The increase in postpartum hemorrhage among high-income countries such as Australia, Canada, Ireland, Norway, Scotland, and the United States over the past two decades remains unexplained.1–7 The subtype of postpartum hemorrhage underlying the increase has been identified as atonic postpartum hemorrhage2,4,7; in Canada, the rate of atonic postpartum hemorrhage increased by 29% (95% confidence interval [CI] 26–33%) from 3.9% in 2003 to 5.0% in 2010.8 However, this increase could not be explained by changes in maternal age, obesity, multifetal gestation, labor induction, oxytocin augmentation, and other maternal, fetal, infant, and obstetric factors.9 The International Postpartum Hemorrhage Collaborative Group recommended an investigation into the role of placenta accreta,3 a severe and potentially life-threatening pregnancy complication, reported to have increased as a consequence of the increase in cesarean deliveries over the past 20–30 years.10,11 The incidence of placenta accreta ranges between 1 and 90 per 10,000 deliveries and varies based on the population frequency of previous cesarean deliveries.12
The purpose of this study was to describe the association between placenta accreta and postpartum hemorrhage and severe postpartum hemorrhage (eg, postpartum hemorrhage with blood transfusion, hysterectomy, or other procedures to control bleeding) and to determine the population-attributable fraction of placenta accreta related to postpartum hemorrhage and severe postpartum hemorrhage (ie, to determine the incidence of postpartum hemorrhage attributable to placenta accreta). We also explored the relationship between placenta accreta and risk factors for placenta accreta such as older maternal age, grand multiparity, multifetal gestation, cesarean delivery, and placenta previa to provide a context for anticipating future temporal trends in placenta accreta incidence.
MATERIALS AND METHODS
The study was based on all women delivering a live birth or a stillbirth in the hospital in Canada (excluding Quebec) in 2009 and 2010. Data for the study were obtained from the Canadian Institute for Health Information's Discharge Abstract Database. This database contains demographic information, medical history, and diagnoses and procedures associated with all hospitalizations in Canada excluding Quebec. The database has been validated for the accuracy of maternal and other perinatal information; postpartum hemorrhage had high sensitivity (90.2%, 95% CI 86.2–93.3) and specificity (98.2%, 95% CI 97.8–98.5) and procedures such as blood transfusion had reasonably high sensitivity 85.7% (42.1–99.6) and very high specificity (99.8%, 95% CI 99.6–99.9).13 Our study used a retrospective cohort design, including all hospital deliveries in Canada (excluding Quebec) that resulted in a live birth or a stillbirth between April 2009 and March 2011 (referred to as calendar years 2009–2010). The years of study were chosen since the Canadian Institute for Health Information began to routinely collect diagnostic information on placenta accreta in 2009.
Placenta accreta was defined using the International Statistical Classification of Diseases and Related Health Problems (Canadian version) code for morbidly adherent placenta (O43.2) and included placenta accreta, increta, and percreta. The outcome of postpartum hemorrhage was defined (using International Statistical Classification of Diseases and Related Health Problems [Canadian version] codes O72.0 to O72.3) as a blood loss of 500 mL or greater after vaginal delivery or 1,000 mL or greater after cesarean delivery or as noted in the medical record by the health care provider. Subtypes of postpartum hemorrhage were identified based on the clinical diagnosis noted in the medical record and included atonic postpartum hemorrhage (occurring within 24 hours after delivery), third-stage hemorrhage (which included retained, trapped, or adherent placenta occurring in the third stage of labor), delayed and secondary postpartum hemorrhage (occurring between 24 hours and 6 weeks after delivery), and postpartum hemorrhage resulting from coagulation defects. Severe postpartum hemorrhage was defined as postpartum hemorrhage occurring in conjunction with blood transfusion, hysterectomy, or other procedures to control bleeding (ie, uterine suturing, pelvic ligation, or pelvic embolization). Retained placenta without hemorrhage referred to retained placenta or portions of the placenta after delivery that was not accompanied by hemorrhage.
Risk factors examined for placenta accreta included maternal age (younger than 20, 20–24, 25–29, 30–34, 35–39, 40 years or older), parity (nulliparous, 1, 2–4, 5 or more, missing), multifetal gestation, placenta previa plus previous cesarean delivery, placenta previa without previous cesarean delivery, and previous cesarean delivery without placenta previa. International Statistical Classification of Diseases and Related Health Problems (Canadian version) codes used for the study are summarized in Appendix 1. Rates and 95% CIs were calculated for placenta accreta, placenta accreta with postpartum hemorrhage, placenta accreta with severe postpartum hemorrhage, and placenta accreta by subtype of postpartum hemorrhage. Logistic regression analyses were used to estimate crude and adjusted odds ratios for risk factors of placenta accreta. The rate ratios and 95% CIs for the association between placenta accreta and postpartum hemorrhage and severe postpartum hemorrhage were estimated as was the population-attributable fraction of postpartum hemorrhage and severe postpartum hemorrhage resulting from placenta accreta. The population-attributable fraction refers to the proportion of the incidence of any disease or outcome (eg, postpartum hemorrhage) in the population that is attributable to the determinant (eg, placenta accreta). It is affected by both the population-level rates of the determinant and the strength of association between the determinant and the disease or outcome. The population-attributable fraction in this context provides an estimate of the fraction of postpartum hemorrhage or severe postpartum hemorrhage that would be prevented if placenta accreta was eliminated from the population. Confidence intervals for the population-attributable fractions were calculated using OpenEpi (www.OpenEpi.com) as described by Abramsom and Gahlinger.14
The statistical software SAS 9.3, Stata 12.1, and OpenEpi (www.OpenEpi.com) 3.01 were used for the analyses. This study was carried out under the auspices of the Public Health Agency of Canada's Canadian Perinatal Surveillance System. The data source included denominalized information, and the Public Health Agency of Canada, which is mandated to monitor maternal and infant health in the Canadian population, is not required to obtain ethics review board approval for such studies.
The incidence of placenta accreta was 14.4 (95% CI 13.4–15.4) per 10,000 deliveries (819 cases among 570,637 deliveries), and the case fatality rate for placenta accreta was 0.12%. The incidence of placenta accreta with postpartum hemorrhage was 7.2 (95% CI 6.5–8.0) per 10,000 deliveries (Table 1). One fourth of all patients with placenta accreta required a blood transfusion (incidence 3.5/10,000 deliveries, 95% CI 3.1–4.1/10,000); 19% had postpartum hemorrhage and a blood transfusion (incidence 2.8/10,000 deliveries, 95% CI 2.3–3.2/10,000; Table 1). Among patients with placenta accreta, 17% underwent a hysterectomy with or without postpartum hemorrhage (incidence 2.4/10,000, 95% CI 2.0–2.8/10,000), whereas 11.2% had both postpartum hemorrhage and a hysterectomy (incidence 1.6/10,000 deliveries, 95% CI 1.3–2.0/10,000). Retained placenta without hemorrhage occurred in 17% of patients. Although 50% of women with placenta accreta had postpartum hemorrhage, women with placenta accreta predominantly received a clinical diagnosis of third-stage postpartum hemorrhage (41% of all cases), whereas a minority (7% of all cases) were labeled with a clinical diagnosis of atonic postpartum hemorrhage. Most cases of atonic postpartum hemorrhage with placenta accreta were severe; 0.63 (95% CI 0.42–0.87) per 10,000 women had atonic postpartum hemorrhage with blood transfusion, hysterectomy, or other procedures to control bleeding. Table 1 summarizes the rate of placenta accreta, postpartum hemorrhage, and related complications.
Risk factors for placenta accreta were older maternal age (30–34 years, 35–39 years, and 40 years or older compared with 20–24 years; Table 2) and low and high parity (nulliparity, 2–4, and 5 or more). Multifetal gestation was also associated with a higher risk of placenta accreta. Placenta previa (in the current pregnancy) in conjunction with a previous cesarean delivery was most strongly associated with placenta accreta (adjusted odds ratio [OR] 91.6, 95% CI 70.5–119.1) followed by placenta previa without previous cesarean delivery (adjusted OR 13.3, 95% CI 10.0–17.8). Previous cesarean delivery without placenta previa was also associated with a significantly increased risk of placenta accreta (adjusted OR 1.44, 95% CI 1.17–1.77).
Although placenta accreta was strongly associated with postpartum hemorrhage (rate ratio 8.3, 95% CI 7.7–8.9; Table 3), the population-attributable fraction was only 1.0%; that is, only 1.0% of the incidence of postpartum hemorrhage was attributable to placenta accreta. On the other hand, placenta accreta was strongly associated with postpartum hemorrhage with hysterectomy (rate ratio 286, 95% CI 226–361) and had a population-attributable fraction of 29%; that is, 29% of the incidence of postpartum hemorrhage with hysterectomy was the result of placenta accreta. Placenta accreta was strongly associated with all severe forms of postpartum hemorrhage; however, the population-attributable fraction was only 7.6% for postpartum hemorrhage with procedures to control bleeding and 5.5% for postpartum hemorrhage with blood transfusion. Table 3 presents the rate ratios (RRs) and population-attributable fractions for postpartum hemorrhage and severe postpartum hemorrhage by subtype.
Of all the subtypes of postpartum hemorrhage, placenta accreta was most strongly associated with third-stage postpartum hemorrhage (RR 52.4, 95% CI 48.0–57.1), and the population-attributable fraction was the highest at 6.9%. The population-attributable fraction for third-stage postpartum hemorrhage with hysterectomy was 63.2%, for third-stage postpartum hemorrhage with other procedures to control bleeding it was 36.8%, and for third-stage postpartum hemorrhage with blood transfusion it was 18.1%.
Our study showed that approximately 50% of the patients with placenta accreta experienced postpartum hemorrhage, and 22.6% experienced a severe form of postpartum hemorrhage (postpartum hemorrhage with blood transfusion, hysterectomy, or other procedures to control bleeding). Placenta accreta was strongly associated with postpartum hemorrhage and most strongly associated with third-stage hemorrhage. The relative infrequency of placenta accreta (14.4/10,000 deliveries) resulted in a low population-attributable fraction for postpartum hemorrhage (1%). Given the modest association between placenta accreta and atonic postpartum hemorrhage (RR 1.45, 95% CI 1.14–1.86; population-attributable fraction 1%, 95% CI 0.0–10%), increases in placenta accreta cannot explain the recent increases in postpartum hemorrhage. The frequency of placenta accreta and postpartum hemorrhage (7.2/10,000 deliveries) is relatively low compared with the magnitude of the absolute temporal increase in atonic postpartum hemorrhage (absolute increase of 1.1% from 3.9% in 2003 to 5.0% in 2010).8 Together, these findings suggest that potential temporal increases in placenta accreta frequency cannot explain the recent rise in atonic postpartum hemorrhage.
Our study showed that placenta accreta accounted for 29% of the incidence of postpartum hemorrhage with hysterectomy, which increased significantly in Canada from 4.9 to 5.8 per 10,000 deliveries between 2003 and 2010.8 Thus, an increase in placenta accreta could have contributed significantly to the rising incidence of postpartum hemorrhage with hysterectomy in Canada. Our study also suggests that increases in placenta accreta could have partly contributed to changes in the incidence of other severe forms of postpartum hemorrhage such as postpartum hemorrhage with blood transfusion and postpartum hemorrhage with other procedures to control bleeding. Changes in placenta accreta rates may have also contributed to severe forms of third-stage hemorrhage, although the absolute rates of such postpartum hemorrhage were low overall.
The strengths of our study were the population-based design that included all hospital deliveries in Canada (excluding Quebec), whereas most previous studies on this topic have been restricted to tertiary care centers.12,15–17 Our data source has been previously validated and postpartum hemorrhage and several other obstetric outcomes have been shown to have high validity in our database.13 Some of the limitations of our study included potential underdiagnoses or underreporting of placenta accreta, which could have led to its underrepresentation among cases of postpartum hemorrhage. It is likely that pathologic confirmation may be more commonly undertaken after severe cases of placenta accreta that require hysterectomy. Nevertheless, the rate of placenta accreta in our study (14.4/10,000 deliveries) was higher than that reported in another population-based study from Ireland, which reported a rate of 8.5 per 10,000 deliveries between 2005 and 2009.7 The rate of placenta accreta is expected to vary based on the prevalence of previous cesarean deliveries combined with the fertility rate of women with previous cesarean deliveries.12 Our study relied on clinical diagnoses of placenta accreta recorded in medical charts and this may have led to an under- or overdiagnosis of placenta accreta. In addition, the validity of subcategories of postpartum hemorrhage is unclear with some evidence that the contribution of atonic postpartum hemorrhage is overestimated.18 Other study limitations included the restricted timeframe for the study, which precluded our ability to study changes in placenta accreta frequency over time. This limitation was the result of routine collection of placenta accreta diagnosis only becoming instituted in our data source in 2009. Finally, we could not ascertain whether postpartum hemorrhage occurred before or after the hysterectomy, the procedures to control bleeding, or blood transfusion. For instance, some hysterectomies, uterine suturing, and ligation and embolization of pelvic blood vessels may have represented preemptive strategies to control bleeding or could have followed antepartum hemorrhage. Regardless, our interest was in the population rates of placenta accreta and associated postpartum hemorrhage, and in all cases, it is clear that placenta accreta preceded postpartum hemorrhage.
Our study revealed a case fatality rate of 0.12%, whereas previous studies have reported mortality rates that range from 0% to 0.3% and 7% for the most severe forms of placenta accreta.12,15,16 In addition, half the patients in our study did not have accompanying postpartum hemorrhage. This finding may be a result of less severe forms of placenta accreta being diagnosed, prenatal diagnosis of placenta accreta, and prompt or prophylactic intervention to prevent bleeding.15,19 Other studies have shown similar rates of postpartum hemorrhage (51.5%) associated with placenta previa that was managed conservatively.20 Currently, the appropriate management of placenta accreta remains controversial, particularly with regard to the use of treatments designed to preserve fertility.21
Similar to other research, our study found that concurrent placenta previa with a previous cesarean delivery was most strongly associated with placenta accreta (OR 91.6, 95% CI 70.5–119.1). This finding supports the American College of Obstetricians and Gynecologists' recommendation to offer ultrasound screening for placenta accreta to women with a uterine scar and placenta previa.21 However, only 8.8% of the patients with placenta accreta in our study had both risk factors, whereas a larger proportion of the patients with placenta accreta (15%) had only a previous cesarean delivery with no placenta previa (implying that such risk-based screening would miss most cases). This finding supports the recommendation of the Confidential Enquiries into Maternal Deaths in the United Kingdom to screen all women with a previous cesarean delivery to have their placental site determined.22 Our data source, the discharge abstract database, did not have information on the number of previous cesarean deliveries for each woman, but previous research has found an increasing risk of placenta accreta with increasing number of previous cesarean deliveries. For example, women with placenta previa had a four times greater risk of placenta accreta given one previous cesarean delivery, and this increased to an 11 times higher risk with two or more previous cesarean deliveries.16 Risk factors other than placenta previa and previous cesarean delivery examined in previous studies (some of which were not available in the discharge abstract database) included previous curettage, Asherman's syndrome, history of abortion or miscarriage, other prior uterine surgery, or trauma and also female fetal sex.11,15–17,23
The finding that nulliparity was significantly associated with placenta accreta in adjusted analyses was unexpected and was possibly an artefact of adjustment for previous caesarean delivery (a variable in the causal pathway between other risk factors and placenta accreta). Another unexpected finding in our study was related to postpartum hemorrhage associated with placenta accreta and blood transfusion; not all women who had severe postpartum hemorrhage received a blood transfusion. This may have been because severe postpartum hemorrhage in our study was defined to include patients in whom any procedure was used to control bleeding, some of which may have been prophylactic intervention in patients in whom the accreta was diagnosed antenatally. Of note, rates of blood transfusion in Canada have remained low relative to other countries.24
In conclusion, placenta accreta is too infrequent a condition to account for the increase in postpartum hemorrhage observed in Canada and elsewhere in recent decades. However, potential temporal increases in placenta accreta may have contributed to temporal increases in postpartum hemorrhage with hysterectomy and other severe forms of postpartum hemorrhage. Placenta previa was most strongly associated with placenta accreta, but the majority of patients with placenta accreta had a previous cesarean delivery without concurrent placenta previa, supporting the practice of assessing placentation among all women with a previous cesarean delivery. Future studies and population surveillance are necessary for monitoring the rates of placenta accreta and its contribution to severe postpartum hemorrhage. In addition, studies should explore other potential causes of the increase in postpartum hemorrhage observed in high-income countries.
1. Ford JB, Roberts CL, Simpson JM, Vaughan J, Cameron CA. Increased postpartum hemorrhage rates in Australia. Int J Gynaecol Obstet 2007;98:237–43.
2. Joseph KS, Rouleau J, Kramer MS, Young DC, Liston RM, Baskett TF; Maternal Health Study Group of the Canadian Perinatal Surveillance System. Investigation of an increase in postpartum haemorrhage in Canada. BJOG 2007;114:751–9.
3. Knight M, Callaghan WM, Berg C, Alexander S, Bouvier-Colle MH, Ford JB, et al.. Trends in postpartum hemorrhage in high resource countries: a review and recommendations from the International Postpartum Hemorrhage Collaborative Group. BMC Pregnancy Childbirth 2009;9:55.
4. Callaghan WM, Kuklina EV, Berg CJ. Trends in postpartum hemorrhage: United States, 1994–2006. Am J Obstet Gynecol 2010;202:353.e1–6.
6. Rossen J, Okland I, Bjarte Nilsen O, Eggebø TM. Is there an increase of postpartum hemorrhage, and is severe hemorrhage associated with more frequent use of obstetric interventions? Obstet Gynecol Surv 2010;89:1248–55.
7. Lutomski JE, Byrne BM, Devane D, Greene RA. Increasing trends in atonic postpartum haemorrhage in Ireland: an 11-year population-based cohort study. BJOG 2012;119:306–14.
8. Mehrabadi A, Liu S, Bartholomew S, Hutcheon J, Kramer M, Liston R, et al.. Temporal trends in postpartum hemorrhage in Canada from 2003 to 2009. J Obstet Gynaecol Can 2014;36:21–33.
9. Mehrabadi A, Hutcheon JA, Lee L, Kramer MS, Liston RM, Joseph KS. Epidemiologic investigation of a temporal increase in atonic postpartum haemorrhage: a population-based retrospective cohort study. BJOG 2013;120:853–62.
10. Timor-Tritsch IE, Monteagudo A. Unforeseen consequences of the increasing rate of cesarean deliveries: early placenta accreta and cesarean scar pregnancy. A review. Am J Obstet Gynecol 2012;207:14–29.
11. Wu S, Kocherginsky M, Hibbard JU. Abnormal placentation: twenty-year analysis. Am J Obstet Gynecol 2005;192:1458–61.
12. Gielchinsky Y, Rojansky N, Fasouliotis SJ, Ezra Y. Placenta accreta—summary of 10 years: a survey of 310 cases. Placenta 2002;23:210–4.
13. Joseph KS, Fahey J; Canadian Perinatal Surveillance System. Validation of perinatal data in the Discharge Abstract Database of the Canadian Institute for Health Information. Chronic Dis Can 2009;29:96–100.
14. Abramson JH, Gahlinger PM. Computer programs for epidemiologists: PEPI version 4.0. Salt Lake City (UT): Sagebrush Press; 2001.
15. Eller AG, Porter TF, Soisson P, Silver RM. Optimal management strategies for placenta accreta. BJOG 2009;116:648–54.
16. Miller DA, Chollet JA, Goodwin TM. Clinical risk factors for placenta previa-placenta accreta. Am J Obstet Gynecol 1997;177:210–4.
17. Morlando M, Sarno L, Napolitano R, Capone A, Tessitore G, Maruotti GM, et al.. Placenta accreta: incidence and risk factors in an area with a particularly high rate of cesarean section. Acta Obstet Gynecol Scand 2013;92:457–60.
18. Ford JB, Algert CS, Kok C, Choy MA, Roberts CL. Hospital data reporting on postpartum hemorrhage: under-estimates recurrence and over-estimates the contribution of uterine atony. Matern Child Health J 2012;16:1542–8.
19. Chantraine F, Braun T, Gonser M, Henrich W, Tutschek B. Prenatal diagnosis of abnormally invasive placenta reduces maternal peripartum hemorrhage and morbidity. Acta Obstet Gynecol Scand 2013;92:439–44.
20. Sentilhes L, Ambroselli C, Kayem G, Provansal M, Fernandez H, Perrotin F, et al.. Maternal outcome after conservative treatment of placenta accreta. Obstet Gynecol 2010;115:526–34.
21. ACOG committee opinion. Placenta accreta. Number 266, January 2002. American College of Obstetricians and Gynecologists. Int J Gynaecol Obstet 2002;77:77–8.
22. Cantwell R, Clutton-Brock T, Cooper G, Dawson A, Drife J, Garrod D, et al.. Saving mothers' lives: reviewing maternal deaths to make motherhood safer: 2006–2008. The Eighth Report of the Confidential Enquiries into Maternal Deaths in the United Kingdom. BJOG 2011;118(suppl 1):1–203.
23. Read JA, Cotton DB, Miller FC. Placenta accreta: changing clinical aspects and outcome. Obstet Gynecol 1980;56:31–4.
24. Joseph KS, Liu S, Rouleau J, Kirby RS, Kramer MS, Sauve R, et al.. Severe maternal morbidity in Canada, 2003 to 2007: surveillance using routine hospitalization data and ICD-10CA codes. J Obstet Gynaecol Can 2010;32:837–46.
International Classification of Diseases, 10th Revision, and the Canadian Classification of Interventions Diagnosis and Procedure Codes Used in the Study