Kayem, Gilles MD, PhD; Kurinczuk, Jennifer J. MSc, MD; Alfirevic, Zarko MD; Spark, Patsy BSc; Brocklehurst, Peter MBChB, MSc; Knight, Marian MPH, DPhil; on behalf of the U.K. Obstetric Surveillance System (UKOSS)
Obstetric hemorrhage is the single most important cause of maternal mortality worldwide, accounting for 25% of all direct maternal deaths.1 It remains a major cause of maternal mortality in developed countries and accounts for 11% of direct maternal deaths in the United Kingdom and 9% in the United States.2,3 Several specific second-line therapies to control postpartum hemorrhage are used worldwide in women in whom uterotonic therapy fails. After an initial report by B-Lynch et al,4 the frequency of use of uterine compression sutures has increased since their introduction 13 years ago. The primary indication for a compression suture is hemorrhage resulting from uterine atony, but it can be used any time uterine compression is considered necessary to control uterine bleeding. This technique is of particular interest because it can be performed as soon as an operating room is available, including in settings with low technical resources, as a technique aimed at preventing hysterectomy. Uterine compression sutures are technically easier to perform than pelvic arterial surgical ligation and do not need additional facilities such as interventional radiology services.
Several modifications of the B-Lynch method as well as new methods using penetrating sutures have been introduced.4–12 Little is known, however, about the indications for their use or success rates because data come largely from small retrospective studies or case reports. Systematic data at a population level to assess how these surgical procedures are used and the clinical outcomes after their use are also lacking.
The objective of this nationwide population-based study was to assess maternal outcomes of uterine compression sutures and to characterize the risk factors for failure, defined as cases that proceeded to hysterectomy.
MATERIALS AND METHODS
Cases were defined as any woman giving birth who was treated with a uterine compression suture to treat postpartum hemorrhage. We identified cases on a national basis through the monthly mailing of the U.K. Obstetric Surveillance System (UKOSS) between September 2007 and March 2009 in a study that was designed to identify women treated with a range of novel therapies for postpartum hemorrhage, including uterine compression sutures. The UKOSS methodology has been described in detail elsewhere.13 In brief, we sent UKOSS case notification cards every month to between one and four nominated reporting clinicians in each hospital in the United Kingdom with a consultant-led maternity unit with a tick box list to indicate whether they had seen any women, including women who died, treated with uterine compression sutures. Cases were defined on the basis of uterine compression suture use and not on the basis of measured blood loss to ensure that we did not introduce any severity bias into the assessment of efficacy. We also asked them to return cards indicating a “nil report” so that we could monitor rates of cards returned and confirm the denominator maternity population to calculate use rates. When a clinician returned a card indicating a case of uterine compression suture use, he or she was sent a data-collection form requesting further details of prognostic factors, other management, and outcomes. Up to five reminders were sent if completed forms were not returned.
To ensure that all cases were identified, we independently contacted intensive care units and asked them to report women treated with uterine compression sutures. No additional cases were identified through this route. In addition, we wrote to the clinical director of each unit at the end of the study asking him or her to confirm whether the reported number of cases was correct. Two additional cases were identified through this route.
All women in whom uterine compression sutures were used to treat a postpartum hemorrhage were included whatever the type of suture. Sutures were classified as follows: B-Lynch,4 modified B-Lynch with two vertical sutures but the uterus not opened (Hayman procedure or similar),5 or other, including multiple vertical, squared compression sutures, or other techniques.
We calculated the use rates with 95% confidence intervals (CIs) by using denominator data from the most recently available hospital episode statistics and birth registration data (2007) as a proxy for September 2007 to March 2009.14 This study covered the entire cohort of U.K. births (an estimated 1,210,279 women delivering) for this period.
Categorical variables were compared using the chi square test or Fisher's exact test as appropriate. We investigated factors associated with compression suture failure, defined as a hysterectomy, in a multivariable logistic regression model. We built a pragmatic model by including all potential explanatory factors identified from the literature or by prior hypothesis in a core model. The following factors were included: women's demographic, medical, and pregnancy characteristics; mode of delivery; cause of postpartum hemorrhage; timing of the use of compression suture; and the delay between delivery and the uterine compression suture placement. Where data were missing for more than 10% or cases, a proxy variable was included in the model to account for the missing points and maximize data availability. We tested for all possible interactions among the following variables: age, parity, body mass index, socioeconomic group, and ethnicity. We used Stata 10 software for all analyses.
This study (07/MRE02/24) was approved by the London Research Ethics Committee.
All 226 U.K. hospitals with consultant-led maternity units contributed data to UKOSS (100%). The mean monthly card return rate over the 19 months of the study was 90%. Five hundred twenty-four cases were reported initially; data collection was completed for 91% of these cases. One hundred thirty-four cases were subsequently reported by clinicians as not cases after initial notification; there were 11 duplicate reports and 12 women whose records were lost. Completed data-collection forms were received for 101 women who were treated with other therapies for postpartum hemorrhage but not treated with uterine compression sutures (Fig. 1). These women are not described in this analysis.
A total of 213 eligible cases were reported in an estimated 1,210,279 women who delivered, representing an estimated rate of use of compression sutures of 18 cases per 100,000 women delivering (95%CI, 15–20 per 100,000). We had minimal information about two cases and we thus excluded them from further analysis. Two hundred eleven women who had a uterine compression suture to treat a postpartum hemorrhage were thus included in this analysis. Seventy-nine (37%) had a B-Lynch procedure; 48 (23%) a modified B-Lynch with two vertical sutures; 32 other techniques including figure-of-eight sutures (n=16), multiple compression sutures (n=7), square sutures (n=1), B-Lynch associated with another uterine suture (n=4), and atypical hemostatic uterine sutures (n=4); and 52 (25%) had an unspecified technique of uterine compression suture. The overall rate of subsequent hysterectomy was 25% (95% CI, 19–31%). Rates of hysterectomy were not significantly different according to the method of uterine compression suture used (Table 1). One hundred twenty-nine women (61%) had a hemorrhage resulting from atony. For these, hysterectomy rates according to the different methods of uterine compression suture were also not significantly different (B-Lynch: 25% [95% CI, 14–38%]; modified B-Lynch: 33% [95% CI, 16–50%]; other methods: 6% [95% CI, 0–30%]; unspecified: 29% [95% CI, 14–48%]).
Four women (2%), of whom one had a hysterectomy, did not receive uterotonic drugs (either oxytocin or oxytocin in combination with ergometrine) prophylactically for the third stage of labor. Ninety-eight percent of women (n=208) were treated with uterotonic therapy before compression suture placement. This treatment followed uterotonic prophylaxis in 98% (n=204) of women. All the women received uterotonic treatment either for prophylaxis or treatment of postpartum hemorrhage.
Uterine balloon or packing was used in addition to the uterine compression suture for 66 women (31%) overall, in 41 (62%) before and in 25 (38%) after the uterine compression sutures were placed. Twenty-one (32%; 95% CI, 21–44%) of the women treated with both procedures in combination went on to have a hysterectomy. Ten women (5%) were treated with arterial embolization or ligation before the uterine compression suture and an additional two (1%) were treated with recombinant factor VIIa. Six (50%; 95% CI, 21–79) of these women went on to have a hysterectomy. Eighteen women (9%) were treated with arterial embolization or ligation after the compression suture, and an additional nine (4%) were treated with recombinant factor VIIa. Thirteen (48%; 95% CI, 29–68) of these women went on to have a hysterectomy. The use of an additional second-line therapy after the use of a uterine compression suture was more frequent in the hysterectomy group (unadjusted odds ratio [OR], 3.09; 95% CI, 1.46–6.56).
The reported causes of postpartum hemorrhage and the subsequent risk of hysterectomy are described in Table 2. There were no statistically significant differences in the risk of hysterectomy depending on the cause of the hemorrhage, although the numbers of women in most categories are small; hence, this analysis has limited statistical power.
The association between the risk of hysterectomy and the delay between delivery and uterine compression suture was not linear (Fig. 2). Therefore, in further analyses, we have included a categorical variable to describe this delay, which allows for the possibility that the women having uterine compression suture therapy more than 6 hours after delivery may be a clinically different group. Time categories were thus formed at 0–1 hours, 1–2 hours, and 2–6 hours to investigate the observed gradient while also including a cut point at 6 hours to account for the main change in risk. Factors associated with hysterectomy in women treated with uterine compression suture are described in Table 3. Having adjusted for all the characteristics in the table, older maternal age (35 years or older), multiparity, unemployed and routine or manual occupational status, vaginal delivery, and a delay between delivery and uterine suture compression of 2–6 hours were significantly associated with an increased risk of hysterectomy. Women with hemorrhage resulting from an “other” cause (including placental abruption [n=14], amniotic fluid embolism [n=2], infection [n=2], bleeding in left broad ligament [n=1], and unspecified causes in three cases) were significantly less likely to have a hysterectomy than women with hemorrhage resulting from uterine atony. The rate of hysterectomy was, however, similar when the causes of the postpartum hemorrhage leading to the uterine compression suture were divided into two categories: atony compared with others. The hysterectomy rate was 26% (95% CI, 18–34) in cases with atony and 23% (95% CI, 15–34) in cases with other causes, including disorders of placentation (accreta, praevia) and uterine tear (P=.69). After adjustment for all the variables included in Table 3, the risk of hysterectomy was no different in women with atony compared with other causes of hemorrhage (adjusted OR, 0.48; 95% CI, 0.20–1.18).
One hundred twenty-seven women labored before delivery; 31 (24%) had a hysterectomy. Hysterectomy in this subgroup was not associated with induction of labor (unadjusted OR, 1.42; 95% CI, 0.71–2.82), induction with prostaglandin (unadjusted OR, 1.39; 95% CI, 0.37–5.19), or the use of oxytocin augmentation during labor (unadjusted OR, 0.54; 95% CI, 0.23–1.26).
Randomized trials are challenging to perform in the context of emergency “near miss” events such as severe postpartum hemorrhage. As a consequence, detailed prospective population studies can play an important role in investigating the use of specific surgical therapies for these and similar conditions, providing an estimate of their success, failure, or complication rates and enabling international comparisons. This national population study describes the use of uterine compression sutures for postpartum hemorrhage.
We found an overall risk of hysterectomy of 25% in women treated with uterine compression sutures, which is generally higher than previously reported. In cases treated with brace sutures, reported success rates have been close to 100%5,6,8,15; small retrospective studies including between 11 and 28 women have reported risks of hysterectomy of between 9% and 28%.6,15–18 Risks of hysterectomy reported in series of women treated with other uterine compression sutures (excluding B-Lynch or modified B-Lynch) are zero out of 23 women (0%) for the Cho technique, one out of 20 women (5%) for the Ouahba technique, and Hackethal et al report zero hysterectomies among seven women (0%).9,11,12 One weakness of these retrospective center studies is their generalizability, because the procedure is performed by obstetricians particularly skilled with the specific technique, which therefore may increase the apparent success rate. Moreover, selective population recruitment may bias such study results. By identifying all the cases within a population, national observational studies effectively avoid such case selection bias.
A prolonged delay of 2–6 hours between delivery and uterine compression suture was independently associated with a fourfold increase in the odds of hysterectomy. Factors that prolong the delay between delivery and the surgical procedure may explain this observation. The delay is unlikely to be the result of expectant management after delivery because nearly all the women had uterotonic drugs at delivery. We adjusted for other factors that may have prolonged the delay in suture placement such as the use of additional procedures to control hemorrhage and the association between delay and hysterectomy remained significant. One possible explanation may be that unrecognized bleeding that prolongs the delay between the delivery and the treatment increases the risk of hysterectomy. A higher blood loss and disseminated intravascular coagulation would lead to clinical conditions that render hysterectomy almost inevitable. The increased risk of hysterectomy associated with a delay between delivery and the surgical procedure emphasizes the need for a careful evaluation of blood loss after delivery to avoid any prolonged delay in recognition of hemorrhage.
Interestingly, vaginal delivery was associated with a higher risk of hysterectomy in women treated with uterine compression sutures and indeed of all the factors we explore was associated with the highest odds (adjusted OR, 6.08). It is possible that the obstetrician is more reluctant to perform a laparotomy to insert a compression suture after excessive bleeding after a vaginal delivery than after a caesarean delivery and that, therefore, only the women with the most severe hemorrhage were selected by the obstetrician to have a uterine compression suture after a vaginal delivery. It is also possible that other methods such as intrauterine balloon or uterine packing were used to control postpartum hemorrhage after vaginal delivery. These cases would not have been recorded by our study when successful. Therefore, cases that have uterine compression sutures after a vaginal delivery may be the most serious, for whom there were no other possible treatment modalities available.
In the present study, maternal age of 35 years or older and multiparity were associated with a threefold increase in the odds of hysterectomy. The association with an older maternal age and multiparity has been found in several studies in developed countries focusing on hysterectomy risk factors in cases of postpartum hemorrhage.19–22 A matched retrospective case–control study in Washington state found that 38% of women who had a hysterectomy were older than 34 years compared with 13% in the control group (P<.01).23 Similar results were also found in a Dutch population study.21 One possible explanation for these results may be that the decision to perform a hysterectomy is taken more readily if a woman is older and has no wish for further pregnancies compared with younger women and those having their first baby.
Maternal complications directly related to the uterine compression suture placement are known to occur but we were unable to follow-up women beyond their initial hospital discharge to investigate this. Short- and long-term complications have been described after the use of uterine compression sutures, including uterine erosion and necrosis of the uterus, septic pyometra, and uterine myometrial defect presenting in a subsequent pregnancy.24–27 These complications emphasize the importance of studying long-term outcomes.
This is not a randomized study and therefore comparisons between the different suture methods are limited because the baseline population treated may differ. Nevertheless, the main strength of this study is that we have collected comprehensive population-based national information about women treated with uterine compression sutures for postpartum hemorrhage. Data for this study were collected by clinicians using information from hospital case notes; thus, we have only been able to investigate factors that are adequately recorded in these notes. However, before starting the study, we did not identify any hypothesized factors that would be insufficiently well recorded to investigate using this methodology. The information on risk factors is all recorded prospectively in obstetric case notes, before delivery or hysterectomy; we believe, therefore, that any information bias will be minimal. Although we used several methods to maximize case ascertainment, we also cannot exclude the possibility of case underascertainment, because UKOSS was the main source of cases included in this study. However, previous studies using similar methodology have demonstrated a high case ascertainment28 and we believe systematic underascertainment is unlikely to have led to any significant bias in the reported results.
A prolonged delay of 2–6 hours between delivery and uterine compression suture was independently associated with a fourfold increase in the odds of hysterectomy. These data emphasize the need for a careful evaluation of blood loss after delivery to avoid any prolonged delay in recognition of hemorrhage.
1. Goodburn E, Campbell O. Reducing maternal mortality in the developing world: sector-wide approaches may be the key. BMJ 2001;322:917–20.
2. Confidential Enquiry Into Maternal and Child Health. Saving Mothers' lives. 2003–2005. London (UK): Royal College of Obstetrician and Gynecologists; 2007.
3. Kung HC, Hoyert DL, Xu J, Murphy SL. Deaths: final data for 2005. Natl Vital Stat Rep 2008;56:1–120.
4. B-Lynch C, Coker A, Lawal AH, Abu J, Cowen MJ. The B-Lynch surgical technique for the control of massive postpartum haemorrhage: an alternative to hysterectomy? Five cases reported. Br J Obstet Gynaecol 1997;104:372–5.
5. Hayman RG, Arulkumaran S, Steer PJ. Uterine compression sutures: surgical management of postpartum hemorrhage. Obstet Gynecol 2002;99:502–6.
6. Bhal K, Bhal N, Mulik V, Shankar L. The uterine compression suture—a valuable approach to control major haemorrhage at lower segment caesarean section. J Obstet Gynaecol 2005;25:10–4.
7. Nelson GS, Birch C. Compression sutures for uterine atony and hemorrhage following cesarean delivery. Int J Gynaecol Obstet 2006;92:248–50.
8. Nelson WL, O'Brien JM. The uterine sandwich for persistent uterine atony: combining the B-Lynch compression suture and an intrauterine Bakri balloon. Am J Obstet Gynecol 2007;196:e9–10.
9. Cho JH, Jun HS, Lee CN. Hemostatic suturing technique for uterine bleeding during cesarean delivery. Obstet Gynecol 2000;96:129–31.
10. Hwu YM, Chen CP, Chen HS, Su TH. Parallel vertical compression sutures: a technique to control bleeding from placenta praevia or accreta during caesarean section. BJOG 2005;112:1420–3.
11. Ouahba J, Piketty M, Huel C, Azarian M, Feraud O, Luton D, et al. Uterine compression sutures for postpartum bleeding with uterine atony. BJOG 2007;114:619–22.
12. Hackethal A, Brueggmann D, Oehmke F, Tinneberg HR, Zygmunt MT, Muenstedt K. Uterine compression U-sutures in primary postpartum hemorrhage after Cesarean section: fertility preservation with a simple and effective technique. Hum Reprod 2008;23:74–9.
13. Knight M, Kurinczuk JJ, Tuffnell D, Brocklehurst P. The UK Obstetric Surveillance System for rare disorders of pregnancy. BJOG 2005;112:263–5.
14. Office for National Statistics. Key Population and Vital Statistics 2007. Newport, South Wales (UK): Office for National Statistics; 2008.
15. B-Lynch C. Conservative surgical management. In: B-Lynch C KL, Lalonde AB, Karoshi M, editors. A textbook of postpartum hemorrhage. A comprehensive guide to evaluation, management and surgical intervention. London (UK): Sapiens Publishing; 2006. p. 287–98.
16. Wohlmuth CT, Gumbs J, Quebral-Ivie J. B-Lynch suture: a case series. Int J Fertil Womens Med 2005;50:164–73.
17. Allahdin S, Aird C, Danielian P. B-Lynch sutures for major primary postpartum haemorrhage at caesarean section. J Obstet Gynaecol 2006;26:639–42.
18. Sentilhes L, Gromez A, Razzouk K, Resch B, Verspyck E, Marpeau L. B-Lynch suture for massive persistent postpartum hemorrhage following stepwise uterine devascularization. Acta Obstet Gynecol Scand 2008;87:1020–6.
19. Knight M; UKOSS. Peripartum hysterectomy in the UK: management and outcomes of the associated haemorrhage. BJOG 2007;114:1380–7.
20. Whiteman MK, Kuklina E, Hillis SD, Jamieson DJ, Meikle SF, Posner SF, et al. Incidence and determinants of peripartum hysterectomy. Obstet Gynecol 2006;108:1486–92.
21. Zwart JJ, Dijk PD, van Roosmalen J. Peripartum hysterectomy and arterial embolization for major obstetric hemorrhage: a 2-year nationwide cohort study in the Netherlands. Am J Obstet Gynecol 2010;202:150.e1–7.
22. Zelop CM, Harlow BL, Frigoletto FD Jr, Safon LE, Saltzman DH. Emergency peripartum hysterectomy. Am J Obstet Gynecol 1993;168:1443–8.
23. Bodelon C, Bernabe-Ortiz A, Schiff MA, Reed SD. Factors associated with peripartum hysterectomy. Obstet Gynecol 2009;114:115–23.
24. Ochoa M, Allaire AD, Stitely ML. Pyometria after hemostatic square suture technique. Obstet Gynecol 2002;99:506–9.
25. Joshi VM, Shrivastava M. Partial ischemic necrosis of the uterus following a uterine brace compression suture. BJOG 2004;111:279–80.
26. Price N, Lynch C. Uterine necrosis following B-Lynch suture for primary postpartum haemorrhage. BJOG 2006;113:1341; author reply 1342.
27. Wu HH, Yeh GP. Uterine cavity synechiae after hemostatic square suturing technique. Obstet Gynecol 2005;105:1176–8.
28. Knight M; UKOSS. Antenatal pulmonary embolism: risk factors, management and outcomes. BJOG 2008;115:453–61.