A Resolved Placenta Previa is Still Associated with Postpartum Hemorrhage

Abstract

Objective

Our objective was to determine whether resolution of a low-lying placenta or placenta previa is associated with postpartum hemorrhage (PPH).

Study Design

This is a retrospective, matched-control cohort study of women who underwent transvaginal sonography during fetal anatomic survey between 18 and 24 weeks of gestation at the University of Pennsylvania from January 2017 to May 2019. Exposure was defined as low-lying placenta (≤1 cm from the internal cervical os) or placenta previa (covering the os) at anatomic survey that was found to be resolved by transvaginal ultrasound in the third trimester. For each exposure, we identified a control patient whose placenta was > 1 cm from internal os at anatomic survey performed on the same day. The primary outcome was PPH at delivery, defined as estimated blood loss ≥ 1,000 mL.

Results

A total of 450 women were included (225/group). The exposed group of resolved placental previa included 85.0% with resolved low-lying placenta and 15.0% with resolved previa. The rate of PPH was significantly higher in the exposed group versus controls (9.8% vs. 4.4%, p = 0.03). Women with resolved previa were 2.5 times more likely to experience PPH than controls (adjusted odds ratio = 2.58, 95% confidence interval: 1.17–5.69), even when controlling for parity, prior cesarean, and delivery mode. Women with resolved previa were also more likely to present to triage with bleeding (16.4% vs. 8.0%, p = 0.006), receive antenatal corticosteroids, (9.3% vs. 3.1%, p = 0.006), and receive intravenous iron postpartum (7.6% vs. 3.1%, p = 0.04).

Conclusion

Our data demonstrate that women with a resolved low-lying placenta or placenta previa remain at significantly increased risk of bleeding-related complications in pregnancy and during delivery when compared with those who never had a previa. Clinicians should consider this association when counseling patients and performing hemorrhage risk stratification.

Postpartum hemorrhage (PPH) is one of the leading preventable causes of maternal mortality.¹ In addition, PPH is a significant contributor to maternal morbidity. PPH can lead to fatigue, coagulopathies, pituitary ischemia, and the need for blood transfusion.²

Low placentation, including a placenta previa (placental tissue covering the internal cervical os) and low-lying placenta (a placenta whose leading edge is within 1 or 2 cm of the os), is a well-established risk factor for hemorrhage, often requiring a cesarean delivery to avoid intrapartum hemorrhage. While low placentation is diagnosed at a rate of 0.28 to 2.8%, most resolve prior to delivery.^3,4 The current literature is conflicted regarding whether sonographic resolution of placenta previa remains a risk factor for PPH, with some studies showing an increased risk of PPH⁵ and others reporting a decreased risk.^6,7

An improved ability to identify women who are at higher risk of hemorrhage can enable providers to better prepare for potential hemorrhage. Currently, many commonly used hemorrhage risk prediction tools do not consider resolved placenta previa as a risk factor.^8–10 The purpose of this study is to determine whether resolution of a low-lying placenta or placenta previa is associated with PPH, with the goal of further optimizing hemorrhage prediction.

Materials and Methods

This is a retrospective matched-control cohort study of women who underwent transvaginal sonography during fetal anatomic survey at 18 to 24 weeks’ gestation at the University of Pennsylvania from December 2016 to May 2019. This study was approved by the Institutional Review Board at the University of Pennsylvania.

Exposure was defined as having a low-lying placenta (≤1cm from cervical os) or placenta previa at anatomic survey followed by resolution on third trimester ultrasound. Resolution of the low-lying placenta or placenta previa was defined as at least 1 cm from the internal os on transvaginal scan. For each exposure, we identified a control subject with placenta > 1 cm from internal os at anatomic survey performed on the same day as the exposed patient. Exclusion criteria were: multiple gestation, malpresentation at delivery, bleeding disorder (not including gestational thrombocytopenia or preeclampsia), anticoagulation therapy, placenta accreta, and persistent low placentation.

The primary outcome was PPH at delivery, defined as estimated blood loss (EBL) ≥ 1,000 mL.⁹ EBL was obtained by best visual estimation by the delivering provider at the time of delivery. Secondary outcomes included EBL, rate of preterm birth (< 37 weeks of gestation), number of triage visits for bleeding complications, induction of labor for bleeding, antenatal corticosteroid use for any indication, mode of delivery, postpartum intravenous iron transfusion, intrapartum and postpartum blood transfusion, and uterotonic use beyond routine postpartum oxytocin.

The baseline PPH rate at our institution is 9%.² To detect a two-fold increase in the risk of PPH in resolved previa/low-lying placenta compared with those who never have a low-lying placenta with 80% power and a two-sided α value of 0.05, would require a total sample of 450 subjects (225 exposed and 225 unexposed). Cases and controls were collected retrospectively through the study period until the desired sample size was reached.

Bivariate comparisons of demographic and baseline clinical characteristics, as well as labor and delivery outcomes, were performed with Fisher’s exact tests and Chi-square tests for categorical variables, and t-tests or Wilcoxon’s rank sum tests for continuous variables, where appropriate. Multivariable logistic regression was utilized to estimate the independent association between outcomes after controlling for confounders. For the purposes of the primary analysis, resolved low-lying placenta and resolved placenta previa were grouped and referred to as “resolved placenta previa.” Sensitivity analyses were also performed comparing just resolved low-lying placenta with the never previa group. We determined covariates in the model in two ways. First, we evaluated for statistical confounders: demographic and clinical characteristics that were associated on bivariate tests (p < 0.10) with both the group/exposure variable (resolved placenta previa vs. never previa) and the outcome variable (PPH). Notably, no covariates met these criteria. Only anterior placental location was associated with exposure, and it was not associated with PPH. Thus, directed acyclic graphs were utilized to select for biologically plausible confounders. As a result, parity and prior cesarean were included in modeling for antepartum outcomes. Mode of delivery was added as a potential confounder for postpartum outcomes. Backward stepwise elimination of covariates (with p-value > 0.10 for removal) was performed to determine which covariates would be retained in the final model. Statistical analyses were performed with Stata 15 (StataCorp, College Station, TX). Statistical significance was set at p < 0.05.

Results

A total of 450 women (225 exposed and 225 unexposed) who underwent their fetal anatomic survey from December 2016 to May of 2019 were included in the study. Of exposed, 85.0% had resolved low-lying placenta, while 15.0% had resolved previa; these were grouped for analysis as “exposed.”

Demographic and clinical characteristics among exposed and unexposed are shown in ►Table 1. The exposed group was more likely to have anterior placental location at the time of anatomy scan. The groups did not differ significantly in terms of race, parity, body mass index, gestational age at diagnosis, and history of cesarean delivery (►Table 1).

Table 1.

Demographics

		Exposed N = 225	Unexposed (no previa) N = 225
		n (%)	n (%)	p-Value
Race	Non-Black	114 (50.7)	100 (44.4)	0.19
	Black	111 (49.3)	125 (55.6)
Nulliparous		133 (59.1)	119 (52.9)	0.18
BMI, mediana		25.0 [21.6, 29.8]	25.4 [21.1, 31.2]	0.52
Gestational age at time of anatomic survey (wk)a		20 [20, 20]	20 [20, 20]	0.51
Placenta location	Not anterior	73 (34.8)	106 (47.3)	0.008
	Anterior	137 (65.2)	118 (52.7)
History of cesarean	No CS	199 (88.4)	193 (85.8)	0.40
	Prior CS	26 (11.6)	32 (14.2)

Abbreviations: BMI, body mass index; CS, cesarean; IQR, interquartile range.

Note: Bold p-values are statistically significant.

^aMedian [IQR]; Exposed group includes resolved previa or low-lying placenta; Unexposed group never had a previa or low-lying placenta.

The rate of PPH was significantly higher with exposed group versus unexposed (9.8% vs. 4.4%, p = 0.03; ►Table 2). Women with resolved previa or low-lying placenta were 2.5 times more likely to experience PPH than controls (adjusted odds ratio [aOR] = 2.58, 95% confidence interval [CI]: 1.17–5.69), even when controlling for parity, prior cesarean, and delivery mode (►Table 2).

Table 2.

Primary and secondary outcomes

	Exposed (n = 225) n (%)	Unexposed (n = 225) n (%)	p-Value	Unadjusted OR [95% CI]	Adjusted OR [95% CI]
Postpartum hemorrhage (PPH)a	22 (9.8)	10 (4.4)	0.03	2.33 [1.08–5.04]	2.58 [1.17–5.69]b
Secondary outcomes
Triage visit for bleeding	37 (16.4)	18 (8.0)	0.006	2.26 [1.25–4.11]	2.27 [1.25–4.13]c
Antenatal corticosteroid use	21 (9.3)	7 (3.1)	0.006	3.21 [1.33–7.70]	3.33 [1.28–8.04]c
Induction of labor for bleeding	13 (5.8)	3 (1.3)	0.02	4.54 [1.28–16.15]	4.57 [1.28–16.34]c
Preterm birth (<37 weeks’ gestation)	20 (8.9)	18 (8.0)	0.74	1.12 [0.58–2.18]	1.13 [0.58–2.20]c
Gestational age at deliveryd	39 [38–39]	39 [38–39]	0.43	NA	NA
Mode of delivery			0.35	NA	NA
Spontaneous vaginal	150 (66.7)	154 (68.4)
Operative vaginal	12 (5.3)	6 (2.7)
Cesarean	63 (28.0)	65 (28.9)
EBLd	400 [300–800]	400 [250–700]	0.04	NA	NA
Unanticipated uterotonic use	51 (22.7)	34 (15.1)	0.04	1.65 [1.02–2.66]	1.70 [1.04–2.78]b
Blood transfusion intrapartum and postpartum	16 (7.1)	13 (5.8)	0.57	1.25 [0.59–2.66]	1.28 [0.59–2.78]b
Intravenous iron transfusion postpartum	17 (7.6)	7 (3.1)	0.04	2.54 [1.03–6.26]	2.62 [1.05–6.55]b
Intra-amniotic infection	12 (5.3)	21 (9.3)	0.10	NA	NA

Abbreviations: CI, confidence interval; EBL, estimated blood loss; IQR, interquartile range; OR, odds ratio.

^aDefined as estimated blood loss (EBL) ≥ 1,000 mL.

^bAdjusted for parity, prior cesarean, and delivery mode.

^cAdjusted for parity and prior cesarean.

^dMedian [IQR]; Exposed group includes resolved previa or low-lying placenta; Unexposed group never had a previa or low-lying placenta.

The exposed group was also more likely to present to triage with bleeding (16.4% vs. 8.0%, p = 0.006), be induced for bleeding (13 vs. 3%, p = 0.02), receive antenatal corticosteroids (9.3 vs. 3.1%, p = 0.006), and require intravenous iron postpartum (7.6 vs. 3.1%, p = 0.04). These outcomes also remained significant when controlling for confounders. There was no significant difference between the groups in gestational age at delivery, preterm birth, mode of delivery, or need for blood transfusion (►Table 2). Of the 38 preterm births, 14 were spontaneous and 24 were medically indicated. There was no difference by exposed or unexposed in either spontaneous or medically indicated preterm births. There was also no difference by exposed or unexposed in the rate of blood transfusion intrapartum or postpartum.

A sensitivity analysis was performed comparing only resolved low-lying placenta, excluding cases with resolved placenta previa, to the control group (►Table 3). The results demonstrated overall similar findings. Significantly increased rate of PPH remained in the resolved low-lying placenta when compared with controls after adjusting for the confounders (aOR = 2.34 [1.02–5.37]). Women with resolved low-lying placenta were more likely to present to triage with bleeding (15.7 vs. 8.0%, p = 0.03), be induced for bleeding (5.2 vs. 1.3%, p = 0.03), and receive antenatal corticosteroids (9.4 vs. 3.1%, p = 0.007). These outcomes also remained significant when controlling for confounders (►Table 3).

Table 3.

A sensitivity analysis

	Resolved low-lying placenta (n = 191) n (%)	Unexposed (n = 225) n (%)	p-Value	Unadjusted OR [95% CI]	Adjusted OR [95% CI]
Postpartum hemorrhage (PPH)a	17 (8.9)	10 (4.4)	0.07	2.10 [0.94–4.70]	2.34 [1.02–5.37]b
Triage visit for bleeding	28 (15.7)	18 (8.0)	0.03	1.98 [1.06–3.70]	1.99 [1.06–3.73]c
Antenatal corticosteroid use	18 (9.4)	7 (3.1)	0.007	3.24 [1.32–7.93]	3.33 [1.35–8.20]c
Induction of labor for bleeding	10 (5.2)	3 (1.3)	0.03	4.09 [1.11–15.08]	4.12 [1.11–15.33]c
Preterm birth (<37 weeks’ gestation)	19 (10.0)	18 (8.0)	0.49	1.27 [0.65–2.50]	1.28 [0.65–2.53]c
Gestational age at deliveryd	39 [38–39]	39 [38–39]	0.38	NA	NA
Mode of delivery			0.36	NA	NA
Spontaneous vaginal	131 (68.6)	154 (68.4)
Operative vaginal	10 (5.2)	6 (2.7)
Cesarean	50 (26.2)	65 (28.9)
EBLd	400 [300–700]	400 [250–700]	0.26	NA	NA
Unanticipated uterotonic use	40 (20.9)	34 (15.1)	0.12	1.49 [0.90–2.46]	1.51 [0.90–2.52]b
Blood transfusion intrapartum and postpartum	14 (7.3)	13 (5.8)	0.52	1.29 [0.59–2.82]	1.40 [0.63–3.11]b
Intravenous iron transfusion postpartum	11 (5.8)	7 (3.1)	0.19	1.90 [0.72–5.01]	2.01 [0.75–5.37]b
Intra-amniotic infection	12 (6.3)	21 (9.3)	0.25	NA	NA

Abbreviations: CI, confidence interval; EBL, estimated blood loss; IQR, interquartile range; OR, odds ratio.

^aDefined as estimated blood loss (EBL) ≥ 1,000 mL.

^bAdjusted for parity, prior cesarean, and delivery mode.

^cAdjusted for parity and prior cesarean.

^dMedian [interquartile range].

Discussion

Women with resolved previa or low-lying placenta were 2.5 times more likely to experience PPH than controls, even when controlling for parity, prior cesarean, and delivery mode. These women were also more likely to present to a triage unit with complaint of bleeding, receive antenatal corticosteroid, be induced for bleeding, receive uterotonics at the time of delivery, and require intravenous iron postpartum.

The previously published literature on this topic has reached conflicting conclusions regarding the effect of resolved previa on the incidence of PPH. For example, Magann et al analyzed outcomes data on second trimester placenta implantation sites from 3,336 pregnancies, 93 of which had “low” placentas. Persistent placenta previas were excluded.⁵ PPH was defined as blood loss greater than or equal to 1,000 mL, a drop in hematocrit of 10 points, or need for blood transfusion. Using fundal placentation as a reference group, they found significantly reduced risk of PPH in the low placenta group (aOR: 0.56, 95% CI: 0.46–0.95). However, although the authors concluded that low placental implantation is associated with a reduced risk of PPH, it is important to note that only 19 of their “low placentas” were within 2 cm of the internal os at the anatomic survey.⁵

On the other hand, Ogueh et al found that pregnant women with an initially low-lying placenta had a higher incidence of PPH (3.6 vs. 2.1%; aOR = 1.77, 95% CI: 1.14–2.75).⁶ This difference persisted even after excluding cases of persistent placenta previa at time of delivery (aOR = 1.62, 95% CI: 1.04–2.54). However, transvaginal sonography was not universally used to evaluate placental location and low-lying placenta was defined as a placenta anywhere in the lower uterine segment, including covering the cervical os. We have previously reported of the potential limitations of transabdominal sonography in evaluating placental location.¹¹ In addition, the report did not include data on routine third trimester transvaginal sonography for reassessment of placental location. Finally, their definition of PPH was EBL greater than 500 mL for a vaginal delivery and greater than 1,000 mL for a cesarean delivery.

Osmundson et al performed a retrospective study to evaluate whether low placentation in the second trimester is an independent risk factor for PPH.⁷ The study defined PPH as EBL greater than 500 mL for a vaginal delivery and greater than 1,000 mL for a cesarean delivery. The study showed that, compared with computer-generated controls, the rate of PPH was significantly higher for women with resolved low placentation (aOR = 2.72, 95% CI: 1.46–5.07). Similar to our study, second trimester placental location was confirmed by transvaginal ultrasound; however, they did not report whether transvaginal scan was used to confirm resolution prior to delivery. In addition, the study defined low placentation as any placenta within 2.5 cm of the os. Notably, our study defined low-lying placenta as 1 cm from the internal os which is supported by Vergani et al who showed that a placental edge > 10 mm from the cervical os was not associated with emergent delivery due to hemorrhage and had significantly lower risk of PPH and EBL compared with placentas 1 to 10 mm from the os.¹² Finally, we also used the most recent American College of Obstetricians and Gynecologists (ACOG) definition of PPH, greater than or equal to 1,000 mL, regardless of mode of delivery.

Maternal hemorrhage requiring blood transfusion is one of the leading causes of maternal morbidity and therefore properly identifying risk factors that may contribute to maternal hemorrhage is of paramount importance. Given the principal finding of this study that patients with resolved placenta previa is 2.5 times more likely to experience PPH, the first step is educating providers that resolved placenta previa is still a significant risk factor for hemorrhage. Yet, a recent systematic review on risk factors for atonic PPH did not include resolved previa as a candidate risk factor to explore.¹³ In addition, current PPH risk assessment tools do not include resolved placenta previa as a risk factor.^8,10 Thus, the addition of this significant risk factor to these risk assessment tools may help improve the prediction of PPH and enable clinicians to better counsel patients and prepare.

Our results also demonstrate that pregnancies with a resolved previa are also at increased risk of antepartum bleeding events. They are more likely to present to obstetric triage with bleeding concerns, receive antenatal corticosteroids, and ultimately be delivered for a bleeding event. Therefore, providers treating these women in the antenatal period should also be cognizant of the status of the resolved placenta previa.

Finally, these increased risks of bleeding of women with resolved placenta previa highlight the importance of optimizing hemoglobin level in the antenatal period with adequate screening and prompt and appropriate treatment of anemia. Our result showed a significant increase in the use of postpartum intravenous iron treatment for postpartum anemia in patients with resolved previa. However, by recognizing the increased risk of hemorrhage in patient with resolved previa, it is possible that some of these women could have benefited from a more aggressive antenatal treatment for anemia in preparation for delivery.

Strengths and Limitations

The primary strength of the study is that it is composed of a large cohort of women who all underwent transvaginal ultrasound, the gold standard for the diagnosis of placenta previa. In addition, this study also utilized the most recent definition of PPH by ACOG. Another strength of the study is our evaluation of secondary outcomes such as needing a triage visit for bleeding, being administered corticosteroid injections, and being induced for bleeding complications which expands the significance of the result from the postpartum period to the antepartum period.

Limitations include its retrospective nature. However, each exposure was matched to a control that was scanned in the same ultrasound session on the same day to minimize the risk of confounding.

It is notable that the rate of PPH in the unexposed group was 4.4%, whereas a previous study done in our institution found that the baseline PPH rate was 9%. This difference is likely explained by the fact that the current work excluded some women at highest risk for hemorrhage including women with multiple gestations, malpresentation at delivery, bleeding disorder, anticoagulation therapy, placenta accreta, and persistent previa. Despite a lower than predicted rate of PPH in the unexposed group, a statistically significant difference between the two groups was noted for the primary outcome. In addition, while the median EBL for both exposed and unexposed group in our study was similar, this is likely because most patients in both groups have normal EBLs. However, the significant differences seen in more clinically relevant outcomes, such as PPH and triage or delivery for bleeding, are better indicators of the risk associated with a resolved previa.

Finally, the study population was drawn from one institution which may limit generalizability to other institutions. Specifically, our study used a cutoff of 1 cm since, at our institution, clinical practice is to forego further surveillance and allow trial of labor for placentas that are 1 cm or greater from the os at anatomy scan. This may limit extension of our results to centers that may use a larger cutoff distance for managing surveillance and delivery planning. However, our institution is composed of a diverse, urban population and therefore may be generalizable to other similar academic institutions.

Conclusion

Women with a resolved placenta previa and low-lying placenta remain at significantly increased risk of bleeding-related complications in pregnancy and during delivery when compared with those who never had a previa. This association should be considered when performing hemorrhage risk stratification.

Key Points.

• Women with resolved previa or low-lying placenta were 2.5 times more likely to experience PPH.

• Women with resolved previa or low-lying placenta were more likely to be induced for bleeding.

• Resolved previa or low-lying placenta is still associated with adverse hemorrhage-related outcomes.