Abstract
Objectives
To determine whether physical function (PF) before menopause is related to cardiovascular disease (CVD) risk.
Methods
Participants were N = 2950 pre-/early peri-menopausal women (median age 46, (25th–75th percentile: 43–48 years). Physical function was assessed at baseline using the Physical Function subscale of the SF-36 and scores were trichotomized (no, some, or substantial limitations). Clinical CVD events were ascertained at annual/biennial clinical assessments through the 15th follow-up visit. Risk of CVD was determined with Cox proportional hazards models.
Introduction
Poor physical function (PF) in late life is associated with a host of adverse clinical outcomes, including cardiovascular disease (CVD) (Corti et al., 1996; Newman et al., 2006; Saquib et al., 2013; Sergi et al., 2015). However, whether PF assessed at younger ages, for example, during midlife (ages 40–64 years), is related to future CVD risk has been less studied. The midlife period is increasingly recognized as a critical timeframe for PF trajectories in aging (Murray et al., 2011; Pope et al., 2001). Physical function often begins to decline in midlife, particularly for women, who have steeper declines in PF compared to midlife men (Murray et al., 2011). Midlife is also a crucial period for cardiovascular health in women as CVD risk is impacted both by chronological and ovarian aging, with CVD risk accelerating after menopause (El Khoudary et al., 2020).
Studies of PF and CVD risk in postmenopausal women suggest that worse PF may be indicative of poor cardiovascular health. In the Women’s Health Initiative, postmenopausal women (mean age 62.8 ± 6.9 years) in the lowest quintile of the physical component score (which encompasses physical function) had nearly double the risk of a CVD event and 4.7 times higher risk of CVD mortality compared to women in the highest (best) quintile, even after adjusting for baseline demographics, health behaviors, health conditions, and CVD risk factors (Saquib et al., 2013). Cross-sectionally, late-midlife women (mean age 59.6 ± 2.7 years) from the Study of Women’s Health Across the Nation (SWAN) with worse performance-based PF had wider adventitial diameter, an indicator of subclinical CVD, independent of other CVD risk factors, medications, and physical activity (El Khoudary et al., 2015). Assessing the relationship between PF at younger ages and clinical CVD events later in life could have implications for prevention efforts, particularly for identifying women at high risk for CVD as they age. In clinical settings, functional limitations assessed for patients aged 65 years and older are considered as part of overall health status, but whether PF at younger ages—and specifically, in pre-/early peri-menopause—could be informative of future health status is unknown. The purpose of this study was to determine whether PF during the pre-/perimenopause is associated with clinical CVD risk later in life, and the extent to which the association is independent of health and cardiovascular risk factors. We hypothesized that worse PF duirng the pre- and peri-menopause would be associated with greater risk of CVD events later in life.
Methods
Participants
Participants were pre- and early peri-menopausal women from SWAN, a longitudinal, multi-ethnic/multi-racial cohort study of health during and following the menopausal transition (Sowers et al., 2000). Briefly, SWAN recruited 3302 women between the ages of 42 and52 years old (baseline assessment in 1996/97) from seven field centers across the United States (Boston, MA, Chicago, IL, southeast MI, Los Angeles, CA, Newark NJ, Oakland, CA, and Pittsburgh, PA). Each study site recruited non-Hispanic White women and a designated racial/ethnic group (African American, Chinese, Japanese, or Hispanic) based upon location. Women completed annual or biennial assessment visits with the most recently completed visit (Visit 15) occurring between 2015 and 2017. At study baseline, women were eligible for SWAN if they had an intact uterus and at least one ovary, were premenopausal or early perimenopausal (i.e., had at least one menstrual period in the past 3 months), had not used reproductive hormones in the previous 3 months, and were not pregnant, lactating, or breastfeeding. Participants were eligible for this analysis if they were free of CVD at baseline (based upon a self-reported questionnaire of history of CVD events), provided data on baseline PF, and completed ≥1 follow-up visit. Of the initial 3302 participants, 2950 (89%) were included in this study. We excluded data from 92 women who reported having a heart attack/angina or stroke prior to baseline, 60 who did not provide baseline PF data, 181 who did not provide any follow-up data, and 19 who did not provide data on key covariates. The SWAN protocol was approved by each site’s institutional review board. All women provided written informed consent at each study visit.
Physical Function Assessment
Self-reported PF from the baseline assessment visit, which was assessed using the Physical Function subscale of the Medical Outcomes Study Short Form 36 (SF-36) (Ware & Sherbourne, 1992), was used for this analysis. The subscale has been validated against objective measures of PF in midlife and older adults for assessing mobility disability (Syddall et al., 2009). It assesses perceived limitation (“limited a lot,” “limited a little,” or “not limited at all”) in vigorous activities, lifting or carrying groceries, climbing several flights of stairs, climbing one flight of stairs, bending/kneeling/stooping, walking more than one mile, walking several blocks, walking one block, and bathing/dressing (McHorney et al., 1994; Ware & Sherbourne, 1992). Women were first asked if they were “limited in any way in activities because of any impairments or health problems.” If they reported no limitations, the PF questionnaire was not administered and the women were categorized as having no limitations. Women who had a positive response completed the questionnaire. Scores were categorized using cutpoints indicating no limitations, some limitations, or substantial limiations as previously described (Rose et al., 1999; Sowers et al., 2001; Tseng et al., 2012).
Cardiovascular Outcomes
Clinical cardiovascular events included heart attack/myocardial infarction, cerebrovascular accident/stroke, percutaneous coronary intervention, coronary artery bypass graft, congestive heart failure, and CVD death (Thurston et al., 2021). Non-fatal events and the date they occurred were self-reported at follow-up visits. At the 12th, 13th, and 15th follow-up visits, routine adjudication of cardiovascular events began. Attempts were made to collect medical records for each self-reported CVD event for these, and prior SWAN visits. Detailed information about these events was collected, including hospital admission history, physical examination discharge summary, laboratory data, diagnostic results, and reports on any operations or procedures, which were compiled by the SWAN Coordinating Center. Two cardiologists, masked to participant details, independently reviewed the medical records, and determined the diagnosis. If the two cardiologists disagreed regarding the diagnosis, a third member resolved the differences. Fatal CVD events were ascertained through systematic review of death certificates, requested by each site following knowledge of a participants’ death.
Covariates
Covariates were all from the baseline visit, though data was imputed using the same variable from the next available visit for missing baseline variables. Demographic characteristics included age, self-identified race/ethnicity (non-Hispanic White, African American, Chinese, Hispanic, and Japanese), study site, and financial strain (“How hard is it for you to pay for the very basics like food, housing, medical care, and heating?” Responses included “not hard at all,” “somewhat hard,” and “very hard”; somewhat and very hard combined for analyses (Hall et al., 2009). At baseline, all women were either premenopausal (menses in last 3 months with no irregularity) or early perimenopausal (menses in last 3 months with irregularity). Hormone use included baseline self-reported lifetime use of any female reproductive hormones (including birth control pills), as SWAN eligibility included not taking hormones within 3 month of study enrollment. Current self-rated general health status was self-reported and categorized as “excellent,” “very good,” “good,” “fair,” or “poor” (excellent/very good, good, or fair/poor combined for analyses). Health behaviors included smoking (never, former, current; combined current/former vs. never in analyses) and physical activity, which was assessed using the total score of Kaiser Physical Activity Survey (KPAS) (Sternfeld et al., 1999). The KPAS consists of 38 items querying about physical activity participation over the past year in sports/exercise, active living, occupational, and household/caregiving domains. Domain-specific scores were calculated from ordinal Likert scale responses regarding the perceived intensity, frequency, and duration of participation. In the sports/exercise section participants also identify the two most common physical activities they participated in over the prior year, and the intensity based upon metabolic equivalents, (METS; <4/4–6/>6 METs) was used for scoring. Domain-specific scores (ranging from 1 to 5, with higher scores indicating greater participation) were then summed to derive the total score.
Body mass index (BMI; kg/m2) was calculated using measured height and weight. Total blood cholesterol was assessed after a 12-hr fast, with hypercholesterolemia being defined as total cholesterol of ≥240 mg/dL or use of a lipid lowering medication. Diabetes mellitus was defined as self-reported use of anti-diabetic medications, fasting glucose ≥126 mg/dL while not taking corticosteroids, or self-report clinician diagnosis. Hypertension was defined as self-report of a clinician diagnosis, measured blood pressure ( ≥130 mm Hg systolic and/or ≥80 mm Hg diastolic (Whelton & Carey, 2018) or self-reported use of antihypertensive medication. Other CVD medications included anti-coagulants, anti-arrhythmics, vasodilators, and platelet inhibitors (combined in analyses).
Analytic Methods
Baseline participant characteristics were calculated for the full sample and stratified by PF limitation category. A series of marginal Cox proportional hazards models with Sandwich variance estimators were used for assessing risk of CVD events based upon baseline PF category (none, some, or substantial limitations). For women with a CVD event (including CVD death), time to event was calculated as time from baseline to the first CVD event. If the exact event date was not available, an imputed date was utilized (midpoint of the year or month that the participant reported), or if a participant gave a potential event date range, the upper limit (most recent) of that range was used. Women without an event were censored at their last study visit where CVD status was recorded. As of the 15th follow-up visit, 161 women were deceased, with death certificates and cause of death obtained for 154 deaths. In total, 24% of the non-fatal events were confirmed through adjudication. Some women had multiple CVD events during follow-up; only the first event was used in analyses.
In all models, clinical site was included as a random effect to account for clustering based upon site. Model 1 was minimally adjusted for baseline sociodemographic factors (age, race/ethnicity, and financial strain). Model 2 was further adjusted for baseline menopausal stage, hormone use, BMI, smoking status, and physical activity. Model 3 (fully-adjusted) additionally included baseline cardiovascular health measures (hypertension, diabetes, hypercholesterolemia, and CVD medication use). Proportional hazards assumptions were tested using cumulative martingale residuals and the supremum tests. All tests resulted in p-values >.05, indicating no violation of the proportional hazards assumption. The variance inflation factor (VIF) was also determined in order to assess multicollinearity; all VIFs were <2, indicating no evidence of multicollinearity.
To ensure robustness of our results, we repeated the above analyses using proportional subdistribution hazards models with non-cardiovascular death as a competing risk and clinical site included as a random effect. Three sensitivity analyses were performed: (1) repeating analyses using only adjudicated cardiovascular events, (2) excluding women who had a CVD event within the first two years to reduce the risk of reverse causation, and (3) adding in self-rated health. Self-rated general health, along with PF, are thought to be the same construct of physical well-being (Saquib et al., 2013), and are correlated in our sample (r = −.28, p < .001). All statistical analyses were run using SAS version 9.4 (SAS Institute Inc., Cary, NC).
Results
Women had a median baseline age of 46 years (25th–75th percentile: 43–48), were followed for a median of 19.1 years (25th–75th percentile: 15.5–19.5) and had a median age of 63.5 years (25th–75th percentile: 60.4–66.2) at the end of the study. In total, 220 women had a CVD event, including 23 women whose first CVD event was fatal. Of the non-fatal first CVD events, 82 women experienced a stroke, 46 women myocardial infarction, 23 percutaneous coronary intervention, 19 congestive heart failure, 4 coronary artery bypass graft, and 23 had multiple events at once.
Baseline Participant Characteristics Overall and by Baseline Physical Function Limitation Category.
Note. *Interquartile range, †Mean score from the Kaiser Physical Activity Survey (KPAS), ‡Cardiovascular disease medications (anti-coagulants, arrhythmic, vasodilators, and platelet inhibitors). Percentages are calculated out of the total number of participants with the measure, missing data <5%.
Baseline Physical Function by Cardiovascular Disease Event (CVD) Status.
Risk of Cardiovascular Disease Events (CVD) by Baseline Physical Function Limitation Category.
Note. Bolded estimates are statistically significant at p < .05.
Unadjusted Model: Testing effect of physical function only.
Model 1: Adjusted for age, race/ethnicity, and financial strain.
Model 2: Additionally adjusted for menopausal status, hormone use, body mass index, smoking, and physical activity.
Model 3: Additionally adjusted for hypertension, diabetes, hypercholesterolemia, and CVD medication use.
Results remained similar in models including non-CVD death as a competing risk factor. Results also remained consistent in senstivity analyses examining only adjudicated events, and after exclusion of the 18 women who had a CVD event within the first 2 years of the study (not shown). Self-rated health was associated with CVD events—women with good (HR: 1.53, 95% CI: 1.37–1.72) and fair/poor health (HR: 2.08, 95% CI: 1.49–2.98) had a significantly higher liklihood of a CVD event compared to women reporting excellent or very good health. After adjustment for self-rated general health status, the association of substantial limitations was no longer statistically significant (HR: 1.06, 95% CI: 1.04–1.08).
Discussion
In this cohort of initially pre- and early peri-menopausal women, women reporting substantial limitations in PF at baseline had 1.62 times higher risk of CVD events over 16 years of follow up compared to women with no limitations, even after accounting for demographics, menopausal factors, health behaviors, and CVD risk factors. Several CVD risk factors, particularly BMI, smoking, diabetes, and hypertension are known to be associated with PF, and our results cannot rule out that CVD and PF limitations share a common causal pathway. The association of substantial limitations was independent of these other risk factors, though was attenuated compared to the initial, unadjusted model. Importantly, several factors that are known to be associated both with worse PF and CVD events are modifiable (BMI, physical activity, and smoking status) or preventable and treatable (diabetes and hypertension) and should be targeted both for promoting cardiovascular health and PF into late life.
Prior studies examining future health risks associated with poor self-reported PF have largely focused on older adults (age 65+ years), or even late-midlife (age 50–64 years) adults. Few have focused specifically on midlife PF, and particularly among women who had not yet gone through menopause. Using the National Medicare Expenditure Survey, Franks and colleagues examined the risk of 5-year mortality based upon self-reported PF as assessed by the PF subscale of the SF-20 (Franks et al., 2003). Within the 45–54-year-old age group, each 10-point higher PF score was associated with a 12% lower risk of death. However, men and women were not examined separately, ignoring the established sex-differences in both life expectancy and the disablement process (Luy & Gast, 2014; Murray et al., 2011).
As previously mentioned, a study of the control participants from the Women’s Health Initiative utilized the SF-36 to assess health-related quality of life, including self-reported PF, among a large sample of postmenopausal women aged 50–79 years (N = 20,308; mean age 62.8 ± 6.9) at baseline. The authors found that that the PF and general health subscales were the only ones of the eight subcales to be associated with CVD incidence, CVD mortality, and all-cause mortality among their sample (Saquib et al., 2013). This is also consistent with our sensitivity analysis showing that self-rated general health status was associated with higher risk of CVD events, and the effect of substantial limitations on CVD events was not independent of self-rated general health. As these measures both assess the physical health domain, caution should be used when assessing them together. Further, though there is some overlap in the age of participants between this study and ours; importantly, all WHI participants were post menopausal, while the SWAN women were pre- and early peri-menopausal at baseline when PF was assessed. The authors concluded that physical health screening should be considered for identifying older women at high risk for CVD, in line with other studies recommending assessing aspects of PF in clinical settings (Parry et al., 2017; Middleton et al., 2015). Still, our results suggest that PF in midlife, prior to menopause, may have clinical relevance, not just PF at older ages.
Performance-based PF is predictive of future CVD events, CVD mortality, and all-cause mortality among older adults (Elbaz et al., 2005; Hamer et al., 2010) and in patients with existing CVD (Hülsmann et al., 2004; Kamiya et al., 2015). At the 12th clinic follow-up visit, when SWAN women were on average age 60 years (97% postmenopausal), women with slower gait speed had higher carotid artery plaque burden and carotid adventitial diameter, indicative of worse vascular health (El Khoudary et al., 2015). The Cardiovascular Health Study (mean age 78.7 years) found relationships between gait speed and subclinical CVD among older women, but not older men (Inzitari et al., 2008). In our study, the PF assessment tool was prone to ceiling effects, with most women reporting no limitations. A sensitive performance-based tool may have had the ability to further distinguish women based upon the spectrum of PF during this life stage, potentially revealing associations not captured here and recent and ongoing collection of performance measures in the cohort will allow these analyses in the future.
We investigated associations between PF and CVD risk, but PF is also known to decline as a result of CVD (Keeney et al., 2019; Levine et al., 2014). Previous work in SWAN demonstrated that women experience declines in PF in conjunction with the development of new chronic conditions, (Lange-Maia et al., 2020b) with incident stroke being associated with a 12% subsequent drop in PF, and an accelerated decline in PF concurrent with incident heart disease (Lange-Maia et al., 2020a). From a prevention standpoint, there are shared intervention targets that are related to PF improvement as well as CVD risk reduction and diabetes prevention and control. Potentially, reductions in PF may be an important hallmark of those women to target for intervention, and further research is needed to determine whether lifestyle interventions targeting midlife women with PF limitations can lead to improved age-related outcomes—especially women reporting substantial limitations. Physical activity and improved fitness, (Beavers et al., 2014; Blair et al., 1996; Manson et al., 2002; Pahor et al., 2014; Rejeski et al., 2012) weight loss, (Beavers et al., 2014; Rejeski et al., 2012; Wing et al., 2011) and smoking cessation (Pignataro et al., 2012; van den Borst et al., 2011) are all powerful intervention targets for promoting healthy aging, but further intervention and policy work is needed to develop optimal strategies for widescale adoption of these behaviors on a population level, particularly in mid-life prior to the increase in CVD known to occur with aging.
Strengths of this work include the long follow up (median 19 years) of a well-characterized cohort of midlife women, including annual/biennial assessment of CVD events at follow-up clinic visits with a portion confirmed by adjudication or death certificates. Importantly, this study extended prior research in postmenopausal women and older adults to a younger age group. A further strength is the racial and ethnic diversity of the cohort. Compared to non-Hispanic white women, non-Hispanic Black women had a higher CVD event risk, but Chinese and Japanese women had lower risk. These trends are broadly consistent with prior literature (Holland et al., 2011; Karnati et al., 2020; Winkleby et al., 1998). In addition to higher CVD risk, non-Hispanic Black midlife and older women generally have worse PF compared to non-Hispanic white women (Sowers et al., 2006). Intervention and prevention efforts regarding CVD risk reduction and for PF limitations are particularly needed among non-Hispanic Black women in mid-life to target these disparities.
This study also had limitations. The ascertainment of CVD events was primarily based upon self-report, and only a subset of events was adjudicated through medical records, potentially leading to outcome misclassification. We do not have reason to believe that events would have been differentially reported by level of PF, and thus the relationships observed in this study may have underestimated the true effects. In addition, this study only included midlife women, and results may not be generalizable or applicable to men. PF is known to be highly dynamic during midlife—women can experience declines but also improvements over time (Ylitalo et al., 2013)—never the less, demonstrating that a single measure of PF during pre- or peri-menopause is predictive of future outcomes has potential clinical implications. Specifically, clinicians should consider the functional status of their midlife patients, particularly in relation to other CVD risk factors, as women with low PF may deserve special attention at reducing known CVD risk factors.
In conclusion, we found associations between worse PF and risk of CVD outcomes in initially pre- and peri-menopausal women, consistent with work in older adults. Future prevention efforts focused on shared risk factors for poor PF and CVD risk could be particularly fruitful at helping midlife women age healthily, both in terms of PF and CVD health, and should be investigated.
Footnotes
Acknowledgments
We thank the study staff at each site and all the women who participated in SWAN.
Author’s Note
Clinical Centers: University of Michigan, Ann Arbor – Carrie Karvonen-Gutierrez, PI 2021–present, Siobán Harlow, PI 2011–2021, MaryFran Sowers, PI 1994–2011; Massachusetts General Hospital, Boston, MA – Sherri-Ann Burnett-Bowie, PI 2020–Present; Joel Finkelstein, PI 1999–2020; Robert Neer, PI 1994–1999; Rush University, Rush University Medical Center, Chicago, IL – Imke Janssen, PI 2020–Present; Howard Kravitz, PI 2009–2020; Lynda Powell, PI 1994–2009; University of California, Davis/Kaiser – Elaine Waetjen and Monique Hedderson, PIs 2020–Present; Ellen Gold, PI 1994–2020; University of California, Los Angeles – Arun Karlamangla, PI 2020–Present; Gail Greendale, PI 1994–2020; Albert Einstein College of Medicine, Bronx, NY – Carol Derby, PI 2011–present, Rachel Wildman, PI 2010–2011; Nanette Santoro, PI 2004–2010; University of Medicine and Dentistry – New Jersey Medical School, Newark – Gerson Weiss, PI 1994–2004; and the University of Pittsburgh, Pittsburgh, PA – Rebecca Thurston, PI 2020–Present; Karen Matthews, PI 1994–2020. NIH Program Office: National Institute on Aging, Bethesda, MD – Rosaly Correa-de-Araujo 2020–present; Chhanda Dutta 2016–present; Winifred Rossi 2012–2016; Sherry Sherman 1994–2012; Marcia Ory 1994–2001; National Institute of Nursing Research, Bethesda, MD – Program Officers. Central Laboratory: University of Michigan, Ann Arbor – Daniel McConnell (Central Ligand Assay Satellite Services). Coordinating Center: University of Pittsburgh, Pittsburgh, PA – Maria Mori Brooks, PI 2012–present; Kim Sutton-Tyrrell, PI 2001–2012; New England Research Institutes, Watertown, MA – Sonja McKinlay, PI 1995–2001. Steering Committee: Susan Johnson, Current Chair; Chris Gallagher, Former Chair.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The Study of Women’s Health Across the Nation (SWAN) has grant support from the National Institutes of Health (NIH), DHHS, through the National Institute on Aging (NIA), the National Institute of Nursing Research (NINR) and the NIH Office of Research on Women’s Health (ORWH) (Grants U01NR004061; U01AG012505, U01AG012535, U01AG012531, U01AG012539, U01AG012546, U01AG012553, U01AG012554, U01AG012495, and U19AG063720). The content of this article is solely the responsibility of the authors and does not necessarily represent the official views of the NIA, NINR, ORWH, or the NIH.
