Hypertensive Disorders of Pregnancy and Risk of Cardiovascular Disease-Related Morbidity and Mortality: A Systematic Review and Meta-Analysis

Approximately 15% of parous women have at least 1 pregnancy complicated by a hypertensive disorder [1]. As reported, hypertensive disorders of pregnancy (HDP) are associated with maternal and fetal complications [2-7]. Given that approximately 30,000 maternal and 500,000 perinatal deaths annually can be attributed to HDP, these disorders have been regarded as one of the major contributors to maternal and fetal mortality and morbidity globally [8-10].

Historically, HDP were considered to be self-limiting and have little long-term effect on health since the blood pressure of most of the women affected would return to normal within 12 weeks after delivery [11]. However, residual anomalies and increased cardiovascular disease (CVD) risk factors (e.g., hypertension and diabetes) in women with a history of HDP are likely to render these people becoming increasingly at risk of developing CVD [12, 13]. Interest in testing this hypothesis has grown rapidly in the past years. However, no consensus has been reached even today. Several studies have reported a significantly higher risk of composite cardiovascular events in women with a history of HDP compared with a reference group [14, 15], but others have not found a significant association between HDP and a long-term risk of developing CVD [16, 17]. Notably, even in studies providing a supportive evidence, effect sizes varied greatly. Given the increasing prevalence of HDP, it is necessary to summarize the existing evidence of the association of HDP with long-term CVD-related morbidity and mortality through integrated approaches.

To this end, our objective here was to perform an updated meta-analysis regarding the association between HDP and the risk of developing CVD-related morbidity and mortality. Our results may lead to a better understanding of CVD risk in women with HDP, which will help to guide future management and contribute to guidelines for clinicians.

This systematic review was conducted following the proposed Systematic Reviews and Meta-Analyses (PRISMA) [18] and the Meta-Analysis of Observational Studies in Epidemiology (MOOSE) [19] reporting guidelines.

Two authors independently identified observational studies published in English prior to 8 July 2019 and reported data on cardiovascular outcomes in women with and without HDP. PubMed, Embase, and Web of Science were systematically searched. The search terms used in combination were: (1) pregnancy, pregnant, gestational, gestation, hypertensive, hypertension, pre-eclampsia (PE), preeclampsia, eclampsia, edema-proteinuria-hypertension gestosis, toxemia of pregnancy, and hemolysis-elevated liver enzymes-low platelets (HELLP) syndrome; and (2) cardiovascular diseases, CVD, coronary disease, coronary thrombosis, coronary stenosis, coronary restenosis, coronary artery disease, coronary heart disease, acute coronary syndrome, ischemic heart disease, myocardial ischemia, myocardial infarction; heart failure, cardiac failure, left ventricular systolic dysfunction, cardiomyopathy, peripheral vascular disease, pulmonary embolism, venous thromboembolism, deep vein thrombosis, cerebrovascular disorders, cerebrovascular disease, cerebrovascular accident, stroke, death, and mortality. In addition, the reviewers manually searched the reference lists of the articles selected to identify any relevant studies missed in the initial search.

The exposure of interest was any type of HDP, including gestational hypertension (GH), PE, eclampsia, HELLP syndrome, and chronic hypertension with superimposed PE. Primary outcomes included any type of CVD and CVD-related mortality. CVD was defined as any of the following events: coronary artery heart disease, heart failure, peripheral vascular disease, and stroke. Coronary artery heart disease included coronary heart disease, ischemic heart disease, myocardial infarction, and coronary artery disease. Peripheral vascular disease was defined as peripheral arterial disease, deep vein thrombosis, and any thromboembolic events. Stroke was defined as any composite stroke, stroke unspecified, and stroke/transient ischemic attack. The outcome CVD-related mortality was defined as death due to any CVD. Although hypertension is outside the range of CVD, we included hypertension as the secondary outcome given its importance as a risk factor for CVD.

At the title and abstract screening stage, we purposely broadened the inclusion criteria to obtain all relevant studies. Studies were considered for inclusion if they were published in English and reported on CVD, CVD-related mortality, and hypertension among women with HDP. The full texts of all selected studies were then reviewed. Studies were included if they (1) were original articles, (2) included at least 2 groups (1 with HDP and 1 without HDP), (3) provided sufficient information to allow for accurate risk estimates to be calculated, and (4) the full-length article was available. Conversely, studies were excluded if they (1) were review papers, conference abstracts, case reports, experimental studies, or qualitative studies, (2) had incomplete or unclear data, or (3) were duplicate publications. There was no restriction based on cohort type, study design, or duration of follow-up. When there was >1 study that involved the same population of HDP patients, only the most recent published or comprehensive one was included.

Two reviewers independently extracted and evaluated the data for each included article using a self-designed data abstraction form. Disagreements were resolved through discussion or consultation with a third reviewer when consensus could not be achieved. The following data were extracted: first author, year of publication, geographic region, study period, sample source (population- or hospital-based), type of exposure (HDP, GH, PE, eclampsia, HELLP syndrome, or chronic hypertension with superimposed PE), any adjustments or matches made, sample size, mean age, reported outcomes, and risk estimates with 95% confidence intervals (CIs; the adjusted ones were collected when available). In cohort studies, the duration of follow-up was also extracted.

The methodological quality of studies was assessed using the Newcastle-Ottawa Scale (NOS) for quality assessment of cohort and case-control studies by 2 researchers (R.W. and T.W.) independently. The NOS was composed of 8 items. It ranged from 1–9 stars and assessed the quality of each study based on 3 modules: selection, comparability, and outcome (cohort studies) or exposure (case-control studies). A final median score of ≥6 was regarded as high quality.

The risk ratio (RR) was used as the measure of the association between the history of HDP and a risk of any CVD, coronary artery heart disease, heart failure, peripheral vascular disease, stroke, CVD-related mortality, and hypertension. The hazard ratios (HRs) and odds ratios (ORs) were directly considered as RRs.

RevMan v5.3 (Nordic Cochrane Center) and Comprehensive Meta-Analysis v2.2 were used to perform all analyses in this study. The Cochran Q test and the I² statistic were used to assess the heterogeneity of RRs across studies. The Cochran Q test was used to evaluate whether the variation across studies was compatible with chance, and p < 0.1 was considered to indicate significant heterogeneity. The I² statistic was a quantitative indicator used to evaluate the percentage of total variance in risk estimates due to statistical heterogeneity rather than chance, or sampling error (I² >75%, high heterogeneity; 51–75%, substantial heterogeneity; 26–50%, moderate heterogeneity; ≤25%, low heterogeneity). The pooled RRs and 95% CIs were calculated using random-effects meta-analyses. To explore possible sources of heterogeneity, subgroup analyses were performed according to geographic region (Asia vs. Europe vs. North America), sample source (hospital-based vs. population-based), type of exposure (GH vs. PE), whether the confounders were adjusted (adjusted vs. unadjusted), and median/mean follow-up time (≤10 vs. 10–20 vs. ≥20 years). Then, a Q test for heterogeneity was used to compare the subgroup differences under the random-effects model (here, Q would be distributed as χ² with df = 1 and p < 0.05 indicating statistically significant differences) [20].

Sensitivity analysis was conducted to examine the influence of individual studies on the overall RRs by repeating the meta-analysis after the exclusion of each included study. Sensitivity analyses were also performed to examine the influence of case-control studies on the overall RRs. Publication bias was evaluated using Begg’s test (p < 0.05 indicated statistically significant differences).

In total, 24,395 unique citations were identified after an initial search. Of these, 24,290 were excluded after screening titles and abstracts, mainly because they were duplications, reviews, or not related to our study (Fig. 1). Then, the full texts of 105 articles were reviewed; a total of 73 studies [12, 14-17, 21-88] were considered eligible and were included in this meta-analysis.

Of the 73 observational studies included, 66 were cohort studies and 7 were case-control studies. The characteristics of the 66 cohort studies with a total sample size >13 million are shown in Table 1. There was a large variance in sample sizes (40–2,066,230). The year of publication was between 1995 and 2019, within which nearly 57.6% (38/66 studies) were published in 2010–2019. Forty studies (60.6%) were conducted in Europe, 14 (21.2%) in North America, 9 (13.6%) in Asia, 2 (3.0%) in South America, and 1 (1.5%) in Oceania. Studies reported a mean or median age of participants in the range of 25–75 years, whereas median or mean follow-ups ranged from 0 to 42 years. For sample sources, 44 studies (66.7%) were population-based, and 22 (33.3%) were hospital-based. Seventeen studies (25.8%) did not adjust for any confounder when estimating the risk of CVD associated with HDP, but the remainder (74.2%) adjusted or matched for age, smoking, or other potential confounders.

The characteristics of the 7 case-control studies with sample sizes ranging from 158 to 3,678 women and published between 2004 and 2019, are shown in Table 2. Study populations were selected from Europe (n = 3), North America (n = 3), and Asia (n = 1). The age of the participants ranged from 15 to 66 years. For sample sources, 2 studies were population-based and 5 were hospital-based. Adjustments and matches were made in 6 studies.

The quality of the 73 studies included here was evaluated using the NOS as shown in online supplementary Tables 1 and 2 (for all online suppl. material, see www.karger.com/doi/10.1159/000508036). The quality was generally good, as 66 studies (90.4%) got 6–9 stars. The long-term outcomes reported in the 73 studies were: 43 cases of any CVD, 25 of coronary artery heart disease, 10 of heart failure, 8 of peripheral vascular disease, 14 of stroke, 17 of CVD-related mortality, and 36 of hypertension. Several studies reported more than 1 outcome of interest.

The risk estimate of any CVD associated with HDP is summarized in Figure 2. The RRs for the association reported by included studies ranged from 0.88 to 13.18. Meta-analytic pooling of these risk estimates yielded a summary RR of 1.80 (95% CI 1.67–1.94), with substantial heterogeneity (I² = 92%, p < 0.001). Begg’s test did not indicate a potential publication bias (Z = 0.283, p = 0.778).

The relationships between HDP and coronary artery heart disease, heart failure, peripheral vascular disease, and stroke are summarized in online supplementary Figures 1–4. The overall RR in relation to HDP was 1.66 (95% CI 1.49–1.84; I² = 86%, p < 0.001) for coronary artery heart disease, 2.87 (95% CI 2.14–3.85; I² = 90%, p < 0.001) for heart failure, 1.60 (95% CI 1.29–2.00; I² = 76%, p < 0.001) for peripheral vascular disease, and 1.72 (95% CI 1.50–1.97; I² = 81%, p < 0.001) for stroke. No evidence of publication bias was detected by using Begg’s test (coronary artery heart disease: Z = 0.701, p = 0.484; heart failure: Z = 0.626, p = 0.531; peripheral vascular disease: Z = 1.485, p = 0.138; stroke: Z = 1.478, p = 0.139).

Subgroup analyses for risk of any CVD, coronary artery heart disease, heart failure, peripheral vascular disease, and stroke are shown in online supplementary Table 3. After these analyses were conducted, the variables including geographic region, sample source, type of exposure, and median/mean follow-up time were identified as the relevant heterogeneity moderators for any CVD (χ² range: 6.79–22.12; all p < 0.05). Notably, in hospital-based studies, the risk estimate of any CVD (2.54, 95% CI 2.16–2.99) was higher than in studies targeted at the general population (1.66, 95% CI 1.54–1.78). When stratified according to the type of exposure, the pooled risk estimate of any CVD in women with a history with GH (1.64, 95% CI 1.43–1.89) was lower than that in women with a history of PE (2.07, 95% CI 1.86–2.30). The risk of developing any CVD associated with a history of HDP was higher in studies with a follow-up time ≤10 years (2.64, 95% CI 2.15–3.25) than in studies with a follow-up time of 10–20 years (1.59, 95% CI 1.43–1.75) or a follow-up time ≥20 years (1.68, 95% CI 1.62–1.75). For coronary artery heart disease, variables including geographic region, type of exposure, and median/mean follow-up time partially explained the between-study heterogeneity (χ² range: 7.00–23.38; all p < 0.05). The 2 variables geographic region and median/mean follow-up time also partially explained the between-study heterogeneity for heart failure and stroke (χ² range: 13.13–34.62; all p < 0.001). Only the sample source was identified as the heterogeneity moderator for peripheral vascular disease (χ² = 9.38, p = 0.002).

The risk estimate of CVD-related mortality associated with HDP is summarized in Figure 3. The RR for the association reported in the included studies was in the range of 1.28–5.04. Meta-analytic pooling of these risk estimates yielded a summary RR of 1.78 (95% CI 1.58–2.00) with substantial heterogeneity (I² = 77%, p < 0.001). Begg’s test did not indicate a potential publication bias (Z= 0.659, p = 0.510).

Subgroup analyses of risk of CVD-related mortality appear in online supplementary Table 3. After these analyses, only geographic region was identified as the relevant heterogeneity moderator for CVD-related mortality (χ² = 5.61; p = 0.018). The risk estimate of CVD-related mortality was higher in studies conducted in Asia (2.63, 95% CI 1.90–3.65) than in those conducted in Europe (1.72, 95% CI 1.51–1.96).

Online supplementary Figure 5 shows the result from the random-effects model combining the RRs for hypertension in women with a history of HDP. The RRs for the association varied from 1.32 to 23.74. Overall, women with a history of HDP compared to the reference group had a significantly higher risk of developing hypertension (RR 3.16 [95% CI 2.74–3.64]). There was high heterogeneity across studies (I² = 98%, p < 0.001). Begg’s test also did not indicate a potential publication bias (Z = 1.716, p = 0.086). After subgroup analyses, the variables including sample source and median/mean follow-up time were identified as the relevant heterogeneity moderators for hypertension (χ² = 9.37, 12.68; all p = 0.002; online suppl. Table 3).

Sensitivity analyses were performed to examine the influence of individual studies on the overall risk estimate for CVD, CVD-related mortality, and hypertension associated with HDP (online suppl. Table 4). Exclusion of any single study did not materially alter the overall combined RR, which ranged from 1.76 (95% CI 1.64–1.89) to 1.85 (1.69–2.03) for any CVD, 1.60 (1.44–1.77) to 1.71 (1.53–1.90) for coronary artery heart disease, 2.59 (1.95–3.44) to 3.16 (2.29–4.37) for heart failure, 1.48 (1.24–1.77) to 1.72 (1.44–2.06) for peripheral vascular disease, 1.60 (1.41–1.81) to 1.87 (1.54–2.28) for stroke, 1.72 (1.54–1.92) to 1.83 (1.62–2.08) for CVD-related mortality, and 3.05 (2.64–3.52) to 3.26 (2.82–3.77) for hypertension. Furthermore, sensitivity analyses were conducted to examine the influence of case-control studies on overall RR (online suppl. Table 5). Results showed that exclusion of case-control studies did not alter the risk estimates for CVD and CVD-related mortality.

In this systematic review and meta-analysis, 73 studies involving >13 million women were included. We showed an association of HDP with the long-term risk of developing CVD, CVD-related mortality, and hypertension. The mean risk estimates were 1.80 for CVD, 1.66 for coronary artery heart disease, 2.87 for heart failure, 1.60 for peripheral vascular disease, 1.72 for stroke, 1.78 for CVD-related mortality, and 3.16 for hypertension, when compared with participants without a history of HDP. To the best of our knowledge, this is the most comprehensive meta-analysis to assess the association between HDP and risk of CVD. Our findings supply helpful information to both clinicians and women with HDP and can contribute to the guidelines for the clinical management of HDP in the future.

Five systematic reviews and meta-analysis [89-93] have been previously performed to investigate the association between PE and future CVD, death due to CVD, and hypertension. A recent review, performed by Grandi et al. [94], examined the risk of CVD associated with GH and PE. Although the results of our current meta-analysis are generally in line with these previous reviews, we also provide a contemporary synthesis of the evidence for all HDP with a rigorous assessment of study quality. Given that different types of specific CVD may have different etiologies, it may be conceivable that the risk of developing each specific CVD is inconsistent. Therefore, we combined the risk of specific CVDs in women with HDP, such as coronary artery heart disease, heart failure, etc. Moreover, we synthesized the literature on peripheral vascular disease, which has not been previously performed. By comparison, our risk estimates of CVD associated with PE were slightly lower than those in a meta-analysis published in 2017 [91], possibly due to the inconsistencies in the included studies. The meta-analysis published in 2017 included 1 study [95] which was not included in our meta-analysis due to the overlap of the study population. Notably, after considering the effects of hypertension in the future coronary heart disease outcome, the link between preeclampsia and future coronary heart disease was no longer statistically significant in 2017. However, the abovementioned meta-analysis was published earlier than 9 large cohort studies from Norway [77, 78], the UK [21, 79, 83], Denmark [23], the USA [85], Canada [87], and Australia [22], all of which were included in our study.

In our study, the increased risk for all CVD and hypertension was greater at ≤10 years compared with at >10 years postpartum. This may be because of a higher absolute risk in the control group during longer follow-up periods (i.e., >10 years). Therefore, the reduction in the RR may be a product of the higher baseline risk. Furthermore, the effect is exacerbated by the small sample size and number of events in the control group reported [39, 73, 88]. Historically, it was thought that GH is a milder form of PE; our study showed that the risk of developing any CVD in GH was lower than that in PE. However, it is only significantly lower in the outcome of any CVD and coronary artery heart disease. Among other outcomes such as heart failure, stroke, hypertension, and so on, the difference was not statistically significant in the GH and PE groups.

The strength of this study is the large sample size; it consisted of contemporary studies on >13 million study participants. This helps to enhance statistical power providing more reliable and precise risk estimates. At the beginning of the study, a comprehensive search strategy was used to identify relevant studies. Moreover, the process including literature retrieval and screening, and data extraction were performed by 2 reviewers independently. Finally, most of the studies were designed to examine future CVDs as their main outcome (n = 43).

One potential limitation of this meta-analysis was the significant heterogeneity across studies for the association between HDP and the risk of developing CVD. This is not surprising, given the different study designs and characteristics of study populations. Fortunately, our subgroup analyses identified several variables associated with the between-study heterogeneity, including geographic region, sample source, type of exposure, and follow-up time. In addition, the variability and complexity of HDP might have contributed to the heterogeneity, given the underlying mechanisms involved in the association between HDP and CVD. However, we were unable to obtain adequate information to test the hypothesis. A second limitation was that significant unmeasured confounding factors may have contributed to the observed association between HDP and CVD. Although most of the studies included here attempted to control for some potential confounding factors, only a few adequately controlled all the conventional cardiovascular risk factors reported in previous studies, such as age, body mass index, gestational diabetes, blood pressure, cholesterol, and family history of CVD. Finally, the duration of follow-up in several of the included studies may not have been sufficient to capture all cardiovascular-related events of interest [15, 30, 50, 61]. Last, but not least, according to the previously published studies [91, 93], we used both HRs and ORs to approximate the RRs. Given the lack of information on incidence rates from case-control studies, however, we could not determine the RR from the pooled data. From individual cohort studies, it was suggested that the risk of CVD incidence was low [77-78].

This study demonstrated that all women who present with any of the HDP subtypes are at significant risk of future CVD, CVD-related mortality, and hypertension, when compared with women who remain normotensive during their pregnancy. In view of the burden and impact of CVDs on women in our society, we recommend a detailed cost-benefit analysis to determine the postnatal timing for a screening program in this high-risk population.

Our meta-analysis reports an increased risk of CVD, CVD-related mortality, and hypertension in women with HDP compared with the reference population. In keeping with current recommendations, our findings highlight the importance of education and lifestyle modifications to reduce risks. Regular monitoring of cardiovascular risk factors in women with a history of HDP is also strongly advised.

The authors would like to thank the editors and reviewers for their suggestions and all colleagues working in the Department of Epidemiology and Health Statistics, Xiangya School of Public Health, Central South University, Hunan, China.

This systematic review was conducted following the proposed Systematic Reviews and Meta-Analyses (PRISMA) [18] and the Meta-Analysis of Observational Studies in Epidemiology (MOOSE) [19] reporting guidelines.

The authors have no conflicts of interest to declare.

The research was supported by the Major Research and Development Project in Hunan Province (2018SK2062) and the Fundamental Research Funds for the Central Universities of Central South University.

L.C., R.W., and T.W. contributed to the conception or design of the work. R.W., T.W., R.G., C.Y., Y.C., X.L., and D.X. contributed to the acquisition, analysis, or interpretation of data for the work. R.W. and T.W. drafted the manuscript. L.C. critically revised the manuscript. All authors gave their final approval and agreed to be accountable for all aspects of work ensuring integrity and accuracy.

L.C. and T.W. are both corresponding authors.