Hemorrhage After Manual Removal of the : Weighing Risk Factors and the Role of the Third Stage of Labor

The Harvard community has made this article openly available. Please share how this access benefits you. Your story matters

Citation Perlman, Nicola. 2018. Hemorrhage After Manual Removal of the Placenta: Weighing Risk Factors and the Role of the Third Stage of Labor. Doctoral dissertation, Harvard Medical School.

Citable link http://nrs.harvard.edu/urn-3:HUL.InstRepos:36923348

Terms of Use This article was downloaded from Harvard University’s DASH repository, and is made available under the terms and conditions applicable to Other Posted Material, as set forth at http:// nrs.harvard.edu/urn-3:HUL.InstRepos:dash.current.terms-of- use#LAA Abstract: Introduction: Manual removal of the placenta, often due to or an abnormally adherent placenta, can be accompanied by severe maternal postpartum hemorrhage. The objective of this study was to identify risk factors for postpartum hemorrhage that would allow triaging of most morbid patients prior to manual removal.

Methods: This was a retrospective case control study of patients who had undergone manual removal of the placenta after vaginal delivery at Brigham and Women’s Hospital between January 1, 2007 and May 29, 2015. We evaluated risk factors for postpartum hemorrhage independently and by presumed causative role. Causative groups (and risk factors) included Atony (multiple gestation, prolonged second stage of labor, birth weight >4000g, parity>4, or BMI>40kg/m2), Abnormal Placentation (>2 prior D&Es, suspicion for accreta on ultrasound, ART, age>40 years, prior cesarean, prior accreta, prior retained placenta, prior uterine surgery, resolved low lying placenta, or Ashermans syndrome), or Other Postpartum Hemorrhage Risk Factors (preeclampsia, acute abruption, fibroids >6cm, or preterm delivery). Risk factors were also classified as either major (concern for accreta on ultrasound or signs of abruption at admission) or minor (any other variable), and we analyzed whether any one major or two minor risk factors, regardless of causative grouping, were associated with postpartum hemorrhage after manual removal. Length of third stage of labor was analyzed in relation to patient characteristics. All significant variables were put into a multivariable analysis to test for confounding.

Results: Of the 997 women identified with manual removal of the placenta during our study period, 172 experienced severe postpartum hemorrhage and were one-to-one matched with controls without hemorrhage. Case patients were more likely than controls to have any one risk factor in either the Atony (49% vs. 37% respectively , p = 0.01) or Abnormal Placentation risk group (58% vs. 38%, respectively, p<0.01) and additionally were more likely to have any one major or two minor risk factors (51% vs. 26% controls; p>0.01). Postpartum hemorrhage patients had manual removal of the placenta later in the third stage of labor (p<0.01). Characteristics associated with rapid manual removal of the placenta within 30 minutes after delivery of the infant included cord avulsion and epidural analgesia (p<0.01), whereas chorioamnionitis was associated with delay until manual removal (p=0.03). Increasing length of third stage of delivery showed strong association with postpartum hemorrhage in our multivariate analysis, along with risk factors for abnormal placentation. Though significant in the univariate analysis, , a risk factor for uterine atony, showed no association with postpartum hemorrhage after controlling for other factors in our multivariable regression.

Conclusion: Patients with concern for uterine atony should have rapid removal of the placenta, regardless of chorioamnionitis and epidural analgesia, in order to minimize risk of postpartum hemorrhage. Patients with risk factors for abnormal placentation or placenta accreta may not have risk of hemorrhage mitigated by rapid manual delivery.

2 Table of Contents:

Contents: Page Abstract 2 Glossary 4 Introduction 5 Methods 12 Results 16 Discussion 19 Summary 24 References 26 Tables 29 Figures 35

3

Glossary listing abbreviations that appear in your thesis MROP: Manual removal of the placenta PPH: Postpartum Hemorrhage ACOG: American College of Obstetricians and Gynecologists MAP: Morbidly adherent placenta Dilation and Curettage: D&C Dilation and Evacuation: D&E Body Mass Index: BMI In Vitro Fertilization: IVF Assisted Reproductive Technology: ART

4 Introduction:

Postpartum hemorrhage (PPH), occurring in approximately 2-6% of deliveries, is a major cause of maternal mortality worldwide.(1-4) In the last two decades, there has been an observed increase in post partum hemorrhage, which currently accounts for one of the most common reasons for maternal intensive care unit admission.(1, 3)

Uterine atony and retained placenta (including placenta accreta) are the two leading causes of postpartum hemorrhage. (1, 3-6) Both etiologies can also be indications for manual removal of the placenta (MROP), which can be followed by severe bleeding. (1) While studies have examined risk factors for PPH, rarely have risk factors specific to manual removal been proposed. Triaging patients according to hemorrhage risk with manual removal of the placenta may allow physicians to proactively avoid or manage this morbid complication.

Postpartum hemorrhage: Postpartum hemorrhage is a widely recognized cause of maternal morbidity and mortality.(4, 7) As one of the leading causes of Intensive Care Unit admission following delivery in the United States, there is great interest in examining risk factors and methods to prevent morbidity from PPH.(7-10) The trend is particularly prescient in the United States, but other developed countries have seen a similarly rise in PPH, not explained by temporal changes in risk factors.(11, 12) And while even in developed countries with resources and adequate personnel the risk of morbidity from maternal hemorrhage is high, giving birth in developing countries poses a staggering risk of death due to complications. Some reports cite maternal morbidity ratios (maternal deaths per 100,000 live births) to be as high as one in six for Afghanistan and Sierra Leone, versus a lowest estimate of one in 30,000 in Sweden.(13)

The import of postpartum hemorrhage in contributing to maternal morbidity has created a demand for better systems for predicting and triaging massive blood loss at delivery.(9) The epidemiology of postpartum hemorrhage has been well studied. The American College

5 of Obstetricians and Gynecologists (ACOG) warns physicians in their Practice Bulletin that primary PPH, defined by the College as greater than 1000cc of blood loss (a recent increase from 500cc), occurs in 4-6% of all , citing a 1991 study by Combs et al that lists PPH as a leading cause of maternal morbidity worldwide.(1, 4) Research by Callaghan et al has notably emphasized that while PPH is increasing, maternal death from PPH is decreasing due to increased rates of transfusion and peripartum hysterectomy.(14-16)

However, while death during labor hospitalization has decreased, severe maternal morbidity has in turn become up to 50 times more common.(16) It is widely recognized that PPH is due primarily to poor contraction of the , or atony, following delivery of the infant, and secondarily to retained or abnormally adherent placenta during the third stage of labor.(1, 4) Other recognized causes include inherited or acquired coagulopathies, and obstetrical complications—such as cervical lacerations episiotomies, and uterine inversion.(4)

In order to effectively and efficiently treat severe PPH, hospitals and healthcare systems are seeking to develop novel ways to triage and prepare for this adverse outcome.(4, 17) Risk scores and algorithms are becoming increasingly commonplace, and often employ a multidisciplinary team including obstetricians, anesthesiologists, nursing, and possibly surgical or radiologic subspecialists. Risk prediction may be particularly useful in settings where treatment must happen quickly and involves many team members and treatment modalities. For instance, when a patient requires manual extraction of the placenta with subsequent transfer to the operating room with severe postpartum hemorrhage, set up of cross-matched blood and surgical equipment, and early notification of a multidisciplinary hemorrhage team would be useful.

Vaginal Delivery with Manual Extraction of the Placenta: Manual removal of the placenta, or manually creating a cleavage plane between the placenta and the uterus, is generally performed in absence of delivery of the placenta during the third stage of labor, and typically after other management options (oxytocin, uterine massage, and cord traction) have been exhausted.

6

According to the literature, retained placenta requiring manual removal is generally assumed to be due to (1) an atonic uterus with poor contraction disallowing normal contractile expulsion of the placenta, (2) abnormal placentation or accreta, in which the placenta has invaded into the maternal myometrium, or (3) trapped or incarcerated placenta, in which the has closed prior to delivery of the placenta.(5, 18)

Normal placentation begins with blastocyst implantation into the maternal endometrium. In preparation for this implantation, the endometrium develops the decidua under the influence of progesterone and estrogen of early . As the blastocyst invades this decidua, the layer of cells forming the surface of the blastocyst develops into the chorionic membrane. Cytotrophoblast cells proliferate from the chorionic membrane and form multinucleated aggregates called syncytiotrophoblast cells. These cells form the placental villi, with cytotrophoblasts at the core and syncytiotrophoblasts forming the epithelial outer layer, allowing fetal-maternal interchange between the villi-decidual interaction. After the second stage of labor, with delivery of the infant, both a hormonal cascade as well as uterine contractions allow for separation of these layers and expulsion of the placenta.

If any part of this process is disrupted then the placenta may not deliver spontaneously and could be retained. For instance, if placental villi have invaded beyond the decidua and into the maternal myometrium causing adherent placenta due to this abnormal placentation, the placenta will not detach in the third stage. Alternatively, in the setting of poor uterine contractions due to atony, the placenta may not be subject to expulsive forces required for delivery.

In developed countries, retained placenta affects around 3% of all vaginal deliveries.(5, 19). Risk factors for retained placenta requiring MROP were elucidated by Endler and colleagues in a 2012 case control study.(20) The authors showed that these independent risk factors included prior retained placenta, preterm delivery, prolonged use of oxytocin, preeclampsia, two or more prior , and one or more prior abortions.(20) The authors also showed that MROP was significantly associated with PPH.(20) In a follow up

7 study Endler et al confirmed these risk factors using a large national registry database as cohort.(21) Their findings showed a particularly strong relationship between preeclampsia and retained placenta, leading the authors to hypothesize that disorders of placentation may be positively associated with MROP. (21)

A prolonged third stage of labor, or time between delivery of the infant and delivery of the placenta, is also recognized as a significant risk factor for postpartum hemorrhage, and studies have shown that reducing the length of third stage of labor via manual extraction of the placenta reduces the risk of postpartum hemorrhage.(22-24) In order to minimize this risk, failure of delivery of the placenta in a timely manner is therefore a common indication for MROP.

In the US, however, there is no one guideline for managing MROP.(25) While ACOG recognizes the increased risk of PPH with a long third stage of labor and thus recommends active management (with oxytocin, cord traction, and maternal effort to expel placenta), they do not suggest a firm timeline for manual removal.(4) While majority of are spontaneously expelled well before 30 minutes, most of the sparse literature on MROP suggests removal in absence of spontaneous delivery at this time.(5, 25)

Even though the general consensus is that MROP should be attempted in absence of spontaneous delivery of the placenta within 30 minutes of delivery of the infant or earlier in the presence of PPH, there is little data to support this guideline.(5, 19) In a study by Deneux-Tharaux et al, surveys from 14 European countries exhibited wide variations in wait time prior to MROP, largely by country but also by hospital.(19) In countries such as Finland and Denmark, obstetricians tended to wait 60 minutes or more prior to MROP, versus in countries such as Spain and France, where providers removed the placenta after 30 minutes.(19) Practices also varied considerably depending on whether or not the patient in question had prior epidural anesthesia.(19) Similarly, the National Institute for Health and Clinical Excellence (NICE) suggests a wait time of 30 minutes in the United Kingdom prior to MROP,(26) while the World Health Organization guidelines propose a wait time of 60 minutes.(13)

8

In a 1991 study by Combs et al, authors tried to define how duration of third stage is related to complications.(25) In the study they found that the third stage had a log normal distribution, with a mean length of 6.8 minutes, with only 3.3% of deliveries having greater a greater than 30 minute third stage. Interestingly, the authors calculated that the incidence of PPH, transfusion, and dilation and curettage remained constant during this period, increasing only after 30 minutes and plateauing at 75 minutes for both manually- and spontaneously-delivered placentas. Because PPH incidence did not increase until after 30 minutes, Combs et al recommended this timing as guidance for when to initiate MROP.

A subsequent study by Dombrowski et al in 1995 tried to determine gestational age specific data for length of third stage, retained placenta, hemorrhage, and manual removal.(27) The authors found that both MROP and PPH decreased with increasing gestational age, and that the two were related. However, causal association could not be determined. Alternatively, while not studying manual removal per se, Rogers et al, in 1998, found that active management of the third stage resulted in significantly reduced risk of PPH.(22)

Later studies by Magann et al sought to strengthen the relationship between length of third stage and PPH. In 2005, Magann and colleagues undertook a prospective observational study in which all women delivering vaginally were assessed for PPH.(24) Using receiver operating characteristic curves, the authors showed that a third stage of labor longer than 18 minutes was associated with a significant risk of PPH.(24) The authors followed up this paper in 2012 with a randomized controlled trial assigning vaginal deliveries to manual removal at either 10 or 15 minutes (as opposed to the traditional 30) if the placenta had not yet spontaneously delivered.(23) The findings supported the authors’ initial study, showing that removal at 15 minutes had a significantly greater likelihood of hemorrhage compared to 10 minutes, augmenting the discussion of MROP’s appropriate timing, and indications.(23)

In their 2014 review, Urner et al emphasize that guidelines for management of retained placenta requiring MROP do not exist, aside from poorly defined timelines for

9 intervention.(5) The authors cite similar risk factors as those studied by Endler et al, and additionally list risk factors for invasive placenta including prior cesarean section and uterine surgery.(5) Urner and colleagues suggest that routine management of retained placenta should include (in addition to MROP at 30 minutes) ultrasound and a multi- specialist team approach, as well as future development of strict guidelines for the third stage in order to reduce complications such as severe PPH.(5) Corviello et al, using a large retrospective cohort from the United States, echoed the above risk factors, particularly stressing stillbirth as an independent risk factor, and proposing that some intrinsic pathophysiology of intrauterine fetal demise leads to increased risk of retained placenta.(28) Titiz and colleagues support these findings in a large cohort from Australia.(2)

While abnormal placentation, such as accreta, increta or percreta, is a known significant risk factor for adherent placenta at cesarean section, very little has been written on its contribution to retained placenta at vaginal delivery. This condition is best understood in the setting of placenta previa with prior cesarean.(29, 30) In this clinical scenario, it is recognized that low lying placenta overlying a uterine scar and prior uterine surgery are major risk factors for hemorrhage morbidity at cesarean delivery. Because of it’s association with low lying placenta and cesarean, radiographic evidence and clinical suspicion for accreta are often lacking in patients who present without placenta previa, which includes all patients undergoing a vaginal delivery.

Interestingly, in a 2017 paper by Bjurstrom et al, the authors report six cases of failed manual removal at vaginal delivery due to clinically adherent placenta—referred to as morbidly adherent placenta, or MAP—only one of which had histopathologic evidence of accreta.(31) Additionally in 2017, Roeca et al reported 339 patients who completed a pregnancy with pathologically diagnosed placenta accreta without hysterectomy with 39 returning for subsequent pregnancy.(32) Of the 39 returning patients, 16 underwent manual removal of the placenta. This suggests that the contribution of accreta to MROP may be under recognized.

10 The mechanisms behind and outcomes of abnormally adherent placenta at vaginal delivery have not been well reported. A few studies have addressed relationships between in assisted reproductive technology (ART) and in vitro fertilization (IVF) and disorders of placentation, though its relationship to retained placenta and PPH has not been well defined. First reported in 2011, Esh-Broder et al showed a significant relationship between IVF and placenta accreta, proposing environmental differences between the endometrium of patients conceiving with IVF and versus those with spontaneous conception.(33) In 2014, Ishihara et al also concluded via retrospective analysis that cryopreserved blastocyst transfer was associated with both accreta and pregnancy induced hypertension, though they had limited ability to control for confounding.(34) A 2015 study by Kaser and colleagues that, after controlling for multiple confounders, cryopreserved embryo transfer was significantly associated with placenta accreta, and proposed relatively thinned endometrium and lower estradiol levels as potential mechanisms.(3)

Others have evaluated IVF with relation to the third stage of labor. Elenis et al, in a 2015 study from Sweden, looked specifically at oocyte donation IVF and the risk of poor obstetrical outcomes in otherwise healthy women.(35) The authors found a positive association between retained placenta and oocyte donation, as well as between PPH and oocyte donation.(35) Finally, in a recent 2016 study by Aziz et al, seeking to determine whether or not length of third stage was related to IVF, the authors concluded that cryopreserved embryo transfer (donated or autologous) without controlled ovarian hyperstimulation was not related to longer third stage, but did significantly increase the risk for MROP.(36)

Purpose of Inquiry: While individual risk factors for MROP have been proposed and studied, no study has determined factors associated with hemorrhagic morbidity in this context. Additionally, because it is mostly studied in the context of cesarean with known previa, few studies have looked at morbidly adherent placenta and placenta accreta risk factors (particularly resolved low lying placenta, prior uterine surgery, history of previous retained placenta, prior cesarean section, and ART) as predictors of PPH after manual extraction. The goal of

11 this study was to determine predictors of primary, severe postpartum hemorrhage after MROP, defined as an estimated blood loss (EBL) greater than or equal to 1500 cc occurring within the first 24 hours after delivery, need for transfusion, greater than or equal to a 10 point drop in hematocrit.(37)

We hypothesized that one major risk factor or multiple minor risk factors for post partum hemorrhage will increase risk of severe hemorrhage after manual extraction. In defining risk factors, we grouped variables according to their association with atony, invasive placentation, or postpartum hemorrhage in the absence of these other factors (other). Due to worldwide increased use of IVF and pathologic placenta accreta, and the paucity of research connecting this procedure to third stage abnormalities, we were particularly interested in this etiology as a risk factor for PPH.(3) We secondarily hypothesized that patients with risk factors for abnormal placentation would show the strongest association with PPH, based on the severity of this condition and lack of timely recognition and management at vaginal delivery.

Elucidating independent risk factors for postpartum hemorrhage will allow obstetricians to better prepare for and recognize severe morbidity at vaginal delivery, with the goal of reducing maternal morbidity and mortality in both developed and developing countries.

Methods

This retrospective case-control study has been approved by Partners/Brigham and Women’s Hospital Institutional Review Board. We used billing (ICD-9 and CPT) codes in order to identify all patients with vaginal deliveries between January 1, 2007 and May 29, 2015 who underwent manual removal of the placenta (MROP). This was supplemented with data from our hospital’s official electronic delivery record, which categorizes placental deliveries as “manual” or “spontaneous.” Patient data was collected from electronic and paper charts. MROP was defined as need for manual creation of a separation plane between uterus and placenta postpartum for an either partially or undelivered placenta. Two authors (NP and DC) reviewed all charts to confirm manual removal.

12

While practices vary, manual removal of the placenta is typically attempted if the patient has not delivered the placenta within 30 minutes, if a delivered placenta is visually incomplete, or if heavy bleeding mandates MROP sooner than 30 minutes. At Brigham and Women’s Hospital, obstetricians use their discretion in deciding when to manually extract the placenta.

Our case group included patients experiencing severe hemorrhage after manual extraction. Severe postpartum hemorrhage was defined in this study as estimated blood loss (EBL) ≥1500cc, hematocrit drop≥10 percentage points, or blood product transfusion for the indication of hemorrhage. Our EBL definition was based on the literature showing that patients with >1500cc blood loss experience the most morbidity and mortality peripartum.(10) Our control group was randomly selected from those patients who did not experience severe postpartum hemorrhage. Controls were one-to-one matched based on year of delivery in order to minimize biases caused by changes in practice over the course of the study period. Any patient in the cohort who delivered extramurally, prior to 24 weeks gestational age, or had delayed extraction of the placenta for more than 24 hours after delivery was excluded from the analysis.

We collected detailed demographic, medical and historical data for each subject, including age, body mass index (BMI) at the end of pregnancy, race, parity, number of prior spontaneous or therapeutic abortions (medical or surgical), prior history of retained placenta (placenta requiring manual or surgical removal in prior pregnancy), and prior history of pathologically-confirmed placenta accreta. Surgical historical variables included prior cesarean, dilation and curettage (D&C) performed in the setting of pregnancy (spontaneous or therapeutic abortion, or treatment of postpartum hemorrhage or retained products of conception), or uterine surgical procedures (myomectomy, any operative hysteroscopy, or endometrial ablation). Prior D&C on a non-pregnant uterus or uterine polypectomies were not considered.

13 Along with demographics and history above, we collected characteristics associated with current pregnancy. These included conception achieved with Assisted Reproductive Technology (ART; defined as either in vitro fertilization or intracytoplasmic sperm injection), multiple gestation in the current pregnancy, a prolonged second stage of labor (time between full cervical dilation of 10cm and delivery of the infant), birth weight greater than 4000g, signs of accreta on ultrasound (vascular lakes, loss of boundary between placenta and myometrium, absence of sonolucent zone, or signs of deeper invasion such as percreta), resolved low lying placenta (defined as a placenta that had come within 2cm or covered the cervical os during pregnancy via ultrasonography records), Asherman syndrome or uterine synechiae (intrauterine adhesions or bands as indicated by ultrasound, hysteroscopy, or as noted by patient’s provider in prenatal records ), preeclampsia (as indicated by patient’s provider in prenatal records), acute (as mentioned by the patients provider, or by abnormally heavy bleeding at presentation), fibroids greater than 6cm on pregnancy ultrasound, and gestational age (with preterm delivery defined as delivery prior to 37 weeks gestational age).

We also collected labor variables of present pregnancy including operative vaginal delivery (vacuum assisted or forceps assisted vaginal delivery) lower genital tract laceration, degree of perineal laceration (as indicated by provider in electronic delivery record), episiotomy, cord avulsion (complete separation of the umbilical cord prior to delivery of placenta, as noted by delivery provider), and chorioamnionitis (fever greater than 100.4F treated with antibiotics during first or second stage of labor).

Pregnancy outcomes and interventions collected included atony (provider noted atony in chart or used greater than or equal to two uterotonics—misoprostol, methylergonovine, prostaglandin F-2α, or oxytocin), clinical accreta (provider reported suspected clinical accreta or noted abnormal adherence of the placenta to the uterine wall with unusually difficult separation), uterine inversion, hysterectomy, intrauterine balloon placement, uterine artery embolization, need for dilation and curettage in current pregnancy, intensive care unit admission, postpartum readmission, and any pathologic accreta diagnosis

14 (defined as placental fibers adherent to the myometrium with absent intervening decidua, as indicated by the pathologist).

We analyzed risk factors for hemorrhage after manual delivery both individually and grouped based on their presumed causative role in PPH associated with MROP. The first group included variables associated with atony, including birth weight greater than 4000g, morbid obesity (BMI greater than 40kg/m2), chorioamnionitis, multiple gestations, high parity (greater than 4) and prolonged second stage of labor (greater than two hours). The next group included variables associated with abnormal placentation or placenta accreta, including previous cesarean delivery, prior uterine surgery, resolved previa or low lying placenta in the current pregnancy, Asherman syndrome or uterine synechiae, history of pathologic accreta, conception achieved via ART, or more than 2 prior D&C procedures as defined above. Our final group consisted of other PPH risk factors, including uterine leiomyoma greater than or equal to 6cm, acute placental abruption, or preeclampsia.(1, 3, 20, 37-43).

We also examined whether any one major risk factor or any two minor risk factors for PPH, regardless of grouping, were associated with hemorrhage. Major risk factors for severe hemorrhage were defined as signs of accreta on antepartum radiograph or presence of acute abruption.(40) Any other risk factor was considered minor.

To analyze the relationship between time and hemorrhage, we recorded every subject’s duration of third stage of labor, defined as time in minutes between delivery of the infant and delivery (manually in our cohort) of the placenta, as officially recorded in the electronic delivery record. To evaluate the relationship between hemorrhage and duration of the third stage, we divided time until placental delivery into three categories (60 minutes). Finally, to better understand the role of time in the development of PPH, we explored whether patient and labor characteristics were associated with varying lengths of third stage via our three time categories above.

15 The data analysis for this paper was generated using SAS software, version 9.4. Copyright © 2016, SAS Institute Inc. Continuous variables were analyzed with Wilcoxon tests, while categorical variables were analyzed with conditional logistic regression or Fisher Exact tests if a variable had a null entry. P-values less than 0.05 were considered significant with two-tailed tests. Finally, in order to test for independent associations and control for confounding, we entered risk factor variables into a stepwise logistic regression model. Any variable associated with hemorrhage with a p-value of ≤ 0.2 was included in the model. Variables were excluded from the model if final p-values were >0.35.

Results:

Between January 1, 2007 and June 30, 2015, 997 women with manual removal of the placenta were identified. Of these subjects, 172 patients were identified as having a major postpartum hemorrhage. An additional 172 matched control patients with MROP but without major PPH were randomly selected from the remaining patients.

Characteristics of case (severe PPH) and control (no PPH) groups are shown in Table 1. The case group showed lower median BMI (33.4 vs. 35.2, p<0.01), longer second stage of labor (48 kg/m2 vs. 38 kg/m2; p=0.04), and longer third stage of labor (38 minutes vs. 23 minutes; p<0.01). The case group was slightly younger than the controls (33.4 years vs. 35.2 years; p<0.01), and the median number of prior pregnancies was one less (0 vs. 1, p<0.01). They also tended to deliver under epidural analgesia less frequently than the control group (145 (85%) vs. 159 (93%), respectively, p=0.02). Ethnicity, prior number of abortions (both therapeutic and spontaneous), operative vaginal delivery, cord avulsion, episiotomy, laceration (including high degrees), and gestational age did not significantly differ between PPH and no PPH groups.

Obstetrical outcomes for case and control groups are shown in Table 2. Average blood loss in our severe PPH group was 1000cc (100-4000cc) and average hematocrit drop was 12.3% (0.3-23.3%). In the case group 84 patients out of 173 were transfused with at least

16 one blood product. Average blood loos in our control group was 400cc (75-1400cc), and average hematocrit drop was 4.1 (0.1-9.7) and no patient had a transfusion (per case inclusion criteria). Hematocrit drop data was frequently unavailable, particularly for our control group, often due to no blood draw postpartum.

Of the 172 patients with significant PPH, 51% of our PPH group was noted to have uterine atony by obstetricians or providers, in comparison to 12% of our non-PPH control group (p<0.01). The PPH group and the control group did not differ in rates of pathologic accreta diagnosis (17% vs. 11%; p=0.09). Similarly, the percentage of PPH patients with clinically adherent placenta (20%) did not differ from the control group (19%; p=0.79), nor did they differ when the clinical and pathologic diagnoses were combined (34% in the accreta group versus 29% in the control group, p=0.36).

Individual predictors within the three hemorrhage risk factor categories and their associations with our MROP with PPH and control groups are shown in Table 3. Within the Atony category, 17% of PPH patients experienced chorioamnionitis, versus the control group which had 8% with chorioamnionitis (p=0.01). Multiple gestation (10% PPH vs. 3% controls; p=0.01) and a second stage of labor longer than two hours (28% PPH cases vs. 17% controls; p<0.01) were also significantly associated with PPH. High birth weight, high parity, and morbid obesity were not significantly different between the case and control groups.

Within the Abnormal Placentation group, prior uterine surgery (24% cases vs. 6% controls; p<0.01), history of retained placenta (13% cases vs. 6% controls; p=0.03) and resolved low-lying placenta (15% cases vs. 8% controls; p=0.03) were significantly associated with PPH after MROP. The percentage of our PPH case group who underwent ART (19%) did not differ from the respective percentage of controls (12%; p=0.09). In our case group there were four patients (2%) who had a prior pathologic accreta without hysterectomy, versus none in our control group (NS). Similarly three patients (2%) of our case group had evidence of accreta on ultrasound, versus none in our control group (NS). None of the potential risk factors in the Other PPH Risk Factors group (large fibroids, acute abruption,

17 high order laceration or preeclampsia) showed a significant association with PPH in the setting of MROP.

Case patients were significantly more likely than controls to have any one risk factor in either the Atony (49% vs. 37% respectively, p = 0.01) or Abnormal Placentation risk group (58% vs. 38%, respectively, p<0.01, Table 3). Similarly, PPH case patients were significantly more likely to have any one major (sign of accreta on ultrasound or signs of abruption at admission) or two minor (any other variable) risk factors (51% vs. 26% controls; p>0.01) or any risk factor regardless of group or severity stratification (83% vs. 66% controls; p<0.01).

The third stage of labor by quartile in our PPH case group and our no PPH control group is shown in Table 4. Within the PPH case group, third stage ranged from 1 to 213 minutes, with 95% of placental deliveries occurring within 2.5 hours and a median of 38 minutes. This varied compared to the control group, in which 95% of placentas delivered within 76 minutes, with a median of 23 minutes.

Table 5 also shows the length of third stage of labor broken down by time category (60 minutes). Patients in the PPH case group had their placentas manually removed later than controls. Within the first 30 minutes, 38% of our PPH cases had MROP versus 68% of controls. Between 31 and 60 minutes, 30% of PPH cases and 22% of controls had MROP. Finally, after waiting 60 minutes, 32% of the PPH case group had their placentas delivered during this time, as opposed to 10% of the controls (p<0.01) Figure 1 illustrates the percentage of case and control patients delivering their placentas over time.

Characteristics of patients delivering at different intervals of third-stage are shown in Table 5. Patients experiencing cord avulsion had significantly shorter time-to-delivery intervals, with 24% of 60 minute deliveries; p<0.01). In contrast,

18 chorioamnionitis during labor was significantly associated with a longer third stage, with 9% of the 60 minute deliveries associated with chorioamnionitis during labor (p=0.03). Our three time categories did not show significant association with atony or accreta, both characteristics tended to have placental delivery earlier in our cohorts range of third stage.

Results from our stepwise logistic regression are shown in Table 6, which shows all variables retained in the final model. After adjusting for significant variables in the univariate analysis (length of third stage of labor, prior uterine surgery, maternal age by year, multiparity, multiple gestations, BMI by quartile, low lying placenta, history of retained placenta, episiotomy, laceration, epidural anesthesia, ART, prematurity, second stage of labor >2 hours, cord avulsion, chorioamnionitis, and parity >4), chorioamnionitis, cord avulsion, and parity>4 fell out of the model. Increasing length of third stage showed the strongest association to PPH (OR=3.1, 95% CI 1.65-5.82 95%) for 3rd stage of 31-60 minutes vs. 60minutes vs.

Discussion:

The purpose of this study was to define characteristics associated with major hemorrhage in the event of manual extraction of the placenta. Our results led us to conclude that length of time until delivery of the placenta is of greatest import in preventing postpartum hemorrhage in the setting of potential uterine atony. We also concluded that patients with

19 risk factors for morbidly adherent placenta are at risk of hemorrhage with manual extraction, regardless of length of the third stage. This inquiry is important for triaging and treating morbid patients on the labor and delivery floor. If characteristics associated with massive hemorrhage are known prior to MROP, physicians can adequately prepare by transferring the mother to the operating room, calling for type and crossed blood products, and notifying a team trained in surgical response to massive blood loss.

This study is unique in that we collected both historical risk factors for PPH without MROP as well as novel risk factors rarely studied in the literature, including resolved low lying placenta, prior history of a retained placenta, and prior uterine surgery. We hypothesized that any one major risk factor or multiple minor risk factors would be associated with an increased risk of significant hemorrhage after manual removal of the placenta. This hypothesis held true with approximately half of our severe postpartum hemorrhage patients having any one major or multiple minor risk factors, versus about a fourth of our controls. However, clinically, these results show that triaging MROP patients via these criteria will only catch around half of the postpartum hemorrhage cases, and will also incorrectly identify at least a quarter of patients who will not go on to hemorrhage after MROP. This suggests that in the setting of MROP, most patients, regardless of whether they experience severe PPH, are morbid a priori, and traditional risk stratification for PPH may not be as useful.

Of more use was evaluating the role of time in the third stage of labor in relation to risk factors for the development of atony or abnormal placentation and PPH. We confirmed that patients with longer third stages of labor—or delayed manual delivery of the placenta— had higher odds of experiencing severe hemorrhage. This result persisted after controlling for potential confounding. We also found that, when evaluated categorically, risk factors for atony and abnormal placentation were associated with PPH. However, with regard to atony risk factors, only multiple gestations remained significantly associated with PPH in our multivariable analysis. These findings suggest that atony risk factors may be modulated by shortening the time until delivery of the placenta. Put another way, even if a patient has exposures associated with atony and PPH, timely delivery of the placenta may lessen or

20 even eliminate a patient’s risk of hemorrhage. The fact that multiple gestations remained significant in the controlled analysis suggests that atony due to uterine over-distention behaves differently than that related to inflammation (as seen with chorioamnionitis), and may be less sensitive to third stage duration. These findings are novel and unreported in the literature.

In contrast to atony risk factors, multiple individual risk factors for morbidly adherent placenta had a strong association with PPH when analyzed in the multivariable analysis. These included resolved low lying placenta, prior uterine surgery, or history of retained placenta in a prior pregnancy, but not use of ART or prior cesarean section. This implies that, unlike with atony, time until manual delivery of the placenta is not a confounder in the setting of risk factors for abnormal placentation. Thus, our data suggests that shortening the third stage of labor when a physician plans MROP due to concern for abnormal placentation may not help lessen the mother’s risk of PPH. This is not to say that providers should delay manual removal in the setting of abnormally adherent placenta. Instead, a more rapid manual removal may not help to avoid maternal hemorrhage.

Our findings support prior studies by Magann et al (2005) and Combs et al (1991) that linked prolonged third stage of labor with PPH(24). Our results expand upon these studies, suggesting that even for deliveries in which the obstetrician chooses to manually remove the placenta as opposed to waiting for spontaneous delivery without intervention, timely removal may particularly reduce hemorrhage in patients with certain risk factors, and that perhaps our limited guidelines on timing should be reconsidered.

Given the observed importance of third stage duration with regards to PPH risk, we explored clinical factors that were associated with a short or longer time interval from delivery of infant to manual delivery of the placenta. We observed that umbilical cord avulsion and epidural analgesia were associated with a shorter third stage, and chorioamnionitis with a longer third stage. This observed relationship might suggest that avulsion of umbilical cord, disallowing its use as traction on the placenta, is often used as an indication for rapid MROP, while chorioamnionitis is not, despite its relationship to

21 atony and PPH. No other risk factors in our study were individually related to a longer third stage of labor.

The relationship between chorioamnionitis and a longer third stage may be due to obstetricians wariness of introducing a hand into the uterus in the setting of known intrapartum infection, thus delaying time until manual removal of a retained placenta. This wariness may explain why this risk factor for atony alone is related to prolonged third stage of labor. Interpreting the epidural analgesia relationship, it is possible that physicians are more likely to perform a potentially uncomfortable procedure faster (rather than wait for potential spontaneous placental delivery) under conditions of analgesia. Alternatively, it may be that patients without an existing epidural must wait for either conscious sedation or spinal anesthesia to be placed before a physician can attempt MROP, lengthening their third stage of labor.

When conceiving of this study, ART was a variable of particular interest due to its known association with placenta accreta, a condition associated with both hemorrhagic morbidity and retained placenta during delivery.(3, 33, 34) While we found that ART is likely associated with MROP, as the 19% rate among our patients with PPH and 12% among our patients without PPH reflect very high rates of ART in this MROP population. However, we found that ART showed no significant relationship to PPH in this clinical setting. This is likely because although the literature has shown ART to be significantly related to pathologic accreta, ART pregnancies may not clinically be more likely to have hemorrhagic morbidity due to adherent placenta.(3, 33, 34) Of note, standardization of definitions for placenta accreta does not exist, and not all adherent placentas or pathologic accreta cause massive hemorrhage, further confusing this relationship between ART and morbid adherence. These findings support Aziz et al’s study, which while showing higher rates of MROP in ART patients, also concluded that there is no relationship between ART and hemorrhagic complications during the third stage of labor.(36)

As opposed to ART, other variables within our morbidly adherent placenta category were positively associated with postpartum hemorrhage. These significant exposures included

22 resolved low lying placenta, prior retained placenta, and prior uterine procedure— variables that are not generally considered in studies of PPH risk in the literature. A clinician planning MROP in a patient with any of these risk factors for abnormal placentation may thus want to consider preparing for hemorrhage. Interestingly, in the morbidly adherent placenta category, prior cesarean section was not associated with the PPH group. This is striking in comparison to the literature, where prior cesarean section is associated with placenta accreta in patients with previa. This suggests that only in the setting of placental implantation over a low transverse cesarean scar is prior cesarean associated with morbid accreta—generally not the case with vaginal deliveries.

A secondary finding of interest was the high rate of pathologic placenta accreta in the cohort. In our study, the overall rate of placenta accreta on pathology in both the case and control groups was around 20%. This is much higher than the rate in the general population (~0.2%)(44), and suggests that the contribution of abnormal placentation to manual extraction may be higher than previously estimated.

Strengths, Limitations, and Further Research Our study draws strength from our use of detailed chart review to collect individual patients’ data, which allowed us to confirm and conceive of accurate criteria for PPH (>1500cc blood loss, >/=10 point hematocrit drop, or transfusion), ensuring that our case group captured true clinical morbidity. Our data collection additionally allowed us to collect previously unstudied exposures, such a resolved low-lying placenta, prior uterine surgery, and prior retained placenta. This is in contrast to most literature on PPH which draws risk factors from large national databases and discharge coding that are necessarily limited in available diagnoses. Additionally, our relatively large number of cases with severe PPH in the setting of manual removal of the placenta allowed us to determine significant relationships between many exposures and PPH.

A major limitation of our study is that it was completed retrospectively, making it difficult to find, confirm, and standardize all data, and restricting chart review to what providers documented. Case-control studies are also limited because while exposures and their

23 relationships to outcomes can be determined, PPH incidence cannot. Additionally, while hematocrit drop was used as criteria for whether or not a patient hemorrhaged, not all patients had postpartum blood draws, meaning that hematocrit drop could not be calculated for some of our study subjects. This may have underestimated our number of cases with severe PPH, biasing our hypothesis to the null. Finally, the indication for manual removal was often not indicated in the chart; therefore we are unable to definitively state why a placenta was removed at a given time.

Future research should focus on validating our prediction variables with a standardized protocol for placenta management and delivery in a prospective trial. A retrospective study from another institution validating our criteria in their MROP patients with PPH would be useful as well. Considering the delay in manual removal in the setting of chorioamnionitis, a prospective study examining the balance of risk of postpartum hemorrhage versus risk of infectious morbidity may be useful as well. Additionally, we could evaluate the use of other modalities for predicting and triaging patients likely to hemorrhage, such as bedside ultrasound or quantitative blood loss, in order to further help physicians contemplating morbidity after manual extraction of the placenta.

Summary: The most important predictor of hemorrhage after manual removal of the placenta is length of the third stage of labor. Major and minor risk factors grouped by atony and morbidly adherent placenta are clinically useful in predicting hemorrhage after MROP. If manual removal of the placenta seems likely, patients with risk factors for atony should have their placentas removed in a timely fashion, even in the setting of intrapartum chorioamnionitis. While providers may hesitate to manually remove a placenta without epidural analgesia or in the setting of chorioamnionitis, earlier placental delivery is likely beneficial in order to lessen risk of hemorrhagic morbidity.

In comparison, in the setting of patients undergoing manual removal with risk factors for morbidly adherent placenta, hemorrhage risk did not appear to be mitigated by faster removal of the placenta. This suggests that obstetricians should consider assembly of

24 appropriate team members and type and crossed blood products, and request transfer to an operating room before performing manual removal of the placenta in a patient with concern for abnormal placentation. Finally, while patients with retained placenta have a high rate of ART use, this factor does not appear to predict hemorrhage risk in this specific clinical scenario.

While time appears to be the most important factor in triaging patients at risk for severe postpartum hemorrhage after manual removal of the placenta, patients with multiple gestation, resolved low lying placenta, prior uterine surgery, or history of prior retained placenta should be considered at risk for severe postpartum hemorrhage even with timely manual removal of the placenta.

25 References

1. Combs CA, Murphy EL, Laros RK, Jr. Factors associated with postpartum hemorrhage with vaginal birth. and gynecology. 1991;77(1):69-76. 2. Titiz H, Wallace A, Voaklander DC. Manual removal of the placenta--a case control study. The Australian & New Zealand journal of obstetrics & gynaecology. 2001;41(1):41-4. 3. Kaser DJ, Melamed A, Bormann CL, Myers DE, Missmer SA, Walsh BW, et al. Cryopreserved embryo transfer is an independent risk factor for placenta accreta. Fertil Steril. 2015;103(5):1176-84 e2. 4. American College of O, Gynecologists. ACOG Practice Bulletin: Clinical Management Guidelines for Obstetrician-Gynecologists Number 76, October 2006: postpartum hemorrhage. Obstetrics and gynecology. 2006;108(4):1039-47. 5. Urner F, Zimmermann R, Krafft A. Manual removal of the placenta after vaginal delivery: an unsolved problem in obstetrics. J Pregnancy. 2014;2014:274651. 6. Bateman BT, Berman MF, Riley LE, Leffert LR. The epidemiology of postpartum hemorrhage in a large, nationwide sample of deliveries. Anesth Analg. 2010;110(5):1368- 73. 7. Kramer MS, Berg C, Abenhaim H, Dahhou M, Rouleau J, Mehrabadi A, et al. Incidence, risk factors, and temporal trends in severe postpartum hemorrhage. American journal of obstetrics and gynecology. 2013;209(5):449 e1-7. 8. Molina RL, Pace LE. A Renewed Focus on Maternal Health in the United States. The New England journal of medicine. 2017;377(18):1705-7. 9. Mahutte NG, Murphy-Kaulbeck L, Le Q, Solomon J, Benjamin A, Boyd ME. Obstetric admissions to the intensive care unit. Obstetrics and gynecology. 1999;94(2):263-6. 10. Shields LE, Wiesner S, Fulton J, Pelletreau B. Comprehensive maternal hemorrhage protocols reduce the use of blood products and improve patient safety. American journal of obstetrics and gynecology. 2015;212(3):272-80. 11. Mehrabadi A, Liu S, Bartholomew S, Hutcheon JA, Kramer MS, Liston RM, et al. Temporal trends in postpartum hemorrhage and severe postpartum hemorrhage in Canada from 2003 to 2010. J Obstet Gynaecol Can. 2014;36(1):21-33. 12. Joseph KS, Rouleau J, Kramer MS, Young DC, Liston RM, Baskett TF, et al. Investigation of an increase in postpartum haemorrhage in Canada. BJOG : an international journal of obstetrics and gynaecology. 2007;114(6):751-9. 13. Ronsmans C, Graham WJ, Lancet Maternal Survival Series steering g. Maternal mortality: who, when, where, and why. Lancet. 2006;368(9542):1189-200. 14. Callaghan WM, Kuklina EV, Berg CJ. Trends in postpartum hemorrhage: United States, 1994-2006. American journal of obstetrics and gynecology. 2010;202(4):353 e1-6. 15. Creanga AA, Berg CJ, Ko JY, Farr SL, Tong VT, Bruce FC, et al. Maternal mortality and morbidity in the United States: where are we now? Journal of women's health. 2014;23(1):3-9. 16. Callaghan WM, Mackay AP, Berg CJ. Identification of severe maternal morbidity during delivery hospitalizations, United States, 1991-2003. American journal of obstetrics and gynecology. 2008;199(2):133 e1-8. 17. Main EK, Goffman D, Scavone BM, Low LK, Bingham D, Fontaine PL, et al. National Partnership for Maternal Safety: Consensus Bundle on Obstetric Hemorrhage. Obstetrics and gynecology. 2015;126(1):155-62.

26 18. Greenbaum S, Wainstock T, Dukler D, Leron E, Erez O. Underlying mechanisms of retained placenta: Evidence from a population based cohort study. European journal of obstetrics, gynecology, and reproductive biology. 2017;216:12-7. 19. Deneux-Tharaux C, Macfarlane A, Winter C, Zhang WH, Alexander S, Bouvier-Colle MH, et al. Policies for manual removal of placenta at vaginal delivery: variations in timing within Europe. BJOG : an international journal of obstetrics and gynaecology. 2009;116(1):119-24. 20. Endler M, Grunewald C, Saltvedt S. Epidemiology of retained placenta: oxytocin as an independent risk factor. Obstetrics and gynecology. 2012;119(4):801-9. 21. Endler M, Saltvedt S, Cnattingius S, Stephansson O, Wikstrom AK. Retained placenta is associated with pre-, stillbirth, giving birth to a small-for-gestational-age infant, and spontaneous : a national register-based study. BJOG : an international journal of obstetrics and gynaecology. 2014;121(12):1462-70. 22. Rogers J, Wood J, McCandlish R, Ayers S, Truesdale A, Elbourne D. Active versus expectant management of third stage of labour: the Hinchingbrooke randomised controlled trial. Lancet. 1998;351(9104):693-9. 23. Magann EF, Niederhauser A, Doherty DA, Chauhan SP, Sandlin AT, Morrison JC. Reducing hemodynamic compromise with placental removal at 10 versus 15 minutes: a randomized clinical trial. American journal of perinatology. 2012;29(8):609-14. 24. Magann EF, Evans S, Chauhan SP, Lanneau G, Fisk AD, Morrison JC. The length of the third stage of labor and the risk of postpartum hemorrhage. Obstetrics and gynecology. 2005;105(2):290-3. 25. Combs CA, Laros RK, Jr. Prolonged third stage of labor: morbidity and risk factors. Obstetrics and gynecology. 1991;77(6):863-7. 26. Intrapartum Care: Care of Healthy Women and Their Babies During . National Institute for Health and Care Excellence: Clinical Guidelines. London2014. 27. Dombrowski MP, Bottoms SF, Saleh AA, Hurd WW, Romero R. Third stage of labor: analysis of duration and clinical practice. American journal of obstetrics and gynecology. 1995;172(4 Pt 1):1279-84. 28. Coviello EM, Grantz KL, Huang CC, Kelly TE, Landy HJ. Risk factors for retained placenta. American journal of obstetrics and gynecology. 2015;213(6):864 e1- e11. 29. Thurn L, Lindqvist PG, Jakobsson M, Colmorn LB, Klungsoyr K, Bjarnadottir RI, et al. Abnormally invasive placenta-prevalence, risk factors and antenatal suspicion: results from a large population-based pregnancy cohort study in the Nordic countries. BJOG : an international journal of obstetrics and gynaecology. 2016;123(8):1348-55. 30. Miller DA, Chollet JA, Goodwin TM. Clinical risk factors for placenta previa-placenta accreta. American journal of obstetrics and gynecology. 1997;177(1):210-4. 31. Bjurstrom J, Collins S, Langhoff-Roos J, Sundberg K, Jorgensen A, Duvekot JJ, et al. Failed manual removal of the placenta after vaginal delivery. Arch Gynecol Obstet. 2017. 32. Roeca C, Little SE, Carusi DA. Pathologically Diagnosed Placenta Accreta and Hemorrhagic Morbidity in a Subsequent Pregnancy. Obstetrics and gynecology. 2017;129(2):321-6. 33. Esh-Broder E, Ariel I, Abas-Bashir N, Bdolah Y, Celnikier DH. Placenta accreta is associated with IVF pregnancies: a retrospective chart review. BJOG : an international journal of obstetrics and gynaecology. 2011;118(9):1084-9.

27 34. Ishihara O, Araki R, Kuwahara A, Itakura A, Saito H, Adamson GD. Impact of frozen- thawed single-blastocyst transfer on maternal and neonatal outcome: an analysis of 277,042 single-embryo transfer cycles from 2008 to 2010 in Japan. Fertil Steril. 2014;101(1):128-33. 35. Elenis E, Svanberg AS, Lampic C, Skalkidou A, Akerud H, Sydsjo G. Adverse obstetric outcomes in pregnancies resulting from oocyte donation: a retrospective cohort case study in Sweden. BMC pregnancy and childbirth. 2015;15:247. 36. Aziz MM, Guirguis G, Maratto S, Benito C, Forman EJ. Is there an association between assisted reproductive technologies and time and complications of the third stage of labor? Arch Gynecol Obstet. 2016;293(6):1193-6. 37. Committee on Practice B-O. Practice Bulletin No. 183: Postpartum Hemorrhage. Obstetrics and gynecology. 2017;130(4):e168-e86. 38. Mehrabadi A, Hutcheon JA, Liu S, Bartholomew S, Kramer MS, Liston RM, et al. Contribution of placenta accreta to the incidence of postpartum hemorrhage and severe postpartum hemorrhage. Obstetrics and gynecology. 2015;125(4):814-21. 39. Wetta LA, Szychowski JM, Seals S, Mancuso MS, Biggio JR, Tita AT. Risk factors for uterine atony/postpartum hemorrhage requiring treatment after vaginal delivery. American journal of obstetrics and gynecology. 2013;209(1):51 e1-6. 40. Dilla AJ, Waters JH, Yazer MH. Clinical validation of risk stratification criteria for peripartum hemorrhage. Obstetrics and gynecology. 2013;122(1):120-6. 41. Allen VM, Baskett TF, O'Connell CM, McKeen D, Allen AC. Maternal and perinatal outcomes with increasing duration of the second stage of labor. Obstetrics and gynecology. 2009;113(6):1248-58. 42. Practice ACoO. ACOG Committee opinion. Number 266, January 2002 : placenta accreta. Obstetrics and gynecology. 2002;99(1):169-70. 43. Eshkoli T, Weintraub AY, Sergienko R, Sheiner E. Placenta accreta: risk factors, perinatal outcomes, and consequences for subsequent births. American Journal of Obstetrics & Gynecology. 2013;208(3):219.e1-7. 44. Committee on Obstetric P. ACOG committee opinion. Placenta accreta. Number 266, January 2002. American College of Obstetricians and Gynecologists. International journal of gynaecology and obstetrics: the official organ of the International Federation of Gynaecology and Obstetrics. 2002;77(1):77-8.

28 Tables and Figures:

Table 1: Maternal and labor and delivery Characteristics in both Case (PPH) and Control (no PPH) Groups.

PPH no PPH P-Value N=172 N=172 Age (years) 33.4 35.2 <0.01 (17.3-48.9) (17.0-50.4) BMI at end of pregnancy 28 (20-54) 30 (21-50) 0.02 (kg/m2) Ethnicity 0.94 White 110 (64%) 105 (61%) Black 23 (13%) 27 (16%) Hispanic 23 (13%) 24 (14%) Asian 15 (9%) 14 (8%) Other 1 (1%) 2 (1%) Parity 0 (0-6) 1 (0-6) <0.01 Gestational age (weeks) 38.8 39.1 0.39 (25.1-442.0) (24.7-41.9) Number of prior D&Ea 0 (0-4) 0 (0-4) <0.01 proceduresb Prior cesarean 14 (8%) 15 (9%) 0.84 Prior abortions (SABc/TABd) 0 (0-6) 0 (0-7) 0.79 Epidural analgesia 145 (85%) 159 (93%) 0.02 Operative vaginal delivery 24 (14%) 19 (11%) 0.42 Cord avulsion 24 (14%) 34 (20%) 0.16 Episiotomy 16 (9%) 9 (5%) 0.15 Laceration (any) 129 (75%) 112 (65%) 0.06 Length of 2nd Stage of Labor 48 (1-345) 38 (1-255) 0.04 (min) Length of 3rd Stage of Labor 38 (1-213) 23 (1-169) <0.01 (min) a) D&E- Dilation and Evacuation preformed in the setting of prior pregnancies (spontaneous or therapeutic abortion, or treatment of postpartum hemorrhage or retained products of conception). b) While median prior number of D&E procedures was the same for case and control group, overall our PPH case group were significantly more likely to have had more D&Es prior to index pregnancy. c) SAB – spontaneous abortion. d) TAB – therapeutic abortion, either with medical or surgical intervention.

29 Table 2: Delivery outcomes in both Case (PPH) and Control (no PPH) groups.

PPH no PPH P-Value N=172 N=172 Estimated blood loss 1000 (100-4000) 400 (75-1400) n/a (cc) Hematocrit drop 12.3 (0.3-23.3) 4.1 (0.1-9.7) n/a (percent) Transfusion 84 (49%) 0 n/a Dilation and Curettage 103 (60%) 14 (8%) <0.01 Uterine inversion 3 (2%) 0 0.25 Hysterectomy 8 (5%) 0 <0.01 Bakri Balloon <0.01 Placed 40 (23%) 0 Failed placement 3 (2%) 1 (1%) None 129 (75%) 171 (99%) Postpartum 15 (9%) 3 (2%) <0.01 readmission Delayed Postpartum 16 (10%) 10 (6%) 0.18 procedure Pathologic diagnosisa Accreta 30/152 (20%) 19/118 (16%) 0.72 Marginal or velamentous 28/149 (19%) 21/117 (18%) 1 cord insertion Infection / 29/149 (19%) 19/117 (16%) 0.73 inflammation a) Denominators in the individual cells reflect the number of subjects that has a specimen sent for pathology. For accreta this reflects a placenta or D&C specimen. For the other diagnoses this reflects placenta specimens only.

30

Table 3: Risk Factor association with postpartum hemorrhage, categorized by role in the development of retained placenta and PPH.

PPH no PPH P- Value N = 172 N = 172 Atony: Birth weight>4000g 12 (7%) 14 (8%) 0.67 BMI > 40 kg/m2 5 (3%) 11 (6%) 0.14 Chorioamnionitis 29 (17%) 13 (8%) 0.01 Multiple gestations 17 (10%) 5 (3%) 0.01 2nd stage > 2 hr 49 (28%) 29 (17%) <0.01 Parity > 4 1 (1%) 5 (3%) 0.14 Any Atony Risk Factor 85 (49%) 63 (37%) 0.01 Abnormal Placentation Prior cesarean 14 (8%) 15 (9%) 0.84 Prior uterine surgerya 42 (24%) 10 (6%) <0.01 Resolved low lying 26 (15%) 13 (8%) 0.03 placentab Ashermans/ Synechiae 1 (1%) 1 (1%) 1 Prior pathologic accreta 4 (2%) 0 0.99 ART 33 (19%) 21 (12%) 0.09 >2 Prior D&C 5 (3%) 9 (5%) 0.29 Signs of accreta on 3 (2%) 0 0.99 ultrasound Prior retained placenta 23 (13%) 11 (6%) 0.03 Any abnormal 99 (58%) 65 (38%) <0.01 placentation Risk Factor Other PPH risk factor Leiomyoma >= 6 cm 3 (2%) 4 (2%) 0.71 Acute Abruption 1 (1%) 4 (2%) 0.22 Preeclampsia 17 (10%) 13 (8%) 0.43 Laceration (high 9 (5%) 4 (2%) 0.15 order)c Any Other PPH Risk 28 (16%) 25 (15%) 0.66 Factor Risk Factors Independent of grouping Any risk factor 143 (83%) 114 (66%) <0.01 One major or two 88 (51%) 45 (26%) <0.01 minord a) Prior uterine surgeries included myomectomy, any operative hysteroscopy, or endometrial ablation.

31 b) Low lying placenta was defined as ultrasonographic diagnosis of placenta previa or the placenta coming within 2cm of the cervical os during the current pregnancy. c) High order lacerations included 3rd and 4th degree vaginal tears after delivery of the infant. d) Major risk factors include acute abruption and radiologic signs of accreta. All other risk factors are included as minor.

32 Table 4: Third stage of labor by centiles in both Case (PPH) and Control (no PPH) groups.

Centiles PPH (minutes) No PPH (Minutes) Maximum 213 169 95th percentile 156 76 75th percentile 79 38 50th percentile 38 23 25th percentile 18 10 5th percentile 6 5 Minimum 1 1

Table 5: Characteristics associated with 30 minute interval categories of the third stage of labor.

60 P-Value minutes minutes minutes N (%) N (%) N (%) Postpartum hemorrhage 66 (36%) 51 (57%) 55 (76%) <0.01 Atony 53 (29%) 33 (37%) 21 (29%) 0.37 Accreta 55 (30%) 23 (26%) 30 (42%) 0.08 Cord avulsion 43 (24%) 7 (8%) 8 (12%) <0.01 Operative vaginal delivery 25 (14%) 8 (9%) 10 (14%) 0.51 Multiple gestation 13 (7%) 5 (6%) 4 (6%) 0.88 Multiparous 106 (58%) 53 (60%) 31 (43%) 0.06 History of retained placenta 13 (7%) 11 (12%) 10 (14%) 0.17 Chorioamnionitis 16 (9%) 9 (10%) 17 (24%) <0.01 Epidural analgesia 171 (93%) 81 (91%) 54 (75%) <0.01 Prolonged 2nd Stagea 36 (20%) 24 (27%) 18 (25%) 0.35 a) Second stage of labor (time between full cervical dilation and delivery of the infant) lasting more than 2 hours.

33 Table 6: Variables associated with postpartum hemorrhage after controlling for confoundersa:

P-Value OR 95% CI Length of 3rd stage <0.01 31-60 vs. 60 vs. 4 were run in this multivariable analysis but did not retain significance.

34 Figures:

Figure 1: Percentage of Case (PPHa) and Control (No PPHb) patients delivering their placentas over time.

Percentage of Case (PPH) and Control (No PPH) patients delivering their placentas over time

30

25

20

15

10

5

0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 a) Blue bars represent Case (PPH) placenta deliveries b) Red bars represent Control (no PPH) placenta deliveries

35