Key Points

• Six small, single studies of fair and poor quality addressed various pharmacologic interventions not including recombinant activated factor VIIa (rFVIIa) with mixed results.
• In one fair quality retrospective cohort study assessing oxytocin and other uterotonics, bleeding was controlled with uterotonic medications without need for further procedures/surgeries in 45 of 91 women (49% success rate).
• In one RCT of tranexamic acid (TXA), blood loss, progression to severe PPH, and need for transfusion were reduced in the TXA arm compared with the non-TXA control arm, but need for further interventions did not differ.
• Need for transfusion or further interventions did not differ in a retrospective cohort study comparing misoprostol and methylergonovine maleate.
• In a small, population-based case series, sulprostone stopped bleeding in 83 percent of participants without need for further intervention.
• Carboprost tromethamine controlled bleeding in 88 percent of women in a small, population-based case series.
• Blood loss and transfusion in women with PPH and disseminated intravascular coagulation (DIC) did not differ in a retrospective study comparing women who received recombinant thrombomodulin with matched controls who did not receive the drug.
• Six small studies of rFVIIa also had mixed results. In one retrospective cohort study, women in the rFVIIa group required more blood products and had greater blood loss than women not receiving the treatment. Differences in change in prothrombin time were not significant between women treated with rFVIIa and those who were not in a case-control study. rFVIIa used as a second-line intervention controlled bleeding without need for further procedures or surgeries in 27 to 31 percent of women in one cohort study, a rate that was similar to treatment with other second-line interventions in that study. In registry studies bleeding was considered improved after one or multiple doses of rFVIIa in 64 to 80 percent of women after the final dose. No study included more than 177 women receiving rFVIIa.
• Strength of the evidence is insufficient for all outcomes of oxytocin and other uterotonics, misoprostol, tranexamic acid, carboprost tromethamine, thrombomodulin, and rFVIIa for PPH management due to the study sizes and lack of studies addressing each agent.

Overview of the Literature

Twelve studies addressed pharmacologic agents for the treatment of PPH:^69-80one RCT,⁶⁹ five cohort studies,^{72, 73, 77-79} one case-control study,⁷⁴ and five population-based case series or registry studies.^{70, 71, 75, 76, 80} Studies were conducted in France,^{69, 70} the United States,^{71, 78} Finland,⁷³ Ireland,⁷⁴ Japan,⁷² the United Kingdom,⁷⁷ Hong Kong,⁷⁹and Australia and New Zealand.^{76, 80} These studies from Australia and New Zealand report on data collected from one registry over differing time periods, but because the overlap in data is not clear, we have presented results from both studies but note that the populations likely overlap to some extent. Another registry study reported data from various northern European countries.⁷⁵

Six of these studies (two cohort studies,^{73, 77}one case-control,⁷⁴ and three registry studies^{75, 76, 80}) addressed rFVIIa. Atony accounted for many of the cases of PPH in studies reporting etiology (range = 18 to 56% of cases).

Other agents were each addressed in one study: tranexamic acid (one RCT, n = 144),⁶⁹ oxytocin and other (unspecified) uterotonics (one retrospective cohort, n=49),⁷⁹ misoprostol compared with methylergonovine maleate (one retrospective cohort, n = 58),⁷⁸ sulprostone (one population-based case series, n = 1,370),⁷⁰ carboprost tromethamine (one registry study, n = 236),⁷¹ and recombinant human soluble thrombomodulin (rTM; one cohort study, n = 36).⁷² Two studies included only women with atonic PPH,^{70, 71} and, where reported, atony accounted for 27 to 65 percent of cases. We rated the RCT as poor quality for all effectiveness outcomes and the five cohort and case-control studies as fair quality. The case series were considered poor quality by default. Table 6 provides an overview of key outcomes in studies with comparison groups. We note that one additional cohort study reported only harms of methylergonovine maleate and is discussed in KQ3.⁶⁰

Table 6

Key outcomes in comparative studies of pharmacologic agents.

Detailed Analysis

Recombinant Activated Factor VIIa (rFVIIa)

A fair quality retrospective cohort study in Finland compared the effectiveness of rFVIIa versus standard management (no rFVIIa) among women with PPH (defined as loss of 1.5 times patient's blood volume).⁷³ Eligible participants were identified using medical records at a single tertiary referral hospital. Of the 48 women identified, 26 were treated with rFVIIa and 22 were not. There were no statistically significant differences in age, body mass index (BMI), obstetrical course (cause of PPH, mode of birth, length of hospital stay after birth), lowest hemoglobin, or lowest platelet count between the two groups. Activated partial thromboplastin time, liters of total bleeding (11.3 vs. 8.0, p = 0.005), units of RBC (20 vs. 13, p = 0.003), units of platelets (23 vs. 14, p = 0.014), and number with fibrinogen concentrate transfused (15 vs. 5, p = 0.014) were significantly greater among women treated with rFVIIa than among untreated women. There was no statistical comparison of maternal or fetal outcomes between the groups.

A retrospective case-control study in Ireland compared the effectiveness of rFVIIa in reversing coagulopathy associated with massive PPH versus standard management (no rFVIIa) between 2003 and 2006.⁷⁴ Twenty-eight women with massive PPH (defined as transfusion of > 5 units of PRBC in 24 hours) were identified using medical records at a single Irish hospital. Of these, six women who were treated with rFVIIa and had a prolonged prothrombin time (PT) were matched with six women with the largest number of PRBC units transfused and prolonged PTs who were not treated with rFVIIa. There were no statistically significant differences in age, obstetrical factors (gestation, parity, cause of massive PPH, or number of hysterectomies), or coagulopathy factors (PRBC, platelets, fresh frozen plasma [FFP], or cryoprecipitate transfused, or worst PT or fibrinogen levels) between the two groups. The PT improved with management in both groups, and there was no significant difference in the magnitude or absolute value of improvement (p = 0.9). There was no statistical comparison of maternal or fetal outcomes between the groups.

One fair quality cohort study used data from the U.K. Obstetric Surveillance System (UKOSS). The UKOSS includes all hospitals with a consultant-led maternity unit in the United Kingdome. Clinicians in these hospitals reported data on PPH cases and treatment to the UKOSS using case notification cards completed monthly. UKOSS personnel also followed up with hospitals to identify potential missed cases. In this study, 31 women received rFVIIa as the initial second-line therapy after failure of conservative PPH management approaches. Sixteen received rFVIIa after uterotonic failure, and 15 received it after failure of uterotonics plus intrauterine balloon tamponade (either with balloon or packing). Among the 16 who had received only uterotonics plus rFVIIa, 11 had successful cessation of bleeding. One required compression sutures, two had ligations, one had interventional radiology, and seven required hysterectomy to control bleeding. Thus, the success rate (control of bleeding without further procedures or surgeries) for rFVIIa was 31 percent. Among the 15 who had rFVIIa after intrauterine tamponade with balloon or packing plus uterotonics, seven required hysterectomy while interventional radiology controlled bleeding after rFVIIa in four (27% success rate for rFVIIa plus uterine tamponade).⁷⁷

Three registry studies also assessed use of rFVIIa. A voluntary registry study described outcomes of treatment of PPH with rFVIIa in nine Northern European countries.⁷⁵ Eligible women (128 total identified, 108 included in the analysis) were identified differently in each country, with most identified by physicians or pharmacists who responded to requests for information about use of rFVIIa for treatment of PPH. In Finland and the Netherlands, information was collected for national surveys prior to initiation of this study, and those data were provided to the study group. Information on study endpoints was gathered retrospectively via standardized surveys completed by local practitioners in some instances and via national survey data in others. The registry gathered information on hematologic parameters after the use of rFVIIa as the primary treatment for PPH and as secondary prophylaxis if other interventions were used prior to rFVIIa. Clinicians noted improvements in bleeding after a single dose in 80 percent of the 92 women receiving rFVIIa to treat PPH and in 75 percent of the 16 women receiving it as secondary prophylaxis. Clinicians judged rFVIIa as failing to control bleeding in 15 cases overall (13.8%) Hemoglobin increased in 51 percent of cases in which bleeding was reduced after rFVIIa and showed no significant change in 32 percent of cases. Hemoglobin levels dropped post-administration in 17 percent of cases.

Two comprehensive registry studies were performed to describe outcomes of off-label use of rFVIIa for treatment of PPH in Australia and New Zealand.^{76, 80} Cases were identified between 2002 and 2008 from the Australian and New Zealand Haemostasis Registry (developed using unrestricted educational grant funds from Novo Nordisk Pharmaceuticals, the maker of rFVIIa), representing 38 hospitals in those countries. Data were collected via standardized data forms from 105 case medical records and treating clinicians of women with acute obstetric hemorrhage who received rFVIIa. Overall, bleeding stopped or decreased in 76 percent of women. Most (78%) women received a single dose of rFVIIa, and 64 percent of these women had decrease or cessation of bleeding. Median dose of rFVIIa was 92 micrograms/kg (range 9 to 139). Most women (76%) required < 6 units PRBC transfusion after receiving rFVIIa, and 13 women (21%) required hysterectomy after rFVIIa failed to control bleeding.

In the second registry study, which includes some of the same women in study summarized above, cases with off-label use of rFVIIa (non-hemophilia indications) were identified at 96 hospitals between 2000 and 2009 in the Australian and New Zealand Haemostasis Registry.⁸⁰ The registry included 95 percent of off-license use of rFVIIa during that time frame. Of 3,446 cases of off-label rFVIIa use identified, 177 were obstetric cases from 175 women with PPH. Data were collected both retrospectively (2000-2005) and prospectively (2005-2009) by trained data collectors, and were validated by central registry staff. A single dose of rFVIIa was used in 134 (76%) of women, and bleeding stopped or decreased in 99 (56%) of women after a single dose, and 114 (64%) of women after the final dose was given. Table 7 outlines key outcomes in comparative studies.

Table 7

Key outcomes in comparative studies of rFVIIa.

Transfusion for Supportive Management of Ongoing PPH

Detailed Analysis

A fair quality, single-center, retrospective cohort study conducted in the United States compared complication rates between whole blood transfusion, PRBC transfusion alone, and combination blood product transfusion for supportive management of PPH.⁸³ Eligible participants with PPH (defined as hypovolemia sufficient to provoke hemodynamic instability) were identified using a database of obstetric and neonatal outcomes. Of 1,540 women identified, 659 received whole blood transfusion, 593 received PRBC only, and 288 received a combination of blood products. There were no statistically significant differences between groups in age, race, or parity, but women in the combination blood product group were more likely to have perineal trauma, placenta previa or abruption, and hysterectomy than the other groups. Mean units of blood product transfused was significantly greater among women getting a combination of blood products when compared with women receiving whole blood or PRBC only (5.5, 2.2, and. 2.3 units in the combination blood products, whole blood, and PRBC groups, respectively, p < 0.001). Women in the combination transfusion group were also significantly more likely to be transferred to the ICU (23%, 4%, and 7% in the combination blood products, whole blood, and PRBC alone groups, respectively, p < 0.05) and to die (2%, 0%, and 1% in the combination blood products, whole blood, and PRBC alone groups, respectively, p = 0.03) than women in the other two groups.

Another fair quality, single-center, retrospective cohort study used electronic medical records at a Korean academic hospital to determine whether patients with an elevated shock index at the time of presentation with PPH would be more likely to require massive transfusion.⁸⁵ Women with PPH (defined as blood loss ≥ 500 mL) were identified as part of the massive transfusion group (defined as receiving transfusion of ≥ 10 units PRBC within 24 hours of birth, n=26) or the non-massive transfusion group (n=100). Groups did not differ in terms of age, parity, mode of birth, bleeding time. Significantly fewer women in the massive transfusion group had an alert mental status (18 vs. 95, p < 0.01) and underwent embolization (22 vs. 36, p < 0.01), and significantly more women in this group required ICU stay (11 vs. 5, p < 0.01) and died (3 vs. 0, p < 0.01). Additionally the median systolic and diastolic blood pressures and hemoglobin levels were significantly lower (5.9 vs. 9.5, p < 0.01), and the median shock index (1.3 vs. 0.8, p < 0.01) and length of hospital stay (4.0 vs. 2.0, p < 0.01) were significantly higher in the massive transfusion group than in the non-massive transfusion group. Transfusion requirements were significantly higher in the first 24 hours and during the hospitalization among the massive transfusion group than the non-massive transfusion group (18.0 units and 3.0 units in the first 24 hours, respectively, and 20.0 units and 4.0 units during the hospitalization, respectively). These finding are confounded by indication as the massive transfusion group was presumably experiencing more severe PPH given their lower median hemoglobin and lower median systolic and diastolic blood pressures than the non-massive transfusion group.

A fair quality, single-center, retrospective cohort study from Ireland compared the effectiveness of transfusion with cryoprecipitate (n = 14) versus fibrinogen concentrate (n = 20) for supportive management of PPH.⁸⁴ Women were identified for inclusion in a major obstetric hemorrhage database if they experienced PPH (defined as blood loss of ≥ 2.5 L, transfusion of ≥ 5 units PRBC, or treatment of a coagulopathy in the acute event). Eligible participants from the database were women treated with either cryoprecipitate or fibrinogen concentrate between 2009 and 2011. There were no statistically significant differences between groups in age, race, BMI, parity, gestation at birth, birth weight, or cause of PPH, but women in the cryoprecipitate group were more likely have previous cesarean birth. There was no statistically significant difference between groups in mean estimated blood loss; number of units of PRBC, Octaplas/fresh frozen plasma, or platelets transfused; medical and surgical treatments administered; and mean length of hospital stay.

Finally, one poor quality pre-post study from the United Kingdom compared the effectiveness of fibrinogen concentrate (n=51) versus fresh frozen plasma (n=42) for management of PPH-associated coagulopathy.⁸⁶ Eligible participants were identified within a single hospital if they had major obstetric blood loss (defined as > 1500 mL) associated with coagulopathy between April 2011 and June 2013, with participants treated between April 2011 and March 2012 receiving treatment with a major obstetric hemorrhage algorithm that included fresh frozen plasma, and participants included from July 2012 through June 2013 receiving treatment with fibrinogen concentrate. Women treated with fibrinogen concentrate received significantly fewer total blood components (3.0 vs 8.0, for the fibrinogen concentrate group vs. the plasma group, p= 0.0004), pooled bags of cryoprecipitate (numbers not reported), total quantity of fibrinogen (0 vs. 3.2, for the fibrinogen concentrate group vs. the fresh frozen plasma group, p= 0.0005), and doses of platelets (numbers not reported). Units of red blood cells given to the two groups did not differ significantly, nor did ICU admission, transfusion-related acute lung injury (n=0 in both periods), or hysterectomy. There was a significantly higher rate of transfusion-associated circulatory overload in the fresh frozen plasma group (p=.04). Table 8 outlines key outcomes.

Table 8

Key outcomes in comparative studies of transfusion for supportive management of PPH.

Key Points

• No good quality studies addressed uterine balloon tamponade.
• In one fair quality pre-post study, 86% of women who had balloon tamponade did not require further procedures or surgeries.
• Case series reported a decrease or cessation of bleeding in 75 to 98 percent of patients treated with a balloon tamponade device, with and without prior or subsequent surgeries or procedures.
• Strength of the evidence for outcomes related to uterine balloon tamponade is insufficient given the small number of studies and small sample sizes.

Overview of the Literature

Five studies, one pre-post study, one retrospective cohort study, two retrospective case series, and one population-based case series, addressed the use of intrauterine balloon tamponade for the management of PPH.^{79, 87-90} The pre-post study was conducted in France, cohort study in Hong Kong, and case series in the United States, Finland, and Italy. Many of the women in these studies had atony (100% in pre-post study, 57.2% in the cohort study, and 16%-72.7% in case series). A total of 208 women had intrauterine tamponade using Bakri,^{87, 90}Sengstaken-Blakemore,⁷⁹ Rusch,⁸⁹ or Belfort-Dildy Obstetrical Tamponade System⁸⁸ balloons. .

Detailed Analysis

One fair quality pre-post study examined the rate of invasive procedures (embolization and surgery) after adding balloon tamponade to the protocol for PPH management in a maternity unit at a tertiary care university hospital in France.⁸⁷ The new protocol required that intrauterine balloon tamponade be performed prior to any invasive intervention in cases of PPH due to uterine atony that were nonresponsive to sulprostone. Data were collected prospectively for 30 months after implementation of the new protocol. The patients in the control group (n = 290, none of whom had balloon tamponade) were identified from electronic medical records as women admitted to the hospital with PPH due to atony requiring sulprostone therapy in the 30 months prior to the new protocol implementation. During the study period, 395 women with PPH required sulprostone therapy, which was unsuccessful in 72 women. Of these women who needed additional procedures or surgeries, 43 had intrauterine balloon tamponade as the initial second-line therapy. No additional procedures or surgeries were required after balloon tamponade in 92% (11/12) of the women who had cesarean births and 84% (26/31) of the women who had vaginal births. Among the six women for whom balloon tamponade was unsuccessful, three had embolization, two had conservative surgical interventions (defined as artery ligations and/or uterine compression sutures), and one had hysterectomy. The overall success rate of balloon tamponade was 86% (37/43 women). Adding balloon tamponade to the protocol decreased the rates of arterial embolization (8.2% pre vs. 2.3% post, p = 0.006, OR 0.26, 95 percent CI: 0.09-0.72) and conservative surgical procedures (5.1% pre vs. 1.4% post, p = 0.029, OR 0.26, 95% CI: 0.07-0.95) among women with vaginal births. Hysterectomy and transfusion rates were unchanged. Rates of invasive interventions and transfusion were unchanged among women with cesarean births (Table 9).

Table 9

Key outcomes in studies of uterine balloon tamponade.

In a fair quality cohort study (see full description in Oxytocin and Other Uterotonics section above), 42 of 91 women with massive PPH required second-line procedures or surgeries to control bleeding.⁷⁹ Procedures included balloon tamponade (n=12), embolization (n=5), and sutures (n=26), and women receiving second-line therapies did not differ in terms of age, BMI, parity, mode of birth, or causes of PPH. Twelve women received uterine balloon tamponade with a Sengstaken-Blakemore tube, with successful control of bleeding in 9 (75%). One woman required subsequent embolization, and two required hysterectomy to control bleeding.

One population-based case series examined the outcomes of women with PPH treated with a dual-balloon catheter tamponade device, the Belfort-Dildy Obstetrical Tamponade System, using postmarketing surveillance data from medical records and clinician interviews at 11 hospitals in the United States.⁸⁸ During the study period (September 2010 – October 2012), 51 women with PPH were treated with the balloon tamponade device. Of these, 28 women had vaginal births and 23 had cesarean births. The median time interval between birth and insertion of the balloon was 2.2 hours (range 0.3-210 hours). Estimated median blood loss was 2000mL (range 855-8700). Thirty-nine (77%) patients required PRBC transfusion, and 12 (24%) were admitted to the ICU. Bleeding was considered to be decreased in 22 (43%) women and stopped in 28 (55%). Eight patients (16%) required additional procedures or surgeries after the balloon placement including hysterectomy (n = 4), uterine artery embolization (n = 4), and surgical repair (n = 3); some required more than one intervention. The overall success rate of balloon tamponade in controlling or decreasing bleeding was 98% (50/51 women, who also had prior medical or surgical interventions). Table 9 outlines key outcomes in studies of uterine balloon tamponade.

A retrospective case series evaluated uterine tamponade conducted with a Rusch balloon between 2002 and 2012 at one Italian center.⁸⁹ All 52 women who had balloon tamponade (mean age=34.4±4.4, 39% multiparous, 60% with atony) received initial uterotonics and other conservative management. Oxytocin was continuously infused in conjunction with tamponade (20 IU for 24 hours). Tamponade balloons were filled with 200 mL in cases of abnormal placentation and 400mL in cases of atony. Women also received antibiotics for 24 hours, and those receiving balloon tamponade after vaginal birth had vaginal packing. Balloons were left in place for a mean of 23.1±9.0 hours (range: 3.5-40 hours). Sixty-three percent of women also received red blood cell transfusion. Balloon tamponade as the initial second-line procedure successfully controlled bleeding in 39 of 52 women (75%, success in 11 of 14 cases of PPH following vaginal birth and 28 of 38 cesarean births). Two women had subsequent uterine artery ligation, one had compression sutures, and 10 had hysterectomies. More failures of balloon tamponade requiring hysterectomy occurred in cases of PPH due to placenta previa and accreta (success in 2 of 5 cases) and in cases due to atony accompanied by placenta previa and/or accreta (success in 3 of 7 cases).

A final retrospective case series reported on 50 women with PPH (n=44) or at risk of PPH (n=6) receiving a Bakri uterine balloon after conservative management including uterotonics, laceration repair, and curettage as needed.⁹⁰ Overall, 29 women had vaginal births and 21 had cesarean births (N primigravid=30). PPH was most often due to placental retention (30% of cases) or vaginal rupture/paravaginal hematoma (22%). Uterine balloons were inserted in the vagina or lower uterine segment and left in situ for a mean of 12.7 hours (range 1-28 hours). Four women had compression sutures or ligation concomitantly with uterine balloon tamponade, and the study reports data on successful control for all women (i.e., not separately for those women who received tamponade alone). In all, uterine balloon tamponade successful controlled bleeding in 43 of 50 women (86%). Three women required subsequent embolization, two required supravaginal uterine amputation, one required compression sutures plus supravaginal uterine amputation, and one had embolization followed by hysterectomy. Because success data are not extractable for women who received uterine balloon tamponade alone, this study is not included in Table 10, which reports rates of successful control of bleeding following uterine tamponade.

Table 10

Success rates after uterine balloon tamponade as the initial second-line procedure.

Table 10 reports rates of successful control of bleeding after uterine tamponade.

Key Points

• No good quality studies addressed embolization.
• Embolization materials, arteries embolized, and interventions used prior to and concomitantly with embolization varied across studies.
• Success (control of bleeding without further procedures or surgeries) rates for embolization as the initial procedure after conservative management ranged from 58 to 98 percent (success in 1251/1435 women), with a median rate of 89 percent.
• Strength of the evidence is low for embolization controlling bleeding without additional procedures or surgeries.

Overview of the Literature

Nineteen studies addressed embolization to treat PPH.^{49, 77, 79, 91-108} Seven studies had explicit comparison groups: one poor quality case-control study⁹¹ and six fair quality cohort studies (reported in multiple publications), five of which were retrospective^{49, 79, 92-96} and one prospective.⁷⁷ Four studies were conducted in France in tertiary care hospitals,^{91, 92, 95, 96}one in Korea,⁴⁹ in a hospital that serves Jehovah's Witnesses, one in the United Kingdom,⁷⁷ which reported data collected via the UKOSS (described in the section on rFVIIa), and one in Hong Kong.⁷⁹ Ten women in one cohort study also had concomitant vessel ligation and/or uterine compression sutures,^92-94 one woman in each of two studies had prior or concomitant artery ligation,^{49, 95} and three in another study⁷⁷ also had intra-arterial balloon placement along with embolization. Eighty-one percent of the cases of PPH reported in the case-control study were due to atony.⁹¹ Rates of atony in the cohort studies ranged from 9 to 69.5 percent. Other causes in all populations included placenta accreta, percreta, and/or previa (range: 9.4 to 22%); thrombus, vascular anomaly, or coagulopathy (range: 2 to 10%); and genital tract lacerations or uterine tears (range: 1to 14%). The case-control study and two retrospective cohort studies reported primarily on longer-term fertility with followup of participants at ≥ 12 months post-embolization (fertility data reported in KQ3).^{91, 92, 95} The prospective cohort study reported primarily success of embolization and the need for additional second-line interventions⁷⁷ as did one retrospective cohort study.⁹⁶ Remaining studies also reported primarily on the rate of success (i.e., controlling bleeding without further procedures or surgical interventions) of embolization.

Twelve retrospective case series also addressed embolization.^97-108 Studies were conducted in France (n = 4), Asia (n = 7), the United States (n = 1). Most cases of PPH were due to atony (range = 43 to 100%), and most studies reported primarily on rates of success. One study¹⁰⁷ reported on embolization to control secondary PPH, and one case series included 50 women in who embolization was performed because of high risk for PPH.¹⁰⁸

Detailed Analysis

One fair quality retrospective cohort study reported in three publications^92-94 included all 101 women who had pelvic artery embolization for PPH from 1994 to 2007 at a tertiary care facility in France. Embolization failed to control bleeding in 11 of 101 women, seven of whom required a postpartum hysterectomy. Failure was associated with increased blood loss as 100 percent of failed cases had blood loss greater than 1500 ml (p < .001). Failure was also associated with increased rate of transfusion with 90 percent of women in whom embolization failed receiving more than 5 units PRBC compared with 43 percent of the successful embolizations (p < .004). Cases of failed embolization were more likely to be complicated by wound infection (27% vs. 6 % in the success group, p < .04).

A second fair quality retrospective cohort study conducted in France assessed outcomes in 52 women undergoing selective embolization using gelfoam (n = 41, mean age = 29.2 ± 4.65 years, 9 primiparous, 11 vaginal births), hysterectomy (n = 6, mean age = 30.1 ± 4.11, 2 primiparous, 2 vaginal births), or both embolization and hysterectomy (n = 5, mean age = 36.6 ± 4.56, 0 primiparous, 0 vaginal births).⁹⁵ All women were treated between 1996 and 2005, and atony was the most frequent cause of PPH across groups (69.5%). All women had medical management (oxytocin, manual placenta removal, uterine massage, prostaglandins, transfusion) prior to embolization or hysterectomy. Embolization successfully stopped bleeding in 41 of 46 cases (89.1%). Five women required additional embolization procedures (insertion of coil to correct injury sustained in cesarean birth, ovarian artery embolization, embolization beyond gluteal artery, embolization of internal iliac artery, embolization of ligated hypogastric arteries). Among five women proceeding to hysterectomy following failed embolization, two women had placenta accreta, one had percreta, and one had sustained arterial injury during embolization. The study also assessed fertility in women who had had embolization (n = 37 available for followup) 2 to 11 years earlier: of the 16 women who desired a future pregnancy, all became pregnant 1 to 11 months following the decision to try to conceive (total of 19 pregnancies in the followup period).

In one fair quality retrospective cohort study reporting outcomes after embolization, ligation, or hysterectomy (see full study description in Ligation section), eight of 61 women with PPH underwent embolization using gelatin sponge or coils as the first secondary procedure.⁹⁶ Embolization failed in three cases: one woman undergoing embolization also required methotrexate, one required subsequent ligation, and one required hysterectomy (63% success rate for embolization alone). This study also reported intervention by cause of PPH: among eight cases treated with primary embolization, three women had PPH due to atony (one cesarean birth). Embolization failed in one case, which resulted in hysterectomy and subsequent death. Embolization was successful in two cases of PPH due to accreta (one cesarean birth) and in one case due to placental abruption (vaginal birth). The procedure failed in one case of PPH due to genital tract laceration (instrumented vaginal birth), leading to subsequent ligation, and successfully controlled bleeding in another case following lacerations.⁹⁶

Another fair quality retrospective cohort study reported outcomes after second-line procedures (see full description in the Oxytocin and Other Uterotonics section) in 42 women with PPH.⁷⁹ Procedures included balloon tamponade (n=12), embolization (n=5), and sutures (n=26), and women receiving second-line therapies did not differ in terms of age, BMI, parity, mode of birth, or causes of PPH. Although five women underwent embolization after the failure of conservative management including oxytocin and other uterotonics, the paper reports etiology only for the four women who had embolization alone (i.e., not followed by another second-line approach). Two women had atony, one had placenta previa, and one had placenta accreta. Embolization successfully controlled bleeding without need for further procedure or surgery in three of the five women receiving embolization (60%). One woman required subsequent compression sutures and one required hysterectomy to control bleeding.

One poor quality case-control study conducted in France assessed the effects of embolization on fertility in 53 women exposed to embolization following PPH and 106 women who had not undergone embolization and were matched on date of birth, age, gravidity and parity, fertility assistance, and mode of birth.⁹¹ Women (mean age = 34.3, range 19-44) had undergone embolization (78.5% using absorbable gelatin, 1.8% using coils, 7.1% using microparticles, 12.6% using gelatin+other) between 2000 and 2006, and the primary cause of PPH was atony (81.1%). Embolization successfully controlled bleeding in 100 percent of women, but three required more than one embolization procedure.

One fair quality prospective cohort study reported UKOSS data collected between 2007 and 2009.⁷⁷ The study reported an analysis of outcomes of second-line therapies (i.e., interventions received after uterotonics alone or with intrauterine tamponade via balloon or packing). Second-line interventions included interventional radiology (defined as embolization or occlusion with an intra-arterial balloon), ligation (of any of the internal iliac, uterine, hypogastric, or ovarian arteries), compression sutures (including B-lynch, modified B-lynch, multiple vertical or horizontal sutures, squared compression sutures, and others), or rFVIIa. Among an estimated 1,237,385 births in the study period, 272 women had PPH treated with the interventions of interest as a second-line intervention. More than 50 percent of PPH cases (53%) were primarily due to atony. Other causes included placenta previa (9%), placenta accreta (10%), uterine tears (13%), and other (15%, includes placental abruption, genital bleeding, amniotic fluid embolism, infection, clotting abnormalities, undetermined causes). Women who had a cesarean birth (n = 230) were treated with a surgical method in 199 (87%) of the cases, and those who gave birth vaginally (n = 42) were more likely to be treated by interventional radiology or rFVIIa (52%, p < 0.001). Among the 272 cases of PPH, 205 women received uterotonics alone, and 67 had uterotonics plus intrauterine tamponade as first-line procedures. Data for each of the second-line therapies addressed in the study are reported under the appropriate intervention type (suture, etc.). Among the 22 women treated with interventional radiology, 19 had embolization alone, two had embolization plus balloon, and one had balloon only. Fourteen of the 22 women received uterotonics prior to interventional radiology. The interventional radiology procedures failed to control bleeding in two women (14%; 95% CI: 0 to 43), who required hysterectomy. Among the eight of 22 women who received uterotonics and intrauterine tamponade prior to interventional radiology, bleeding was controlled in seven cases, and one woman (12%, 95% CI: 0 to 53) required an additional (unstated) intervention. The study does not report the success of embolization alone but only the success of both interventional radiology procedures together.

One fair quality retrospective cohort conducted at a hospital that treated Jehovah's Witnesses in Korea reported results from women treated with embolization or hysterectomy between 2002 and 2009 (see Hysterectomy section for results from that arm).⁴⁹ All women were initially treated with uterotonics (oxytocin, ergots, prostaglandins), uterine massage, transfusion (in patients who were not Jehovah's Witnesses), and fluid replacement. Among the 124 women (eight Jehovah's Witnesses) experiencing primary PPH, 60 (mean age 31.0 ± 4.8 years, 17 primiparous, 23 vaginal births) underwent selective embolization using gelfoam. PPH was most frequently due to atony (92.4%), and mean blood loss prior to embolization was 676.7 ml. Embolizations were performed by the same two interventionists across the study period. Mean ICU stay in the embolization group was 5 days (mean overall LOS = 8.6 days). Two women in the embolization group required hysterectomy due to continued bleeding from the cesarean uterine wound and from vaginal and cervical lacerations after vaginal birth.

In case series, rates of success (control of bleeding after embolization without further procedures or surgeries) ranged from 58 to 98 percent. In some cases, women had a procedure such as ligation or balloon tamponade prior to embolization. Five studies also reported on resumption of menses and/or pregnancies achieved (see discussion in KQ3).

One population-based case series reported on 211 women undergoing embolization either to control ongoing PPH (n=161, mean age=32.4±4.8 years, primipara=47.2%) or prophylactically (n=50, mean age=30.1±6.1 years, primipara=50%).¹⁰⁸ Of note, this study included 56 women (37 in the emergency embolization group and 19 in the prophylactic group) who were <22 weeks gestation at the time of treatment. Most cases of prophylactic embolization were performed for retained placenta (n=37), while most cases of emergency embolization were for atony (n=73). Embolic materials included gelatin sponge in most cases (n=193 cases), but metal coils (n=11) and other materials including N-butyl-2-cyanoacrylate (n=7) were also used. Embolization successfully controlled bleeding in 181 of 211 women (86%); 12 women required a second embolization procedure, and 18 required hysterectomies. Because the study does not clearly report how women who had second embolizations also had hysterectomies, we do not include this study in the success rates in Table 12.One retrospective case series reported on 117 cases of embolization (mean age=32.0±5.0, 69 vaginal births, 56 primiparous) for PPH performed between 2006 and 2013 at a Korean hospital.¹⁰⁶ More than half of the cases of PPH in the embolization group (54.7%) were due to atony, and women were treated initially with fluids, uterotonics, uterine massage, suture of lacerations, and uterine evacuation as needed. Embolization was performed with gelatin particles, coils, glue, or polyvinyl alcohol particles and was successful overall at controlling bleeding without further procedural or surgical intervention in 103 of 117 women (88%). Ten women required additional embolization, and four had hysterectomies. Embolization failure was associated with DIC (OR 3.364, 95% CI: 0.838 to 13.503, p=.08), greater than 10 RBC units transfused (OR 8.011, (95% CI: 1.531 to 41.912, p=.014), and embolization of uterine and ovarian arteries (OR 20.472, (95% CI: 2.715 to 154.365, p=.003). Nineteen of the 117 cases of PPH were secondary (12 cesarean births, p=.03 compared with primary PPH group), and embolization successfully controlled bleeding in 18 of these cases. This study includes data on 20 women who underwent hysterectomy but no outcomes of interest for the current review were reported; thus we did not include the hysterectomy data.

Table 12

Success rates after embolization as the initial second-line procedure.

One retrospective case series included 56 women (median age = 33 years, median gravida = 2, median para = 2) with severe PPH (defined as ≥ 1000mL blood loss via clinical estimation or weighing of blood collecting bag; ≥ 500mL blood loss with poor clinical signs; continued bleeding; need for transfusion; or DIC) undergoing embolization at a French tertiary care hospital between 1995 and 2005.⁹⁷ All women received initial medical treatment including suturing of vaginal or cervical lesions, oxytocin, uterine massage, and sulprostone. Thirty births were vaginal without instrumentation (54.5%), nine were instrumented vaginal (16.5%), and 16 were cesarean (29%). All women had atony, and 36 required transfusion (64.3%). Embolization was performed with gelfoam or sponge. Embolization successfully stopped bleeding in 55 cases (98% success rate). One woman required a second embolization session to control bleeding, and none needed further surgical interventions for bleeding.

Another French retrospective case series including 113 women (mean age = 31 years, 67 cesarean births) reported on menses and fertility outcomes and success of the embolization procedure.⁹⁸ PPH was most frequently due to atony (75% of cases), and all women received medical management prior to embolization. Embolization materials included gelatin sponge, powder, and microparticles. Eighteen women required surgery prior to embolization (sutures, n = 11; ligation, n = 7). Embolization successfully controlled bleeding in 111 cases (results not reported for women who had embolization without a prior surgical procedure). Two women required hysterectomy post-embolization.

In a Korean retrospective case series reporting on 251 women with primary PPH (mean age 32 ± 4 years, 139 nulliparous, 141 vaginal births), most cases of PPH were due to atony (78.9%).¹⁰¹ The study reviewed data from women treated between 2000 and 2011. All women had medical management prior to embolization, and 22 had surgical interventions prior to embolization (hysterectomy, n = 15; uterine artery ligation, n = 2; laparotomy, n = 2; suture or uterine wall repair, n = 2; dilatation and curettage, n = 1). Embolization was performed with gelatin sponge or multiple particles. Embolization successfully controlled bleeding in 201 of the 229 women for whom embolization was the first second-line procedure (88%). Among all 251 women, embolization successfully controlled bleeding in 217 (87%). Twelve women required a repeat embolization (success in nine cases, one hysterectomy, one laparotomy, one death), nine required hysterectomy, six required laparotomy (one death), three required additional conservative management, one required uterine artery ligation, and three died after the first embolization session. Successful embolization was associated with vaginal birth, absence of DIC, and absence of need for transfusion of > 10 PRBC units (p values < .05).

A retrospective review of embolization for PPH conducted at two Korean hospitals between 2006 and 2011 included data from 176 women (mean age = 33.9 years, 105 vaginal births, 73 primiparous) undergoing 189 embolization procedures.¹⁰⁵ Women who had cesarean births were significantly older than those with vaginal births (p = 0.035). Twenty-five cases of PPH were secondary, and overall, PPH was most frequently due to atony (57.6% of cases). Embolizations were done with gelatin sponge, particles, coils, or a combination. Bleeding successfully stopped after embolization in 158 cases (89.7%). Twelve women needed a repeat embolization, 11 needed a surgical procedure (five hysterectomies), and one needed vascular ligation.

One retrospective case series reporting data from a U.S. tertiary care hospital included 76 women (mean age = 33 years, 18 cesarean births) who had PPH.⁹⁹ Ten women were excluded from analysis because they had interventions prior to or concomitant with embolization or had an ectopic pregnancy. Embolization (performed with gelfoam and/or coils) successfully controlled bleeding without further procedures or surgeries in 63 of 66 women (95%). Three women required a subsequent hysterectomy. Embolization was successful in 98% (49/50) of the women with primary PPH and 88% (14/16) of the women who had secondary PPH (presentation 4 to 72 days post-birth, mean = 25 days). Women required a mean 0.4 units PRBC after embolization, and the mean hospital stay overall was 3.5 days (range 1-12 days). Among those with primary PPH, mean hospital stay was 3.9 days and was 2 days in the secondary PPH group.

One Japanese retrospective case series included data from 55 women (median age 33 years, 34 vaginal births, median parity = 1, range 0-3) with PPH treated with embolization between 2003 and 2013.¹⁰⁴ Most cases of PPH were due to atony (n = 41), and all women had initial conservative management including uterine massage, packing, and uterotonics. The embolization material was gelatin sponge, and embolization successfully stopped bleeding without an additional intervention in 46 women (84%). Bleeding stopped in two women who went on to hysterectomy after embolization due to uterine necrosis. The study does not report the interventions performed for the other seven women who required another procedure after embolization. Advanced maternal age and retained placenta were independent risk factors for failure of embolization (OR 1.46, 95% CI: 1.12 to 2.18 and OR 15.48, 95% CI: 2.04 to 198.12, respectively).

One French retrospective case series reported outcomes among 102 women (mean age 31.8 ± 5.9 years, 82 vaginal births, mean parity 2.01 ± 1.11) undergoing embolization at an academic medical center between 1998 and 2002.¹⁰³ Women may have had medical management including uterine massage and oxytocin prior to embolization. PPH was due to atony in 43 percent of women. Mean ICU stay was 2.07 ± 1.2 days, and units of whole blood, platelets, and fresh frozen plasma transfused ranged from 0 to 31. Embolization was successful without further surgical procedure in 59 women. Fourteen women required a second embolization to control bleeding, and 29 required surgery (nine laparotomies, two uterine artery ligations, seven hysterectomies, 11 genital tear repairs plus subsequent embolization). Embolization was more successful in women with vaginal births (success in 63/81 vaginal births) compared with cesarean (success in 11/21 cesarean births, p = 0.017; OR for poor outcome associated with cesarean birth: 0.16, 95% CI: 0.04 to 0.5). Atony as the cause of PPH was also associated with greater success (success in 39/44 women; OR 4.13, 95% CI: 1.35 to 12.6).

Another retrospective case series conducted in a French tertiary care hospital reported on success rates for embolization in 98 women with PPH (33 considered “major” PPH, defined as change in peripartum hemoglobin level of ≥ 4 g/dL and/or hemodynamic instability and/or hypovolemic shock).¹⁰² All women had treatment (resuscitation, uterotonics, manual placenta removal, surgical repair of tears as indicated) prior to embolization, and most cases of PPH were due to atony. Forty-five women had vaginal births, 14 had instrumented vaginal births, and 28 had cesarean births. Embolization was performed with gelatin sponge pledgets and coils as needed. Twenty-six women had a surgical procedure prior to embolization (vaginal or cervical suture, n = 17; uterine suture, n = 1; artery ligation, n = 3; hysterectomy, n = 9; packing, n = 2). Embolization successfully controlled bleeding in 90 of the 98 cases of PPH. Women in whom PPH failed to control bleeding required subsequent uterine suture (n = 4), laparotomy for vessel ligation (n = 2), and repair of genital tears (n = 2). Embolization plus uterine sutures failed in three cases, leading to hysterectomy.

In another large retrospective case series from Korea, 257 women (mean age = 32 years, 162 primiparas, 112 cesarean births) underwent embolization for PPH between 2004 and 2011.¹⁰⁰ PPH was most often caused by atony (n = 156 cases), and embolization materials included gelatin sponge, N-butyl-cyanoacrylate, or both. Nineteen cases of PPH were secondary. Nine women had a surgical procedure prior to embolization (eight hysterectomies, one artery ligation). Embolization successfully stopped bleeding in 233 women overall (91%). In the 248 women for whom embolization was the first second-line procedure, embolization was successful in 226 (91%). Women for whom embolization failed to control bleeding were more likely to have DIC (OR 6.57, 95% CI: 1.60 to 26.9, p = .009), and the rate of major complications was significantly greater among failed embolizations vs. successful (9.4% vs. 37.5%, p < .01).

Finally, one retrospective case series conducted in Korea included 52 women (mean age 31.6 years, range=25-40) with secondary PPH.¹⁰⁷ Bleeding began a median 10 days post-birth (range 1-39 days) and was most frequently related to retained placental tissue (44.2% of cases). All women had initial conservative management prior to embolization, which was conducted with gelatin particles, N-butyl cyanoacrylate, and/or microcoils. Embolization successfully controlled bleeding without further procedure or surgery in 47 of 52 women (90.4%). In univariate analyses, successful control of bleeding was not associated with obstetric characteristics, mode of birth, onset of bleeding post-birth, length of stay, amount of transfusion, or cause of bleeding (all p values=ns). One woman needed repeat embolization, one had further conservative management, and three women had subsequent hysterectomy. In the 44 women available for followup at a mean of 12.6 months post-procedure (range 1-62 months), all women had regular menstruation and five had pregnancies, although the number desiring pregnancy was not reported. The investigators note that no complications occurred. Table 11 outlines key outcomes in all studies of embolization.

Table 11

Key outcomes in studies of embolization.

Key Points

• No good quality studies addressed uterine compression sutures.
• In one fair-quality prospective cohort study, sutures were effective in controlling bleeding without further procedures or surgeries in 140 of 199 women, all of whom received uterotonics and/or intrauterine balloon tamponade prior to sutures (70% success rate). Sutures were successful in 15 of 21 women in another study (71%).
• Strength of the evidence is insufficient for the success of uterine compression sutures in controlling bleeding given the few studies available.

Overview of the Literature

Three studies addressed uterine compression sutures, one prospective cohort study (reported in two publications), one retrospective cohort study, and two retrospective case series.^{77, 79, 82, 111} The prospective cohort study, rated as fair quality, reported data collected via the UKOSS.^{77, 82} Two-hundred and eleven cases of PPH were treated with sutures in the study period. One retrospective cohort study reported on 26 women with massive PPH in Hong Kong.⁷⁹ The case series reported data from interventions performed by a single surgeon in Argentina¹¹¹ The study reports on 539 cases of PPH treated with ligation or suture and does not clarify how many women received each technique. Two additional studies of compression sutures reported harms outcomes only and are described under KQ3.^{62, 66}

Detailed Analysis

One fair quality prospective cohort study reported UKOSS data collected between 2007 and 2009.⁷⁷ The study reported an analysis of outcomes of second-line therapies (i.e., interventions received after uterotonics alone or with intrauterine tamponade via balloon or packing. Among women who were initially treated with uterotonics alone, 161 went on to require compression sutures, which were successful in controlling bleeding in 120 cases (74.53% success rate). Twenty-five women required hysterectomy (without another intervening procedure) after sutures. Three women had ligation after suture; seven had either embolization or balloon placement (three of these went on to require hysterectomy); and six had rFVIIa (four ultimately required hysterectomy). Thus, compression sutures with or without subsequent procedures failed to control bleeding in 32 women, leading to hysterectomy. Among 38 women who required sutures after failure of uterotonics plus intrauterine tamponade, 14 went on to require hysterectomy (eight immediately, two after ligation and/or rFVIIa, two after interventional radiology and/or rFVIIa, and two after rFVIIa alone). Overall (among women who received uterotonics and intrauterine tamponade), sutures successfully controlled bleeding in 70 percent of cases (n = 140/199 cases)⁷⁷

Another publication from this study,⁸²which includes data from the majority (n = 199/211) of the participants who received sutures described above,⁷⁷ reported on 211 women receiving compression sutures (B-lynch, n = 79; modified B-lynch, n = 48; other, including square sutures or combination sutures, n = 32; unspecified, n = 52) to treat PPH in the study period. The most common reason for the hemorrhage was uterine atony (n = 129, 61%). As in the first study, all women had prior uterotonic treatment either for prophylaxis or treatment of PPH. Ten women had embolization or ligation, 41 had uterine balloon or packing, and two had rFVIIa prior to sutures. Embolization or ligation following sutures was required in 18 cases, rFVIIa in nine, and uterine packing or balloon in 25. Overall, sutures as the initial second-line therapy failed to control bleeding, leading to subsequent hysterectomy, in 46 cases and successfully controlled hemorrhage in 153 cases (sutures were not the initial second-line therapy in 12 cases). Fifty-two women (25%) of all women (those who received sutures as the initial second-line therapy and those who received sutures in combination with or after another second-line procedure) required hysterectomy to control bleeding. More women who required an additional second-line intervention went on to require hysterectomy (OR 3.09, 95% CI: 1.46 to 6.56).

In a fair quality retrospective cohort study (see full description in Oxytocin and Other Uterotonics section), 42 of 91 women with massive PPH required second-line procedures or surgeries to control bleeding.⁷⁹ A total of 26 women received sutures (including B-Lynch, Hwu, Cho square, and Hayman), 21 of whom received sutures alone, and five of whom also had sequential embolization. In the 21 women receiving sutures alone, bleeding was successfully controlled in 15 (71.4%). Six women required subsequent hysterectomy. None of the women who had both sutures and embolization required hysterectomy. One retrospective case series reported data on 539 cases of PPH treated with either uterine sutures or arterial ligation in hospitals in Argentina between 1989 and 2009.¹¹¹ Sutures were placed by a single surgeon, and suture types included B-lynch, Cho, Hayman, and Pereira. The number of sutures compared with ligations, and potential overlap between interventions, is not clear. Overall, the study reports cessation of bleeding in 499 cases. Forty women required hysterectomy, but whether this occurred after suture or ligation or a combination is not clear. B-lynch sutures were reported as successful in 81 of 86 cases, Hayman sutures in 34 of 37, Cho sutures in 281 of 313 cases, and Pereira in 11 of 11 cases, but again, prior or subsequent interventions are not clear. Because the number of women who received sutures as the initial second-line intervention is clearly reported in only two studies,^{77, 79, 82} we do not include a success rate table for uterine compression sutures. Table 13 outlines data from studies with comparison groups.

Table 13

Key outcomes in studies of uterine compression sutures.

Key Points

• No good quality studies addressed uterine and other pelvic artery ligation (hereafter, ligation).
• Rates of successful control of bleeding without further procedures or surgeries ranged from 36 to 96 percent with a median of 92 percent in three studies.
• Strength of the evidence is low for ligation controlling bleeding without further procedures or surgeries.

Overview of the Literature

Five studies reported data on ligation.^{77, 96, 111-113} Studies include two fair quality cohort studies, one conducted in the U.K.,⁷⁷ and one in France.⁹⁶ In the prospective study of ligation of pelvic vessels (unspecified), 25 percent of cases of PPH were due to atony, 30 percent due to uterine tears, 20 percent due to accreta, and 25 percent due to other causes, and most women were under age 35 (60%).⁷⁷ Nearly 40 percent of cases of PPH in the retrospective cohort study, which included cases of bilateral hypogastric artery ligation, were due to atony, and participant age was not reported.⁹⁶ Studies primarily reported rates of success for ligation. Three retrospective case series also reported data on ligation: one reported cases of bilateral uterine artery ligation or selective pelvic pedicle ligation, performed by a single surgeon in Argentina,¹¹¹ one reported on outcomes after uterine artery ligations over 30 years in a U.S. center,¹¹² and the final study reported on triple uterine artery ligation conducted over 9 years in France¹¹³. Case series primarily reported success rates and provide little data on participant characteristics.

Detailed Analysis

Outcomes of ligation were reported in a fair quality UKOSS cohort study described fully above.⁷⁷ Fourteen women required vessel ligation as second-line procedure following uterotonics alone. Ligation successfully controlled bleeding in five women, and five required sutures (followed by hysterectomy in three), two required rFVIIa (followed by hysterectomy in one), and two required hysterectomy immediately after ligation. Six women had ligation after uterotonics and intrauterine tamponade failure, and three went on to hysterectomy to control bleeding (two after sutures plus rFVIIa, one after sutures alone).⁷⁷

Another fair quality retrospective cohort study reported data from women with PPH admitted to a French ICU between 1983 and 1998 and included some data on future fertility.⁹⁶ Sixty-one cases of PPH occurred in the time period, 48 of which were treated with bilateral ligation of the hypogastric arteries, eight with embolization using gelatin sponge or coils, and five with hysterectomy as the primary procedure. Across groups, 39 women required transfusion of four or more blood units. Most of the 56 women requiring either ligation or embolization as a primary procedure had cesarean births (n = 41). The women requiring primary hysterectomy all had hemorrhagic shock. The primary procedure failed in eight cases (described under each intervention). Among the 48 women undergoing primary ligation, four required hysterectomy to correct bleeding (92% success rate for primary ligation). This study also reported intervention by cause of PPH: 20 women had PPH due to atony and received ligation as the primary intervention. Nineteen of these 20 had cesarean births (elective or emergency). Ligation was successful in controlling bleeding in 18 of 20 cases, with two women requiring subsequent hysterectomy (one vaginal birth and one cesarean birth). Eleven women (10 cesarean births) had PPH due to accreta. Ten ligations were successful in this group; one woman who had a cesarean birth required hysterectomy and subsequently died. Seven women had PPH due to genital tract laceration (seven vaginal births, 4 instrumented), and ligation was successful in all cases. Six women had placental abruption (six cesarean births), and ligation was successful in all cases. Two women had uterine rupture or pre-rupture (two cesarean births) with bleeding controlled successfully by ligation in both cases. Two women had PPH due to uterine artery injury, presumably incurred during cesarean birth. Ligation successfully controlled bleeding in one case, and the other women died. Finally, one woman with a cesarean birth had PPH related to placenta previa. Ligation failed to control bleeding, leading to subsequent hysterectomy.⁹⁶

One French retrospective case series included 56 women with PPH (median age=31.5, median parity=0.5) who underwent triple uterine artery ligation with (n=43) or without (n=13) concomitant uterine compression sutures.¹¹³ The PPH treatment protocol in the hospital studied included oxytocin followed by sulprostone followed by ligation as needed, sutures as needed, and other procedures including hysterectomy or embolization if bleeding remained uncontrolled. Most cases (80.4%) of PPH were due to atony. All women received initial oxytocin, and 83.9 percent also received sulprostone. Overall, ligation alone and ligation with suture controlled bleeding in 51 of 56 women (91.1%). Four women had a subsequent hysterectomy and one required embolization. Failure of ligation with or without suture occurred more often in cases of PPH due to accreta (4 cases) compared with atony (1 case, p=.0004, [OR for failure of ligation ± suture=15.07, 95% CI: 1.12 to 201.9, p=.041]). Ligation with or without suture was also significantly less likely to fail when women had first received sulprostone (p=.025).

One retrospective case series reported data on 539 cases of PPH treated with either uterine sutures or arterial ligation in hospitals in Argentina between 1989 and 2009.¹¹¹ Interventions were conducted by a single surgeon. The number of sutures compared with ligations, and potential overlap between interventions, is not clear. Overall, the study reports cessation of bleeding in 499 cases. Forty women required hysterectomy, but whether this occurred after suture or ligation or a combination is not clear. Ligation was reported as successful in 68 of 105 cases, but again, prior or subsequent interventions are not clear.

Another retrospective case series reviewed data from 29 years (1963-1992) of ligations performed in a U.S. hospital.¹¹² Women received initial medical therapy including uterotonics, and 265 underwent bilateral uterine artery ligation after cesarean birth. Atony accounted for most cases of PPH across the study period (n = 135), and the rate of PPH treated with ligation declined across decades (n = 124, 60, 81 per each decade from 1963-1992). Overall, ligation failed to control bleeding in 10 women, eight of whom had abnormal placentation. Six of these 10 women had total hysterectomies, three had sutures, and one had ovarian artery ligation. Most treatment failures (n = 7) occurred in the first decade reviewed. The study reports that menstrual flow was not affected, but method and timing of followup is not clear. Table 14 outlines key outcomes of studies.

Table 14

Key outcomes in studies of uterine and other pelvic artery ligation.

Ligation Success Rates

Ligation was performed on multiple sites (e.g., internal iliac, uterine arteries) within and across studies, and rates of successful control of bleeding ranged from 36 to 96 percent with a median of 92 percent (Table 15).

Table 15

Success rates after uterine and other pelvic artery ligation as the initial second-line procedure.

Key Points

• One study compared embolization and hysterectomy.
• Embolization failed to control bleeding in 20 cases (18%), leading to 17 hysterectomies.
• Women in the hysterectomy group had significantly more ICU admissions compared with the embolization group (RR 1.6, 95% CI: 1.1 to 2.4) and had a greater median length of stay (LOS, 10 days vs. 7 days).
• Strength of the evidence was low for embolization controlling bleeding without additional procedures or surgeries and insufficient for the effects of hysterectomy.

Overview of the Literature

One fair quality prospective cohort study conducted in the Netherlands¹⁰⁹ compared outcomes following embolization or hysterectomy. The 205 women in the study most frequently had PPH related to atony (33%), and 43.4 percent were age 40 or older.

Detailed Analysis

One fair quality cohort study (Table 16) conducted in the Netherlands (LEMMoN: Nationwide Study into Ethnical Determinants of Maternal Morbidity in the Netherlands) prospectively collected data on severe maternal morbidity from all 98 Dutch maternity hospitals between 2004 and 2006 using a standardized collection form.¹⁰⁹ Two hundred and five women required either embolization (n = 114) or hysterectomy (n = 108) or both (n=17) during the study period. More than 40 percent (43.4%) of women in both groups were age 35 or older, 39.5 percent were nulliparous, and 49.8 percent had cesarean births. The most frequent cause of PPH in the embolization arm was atony (33%) and disorders of placentation (placenta previa, morbidly adherent placenta) in the hysterectomy group (35%). Women in both arms had other interventions prior to either embolization or hysterectomy including oxytocin ( > 80% of both groups); sulprostone ( > 50% of both groups); plasma replacement, frozen plasma, or red blood cell transfusion ( > 78% of both groups); and other surgical interventions including arterial ligation, B-lynch suture, inspection (6 women in embolization and 11 in hysterectomy groups).

Table 16

Key outcomes in studies of embolization and hysterectomy.

Embolization failed to control bleeding in 20 cases (18%): 17 women in the embolization group also ultimately required hysterectomy to control PPH (two of these were due to uterine necrosis) and one case was resolved with balloon tamponade. In sub-analyses of these failed cases, embolization had a failure rate of 25 percent following cesarean birth. Women in the hysterectomy group required more transfusions (median 14 vs. 10, p = 0.002) and more massive transfusions (≥ eight units of red blood cells) compared with women undergoing embolization (RR 1.5, 95% CI: 1.1 to 2.1); however, timing of transfusion (i.e., pre- or post-embolization or hysterectomy) is not clear. Women in the hysterectomy group also had significantly more ICU admissions compared with the embolization group (RR 1.6, 95% CI: 1.1 to 2.4) and had a greater median LOS (10 days vs. 7 days).¹⁰⁹

Key Points

• Two of eight studies reported data to calculate control of bleeding without additional procedures or surgeries. In these two studies bleeding was controlled after hysterectomy as the initial second-line intervention in a median of 57 percent of cases.
• In one case series analyzing data by hospital volume, there was no difference in transfusion, intraoperative injury, length of stay, or medical complications based on hospital volume after adjusting for age, race, hospital size, year of diagnosis, and hospital type.
• Strength of the evidence is insufficient for all hysterectomy outcomes given the few studies available.

Overview of the Literature

Eight studies reported outcomes of hysterectomy.^{45, 49, 96, 110, 114-117} Studies included two retrospective cohort studies of fair quality, one conducted in France (n=10)⁹⁶ and the other in Korea (total n = 61).⁴⁹ Atony accounted for 75 percent of the 61 cases in one study,⁴⁹ while PPH in the 10 women undergoing hysterectomy in the second was due to genital tract lacerations in three cases, atony in three cases, placenta accreta or previa or placenta abruption in three cases, and uterine rupture in the final case. Four population-based case series also reported on outcomes following hysterectomy. Case series were conducted in Canada,¹¹⁴ Denmark,¹¹⁵ the U.K.,¹¹⁶ and the United States.⁴⁵. One retrospective case series reported on 55 peripartum hysterectomies conducted at one U.S. hospital.¹¹⁷ Finally, another retrospective case series conducted at a university hospital in the U.K. and including data from 52 cases of PPH also reported risk factors for hysterectomy.¹¹⁰Participant ages ranged from 14 to 54 years in the studies reporting age,^{45, 110, 114} and PPH was typically due to atony (range 30 to 56% of cases) or placenta previa or accreta (range: 20 to 38% of cases). One additional case series assessing hysterectomy reported only harms data and is addressed in KQ3.⁶⁴

Detailed Analysis

In one fair quality cohort study including women undergoing embolization (results described in embolization section) or hysterectomy, all women were initially treated with uterotonics (oxytocin, ergots, prostaglandins), uterine massage, transfusion (in patients who were not Jehovah's Witnesses) and fluid replacement.⁴⁹ Among the 124 women (eight Jehovah's Witnesses) experiencing primary PPH, 61 (mean age 31.8 ± 4.0 years, 22 primiparous, 33 vaginal deliveries) underwent hysterectomy. PPH was most frequently due to atony (75.4%), and mean blood loss prior to procedure was 1288.3 ml. Significantly more women in the hysterectomy group had DIC, hypotension, elevated heart rate, greater blood loss before intervention, and greater total transfusion requirements than in the comparison arm of women undergoing embolization (all p values < 0.001). Mean total LOS was 11.5 days. Thirty-nine women in the hysterectomy group required ICU care; however, the study does not report mean ICU stay. Fifty-seven women in the hysterectomy group required transfusion after surgery, and four also required embolization post-hysterectomy.

In another fair quality retrospective cohort study reporting outcomes after embolization, ligation, or hysterectomy (see full study description in Ligation section above), five of 61 women received hysterectomy as the primary procedure. The women requiring primary hysterectomy all had hemorrhagic shock, and the procedure was not successful at controlling bleeding in four cases. One woman also required subsequent embolization. This study also reported intervention by cause of PPH: hysterectomy was the primary procedure in three cases of PPH due to genital tract laceration (three vaginal births). As noted, one woman required subsequent embolization, and the other two died. Similarly, one woman who had a cesarean birth died after hysterectomy for PPH due to uterine rupture. Hysterectomy successfully controlled bleeding in one case of PPH due to placental abruption.⁹⁶

One population-based case series reported on outcomes following peripartum hysterectomy due to PPH.¹¹⁶ In this study there were 315 cases of PPH that resulted in hysterectomy identified via UKOSS between 2005 and 2006. The median ICU stay was 2 days. Sixty-two women had a return to the operating room for a second surgery after hysterectomy. Fourteen percent of these women had a second surgery due to continued bleeding and 6 percent had return due to damage to other organs during hysterectomy. The median number of blood units transfused ranged from nine to 12 depending on etiology of transfusion.

Another population-based case series from the United States was conducted with data from a nationwide validated database that collected quality and resource utilization data (Perspective) data from 500 facilities in the United States.⁴⁵ The main hypothesis of this study was that hospital volume affects outcomes of postpartum hysterectomy. Among the 2,209 patients identified, overall maternal mortality was 1.2 percent among low, intermediate, and high volume facilities, reoperation rates were 3.2 to 6.4 percent (p = 0.02). Intensive care use rates were 45 percent, 39.6 percent and 27.4 percent for low, medium and high-volume institutions, respectively (p < 0.001). The mean length of stay was 3.5 to 4.1 days. After adjusting for age, race, hospital size, year of diagnosis and hospital type, there was no difference in transfusion or length of stay based on hospital volume. Perioperative death was higher at low volume facilities (1.8% compared with 0.9 and 0.8% at medium and high volume hospitals, p = 0.02). Adjusted OR for perioperative death was 0.22 at high volume facilities.

A population-based case series in Denmark collected peripartum hysterectomy data from 1995 to 2004 using the Danish Medical Birth Register, which records information on all births in the country since 1973.¹¹⁵ Peripartum hysterectomy was defined in this study as a hysterectomy taking place immediately after and up to one month after birth. Out of 653,482 births, there were 152 peripartum hysterectomies to control hemorrhage; thirty percent of cases of PPH were due to atony. Prior to hysterectomy, 80 percent of women received oxytocin, 73 percent prostaglandins, 43 percent misoprostol, and 43 percent ergot alkaloid. Ligation was performed in 21 percent of patients and B-lynch suture was also done in 21 percent prior to hysterectomy. Hysterectomy was more often performed after cesarean birth (n = 101, RR for hysterectomy after cesarean compared with vaginal birth = 11.1, 95% CI: 7.9 to 15.6, p < .0001). Sixteen women (11%) needed reoperation.

An additional population-based case series reported on all cases of postpartum hysterectomy done between 1999 and 2006 in a Canadian hospital.¹¹⁴ All obstetric care in the region is linked to a regional database. Investigators identified all hysterectomies that occurred within 24 hours of birth. A total of 87 peripartum hysterectomies were performed in the study period, a rate of 0.8 per 1,000 births. Thirty-four percent of women in the series had placenta previa or accreta. All women received uterotonics prior to hysterectomy, and 86 percent received blood transfusion. Pelvic vessels were ligated in 33 percent of cases. B-lynch suture was done 3 times. Forty-six women (53%) were admitted to the ICU, and mean length of stay after birth was 6 days (range 2 to 16). Eighty-one percent of hysterectomies took place after cesarean birth (n = 70).

Two retrospective case series reported on emergency hysterectomy outcomes and were conducted in the U.K.¹¹⁰ and the U.S.¹¹⁷ In the U.K. series, most (n=50/52) women had primary PPH and all had numerous interventions, including uterotonics, packing, balloon tamponade, and sutures, prior to hysterectomy to control bleeding.¹¹⁰ In multivariate analyses, multiparity, placenta previa, primary PPH, and failed induction were significant risk factors for hysterectomy (all p values <.02). The U.S. series reported on 55 peripartum hysterectomies (17 vaginal births, 38 cesarean; mean age=29 ± 6.8), typically for PPH due to atony (56.4% of cases).¹¹⁷ Mean overall length of stay was 11 ± 7.9 days, mean number units transfused was 6.9±5.3, and mean estimated blood loss was 3325.6 ± 1839.2 mL. Table 17 outlines outcomes.

Table 17

Key outcomes in studies of hysterectomy.

Pharmacologic Interventions

Tranexamic acid. In an RCT that compared women who received tranexamic acid with women who did not (n = 72 per group), serious side effects did not differ between the two groups. Two women in the tranexamic acid group and one in the control group had deep vein thrombosis (p = 0.37). None of the women experienced renal failure, seizures, or death. Mild, transient adverse effects occurred more often in the tranexamic acid group than in the control group (24% vs. 6%, p = 0.03). These side effects included nausea and vomiting (15% vs. 2%, p = 0.002), phosphenes (11% vs. 3%, p = 0.02), and dizziness (6% vs. 4%, p = 0.28). The trial was not adequately powered to report safety but was good quality for harms reporting.⁶⁹

Sulprostone. In one population-based case series of 1,370 women treated with sulprostone, 51 women (3.7%) experienced at least one side effect.⁷⁰ These side effects included digestive effects (n = 34), hyperthermia and chills (n = 7), cardiac effects (n = 5), high blood pressure (n = 2), respiratory effects (n = 2), and dizziness (n = 2). The cardiac side effects (tachycardia, n = 1; atypical chest pain, n = 1; ischemia, n = 3) were considered severe by the investigators and resolved with cessation of sulprostone. Other severe harms included acute hypertension in one woman and acute cyanosis in a woman with asthma, both of which also resolved with cessation of sulprostone. This study, which is part of family of studies reporting on a systems-level intervention for PPH,^{37, 70, 81} was rated as poor quality for harms reporting.

Methylergonovine maleate. One cohort study (rated good quality for harms reporting) used data from U.S. hospital admissions collected over 4 years to identify women who had been given methylergonovine maleate during hospitalization for birth (n = 139,617) and those who had not (n = 2,094,013).⁶⁰ The study compared rates of myocardial ischemia and infarction in the exposed and unexposed women. Six women in the methylergonovine maleate group and 52 in the non-methylergonovine maleate group had an acute coronary syndrome (composite of acute myocardial infarction and unstable angina). The adjusted relative risk of developing an acute coronary syndrome associated with methylergonovine maleate exposure was 1.67 (95% CI: 0.40 to 6.97), and the risk difference was 1.44 per 100,000 patients (95% CI: -2.56 to 5.45). Four women in the methylergonovine maleate group and 44 in the non-exposed group had an acute myocardial infarction (RR for infarction associated with methylergonovine maleate = 1.00m 95% CI: 0.20 to 4.95, risk difference per 100,000 patients = 0, 95% CI: -3.47 to 3.47).

Carboprost tromethamine. One-fifth (n = 48/237) of the participants in a population-based case series experienced a side effect attributed to the drug. Harms reported included diarrhea (11.4%), elevated blood pressure (6.8%), vomiting (6.8%), elevated temperature (2.1%), flushing (1.7%), and tachycardia (1.7%). Quality for the reporting of harms was assessed as poor.⁷¹

Recombinant activated factor VIIa (rFVIIa). Five studies (one good⁷⁶ and four poor quality for harms reporting^{73-75, 80}) with rFVIIa as an intervention reported harms. Two women who received rFVIIa in a retrospective cohort study⁷³ (n = 26) experienced adverse events that may be related to the medication. These included pulmonary edema (n = 1) and PE (n = 1). Neither of these events occurred in women who did not receive rFVIIa (n = 22), but this may be due to the small sample size rather than evidence of an effect of the medication.⁷³ One case-control study reported one case of acute respiratory distress syndrome (ARDS) among the six women who received rFVIIa. There were no long term sequelae, though exact long term complications of interest were not described.⁷⁴ In a population-based case series, adverse events potentially related to rFVIIa in the 92 women to whom it was administered included thromboembolism (n = 4; 2 had PE, one had bilateral ovarian vein thrombosis, and one had a thrombus involving the jugular and subclavian vein, upper arm, and axilla that was not thought to be related to rFVIIa), myocardial infarction (n = 1), and allergic reaction (n = 1). None of these events occurred in women who did not receive rFVIIa (n = 16), but this may be due to the small sample size.⁷⁵

Two studies reported data from the same rFVIIa registry for differing time periods; however, because the overlap between studies is not clear, we report these studies separately. In one study, rated as good quality for harms reporting, and including 105 women with PPH, adverse events potentially related to rFVIIa included cerebrovascular accident (n = 1), deep venous thrombosis (n = 1), and pulmonary embolism (n = 1).⁷⁶ The other study reporting data from this registry included 175 cases of rFVIIa use for PPH and reported that 15 women (8.6%) had thromboembolic adverse events, the most common of which were venous thrombosis among five women (2.9%), disseminated intravascular coagulation in nine (5.1%), and other thrombosis in three (1.7%). There were two arterial thrombotic events including one (0.6%) myocardial infarction. ⁸⁰

Transfusion for Supportive Management of PPH

Seven studies reported harms of transfusion for PPH management.^{37, 61, 63, 65, 70, 81, 83, 84, 86} One retrospective cohort study included 659 women who received whole blood transfusion, 593 who received packed red blood cells (PRBC) only, and 288 who received a combination of blood products. There was a significant difference in the number of women who experienced acute tubular necrosis (0.3% whole blood only vs. 2% PRBC only vs. 4% combinations), acute respiratory distress (0.5% vs. .3% vs. 2%), pulmonary edema (7% vs. 4% vs. 14%), and hypofibrinogenemia (0.2% vs. 0.3% vs. 16%).⁸³ In another retrospective cohort study, there were no thrombotic complications or adverse reactions to cryoprecipitate or fibrinogen concentrate among 34 women receiving either treatment.⁸⁴ In a population-based case series addressing the thromboembolic risk associated with severe PPH and blood replacement therapies in 317 women with severe PPH (defined as uterine bleeding in the first 24 hours after birth, persisting after manual exploration of the uterine cavity and requiring IV uterotonics with a decrease of hemoglobin > 40g/l^-1, or > 4 U RBCs, hemostatic intervention or death), none of the women developed symptomatic deep vein thrombosis (DVT) or PE.⁶¹ Three women developed superficial venous thrombosis (SVT). Severe PPH or packed RBC unit transfusions were found to be a risk factor for SVT. Other variables, such as cesarean birth, absence of low molecular weight heparin use, pre-eclampsia, severe pre-eclampsia, HELLP syndrome, placenta abruption, pregnancy loss, unexplained pregnancy loss, or F12C46T polymorphism were found to be significant risk factors for SVT. In one report from a larger, systems-level RCT ^{37, 70, 81} that included 660 women who received a transfusion, five transfusion-related adverse events (not described) occurred. The investigators considered one case of pulmonary edema to be a severe harm.⁸¹ A pre-post study comparing transfusion with a combination of red blood cells, fresh frozen plasma, and platelets vs. a combination of red blood cells, platelets, and fibrinogen concentrate in 93 women with PPH reported the development of transfusion-associated circulatory overload in four women in the non-fibrogen period and none in the fibrinogen period (p=.04).⁸⁶

Another retrospective case series including 104 women requiring transfusion for PPH reported pulmonary complications in 2.8 percent of women and cardiac complications in 1 percent but did not describe complications further.⁶³ A final series included 71 women with PPH and assessed the risk of developing transfusion-related acute lung injury (TRALI) associated with transfusion.⁶⁵ Of these 71 women, 13 met criteria for a diagnosis of TRALI as they developed new-onset hypoxemia within 6 hours of transfusion without cardiogenic or other cause, and one woman met criteria for possible TRALI with the same symptoms but an alternative risk factor as a possible cause of symptoms. Women with pregnancy-related hypertensive disorders were more likely to develop TRALI (36% vs. 5% in the TRALI vs. no TRALI groups, p=0.006). Age, smoking status, pre-existing morbidities, non-pregnancy related hypertensive disorders, parity, caesarean section, and the need for surgical intervention were not associated with the development of TRALI.

We rated one study as good quality for harms reporting,⁶⁵ and six as poor quality for harms reporting.