Placenta previa, defined as placenta that implants at or over the cervical os,1 is present in approximately 0.3–0.5% of all pregnancies at delivery.2,3 The possibility of considerable feto-maternal hemorrhage, and its sequelae, contributes significantly to both maternal and perinatal morbidity.4 Twin pregnancies may be at higher risk for placenta previa than singleton gestations, either because of having two implantation sites, as in dichorionic pregnancies, or simply because of the increased placental mass, as in monochorionic pregnancies. Current data on the incidence of placenta previa in twin pregnancies relative to singleton pregnancies, based on natality data or relatively small populations, are conflicted.5–9
Additionally, information on the natural history of placenta previa in twin pregnancies is limited. Previous studies in singleton pregnancies have demonstrated that 83% of all previa diagnosed in the second trimester resolve by delivery.10 For twin pregnancies, this rate is unknown, as are estimates of the gestational age at which resolution occurs. Knowledge of the characteristics for placenta previa resolution in twin pregnancies may be helpful in managing twin pregnancies and in counseling patients. Therefore, we aimed to estimate the incidence of placenta previa in twin pregnancies compared with singleton pregnancies. Additionally, we sought to estimate the rate and gestational age of placenta previa resolution in twin pregnancies.
MATERIALS AND METHODS
This is a retrospective cohort study of all consecutive patients who underwent routine second-trimester (15–22 weeks) ultrasonography for anatomic survey at a single tertiary care center. Institutional Review Board approval was obtained from Washington University School of Medicine. Data were collected prospectively by dedicated nurses from 1990 to 2008. Each patient undergoing ultrasonography in our center receives a standardized form requesting information regarding pregnancy complications, delivery complications, and neonatal outcomes, and it is to be completed and returned after delivery. If the form was not returned within 4 weeks of the expected delivery date, then the coordinator called the patient. In cases in which the patient could not be reached, the coordinator contacted the referring physician to obtain outcome data.
Patients were included in this study if they had a confirmed singleton or twin gestation. They were excluded if they had fetal demise at the time of presentation for anatomic survey, if they had a higher-order multiple gestation, or if delivery data were unavailable. Fetal anomalies were not excluded. Gestational age was determined either by last menstrual period if known and concordant with ultrasonography (within 7 days of first-trimester ultrasound examination or 14 days of second-trimester ultrasound examination) or by the earliest ultrasound examination available when the last menstrual period was unknown or discordant with ultrasonography. Chorionicity was determined by the earliest ultrasonography available. First-trimester diagnosis of chorionicity was based on the number of gestational sacs, amnions, and yolk sacs present, or on the presence of a lambda sign.11 Second-trimester determination of chorionicity was based on gender discordance, presence of two placental masses, and characteristics of the intertwin dividing membrane (twin peak sign, T-sign, thickness of membrane).12 Final assignment of chorionicity was performed by an attending physician dedicated to obstetric ultrasonography.
At our institution, second-trimester ultrasound examinations routinely involve assessment of placenta location.13,14 Suspected placenta previa by transabdominal scanning is confirmed with transvaginal ultrasonography. All subsequent scans for placental location are transvaginal. Placenta previa diagnosed by transvaginal ultrasonography was coded as complete (placenta covering the entire cervical os), partial (placenta covering part of the internal cervical os), or marginal (placenta within 2 cm of the internal cervical os). Low-lying placenta often is used to describe a placenta extending into the lower uterine segment when seen on transabdominal ultrasonography or for previa identified before the third trimester; we do not use this description at our institution because of its variable definitions and implications.4 Previa resolution was defined as the absence of a previously diagnosed previa that was not attributable to delivery. If delivery occurred during a defined period, the previa was only counted as resolved if documented as such in the delivery record.
The incidence of previa in twin and singleton gestations at the time of second-trimester ultrasound examination was compared. Because complete and partial previa are less likely to resolve, we examined the incidence of complete and partial previa as a secondary analysis.10 We also analyzed the effect of chorionicity on the incidence of placenta previa.
To estimate the rate and timing of previa resolution, we performed a subanalysis of all twin gestations that underwent serial ultrasound examinations between 15 and 40 weeks of gestation. We examined the number of all twin pregnancies with previa resolving at 24–28, 28–32, 32–36, and 36 or more weeks of gestation as documented by subsequent ultrasonography or the delivery record. These gestational age ranges were selected because twins are scanned at least every 4 weeks, typically beginning by 28 weeks of gestation at our institution. Consequently, most twins will be scanned at least once during the predefined time periods. When a woman did not undergo an ultrasound examination in the predesignated time period, a previa was assumed to have continued during that time period. For example, if a woman had placenta previa diagnosed at 18 weeks and did not undergo a repeat ultrasound examination documented until 30 weeks, then the previa was considered resolved between 28 and 32 weeks.
Singleton and twin gestations were compared with univariable analyses using unpaired Student t or Mann-Whitney U tests for continuous variables, and χ2 or Fisher exact tests for categorical variables, as appropriate. Potentially confounding variables of the exposure–outcome association were identified in stratified analyses. Multivariable logistic regression models were then developed to better-estimate the effects of twin pregnancy on previa while adjusting for potentially confounding effects. Clinically relevant covariates (age, gravidity, parity, race, prepregnancy body mass index, current smoking status, history of termination, previous cesarean delivery) for initial inclusion in multivariable statistical models were selected using results of the stratified analyses, and factors were removed in a backward stepwise fashion, based on significant changes (10% or more) in the exposure adjusted odds ratio (OR). All statistical analyses were performed using STATA 11.1.
Of the 86,241 second-trimester ultrasound examinations performed for 76,495 women, 67,895 pregnancies were included in the analysis after exclusion of pregnancies with missing data (8,425) and higher-order multiple gestations (151 triplets, 24 quadruplets). Of the pregnancies included in the primary analysis, 2,194 were twin gestations (3.2%, 95% confidence interval [CI] 3.1–3.4). The twin group had a higher incidence of gestational diabetes and preeclampsia, and was less likely to have had a previous cesarean delivery (Table 1). The two groups were similar with respect to age, body mass index, gravidity, parity, race, tobacco use, and hypertension.
Of the 2,194 women with twin gestations who underwent second-trimester ultrasound examinations, 56 (2.5%, 95% CI 1.9–3.3) had placenta previa of any type, compared with 1,381 (2.1%, 95% CI 2.0–2.2) in patients with a singleton pregnancy (relative risk [RR] 1.21, 95% CI 0.93–1.58) (Fig. 1). After adjusting for gravidity, race, and history of cesarean delivery, the risk of any type of previa remained similar (adjusted OR 1.25, 95% CI 0.95–1.66). When including only complete or partial previa, there remained no statistically significant difference in the rate of previa between the singleton and twin groups (adjusted OR 1.34, 95% CI 0.90–2.01). However, when only dichorionic twins were considered, the incidence of placenta previa was significantly greater than in singletons (adjusted OR 1.66, 95% CI 1.09–2.53). The difference between monochorionic twins and singletons was not statistically significant (RR 0.20, 95% CI 0.03–1.43). The findings were similar when only marginal and complete previa were considered (Fig. 2).
The incidence of previa was further analyzed by chorionicity among the twin pregnancies (Table 2). These findings were statistically significant, displaying an increased incidence of both complete and partial previa (RR 7.90, 95% CI 1.07–58.24) and all types of previa (RR 3.29, 95% CI 1.32–8.21) for dichorionic pregnancies when compared with monochorionic twin pregnancies.
Finally, in the subanalysis of the 1,738 twin pregnancies undergoing serial ultrasound examinations at between 15 and 40 weeks of gestation, the portion of placenta previa resolving at predetermined gestational age ranges (24–28, 28–32, 32–36, and 36 weeks or more) was evaluated (Fig. 3). Women with previa who underwent delivery for obstetric indications (other than maternal bleeding) are listed as scheduled deliveries. The number of ultrasound examinations undergone by each patient (starting in the second trimester) ranged from two to nine; the number of twin pregnancies that underwent ultrasound examination at each successive interval decreased because of delivery. The previa were not subdivided by type because the groups were relatively small. Of 1,738 twin pregnancies undergoing serial ultrasound examinations, 51 (2.93%, 95% CI 2.19–3.84) were noted to have a previa present on second-trimester ultrasonography. Of the placenta previa diagnosed on second-trimester ultrasonography, 43% resolved by 28 weeks, 69% resolved by 32 weeks, and 82% resolved by 36 weeks. Forty-seven percent (7 of 15) of the previa cases that were still present at 32 weeks resolved before 36 weeks. Of the nine previa cases that persisted until delivery, five were complete, two were partial, and two were marginal (seven occurred in dichorionic pregnancies, two occurred in monochorionic pregnancies). The six placenta previa cases that were noted beyond 36 weeks did not resolve (four of six were complete, one of six was partial, one of six was marginal or vasa previa).
Risk factors for persistent placenta previa were investigated. Among the 42 patients with resolved previa, 12 (28.6%) had reported a previous cesarean delivery. Only one (11%) of the nine previa cases persisting until delivery occurred in a patient with a previous cesarean delivery. None of the five patients who underwent delivery because of vaginal bleeding had a previous cesarean delivery. No placenta accreta cases were observed. Marginal previa cases were most likely to resolve and complete previa cases were least likely to resolve; no other specific predictors of previa resolution were noted (data available on request).
Although twin pregnancies do not appear to be at increased risk for placenta previa as diagnosed on second-trimester ultrasonography when compared with singletons, dichorionic twin pregnancies have a greater incidence of placenta previa than singletons or monochorionic twins. We hypothesize that the increased incidence of placenta previa in dichorionic twins is attributable to the presence of two implantation events, as opposed to simply an increased placental mass. This hypothesis is further supported by the low previa rate seen in our monochorionic twins, which is comparable with that of singletons, because the likely risk factor for previa is the number of implantation sites. Similar to singleton gestations, a significant proportion of placenta previa diagnosed in twins on the second-trimester ultrasound scan (82%) will resolve before delivery.10
Current data available for the incidence of placenta previa in twin pregnancies are conflicted.4 Two case control studies by Mizrahi et al and Spellacy et al report either lower or similar incidences of placenta previa between singletons and twins.5,6 In contrast, two large retrospective studies on the topic both found an increased incidence of placenta previa in twins compared with singletons. Ananth et al7 used the United States natality data file to characterize the incidence of previa in twin compared with singleton pregnancies. In their study of 3,793 twin and 104,754 singleton births, the incidence of placenta previa in twins was 0.39%, a 40% increase compared with the singleton population.7 Because this study was based on natality data, it only considers placenta previa present at the time of cesarean delivery. Although this method is specific for placenta previa, the majority of placenta previa cases now are diagnosed at second-trimester ultrasound examination. Strong et al8 published a cohort study comparing 148,197 singleton and 1,464 twin pregnancies. Their study defined placenta previa by diagnoses listed in the delivery statistics of the hospital. They found a previa incidence of 0.55% for twins, which is significantly higher than the 0.31% they reported for singleton pregnancies.8 Neither of these studies was able to consider chorionicity in their study, which we have demonstrated has an effect on the incidence of placenta previa. Perhaps their observed difference in previa incidence between twins and singletons is attributable to the fact that the majority of twins are dichorionic.15 Additionally, these previous studies are unable to provide information regarding the natural course of placenta previa in twin pregnancies. A fifth study, a retrospective cohort by Francois et al,9 reported no difference in the incidence of placenta previa among singleton and multiple gestations. However, they did find a strong association between a history of cesarean delivery and placenta previa development in multiple gestations.9
Our study offers several strengths. The relatively large sample size, including 1,738 twin pregnancies with serial ultrasound examinations, enabled us to estimate the incidence of a rare entity (placenta previa). Additionally, we were able to stratify our previa by chorionicity of the affected pregnancy, showing an association between dichorionicity and placenta previa. Further, our use of ultrasound-diagnosed placenta previa with access to serial ultrasound examinations, as opposed to previa at delivery, makes our study clinically relevant to practitioners. The majority of patients now undergo routine second-trimester ultrasound examinations.14 The finding of placenta previa on ultrasonography can be distressing to patients and requires delivery planning and counseling on the part of practitioners. The majority of patients underwent ultrasound assessment every 3–4 weeks. Only four patients in our cohort had a longer time period between ultrasound examinations. Consequently, we have the unique ability to assess the timing of resolution of placenta previa after second-trimester diagnosis, making our study relevant to the practicing physician.
Our study is not without limitations. One potential limitation to consider is the possibility of selection bias because complete outcome data were only available for 89% of the pregnancies in our cohort. The portion of the cohort excluded for having incomplete data was, on average, more likely to be younger, more likely to have higher parity, smoke, and more likely to be of nonwhite race (Table 1). However, the incidences of twins (3.2% compared with 3.0%) and previa (2.1% compared with 2.2%) were nearly identical between the outcome and no outcome available groups, suggesting that any selection bias would have a minimal effect on our findings. The twin pregnancies with serial ultrasound examinations represented 79% of our cohort. Of the twin pregnancies not followed-up by serial ultrasound examination at our institution, they were more likely to smoke, to be of white race, and to have a higher body mass index, and were less likely to have placenta previa (0.9%, data not shown) or preeclampsia when compared with those with serial ultrasound examinations.
Finally, although our dataset was large, the relative rarity of placenta previa precluded an analysis specific to previa type. However, we were able to demonstrate that the majority of placenta previa cases that did not resolve were complete previa, suggesting that, similar to singletons, complete previa cases diagnosed in the second-trimester are less likely to resolve.10 Additionally, despite our large sample size, only six previa cases remained at 36 weeks, and we are unable to assess the natural history of persistent previa after this. However, this is unlikely to be important clinically because the mean gestational age of delivery for twins is 35 weeks.4
In conclusion, our data demonstrate dichorionic pregnancies to be at significantly increased risk for ultrasound-diagnosed previa when compared with singleton or monochorionic pregnancies. These findings emphasize the importance of not only following recommendations to assess placental location but also locating both placental locations in dichorionic pregnancies.13 Patients with twin pregnancies can be reassured that 82% of the previa in our study resolved before delivery, with a substantial portion resolving by 28 weeks and the majority resolving by 32 weeks. This pattern of resolution is similar to that observed in singleton gestations.4,10 When patients with twins have placenta previa diagnosed at their second-trimester scans, they may be reassured that the majority of cases will resolve. Because of the high rate of resolution throughout pregnancy, it is advisable to assess the locations of both placentas at every ultrasound examination and before delivery because both the timing and the route of delivery may be affected by previa resolution.
1. Oyelese Y, Smulian JC. Placenta previa, placenta accreta, and vasa previa. Obstet Gynecol 2006;107:927–41.
2. Iyasu S, Saftlas AK, Rowley DL, Koonin LM, Lawson HW, Atrash HK. The epidemiology of placenta previa in the United States, 1979 through 1987. Am J Obstet Gynecol 1993;168:1424–9.
3. Martin JA, Hamilton BE, Ventura SJ, Menacker F, Park MM, Sutton PD. Births: final data for 2001. Natl Vital Stat Rep 2002;51:1–102.
4. Hull AD, Resnik R. Placenta previa, placenta accreta, abruptio placentae, and vasa previa. In: Creasy RK, Resnik R, Iams JD, Lockwood CJ, Moore TR, editors. Creasy and Resnik's maternal-fetal medicine: Principles and practice. 6th ed. Philadelphia (PA): Saunders/Elsevier; 2009. p. 453–37, 725–37.
5. Mizrahi M, Furman B, Shoham-Vardi I, Vardi H, Maymon E, Mazor M. Perinatal outcome and peripartum complications in preterm singleton and twins deliveries: a comparative study. Eur J Obstet Gynecol Reprod Biol 1999;87:55–61.
6. Spellacy WN, Handler A, Ferre CD. A case-control study of 1253 twin pregnancies from a 1982–1987 perinatal data base. Obstet Gynecol 1990;75:168–71.
7. Ananth CV, Demissie K, Smulian JC, Vintzileos AM. Placenta previa in singleton and twin births in the United States, 1989 through 1998: a comparison of risk factor profiles and associated conditions. Am J Obstet Gynecol 2003;188:275–81.
8. Strong TH Jr, Brar HS. Placenta previa in twin gestations. J Reprod Med 1989;34:415–6.
9. Francois K, Johnson JM, Harris C. Is placenta previa more common in multiple gestations? Am J Obstet Gynecol 2003;188:1226–7.
10. Harper LM, Odibo AO, Macones GA, Crane JP, Cahill AG. Effect of placenta previa on fetal growth. Am J Obstet Gynecol 2010;203:330.e1–5.
11. Sepulveda W, Sebire NJ, Hughes K, Odibo A, Nicolaides KH. The lambda sign at 10–14 weeks of gestation as a predictor of chorionicity in twin pregnancies. Ultrasound Obstet Gynecol 1996;7:421–3.
12. Callen PW. Ultrasound evaluation of multiple pregnancies. In: Ultrasonography in obstetrics and gynecology. 5th ed. Philadelphia (PA): Saunders/Elsevier; 2008. p. 272–6.
13. American Institute of Ultrasound in Medicine. AIUM Practice Guideline for the performance of an antepartum obstetric ultrasound examination. J Ultrasound Med 2003;22:1116–25.
14. Ultrasonography in pregnancy. Practice Bulletin No. 101. American College of Obstetricians and Gynecologists. Obstet Gynecol 2009;113(Pt 1):451–61.
© 2012 The American College of Obstetricians and Gynecologists
15. Cameron AH, Edwards JH, Derom R, Thiery M, Boelaert R. The value of twin surveys in the study of malformations. Eur J Obstet Gynecol Reprod Biol 1983;14:347–56.