Article Text
Abstract
Background and objective Hypertensive disorders of pregnancy (HDPs) affect 5%–10% of pregnancies and have been associated with excess maternal cardiovascular disease (CVD) risk. The primary aim of this study was to reliably estimate absolute and relative risks of CVD after HDP.
Methods A retrospective cohort of women who had singleton pregnancies in New South Wales, Australia, between 2002 and 2016 and identified using linked population health administrative databases. The primary exposure was new-onset HDP (pre-eclampsia/eclampsia and gestational hypertension), and the endpoint was hospitalisation or death due to ischaemic or hypertensive heart disease, or stroke. Kaplan-Meier analysis estimated risks among mothers following their first birth, and multivariable time-dependent Cox regression estimated the association between HDP and CVD.
Results Among 528 106 women, 10.3% experienced HDP in their first pregnancy. The 10-year estimated risk of CVD was 2.1 per 1000 if no HDP and 5.5 per 1000 following HDP. Adjusting for demographics, gestational diabetes, small for gestational age and preterm birth, we found that there was an interaction between smoking and HDP, and a larger effect of early-onset (<34 weeks) HDP, compared with late-onset HDP. The HR for women with early-onset HDP who did not smoke was 4.90 (95% CI 3.00 to 7.80) and the HR for those who did smoke was 23.5 (95% CI 13.5 to 40.5), each compared with women without HDP who did not smoke.
Conclusion In this nationally representative Australian cohort, HDP, especially early onset, conferred a clear increase in the risk of CVD, with amplification by smoking. Targeted preventive health, during and after pregnancy, could prevent a substantial burden of CVD among childbearing women.
- pregnancy
- hypertension
- smoking cessation
- diabetes
Statistics from Altmetric.com
Introduction
Hypertensive disorders of pregnancy (HDPs) affect 5%–10% of pregnancies worldwide.1 Beyond the high rates of acute maternal and foetal mortality and morbidity,2 HDPs are independently associated with the development of postpartum maternal cardiovascular disease (CVD) and cerebrovascular disease.3–7
In Australia, over 7% of pregnancies are affected by HDP with around 3% complicated by pre-eclampsia.8 Few large, population-based studies have been undertaken to understand the CVD risk for Australian women associated with HDP and the public health ramifications.9 Additionally, while a number of studies have controlled for the presence of traditional risk factors for CVD in women with HDP, finding that HDP is an independent risk factor for CVD, few studies have examined the interaction between HDP, traditional CVD risk factors and other pregnancy complications with respect to subsequent CVD risk.10
A greater understanding of these interactions would have important public health implications and may necessitate updates to clinical guidelines on cardiovascular follow-up and management following a pregnancy complicated by hypertension, pre-eclampsia or eclampsia. Currently, there is no systematic, broadly accepted follow-up or postpartum pathway for these at-risk women in Australia, Europe or the USA.1
Accordingly, we sought to estimate the absolute and relative risks of a maternal cardiovascular event in a nationally representative cohort of women. We further aimed to define the differential effect of early-onset versus late-onset HDP and the interplay between pregnancy-associated risk, other pregnancy factors and smoking, a traditional cardiovascular risk factor which is universally discouraged in pregnancy.
Methods
Cohort and data linkage
A retrospective cohort study using linked population health administrative datasets was undertaken in New South Wales (NSW), Australia. This included women aged 12–55 years who were NSW residents at their first birth with a complete birth history in NSW of singleton births between 1 January 2002 and 31 December 2016. NSW has approximately 95 000 confinements per year, representing 32% of all births in Australia.
The analysis used the Maternal and Child Health Register (MCHR). The MCHR is a linked, deidentified dataset containing information for children and young people aged less than 16 years and women aged 12–55 years and includes linked records from the NSW Perinatal Data Collection (PDC), NSW Admitted Patient Data Collection (APDC), NSW Emergency Department Data Collection, Australian Coordinating Registry Cause of Death Unit Record File (COD URF) and the NSW Registry of Births, Deaths and Marriages death registrations (RBDM). Record linkage for the MCHR was performed by the NSW Centre for Health Record Linkage (CHeReL).
The NSW PDC is a population-based data collection that covers all births in NSW public and private hospitals, as well as home births in NSW. It encompasses all births, including stillbirths of at least 20 weeks ’ gestation or at least 400 g birth weight, and all live births, and includes demographic information on the mother and baby and information on the mother’s health, her pregnancy, labour and birth, and the health of the baby. The NSW APDC comprises a census of all admitted patient services provided by NSW public hospitals, public psychiatric hospitals, public multipurpose services, private hospitals and private day procedure centres. It covers demographic information and information on diagnoses, procedure and hospital care for every hospital separation in NSW. Diagnosis information is coded according to the International Classification of Diseases and Related Health Problems, Australian Modification (ICD-10-AM).11 The COD URF includes death registration information pertaining to all deaths occurring in NSW and includes demographic information and particulars of the death, including cause of death and place of death, which are recorded either through the death registration process or by the coronial information system. Cause of death information is coded according to the ICD-10. The RBDM death registration data cover all deaths occurring in NSW. Demographic information and particulars of the death, including cause and place of death, are recorded on the death registration. The CHeReL uses probabilistic record linkage software, with procedures designed to ensure fewer than 5/1000 false-positive links and 5/1000 false-negative links.12
All linked records for a woman were excluded where there was missing covariate data, inconsistent date information, no linked birth hospitalisation record, or a study outcome was recorded on the first birth hospitalisation record.
Primary exposure
HDP was defined as pre-eclampsia/eclampsia or gestational hypertension recorded on a PDC record or linked birth hospitalisation record for the mother (ICD-10-AM O11, O13-16).11 HDP was further categorised into early-onset (birth≤34 weeks’ gestation) or late-onset HDP. Pre-eclampsia is defined as new onset of hypertension after 20 weeks’ gestation with evidence of proteinuria and/or acute kidney injury, liver dysfunction, neurological involvement, thrombocyotopaenia, haemolysis or foetal growth restriction.12 Eclampsia is a severe form of pre-eclampsia with maternal seizure. Gestational hypertension is defined as de novo hypertension after 20 weeks’ gestation without any of the additonal features seen in pre-eclampsia. Chronic hypertension, defined as hypertension known prior to pregnancy or hypertension that manifests before 20 weeks’ gestation,13 was not an exposure for this analysis.
Covariates
The covariates of clinical interest in this study were age, small for gestational age (SGA, birthweight<10th percentile for Australian reference for sex and gestational age,14 smoking during pregnancy, gestational diabetes mellitus (GDM) (PDC record or linked birth hospitalisation with diagnosis O24), non-hypertensive preterm birth (<37 weeks), social status (Index of Relative Socio-Economic Advantage and Disadvantage (Socio-Economic Indexes for Areas (SEIFA)) score, based on postcode of residence at first birth,15 number of previous pregnancies of at least 20 weeks’ gestation and number of HDP exposed pregnancies.
Outcome
The outcome of interest was hospitalisation or death due to ischaemic heart disease (ICD-10-AM32/ICD-1033 I20-I25), stroke (ICD-10-AM32 ICD-1033 I61, I62.9, I63, I64 where not also I62.0, I62.1, S06.5 and S06.6) or hypertensive heart disease (I11–I13).
Statistical analysis
Following women from their first birth only, the cumulative risk of CVD hospitalisation (presented as a rate per 1000 women) was estimated from Kaplan-Meier survival analysis, following women from their first birth. HDP status, as well as other predictive variables, was as recorded at this birth.
Adjusted relative hazards for the association between HDP and CVD were estimated from a multivariable time-dependent Cox regression using information recorded at all births. Once a mother was exposed to any HDP, they were considered exposed for all future time points. Once exposed to early-onset HDP, they were considered exposed to early-onset HDP for all future time points. Similarly, for exposures to SGA, smoking, non-hypertensive preterm birth and GDM. Age (in single years) was modelled using a restricted cubic spline to allow for a potential non-linear relationship between age and relative risk of CVD (see online supplementary figure 2). Number of births was treated as a categorical variable due to the highly skewed distribution; three or more births were grouped into a single category. HDP, SGA, smoking, gestational diabetes, non-hypertensive preterm birth and SEIFA (split at the second lowest quintile) were taken as binary variables. To develop a parsimonious prediction model, unadjusted analyses were performed for the exposure and all covariates. Variables significant at p=0.25 were considered for the multivariable model. All variables were included in a preliminary multivariable model. Following an initial deletion of covariates, first-order interactions of clinical interest (between HDP and smoking, HDP and gestational diabetes, and smoking and gestational diabetes) were considered and included. Variables were deleted using backwards elimination16 at p=0.05 (see online supplementary table 2). The final model included age, gestational diabetes, preterm non-hypertensive birth, SEIFA, and an interaction between HDP onset and smoking.
Supplemental material
To obtain a picture of the joint effects of HDP, GDM and preterm birth in the context of smoking status, coefficients from the final multivariable Cox model were added to estimate the relative risks of these combinations for individuals.
Data preparation was performed using SAS V.9.3. Statistical analyses were performed using the survival and rms packages in R. Figures were created using ggplot217 and STATA V.15.
Patient and public involvement
Neither patients nor the public were involved in this retrospective analysis.
Results
Baseline characteristics
Of 544 695 women in the cohort, 528 106 (97%) were included in the analyses. Among the included women, 54 323 (10.3%) experienced HDP in their first pregnancy. The demographic characteristics of these women at first birth are outlined in table 1. Women who experienced HDP were slightly older at first birth (28.7 vs 28.5 years), more likely to have GDM (9.3% vs 7.5%) and to deliver an SGA baby (15.8% vs 12.4%). Those with HDP were also more likely to be born in Australia (74.5% vs 64.3%) and to be reported as Aboriginal or Torres Strait Islander (3.6% vs 2.9%). Conversely, women with HDP were less likely to be current smokers (8.1% vs 9.4%).
Estimated risk of CVD following a first birth
Overall, 69.9% of women had a follow-up of 10 years or more following their first birth. The median follow-up was 7.0 years (interquartile interval: 3.4–10.8) in those without HDP and 7.3 years (3.6–11.2) in those with a HDP history, with no difference between those with early-onset or late-onset disease. Altogether, 925 (0.2%) of women had a cardiovascular event: 26 among those with an early HDP (1.1%), 196 among those with a late HDP (0.4%) and 703 among those with no HDP (0.1%).
The estimated 10-year risk of a major cardiovascular event was 5.5 per 1000 women in those with HDP and 2.1 per 1000 in those without (table 2). Maternal age increased the risk of a cardiovascular event in both groups, but to a greater extent in those with HDP. In women aged 35 years and over, the risk of a cardiovascular event was 10.4 per 1000 in the HDP group vs 4.0 per 1000 in the non-HDP group. Smoking tripled the risk of a cardiovascular event in those with HDP as compared with those with a normotensive pregnancy (9.0 per 1000 patient years risk vs 3.0 per 1000). The presence of both diabetes and HDP resulted in a more than doubling of the risk of a fatal or non-fatal cardiovascular event as compared with diabetes alone (10.9 vs 3.6 per 1000 risk). Similarly, the coexistence of HDP and an SGA fetus more than doubled the risk of a cardiovascular event (6.2 vs 2.3 per 1000). Early-onset HDP (≤34 weeks’ gestation) was the most powerful risk factor identified, with a 15.9 per 1000 risk of fatal or non-fatal cardiovascular event at 10 years (figures 1 and 2 and online supplementary figures 3–6).
Association between HDP in any pregnancy and a major cardiovascular event
Of the 528 106 women in the study cohort, 62 970 (11.4%) ever had HDP and 2 927 ever had early-onset HDP (0.6%) (see online supplementary table 3). Women who experienced an HDP had more births in total during study follow-up than those who had not. SEIFA scores at the time of first pregnancy were similar between those who did and did not experience HDP. Rates of HDP were higher among older (17.5% among women aged over 45 years) versus younger mothers (13% among women aged 20–24) (see online supplementary table 3).
In unadjusted analyses, the presence and timing of onset of HDP, number of exposed births, greater age, smoking, SGA, GDM, preterm birth and relatively high deprivation were all associated with increased risk of a cardiovascular event, but there was no association with number of prior births (p<0.05) (see online supplementary table 4). All the pregnancy complications incurred a greater relative risk than smoking during pregnancy, except SGA, while early-onset of HDP outweighed the risk from late-onset considerably.
In the adjusted model, timing of onset of HDP, greater age, smoking, GDM, preterm birth and high deprivation were independently associated with increased risk of CVD. The interaction between smoking and time of onset of HDP was also independently associated with CVD (figure 3). After adjusting for smoking, age, GDM, preterm birth and deprivation status, compared with non-smokers with no HDP, the risk of CVD for non-smoking women with late-onset HDP was 2.5 (95% CI 2.3 to 2.9) times as high, and that for early onset was 4.9 (95% CI 3.0 to 7.8) times as high. Compared with the same reference group, for women who smoked, late-onset HDP multiplied the risk by 4 (95% CI 2.8 to 5.7) and late onset by 23.5 (95% CI 13.5 to 40.8).
Among those not smoking in pregnancy, when HDP was considered in tandem with other pregnancy complications, late onset together with GDM incurred an estimated relative risk of 3.8 (95% CI 3.0 to 4.8) and early onset of 7.4 (95% CI 4.5 to 12.2), compared with no HDP, GDM or preterm birth (figure 4). Similarly, early-onset HDP with preterm birth had an estimated relative risk of 5.2 (95% CI 4.0 to 6.8), and late onset of 10.1 (95% CI 6.1 to 16.8), compared with the same reference. When smoking was added to the mix of positive risk factors, the relative risks, compared with the same reference, were considerably higher, as would be expected from figure 3.
Discussion
In a nationally representative cohort of women, HDP conferred a markedly elevated relative risk of maternal CVD or death to a greater magnitude than that seen with the traditional risk factors evaluated. While estimated absolute risk was low in these young women, the relative risk, as compared with matched women with an unexposed pregnancy, was dramatic: a more than doubling of the risk of a cardiovascular event or death in those with late-onset HDP and almost five times the risk following early-onset disease. In comparison, current smoking alone conferred a less than doubling of the cardiovascular risk. Moreover, there was a clear synergistic and additive effect between HDP and smoking, such that smokers with late-onset HDP had four times the risk of CVD, and smokers with early-onset HDP had more than 23 times the risk of CVD, compared with non-smokers with no HDP. The combination of early-onset HDP, gestational diabetes and smoking resulted in a relative risk of almost 50 for a maternal cardiovascular event.
While only 0.6% of the studied cohort had early-onset HDP, reflective of the broader prevalence of the condition, these women represent an important at-risk group. An estimated absolute risk of 16 per 1000 patient years is consistent with low absolute risk as per American Heart Association guidelines18; however, this group’s absolute risk will significantly increase with advancing age and concomitant risk factors. These data highlight the importance of ensuring systematic postpartum cardiovascular follow-up with a focus on addressing modifiable risk factors, such as smoking.
Women with late-onset HDP cannot be forgotten. These data suggest that late-onset HDP is associated with an independent risk of an adverse cardiovascular event in women, in an order of magnitude greater than smoking. Given that more than 80% of HDP is late onset, these women should be identified as an important research and clinical priority, with an opportunity to alter the cardiovascular trajectory of a large population of women. On a global scale, estimates suggest that >300 million women have previously been exposed to pre-eclampsia and thus are currently at risk of an adverse cardiovascular event.3 Moreover, with the rapidly escalating rates of gestational diabetes worldwide,19 20 we are likely to see an increase in women with both late-onset HDP and gestational diabetes, a group at almost four times the risk of a cardiovascular event. These women are an important focus for public health and primary prevention initiatives.
Previous studies have reported that women who experienced vascular disorders of pregnancy are at increased risk of subsequent CVD, though the pathophysiology linking these conditions, and whether it is a causative association, is less well understood. These studies, however, are heterogenous with respect to the populations studied, follow-up period and the magnitude of the risks identified.21–25 A recent meta-analysis of 22 studies and >6 million women reported that pre-eclampsia was independently associated with heart failure, coronary artery disease, stroke and cardiovascular death. Many of the included studies, however, reported on smaller cohorts with very few delineating between early-onset and late-onset disease.6 This association between HDP and CVD was reinforced in a recent Leon et al UK Pregnancy Cohort study of 1.3 million women.7 Knowledge with respect to the interaction between sex-specific risk factors and traditional risk factors, however, is very limited in these reports.
Previous studies, including a Norwegian linkage study of similar size to the current analysis, report the importance of SGA in addition to pre-eclampsia. The risk of a major cardiovascular event was more than double in women with both SGA and pre-eclampsia.10 While we did identify an increase in absolute risk in women with both HDP and SGA in our analyses, the effect of SGA did not remain separately significant in our multivariable relative risk model. There are several possible explanations for this discrepancy, with the most likely being that the NSW population encompasses all HDP rather than just pre-eclampsia. While the Norwegian study did not specifically separate HDP into early and late onset, a subsidiary analysis of preterm delivery prior to 34 weeks reported a relative risk of a cardiovascular event of 7. This supports our findings with respect to early-onset HDP, although not to the same magnitude as seen in our cohort of NSW women.
From a pathophysiological perspective, data suggest that early-onset and late-onset pre-eclampsia may represent either different phenotypes of the same conditions or different underlying processes.26–28 Early-onset disease appears to be mediated by abnormal placentation with ensuing microthrombi, oxidative stress, inflammation and a systemic vascular response,29 while late-onset disease may represent an unmasking of underlying maternal CVD in predisposed individuals. Thus, it is certainly plausible that they may portend different cardiovascular risk profiles and require different long-term management.
The key strength of this analysis is the population included. Through reporting on all women who gave birth within NSW during well over a decade, we are ensuring a diverse population and results that are broadly applicable. As the data sources capture routinely collected healthcare data, they are not affected by inclusion bias. This study is also unique in its ability to report extensively on the important interactions between pregnancy risk factors and traditional risk factors, which make these data clinically relevant and important. Finally, the ability to report on early-onset versus late-onset HDP provided a valuable opportunity to identify a subgroup at higher risk.
Limitations of this study include that we were unable to report with confidence on the different subgroups of HDP (gestational hypertension, pre-eclampsia and eclampsia) with the available data due to small numbers. When dealing with routinely collected data, there is also the risk of misclassification, particularly of chronic hypertension as a HDP. However, this would lead to underestimation of relative risks. Further, we did not have access to some important covariates, such as body mass index and cholesterol level, which are important predictors of cardiovascular risk.
Conclusion
HDPs are associated with a marked increase in the risk of a maternal fatal or non-fatal cardiovascular event. Those at highest risk are women with early-onset HDP, those who concurrently smoke, have concomitant GDM, or delivery of a preterm or SGA baby. Pregnancy is an ideal opportunity to identify at-risk women and target modifiable risk factors such as smoking. These data suggest that clear guidance on postpartum follow-up for at-risk women needs to be urgently developed and effectively implemented.
Key questions
What is already known on this subject?
Hypertensive disorders of pregnancy (HDPs) have been demonstrated to be associated with an increased risk of maternal cardiovascular disease (CVD). Similarly, gestational diabetes, preterm delivery and bearing small-for-gestational age babies have also been linked with increased CVD in mothers. What is unclear, however, is whether there are interactions between these factors and traditional CVD risk factors such as diabetes and smoking.
What might this study add?
In a large, nationally representative cohort of women followed up for almost a decade, some pregnancy-related cardiovascular risk factors had a stronger association with CVD than smoking, while HDP synergistically interacts with smoking. Among non-smoking women, late- onset HDP had 2.5 times, and early-onset HDP had 4.9 times, the risk of a major cardiovascular event compared with a normotensive pregnancy. Early-onset HDP and smoking together were associated with over 23 times the risk, compared with not smoking and no HDP, which is an exceptionally large effect in the context of CVD.
How might this impact on clinical practice?
Currently, there is no systematic approach to postpartum follow-up in women who have experienced a hypertensive disorder of pregnancy in the UK, USA or Australia nor is adequate attention given to addressing modifiable risk factors in this group. This study highlights that these women present an important focus for public health and primary prevention initiatives.
References
Footnotes
CA and MN are joint first authors.
Contributors All authors were involved in the design of the study. MN carried out the statistical analyses. CA and MW wrote the first complete draft of the paper. All other authors contributed to further drafts and approved the final manuscript. The corresponding author attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted.
Funding MW is supported by the National Health and Medical Research Council of Australia (grant numbers 1080206 and 1149987). AH is supported by a National Health and Medical Research Council (Australia) Early Career Fellowship (1141570).
Competing interests MW has received consulting fees from Amgen, Inc, and Kyowa Kirin Co, Ltd.
Patient and public involvement Patients and/or the public were not involved in the design, conduct, reporting or dissemination plans of this research.
Patient consent for publication Not required.
Ethics approval Ethics approval was obtained from the NSW Population and Health Services Research Ethics Committee (reference number: 2018HRE0902). The study was conducted in accordance with the principles of the Declaration of Helsinki.
Provenance and peer review Not commissioned; internally peer reviewed.
Data availability statement All data relevant to the study are included in the article or uploaded as supplementary information. No additional data are available.