OBJECTIVE: To identify factors related to retained placenta in the context of contemporary obstetric practice.
METHODS: This was a case-control study comparing 408 cases of retained placenta and an equivalent number of control individuals. Epidemiological and delivery-related variables were registered in computerized prenatal and in-hospital medical records. Univariable and multivariable logistic regressions were used for estimation of risk ratios and statistical significance.
RESULTS: Independent risk factors for retained placenta were: previous retained placenta (odds ratio [OR] 12.61, 95% confidence interval [CI] 3.61–44.08); preterm delivery (OR 3.28, 95% CI 1.60–6.70); oxytocin use for 195–415 minutes (OR 2.00, 95% CI 1.20–3.34); oxytocin use more than 415 minutes (OR 6.55, 95% CI 3.42–12.54, number needed to harm 2.3); preeclampsia (OR 2.85, 95% CI 1.20–6.78); two or more previous miscarriages (OR 2.62, 95% CI 1.31–5.20); and one or more previous abortion (OR 1.58, 95% CI 1.09–2.28). Parity of two or more had a seemingly protective effect (OR 0.40, 95% CI 0.24–0.70), as did smoking at the start of pregnancy (OR 0.28, 95% CI 0.09–0.88). Retained placenta was significantly associated with an increased risk of postpartum hemorrhage. The OR related to blood loss exceeding 500 mL, 1,000 mL, and 2,000 mL and the need for blood transfusion was 33.07 (95% CI 20.57–53.16), 43.44 (95% CI 26.57–71.02), 111.24 (95% CI 27.26–454.00), and 37.48 (95% CI 13.63–103.03), respectively. Diabetes was numerically overrepresented in the case group, but the power of the study to detect a significant difference in risk outcome was insufficient.
CONCLUSION: Identifying risk factors for retained placenta is important in the assessment of women after delivery. The increased risk associated with duration of oxytocin use is of interest, considering its widespread use.
LEVEL OF EVIDENCE: II
Awareness of risk factors for retained placenta, particularly prolonged oxytocin use, is important in identifying women at risk of this condition and subsequent risk of hemorrhage.
From the Department of Obstetrics and Gynecology, Stockholm South General Hospital, Stockholm, Sweden.
Funded by Karolinska Institutet, Department of Clinical Science and Education, Department of Obstetrics and Gynecology, South General Hospital, Stockholm, Sweden.
The authors thank Pardon Masuka, medical statistician from Cape Town, South Africa, for statistical review.
Corresponding author: Margit Endler, MD, Sjukhusbacken 10, 118 83 Stockholm, Sweden; e-mail: email@example.com.
Financial Disclosure The authors did not report any potential conflicts of interest.
Retained placenta occurs as a complication to delivery in 2–3% of vaginal deliveries in developed countries.1 The exact etiology behind retained placenta is not known and probably complex.2 Untreated retained placenta can result in fatal hemorrhage and is the second leading cause of postpartum hemorrhage after uterine atony.3 The condition usually requires manual removal under anesthesia, which has been shown to increase the risk of postpartum endometritis and possibly exacerbate blood loss.4–6 The incidence appears to be lower in less industrialized nations.1 This may be attributable to an unknown number of undiagnosed cases, but could also suggest that certain delivery-related or epidemiological factors, more common in high-resource settings, influence the condition unfavorably.
Previous studies have identified several epidemiological and delivery-related risk factors pertaining to the condition, and three previous studies have used a multivariable approach with a case number exceeding 100 patients. Risk factors for which there is some consensus between studies are: previous retained placenta, preterm delivery, grandmultiparity (parity of five or more), previous dilatation and curettage (D&C), and small placental weight.7–9 Other risk factors that have been indicated are previous cesarean delivery, previous abortions, induced labor, older maternal age, delivery in a labor bed, midwife delivery, preeclampsia, and oxytocin use.7–9 To our knowledge, however, there has been only one study using multivariable regression analysis in a large patient sample in relation to risk factors for retained placenta in the context of modern contemporary obstetric practice since or before 1991.8 Since that time, both demographic characteristics of parturients (age, weight) and obstetric management (use of oxytocin, regional anesthesia, induction of labor) have undergone a significant change. In light of these changes, we wanted to explore the epidemiology of retained placenta in a Swedish population with a particular focus on parameters relating to dystocia and the use of oxytocin for labor augmentation.
MATERIALS AND METHODS
This case-control study was conducted at Stockholm South General Hospital, a tertiary-level obstetric department, in all cases of retained placenta between February 2007 and March 2010. All women meeting the inclusion criteria of vaginal delivery and total or partial placental retention were identified. Cases of intrauterine fetal death and multiple pregnancies were excluded. Retained placenta was defined as the need of manual removal of the placenta or parts of it under general or regional anesthesia. Cases thus fulfilled the Nordic Surgical Classification Code MBA30 (manual removal of fetal membranes, placenta, or rest thereof). According to the guidelines of the department, the third stage of labor is managed by administration of 5–10 units of oxytocin (intravenously or intramuscularly) immediately after the delivery of the newborn. Thereafter, controlled cord traction is applied and the cord is clamped after 2–3 minutes. If the placenta has not been delivered within 30 minutes, manual removal is initiated; it is initiated earlier in case of heavy bleeding. A control group consisting of vaginal deliveries (intrauterine death, multiple pregnancy, and retained placenta excluded) within the same time period was randomly picked from the electronic delivery record in the following manner: the chronologically first 10 or 11 deliveries every month during the study period were selected electronically, resulting in an equivalent number of control individuals and giving an even distribution of births over time.
In the Stockholm region, all pregnancy and delivery data are registered in computerized prenatal and in-hospital medical records. Background data and information about the delivery were retrieved from these records and registered in a database. All diagnoses listed in patient records are double-checked and, if needed, completed by specially trained diagnostic registering nurses. In the case of labor dystocia (International Classification of Diseases, Tenth Revision, code 062.0 and 062.1), which was of particular interest in this study, an extra assessment of each patient journal and partogram was performed by one of the authors (M.E.) to confirm that a diagnosis was not missed or incorrectly made. A special focus was likewise given on the duration of the phases of labor and oxytocin use. Time and date of first contractions, of the beginning of the active phase of the first stage of labor (defined as regular painful contractions and a cervical dilatation of 3 cm), of the start of the second stage of labor (cervix fully dilated), of delivery, and of placental release were registered as well as when and for how long oxytocin was administered before delivery. Latency phase was defined as the period from the start of contractions until the start of the active phase. Active labor in this study was defined as the period from the start of the active phase until the delivery of the child. A diagnosis of labor dystocia was made if the active phase of the first stage of labor progressed at an average rate of cervical dilatation of less than 1 cm per hour within a margin of plus 2 hours, a lack of progression during the first stage for more than 2 consecutive hours after a previously normal progression, or a second stage of labor of more than 3 hours. In the regression analysis, we also separately used the definition of active labor for 12 hours or more. The study was approved by the Local Ethics Committee at Karolinska Institute, Stockholm, Sweden (2010/438-31/4) and the Institutional Review Board of the Department of Obstetrics and Gynecology, South General Hospital.
Our sample size was calculated to attain a power more than 80% in the test of association between retained placenta and exposure to oxytocin during labor. The hypothesized odds ratio (OR) was based on the results relating to oxytocin of a previous study (OR 1.47, assumed prevalence of oxytocin use of 40%, α=0.05).8 Univariable and multivariable logistic regressions were used for estimation of risk ratios and statistical significance. Variables that were significantly associated with retained placenta in the univariable analysis were added into the multivariable regression model in order of strength (combined assessment of OR and significance level). Variables losing power (a 95% confidence interval [CI] incorporating 1) were excluded. For risk analysis relating to time durations, patients were grouped according to percentile categories. In scale variables without a normal distribution, the nonparametric Mann–Whitney test was used for comparison of medians. Smoking and preeclampsia were not added in the same model because this inflated and distorted results because of the fact that there were no smokers among the patients with preeclampsia. With this exception, all epidemiological and delivery-related variables were subjected to logistic regression in the same model. The regression analysis pertaining to blood loss was performed separately with blood loss as the dependent variable. The regression model and statistical analysis were performed using the PASW Statistics 18.0.0 (SPSS).
During this study period there were 427 cases of retained placenta. In all but three cases the placenta was retained more than 30 minutes. The total number of vaginal deliveries during the study period was 16,209, giving an incidence of retained placenta of 2.6% (427 of 16,209, 95% CI 2.39–2.89%). Four hundred eight cases fulfilled inclusion criteria and were subsequently compared with an equal number of incidents in the control group.
The case and control group women were similar in age (median age 33 years and 33 years respectively; P=.95). The average number of previous pregnancies was the same in the case and control groups (mean 2.3 in both groups). Neither body mass index at the start of pregnancy nor weight gain during pregnancy was independently associated with retained placenta. There was a strong association between retained placenta and previous retained placenta. Previous miscarriages as well as previous abortions were likewise independent risk factors. A previous parity of two or more had an independent protective effect, as did smoking at the start of pregnancy. There was no association with previous cesarean delivery. An overview of results relating to maternal characteristics and prepregnancy history is given in Table 1.
We tested for association with a number of intercurrent diseases and pregnancy-related disorders. Preeclampsia was the only independently significant risk factor for retained placenta in our model. An increased risk associated with hypothyroidism was borderline significant. Diabetic individuals were numerically overrepresented in the case group, but the total sample was small (five in the case group compared with one in the control group). These results are summarized in Table 2.
The only delivery-related risk factors independently associated with retained placenta in the multivariable model were oxytocin use during labor and preterm delivery. Oxytocin augmentation was used in 54.2% (221 of 408) of case group individuals and 36.0% (147 of 408) of control group individuals, with a median duration of 4.5 and 2.5 hours, respectively (P<.001). The association with oxytocin was stronger when oxytocin use was stratified according to duration of administration. There was no increased risk of retained placenta associated with oxytocin administration for 194 minutes or less (50th percentile of duration); however, thereafter, the risk increased noticeably (Fig. 1). Augmented labor for more than 415 minutes was associated with an adjusted OR of 6.55 (95% CI 3.42–12.54). With an assumed expected event rate of retained placenta of 2.6%, this would mean that one extra case of retained placenta over this baseline incidence would occur for every three patients receiving oxytocin for more than 415 minutes (number needed to harm 2.3). Conflating risk factors augmented the associated risk of retained placenta. Although preeclampsia has an associated OR of 3.54 with retained placenta (95% CI 1.59–7.90) and oxytocin use for more than 194 minutes has an associated OR of 2.42 (95% CI 1.60–3.66), a woman with both these risk factors had an unadjusted associated OR of 18.84 (95% CI 2.50–141.84). Inadvertently, the study showed that out of all patients receiving oxytocin, 36% did not qualify for a diagnosis of dystocia before the initiation of medication. Twelve percent of patients with a diagnosis of dystocia received no oxytocin.
Gestational age at delivery was not different between those in the case and control groups, with a median of 281 days (range 105 days, minimum 191, maximum 296) for those in the case group and 280 days (range 80 days, minimum 217, maximum 297) for those in the control group corresponding to 40 weeks plus 1 day and 40 weeks plus 0 days, respectively (P=.14). Both preterm delivery and delivery at more than 41 weeks were, however, overrepresented in the case group, with preterm delivery being independently associated with retained placenta. Twenty-six percent of those in the case group compared with 15.4% of those in the control group had induced labor, 22.3% compared with 12.5% were instrumental deliveries, and 45.8% compared with 37.7% had epidural analgesia. None of these variables was independently associated with retained placenta, however. There was no difference in the average duration of the latency phase between the two groups (median duration 5 hours for both groups, P=.74, range for case group 181 minutes, range for control group 183 minutes). The active phase of the first stage of labor was longer in the case group (median duration 5.5 hours for cases, range 29.0, compared with 4.5 hours for control group, range 34.7; P<.04), as was the duration of the second stage (median duration 60 minutes in case group, range 9.2 hours, 30 minutes in control group, range 6.6 hours; P<.001). All variables reflecting prolonged or inefficient labor at various stages of the delivery process, ie, a diagnosis of dystocia, a total labor duration 12 hours or more, or a prolonged second stage, came out as risk factors in the univariable analysis but were not independently associated with retained placenta after adjusting for other factors in the multivariable regression model. Active labor was also stratified into percentiles of duration in the univariable analysis (less than 50th percentile, referent value, less than 6.17 hours, 50th-75th percentile 6.17–10.5 hours, 75th percentile more than 10.5 hours). The associated risk to retained placenta increased somewhat with increasing duration of labor but was only significant for labor over the 75th percentile of duration (OR 1.15, 95% CI 0.80–1.63 and OR 1.96, 95% CI=1.36–2.81, respectively). An overview of results relating to delivery-related factors is given in Table 3.
The risk for blood loss of 500 mL or more, 1,000 mL or more, and 2,000 mL or more, respectively, and the risk of requiring a blood transfusion were increased in the case of retained placenta. Among those in the control group, median blood loss was 400 mL (range 3,950 mL); among those in the case group median blood loss was 1,600 mL (range 5,800 mL; P<.001). Prolonged labor is a known risk factor for blood loss;10 therefore, all categorizations of blood loss were corrected for labor 12 or more hours in their association with retained placenta. Results pertaining to blood loss are presented in Table 4.
Our study indicates that previous placental retention, prolonged use of oxytocin, preterm delivery, preeclampsia, repeated miscarriages, and at least one previous abortion independently increase the risk of retained placenta, whereas having given birth twice or more before and smoking at the start of pregnancy lower the risk of this condition. A summary of significant results can be found in Table 5.
The incidence of retained placenta of 2.6% is consistent with the average incidence of 2.67% in developed countries, as presented in a previous systematic review.1 We have confirmed and quantified the increased risk of clinically significant blood loss as well as the need for blood transfusion after retained placenta. Our assessment in this regard is compatible with previous research.3,8
The magnitudes of the risk factors found in this study are similar to those in the Combs and Laros study from 1991,8 which resembles ours in terms of patient sample size, demographic profile, and obstetric setting. Our findings are in agreement with theirs in that induction of labor, (supposed) amnionitis, premature rupture of membranes, and previous cesarean delivery are not significant risk factors for retained placenta.
Oxytocin use previously has been identified as a weak risk factor for retained placenta.8 We found an exponential increase in risk for retained placenta associated with increasing duration of oxytocin use above a certain cut-off time. Twenty-five percent of patients receiving oxytocin did so for more than 415 minutes, which was associated with a 6.55-fold increase in the risk of retained placenta. There was, however, no risk associated with oxytocin use less than 194 minutes. Likewise, it has been shown in previous studies that the routine administration of oxytocin after delivery to shorten the third stage is not associated with an increased risk for retained placenta.11 Thus, arguably it is not oxytocin per se that is an agent in the development of retained placenta, but the consequences of its prolonged use. Our data do not include the total dose of oxytocin administered, although it can be assumed that a fairly good correlation exists between the duration of administration and total dose.
Patients with dystocic labor and patients receiving oxytocin are understandably patient populations that overlap to a large extent. It seemed, however, that the risk of retained placenta pertained to oxytocin and was not attributable to the fact that labor in these patients was prolonged. None of the measurements of dystocia or prolonged labor came out as significant risk factors in the multivariable model. The proportion of patients with dystocia diagnosed who did not receive oxytocin was small (31 of 267 patients), but these patients were evenly distributed in the case and control groups and, in fact, were somewhat overrepresented in the control group (14 patients in the case group, 17 in the control group). However, we cannot exclude that the factor of oxytocin in our results is a proxy for prolonged labor and that retained placenta is in fact related to dystocia and myometrial fatigue. In the study by Owolabi et al9, a duration of first stage of labor of more than 10 hours was not associated with an increased risk of retained placenta when confounding factors were adjusted for. This study apart, to our knowledge, dystocia has not been studied in relation to retained placenta.
Several studies within the field of veterinary medicine have shown that retained placenta in cows is related to oxidative stress and a disruption of the apoptotic process in the placenta.12–14 Many studies of the human placenta have found a similar pathological mechanism in relation to conditions of defective placentation,15 such as preeclampsia, recurrent miscarriages, and preterm delivery as well as intrauterine growth restriction and diabetes mellitus.16–24 The comorbidity between these conditions is well-known.25–27 This suggests that there may be a common underlying pathological mechanism related to retained placenta and some of its main risk factors (preeclampsia, preterm delivery, and recurrent miscarriages). Intrauterine growth restriction was not a significant risk factor for retained placenta, possibly because of a too small sample size and the multifactorial nature of the condition. Likewise, the small number of diabetic patients (pregestational diabetes and gestational diabetes) made the strength of the association with retained placenta impossible to evaluate, although it is noticeable that diabetic patients were overrepresented in the case group (five compared with one, OR 5.05, 95% CI 0.59–43.41). If there is a related etiopathology between retained placenta and some of its most prominent risk factors, namely oxidative stress and a deregulated placental apoptosis, it may be that retained placenta forms part of the spectrum of pregnancy-related disorders of defective placentation. A recent study has demonstrated that oxytocin use during pregnancy decreases gluthatione, an agent in antioxidative defense.28 Prolonged augmentation of labor may lead to oxidative stress, which in turn depletes the antioxidative defenses in the placenta. Hypothetically, this could render the placenta more vulnerable to placental retention.
Smoking had a protective effect in this study on the incidence of retained placenta. Smoking has been shown to lower the incidence of preeclampsia by 32%,29 which may indirectly explain this result. Carbon monoxide has also been shown to reduce the apoptotic process in the placenta.30 Parity of two or more had a protective effect according to our results. Previous studies have shown that this effect reverses over a parity of four. In our patient material, the frequency of multiparity was only 0.3%; hence, this subgroup was too small to take part in the analysis.
Previous abortion was, in our study, associated with an increased risk for retained placenta (OR 1.58, 95% CI 1.09–2.28). Whether D&C was performed, as opposed to medical treatment, is often not specified in patient records, and thus this subgroup might be underestimated. In the univariable analysis, previous D&C is a stronger risk factor than abortion for retained placenta and it is probable that the risk associated with previous abortions is not attributable to the termination of pregnancy but is attributable to the associated D&C. One might expect an increased risk of pathological implantation over a uterine scar leading to retained placenta in the form of partial or total placenta accreta in women with previous D&C or a previous cesarean delivery. In the largest study investigating this, the correlation with previous abortion was, as in our study, relatively weak (OR 1.35; P<.05), and there was no correlation with previous cesarean delivery.8 The correlations with abortion as well as previous cesarean delivery are stronger in a Nigerian study from 2008.9 It is possible that the discrepancy between our results lies in the fact that the incidence of postoperative infections after both previous abortions and cesarean delivery were higher in the Nigerian study population.
Awareness of risk factors for retained placenta can aid in identifying women at heightened risk for this condition and subsequent risk of hemorrhage.
1. Cheung WM, Hawkes A, Ibish S, Weeks AD. The retained placenta: historical and geographical rate variations. J Obstet Gynaecol 2011;31:37–42.
2. Herman A, Weinraub Z, Bukovsky I, Arieli S, Zabow P, Caspi E, et al.. Dynamic ultrasonographic imaging of the third stage of labor: new perspectives into third-stage mechanisms. Am J Obstet Gynecol 1993;168:1496–9.
3. Bateman BT, Berman MF, Riley LE, Leffert LR. The epidemiology of postpartum hemorrhage in a large, nationwide sample of deliveries. Anesth Analg 2010;110:1368–73.
4. Ely JW, Rijhsinghani A, Bowdler NC, Dawson JD. The association between manual removal of the placenta and postpartum endometritis following vaginal delivery. Obstet Gynecol 1995;86:1002–6.
5. Anorlu RI, Maholwana B, Hofmeyr GJ. Methods of delivering the placenta at caesarean section. Cochrane Database Syst Rev 2008;3:CD004737.
6. Morales M, Ceysens G, Jastrow N, Viardot C, Faron G, Vial Y, et al.. Spontaneous delivery or manual removal of the placenta during caesarean section: a randomised controlled trial. BJOG 2004;111:908–12.
7. Adelusi B, Soltan MH, Chowdhury N, Kangave D. Risk of retained placenta: multivariate approach. Acta Obstet Gynecol Scand 1997;76:414–8.
8. Combs CA, Laros RK Jr. Prolonged third stage of labor: morbidity and risk factors. Obstet Gynecol 1991;77:863–7.
9. Owolabi AT, Dare FO, Fasubaa OB, Ogunlola IO, Kuti O, Bisiriyu LA. Risk factors for retained placenta in southwestern Nigeria. Singapore Med J 2008;49:532–7.
10. Allen VM, Baskett TF, O'Connell CM, McKeen D, Allen AC. Maternal and perinatal outcomes with increasing duration of the second stage of labor. Obstet Gynecol 2009;113:1248–58.
11. Jackson KW Jr, Allbert JR, Schemmer GK, Elliot M, Humphrey A, Taylor J. A randomized controlled trial comparing oxytocin administration before and after placental delivery in the prevention of postpartum hemorrhage. Am J Obstet Gynecol 2001; 185:873–7.
12. Kankofer M. Antioxidative defence mechanisms against reactive oxygen species in bovine retained and not-retained placenta: activity of glutathione peroxidase, glutathione transferase, catalase and superoxide dismutase. Placenta 2001;22:466–72.
13. Celi P. Biomarkers of oxidative stress in ruminant medicine. Immunopharmacol Immunotoxicol 2011;33:233–40.
14. Kamemori Y, Wakamiya K, Nishimura R, Hosaka Y, Ohtani S, Okuda K. Expressions of apoptosis-regulating factors in bovine retained placenta. Placenta. 2011;32:20–6.
15. Brosens I, Pijnenborg R, Vercruysse L, Romero R. The “Great Obstetrical Syndromes” are associated with disorders of deep placentation. Am J Obstet Gynecol 2011;204:193–201.
16. Buhimschi IA, Buhimschi CS, Pupkin M, Weiner CP. Beneficial impact of term labor: Nonenzymatic antioxidant reserve in the human fetus. Am J Obstet Gynecol 2003;189:181–8.
17. Heazell AE, Crocker IP. Live and let die—regulation of villous trophoblast apoptosis in normal and abnormal pregnancies. Placenta 2008;29:772–83.
18. Jauniaux E, Poston L, Burton GJ. Placental-related diseases of pregnancy: Involvement of oxidative stress and implications in human evolution. Hum Reprod Update 2006;12:747–55.
19. Hung TH, Skepper JN, Charnock-Jones DS, Burton GJ. Hypoxia-reoxygenation: a potent inducer of apoptotic changes in the human placenta and possible etiological factor in preeclampsia. Circ Res 2002;90:1274–81.
20. Hempstock J, Jauniaux E, Greenwold N, Burton GJ. The contribution of placental oxidative stress to early pregnancy failure. Hum Pathol 2003;34:1265–75.
21. Allaire AD, Ballenger KA, Wells SR, McMahon MJ, Lessey BA. Placental apoptosis in preeclampsia. Obstet Gynecol 2000;96:271–6.
22. Madazli R, Benian A, Ilvan S, Calay Z. Placental apoptosis and adhesion molecules expression in the placenta and the maternal placental bed of pregnancies complicated by fetal growth restriction with and without pre-eclampsia. J Obstet Gynaecol 2006;26:5–10.
23. Roje D, Zekic Tomas SZ, Kuzmic Prusac I, Capkun V, Tadin I. Trophoblast apoptosis in human term placentas from pregnancies complicated with idiopathic intrauterine growth retardation. J Matern Fetal Neonatal Med 2011;24:745–51.
24. Sobrevia L, Abarzúa F, Nien JK, Salomón C, Westermeier F, Puebla C, et al.. Differential placental macrovascular and microvascular endothelial dysfunction in gestational diabetes. Placenta 2011;32(Suppl 2):59–64.
25. Trogstad L, Magnus P, Moffett A, Stoltenberg C. The effect of recurrent miscarriage and infertility on the risk of pre-eclampsia. BJOG 2009;116:108–13.
26. Watson LF, Rayner JA, King J, Jolley D, Forster D, Lumley J. Modelling prior reproductive history to improve prediction of risk for very preterm birth. Paediatr Perinat Epidemiol 2010;24:402–15.
27. Xiong X, Fraser WD, Demianczuk NN. History of abortion, preterm, term birth, and risk of preeclampsia: a population-based study. Am J Obstet Gynecol 2002;187:1013–8.
28. Schneid-Kofman N, Silberstein T, Saphier O, Shai I, Tavor D, Burg A. Labor augmentation with oxytocin decreases glutathione level. Obstet Gynecol Int 2009;2009:807659.
29. Conde-Agudelo A, Althabe F, Belizán JM, Kafury-Goeta AC. Cigarette smoking during pregnancy and risk of preeclampsia: a systematic review. Am J Obstet Gynecol 1999;181:1026–35.
30. Bainbridge SA, Belkacemi L, Dickinson M, Graham CH, Smith GN. Carbon monoxide inhibits hypoxia/reoxygenation-induced apoptosis and secondary necrosis in syncytiotrophoblast. J Pathol 2006;169:774–83.