The use of ultrasound to localize the placenta and identify placenta previa was first described by Gottesfeld et al in 1966.1 Placenta previa complicates only one delivery in 200, but is detected during approximately one in 20 ultrasound examinations performed before 20 weeks' gestation.2,3 Previa detected before midpregnancy persists until delivery in only one in ten cases.4 Resolution or “migration” of early placenta previa is believed to reflect a combination of preferential growth toward the well-vascularized fundus and degeneration of peripheral villi in the lower uterine segment that receives less blood flow.3,5
Though the majority of women with early placenta previa experience resolution by the third trimester, those who have third-trimester previa are at increased risk for persistence until delivery. Presently, data are limited regarding the magnitude of this risk.6 Counseling for such women is further complicated by the fact that many with previa at delivery have other risk factors, including higher parity and prior cesarean delivery.2,7 It is not clear if parity and prior cesarean delivery increase the likelihood of placental implantation over the internal os, or whether they affect the ability of the placenta to migrate away from the os after implantation.
Our objective was to evaluate gestational age at ultrasound detection of placenta previa as a predictor of whether previa would persist until delivery. In addition, we sought to estimate how the type of placenta previa, parity, and prior cesarean delivery would affect the likelihood of previa persistence.
MATERIALS AND METHODS
This was a retrospective cohort study of pregnancies between 15 and 36 weeks' gestation in which there was antepartum ultrasound evidence of placenta previa. Pregnancies were ascertained using a computerized ultrasound database, based on diagnosis of placenta previa on either transabdominal or endovaginal ultrasound performed February 1, 1991, through September 30, 2000. Previa was termed complete if the placenta completely covered the internal os. Previa was termed incomplete if the inferior edge of the placenta partially covered or reached the margin of the internal os, thus combining traditional definitions for partial and marginal placenta previa. Incomplete previa was selected because of the inherent difficulty in reliably distinguishing between partial and marginal placenta previa at ultrasound examination, particularly in the absence of cervical dilatation, as described by other investigators.8,9 Examples of complete and incomplete placenta previa are shown in Figures 1A and 1B, respectively. Our clinical practice is to perform sonography with the maternal bladder empty. Low-lying placentas were not included.
The outcome of the study was persistent placenta previa resulting in cesarean delivery. This diagnosis was based on clinical assessment and ultrasound at time of delivery. Cesarean delivery was chosen because it is the most common outcome associated with placenta previa.4 For study purposes, women with incomplete previa who were delivered vaginally were not considered to have the study outcome because of difficulties in verifying the relationship between placenta and internal os in such cases. Only liveborn singletons 25 weeks' gestation or more were included. Information about parity and prior cesarean delivery was obtained using a computerized obstetric database, which contains selected obstetric and neonatal outcomes for all women delivering infants at Parkland Hospital.
Gestational age at time of ultrasound examination was grouped into 4-week intervals to facilitate comparisons. Persistence was defined as the number of women with previa during each ultrasound interval who later had previa at delivery. Factors considered in evaluating previa persistence within each interval included the type of previa (complete versus incomplete), parity, and whether there had been at least one prior cesarean delivery. No more than one ultrasound examination was included within a particular gestational age interval, and all comparisons were made within these intervals, so that no pregnancy was counted more than once. Over the study period, there were two instances (0.2%) in which two examinations were performed on the same pregnancy during a 4-week interval with differing diagnoses (complete previa and incomplete previa). The ultrasound was categorized as complete previa in these two cases. The study outcome was previa at delivery; whether previa would be present at subsequent ultrasound examinations was not considered.
Statistical analyses were performed using χ2 for categoric outcomes, Mantel-Haenszel χ2 for trend, and goodness-of-fit χ2 to compare women with previa with our overall obstetric population. Logistic regression was used to adjust for the effects of previa type, parity, and prior cesarean delivery on previa persistence within each gestational age interval. In the logistic regression, parity and prior cesarean delivery were modeled using three independent groups: nulliparous women, parous women with no prior cesarean delivery, and parous women with one or more prior cesarean deliveries. P values < .05 were judged statistically significant. Analyses were performed using SAS system 8.0 (SAS Institute, Cary, NC).
A total of 714 women were found to have placenta previa on antepartum ultrasound examination. Of these women, 141 (20%) were nulliparous and 573 (80%) were multiparous. In addition, 132 (23%) of the multiparous women had at least one prior cesarean delivery. A total of 215 women (30%) underwent cesarean delivery for previa that persisted until delivery. Previa that had been detected on ultrasound examination persisted in 28 (20%) of nulliparous women, 187 (33%) of multiparous women, and 65 (49%) of those multiparous women with a prior cesarean delivery.
To estimate the influence of parity and prior cesarean delivery on the prevalence of placenta previa at time of antenatal ultrasound, we used the 15,725 deliveries at our hospital in the year 2000 as a comparison group. As compared with our overall obstetric population, women with previa at antenatal ultrasound were more likely to be multiparous, 80% versus 62%, respectively, P < .001. These women were also slightly more likely to have had prior cesarean delivery than our overall population, 23% versus 19%, respectively, P = .03.
During the study period, previa was diagnosed during 940 ultrasound examinations in the 714 women. Shown in Table 1 is the number of women with placenta previa at ultrasound examination. Data are grouped into 4-week intervals and presented according to type of previa, parity, and prior cesarean delivery. Persistence was evaluated within these subgroups.
Shown in Figure 2 is the percentage of women in whom previa persisted until delivery, according to gestational age at ultrasound detection of previa. Of women diagnosed with previa at 15 through 19 weeks, only 12% had previa at delivery, similar to what others have reported.4,10 However, previa diagnosed at 20–23 weeks was present at delivery in more than one-third, and if detected at 24–27 weeks or 32–35 weeks, persistence at delivery increased to nearly 50% and 75% of cases, respectively.
Also shown in Figure 2 is the percentage of women with persistent previa according to whether previa was complete or incomplete. Regardless of gestational age at detection, complete previas were more than twice as likely to persist until delivery than incomplete previas, all P < .001.
The percentage of women with persistent previa was next evaluated according to parity. Three groups were compared: nulliparous women, women with parity of one, and women with parity of two or more. At the 15–19-week and 28–31-week intervals, women with parity of two or more were significantly more likely than nulliparous women to have persistent previa, 16% versus 5% at 15–19 weeks and 68% versus 43% at 28–31 weeks, respectively, both P < .05. There were no other significant differences between the three groups at any gestational interval.
Persistence of previa according to prior cesarean delivery is shown in Figure 3, also stratified by gestational age at ultrasound detection. Though women with prior cesarean delivery accounted for only a fraction (23%) of those with placenta previa at ultrasound, they were at increased risk for persistence if previa was detected during the second trimester, before 28 weeks' gestation.
Because we had found that complete previa and prior cesarean delivery were risk factors for persistence, we evaluated these factors in combination (Table 2). Four groups were studied: incomplete previa with no prior cesarean delivery, incomplete previa with prior cesarean delivery, complete previa with no prior cesarean delivery, and complete previa with prior cesarean delivery. At each gestational age (ultrasound) interval, there was a significant trend across these groups, such that women with both complete previa and prior cesarean delivery were at highest risk for previa persistence, P for trend < .001.
Finally, logistic regression was used to adjust for the type of previa, parity, and prior cesarean delivery on the likelihood of previa persistence within the gestational age intervals specified. Crude and adjusted odds ratios are presented in Table 3. As shown in Table 3, complete previa was significantly more likely to persist than incomplete previa at each gestational age interval. Prior cesarean delivery was also an independent risk factor for previa persistence among those women diagnosed with previa during the second trimester (before 28 weeks). Interestingly, parity was no longer a risk factor for previa persistence at any gestational age interval after adjusting for prior cesarean delivery and type of previa, all P > .05.
There are three primary findings from this analysis, the largest to date on the topic of previa persistence. The first is that when placenta previa is detected on ultrasound examination after midpregnancy, there is significant risk that the delivery will be complicated by persistent previa. The magnitude of this risk was nearly 50% after 24 weeks' gestation, approaching 75% after 32 weeks. Further, previa was at least twice as likely to persist if complete than if incomplete. Based on our data, it would be appropriate to counsel women differently according to gestational age and type of previa.
The second finding is that parity is a risk factor for previa prevalence during antepartum ultrasound, but parity does not appear to be a risk factor for previa persistence at time of delivery. Thus, once a woman is found to have placenta previa sonographically, counseling regarding her individual risk for persistence need not be altered based on parity. The finding that parous women are more likely to have previa at ultrasound supports a commonly proposed mechanism for previa in parous women, that each pregnancy depletes normal decidua, favoring implantation on the “scanty” decidua of the lower uterine segment.8
The last finding is that women with prior cesarean delivery were more likely to have persistent previa, particularly if previa was diagnosed during the second trimester. It has been suggested that damage to the uterine lining during cesarean delivery predisposes to low implantation of the placenta in the uterus.11 Results of our series support a different explanation, that scarring impairs the ability of placentas that implant in the lower uterine segment to “migrate” with advancing gestation.8
Only limited data are available regarding persistence of placenta previa detected after midpregnancy. In a classic study of 503 women with previa, Comeau et al reported that even when present into the third trimester, asymptomatic previa had a 75% chance of resolution by the time of delivery.6 Such findings initially appear very different from our own. It is somewhat difficult to compare this study with ours, however, as it included women with low-lying placentas and excluded women with symptomatic previa.
More recent series have suggested that if ultrasound is performed just a few weeks later in gestation, prediction of previa at delivery is improved.12,13 Taipale et al compared screening for placenta previa at two intervals, 12–16 weeks' and 18–23 weeks' gestation.12 Screening was more effective at 18–23 weeks because the prevalence of previa had already decreased from 4% to 1% by the later interval. Lauria et al found that if the placenta overlapped the os by 10 mm before 24 weeks' gestation, there was a 38% risk of persistence until delivery, but the risk of persistence increased to 57% with any degree of previa after 24 weeks' gestation, findings in agreement with ours.13 It seems clear that the positive predictive value for previa at delivery increases the later in gestation that previa is detected sonographically.
Limitations of this retrospective review should be mentioned. We do not have information about whether women were symptomatic. It is not known whether women with previa who present with bleeding are more likely to have their delivery complicated by persistent previa, and future studies might address this interesting question. Zelop et al reported that 22 of 43 of women with persistent placenta previa were asymptomatic, confirming the need for sonographic follow-up in all women diagnosed with previa, regardless of bleeding.4
Another limitation is the inherent inaccuracy of sonography to diagnose placenta previa, with false-positive rates of 2–6% for transabdominal ultrasound and 1–2% for endovaginal ultrasound examinations.14 However, if anything, the false-positive rate would be expected to dilute our findings of persistence rather than to increase them. Because our study design did not take into consideration women in whom previa was not detected sonographically, we cannot comment on the effect of false-negatives, though if any previas were missed during early ultrasound examinations, it is likely that the women would come to attention later in gestation.
Our study describes the increased risk for placenta previa persistence with increasing gestational age at ultrasound detection, as well as the effects of parity and prior cesarean delivery on this relationship. Such information may be useful for counseling patients and assisting with future management decisions.
1. Gottesfeld KR, Thompson HE, Holmes JH, Taylor ES. Ultrasonic placentography. Am J Obstet Gynecol 1966;96:538–47.
2. Cunningham FG, Gant NF, Leveno KJ, Gilstrap LC III, Hauth JC, Wenstrom KD, eds. Obstetrical hemorrhage. In: Williams Obstetrics. 21st ed. New York: McGraw Hill; 2001:619–69.
3. Taipale P, Hilesmaa V, Ylostalo P. Diagnosis of placenta previa by transvaginal sonographic screening at 12–16 weeks in a nonselected population. Obstet Gynecol 1997; 89:364–7.
4. Zelop CC, Bromley B, Frigoletto FD, Benacerraf BR. Second trimester sonographically diagnosed placenta previa: Prediction of placenta previa at birth. Int J Gynecol Obstet 1994;44:207–10.
5. King DL. Placental migration demonstrated by ultrasonography. Radiology 1973;109:167–70.
6. Comeau J, Shaw L, Marcell CC, Lavery JP. Early placenta previa and delivery outcome. Obstet Gynecol 1983;61:577–80.
7. Lydon-Rochelle M, Holt VL, Easterling TR, Martin DP. First-birth cesarean and placental abruption or previa at second birth. Obstet Gynecol 2001;97:765–9.
8. Harris RD, Alexander RD. Ultrasound of the placenta and umbilical cord. In: Callen PW, ed. Ultrasonography in Obstetrics and Gynecology. 4th ed. Philadelphia: WB Saunders; 2000:597–625.
9. Dawson WB, Dumas MD, Romano WM, Gagnon R, Gratton RJ, Mowbray RD. Translabial ultrasonography and placenta previa: Does measurement of the os-placenta distance predict outcome? J Ultrasound Med 1996;15:443–6.
10. Wexler P, Gottesfeld KR. Early diagnosis of placenta previa. Obstet Gynecol 1979;54:231–4.
11. Ananth CV, Smulian JC, Vintzileos AM. The association of placenta previa with history of cesarean delivery and abortion: A metaanalysis. Am J Obstet Gynecol 1997;177:1071–8.
12. Taipale P, Hiilesmaa V, Ylostalo P. Transvaginal ultrasonography at 18–23 weeks in predicting placenta previa at delivery. Ultrasound Obstet Gynecol 1998;12:422–5.
13. Lauria MR, Smith RS, Treadwell MC, Comstock CH, Kirk JS, Lee W, et al. The use of second-trimester transvaginal sonography to predict placenta previa. Ultrasound Obstet Gynecol 1996;8:337–40.
14. Leerentveld RA, Gilberts ECAM, Arnold MJCWJ, Wladimiroff JW. Accuracy and safety of transvaginal sonographic placental localization. Obstet Gynecol 1990;76:759–62.