Healthy Lifestyle Index and Risk of Cardiovascular Disease Among Postmenopausal Women With Normal Body Mass Index

Abstract

Background

A lifestyle comprising a healthy diet, light alcohol consumption, no smoking, and moderate or intense physical activity has been associated with reduced risk of cardiovascular disease (CVD). We examined the association of a healthy lifestyle index (HLI), derived from scores for each of these components plus waist circumference, with the risk of incident CVD and CVD subtypes in postmenopausal women with normal body mass index (18.5–<25.0 kg/m²).

Methods and Results

We studied 40 118 participants in the Women's Health Initiative, aged 50 to 79 years at enrollment, with a normal body mass index and no history of CVD. The HLI score was categorized into quintiles. We estimated multivariable adjusted hazard ratios (HR) and 95% CIs for the association of HLI with risk of CVD and CVD subtypes using Cox regression models. A total of 3821 cases of incident CVD were ascertained during a median follow‐up of 20.1 years. Compared with the lowest quintile (unhealthiest lifestyle), higher HLI quintiles showed inverse associations with the risk of CVD (HR _quintile−2 =0.74 [95% CI, 0.67–0.81]; HR _quintile−3 =0.66 [95% CI, 0.60–0.72]; HR _quintile−4 =0.57 [95% CI, 0.51–0.63]; and HR _quintile−5 =0.48 [95% CI, 0.43–0.54], P‐trend=<0.001). HLI was also inversely associated with risks of stroke, coronary heart disease, myocardial infarction, angina, and coronary revascularization. Subgroup analyses, stratified by age (≤63 years vs >63 years), body mass index (</≥ 22.0 kg/m²), and general health status (absence/presence of hypertension, diabetes, or lipid‐lowering drug use) also showed inverse associations between HLI and risk of CVD.

Conclusions

Among postmenopausal women with a normal body mass index, adherence to a healthy lifestyle is associated with a reduced risk of clinical CVD and CVD subtypes, underscoring the cardiovascular benefits of maintaining a healthy lifestyle, even for women with a healthy weight.

Nonstandard Abbreviations and Acronyms

HLI

healthy lifestyle index

OS

observational study

WHI

Women's Health Initiative

Clinical Perspective.

What Is New?

• In this study of 40 118 postmenopausal women participating in the WHI (Women's Health Initiative) study with normal body mass index (18.5–<25.0 kg/m²), higher levels of healthy lifestyle index were associated with reduced risk of cardiovascular disease over a period of 20 years.

What Are the Clinical Implications?

• Women with a normal body mass index are usually considered at a lower risk of cardiovascular disease compared with women with higher body mass index.

• Maintaining a healthy lifestyle comprising a healthy diet, light alcohol consumption, no smoking, and moderate or intense physical activity provides substantial cardiovascular benefits beyond the control of body weight.

Cardiovascular disease (CVD) is the leading cause of morbidity and mortality worldwide and is responsible for reducing the quality and duration of life. ¹ Extensive evidence indicates that excessive body weight is positively associated with risk of coronary heart disease (CHD), heart failure, atrial fibrillation, and stroke. ² This had led to the recommendation to maintain a normal weight throughout life to reduce the risk of developing cardiovascular‐related conditions. ³ , ⁴ , ⁵ A recent study that evaluated the lifetime risk of CVD using data from 10 US cohort studies found that women (age, 40–59 years) with a body mass index (BMI, calculated as weight in kilograms divided by the square of height in meters [kg/m²]) between 18.5 and <25.0 kg/m² had the lowest lifetime risk of developing CVD and the highest number of lived CVD‐free years compared with all the other BMI categories (underweight, <18.5; overweight, 25.0–29.9; obesity, 30.0–39.9; and morbid obesity, ≥40 kg/m²). ⁶

Alcohol consumption, cigarette smoking, ⁷ low physical activity, ⁸ and poor diet, ⁹ entailing low intake of fruits, vegetables, ¹⁰ and whole grains ¹¹ and high consumption of red and processed meat, ¹² , ¹³ are modifiable lifestyle factors associated with increased risk of CVD. ¹⁴ The American College of Cardiology/American Heart Association clinical guidelines indicate that the most effective preventive strategy to reduce the burden of CVD is to maintain a healthy lifestyle, comprising consumption of a high‐quality diet, undertaking adequate physical activity, not smoking, consuming low to moderate levels of alcohol, and maintaining or achieving a normal BMI. ⁵ , ¹⁵

Evaluating the association of individual factors with CVD risk may not be as effective as considering them together in a combined score, ⁴ which summarizes these lifestyle behaviors and may account for the correlations among individual factors. Several studies have shown that a relatively high healthy lifestyle index (HLI), which reflects a high‐quality diet, moderate to high physical activity, low alcohol consumption, a relatively small waist circumference, and no cigarette smoking, is associated with reduced risk of CVD, ¹⁶ , ¹⁷ Although previous studies have examined the association of individual behavioral factors, such as a sedentary lifestyle, with CVD risk among women with a normal BMI, ¹⁸ none has examined whether a combination of these factors, as reflected in a higher HLI, is associated with reduced risk of CVD among individuals with a normal BMI.

In the study reported here, we evaluated the association between the HLI and CVD risk among participants with normal BMI in the WHI (Women's Health Initiative) study. In addition, previous evidence from this cohort of postmenopausal women with normal BMI suggests that body fat distribution is associated with the risk of CVD. ¹⁹ Therefore, we also tested whether the inclusion of variables measuring fat distribution in the analytical model changed the estimates of the association between HLI and CVD risk.

Methods

The data that support the findings of this study are available from WHI at helpdesk@whi.org. Further information is available from the corresponding author with the permission of WHI.

Study Population

Between 1993 and 1998, the WHI enrolled 161 808 postmenopausal women aged 50 to 79 years at 40 clinical sites across the United States to advance understanding of the determinants of major chronic diseases in postmenopausal women. ²⁰ Based on the original WHI protocol, the term “women” indicates individuals who were assigned female at birth and identified as women at the time of the study enrollment. The WHI comprises an observational study (OS) component (n=93 676) and 4 overlapping randomized clinical trials (CTs) (n=68 132), including 2 hormone therapy trials, a dietary modification trial, and a (calcium plus vitamin D) supplementation trial. The primary study was completed in 2005 and was followed by 4 extensions (starting in 2005 and currently extended to 2027) to collect updated information annually on outcomes among those who consented to continue to participate.

The WHI study received ethical approval from the ethics committees at all 40 clinical centers and at the WHI coordinating center (Fred Hutchinson Cancer Center – Seattle, WA). All participants provided written informed consent. All informed consent materials were reviewed by study investigators, local Institutional Review Boards, the WHI Data and Safety Monitoring Board, the Women's Health Initiative Policy Advisory Committee, the National Institutes of Health, and the Institute of Medicine in Washington, DC.

Covariates

Sociodemographic, health, lifestyle, and medication use information was collected at enrollment in the WHI using self‐administered questionnaires. Race and ethnicity were included in the study as reported by the participants at baseline when asked the question: “How would you describe your racial or ethnic group? If you are of mixed blood, which group do you identify with most?”. Participants also completed a semiquantitative food frequency questionnaire in which they reported the frequency of consumption of foods and beverages over the past 3 months. ²⁰ Weight, height, and waist circumference, were measured by trained study personnel at baseline and at periodic follow‐up exams using standardized procedures. Blood pressure (BP) was measured using an appropriate cuff size on the upper right arm and a mercury sphygmomanometer after the participant had been seated and rested for 5 minutes; the average of 2 BP measurements (≥30 minutes apart) was used. Participants attending 3 study centers (Pittsburgh, PA; Birmingham, AL; and Tucson‐Phoenix, AZ) had body fat measured using dual‐energy x‐ray absorptiometry (DXA; QDR2000, 2000+, or 4500W; Hologic Inc., Bedford, MA), ²¹ following a high‐quality assurance program. ²²

Health Lifestyle Index

Information on 5 lifestyle‐related factors, namely, waist circumference, cigarette smoking, alcohol intake, diet quality, and leisure‐time physical activity, was used to generate the HLI score. Waist circumference was measured to the nearest 0.1 centimeter at the narrowest portion of the torso region. Reported information on smoking status and amount smoked was used to generate smoking categories (never, former smokers quit ≥10 years, former smokers quit <10 years, current smoking ≤15 pack‐years, and current smoking >15 pack‐years). Baseline alcohol consumption was obtained from the food frequency questionnaire, which collected information on the type (beer, wine, and hard liquor) and quantity of alcohol consumed; consumption was summarized as drinks per day and then categorized (0, >0‐<0.5, 0.5‐<1, 1‐<2, ≥2 drinks/d). Dietary quality was summarized using the Alternative Healthy Eating Index 2010, ²⁰ which was constructed based on information obtained from the food frequency questionnaire. The Alternative Healthy Eating Index includes food and beverages strongly associated with the risk of chronic disease such as vegetables, fruit, whole grain, nuts and soy protein, sugar‐sweetened beverages, processed red meat, long‐chain polyunsaturated fatty acids (eicosapentaenoic acid+ docosahexaenoic acid), and other polyunsaturated fat, trans fat, and sodium. The Alternative Healthy Eating Index score (0–100, with lower score indicating poor diet quality) ²⁰ was then categorized into quintiles. Leisure time physical activity was reported via a self‐administered questionnaire which ascertained the frequency and duration of various types of physical activity. This questionnaire was specifically designed to measure physical activity in women, including minorities and those with special needs. Details describing the individual activities and related scores listed on the questionnaire and its validity have been previously described. ²³ A metabolic equivalent value was assigned to each category of physical activity (mild, moderate, and vigorous) and multiplied by its duration. ²⁴ The sum of the weekly metabolic equivalent‐hours of all activities (metabolic equivalent‐hours per week [h/wk]) was categorized into quintiles.

A score (0–4) was assigned to individual categories of each lifestyle component, with higher scores reflecting healthier behaviors. Based on the nonlinear association between alcohol intake and CVD in this and in other cohorts, ²⁵ the category of 0.5 to <1 drink/d was considered the one with lowest risk and was assigned the highest score (namely, 4). The scores for each factor were summed to generate a composite HLI score (0–20), with higher scores representing healthier behaviors. A similar multilevel score used to assess the association of healthy lifestyle factors with health outcomes has been used previously in this cohort. ²⁶ , ²⁷ Detailed information on HLI components, their categorization, and assigned scores is presented in Table S1.

Outcomes

During the main WHI study (1993–2005), study participants were contacted every 6 months (CT) or annually (OS), to determine whether (among other outcomes) they had been hospitalized or had undergone a procedure for potential CVD events. After completion of the main study, outcome information was collected annually from those subjects who participated in the extension studies. All positive responses were followed by medical records review, and outcome adjudication was performed centrally by WHI physicians until 2010; after this time, of all reported CVD cases, only those in the Medical Records Cohort (n=7828), which included women in the hormone therapy trials, plus all Black and Hispanic participants, were centrally adjudicated, while for the rest of the cohort self‐reported outcome information was collected.

For the present study, the primary outcome was the first occurrence of CVD as defined in the WHI study, which included stroke, CHD, angina requiring hospitalization, and coronary revascularization procedures in women with no prior history of CVD at baseline. CHD included nonfatal myocardial infarction (MI) with overnight hospitalization, possible or definite fatal CHD, and coronary revascularization validated by medical record review. ²⁸ We also used an alternative CVD outcome that included only cases of stroke, CHD and MI, given that angina and coronary revascularization are less reliably reported, which may have led to detection/referral biases. Diagnosis of stroke, which included ischemic and hemorrhagic types, required rapid onset of a neurological deficit that lasted >24 hours, hospitalization, and support from imaging studies when available. During follow‐up through March 2020, 3821 incident CVD events were ascertained and consisted of 1472 cases of CHD (including 1076 cases of clinical MI), 1304 cases of coronary revascularization (596 with concurrent CHD and 42 with simultaneous stroke), 643 cases of angina (72 with concurrent CHD and 40 with simultaneous stroke), and 1436 cases of stroke (151 with simultaneous CHD). Given that some participants experienced >1 event at once, the sum of the component events exceeds the total number of individuals with CVD. Deaths from CVD or other causes were documented with death certificates and centrally adjudicated using medical record review by study physicians.

Analytic Cohort

At baseline, 54937 (34.3%) WHI participants had a BMI between 18.5 and 25.0 kg/m². From this eligible sample, we excluded women who participated in the dietary intervention arm of the DM trial (n=5005), and those who had a positive history of CVD at baseline (n=7857), had missing follow‐up data (n=134), or did not have complete exposure (HLI) information (n=1823). The remaining 40 118 women were included in the present analysis, of whom 2507 women had data on DXA.

Statistical Analysis

Characteristics of the analytical sample were summarized by HLI categories (quintiles) based approximately on quintiles of the observed distribution of HLI values among the noncases of incident CVD. For each HLI quintile, we calculated incidence rates for CVD overall and for specific CVD subtypes by dividing the number of first events by 1000 person‐years at risk. We used Cox proportional hazards models to estimate hazard ratios (HR) and 95% CI of the association of the HLI quintiles with risk of CVD and its subtypes using the lowest quintile as the referent category. In addition, linear increments of the HLI SD were also modeled. Study follow‐up time was used as the underlying time‐at‐risk scale and was measured from the date of enrollment to the date of occurrence of the outcome of interest for cases, and to the end of the follow‐up (February 28, 2020), study withdrawal, or death other than from CVD, whichever came first, for the noncases. No departure from the proportional hazards assumption for the association of HLI with CVD risk was detected using Schoenfeld residuals (P=0.319). Since age at the time of enrollment violated the proportionality assumption, all analyses were stratified by age (5‐year intervals).

The selection of the variables included in the multivariable models was based on whether they reached statistical significance (P<0.05) in univariable analysis, or if they were known cardiovascular risk factors. The following variables were included in the models: age, race and ethnicity (American Indian or Alaska Native, Asian or Pacific Islander, Black, Hispanic or Latino, White, and other race and ethnicity); education (≤high school, some college, and ≥college degree); marital status (never married, divorced/separated, widowed, married or marriage‐like relationship); family income (<$35 000, ≥$35 000–<$50 000, ≥$50 000–<$75 000, ≥$75 000); reported sleep quality in the prior 4 weeks (restless, average, restful); insomnia (WHI Insomnia Rating Scale ≥9), ²⁹ height; total nonalcohol dietary energy intake (derived from the food frequency questionnaire); duration of use of menopausal hormone therapy (never, ≤ 5, 5–<10, 10–<15, ≥15 years); use of aspirin; use of antihypertensive drugs; use of lipid‐lowering drugs; years since menopause; systolic BP; diastolic BP; having quit cigarette smoking for health reasons; and having quit alcohol for health reasons. In addition, we adjusted the models by WHI study component (OS, CT) and trial arms hormone therapy (estrogen‐alone intervention, nonintervention; estrogen+progestin intervention, nonintervention), calcium plus vitamin D (intervention, nonintervention), and DM (nonintervention). Variance–covariance matrices were used to test for collinearity between pairs of variables, but none was found. For categorical variables with missing values, a separate category was created to retain the observations with missing values in the analysis, while for continuous variables, such a systolic and diastolic BP only 23 women had missing values (<0.06%), which were imputed based on factors that affect BP level such as age, diabetes, BMI, waist circumference, and healthy lifestyle. ³⁰ Approximately 8.5% of women had missing data on ≥1 variables. To reduce the possibility of bias, we repeated the analyses with the exclusion of women with these missing values. We tested for trends in the association across quintiles by assigning the HLI median values to each quintile and modeling this as a continuous variable.

A competing‐risk analysis for CVD and its subtypes was performed using Fine–Gray subdistribution hazard models with death from non‐CVD causes (n=6417) as the competing outcome. ³¹ In addition, we performed several sensitivity analyses. First, to reduce the possibility of reverse causality, we analyzed only women with >2 years of follow‐up (n=39 485); second, since postmenopausal years are often associated with weight gain, we restricted the analysis to women who maintained a normal BMI as reported at the year 3 follow‐up exam (n=28 851); and third, since menopausal hormone therapy alone or in combination with calcium and vitamin D supplementation might modify the risk of CVD, ³² we performed an analysis including only women who did not participate in the hormone therapy and calcium plus vitamin D intervention arms (n=35 036). We also conducted subgroup analyses based on several characteristics such as age (≤/>median age, 63 years), and BMI (</≥22.0 kg/m²) (this cutoff was chosen based on reports on BMI and associated mortality risk ³³ ). Additional subgroup analyses were conducted separately in reported sleep quality categories (restless, average, and restful), and in general health status groups (defined by the presence of hypertension or diabetes, or the use of lipid‐lowering drugs). For this purpose, hypertension was defined as having a systolic BP ≥130 mm Hg and a diastolic BP ≥80 mm Hg and/or use of antihypertensive medication, in accordance with the current clinical guidelines. ³⁴ Since the selection criteria and the participant characteristics of the WHI study components (OS and CT) were different, we conducted additional subgroup analyses in each of these study components. We also examined the association of the individual components of the HLI with risk of CVD and tested whether they were determinants of the overall association by removing them one at a time from the HLI composite score.

We examined the relationship between HLI and various measures of body fat distribution in the subcohort of normal BMI women for whom DXA measures were available (n=2507). To evaluate whether this sample was representative of the total analytical cohort, we compared baseline characteristics between this group and the one without DXA information (Table S2). Age, race and ethnicity, and time since menopause‐adjusted partial correlations and multivariable linear regression models adjusted for age, race and ethnicity, total energy intake, height, and time since menopause were used to evaluate the relationship between HLI (linear and quintiles) and each measure of body fat distribution. Furthermore, in the subgroup of 2507 women with DXA data, we examined whether the association between HLI quintiles and CVD was altered by including DXA measures in the model.

All statistical analyses were performed with STATA version 17 (Stata Corp LP, College Station, TX). All P values were 2‐sided.

Results

During a median follow‐up period of 20.1 years (interquartile range, 9.1–22.9 years), 3821 first CVD cases were ascertained among the 40 118 postmenopausal women. Distributions of the baseline characteristics of the study participants by HLI quintiles are shown in Table 1. Women with higher HLI scores were more likely to be White, had a higher level of education, were married or in a marriage‐like relationship, were enrolled in the OS, had menopause <15 years before enrollment, were less likely to have diabetes or to use anti‐hypertensive and lipid‐lowering medications, had on average lower systolic and diastolic BP, and were less likely to have quit smoking and alcohol consumption for health reasons.

Table 1.

Baseline Characteristics of Participants in the Women's Health Initiative Study With Body Mass Index Between 18.5 and 25.0 kg/m²

	Healthy lifestyle index categories
	1st	2nd	3rd	4th	5th
n	7197	7156	8810	8003	8952
HLI range	0–7	8–9	10–11	12–13	≥14
Characteristics
Age at enrollment,* y	62.8 (7.3)	63.3 (7.4)	63.2 (7.5)	63.3 (7.4)	62.6 (7.3)
Age at CVD diagnosis,* , † y	73.2 (8.1)	75.6 (7.8)	76.0 (7.9)	75.8 (7.7)	75.8 (8.2)
Follow‐up time,* y	15.0 (7.3)	16.0 (7.2)	16.7 (7.1)	17.1 (6.9)	17.8 (6.6)
Race and ethnicity, n (%)
American Indian or Alaska Native	31 (0.4)	18 (0.3)	29 (0.3)	12 (0.2)	16 (0.2)
Asian or Pacific Islander	191 (2.7)	311 (4.4)	451 (5.1)	446 (5.6)	422 (4.7)
Black	565 (7.9)	350 (4.9)	326 (3.7)	183 (2.3)	145 (1.6)
Hispanic or Latino	301 (4.2)	260 (3.6)	261 (3.0)	201 (2.5)	188 (2.1)
White	6015 (83.6)	6110 (85.4)	7644 (86.8)	7040 (88.0)	8086 (90.3)
Other	94 (1.3)	107 (1.5)	99 (1.1)	121 (1.5)	95 (1.1)
Education, n (%)
≤High school	1987 (27.6)	1533 (21.4)	1537 (17.4)	1144 (14.3)	934 (10.4)
Some college	2862 (39.8)	2639 (36.9)	3055 (34.7)	2560 (32.0)	2653 (29.6)
≥College degree	2174 (30.2)	2797 (39.1)	3857 (43.8)	3926 (49.1)	4861 (54.3)
Missing	174 (2.4)	187 (2.6)	361 (4.1)	373 (4.7)	504 (5.6)
Marital status, n (%)
Never married	315 (4.4)	343 (3.8)	373 (4.2)	374 (4.7)	345 (3.9)
Divorced or separated	1343 (18.7)	1115 (15.6)	1195 (13.6)	1137 (14.2)	1179 (13.2)
Widow	1268 (17.6)	1165 (16.3)	1320 (15.0)	1168 (14.6)	1182 (13.2)
Married or marriage‐like relationship	4244 (59.0)	4507 (63.0)	5885 (66.8)	5287 (66.1)	6222 (69.5)
Missing	27 (0.4)	26 (0.4)	37 (0.4)*	37 (0.5)	24 (0.3)
WHI study enrollment, n (%)
Observational study	4402 (61.2)	4852 (67.8)	6437 (73.1)	6113 (76.4)	7200 (80.4)
Calcium and vitamin D trial	1515 (21.1)	1222 (17.1)	1302 (14.8)	1048 (13.1)	999 (11.2)
Dietary modification trial‡	1548 (21.5)	1328 (18.6)	1380 (15.7)	1046 (13.1)	962 (10.8)
Hormone therapy trial	1476 (20.5)	1123915.50	1134 (12.9)	942 (11.8)	863 (9.6)
Time since menopause, n (%)
<5 y	851 (11.8)	943 (13.2)	1307 (14.8)	1194 (14.9)	1505 (16.8)
5‐<15 y	2566 (35.7)	2600 (36.3)	3277 (37.2)	3052 (38.1)	3585 (40.1)
≥15 y	3264 (45.4)	3269 (45.7)	3889 (44.2)	3489 (43.6)	3613 (40.4)
Missing	516 (7.2)	344 (4.8)	337 (3.8)	268 (3.4)	249 (2.8)
Quality of sleep, n (%)
Restless	1165 (16.2)	1063 (14.9)	1229 (14.0)	956 (12.0)	1064 (11.9)
Average	3104 (43.1)	3001 (41.9)	3587 (40.7)	3171 (39.6)	3364 (37.6)
Restful	2895 (40.2)	3058 (42.7)	3964 (45.0)	3849 (48.1)	4505 (50.3)
Missing	33 (0.5)	34 (0.5)	30 (0.3)	27 (0.3)	19 (0.2)
Insomnia, WHIIRS, n (%)	2169 (30.1)	2158 (30.2)	2541 (28.8)	2158 (27.0)	2264 (25.3)
History of type 2 diabetes, n (%)	219 (3.1)	162 (2.3)	195 (2.2)	132 (1.70	97 (1.1)
Aspirin use, n (%)	1485 (20.6)	1471 (20.6)	1674 (19.0)	1569 (19.6)	1755 (19.6)
Antihypertensive, n (%)	1556 (21.7)	1302 (18.2)	1405 (16.0)	1205 (15.1)	1097 (12.3)
Lipid‐lowering drug, n (%)	491 (6.8)	465 (6.5)	501 (5.7)	384 (4.8)	306 (3.4)
Vasodilator, n (%)	27 (0.4)	7 (0.1)	7 (0.1)	8 (0.1)	3 (<0.1)
Total nonalcohol dietary energy,* kcal/d	1518.0 (649.8)	1498.0 (646.5)	1480 (578.3)	1504 (583.0)	1515.9 (554.6)
Diabetes, n (%)	219 (3.0)	162 (2.3)	195 (2.2)	132 (1.7)	97 (1.1)
Systolic blood pressure,* mm Hg	124.9 (18.0)	124.4 (18.0)	123.5 (17.9)	122.6 (17.7)	121.2 (17.4)
Diastolic blood pressure,* mm Hg	74.0 (9.4)	73.8 (9.2)	73.5 (9.1)	73.3 (9.0)	72.9 (8.9)
Waist circumference,* cm	79.6 (6.4)	76.5 (6.5)	74.8 (6.5)	73.2 (6.0)	70.7 (5.1)
Alcohol, never, n (%)	605 (8.4)	512 (7.1)	616 (7.0)	420 (5.3)	265 (3.0)
Diet score,* AHEI‐2010	44.8 (7.2)	49.7 (8.0)	53.9 (8.4)	57.9 (8.4)	63.6 (7.8)
Cigarette smoking, never, n (%)	2133 (29.6)	3532 (49.4)	4859 (55.2)	4681 (58.5)	6075 (67.9)
Physical activity,* MET‐h/wk	4.8 (6.7)	9.6 (10.8)	14.0 (12.8)	19.6 (15.2)	27.5 (17.1)
Quit smoking for health reason, n (%)	442 (6.1)	418 (5.8)	430 (4.9)	393 (4.9)	264 (3.0)
Quit alcohol for health reason, n (%)	263 (3.7)	233 (3.3)	238 (2.7)	164 (2.1)	111 (1.2)

Data represent numbers (percentage), unless otherwise specified. AHEI indicates Alternative Healthy Eating Index; CVD, cardiovascular disease; HLI, healthy lifestyle index; MET, metabolic equivalent; WHI, Women's Health Initiative; and WHIIRS, Women's Health Initiative Insomnia Rating Scale.

^*Data represent means (SD).

^† Only for cardiovascular disease cases.

^‡ Including only women in the nonintervention arm of the dietary modification trial.

The association of HLI with risk of CVD and its subtypes is shown in Table 2. In the age‐adjusted model, compared with the lowest quintile, all the other HLI quintiles showed inverse associations with the risk of developing incident disease. After adjusting for potential confounders, we found a statistically significant inverse dose–response association between the HLI and CVD risk. Results were similar for all CVD subtypes analyzed, with strong inverse trends in the association between the HLI and risk of stroke (fully adjusted HR _quintile−5 =0.64 [95% CI, 0.57–0.76], P‐trend <0.001—when compared with the hazard in the first quintile), CHD (fully adjusted HR _quintile−5 =0.38 [95% CI, 0.32–0.45], P‐trend <0.001), MI (fully adjusted HR _quintile−5 =0.40 [95% CI, 0.33–0.49], P‐trend <0.001), angina (fully adjusted HR _quintile−5 =0.46 [95% CI, 0.35–0.61], P‐trend <0.001), and coronary revascularization (fully adjusted HR _quintile‐5 =0.41 [95% CI, 0.34–0.49], P‐trend <0.001). The inverse associations persisted when death was considered as a competing risk (Table 2). The results of the sensitivity analyses supported the primary results. Specifically, the strength and significance of the associations between HLI and risk of CVD were not different for women with a follow‐up longer than 2 years or for women who maintained a normal BMI up to year 3 of the follow‐up (Table 3). Subgroup analysis by BMI categories and by age at baseline also indicated that increasing HLI quintiles were associated with significant inverse associations with CVD risk in all groups. Results were similar for healthy women and for those with previous chronic conditions such as hypertension, diabetes, or hyperlipidemia, when analyzed separately (Table 3). The exclusion of women with missing values on the covariables did not change the results of the analysis (Tables S3 and S4). Similar results were also observed when the relationship between HLI and CVD risk was analyzed separately by study components (OS and CT), or by sleep categories (Table S5).

Table 2.

Incidence Rates, Hazard Ratios, and 95% CIs for Risk of Cardiovascular Disease in Association With the Healthy Lifestyle Index Among Women With Body Mass Index Between 18.5 and 25.0 kg/m²

	Healthy lifestyle index categories
	1st	2nd	3rd	4th	5th	P trend
Outcome
Cardiovascular disease (all outcomes)
Cases	963	767	837	655	599
IR	8.92	6.69	5.69	4.79	3.77
HR (95% CI)*	1.00	0.70 (0.64–0.77)	0.60 (0.54–0.65)	0.49 (0.45–0.55)	0.41 (0.37–0.45)	<0.001
HR (95% CI)†	1.00	0.74 (0.67–0.81)	0.66 (0.60–0.72)	0.57 (0.51–0.63)	0.48 (0.43–0.54)	<0.001
SHR (95% CI)† , ‡	1.00	0.75 (0.68–0.83)	0.65 (0.59–0.71)	0.55 (0.50–0.61)	0.46 (0.41–0.51)	<0.001
Linear regression† , § HRSD (95% CI) 0.74 (0.72–0.77)
Cardiovascular disease (CHD, MI, and stroke)║
Cases	710	548	608	459	432
IR	6.57	4.78	4.13	3.36	2.72
HR (95% CI)*	1.00	0.67 (0.60–0.75)	0.58 (0.52–0.65)	0.46 (0.41–0.52)	0.39 (0.35–0.44)	<0.001
HR (95% CI)†	1.00	0.73 (0.65–0.82)	0.68 (0.61–0.76)	0.56 (0.50–0.63)	0.51 (0.45–0.58)	<0.001
SHR (95% CI)† , ‡	1.00	0.74 (0.66–0.83)	0.66 (0.59–0.73)	0.54 (0.48–0.61)	0.47 (0.41–0.53)	<0.001
Linear regression† , § HRSD (95% CI) 0.66 (0.62–0.70)
Stroke
Cases	328	294	308	255	251
IR	2.96	2.50	2.05	1.83	1.56
HR (95% CI)*	1.00	0.76 (0.66–0.91)	0.63 (0.54–0.74)	0.55 (0.47–0.65)	0.49 (0.42–0.58)	<0.001
HR (95% CI)†	1.00	0.85 (0.72–0.99)	0.74 (0.63–0.87)	0.67 (0.57–0.79)	0.64 (0.57–0.76)	<0.001
SHR (95% CI)† , ‡	1.00	0.87 (0.75–1.02)	0.73 (0.62–0.86)	0.67 (0.57–0.79)	0.61 (0.52–0.73)	<0.001
Linear regression† , § HRSD (95% CI) 0.82 (0.77–0.87)
Coronary heart disease
Cases	428	297	325	226	196
IR	3.86	2.53	2.17	1.62	1.22
HR (95% CI)*	1.00	0.61 (0.52–0.71)	0.52 (0.45–0.60)	0.38 (0.33–0.45)	0.30 (0.25–0.36)	<0.001
HR (95% CI)†	1.00	0.65 (0.56–0.76)	0.60 (0.52–0.70)	0.46 (0.39–0.54)	0.38 (0.32–0.45)	<0.001
SHR (95% CI)† , ‡	1.00	0.66 (0.57–0.77)	0.58 (0.51–0.67)	0.44 (0.38–0.52)	0.35 (0.30–0.42)	<0.001
Linear regression† , § HRSD (95% CI) 0.66 (0.62–0.70)
Myocardial infarction
Cases	304	216	229	173	154
IR	2.75	1.84	1.53	1.24	0.96
HR (95% CI)*	1.00	0.65 (0.54–0.77)	0.52 (0.44–0.62)	0.41 (0.34–0.50)	0.33 (0.27–0.40)	<0.001
HR (95% CI)†	1.00	0.66 (0.55–0.79)	0.58 (0.49–0.69)	0.48 (0.40–0.58)	0.40 (0.33–0.49)	<0.001
SHR (95% CI)† , ‡	1.00	0.68 (0.57–0.81)	0.57 (0.48–0.68)	0.47 (0.39–0.57)	0.38 (0.31–0.46)	<0.001
Linear regression† , § HRSD (95% CI) 0.67 (0.62–0.71)
Angina
Cases	165	122	129	111	87
IR	1.49	1.04	0.87	0.80	0.54
HR (95% CI)*	1.00	0.67 (0.54–0.85)	0.52 (0.42–0.66)	0.48 (0.38–0.61)	0.34 (0.26–0.44)	<0.001
HR (95% CI)†	1.00	0.71 (0.56–0.90)	0.64 (0.51–0.81)	0.62 (0.49–0.80)	0.46 (0.35–0.61)	<0.001
SHR (95% CI)† , ‡	1.00	0.71 (0.56–0.90)	0.60 (0.47–0.76)	0.56 (0.44–0.72)	0.40 (0.31–0.52)	<0.001
Linear regression† , § HRSD (95% CI) 0.75 (0.68–0.82)
Coronary revascularization
Cases	348	265	269	242	180
IR	3.14	2.26	1.79	1.74	1.12
HR (95% CI)*	1.00	0.66 (0.56–0.77)	0.51 (0.44–0.60)	0.50 (0.42–0.59)	0.33 (0.28–0.40)	<0.001
HR (95% CI)†	1.00	0.71 (0.61–0.84)	0.60 (0.51–0.71)	0.60 (0.50–0.71)	0.41 (0.34–0.49)	<0.001
SHR (95% CI)† , ‡	1.00	0.74 (0.63–0.87)	0.59 (0.50–0.70)	0.59 (0.50–0.69)	0.39 (0.33–0.47)	<0.001
Linear regression† , § HRSD (95% CI) 0.70 (0.65–0.74)

Incidence rate expressed per 1000 person‐years. CHD indicates coronary heart disease; HR, hazard ratio; MI, myocardial infarction; IR, incidence rate; and SHR, subhazard ratio.

^*Model stratified by age (5‐year strata) and adjusted for age at baseline.

^† Models stratified by age (5‐year strata) and adjusted for age, total nonalcohol energy daily intake, race, education, income, marital status, insomnia, sleep quality, history of diabetes, participation in Women's Health Initiative clinical trials: enrollment in the dietary trial nonintervention arm, enrollment in the calcium and vitamin D trial and intervention arm, enrollment in the menopausal hormone therapy trial and intervention arm, ever use of menopausal hormone therapy, use of aspirin, use of antihypertensive drugs, use of lipid‐lowering drugs, height, years since menopause, systolic blood pressure, diastolic blood pressure, quit cigarette smoking for health reasons, and quit alcohol for health reasons.

^‡ Estimates obtained using Fine–Gray's competing risk model with non‐CHD deaths as competing outcome.

^§ Linear regression per healthy lifestyle index SD unit increase (3.75 healthy lifestyle index unit).

^║ Cardiovascular disease includes only coronary heart disease, myocardial infarction, and stroke.

Table 3.

Incidence Rates, Hazard Ratios, and 95% CIs for Risk of Cardiovascular Disease in Association With the Healthy Lifestyle Index Among Subgroups of Women Enrolled in the WHI

	Healthy Lifestyle Index categories					P trend
	1st	2nd	3rd	4th	5th
Restricted to women with follow‐up >2 y (n=39 485)
CVD cases	891	717	777	608	561
IR	8.3	6.31	5.34	4.49	3.52
HR (95% CI)*	1.00	0.70 (0.64–0.78)	0.59 (0.54–0.65)	0.49 (0.44–0.54)	0.41 (0.36–0.45)	<0.001
HR (95% CI)†	1.00	0.74 (0.67–0.81)	0.66 (0.59–0.72)	0.56 (0.51–0.62)	0.48 (0.43–0.54)	<0.001
Restricted to women with BMI=18.5–<25.0 kg/m2 at baseline and at year 3 of the follow‐up (n=28 851)
CVD cases	573	528	570	498	486
IR	8.31	6.51	5.24	4.66	3.71
HR (95% CI)*	1.00	0.73 (0.65–0.82)	0.59 (0.53–0.67)	0.52 (0.46–0.59)	0.44 (0.39–0.49)	<0.001
HR (95% CI)†	1.00	0.77 (0.68–0.86)	0.66 (0.58–0.74)	0.60 (0.53–0.68)	0.52 (0.45–0.59)	<0.001
Excluding participants in the intervention arms of the postmenopausal hormone therapy trial and the calcium and vitamin D trial (n=35 036)
CVD cases	713	600	682	551	498
IR	8.17	6.17	5.31	4.53	3.46
HR (95% CI)*	1.00	0.72 (0.64–0.80)	0.62 (0.55–0.68)	0.52 (0.46–0.58)	0.41 (0.37–0.47)	<0.001
HR (95% CI)†	1.00	0.75 (0.67–0.84)	0.68 (0.61–0.76)	0.60 (0.53–0.67)	0.49 (0.44–0.55)	<0.001
Restricted to women with BMI=18.5‐<22.0 kg/m2 (n=13 649)
CVD cases	193	193	230	250	284
IR	9.62	6.83	4.87	4.74	3.61
HR (95% CI)*	1.00	0.65 (0.53–0.79)	0.45 (0.38–0.55)	0.44 (0.36–0.53)	0.34 (0.28–0.41)	<0.001
HR (95% CI)†	1.00	0.67 (0.55–0.83)	0.49 (0.41–0.60)	0.49 (0.40–0.59)	0.40 (0.33–0.48)	<0.001
Restricted to women with BMI ≥22.0‐<25.0 kg/m2 (n=26 469)
CVD cases	770	574	608	405	315
IR	8.80	6.71	6.98	4.83	3.92
HR (95% CI)*	1.00	0.71 (0.64–0.79)	0.65 (0.59–0.73)	0.51 (0.45–0.57)	0.44 (0.39–0.50)	<0.001
HR (95% CI)†	1.00	0.75 (0.67–0.84)	0.72 (0.65–0.81)	0.59 (0.52–0.66)	0.52 (0.45–0.59)	<0.001
Restricted to women age ≤ 63 y (n=21 088)
CVD cases	359	207	204	167	167
IR	5.61	3.24	2.39	2.21	1.82
HR (95% CI)*	1.00	0.55 (0.47–0.66)	0.43 (0.36–0.51)	0.38 (0.32–0.46)	0.33 (0.27–0.39)	<0.001
HR (95% CI)†	1.00	0.66 (0.52–0.73)	0.51 (0.43–0.61)	0.49 (0.40–0.59)	0.43 (0.36–0.52)	<0.001
Restricted to women age >63 y (n=19 030)
CVD cases	604	560	634	488	432
IR	13.84	11.42	10.13	8.19	6.58
HR (95% CI)*	1.00	0.79 (0.70–0.88)	0.69 (0.62–0.77)	0.56 (0.49–0.63)	0.45 (0.40–0.51)	<0.001
HR (95% CI)†	1.00	0.81 (0.72–0.91)	0.75 (0.67–0.84)	0.62 (0.55–0.70)	0.52 (0.46–0.59)	<0.001
Restricted to women without hypertension, or diabetes, or taking antihypertensive and/or lipid‐lowering medications (n=27 231)
CVD cases	456	381	455	350	358
IR	6.50	5.02	4.42	3.59	3.01
HR (95% CI)*	1.00	0.71 (0.62–0.81)	0.63 (0.55–0.72)	0.49 (0.43–0.57)	0.43 (0.37–0.49)	<0.001
HR (95% CI)‡	1.00	0.73 (0.63–0.83)	0.68 (0.60–0.78)	0.55 (0.48–0.64)	0.49 (0.43–0.57)	<0.001
Restricted to women with hypertension, or diabetes, or taking antihypertensive and/or lipid‐lowering medications (n=12 887)
CVD cases	507	386	382	305	241
IR	13.22	10.20	8.73	7.91	6.29
HR (95% CI)*	1.00	0.72 (0.63–0.82)	0.60 (0.53–0.69)	0.55 (0.48–0.63)	0.44 (0.38–0.51)	<0.001
HR (95% CI)‡	1.00	0.76 (0.66–0.86)	0.65 (0.56–0.74)	0.60 (0.51–0.69)	0.48 (0.41–0.57)	<0.001

Incidence rate expressed per 1000 person‐years. BMI indicates body mass index; CVD, cardiovascular disease; HR, hazard ratio; IR, incidence ratio; and WHI, Women's Health Initiative.

^*Model stratified for age (5‐year strata) and adjusted for age at baseline.

^† Models stratified for age (5‐year strata) adjusted for age, total nonalcohol energy daily intake, race, education, income, marital status, insomnia, sleep quality, history of diabetes, participation in WHI clinical trials: enrollment in the dietary trial nonintervention arm, enrollment in the calcium and vitamin D trial and intervention arm, enrollment in the menopausal hormone therapy trial and intervention arm, ever use of menopausal hormone therapy, use of aspirin, use of antihypertensive drugs, use of lipid‐lowering drugs, height, years since menopause, systolic blood pressure, diastolic blood pressure, quit cigarette smoking for health reasons, and quit alcohol for health reasons.

^‡ Models stratified for age (5‐year strata) adjusted for age, total nonalcohol energy daily intake, race, education, income, marital status, insomnia, sleep quality, participation in WHI clinical trials: enrollment in the dietary trial nonintervention arm, enrollment in the calcium and vitamin D trial and intervention arm, enrollment in the menopausal hormone therapy trial and intervention arm, ever use of hormone replacement therapy, use of aspirin, height, year since menopause, systolic blood pressure, diastolic blood pressure, quit cigarette smoking for health reasons, quit alcohol for health reasons.

Analysis of individual HLI components showed inverse dose–response associations with CVD risk for each factor except for alcohol intake (Figure). Furthermore, excluding one component at a time in turn from the HLI showed similar associations between increasing score and reduced risk of CVD (Table 4).

Figure . Association between individual Healthy Lifestyle Index components and risk of cardiovascular disease in women with BMI=18.5–<25.0 kg/m².

Models adjusted for age, total nonalcohol energy daily intake, race, education, family income, marital status, history of diabetes, trial participation, ever use of hormone replacement therapy, use of aspirin, use of antihypertensive drugs, use of lipid‐lowering drugs, use of cardiac glycosides, height, year since menopause, systolic blood pressure, diastolic blood pressure, quit cigarette smoking for health reasons, and quit alcohol for health reasons. Individual healthy lifestyle factors were included in the same model. AHEI indicates Alternative Healthy Eating Index; CVD, cardiovascular disease; HR, hazard ratio; and MET, metabolic equivalent.

Table 4.

Hazard Ratios and 95% CI for Risk of Cardiovascular Disease in Association With the Healthy Lifestyle Index Among Women With Body Mass Index Between 18.5 and 25.0 kg/m², Excluding Each Lifestyle Component in Turn

	Healthy Lifestyle Index categories					P trend
	1st	2nd	3rd	4th	5th
HLI without smoking
IR*	8.72	7.04	6.21	4.68	3.79
HR (95% CI)†	1.00	0.83 (0.76–0.91)	0.75 (0.67–0.84)	0.59 (0.54–0.66)	0.51 (0.46–0.57)	<0.001
HR (95% CI)‡ , §	1.00	0.87 (0.79–0.96)	0.83 (0.74–0.94)	0.69 (0.62–0.76)	0.62 (0.56–0.70)	<0.001
HLI without alcohol
IR*	8.50	6.62	5.61	4.89	3.68
HR (95% CI)†	1.00	0.73 (0.66–0.79)	0.60 (0.54–0.67)	0.53 (0.49–0.59)	0.42 (0.38–0.47)	<0.001
HR (95% CI)‡ , ║	1.00	0.76 (0.70–0.83)	0.65 (0.58–0.73)	0.59 (0.54–0.65)	0.48 (0.43–0.54)	<0.001
HLI without physical activity
IR*	8.22	6.03	5.47	4.73	3.92
HR (95% CI)†	1.00	0.69 (0.63–0.76)	0.61 (0.55–0.69)	0.53 (0.48–0.59)	0.46 (0.41–0.52)	<0.001
HR (95% CI)‡ , ¶	1.00	0.73 (0.66–0.80)	0.66 (0.59–0.75)	0.59 (0.53–0.66)	0.53 (0.47–0.60)	<0.001
HLI without diet
IR *	9.20	5.93	5.43	4.81	3.59
HR (95% CI)†	1.00	0.68 (0.62–0.75)	0.60 (0.54–0.67)	0.54 (0.49–0.60)	0.44 (0.39–0.50)	<0.001
HR (95% CI)‡ , **	1.00	0.72 (0.65–0.79)	0.65 (0.58–0.73)	0.59 (0.53–0.66)	0.50 (0.44–0.57)	<0.001
HLI without waist circumference
IR*	8.30	6.61	5.71	4.93	4.20
HR (95% CI)†	1.00	0.71 (0.65–0.79)	0.60 (0.54–0.66)	0.51 (0.46–0.56)	0.44 (0.40–0.50)	<0.001
HR (95% CI)‡ , ††	1.00	0.75 (0.68–0.82)	0.64 (0.58–0.71)	0.58 (0.52–0.64)	0.51 (0.46–0.57)	<0.001

HR indicates hazard ratio; and IR, incidence rate.

^*Unadjusted incidence rate.

^† Model adjusted for age at enrollment in the study.

^‡ Models adjusted for age, total nonalcohol energy daily intake, race, education, income, marital status, insomnia, sleep quality, history of diabetes, participation in Women's Health Initiative (WHI) clinical trials: enrollment in the dietary trial non‐intervention arm, enrollment in the calcium and vitamin D trial and intervention arm, enrollment in the menopausal hormone therapy trial and intervention arm, ever use of menopausal hormone therapy, use of aspirin, use of antihypertensive drugs, use of lipid‐lowering drugs, height, years since menopause, systolic blood pressure, diastolic blood pressure, quit cigarette smoking for health reasons, and quit alcohol for health reasons.

^§ Model additionally adjusted for categories of smoking.

^║ Model additionally adjusted for categories of alcohol intake.

^¶ Model additionally adjusted for quintiles of physical activity.

^**Model additionally adjusted for quintiles of diet score (Alternative Healthy Eating Index‐2010).

^†† Model additionally adjusted for quintiles of waist circumference.

A moderate negative correlation was observed between HLI and total body fat (r=−0.34), fat percentage (r=−0.30), trunk fat (r=−0.39), and trunk/leg fat ratio (r=−0.30), while the correlation with leg fat was weak (r=−0.11) (Table S6). Analysis of the relationship between HLI and measures of total and regional body fat distribution revealed an inverse association between the HLI and total body fat as an absolute amount (kg) and as a percentage of body weight, and similarly for trunk fat (kg), leg fat (kg) and trunk/leg fat ratio (Table S7). After introducing these variables (separately) into the primary multivariable model, the significant inverse association between HLI quintiles and risk of CVD persisted. However, the estimates for the HR _quintiles were somewhat attenuated after trunk/leg fat ratio adjustment, which also showed a significant association with the risk of CVD in the same model (Table S8).

Discussion

In this study, among postmenopausal women with normal BMI, a relatively high HLI score, based on a combination of modifiable behavioral factors including diet quality, physical activity, smoking history and amount, alcohol consumption, and waist circumference, was significantly inversely associated with the risk of incident primary CVD overall, and with risk of stroke, CHD, MI, angina, and coronary revascularization, over a median follow‐up period of ≈20 years. The results remained unchanged when analyses were restricted to women with > 2 years of follow‐up or to those who maintained a normal BMI over approximately the first 3 years of follow‐up. Similarly, subgroup analyses by BMI, age, health status categories, WHI study components, and sleep quality showed consistent inverse associations between HLI and CVD risk across all subgroups. The exclusion of one HLI component at a time from the combined HLI score did not alter the relationship between a healthy lifestyle and CVD, suggesting that no specific behavioral habit alone was responsible for the observed association.

With the exception of alcohol consumption, analysis of the association between individual HLI components and risk of CVD indicated that healthier behaviors were associated with reduced CVD risk. For alcohol level, nondrinkers had a suggestive increase in risk of several types of CVD compared with those drinking, although no statistically significant difference among the categories was detected. Previous studies of the association between alcohol consumption and CVD risk found that compared with individuals who drank moderately (2–3 alcohol units/d), nondrinkers had an increased risk of several types of CVD, while those exceeding a moderate amount had a reduced risk of certain types of CVD but an increased risk of others. ²⁵ , ³⁵ In the subgroup of women with data on body composition measured using DXA, we found inverse dose–response associations between HLI levels and several measures of body composition and fat distribution. Adding these variables to the primary analysis for the subgroup with DXA measures resulted in a modest attenuation of the association between HLI and CVD, which remained significant in all the models, while the trunk‐to‐leg fat ratio was also significantly associated with an increased risk of CVD.

To the best of our knowledge, this is the first study that focuses on the association of HLI with CVD risk in postmenopausal women with normal BMI. In contrast, the association of lifestyle with risk of CVD has been studied extensively in subjects across the full BMI spectrum. A previous study conducted in the WHI cohort examined a combined cardiovascular health score derived from 7 factors including smoking, BMI, physical activity, diet, total cholesterol, BP, and fasting glucose in association with the risk of CVD and found that compared with women in the highest score, those with the lowest had almost 7 times higher risk of developing CVD. ³⁶ A meta‐analysis including 22 high‐quality prospective/longitudinal studies examined the relationship between at least 3 of smoking, alcohol consumption, physical activity, diet, and body weight, and the risk of overall CVD, or CHD, stroke, and heart failure in subjects CVD‐free at baseline. ¹⁶ The study showed that a higher healthy lifestyle index was associated with a reduced risk of CVD (66%), stroke (60%), and heart failure (69%). A more recent meta‐analysis, which included some of the same studies, ¹⁷ showed a 62% reduction in risk of incident CVD and between 55% to 71% reductions in risk of stroke, atrial fibrillation, CAD, and heart failure for those with higher lifestyle index scores compared with those with the lowest scores. The results were consistent across countries, race and ethnicity, and socioeconomic status.

Among the modifiable risk factors associated with the incidence of CVD in women, cigarette smoking and excessive body weight have the highest attributable risks. ⁶ , ³⁷ However, while in several Western countries smoking prevalence has declined in the past few decades, especially among middle‐aged and older women, excessive weight has been on the rise worldwide, accounting for a higher percentage of CVD cases. ³⁸ This is of great importance during menopause, when weight gain and changes in body composition often occur. ³⁹ The increasing prevalence of elevated BMI has led to the recommendation to adopt a healthy lifestyle which includes being physically active and consuming a healthy diet mainly aimed at weight loss and/or its maintenance within a normal “healthy” range. ⁴⁰ While obtaining and/or maintaining a normal BMI is an important lifestyle goal, our results indicate that modifying lifestyle behaviors may be important to significantly reduce the risk of CVD even among women with normal BMI. Indeed, we found that compared with those with the lowest level of HLI, even those with intermediate HLI scores, resulting from various levels of adherence to different lifestyle behaviors, have a lower CVD risk. This trend was present in various subgroups of postmenopausal women, independently of their BMI, age, or health status, despite the different baseline risks of the referent categories, as indicated by the reported incidence rates. In addition, we showed that the benefits associated with adopting an overall healthy lifestyle did not depend exclusively on the presence of a specific factor, as demonstrated when each HLI component was excluded in turn from the index and included separately in the model.

Several biological mechanisms and metabolic markers might mediate the inverse association between HLI and CVD risk in postmenopausal women with normal BMI. The amount of body fat, especially abdominal, is associated with an increased risk of CVD across the whole spectrum of BMI. ¹⁹ , ⁴¹ A combination of a high‐quality diet (characterized by a high intake of fruits, vegetables, and whole grains and limited consumption of red and processed meat), moderate alcohol consumption, regular moderate‐to‐vigorous physical activity, and smaller waist circumference may be related to less total body and abdominal fat, known risk factors for CVD. ¹⁹ , ⁴² Among a subgroup of women with DXA data on body fat distribution, we showed for the first time that a relatively high HLI is inversely associated with the whole body, trunk, and leg fat. When the variables representing DXA measures of body fat were included separately in the analysis examining the relationship between HLI quintiles and the risk of CVD, only trunk/leg fat ratio was significantly associated with an increased risk of CVD while slightly attenuating the estimate of the association between HLI upper quintiles and the risk of CVD. These results may suggest that trunk/leg fat ratio has an independent effect on the risk of CVD while also being part of the mechanism underlying the relationship between the HLI and this outcome. Healthier lifestyle behaviors are also associated with lower levels of inflammatory markers such as interleukin‐6, tumor necrosis factor, C‐reactive protein, lipid levels, and white cell counts ⁴³ , ⁴⁴ , ⁴⁵ and with reductions in circulating glucose level, fasting insulin concentration and insulin‐resistance, which contribute to the pathophysiology of atherosclerotic CVD. ⁴⁶ , ⁴⁷ In this analytical sample, we did not have a consistent number of women with all biomarker data to address the possible role of these analytes in the HLI‐CVD association.

The present study has several strengths, including the large sample size and the prospective design with an extensive follow‐up period and substantial participant retention. Anthropometric measurements at baseline and follow‐up visits were obtained in the clinical setting by trained personnel. HLI was calculated considering all modifiable lifestyle factors indicated by the American Heart Association to be linked to the risk of CVD. For each behavioral component, we used a multilevel score, which allowed us to test the relationship of HLI with CVD risk in a dose‐dependent manner. For a large portion of the study duration, outcome ascertainment and adjudication were based on a review of medical and hospital records by trained physicians. Several sociodemographic and health‐related characteristics collected at baseline were included in the analysis as potential confounders. Compared with the entire population of WHI participants, the analytical sample for this study included predominantly White women with a lower percentage of Black and Hispanic women. Given that the latter 2 ethnic groups tend to have, on average, a higher BMI, the present results may not be generalizable to the entire population of postmenopausal women. We also tested the association of HLI and CVD risk in a competing risk model using deaths other than from CHD as the competing risk to limit the possibility that the observed findings resulted from selective mortality associated with the exposure over such an extended follow‐up period. A major limitation of the study is a lack of information on changes in lifestyle during follow‐up which might have altered the observed associations. In particular, some of the women in this study population may not have maintained a normal BMI after enrollment in the study. To reduce this possibility, we conducted a sensitivity analysis excluding women who did not have a normal BMI after an average of 3 years of follow‐up; data on BMI at a later exam were available only for a limited number of women. In addition, biomarkers such as lipid profile, fasting glucose, and insulin were available only for a limited number of women and could not be included in the statistical models as covariates. Another limitation of the study is that only a portion of women had DXA data on body fat distribution.

The current study showed that higher HLI is associated with a reduced CVD risk among women with a normal BMI. This group is usually considered to be at lower risk for this disease, so the findings of the present study suggest a potentially critical role of lifestyle behavior in reducing the risk of incident primary CVD beyond control of body weight. From a clinical and public health perspective, this is an important finding. Our findings also suggest that body fat distribution may only in part be involved in the mechanisms that regulate the association between HLI and CVD risk, indicating that other protective mechanisms, not measured in this study, may play a role.

The results of this study suggest that in postmenopausal women with normal BMI, a healthy lifestyle including a high‐quality diet, moderate or intense physical activity, no current smoking, moderate alcohol intake, and a small waist circumference, is inversely associated with the risk of CVD and its subtypes, including stroke, CHD, MI, angina, and coronary revascularization.

Sources of Funding

The WHI program is funded by the National Heart, Lung and Blood Institute, National Institutes of Health, US Department of Health and Humans Services.

Disclosures

Dr T.E. Rohan receives consulting fees (unrelated to the present work) and holds stock options from Health Outlook Corporation. Dr M. La Monte receives financial compensation as associate editor of Medicine & Science in Sports & Exercise (<$1000/year). Dr H.A. Tindle served as 2018 to 2021 Board Member of the North American Quitline Consortium. The remaining authors have no disclosures to report.

Supporting information

Tables S1–S8