Maternal Cardiovascular and Cerebrovascular Health After Placental Abruption: A Systematic Review and Meta-Analysis (CHAP-SR)

Abstract

Placental abruption and cardiovascular disease (CVD) have common etiological underpinnings, and there is accumulating evidence that abruption may be associated with future CVD. We estimated associations between abruption and coronary heart disease (CHD) and stroke. The meta-analysis was based on the random-effects risk ratio (RR) and 95% confidence interval (CI) as the effect measure. We conducted a bias analysis to account for abruption misclassification, selection bias, and unmeasured confounding. We included 11 cohort studies comprising 6,325,152 pregnancies, 69,759 abruptions, and 49,265 CHD and stroke cases (1967–2016). Risks of combined CVD morbidity-mortality among abruption and nonabruption groups were 16.7 and 9.3 per 1,000 births, respectively (RR = 1.76, 95% CI: 1.24, 2.50; I² = 94%; τ² = 0.22). Women who suffered abruption were at 2.65-fold (95% CI: 1.55, 4.54; I² = 85%; τ² = 0.36) higher risk of death related to CHD/stroke than nonfatal CHD/stroke complications (RR = 1.32, 95% CI: 0.91, 1.92; I² = 93%; τ² = 0.15). Abruption was associated with higher mortality from CHD (RR = 2.64, 95% CI: 1.57, 4.44; I² = 76%; τ² = 0.31) than stroke (RR = 1.70, 95% CI: 1.19, 2.42; I² = 40%; τ² = 0.05). Corrections for the aforementioned biases increased these estimates. Women with pregnancies complicated by placental abruption may benefit from postpartum screening or therapeutic interventions to help mitigate CVD risks.

Abbreviations:

CI

confidence interval

CHD

coronary heart disease

CVD

cardiovascular disease

HR

hazard ratio

RR

risk ratio

Globally, an estimated 423 million men and women were diagnosed with cardiovascular disease (CVD) in 2016, of whom 17.9 million (31%) died from CVD-related causes (1). Obstetrical complications, particularly ischemic placental disease (2, 3) (preeclampsia, placental abruption, and fetal growth restriction (4, 5)), gestational diabetes (6, 7), and preterm delivery (8–11), have a profound and lasting impact on the cardiovascular health of women and newborns in adulthood (12–15). Placental abruption is the consequence of acute stimuli at the maternal-fetal interface resulting in the rupture of the decidual artery and leading to premature placental separation (16, 17). Abnormal vascular remodeling, thrombosis/coagulation, infection, and angiogenesis leading to placental inflammation are intermediate pathophysiological effects that lead to poor placentation or accelerated placental detachment (18, 19). Placental abruption affects approximately 1% of pregnancies but has a 10-fold higher recurrence rate (20–22) and is associated with high rates of perinatal and infant mortality (23–27).

Multiple risk factors, including hypertension, hyperlipidemia, diabetes mellitus, smoking, obesity, and older age contribute to the development of CVD and stroke (28). These same risk factors are also implicated in increased risk of placental abruption, suggesting an overlap in epidemiologic risk profiles (23, 29). It is also noteworthy that placental abruption and CVD have common etiologic underpinnings, including thrombosis, coagulation defects, inflammation, and infection. CVD, in general, is more vascular than thrombosis-mediated while stroke is relatively more thrombotic and less vascular-mediated (30). These profiles correspond well with respect to the epidemiology of placental abruption—a majority of abruption cases are vascular-mediated rather than thrombotic-mediated (16). These reasons provide a strong and compelling rationale to support a biologic connection between placental abruption and future CVD complications.

There have been at least 9 meta-analyses (as of October 2020) examining the associations between preeclampsia/eclampsia, which are strong risk factors for placental abruption (23, 26, 31–37), and risks of CVD over the life course (37–44). Of these, only one (7 cohort studies) examined the association between placental abruption and CVD (including stroke) (39). Although insightful, this meta-analysis leaves several unresolved questions, including: 1) whether abruption is associated with increased risks of CVD mortality and nonfatal morbidity separately; 2) whether women with ≥2 placental abruptions across pregnancies are at higher risk of CVD, suggestive of a dose-response association; and 3) whether these associations are affected by exposure (placental abruption) misclassification and unmeasured confounding biases. To address these knowledge gaps, we performed a systematic review and meta-analysis to examine the associations between placental abruption and CVD.

METHODS

Review protocol and registration

This systematic review and meta-analysis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (45). The review protocol was registered in PROSPERO (CRD42019136445) (46). Meta-analysis is not considered human subjects research, so no ethics approval from institutional review board was sought.

Data sources and searches

Systematic literature searches were performed after extensive testing in following databases: PubMed (National Center for Biotechnology Information, Bethesda, Maryland), the Cochrane Library (John Wiley & Sons, Hoboken, New Jersey), CINAHL (Cumulative Index to Nursing and Allied Health Literature) in EBSCOhost (EBSCO Information Services, Ipswich, Massachusetts), Web of Science (Core Collection, the Thomson Corporation, Toronto, Canada), and Scopus (Elsevier BV) (since inception). All the searches were focused on 2 main concepts: placental abruption and cardiovascular diseases. Related terms and MeSH headings were identified and searched using “OR” to connect the results for each concept and then using “AND” to combine the search results from both concepts. The search strategy implemented in PubMed is shown in Web Table 1 (available at https://doi.org/10.1093/aje/kwab206). No filters were applied in the searches. All the search results were exported to an EndNote (Clarivate Analytics, Philadelphia, Pennsylvania) library. Duplicate references were removed before the EndNote library was sent to the subject experts for screening. The number of search results, the number of screened references, and the final set of included studies are displayed in a flow diagram in Figure 1.

Figure 1.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram of study selection. The flow diagram shows the total number of studies identified in the literature search, those that were screened, and those that were found eligible for inclusion in the meta-analysis. CVD, cardiovascular disease.

Study selection

We included observational studies that examined maternal CVD risks in relation to placental abruption. We additionally excluded studies that examined only coronary heart disease (CHD) and stroke risk factors as outcomes, as well as data in abstract form, reviews, editorials, commentaries, and book chapters; papers published in a non-English language were also excluded.

Primary and secondary endpoints

The primary endpoint was CVD morbidity and mortality combined. Hereafter, we refer to heart-related conditions as CHD. The secondary endpoints included: 1) CHD and stroke mortality combined; 2) CHD mortality only; 3) stroke mortality only; and 4) CHD and stroke morbidity combined. Since associations between placental abruption and nonfatal CHD (47) and nonfatal stroke (48) outcomes were reported in only 1 study each, we did not perform a meta-analysis for these outcomes. However, these studies are included the secondary endpoints listed in 1–4 above.

Data extraction and quality assessment

Two authors (C.V.A. and M.S.) reviewed the titles (and the abstract, when necessary) of the 977 paper identified. Both authors also completed an independent assessment of bias on all studies. Two authors (H.S.P. and S.A.) independently extracted the relevant data from all identified studies. The study by Lykke et al. (49) was not included in the meta-analysis to avoid duplication, since more recent data in an expanded Danish population (1978–2010) were recently reported (47, 48). DeRoo et al. (50) reported associations between placental abruption and CHD and stroke-related deaths in Norway and Sweden separately in a single publication, so we undertook the meta-analysis separately for Norway and Sweden.

Data synthesis and statistical analysis

The primary effect measure was the pooled risk ratio (RR) with 95% confidence interval (CI) based on random-effects models and was weighted based on the inverse of the study size. Heterogeneity in the effect measure in the pooled analysis was determined by I² and τ² statistics. The meta-analysis was performed for the primary and 4 secondary outcomes listed above. We looked for evidence of publication bias in the pooled estimates, based on the funnel plot. Since the number of studies for each CHD and stroke outcome was small, publication bias was assessed only for the combined CHD and stroke mortality and morbidity.

Subgroup analysis

In a planned subgroup analysis, we estimated the associations between placental abruption and secondary outcomes, defined based on the specific subtype of CHD, including ischemic heart disease, myocardial infarction, hypertensive heart disease, and congestive heart failure. We also examined specific subtypes of cerebrovascular disease, including ischemic stroke and hemorrhagic stroke.

Sensitivity analysis

We performed 2 sensitivity analyses. First, we examined whether the number of placental abruption episodes across successive pregnancies was associated with increased risks of CHD and stroke morbidity and mortality. Second, some studies did not report raw data in 2 × 2 tables of the abruption by CHD (and stroke) cross-classifications but instead reported a confounder-adjusted hazard ratio (HR) with 95% CI. We extracted the HR and 95% CI and back-calculated the standard errors of the HR. We then pooled the HRs from all studies based on the adjusted HRs.

Quality assessment

We undertook a bias analysis based on the Risk of Bias in Non-randomized Studies of Interventions (ROBINS-I) tool (51). Each of the individual manuscripts was reviewed by 2 authors (C.V.A. and M.S.), both of whom assigned a low, moderate, serious, or critical risk of bias, or no information available, for the following 7 categories: 1) bias due to confounding; 2) bias in selection of participants into the study; 3) bias in classification of interventions; 4) bias due to missing data; 5) bias in measurement of outcomes; 6) bias in selection of the reported results; and 7) overall bias. The minimum set of confounders that required adjustment in each of the studies included maternal age (or age as a time scale in the Cox regression models), a measure of socioeconomic status (e.g., education, income), and smoking, as well as year of delivery. Since placental abruption (exposure) was not an intervention in the studies included, bias due to deviation from intended interventions was not assessed.

Multiple probabilistic bias analysis

We undertook a probabilistic bias analysis to assess the extent to which exposure (placental abruption) misclassification, selection bias, and unmeasured confounding simultaneously may have affected the associations reported in the pooled meta-analysis estimates (52, 53).

Exposure misclassification bias

Correction for exposure misclassification was accomplished based on the following assumptions: 1) the misclassification of placental abruption was assumed nondifferential with respect to the outcome; 2) the sensitivity and specificity for abruption diagnosis was assumed to range between 0.50 and 1.00, and between 0.99 and 1.00, respectively, derived from an earlier validation study (54); and 3) the simulations were performed assuming a uniform distribution. We estimated the bias-corrected median RR and intrinsic 95% CI based on 100,000 replications.

Selection bias

Correction for potential selection bias in the outcome (CHD/stroke) was based on the following assumptions: 1) the selection probabilities were assumed nondifferential with respect to the outcome; 2) the sensitivity and specificity for abruption diagnosis was assumed to range between 0.75 and 1.00, and between 0.99 and 1.00, respectively; and 3) the simulations were performed assuming a uniform distribution. We estimated the bias-corrected median RR and intrinsic 95% CI based on 100,000 replications.

Unmeasured confounding bias

We estimated the impact of unmeasured confounding on the association between placental abruption and CHD and stroke outcomes based on the following assumptions: 1) prevalence of the unmeasured confounder(s) among those with and without abruption ranged between 5% and 30%, and 3% and 20%, respectively (e.g., maternal smoking, a strong confounder of abruption (31, 55, 56)); 2) RR for the confounder-outcome association ranged between 0.50 and 5.00; and 3) simulations were performed assuming a uniform distribution. We estimated the unmeasured confounding bias–corrected median RR and intrinsic 95% CI based on 100,000 replications.

Meta-analysis was performed in RevMan (version 5.3; the Cochrane Collaboration, the Nordic Cochrane Centre, Copenhagen, Denmark) and RStudio (version 1.3.1056; RStudio, PBC, Boston, Massachusetts) with the package “episensr” (57) for the probabilistic bias analysis.

RESULTS

Study overview

The literature search identified a total of 947 unique papers. Both reviewing authors agreed that 12 studies (47, 48, 50, 58–66) were potentially eligible for inclusion in the systematic review and meta-analysis. The data extracted from each of the 12 studies by 2 authors (H.S.P. and S.A.) were cross-checked for between-reviewer accuracy and were found to be 100% accurate. The characteristics of subjects in the 12 papers are described in Table 1; 11 of them were cohort by design (47, 48, 50, 58–64, 66), and 1 was a hospital-based case-control study (65). Since we were unable to combine data from cohort studies with the case-control study, the latter was not included in the meta-analysis (but considered for the systematic review). The meta-analysis included 11 cohort studies comprising 6,325,152 pregnancies, 69,759 placental abruptions, and 49,265 CHD and stroke cases. The median follow-up in the individual studies ranged from 5 to 25 years.

Table 1.

Description of Studies With Cardiac Disease and Cerebrovascular Endpoints Included in the Systematic Review and Meta-Analysis

First Author, Year (Reference No.)	Country	Years	Study Design	Follow-up Median (Range), years	Placental Abruption		Nonabruption	CVD and Stroke Outcomes		Comments
					No.	%	No.	Mortality	Nonfatal Complications
Ananth, 2017 and 2019 (47, 48)	Denmark	1978–2010	Retrospective cohort	16 (8–24)	11,829	0.7	1,644,258	All CVD and stroke	IHD, MI, HHD, CHF; ischemic (TIA and cerebral infarction) and hemorrhagic (subarachnoid and intracerebral) stroke	All births
Cain, 2016 (64)	United States (Florida)	1998–2010	Retrospective cohort	5 (4–6)	3,032	1.1	265,973	Not examined	IHD, PAD, CHF, and stroke	First singleton births only
DeRoo, 2016 (50)	Norway	1967–2002	Retrospective cohort	25 (1–42)	4,732	0.6	831,415	All CVD (and IHD), stroke	Not examined	Primary analysis: first singleton births; and all births
DeRoo, 2016 (50)	Sweden	1973–2003	Retrospective cohort	23 (1–42)	6,249	0.5	1,275,401	All CVD (and IHD), stroke	Not examined	Primary analysis: first singleton births; and all births
Pariente, 2014 (63)	Israel (Negev)	1988–2010	Hospital-based cohort	14	653	1.4	47,585	All causes	CVD-related hospitalizations	All births
Riihimaki, 2017 (62)	Finland	1969–2005	Matched case-cohort	25 (0–45)	7,805		25,523	All CVD (and IHD) and stroke	Not examined	3:1 match of nonabruption to abruption cases
Sia, 2019 (65)	Canada (Alberta)	2005–2009	Hospital, based case-control study	–a	5	2.0	8 (3.3%)	Not examined		Cases included women with CAD or history of MI
Soh, 2016 and 2015 (59, 60)	Sweden	1973–2011	Retrospective cohort	10 (5–22)	40	1.1	3,638	All causes; and MI, PAD, stroke	MI, PAD, stroke	Parous women with SLE
Ray, 2005 (58)	Canada (Ontario)	1990–2004	Retrospective cohort	9 (0–14)	11,156b		950,885	All causes	Not examined	First births only
Ray, 2016 (61)	Canada (Ontario)	1993–2012	Retrospective cohort	5	48	2.4	1,623	Coronary artery revascularization	Not examined	Women with ≥1 previous delivery; those with CVD ≤1 year before or coronary revascularization ≤90 days after any delivery were excluded
Ukah, 2020 (66)	Canada (Quebec)	1989–2016	Prospective cohort	29	24,258	2.0	1,200,717	Not examined	–c	First CVD hospitalization

Abbreviations: CHF, congestive heart failure; CVD, cardiovascular disease; HHD, hypertensive heart disease; IHD, ischemic heart disease; IQR, interquartile range; MI, myocardial infarction; PAD, peripheral artery disease; SLE, systemic lupus erythematosus; TIA, transient ischemic attack.

^a CVD cases: 3.0 (IQR, 2.0–4.0), controls: 3.5 (IQR, 2.0–5.0).

^b 9,303 with infarctions.

^c CVD outcomes examined included heart failure, IHD, MI, angina, cardiac arrest, inflammatory heart disease, conduction defects, valve disorders, cardiomyopathy, and pulmonary heart disease as well as cardiovascular interventions (coronary angioplasty, coronary artery bypass graft, pacemaker, valve surgery, cardiac transplant) and admission to a coronary care unit. Cerebrovascular disease included ischemic and hemorrhagic strokes, atherosclerotic disease, aortic aneurysm or dissection, other aneurysm, and arterial embolism.

Primary endpoint

Combined CHD and stroke morbidity and mortality.

We extracted data from 7 studies (48, 50, 60, 62–64, 66) that reported 8 associations between placental abruption and risks of combined morbidity and mortality from CHD and stroke (Figure 2). Risks of the combined CHD and stroke morbidity and mortality among placental abruption and nonabruption groups were 16.7 and 9.3 per 1,000 births, respectively (RR = 1.76, 95% CI: 1.24, 2.50; I² = 94%; τ² = 0.22). Data on outcomes cross-classified by placental abruption status, and the study weights, are described in Web Table 2.

Figure 2.

Meta-analysis of studies reporting the association between placental abruption and combined risks of total coronary heart disease and stroke mortality and morbidity, multiple countries, 1967–2016. The forest plot shows the effect measure for each study that reported associations between placental abruption and combined risks of total coronary heart disease and stroke mortality and morbidity, with the pooled risk ratio (RR) = 1.76 (95% confidence interval (CI): 1.24, 2.50). There was substantial heterogeneity in the pooled effect estimates (I² = 93%; τ² = 0.21). All risk ratios are weighted by the inverse-variance method.

Quality assessment.

The risk-of-bias analysis showed that the overall bias for most studies was of moderate risk, while the studies by Riihimaki et al. (62) and Sia et al. (65) were deemed to have a serious risk of bias. (Web Table 3). Eight studies were rated as having moderate or serious risk of bias for not having met the minimum set of confounder adjustments, while all studies were rated as having moderate risk of bias for the classification of interventions. There was no clear evidence of publication bias (Web Figure 1).

Secondary endpoints

CHD and stroke mortality.

Associations between placental abruption and mortality from CHD and stroke were reported in 5 studies (48, 50, 60, 62, 63). Placental abruption was associated with an increased risk of the combined endpoint of CHD and stroke mortality, with a pooled RR of 2.65 (95% CI: 1.55, 4.54; I² = 85%; τ² = 0.36; Figure 3). Pooled data from 5 studies (47, 50, 60, 62, 63) showed that abruption was associated with a 2.64-fold (95% CI: 1.57, 4.44; I² = 76%; τ² = 0.31) increased risk of CHD deaths (Web Figure 2). Similarly, data on placental abruption and stroke deaths were reported in 3 studies (48, 50, 62), with a pooled RR of 1.70 (95% CI: 1.19, 2.42; I² = 40%; τ² = 0.05) (Web Figure 3). DeRoo et al. (50) also reported increased risks of death from ischemic heart disease (HR = 2.0, 95% CI: 1.4, 2.9) and stroke (HR = 1.6, 95% CI: 1.0, 2.4) following placental abruption in the first pregnancy.

Figure 3.

Meta-analysis of studies reporting the association between placental abruption and combined risks of coronary heart disease and stroke mortality, multiple countries, 1967–2016. The forest plot shows the effect measure for each study that reported associations between placental abruption and combined risks of coronary heart disease and stroke mortality, with the pooled risk ratio (RR) = 2.65 (95% confidence interval (CI): 1.55, 4.54; I² = 84%; τ² = 0.33). All risk ratios are weighted by the inverse-variance method.

Nonfatal CHD and stroke morbidity.

Data on the association between abruption and nonfatal CHD and stroke morbidity were extracted from 5 studies (48, 60, 63, 64, 66). Placental abruption was associated with increased risk of CHD and stroke morbidity (RR = 1.32, 95% CI: 0.91, 1.92; I² = 93%; τ² = 0.15) (Figure 4). Ananth et al. (47) reported increased risks for ischemic heart disease (HR = 1.6, 95% CI: 1.4, 1.9), acute myocardial infarction (HR = 1.9, 95% CI: 1.4, 2.4), hypertensive heart disease (HR = 2.2, 95% CI: 1.1, 4.5), and congestive heart failure (HR = 1.7, 95% CI: 1.2, 2.3) in relation to abruption. Ananth et al. (48) also reported increased risks of ischemic stroke (HR = 1.4, 95% CI: 1.1, 1.7) and hemorrhagic stroke (HR = 1.4, 95% CI: 1.1, 1.9) of similar magnitudes in relation to placental abruption.

Figure 4.

Meta-analysis of studies reporting the association between placental abruption and combined risks of nonfatal coronary heart disease and nonfatal stroke morbidity, multiple countries, 1967–2016. The forest plot shows the effect measure for each study that reported associations between placental abruption and combined risks of combined risks of nonfatal coronary heart disease and nonfatal stroke morbidity, with the pooled risk ratio (RR) = 1.32 (95% confidence interval (CI): 0.91, 1.92; I² = 81%; τ² = 0.06). All risk ratios are weighted by the inverse-variance method.

Probabilistic bias analysis

The multiple probabilistic bias analysis suggests that once the reported associations were corrected for exposure misclassification and unmeasured confounding, the strength of the RRs for all CHD and stroke outcomes were higher compared with the pooled estimates (Table 2). For instance, for CHD and stroke mortality and morbidity combined, the bias-corrected RR was 3.85 (95% CI: 2.66, 4.85) in contrast to the observed estimate of 1.76 (95% CI: 1.24, 2.50).

Table 2.

Simultaneous Correction for Exposure Misclassification Selection Bias and Unmeasured Confounding (Random Bias) Based on a Probabilistic Bias Analysis of the Association Between Placental Abruption and Cardiac Disease and Stroke, Data From Multiple Countries, 1967–2016

		Placental Abruption			Nonabruption
CVD and Stroke Outcome	No. of Studies							Observed Pooled Estimate		Corrected for Exposure Misclassification and Selection Bias		Corrected for Exposure Misclassification, Selection Bias, and Unmeasured Confounding
		Total No. of Pregnancies	No. of Events a	Risk Per 1,000 Pregnancies	Total No. of Pregnancies	No. of Events a	Risk Per 1,000 Pregnancies	RR	95% CI	RR	95% CI	RR	95% CI
CHD-stroke mortality and morbidity combined	7	58,598	979	16.7	5,292,550	49,265	9.3	1.76	1.24, 2.50	3.87	2.67, 4.82	3.85	2.66, 4.85
Mortality
CHD-stroke combined	5	31,308	121	3.9	3,827,820	5,066	1.3	2.65	1.55, 4.54	4.01	2.77, 4.99	3.95	2.70, 5.18
CHD only	5	31,308	72	2.3	3,827,820	3,308	0.9	2.64	1.57, 4.44	5.10	3.51, 6.34	5.00	3.33, 6.86
Stroke only	3	30,615	61	2.0	3,770,597	3,496	0.9	1.70	1.19, 2.42	3.53	2.43, 4.40	3.46	2.28, 4.83
Nonfatal morbidity
CHD-stroke combined	5	39,813	863	21.7	3,162,171	44,607	4.1	1.32	0.91, 1.92	2.30	1.58, 2.86	2.28	1.58, 2.88

Abbreviations: CHD, coronary heart disease; CI, confidence interval; CVD, cardiovascular disease; RR, risk ratio.

^a Events pertain to coronary heart disease or stroke.

Sensitivity analysis

Three studies (47, 48, 50) provided data on the associations between the number of placental abruptions across pregnancies and risks of CHD and stroke morbidity and mortality. Two studies (47, 50) showed evidence of progressively increasing risk of CHD deaths among women with 1 and ≥2 abruptions, and Ananth et al. (48) showed progressively increasing risk of stroke morbidity among women with 1 and ≥2 placental abruptions (Web Table 4).

We extracted and pooled the adjusted HRs from 5 studies (48, 50, 60, 61, 63) that reported an association between placental abruption and deaths from CHD or stroke. The pooled HR across these studies was 1.93 (95% CI: 1.52, 2.44; I² = 17%; τ² = 0.02) (Web Figure 4). Similarly, we extracted and pooled the adjusted hazard ratios from 5 studies (48, 60, 61, 64, 66) that reported an association between placental abruption and nonfatal CHD or stroke morbidity. The pooled HR was 1.35 (95% CI: 1.23, 1.49; I² = 22%; τ² = 0.00) (Web Figure 5).

DISCUSSION

Women that suffer from placental abruption are at about twice the risk of developing CHD and stroke compared with women without abruption. These risks appear to vary based on the number of prior abruptions. The observed pooled estimates of risk of CHD and stroke associated with placental abruption are underestimates and likely biased by exposure misclassification and unmeasured confounding. This meta-analysis provides more granular analysis of the association between placental abruption and CHD and stroke risks than the study by Grandi et al. (39) did. In particular, this meta-analysis shows increased risks of CHD mortality and stroke mortality in relation to placental abruption, as well as increased risks of nonfatal complications. There is also evidence of increasing risks of CHD and stroke mortality with increasing number of placental abruptions across successive pregnancies. Correction of risk estimates for exposure misclassification and unmeasured confounding biases resulted in many of these associations being stronger.

There are at least 3 potential factors that may shed light on the pathophysiology of the mechanisms linking placental abruption to CHD and stroke. First, it has been suggested that pregnancy complications may simply unmask preexisting CHD risk (67). Women who suffer an ischemic insult to the maternal-fetal interface have a surge in oxidative stress leading to vascular damage and thrombosis (16, 17). Defective trophoblast invasion leading to reduced uteroplacental blood flow also induces “placental stress” (68). This process, in turn, leads to endothelial cell dysfunction and an imbalance in angiogenic factors (69), leading to placental abruption (70, 71). It is not unlike the endothelial damage and thrombosis that precipitate CHD and stroke events. Disruption of the hemostatic and vascular systems manifesting as uteroplacental ischemia and hemorrhage may underscore shared pathways between placental abruption and CHD and stroke complications (48).

Second, women with placental abruption may share similarities with those who develop CHD or stroke later in life with regard to clinical risk factors (e.g., smoking or drug use), metabolic changes including abnormal lipid profiles (11, 72), or inflammatory mediators, suggesting a shared etiology between these conditions (12). Third, given the etiologic similarity in vascular and thrombotic factors between placental abruption and CHD and stroke, it is plausible that genetic imprints and their interactions with other risk factors are likely to shape the risks of both placental abruption and CHD and stroke later in life.

The increased hemodynamic and vascular demands of pregnancy place stress on the cardiovascular system. As women elect to postpone pregnancies, the incidence and prevalence risks of obstetrical complications are likely to increase, which in turn places greater demands on the cardiovascular system and can consequently lead to increased CHD and stroke risks. Whatever the mechanisms through which these associations manifest, they suggest that placental abruption unmasks the risks of CHD and both ischemic and hemorrhagic strokes, similar to what is seen for other pregnancy characteristics and obstetrical complications (73).

Study limitations

The risk of CVD in individual studies is dependent on the length of the follow-up period. Compared with studies with extended periods of follow-up, those with a relatively shorter follow-up can be regarded as an “immature” cohort where eventual CHD and stroke risks are underestimated. Studies of placental abruption, in general, should consider how selective fertility—that is, the couple’s ability to achieve the desired family size (74)—might shape the risks of CHD and stroke. If selective fertility were indeed operating in these populations, then the cumulative risks of CHD and stroke outcomes would have been underestimated, and the reported risks are likely conservative. Although even a single episode of abruption is associated with increased risks of CHD and stroke, the lack of an observed dose-response association between the number of abruptions across pregnancies and nonfatal CHD and stroke complications (Web Table 2) may have been a consequence of selective fertility.

This meta-analysis shows substantial heterogeneity in the pooled effect measures for most outcomes examined. Apparent reasons for this heterogeneity may be several, but importantly, variations in exposure and outcome definitions across studies, as well as the selection of the population, are chief reasons. There is no universally accepted definition for placental abruption, nor can the complication be predicted. Equally, there are a number of subtypes of CHD; while some studies defined the outcome as “any CVD,” others included specific types. Variations in the populations included in each study may have also contributed to this heterogeneity. For example, some studies included only first births (which typically will have lower rates of abruption (75, 76)), were restricted to parous women diagnosed with systemic lupus erythematosus (59, 60), or were among women with a first CVD hospitalization (66). Given the high degree of heterogeneity in some of the pooled estimates of associations, the findings must be cautiously interpreted.

Conclusions

This meta-analysis shows robust associations between placental abruption and the risks of CHD and stroke. The impact of exposure misclassification and unmeasured confounding appears to have substantial effects on the reported associations. Although placental abruption is seldom preventable, efforts to ameliorate risks in subsequent pregnancies (such as optimal pregnancy spacing and adopting healthy behavioral and lifestyle, including smoking and alcohol cessation) remain important. Because prediction and prevention of CHD and stroke remains a global priority, postpartum follow-up of women after placental abruption may be beneficial.