS based on the technique described by Buttram.1 The default skin incision was a suprapubic transverse one. The uterus was exteriorised and the bowels packed away using two large wet swabs. We tried to use a single midline anterior uterine incision whenever possible. Blood loss was assessed by weighing swabs and measuring blood collected by suction. Suction drains were left in situ for 48 hours. All patients received peri-operative thromboprophylaxis with anti-embolic stockings and heparin, and antibiotic prophylaxis with co-amoxiclav (Augmentin, GlaxoSmithKline).
Three tourniquets were applied prior to myomectomy, one to occlude the uterine arteries and two more to occlude the left and right ovarian vessels. The broad ligament was opened anteriorly and the bladder reflected inferiorly. Next, a small opening was made in the avascular space in the posterior leaf of the broad ligament on either side of the uterine isthmus superior to the uterine vessels. A number 1 polyglactin tie was threaded through the two holes and tied tightly anteriorly around the cervix at the level of the internal os using a Roeder slip knot which could be retightened using a laparoscopic knot pusher if required.
A 20-cm length of narrow bore plastic tubing (we used a 3-mm Westcott anaesthetic anti-syphon set) was used for the ovarian tourniquets based on the technique described by Thompson.15 The tubing was passed through the defect in the broad ligament and looped around the infundibulopelvic ligament lateral to the fallopian tube and ovary. The two ends of the tubing were threaded through a short length of 24 Fr Foley catheter tubing, which acted as a cushion, and pulled tight and held with a small clamp to occlude the ovarian vessels. The procedure was repeated on the contralateral side.
Women in the control group did not receive tourniquets.
Our primary end point was intra-operative blood loss. Secondary outcome measures included post-operative blood loss (assessed by surgical drains and changes in haemoglobin), blood transfusion rates, operative morbidity (return to theatre, pyrexia >38°C on more than two occasions 4 hours apart, bowel obstruction, wound infection, wound haemorrhage, chest infection and venous thromboembolism), uterine blood flow (assessed by serial measurements of the resistance index for the uterine artery) and ovarian function (assessed by changes in serum FSH).
Based on a review of the published literature,7,16 we hypothesised that 75% of our control group would lose more than 500 mL of blood compared with 25% of the tourniquet. For the probability of a type 1 statistical error (two-sided) to be less than 0.05 and the probability of a type 2 error to be less than 0.2, we calculated that we would need 14 patients in each group, giving a total of study size of 28 women.
Data were analysed using t test (paired and unpaired), Mann–Whitney U test, Wilcoxon signed rank test and Fisher's exact test using GraphPad Prism 4 software (GraphPad Software, San Diego, California, USA). We assumed significance at the 5% level (P < 0.05).
The 28 patients were recruited from a cohort of 171 patients referred to hospital for management of moderately enlarged uterine fibroids. Of the 30 women for whom open myomectomy was judged appropriate, one patient was excluded because of anaemia due to thalassaemia, and another declined to be randomised (Fig. 1). The two randomisation groups were similar in baseline characteristics, including uterine size (Table 1).
Table 1. Patient baseline characteristics. Data are mean (SD), median [range] or number.
Control group (n= 14)
Tourniquet group (n= 14)
Pre-operative Hb (g/dL)
GnRH analogue pretreatment
Uterine size (weeks)
The operative details are summarised in Table 2. All myomectomies were completed successfully. Mean intra-operative bleeding was 1870 mL greater in the control group compared with those given tourniquets (95% CI 1159–2580 mL). The transfusion rate in the control group was significantly greater than in controls (79% compared with 7%, P= 0.0003) (Table 3).
Table 2. Operative details. Data are mean (SD), median [range] or number.
Control group (n= 14)
Tourniquet group (n= 14)
Transverse incision (n)
Operating time (min)
Tourniquet time (min)
No. of fibroids removed
Weight of fibroids (g)
Blood loss (mL)
Table 3. Post-operative blood loss and transfusion data. Data are mean [SD], median (range), or number (%).
Control group (n= 14)
Tourniquet group (n= 14)
Volume in drains at 20 min (mL)
Total drained volume at 48 hours (mL)
Fall in haemoglobin at day 2 (g/dL)
No. of patients transfused
Blood (total units)
Fresh frozen plasma (total units)
Platelet (total units)
Episodes of defined post-operative morbidity were significantly greater in the control group (8 vs 1 episodes, P= 0.0128). The incidence of pyrexia >38°C was the largest single component of this difference.
There were no significant changes in uterine artery Doppler resistance indices compared with baseline values at any of the follow up time points (five days, six weeks, three months and six months; Fig. 2). Similarly, there was no significant difference in menstrual cycle day two mean serum FSH concentrations measured between three and six months after surgery compared with baseline values in either the control group (7.2 IU [3.1]) or tourniquet group (9.2 IU [7.1]) (Fig. 3).
For the first time in controlled conditions, we have shown that triple tourniquets applied to the uterine and ovarian vessels significantly reduce operative blood loss, the need for blood transfusion and peri-operative morbidity at open myomectomy. Although we did not find a statistically significant benefit to leaving the absorbable uterine tourniquet in situ after the myomectomy, this was not a primary end point and our study proved underpowered for this outcome. Reassuringly, the tourniquets did not appear to affect subsequent ovarian function. As our study was inclusive rather than exclusive (by including patients who had undergone previous laparotomy, including previous myomectomy), these results are as generalisable as possible.
Temporary haemostatic occlusion of the uterine blood supply at open myomectomy was first achieved using metal clamps17 and evolved from earlier practices of caustic packs18 and ligation of the uterine arteries.19 In the original description, both the uterine arteries and the ovarian vessels were occluded. This technique of total vascular occlusion of the uterine blood supply is still taught today.15 Our study evaluated triple tourniquets against a control group. In the future, it would be interesting to evaluate triple tourniquets against the practise of using a single tourniquet.20
Chemical tourniquets using vasopressin have also been described.7,9 One randomised controlled trial comparing intramyometrial vasopressin with uterine tourniquet did report a significant advantage to the former, but ovarian tourniquets were not applied, the uterine tourniquet was released every 20 min, and most importantly, blood loss in swabs was not included in the estimation of operative blood loss.7 Another trial utilised occlusion of both the uterine and ovarian vessels and reported no difference with vasopressin.9 It seems, therefore, that triple tourniquets are effective as vasopressin injection, but without the risks of the latter.21–23
While intra-operative blood loss in the tourniquet group was broadly comparable to previously published data, the results in the control group were considerably in excess of what we expected based on a review of the literature. One explanation could be the use of different methods of blood loss collection and estimation as discussed earlier. Another reason for this discrepancy might be differences in the amount of fibroid tissue removed. Previously published series do not to include data on the weight of fibroids removed, and yet it is known that estimation of uterine size is only a rough estimate of uterine weight and therefore fibroid weight. One study, for instance, found that a uterus, which on bimanual examination feels equivalent to 20 weeks of gestation, has on average weight of about 1000 g, but with a standard deviation of over 500 g.24 As intra-operative bleeding is proportional to the amount of fibroid tissue removed, differences in fibroid weight can account for major differences in blood loss data.9
In our study, we chose only to use pre-operative GnRHa under certain well-defined conditions. Although such pretreatment can be effective in reducing blood loss at open myomectomy,25 these drugs are expensive, are associated with oestrogen deficient side effects, can make the surgery more difficult by causing fibrosis between the tissue planes, and most importantly, have been shown to be associated with an increase in fibroid recurrence at six months.25 Recently, the use of misoprostol as a pre-operative medical agent has been described but this technique requires further evaluation.26
As an alternative to surgical management of uterine fibroids, some women have opted for uterine artery embolisation (UAE).4 However, this technique is still being fully evaluated27 and is not without controversy.28 Current guidelines in the United Kingdom, for instance, advise against its use in women desiring pregnancy.27
The use of an absorbable tourniquet that is left in situ around the uterine arteries to reduce peri-operative bleeding and subsequent pelvic adhesions is a novel idea that we believe may still be effective when examined in an appropriately powered study. It could also be that the use of drain volumes is only a crude outcome measure as it reflects not only blood loss but serous fluid that oozes from the uterine incisions.
Until now, there have been no data on uterine vascular changes following open myomectomy whether tourniquets are or are not utilised. We do know that following UAE there is a marked and permanent increase in uterine artery resistance index and a reduction in peak systolic velocity.29,30 There was no such increase in the resistance index of the uterine artery in our study. However, the study was only powered to detect differences in blood loss, and not resistance indices. A larger study would be required to be confident that there is no effect.
Also for the first time to our knowledge, we attempted to determine the effect of triple tourniquets on ovarian function using changes in serum FSH. Serum FSH is widely used to evaluate this outcome following UAE.31,32 Although the mean tourniquet time was close to 1 hour, there did not appear to be any significant changes observed compared with pre-operative levels. We accept that this is short term data and propose a longer term follow up of gonadotrophins. Similarly, as for the uterine blood flow outcomes, the study was not specifically powered to exclude an effect on ovarian function and our study might have been underpowered to determine this. However, reassuringly though, the first patient in the tourniquet group has recently given birth, having had an uncomplicated full-term pregnancy. This is consistent with data from heterotopic autotransplantation in the ewe (an accepted model in reproductive endocrinology) which suggests that there is only minimal injury to tissue even after 3 hours of ovarian vascular occlusion.33
In summary, our results clearly demonstrate the effectiveness of triple tourniquets and we would strongly advocate their routine use at open myomectomy as a safe and convenient alternative to other haemostatic techniques.