Objective Women with pregnancies complicated by hypertensive disorders of pregnancy (HDP) have increased long-term cardiovascular (CV) risk. We sought to determine if they demonstrate increased short-term CV risk.
Methods Using administrative records, all hospital-based deliveries in Florida from 2004 to 2010 and subsequent readmission to any Florida hospital within 3 years of index delivery were identified. Deliveries and clinical diagnoses were determined using International Classification of Diseases, Ninth Revision, Clinical Modification codes. HDP included pregnancies complicated by gestational hypertension, pre-eclampsia or eclampsia. Outcomes were CV readmission (acute myocardial infarction, stroke or heart failure), non-CV readmission and any readmission within 3 years of delivery excluding subsequent deliveries. Associations were determined using multivariate logistic regression.
Results Among 1 452 926 records from delivering mothers of singleton infants (mean age 27.2±6.2 years; 52% white, 23% African American (AA), 18% Hispanic), there were 4054 CV and 259 252 non-CV readmissions. Women with HDP had higher CV readmission rates (6.4 vs 2.5/1000 deliveries; P<0.001). AA women had higher rates of CV readmission than whites or Hispanics (6.8 vs 1.7 vs 1.0/1000 deliveries, respectively; P<0.001). Women with HDP had higher multivariate risk of CV readmission (OR 2.41; 95% CI 2.08 to 2.80) and any readmission (OR 1.13; 95% CI 1.10 to 1.15). Compared with whites, AA women had higher risk for CV readmission (OR 3.60; 95% CI 3.32 to 3.90) after adjustment for HDP.
Conclusion Women with HDP had twice the risk of CV readmission within 3 years of delivery, with higher rates among AA women. More work is needed to explore preventive strategies for HDP-associated events.
- cardiac risk factors and prevention
- heart disease
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Cardiovascular (CV) disease remains the number one killer of women,1 but current risk prediction models and clinical risk calculators poorly predict a woman’s future CV risk.2–4 Obstetric complications and their relationship to underlying vascular health may be an opportunity for improved CV risk prediction. A single pregnancy complicated by a hypertensive disorder of pregnancy (HDP) has been shown to increase long-term CV risk through population studies.5–7 Accordingly, in 2011 the American Heart Association updated clinical guidelines to include HDP as a major risk factor for CV disease in women.8
The definition of HDP includes gestational hypertension (HTN)—defined as the development of HTN after 20 weeks’ gestation without proteinuria, pre-eclampsia—defined as the development of HTN after 20 weeks’ gestation with proteinuria or evidence of end-organ damage up to 6 weeks postpartum, and eclampsia—defined as the development of grand mal seizures in a woman with pre-eclampsia.9 Current estimates show that HDP complicates about 10%–15% of pregnancies10 and represents a spectrum of disease that can progress from less severe (gestational HTN) to more severe (eclampsia). One mechanism for the association between increased CV risk and HDP is that predisposed women manifest with hypertensive phenotypes during pregnancy due to global endothelial dysfunction which unmasks pre-existing subclinical vascular disease.11–13 Clinically, the treatment for HDP is delivery, but many women who develop HDP have higher rates of CV risk factors before14 and after15 16 pregnancy, suggesting CV risk extends beyond gestation. While the majority of available evidence shows increased risk of CV events decades after a pregnancy complicated by HDP, there is growing evidence suggesting women with HDP may have higher short-term CV risk as well.17
Knowledge of the timing and type of CV events that occur in association with HDP could provide valuable insight for future preventive strategies. The purpose of this analysis was to determine if women with a pregnancy complicated by HDP are at increased risk of CV events in the short term following delivery. We hypothesised women with a singleton pregnancy complicated by HDP would have increased risk of a CV event within 3 years of delivery compared with women who did not have a pregnancy complicated by HDP.
Following approval by our institutional review board, we used administrative hospital discharge data from Florida through the Healthcare Cost and Utilization Project (HCUP) State Inpatient Database and constructed a retrospective cohort that included all in-hospital deliveries from 2004 to 2010. Because data are available publicly, informed consent was waived. Per the HCUP Data Use Agreement, cell sizes under 10 were reported as ‘<10’ for privacy protection. Only a single (index) pregnancy was included for each woman in the analysis. Index pregnancy was defined as the first new singleton delivery occurring between the years 2004 and 2010 regardless of parity. All included records were collected through 2013 to allow 3 years of follow-up. We determined demographics, comorbid conditions, procedures performed during admission and outcomes at the time of discharge from the index admission. Records that listed male sex of the mother or pregnancy with multiples (eg, twins, triplets, and so on) were excluded. Multiples were excluded because of known higher risk of developing HDP compared with singleton pregnancies.18 19 Race and ethnicity were stratified into four broad categories: white, African American (AA), Hispanic and other, which included all remaining races and ethnicities due to small sample sizes. Age was evaluated as both a continuous variable and categorical by decade due to potential for bimodal distribution of comorbidities or adverse outcomes at extremes of maternal age. Parity could not be determined due to lack of direct documentation in administrative data, limited time interval of study and inability to account for births in other states.
Comorbid diagnoses and performed procedures were determined using International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes (online supplementary table 1). Our predictor of interest was HDP, defined as a composite of gestational HTN (ICD-9-CM code 642.3), pre-eclampsia (642.4, 642.5) and eclampsia (642.6) diagnosed during the index admission for delivery. Creating a composite HDP predictor allowed for inclusion of women reclassified from gestational HTN to pre-eclampsia following updated pre-eclampsia guidelines published in 2013.9 Chronic HTN complicating pregnancy (642.0, 642.1, 642.2) was excluded from the HDP exposure group and considered a comorbid condition because it is a known risk factor for CV events.10 20 Additionally, women with individual codes for both chronic HTN and pre-eclampsia or gestational HTN were excluded from the HDP exposure group because of an inability to verify if women had both conditions. This was felt to be the more conservative approach, as chronic HTN independently predicts CV events10 20 and women misclassified as having HDP might affect the observed association between our primary predictor (HDP) and readmission outcomes. Women with a specific diagnosis of pre-eclampsia or eclampsia superimposed on chronic HTN (642.7) were included in the HDP exposure group. There were 1391 deliveries double coded as gestational HTN and pre-eclampsia, 114 deliveries double coded as pre-eclampsia and eclampsia, and 68 deliveries double coded as gestational HTN and eclampsia, likely representing progression to a more severe form of HDP, but were considered a single case of HDP in our data set (ie, women were not double counted). Only women with a diagnosis of HDP prior to hospital discharge for the index delivery were included as we were unable to capture new cases diagnosed in the outpatient setting post partum. Our primary outcome was CV readmission to any hospital in Florida within 3 years of delivery. Secondary outcomes included all-cause readmission, non-CV readmission and time to first readmission. Readmissions for subsequent pregnancies and deliveries were excluded. Primary admission diagnoses were used to determine the reason for readmission and stratified by readmission type: CV readmission—defined as a composite of acute myocardial infarction (MI) (410.x, 414.12), acute stroke (430, 431, 432.x, 433.x, 434.x, 435.x, 436) and acute heart failure (425.4, 425.8, 428.x, 674.5), non-CV readmission and no readmission. Time to first readmission was measured from the index delivery and tracked until the first readmission only. Readmissions within the first 42 days post partum were measured as these were within the diagnostic window and complication period for pre-eclampsia/eclampsia and could represent complications related to the pregnancy.
Supplementary file 1
Analyses were performed using a χ2 test for categorical variables and Student’s t-test for continuous variables. Associations were determined using logistic regression. Kaplan-Meier survival curves were built for time to first readmission with failure defined as first CV readmission and first readmission for any cause. Given the large sample size and anticipation that all covariates would be significant, the multivariate model was built in serial fashion with the following covariates: age, race/ethnicity, obstetric complications (preterm delivery, caesarean delivery, gestational diabetes), diagnoses and complications likely related to the pregnancy (transfusion, anaemia, thrombocytopenia, acute renal failure), comorbid conditions (tobacco, alcohol/drug abuse, chronic HTN, coronary artery disease, hyperlipidaemia, heart failure, diabetes, chronic kidney disease, obesity, chronic obstructive pulmonary disease, asthma, depression, autoimmune disorder) and payer (Medicare, Medicaid, private or unknown), which served as a proxy for socioeconomic status. Interaction testing was performed and significant interaction terms were entered into the multivariate model. Statistical significance was set at a P value of <0.05. All statistical analyses were performed using Stata V.14.1 (StataCorp, College Station, TX).
The overall sample included 1 452 926 singleton in-hospital deliveries that occurred in Florida between the years 2004 and 2010. Over 60% of women were between the ages of 18 and 29 years (mean 27.2±6.2 years). There were 108 875 pregnancies complicated by HDP (7% gestational HTN, 11% pre-eclampsia, <1% eclampsia) at the time of index delivery. A summary of baseline demographics and clinical profiles stratified by HDP status from the index admission for delivery is shown in table 1. Women with HDP were younger and had higher rates of obesity, anaemia, preterm delivery and caesarean delivery. When all women were stratified by readmission status (table 2), women with CV readmissions were older and had higher rates of obesity, heart failure, anaemia, depression and HDP at the time of index delivery. These women also had higher rates of preterm delivery and caesarean delivery. Of the 4054 CV readmissions, 701 (17.3%) were women with HDP during the index delivery, giving an event rate of 6.4 CV readmissions per 1000 women with HDP over 3 years compared with an event rate of 2.5 CV readmissions per 1000 women without HDP over 3 years.
Kaplan-Meier curves showing per cent survival without readmission over the 3 years of follow-up are shown in figure 1. Overall, only 32.7% of all first readmissions occurred within the first year post partum and less than 3% occurred in the immediate 42-day postpartum period (figure 2). This was also true for CV readmissions, where only 35.7% of first CV readmissions occurred within the first year following delivery.
Heart failure was the most common reason for CV readmission accounting for 78.6% of all CV readmissions and 84.4% of CV readmissions in women with HDP. A breakdown of CV event type is shown in online supplementary table 2. AA women made up 57% of the CV readmissions for an event rate of 6.8 CV readmissions per 1000 pregnancies in AA women compared with an event rate of 1.7 and 1.0 CV readmissions per 1000 pregnancies for white and Hispanic women, respectively (P<0.001 overall). Of the event types, AA women made up 62% of all heart failure readmissions (1986 of 3188; P<0.001 overall) and white women made up 52% of acute MI readmissions (205 of 394; P<0.001 overall).
Supplementary file 2
The results for the logistic regression model are shown in table 3. In the unadjusted model, HDP was a significant predictor of CV readmission with an OR of 2.59 (95% CI 2.39 to 2.81; P<0.001). This relationship was only slightly attenuated in the fully adjusted model (OR 2.41, 95% CI 2.08 to 2.80; P<0.001). HDP was also found to be a significant predictor of non-CV readmission (OR 1.11, 95% CI 1.09 to 1.14; P<0.001) and for any readmission (OR 1.13, 95% CI 1.10 to 1.15; P<0.001). Significant predictors for CV readmission in the full multivariate model are shown in figure 3. Heart failure was the strongest predictor for CV readmission (OR 28.75; 95% CI 23.87 to 34.64). Young age (<18 years) and asthma were associated with reduced risk of CV readmission. In the fully adjusted multivariable model, AA women had the highest risk for CV readmission (OR 3.60; 95% CI 3.32 to 3.90) and Hispanic women had the lowest risk (OR 0.59; 95% CI 0.51 to 0.68) when compared with white women after adjustment for HDP (figure 3).
Significant interactions were found between race/ethnicity and HDP (P<0.001) and obesity and HDP (P<0.001). The risk of CV readmission for HDP among white women was 2.96 (95% CI 2.57 to 3.41) compared with 1.74 (95% CI 1.55 to 1.96) in AA women, and 2.41 (95% CI 1.65 to 3.44) among Hispanic women. The risk of CV readmission estimated for HDP among obese women was 1.25 (95% CI 1.01 to 1.55) in comparison to 2.55 (95% CI 2.33 to 2.79) among non-obese women.
In this study of over 1.4 million diverse women, those who experienced a pregnancy complicated by HDP had twice the risk of CV readmission within 3 years of delivery compared with women without a history of HDP. Most readmissions occurred after the first year post partum, suggesting observed CV events were unrelated to direct complications from pregnancy. There were significant racial/ethnic differences in the patterns and risk of CV readmission, such that AA women had higher rates of heart failure and were at threefold greater risk for CV readmission compared with white women after adjustment for HDP. Pregnancy is an ideal time to risk stratify women because four out of five US women will have at least one pregnancy in her lifetime.21 HDP complicates between 10% and 15% of pregnancies,5 10 and with almost 4 million births in the USA each year,22 hundreds of thousands of mothers could be risk stratified without further cost or testing. By recognising HDP as a risk factor for short-term and long-term CV disease, women can be educated about risk, lifestyle modification and surveillance with more aggressive risk factor management post partum and beyond.
Our findings suggest higher short-term risk for CV events following a pregnancy complicated by HDP. Similar findings were reported by Black et al23 who used Kaiser Permanente data and found women free of pre-existing HTN or pre-HTN who experienced HDP had twice the risk for developing HTN within 1 year of delivery (relative risk 2.33; 95% CI 1.94 to 2.80). A recent study used a statewide-linked database to investigate short-term CV outcomes over 5 years in women free of comorbidities after a pregnancy complicated by a placental syndrome (pre-eclampsia, eclampsia, gestational HTN, placental abruption and placental infarction).17 Here, women with placental syndromes had increased risk of CV events (HR 1.19; 95% CI 1.07 to 1.32) which compounded when placental syndromes occurred with preterm birth and/or small for gestational age neonate (HR 1.45; 95% CI 1.24 to 1.71), likely signalling more severe vascular dysfunction.17 Unlike our study, only healthy women were included in the previous work. However, even those very low-risk women demonstrated increased risk of a CV event after a pregnancy complicated by HDP. Additionally, we found significant racial/ethnic differences in the risk and type of CV readmissions, such that AA women were at greater risk for CV events compared with white women. Racial differences in hypertensive disorders, both during and outside of pregnancy, have been observed by prior investigators,24 25 but associations between race and CV readmission are not well elucidated.
Large population studies have shown increased long-term risk for CV disease in women with HDP. A 2001 study using Norwegian medical registries found women with pre-eclampsia had a 20% higher risk of CV death over 25 years compared with women without pre-eclampsia.5 In a meta-analysis, pre-eclampsia was associated with a relative risk of 2.16 (95% CI 1.86 to 2.52) for ischaemic heart disease over 12 years of mean follow-up and a relative risk of 3.70 (95% CI 2.70 to 5.05) for HTN over 14 years of mean follow-up.26 More recently, investigators used 50 years of follow-up data in California and found early-onset pre-eclampsia (before 34 weeks’ gestation), late-onset pre-eclampsia (after 34 weeks’ gestation) and small for gestational age delivery significantly predicted premature CV death in women (HR 3.6, 95% CI 1.04 to 12.19; HR 2.0, 95% CI 1.18 to 3.46; and HR 1.6, 95% CI 1.02 to 2.42, respectively).10 In that analysis, risk of CV death associated with paired pregnancy complications was highest in women with preterm delivery plus pre-existing HTN (HR 7.1; 95% CI 3.49 to 14.55), pre-eclampsia plus pre-existing HTN (HR 5.6; 95% CI 2.09 to 15.18) and gestational HTN plus preterm delivery (HR 5.0; 95% CI 2.64 to 9.60)10—illustrating how multiple complications in a single pregnancy likely reflect more severe vascular dysfunction and compounded risk for the mother. These findings and the results of our study demonstrate increased CV risk in women with HDP beginning soon after delivery and persisting decades later. However, more work is needed to understand the mechanisms underlying HDP and how it relates to future CV disease, including investigation of risk modification in women with HDP and how it impacts CV risk.
There are important limitations to any study using administrative data. ICD-9-CM codes have variable diagnostic accuracy, but have been found to be sufficient for population research.27 Reliably determining parity was not possible using the current database and future efforts should seek to collect this information. In addition, we did not have access to the participants’ obstetrical histories, specifically whether a previous pregnancy was complicated by HDP. To avoid including women with chronic HTN who were misclassified as HDP, we excluded women with individual diagnostic codes for both chronic HTN and HDP from the exposure group. We felt this was the more conservative approach, as chronic HTN is estimated to effect only about 2% of pregnancies10 but is independently associated with CV events.10 20 A sensitivity analysis excluding all women with chronic HTN from the entire cohort was performed and yielded similar results (CV readmission OR 2.44, 95% CI 2.11 to 2.83, P<0.001; non-CV readmission OR 1.11, 95% CI 1.09 to 1.14, P<0.001; any readmission OR 1.13, 95% CI 1.10 to 1.15, P<0.001). Women with pre-existing cardiac conditions, such as heart failure and coronary artery disease, were included in these analyses and could have driven some of the observed CV risk. However, rates for pre-existing cardiac conditions were overall very low (<1% each for heart failure and coronary artery disease) and likely had a limited effect on the observed associations. Only women with a diagnosis of HDP at the time of the index delivery were included in our exposure group, which may have resulted in a smaller HDP sample size because new cases of HDP diagnosed post partum or in subsequent pregnancies (new or recurrent) were not included. This might account for the lower HDP rate in our study (7% of all deliveries) compared with prior studies showing rates of 10%–15%.10 We could not evaluate recurrent HDP cases in our analysis, although women with HDP have a 9%–14% risk of recurrent HDP during subsequent pregnancies28 and recurrent HDP may be associated with higher CV risk.29 Our data are limited to in-hospital deliveries and events only. We did not have access to socioeconomic information such as household income or education level, and so a proxy measure of insurer/payer was used. Due to the large sample size, there were many statistically significant differences in clinical characteristics that may not be clinically significant. Some, including anaemia, obesity, diabetes, chronic obstructive pulmonary disease/asthma, and depression, have previously been associated with CV risk. However, elaboration of their role in HDP-associated CV risk was beyond the scope of this analysis, and targeted investigation of potential interactive mechanisms is warranted.
In this diverse sample, women with HDP had twice the risk of short-term CV readmission within 3 years of delivery compared with women without HDP, demonstrating that increased CV risk associated with HDP appears soon after pregnancy and persists for decades with higher risk in AA women. More work is needed to explore the types and mechanisms of HDP-associated CV events for potential preventive strategies.
What is already known on this subject?
Pre-eclampsia and gestational hypertension are known cardiovascular (CV) risk factors for women, but little is known about how soon after delivery these women are at risk for CV events.
What might this study add?
Our study used hospital admission data to identify women with singleton deliveries complicated by a hypertensive disorder of pregnancy (HDP) and followed them for CV readmissions within 3 years post partum. We found women with HDP had over twice the risk of CV readmission (OR 2.41; 95% CI 2.08 to 2.80) within 3 years of delivery compared with women without HDP.
How might this impact on clinical practice?
Women with pregnancies complicated by HDP can be risk stratified, educated and followed more closely by healthcare providers for CV risk factor management while ongoing efforts continue to learn more about the timing, pathophysiology and types of CV events these women experience.
Contributors JLJ and DPK conceived and designed the study. DPK obtained the data and JLJ performed the analysis under the guidance of DPK. JLJ, DPK, TDM, MBD and JCE helped interpret the results. JLJ drafted the manuscript and all authors reviewed it critically for important intellectual content. All authors participated in the revisions and approved the final version. No individuals who significantly contributed to this work were excluded as an author.
Funding JLJ was supported by a T32 teaching grant (grant number 5T32HL007822). TDM is supported by the National Institute on Child Health and Human Development under award number 5K12HD001271-18. DPK is supported by the Jacqueline’s Research fund from the Jacqueline Marie Leaffer Foundation at the University of Colorado Center for Women’s Health Research and by NIH grant 1K08HL125725.
Disclaimer The content is solely the responsibility of the authors and does not necessarily represent the official view of the National Institute of Health.
Competing interests None declared.
Ethics approval COMIRB.
Provenance and peer review Not commissioned; externally peer reviewed.