Author + information
- Received January 4, 2017
- Revision received March 10, 2017
- Accepted March 12, 2017
- Published online May 15, 2017.
- Philip S. Hall, MDa,
- Gregory Nah, MAa,
- Barbara V. Howard, PhDb,
- Cora E. Lewis, MD, MSPHc,
- Matthew A. Allison, MD, MPHd,
- Gloria E. Sarto, MD, PhDe,
- Molly E. Waring, PhDf,
- Lisette T. Jacobson, PhD, MPA, MAg,
- JoAnn E. Manson, MD, DrPHh,
- Liviu Klein, MD, MSa and
- Nisha I. Parikh, MD, MPHa,∗ ()
- aDivision of Cardiology, University of California, San Francisco, San Francisco, California
- bMedStar Health Research Institute and Georgetown/Howard Universities Center for Clinical and Translational Research, Hyattsville, Maryland
- cDivision of Preventive Medicine, University of Alabama at Birmingham, Birmingham, Alabama
- dDivision of Preventive Medicine, University of California, San Diego, La Jolla, California
- eDepartment of Obstetrics and Gynecology, University of Wisconsin-Madison, Madison, Wisconsin
- fDepartment of Quantitative Health Sciences, University of Massachusetts Medical School, Worcester, Massachusetts
- gDepartment of Preventive Medicine and Public Health, University of Kansas School of Medicine-Wichita, Wichita, Kansas
- hDivision of Preventive Medicine, Brigham and Women's Hospital, Harvard Medical School, Boston, Massachusetts
- ↵∗Address for correspondence:
Dr. Nisha I. Parikh, Division of Cardiology, University of California, San Francisco, 555 Mission Bay Boulevard South 352T, San Francisco, California 94158.
Background Reproductive factors reflective of endogenous sex hormone exposure might have an effect on cardiac remodeling and the development of heart failure (HF).
Objectives This study examined the association between key reproductive factors and the incidence of HF.
Methods Women from a cohort of the Women's Health Initiative were systematically evaluated for the incidence of HF hospitalization from study enrollment through 2014. Reproductive factors (number of live births, age at first pregnancy, and total reproductive duration [time from menarche to menopause]) were self-reported at study baseline in 1993 to 1998. We employed Cox proportional hazards regression analysis in age- and multivariable-adjusted models.
Results Among 28,516 women, with an average age of 62.7 ± 7.1 years at baseline, 1,494 (5.2%) had an adjudicated incident HF hospitalization during an average follow-up of 13.1 years. After adjusting for covariates, total reproductive duration in years was inversely associated with incident HF: hazard ratios (HRs) of 0.99 per year (95% confidence interval [CI]: 0.98 to 0.99 per year) and 0.95 per 5 years (95% CI: 0.91 to 0.99 per 5 years). Conversely, early age at first pregnancy and nulliparity were significantly associated with incident HF in age-adjusted models, but not after multivariable adjustment. Notably, nulliparity was associated with incident HF with preserved ejection fraction in the fully adjusted model (HR: 2.75; 95% CI: 1.16 to 6.52).
Conclusions In post-menopausal women, shorter total reproductive duration was associated with higher risk of incident HF, and nulliparity was associated with higher risk for incident HF with preserved ejection fraction. Whether exposure to endogenous sex hormones underlies this relationship should be investigated in future studies.
Endogenous sex hormones present during a woman’s reproductive period prior to menopause may have unique effects on her lifetime risk of atherosclerosis, hypertension, cardiac remodeling, and heart failure (HF) (1,2). Two physiological phenomena affecting endogenous sex hormone levels during the reproductive period include menstrual cycling and pregnancy. The protective effects of menstrual cycles are frequently hypothesized to explain why pre-menopausal women have a lower risk of cardiovascular disease (CVD) compared with post-menopausal women of similar age, although this is still a controversial hypothesis (2,3). Moreover, data suggest that women with early menopause have an elevated risk of coronary heart disease (CHD) and stroke, and there are conflicting data as to whether early menopause also predicts HF (4–7).
In addition to menstrual cycling, pregnancy involves alterations in cardiovascular hemodynamics, fluid balance, inflammatory and thrombotic pathways, and exposure to endogenous sex hormones, all of which impart a variety of peripartum and post-partum risks to women (8–11). Previous analyses have suggested a relationship between reproductive factors, such as total number of pregnancies and number of live births, with the subsequent development of CVD, including HF (12–21). A variety of mechanisms have been proposed, but the relationship remains incompletely understood and is likely multifactorial. Notably, a higher number of pregnancies leading to live births has been associated with increases in left ventricular (LV) mass and end-diastolic volume, as well as with a decrease in LV ejection fraction, which may lead to HF development (18).
By comparison, infertility has also been associated with increased maternal cardiovascular risk: women who reported at least 5 years of infertility before having a successful pregnancy demonstrated a 19% higher incidence of CVD compared with women not reporting infertility (17). Consistent with these findings, an analysis of the Swedish population registry found a J-shaped relationship between a woman’s number of live births and her subsequent risk of CVD, finding that both nulliparous women and women with at least 5 live births had increased risk compared with women with 2 live births (16). A higher number of pregnancies correlates with an earlier age at first birth, which also has been demonstrated to be independently associated with higher risk for CHD (22). It is unclear to what extent socioeconomic factors may affect this risk (23).
To better characterize the relationship between reproductive factors and subsequent HF risk, we examined post-menopausal women in the WHI (Women’s Health Initiative) to test associations among total number of live births, age at first pregnancy lasting at least 6 months, and total reproductive duration (time from menarche to menopause) with incident HF. Our hypotheses were that nulliparity and multiparity (>2 pregnancies) would be associated with an increased risk of HF, as would a younger age at first pregnancy and a shorter total reproductive duration.
The design of WHI has been described previously (24,25). Briefly, a total of 161,808 women were enrolled in the observational study and clinical trial components from 1993 to 1998 in the United States. We studied the subset of 44,174 WHI participants included in the University of North Carolina (UNC) HF cohort, for whom HF outcomes were centrally adjudicated from enrollment through September 2014. This cohort includes all women randomized to the hormone trial component of WHI (n = 27,347) and all black participants (n = 11,880) and Hispanic participants (n = 4,947) from the other clinical trials and the observational study.
Figure 1 depicts the creation of our analytic sample. All women with pre-existing CVD (n = 467), including CHD, HF, stroke, or myocardial infarction, were excluded from the analysis, as were women with missing data (n = 15,191). Of the remaining participants, 28,516 women had completed menopause at enrollment and were included in the current analyses.
Baseline characteristics, including a detailed reproductive history, were obtained by interviews and questionnaires (24,25). Reproductive factors were defined as the age at first pregnancy lasting at least 6 months, total number of live births, and total reproductive duration (defined as the age at menarche subtracted from age at menopause). Selected covariates included age at screening, household income, education level, ethnicity, U.S. region, body mass index, hypertension (defined by self-report, systolic blood pressure ≥140 mm Hg or diastolic blood pressure >90 mm Hg at screening, or use of antihypertensive medication), diabetes (defined by fasting glucose >126 mg/dl or use of diabetes medication), hyperlipidemia requiring the use of medication, smoking status, history of breastfeeding for at least 1 month, history of pregnancy loss, prior hysterectomy, and usage of oral contraception or menopausal hormone therapy.
Within the cohort, all confirmed cases of HF hospitalization and patient-reported development of HF, angina, or CVD during hospitalization were sent to trained physicians at UNC for adjudication. These physicians reviewed the HF hospitalization medical records and classified cases following the algorithm used in the ARIC (Atherosclerosis Risk in Communities) study (26). Initial and subsequent hospitalizations were included, and information such as ejection fraction and diastolic dysfunction were included, when available, from the medical record. Cases were classified into 1 of the following 5 categories: definite acute decompensated heart failure (ADHF), possible ADHF, chronic stable HF, unclassifiable, or HF unlikely.
The primary outcome for this analysis was the time to development of a first hospitalization for definite or probable ADHF or newly diagnosed stable HF. Additional outcomes included the time to development of heart failure with reduced ejection fraction (HFrEF) (defined as EF <50%) and time to development of heart failure with preserved ejection fraction (HFpEF) (defined as EF ≥50%), analyzed individually. In the 90 cases where a participant developed both HFpEF and HFrEF during the analytic period, the 2 classes of HF outcomes were not considered mutually competing events.
Cox proportional hazards regression models, both age- and multivariable-adjusted, estimated associations among the number of live births, age at first pregnancy, and total reproductive duration with incident hospitalized HF, and with HFrEF and HFpEF individually. We tested the assumption of proportionality of hazards with Kaplan-Meier survival analysis, and we performed sensitivity analyses to test the competing influence of death during follow up. A p value <0.05 was considered statistically significant. All analyses were performed with SAS version 9.4 (SAS Institute Inc., Cary, North Carolina).
In secondary analyses, we separately evaluated the individual determinants of reproductive duration—age at menarche and age at menopause—with incident HF hospitalization. We also investigated effect modification by race/ethnicity by including an interaction term crossing race/ethnicity with reproductive duration in the regression model.
Recognizing that 42% of women in our study sample reported a history of hysterectomy prior to enrollment (Table 1), we separately evaluated the incidence of our primary outcome among women with natural menopause (without history of hysterectomy or oophorectomy) and among women with surgical menopause (with history of hysterectomy and/or oophorectomy). In secondary sensitivity analyses, we evaluated the incidence of our primary outcome among women without oral contraceptive use or menopausal hormone therapy, and after adjusting for infertility.
Of the 44,174 women included in the UNC HF adjudication cohort, 28,516 women had no missing reproductive data, were free of prevalent CVD, and were included in the analysis (Figure 1). Baseline characteristics of this sample are presented in Table 1. The mean age at screening was 62.7 years, and the mean ages at menarche and menopause were 12.6 and 47.1 years, respectively, with a mean total reproductive duration of 34.4 years. The average number of live births was 3.3.
The mean and median follow-up were 13.1 and 15.2 years, respectively. Minimum follow-up was 9 days and maximum follow-up was 19.6 years. There were 1,494 cases of incident adjudicated HF hospitalization, of which 505 were cases with HFrEF and 586 were cases with HFpEF. In the remaining 403 cases, ejection fraction was not available or was undetermined. There were 5,600 deaths during follow up: 827 deaths (15%) were preceded by HF events, including 271 with HFpEF, 204 with HFrEF, and 352 with undetermined ejection fraction.
Shorter total reproductive duration was associated with increased risk of HF, with an age-adjusted hazard ratio (HR) of 0.98 per year (95% confidence interval [CI]: 0.97 to 0.98 per year; p < 0.0001) and a multivariable-adjusted HR of 0.99 per year (95% CI: 0.98 to 0.99 per year; p = 0.02). This equates to a 5-year adjusted HR of 0.95 (95% CI: 0.91 to 0.99) (Table 2, Central Illustration).
Younger age at first pregnancy (age <20 years) and nulliparity were associated with increased risk of incident HF in age-adjusted models (HR: 1.42; 95% CI: 1.16 to 1.75; and HR: 1.80; 95% CI: 1.07 to 3.03, respectively). However, these associations were not statistically significant after multivariable adjustment (younger age at first pregnancy HR: 1.14; 95% CI: 0.91 to 1.43; nulliparity HR: 1.70; 95% CI: 0.95 to 3.03) (Table 2).
Shorter total reproductive duration was associated with increased risk of HFrEF in age-adjusted models (HR: 0.98; 95% CI: 0.97 to 0.99), but not after multivariable adjustment (Table 3). Nulliparity was associated with a statistically significant increased risk of HFpEF in both age-adjusted (HR: 2.57; 95% CI: 1.22 to 5.44) and multivariable-adjusted models (HR: 2.75; 95% CI: 1.16 to 6.52) (Table 4). This association remained significant in sensitivity analyses after additional adjustment for infertility (HR: 2.62; 95% CI: 1.24 to 5.55). Early age at first pregnancy and shorter total reproductive duration were associated with increased risk of HFpEF in age-adjusted but not multivariable-adjusted models (Table 4).
Sensitivity analyses considering death as a competing event were not materially different from the reported results (data not shown).
Table 5 presents the HRs for the individual components of reproductive duration, age at menarche, and age at menopause, both for the entire sample and stratified by surgical versus natural menopause. The association between shorter reproductive duration and increased risk of incident HF was related to an earlier age at menopause (adjusted HR: 0.99 per additional year before menopause; 95% CI: 0.98 to 0.99 per additional year), and particularly among women with natural menopause (adjusted HR: 0.97 per year before natural menopause; 95% CI: 0.96 to 0.99 per year). There was no evidence of statistically significant effect modification by race/ethnicity (p = 0.13).
Early age at first pregnancy (age <20 years) was significantly associated with increased risk of HF hospitalization among women without a history of oral contraceptive or menopausal hormone therapy (Table 6) and among women with natural menopause (Table 7), but not surgical menopause (Table 8).
Overall, our study demonstrated an association between selected reproductive factors and incident HF. Specifically, a shorter total reproductive duration resulted in a modestly increased risk of any HF, likely driven by a younger age at menopause. Moreover, our findings suggested that the association between total reproductive duration and HF is more pronounced in women with natural menopause compared with surgical menopause. Additionally, nulliparity was associated with an increased risk of HFpEF, even after adjustment for cardiovascular and sociodemographic confounders. However, higher parity was not associated with HF risk.
Shorter reproductive duration and HF
Prior data have suggested that women with early menopause, and consequently a shorter reproductive period with fewer reproductive cycles and lower cumulative exposure to endogenous sex hormones, have an increased risk of CHD, stroke, and possibly HF (4–7). An analysis of the Framingham Heart Study population suggested that the increased CHD risk with early menopause may be influenced by a loss of the protective effects of endogenous sex hormones and an increase in thrombotic risk (27). Estrogens have beneficial effects on nitric oxide synthesis and signaling, coronary artery calcification, myocardial contractile reserve, insulin resistance, and lipid metabolism, and progesterone has been demonstrated to decrease blood pressure through vasodilation and decreased angiotensin-II vasoresponsiveness (2,28–33). Unfortunately, randomized trials of sex hormone supplementation have failed to uniformly support the hypothesis that cardiovascular protection can be restored with exogenous estrogen or estrogen/progesterone supplementation (1,2,30,34–36).
Our finding that a shorter total reproductive duration was associated with a modestly increased risk of HF, even after adjustment for traditional cardiovascular risk factors, and particularly among women with natural menopause, might be due to the increased CHD risk that accompanies early menopause. We did not demonstrate a significant association between a shorter reproductive period and HFrEF that would corroborate this hypothesis, but a substantial proportion of HF events had unknown ejection fraction (27%), suggesting that HFrEF was likely under-represented. Importantly, another analysis of the Framingham population raised the possibility of reverse causation, demonstrating that pre-menopausal CHD risk factors, including higher serum cholesterol levels, blood pressure, and weight, were each associated with earlier age at menopause (37). These findings warrant ongoing evaluation of the potential cardioprotective mechanisms of endogenous sex hormone exposure in women.
Number of pregnancies leading to live births and HF
Our findings that nulliparity was associated with a higher risk of incident HFpEF were also consistent with prior data from the Swedish Population Registers, which suggested a J-shaped association between the number of pregnancies and subsequent cardiovascular disease (including HF). We did not observe a statistically significant increased risk of HFpEF with higher numbers of live births in this study.
To our knowledge, the association between nulliparity and HFpEF has not been reported previously. The continuous ovulatory cycles uninterrupted by pregnancy that correlate with nulliparity have been reported to increase the risk of breast, ovarian, and uterine cancers (38), but are generally believed to be protective against adverse cardiac remodeling and diastolic dysfunction (39). We hypothesized that the association between nulliparity and HFpEF might be driven by the established association between infertility, a correlate of nulliparity, and CVD (17,40). Exact causes of infertility have not been closely examined in terms of their association with CVD, and a large cohort analysis did not implicate fertility therapy as a cause of subsequent cardiovascular disease (41). We repeated the analysis controlling for a reported history of infertility, and the association between nulliparity and HFpEF was still present (adjusted HR: 2.62; 95% CI: 1.24 to 5.55), suggesting that the association could not be explained by infertility. It remains possible that residual confounding by hypertension or sociodemographic factors associated with hypertension is still present, despite their inclusion as covariates in the regression model, and this association merits further investigation.
The availability of reproductive data and adjudicated HF outcomes in the UNC HF cohort of WHI allowed us to uniquely evaluate the association between total reproductive duration and number of pregnancies. Only the first occurrence of HF hospitalization was included as an outcome in the initial WHI study, and details such as LV ejection fraction were not uniformly reported. We had a large sample size that was well represented along racial/ethnic, socioeconomic, and geographic lines. A substantial proportion of women in the cohort had missing reproductive data and were excluded from the analysis, which would likely diminish our power to identify true associations. We also relied upon self-reporting in determining our primary exposures and we did not have biomarker-confirmed menopause status. Additional information regarding pregnancy complications, such as pregnancy-induced hypertension, pre-eclampsia, and peripartum cardiomyopathy, would have been helpful and informative to include in the model, but were not available in the WHI cohort at the time of our analysis. A substantial percentage of HF cases had undetermined EF (27%), and more detailed echocardiographic findings or biomarkers, such as brain natriuretic peptide, would have been helpful to confirm the diagnoses. Finally, our decision to analyze HFpEF and HFrEF outcomes individually required additional statistical tests, and our results should be interpreted as hypothesis generating, with the need for corroboration in other populations. We did not account for multiple tests in determining statistical significance.
In this study, we found that a shorter total reproductive duration was associated with higher risk of incident total HF hospitalization in post-menopausal women, and nulliparity was associated with higher risk of incident hospitalized HFpEF. Whether exposure to endogenous sex hormones underlies this relationship should be investigated in future studies.
COMPETENCY IN MEDICAL KNOWLEDGE: Post-menopausal women with longer reproductive duration have a lower risk of hospitalization for HF, and nulliparity is associated with a higher risk of developing HFpEF.
TRANSLATIONAL OUTLOOK: Further studies are needed to determine the mechanisms linking endogenous sex hormone exposure during a woman’s reproductive years to cardiovascular risk after menopause.
This work was supported by the American College of Cardiology Merck Award (to Dr. Hall), American Heart Association Grant 13CRP17350002 (to Dr. Parikh), and National Institutes of Health grant 7R21HL115398 (to Dr. Parikh). The WHI program is funded by the National Heart, Lung, and Blood Institute, National Institutes of Health, U.S. Department of Health and Human Services, through contracts HHSN268201600018C, HHSN268201600001C, HHSN268201600002C, HHSN268201600003C, and HHSN268201600004C. The authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- acute decompensated heart failure
- coronary heart disease
- cardiovascular disease
- heart failure
- heart failure with preserved ejection fraction
- heart failure with reduced ejection fraction
- Received January 4, 2017.
- Revision received March 10, 2017.
- Accepted March 12, 2017.
- 2017 American College of Cardiology Foundation
- Harvey R.E.,
- Coffman K.E.,
- Miller V.M.
- Murphy E.
- Appiah D.,
- Schreiner P.J.,
- Demerath E.W.,
- Loehr L.R.,
- Chang P.P.,
- Folsom A.R.
- Chung E.,
- Leinwand L.A.
- Cain M.A.,
- Salemi J.L.,
- Tanner J.P.,
- Kirby R.S.,
- Salihu H.M.,
- Louis J.M.
- Fraser A.,
- Nelson S.M.,
- Macdonald-Wallis C.,
- et al.
- Ngo A.D.,
- Roberts C.L.,
- Figtree G.
- Parikh N.I.,
- Cnattingius S.,
- Mittleman M.A.,
- Ludvigsson J.F.,
- Ingelsson E.
- Parker D.R.,
- Lu B.,
- Sands-Lincoln M.,
- et al.
- Weinberg E.O.,
- Thienelt C.D.,
- Katz S.E.,
- et al.
- Parikh N.I.,
- Jeppson R.P.,
- Berger J.S.,
- et al.
- Park J.S.,
- Jung I.,
- Youn J.C.,
- Cho H.Y.
- Rosamond W.D.,
- Chang P.P.,
- Baggett C.,
- et al.
- Petre R.E.,
- Quaile M.P.,
- Rossman E.I.,
- et al.
- Christian R.C.,
- Liu P.Y.,
- Harrington S.,
- Ruan M.,
- Miller V.M.,
- Fitzpatrick L.A.
- Kok H.S.,
- van Asselt K.M.,
- van der Schouw Y.T.,
- et al.
- Zhao Z.,
- Wang H.,
- Jessup J.A.,
- Lindsey S.H.,
- Chappell M.C.,
- Groban L.
- Chandra A.,
- Copen C.E.,
- Stephen E.H.
- Udell J.A.,
- Lu H.,
- Redelmeier D.A.