Journal of Perinatology (2009) 29, 422–427 r 2009 Nature Publishing Group All rights reserved. 0743-8346/09 $32 www.nature.com/jp ORIGINAL ARTICLE Identifying women most likely to benefit from prevention strategies for postpartum hemorrhage

MC Lu1,2,3, LM Korst4, M Fridman5, E Muthengi2 and KD Gregory3,4,6 1Departments of and Gynecology, David Geffen School of Medicine at UCLA, Los Angeles, CA, USA; 2Department of Community Health Sciences, Los Angeles, CA, USA; 3Center for Healthier Children, Families and Communities, UCLA School of Public Health, Los Angeles, CA, USA; 4Department of Obstetrics and Gynecology, University of Southern California, Keck School of Medicine, Los Angeles, CA, USA; 5AMF Consulting, Inc, Los Angeles, CA, USA and 6Cedars-Sinai Medical Center and Burns Allen Research Institute, Los Angeles, CA, USA

products of conception, subinvolution of the placental site, Objective: To identify women who are most likely to benefit from primary infection or .2 Serious morbidity arising from PPH prevention strategies for postpartum hemorrhage (PPH). also includes but is not limited to , hypovolemic Study Design: In a retrospective patient cohort, we applied recursive shock, disseminated intravascular coagulopathy, renal failure, partitioning algorithms to identify the most discriminant risk factors and hepatic failure, adult respiratory distress syndrome, and Sheehan’s 3 their interactions, and calculated the ‘number needed to treat’ to prevent syndrome. a single case of PPH (estimated loss >1000 ml). Global initiatives to reduce maternal mortality and to decrease maternal morbidity in obstetrics have focused on strategies for the Result: By delivery category, the highest risk groups with ‘number management of PPH, specifically emphasizing the active needed to treat’ ranging from 4 to 7 were: (1) management of the third stage of labor, including the use of (PPH ¼ 0.7% of 16 218)Fmacrosomia with gestational and immediately after the birth of the baby.4 The need for manual removal of the ; (2) primary cesarean (PPH ¼ 18.7% of standardized treatment protocols has been emphasized recently, 2696)Fmacrosomia and multiparity; and (3) repeat cesarean although which uterotonics should be used remains (PPH ¼ 16.0% of 1832)Futerine incision other than low transverse and controversial.5,6 Many hospitals currently run drills to train their failed vaginal birth after cesarean. staff using standardized protocols to recognize and treat PPH.7,8 Conclusion: Clinical profiles that identify women at risk for PPH can Although the emphasis to date has been on early diagnosis and provide a foundation for the development of primary prevention strategies. treatment, primary prevention of PPH has remained elusive. Journal of Perinatology (2009) 29, 422–427; doi:10.1038/jp.2009.2; Numerous risk factors for PPH have been identified, and include published online 29 January 2009 conditions associated with uterine overdistention (e.g. multiple gestation, polydydramnios), prolonged second and third stages of Keywords: postpartum hemorrhage; prevention; number needed to treat analysis; recursive partitioning algorithms labor, use of or tocolytic agents, abnormal placentation, and instrumental delivery.9 Given these multiple and disparate causes of PPH, primary prevention strategies have been lacking. The objective of this study is to create clinical profiles Introduction of women at risk for PPH for the purpose of identifying those Postpartum hemorrhage (PPH) remains a leading cause of women who are most at risk. Such information could subsequently 1 -related death in the United States. It is largely due to be applied to the generation and evaluation of institutionally , which complicates about 5% of deliveries. Severe specific strategies for PPH prevention. cases are usually recognized intraoperatively or within the first hour after vaginal delivery, although ‘late’ PPH occasionally occurs more than 24 h after delivery and may be because of retained Methods

Correspondence: Dr LM Korst, Department of Obstetrics and Gynecology, University of We conducted a retrospective cohort study of all births that Southern California Women’s and Children’s Hospital, 1240 North Mission Road, Room occurred between August 1995 and February 2004 in a single 5K40, Los Angeles, CA 90033, USA. Southern California hospital. This hospital is a teaching institution E-mail: [email protected] Received 27 October 2008; revised 21 December 2008; accepted 1 January 2009; published with a patient population that is ethnically diverse and largely online 29 January 2009 socioeconomically disadvantaged. The hospital has maintained a Women at risk for postpartum hemorrhage MC Lu et al 423 comprehensive delivery database that included pregnancy, labor Table 1 Clinical risk factors entered in recursive partitioning algorithms and delivery data abstracted from the medical chart review. These Age Pitocin use records were matched to administrative records containing performed during labor International Classification of Diseases, Version 9 codes to obtain Cardiac condition Premature rupture of membranes data regarding maternal and neonatal outcomes during Cervical ripening with Pre-term labor hospitalization. Women with stillborns, weighing <500 g Cesarean during second stage** Pre-term premature rupture of and multiple gestations had been excluded from the original membranes dataset. Because this study was intended to focus on typical Cholestasis of pregnancy Prior cesarean obstetrical patients, additional exclusions specific to this study were Chorioamnionitis Prior uterine surgery other than placenta accreta, percreta or increta, and , as cesarean these diagnoses would likely be a ‘trigger’ associated with altered Disseminated intravascular coagulation Prolonged second stage* management, including heightened surveillance immediately Failed vaginal birth after cesarean** Pulmonary condition postpartum. Gestational age (pre term (<37 weeks), Renal condition term (37–41 weeks), post-term (>41 The primary study outcome is PPH, defined as an estimated weeks)) blood loss (EBL) of 1000 ml or greater, which is consistent with the Rh sensitization recommendation of the World Health Organization, and is based Group B b strep positive Ruptured membranes >24 h on experience showing that blood loss up to 1000 ml may be Hepatitis or other hepatic condition Seizure disorder considered to be physiological, and that, for healthy women HIV 1000 ml is the physiological point at which a woman’s vital signs HSV Substance abuse may be affected.10,11 Nevertheless, PPH has been variably defined as Hypertension (includes gestational Third- or fourth-degree perineal having: (1) an EBL X500 ml;9 (2) an EBL X1000 ml;12 (3) an hypertension, pre-, eclampsia laceration EBL X500 ml for vaginal deliveries and X1000 ml for cesarean and chronic hypertension deliveries13 and (4) an EBL X1500 ml.14 For completeness, we Intra-uterine growth restriction Thyroid condition evaluated the relationships between these different case definitions Macrosomia by clinical estimation Tocolysis or treatment with magnesium for PPH and a composite measure of clinically significant ‘severe’ >4000 grams sulfate Malpresentation Treated with steroids for prematurity outcomes that consisted of , cardiovascular events, Manual extraction of the placenta* Use of forceps or vacuum* renal failure, disseminated intravascular coagulation and Mental health condition Uterine incision (low transverse vs hysterectomy, as defined by International Classification of Diseases, other)* Version 9 codes. Multiparity Uterine or cervical soft tissue condition: Using data abstracted from the medical chart review, we e.g. fibroids, cervical dysplasia identified clinical conditions or risk factors that could be associated with PPH (Table 1). The study population was stratified into three Abbreviations: HIV, human immunodeficiency virus; HSV, herpes simplex virus. clinical groups: vaginal delivery, primary cesarean delivery and Starred factors were relevant only to vaginal (*) or cesarean delivery (**). repeat cesarean delivery. Prolonged second stage of labor was defined as a second stage labor X3orX2 h for primiparous women with and without an epidural, respectively, as per the definition used by the American College of Obstetricians and hemorrhage. Recursive partitioning is a nonparametric technique Gynecologists.15 For multiparous women, this definition changed that produces a classification or decision tree in which subjects are to X2 and X1 h depending on the use of regional anesthesia. assigned to mutually exclusive subsets according to a set of Prolonged second stage could not be included in the cesarean predictor variables.17,18 It identifies subject subgroups with varying models as the majority of women delivered by cesarean were risks and may uncover interactions between predictors that may be delivered before labor, or during the first stage of labor. Instead, the overlooked in the traditional application of logistic regression to variable ‘cesarean performed during second stage’ was used to datasets with numerous predictors, as in our case. The decision tree capture the timing of the cesarean birth. The majority of cesarean was constructed by splitting subsets of the dataset using all deliveries are secondary to dystocia, either ‘arrest of dilatation,’ predictor variables to create two or more child nodes repeatedly, which is during the first stage, or ‘arrest of descent,’ which is beginning with the entire dataset. The predictor having the highest during the second stage. Overall, fewer than 15% of all primary association with the target variable was selected for splitting. The cesareans are performed for .16 association between each predictor variable and the target was For each delivery group, recursive partitioning algorithms computed using the analysis of variance F-test (for ordinal and (RPA) were used to examine the relationship of the clinical factors continuous predictors) or Pearson’s w-square test (for nominal to PPH, and to ascertain which women were most at risk for predictors).

Journal of Perinatology Women at risk for postpartum hemorrhage MC Lu et al 424

Once these clinical factors and their interactions were identified, compared with women who did not have PPH (ORs 19.53; 95% CIs multivariable logistic regression models were built for each group 11.10 to 34.36), and were 40 times more likely to have an using these clinical factors as covariates, and odds ratios (ORs) and erythrocyte transfusion compared with women who did not have 95% confidence intervals (CIs) for each factor were calculated. We PPH (ORs 40.00; 95% CIs 22.73 to 70.42). theorized that these factors could then be used to identify women at Of the 16 218 vaginal deliveries, 122 (0.8%) were complicated by the greatest risk for PPH, and could be used to develop strategies for PPH. Data on the duration of the second stage of labor were prevention, which may include closer supervision and/or earlier missing for 2385 deliveries, largely because many of these women intervention. We calculated the number of at-risk women who entered the hospital during the second stage. Owing to the would need to be ‘treated’ under a prevention protocol to prevent importance of including prolonged second stage in the model, one case of PPH. We report the ‘number needed to treat’ with 95% these deliveries were excluded from further analyses, yielding a CIs for patients with one risk factor and patients with two risk total of 13 833 vaginal deliveries. RPA and multivariable logistic factors (relative to patients without any risk factors) within each of regression equations were then used to identify and quantify the the three delivery groups. The estimated probabilities for PPH used most significant risk factors for PPH complicating vaginal in these calculations were derived from the respective logistic deliveries. regression models. All analyses were performed using the Statistical Five clinical risk factors emerged from this analysis: prolonged Analysis Software (Version 9.1, Statistical Analysis System (SAS) second stage, macrosomia with maternal diabetes, macrosomia Institute Inc., Cary, NC, USA) with the exception of the RPA without maternal diabetes, manual removal of placenta and the specifying the decision tree, which was constructed using the use of magnesium sulfate. These risk factors were then evaluated Answer Tree module of the Statistical Package for the Social simultaneously in the multivariate model described in Table 3, Sciences (SPSS Inc., Chicago, IL, USA). The study was approved by which lists the ORs and 95% CIs for each of the covariates. the Institutional Review Board of the participating hospital. Of the 2696 primary cesarean deliveries, 505 (18.7%) were complicated by PPH. Our analysis identified the four most significant risk factors for PPH: cesarean during second stage, Results macrosomia, chorioamnionitis and multiparity (Table 3). A total of 23 673 deliveries occurred during the study period, and of these, 21 867 (92.4%) were matched to administrative records for Table 3 Odds ratios for risk factors associated with postpartum hemorrhage by both mother and baby. After the clinical exclusions noted above, method of delivery based on multivariate logistic regression models 20 746 deliveries (94.9%) formed the final study population. Of these, Factor OR (95% CI) 16 218 (78.2%) were vaginal deliveries, 2696 (13.0%) were primary cesarean deliveries and 1832 (8.8%) were repeat cesarean deliveries. Vaginal deliverya (N ¼ 13 833; with 92 (0.7%) PPH) Table 2 describes the variation in clinically significant ‘severe’ Prolonged second stage 2.11 (1.28–3.46) maternal outcomes with various definitions of PPH from the Macrosomia without diabetes 2.09 (1.13–3.87) literature. Overall, 4.4% of the study population had an EBL Macrosomia with diabetes 7.85 (3.09–20.0) X1000 ml; among these, 2.5% experienced a severe maternal Manual delivery of placenta 4.60 (2.46–8.61) outcome. Using the definition of EBL X1000 ml, women with PPH Use of magnesium sulfate 2.52 (1.40–4.52) were 19 times more likely to have a severe maternal outcome Primary cesarean deliveryb (N ¼ 2696; with 505 (18.7%) PPH) Macrosomia 2.18 (1.72–2.75) Table 2 The relationships between different case definitions for postpartum Cesarean during second stage 1.48 (1.14–1.92) hemorrhage and severe maternal outcomes (N ¼ 20 746) Chorioamnionitis 1.60 (1.26–2.04) Case definition Number (%) of % With severe maternal Multiparity 1.51 (1.23–1.85) cases outcomea Repeat cesarean deliveryc (N ¼ 1832; with 294 (16.0%) PPH) X EBL 500 ml 5358 (25.8%) 0.8 Uterine incision other than low transverse 2.71 (1.89–3.90) X EBL 500 ml vaginal or 2007 (9.7%) 1.2 Age >35 1.48 (1.13–1.95) X 1000 ml CS Failed vaginal birth after cesarean (VBAC) 1.91 (1.47–2.49) EBL X1000 ml 906 (4.4%) 2.5 EBL X1500 ml 166 (0.8%) 5.4 Abbreviations: CI, confidence interval; OD, odds ratio, PPH, postpartum hemorrhage. aGlobal likelihood ratio P ¼ <0.0001, Hosmer and Lemeshow goodness of fit P ¼ 0.9379, Abbreviation: CS, cesarean; EBL, estimated blood loss. c ¼ 0.657. aSevere maternal outcome was defined using administrative data (International bGlobal likelihood ratio P<0.0001, Hosmer and Lemeshow goodness of fit P ¼ 0.5708, Classification of Diseases, Version 9 (ICD-9) codes) as a composite of maternal death, c ¼ 0.615. cardiovascular events, renal failure, disseminated intravascular coagulation and cGlobal likelihood ratio P<0.0001, Hosmer and Lemeshow goodness of fit P ¼ 0.1808, hysterectomy. c ¼ 0.635.

Journal of Perinatology Women at risk for postpartum hemorrhage MC Lu et al 425

Table 4 The number of at-risk women needed to treat in order to prevent one case of postpartum hemorrhage, by route of delivery and number of risk factors

Characteristic Prevalence per centa Number needed to treat 95% confidence interval

Vaginal delivery One risk factor Macrosomia with gestational diabetes 0.8 33 13–94 Manual removal of placenta 2.8 62 32–129 Use of magnesium sulfate 6.0 146 72–334 Prolonged second stage 10.9 210 112–427 Macrosomia without gestational diabetes 7.2 203 92–549 Two risk factors Macrosomia with gestational diabetes and manual removal of placenta 0.03 7 3–20 Macrosomia with gestational diabetes and use of magnesium sulfate 0.07 13 5–36 Prolonged second stage and macrosomia with gestational diabetes 0.1 17 7–48 Manual removal of placenta and use of magnesium sulfate 0.3 22 10–48 Prolonged second stage and manual removal of placenta 0.5 28 14–58 Manual removal of placenta and macrosomia without gestational diabetes 0.2 27 12–64 Prolonged second stage and use of magnesium sulfate 0.9 56 28–119 Use of magnesium sulfate and macrosomia without gestational diabetes 0.3 53 22–139 Prolonged second stage and macrosomia without gestational diabetes 1.0 71 33–168

Primary cesarean One risk factor Macrosomia 16.0 10 8–13 Chorioamnionitis 19.4 19 14–28 Multiparity 43.6 23 21–26 Cesarean during second stage 14.8 25 16–46 Two risk factors Macrosomia and chorioamnionitis 4.0 5 4–7 Macrosomia and multiparity 8.5 5 5–7 Cesarean during second stage and macrosomia 3.6 6 4–8 Chorioamnionitis and multiparity 4.6 8 6–11 Cesarean during second stage and chorioamnionitis 5.4 8 6–12 Cesarean during second stage and multiparity 4.5 9 7–13

Repeat cesarean One risk factor Uterine incision is not low transverse or low transverse with extension 8.9 7 5–10 Failed vaginal birth after cesarean (VBAC) 28.7 12 10–15 Age >35 years 27.7 23 17–31 Two risk factors Uterine incision other than low transverse and failed VBAC 2.5 4 3–5 Uterine incision is other than low transverse and age >35 years 3.4 5 3–6 Failed VBAC and age >35 6.7 7 5–9

Abbreviation: VBAC, vaginal birth after cesarean. aPer cents calculated out of non-missing values.

Of the 1832 repeat cesarean deliveries, 294 (16.0%) were Table 4 presents the numbers of at-risk women needed to treat complicated by PPH. RPA identified the three most significant risk in order to prevent one case of PPH. In this context, a ‘treatment’ factors for PPH complicating repeat cesarean deliveries: advanced would be equivalent to participating in a prevention strategy. For maternal age (X35 years), failed vaginal birth after cesarean and vaginal delivery, 210 women with prolonged second stage would a uterine incision other than an uncomplicated low transverse need to be treated in order to prevent one case of PPH. However, for incision (Table 3). women with maternal diabetes and fetal macrosomia, only 17 have

Journal of Perinatology Women at risk for postpartum hemorrhage MC Lu et al 426 to be treated in order to prevent one case of PPH. For primary high-risk groups identified here suggest multiple opportunities to cesarean delivery, 19 women with chorioamnionitis would need to guide hospitals in the development of ‘best practices’ for routine be treated in order to prevent one case of PPH. However, for obstetrics. women with chorioamnionitis who underwent a cesarean delivery Under a secondary prevention protocol, high-risk patients might during the second stage (most of which were due to arrest of benefit from increased nursing vigilance immediately postpartum descent), only eight have to be treated in order to prevent one case and/or an empirical prophylactic treatment regimen. of PPH. The frequency and duration of nursing assessments could perhaps be extended, and nursing communication between the staff on labor and delivery and the postpartum ward could specify Discussion that the patient is at increased risk for PPH, and monitored As quality improvement efforts intensify in maternal care, there is a accordingly. Strategies to decrease the prevalence of these need for targeted prevention strategies that can address PPH and key risk factors could also be undertaken, for example, other principal morbidities of . Information regarding concentrating on reducing risks for infection and manual which women are most at risk can be used to develop extraction of the placenta. institutionally specific ‘clinical triggers’ for intervention, to This retrospective study was undertaken in a single institution, decrease the occurrence of PPH, and lead to heightened which limits the generalizability of our findings not only because surveillance of at-risk patients. If PPH can be anticipated and they are dependent on an individual particular study population recognized promptly, maternal morbidity may be lessened. and set of management and prevention practices, but also because Numerous risk factors for PPH have been suggested earlier, and they depend on the nature and accuracy of the data recorded at those we have identified here are not new.13 However, the use of this institution. Some of the CIs of the ORs are wide for these rarer RPA allows us to highlight those clinical risk factors that appear to combinations of risk factors and may limit generalizability to other be most strongly associated with PPH so that those who provide institutions. Nevertheless, the ability to evaluate such a high care in childbirth can identify women at the highest risk. Moreover, volume of obstetrical patients with consistent management has whereas earlier studies examined these numerous clinical factors allowed us to examine the potential impact of numerous risk as independent risk factors for PPH, we were able to examine how factors on PPH, and to estimate the occurrence of rare but severe these factors contribute to the risk of PPH in the presence of all the maternal outcomes. other risk factors, and identify those that were most important. The dependence of the diagnosis of PPH on the subjective EBL Here, we identified women at the highest risk for PPH by estimate inherently limits our findings. As 500 ml is now delivery route. The combination of risk factors that was important understood to be an average EBL for women in normal labor, the for each delivery route depended both on the prevalence of the risk current adoption of 1000 ml provides a more clinically relevant factor and its strength. These methods thus focus on women most definition,10,11 and a definition that is associated with substantial at risk in a typical obstetrical population, so that the hospital staff maternal morbidity in our own data. As noted by the World Health can determine where their vigilance and resources should be Organization, the ‘500 ml limit should be considered an alert line; directed. The risk factors and their combinations uncovered here the action line is then reached when vital functions of the woman are consistent with the literature and with general obstetrical are endangered. In healthy women this usually only occurs after experience. blood loss >1000 ml.10’ Further, it has the advantage of being The clinical nature of PPH provides both primary and secondary standard across delivery method. prevention opportunities for the obstetrical staff. As primary In this paper, we provide clinical profiles of women who might prevention opportunities for PPH are not well understood, our benefit the most from both primary and secondary PPH prevention results can only suggest a foundation for further study of the strategies. This work suggests several avenues by which institutions identified high-risk groups. For example, under a primary may use these results to assist in monitoring at-risk women and in prevention protocol for women with anticipated vaginal deliveries, developing their own prevention strategies to reduce the occurrence those with gestational diabetes and macrosomia could be and morbidity of PPH. Our study suggests that individualized monitored strictly using Friedman’s Curve or a labor partogram to patient-specific risk assessment for PPH may serve as a trigger for confirm that labor is progressing as expected.19–21 These results initiating institutional-specific PPH prevention strategies. As we also recognize that, all women with protraction disorders deserve seek improvement in maternal quality of care, we should not limit prompt recognition, treatment and consideration for cesarean our efforts to improved early diagnosis and treatment of PPH, but delivery sooner rather than later. Furthermore, the data indicate rather, we should seek the development of strategies for primary that compounding protracted labor with chorioamnionitis or a prevention of PPH. Preventing PPH provides a substantial manual placental extraction, two circumstances that may in opportunity to decrease maternal morbidity and mortality and themselves be largely avoidable, greatly increases PPH risk. The thereby improve overall childbirth outcomes.

Journal of Perinatology Women at risk for postpartum hemorrhage MC Lu et al 427

Acknowledgments 10 Technical Working Group. Care in normal birth: a practical guide. World Health Organization: Geneva, 1996 pp 31. Financial support for this work was provided by the California Department of 11 Technical Working Group, Department of Making Pregnancy Safer. WHO Health Services Maternal and Child Health Branch (Contract No. 01-15166). Recommendations for the Prevention of Postpartum Haemorrhage. World Health Organization: Geneva, 2007 pp 17. 12 Burchell RC. Postpartum haemorrhage. In: Quilligan ES (ed). Curr Ther Obstet References Gynaecol. WB Saunders: Philadelphia, PA, 1980. 1 Chang J, Elam-Evans LD, Berg CJ, Herndon J, Flowers L, Seed KA et al. Pregnancy- 13 American College of Obstetricians and Gynecologists. ACOG Practice Bulletin: Clinical related mortality surveillanceFUnited States, 1991–1999. MMWR Surveill Summ Management Guidelines for Obstetrician–Gynecologists Number 76, October 2006: 2003; 52:1–8. postpartum hemorrhage. Obstet Gynecol 2006; 108: 1039–1047. 2 Gilstrap III LC, Ramin SM. Postpartum hemorrhage. Clin Obstet Gynecol 1994; 37(4): 14 Waterstone M, Bewley S, Wolfe C. Incidence and predictors of severe obstetric morbidity: 824–830. case-control study. BMJ 2001; 322: 1089–1094. 3 Bonnar J. Massive obstetric haemorrhage. Baillieres Best Pract Res Clin Obstet 15 American College of Obstetricians and Gynecologists. Operative Vaginal Delivery. Gynaecol 2000; 14: 1–18. American College of Obstetricians and Gynecologists: Washington, DC, 2000 ACOG 4 Prendiville WJ, Elbourne D, McDonald S. Active versus expectant management in the practice bulletin no. 17.. third stage of labour. Cochrane Database Syst Rev 2000; (3): CD000007. 16 Gregory KD, Curtin SC, Taffel SM, Notzon FC. Changes in indications for cesarean doi:10.1002/14651858.CD000007. delivery: United States, 1985 and 1994. Am J Public Health 1998; 88(9): 1384–1387. 5Gu¨lmezoglu AM, Forna F, Villar J, Hofmeyr GJ. for prevention of 17 Breiman L, Friedman JH, Olshen RA, Stone CJ. Classification and Regression Trees. postpartum haemorrhage. Cochrane Database Syst Rev 2007; (3): CD000494. Wadsworth International Group: Belmont, CA, 1984. doi:10.1002/14651858.CD000494.pub3. 18 Sands K, Vineyard G, Livingston J, Christiansen C, Platt R. Efficient identification of 6 McDonald S, Abbott JM, Higgins SP. Prophylactic ergometrine-oxytocin versus post-discharge surgical site infections; use of automated pharmacy dispensing oxytocin for the third stage of labour. Cochrane Database Syst Rev 2004; (1): information, administrative data, and medical record information. J Infect Dis 1999; CD000201. doi:10.1002/14651858.CD000201.pub2. 7 Burtelow M, Riley E, Druzin M, Fontaine M, Viele M, Goodnough LT. 179: 434–441. How we treat: management of life-threatening primary postpartum hemorrhage 19 Friedman EA. Labor. Clinical evaluation and management. Appleton-Century-Crofts: with a standardized massive transfusion protocol. Transfusion 2007; 47(9): 1564–1572. New York, 1967. 8 Mousa HA, Walkinshaw S. Major postpartum haemorrhage. Curr Opin Obstet Gynecol 20 Lavender T, Alfirevic A, Walkinshaw S. Effect of different partogram action lines on 2001; 13(6): 595–603. birth outcomes: a randomized controlled trial. Obstet Gynecol 2006; 108: 9 World Health Organization. The prevention and management of postpartum 295–302. haemorrhage. WHO report of technical working group. World Health Organization: 21 Lavender T, Alfirevic A, Walkinshaw S. Partogram action line study: a randomized trial. Geneva, 1990 Report No. WHO/MCH/90.7. Br J Obstet Gynecol 1998; 105: 976–980.

Journal of Perinatology