Abstract
Background:
Prior research indicates that food insecurity is negatively associated with cardiometabolic health and plays a role in cardiovascular disease (CVD) risk, which also increases with age. Food insecurity disproportionately affects aging women in the United States, who face higher poverty and economic insecurity rates than their male counterparts, increasing their risk for hunger, poor nutrition, and chronic conditions. We examined the association of food security and assistance with CVD morbidity and mortality risks in postmenopausal women through a target trial emulation.
Methods:
A retrospective cohort study was conducted among 41,227 postmenopausal women from the Women’s Health Initiative Observational Study free of CVD at baseline. Approximately 96% reported food security, and 3% reported reliance on food assistance at six years of follow-up (analytic baseline). Furthermore, 16% of 1,861 women reporting food insecurity also reported food assistance. Inverse Probability of Treatment Weighting (IPTW) within Cox regression models was performed using demographic, socioeconomic, lifestyle, and health characteristics collected prior to the analytic baseline.
Results:
Overall, 6,037 eligible women experienced CVD outcomes over ~24 years of follow-up. In IPTW Cox regression models, food security (hazard ratio [HR]=0.65, 95% confidence interval (CI): 0.47, 0.89), and food assistance (HR=0.47, 95% CI: 0.46, 0.48) were inversely related to CVD risk. Similar results were obtained for all-cause and chronic disease-specific mortality risks.
Conclusion:
Food security and assistance are generally associated with lower CVD risks after menopause, with implications for food assistance programs, targeted nutritional counseling, community interventions, and holistic care among aging women in the United States.
Keywords: Cardiovascular disease, Food assistance, Food security, Menopause, Target Trial Emulation
INTRODUCTION:
There is growing evidence that the lack of food security may be detrimental to cardiometabolic health and can promote the development of chronic diseases in a wide range of populations1–3. For instance, a cross-sectional study involving 7,640 adults from the National Health and Nutrition Examination Surveys (NHANES) evaluated the link between household food security and allostatic load (AL), a measure of chronic physiological strain often assessed using biomarkers of cardiovascular, metabolic, immune, and neuroendocrine function, and found that adults living in marginally food-secure or -insecure households had significantly higher mean AL scores than those in food-secure households2. Similarly, elevated AL scores were present among Supplemental Nutrition Assistance Program (SNAP) participants and Hispanic women in marginally food-secure households2.
Food insecurity disproportionately affects aging women in the U.S., who face higher poverty and economic insecurity rates than their male counterparts, increasing their risk for hunger, poor nutrition, and chronic conditions4. Conversely, food security and access to nutrition assistance programs may play a critical role in shaping cardiovascular disease (CVD) risk among aging women, as inadequate access to healthy foods is associated with poorer diet quality, obesity and its associated cardiometabolic risk factors1–3.
A limited number of studies have addressed differences in the relationship between food security, food assistance, and CVD outcomes across demographic groups1 2, and even fewer focused on women5. Notably, no studies have comprehensively examined these relationships in postmenopausal women, a high-risk group for CVD development. We focus on this population and constructs because postmenopausal women face heightened cardiovascular risk due to aging and the decline of hormonal protection6 7. Food insecurity exacerbates this risk via poor diet, chronic stress, obesity, hypertension, diabetes, smoking, and limited healthcare access8 9. The role of dietary patterns in CVD has been previously examined among postmenopausal women who participated in the Women’s Health Initiative (WHI), with evidence suggesting that poor diet quality may increase CVD morbidity and mortality risks10 11. For instance, an analysis of postmenopausal women enrolled in the WHI Observational Study (WHI-OS) showed that women in the best vs. worst category of the Dietary Modification Index12 had hazard ratios (HRs) of 0.88 for incident cardiovascular disease (CVD) and 0.91 for heart failure (HF), while those in the best vs. worst category of the Alternate Healthy Eating Index (AHEI) had HRs of 0.77 for CVD and 0.70 for HF13.
In this target trial emulation of women from the WHI-OS described below, we examined if achieving food security and reliance on food assistance was associated with future CVD morbidity and mortality among postmenopausal women.
METHODS:
Database:
From October 1, 1993—December 21, 1998, the WHI enrolled and collected data on a multiethnic and multi-racial sample of 161,808 postmenopausal women 50 to 79 years of age from 40 clinical centers in 24 U.S. states and the District of Columbia. The study design, eligibility criteria, recruitment methods and measurement protocols have been described elsewhere14 15. Briefly, two components of the WHI were the clinical trials (WHI-CTs) (n=68,132) and the WHI-OS (n=93,676). All WHI participants – WHI-CTs and WHI-OS – completed the same assessments at enrollment that covered background characteristics and some of these characteristics were also evaluated at later follow-up visits (Appendix A). The main WHI study period, including follow-up, occurred between 1993 and 2005. Of 150,076 participants who were in active follow-up after the main WHI study ended in 2005, 115,400 (76.9%) participated in Extension Study 1 (2005—2010) and 93,122 (86.9%) of those eligible participated in Extension Study 2 (2010—Present)16 17. The WHI received institutional review board approval with informed consent from all participants at clinical centers, the Coordinating Center, and the National Institute of Health, in compliance with the Helsinki Declaration as revised in 1983.
Target Trial Emulation Framework:
A target trial emulation uses observational data to mimic a randomized controlled trial (RCT) and estimate causal effects and seeks to improve causal inference from observational data when a RCT is not possible, ethical, or practical18. It involves developing a target trial to address a specific question and employing non-randomized data to replicate its protocol18. The specifics of the target trial emulation approach, including eligibility criteria, treatment strategies, treatment assignment, follow-up, outcomes, causal contrast, and statistical analysis are displayed in Table 1. Our analysis was designed to emulate a RCT whereby eligible postmenopausal women were randomly assigned based on two alternative treatment strategies at year 6 of follow-up (analytic baseline): (1) food security achieved vs. food security not achieved and (2) reliance on food assistance vs. no reliance on food assistance, and were followed over time until first occurrence of an outcome, loss to follow-up, death, or end of follow-up, whichever came first.
Table 1.
Target trial specification and emulation to estimate the intention-to-treat effect of food security and food assistance on CVD morbidity and mortality risks
| Protocol component | Hypothetical target trial | Observational Study |
|---|---|---|
| Eligibility criteria | Participated in the Women’s Health Initiative Observational Study [criterion 1] Responded to year 6 follow-up (analytic baseline) [criterion 2] Reported presence or absence of food security at analytic baseline [criterion 3] Reported reliance or non-reliance on food assistance at analytic baseline [criterion 4] Had no missing data on demographic, socioeconomic, lifestyle, and health characteristics [criterion 5] Followed for at least 12 months after analytic baseline [criterion 6] No history of CVD prior to analytic baseline [criterion 7] |
Same as hypothetical target trial |
| Treatment strategies | Food security (experimental) vs. No food security (control) at analytic baseline Food assistance (experimental) vs. No food assistance (control) at analytic baseline |
Food security (exposed) vs. No food security (non-exposed) at analytic baseline Food assistance (exposed) vs. No food assistance (non-exposed) at analytic baseline |
| Treatment assignment | Random assignment to either experimental or control group | Non-random assignment to either exposed or non-exposed group Propensity scoring (Inverse Probability of Treatment Weighting) between exposed and non-exposed groups on selected demographic, socioeconomic, lifestyle, and health characteristics evaluated at or prior to analytic baseline |
| Follow-up | Maximum follow-up time from analytic baseline is ~24 years, with censoring at loss to follow-up, death, or the latest available follow-up date | Same as hypothetical target trial |
| Outcomes | First occurrence of outcome (fatal or non-fatal CVD, all-cause death, chronic disease-specific death) from analytic baseline until the latest available follow-up date | Same as hypothetical target trial |
| Causal contrast | Hazard ratio | Same as hypothetical target trial |
| Statistical analysis | Cox regression model | Same as hypothetical target trial |
Abbreviations: CVD = Cardiovascular disease.
Population:
Study-eligible women included those who met several criteria of the target trial emulation framework, namely, WHI-OS participants (criterion 1) with year 6 follow-up (analytic baseline) data (criterion 2), no missing data on exposure statuses [food security and food assistance] assessed at analytic baseline (criteria 3–4), no missing data on covariates assessed at WHI enrollment (1993—1998) and/or analytic baseline (criterion 5), >12 months of follow-up from analytic baseline until censoring or end of follow-up (criterion 6), and no self-reported CVD-related history [transient ischemic attack, stroke, angina, myocardial infarction, cardiac arrest, heart failure, cardiac catheterizations, heart or coronary bypass surgery, angioplasty of coronary arteries, carotid endarterectomy/carotid angioplasty, atrial fibrillation, and/or aortic aneurysm] at enrollment (1993—1998) or CVD event between enrollment (1993—1998) and analytic baseline (criterion 7).
Measures:
Treatment strategies and assignment:
Food security and assistance were ascertained at the year 6 follow-up visit among WHI-OS, but not WHI-CTs, participants. The year 6 (analytic baseline for our analysis) follow-up questionnaire was administered to 79,885 WHI-OS participants. When asked [“Which one of these statements best describes the food eaten in your household in the last year?”], 73,803 reported food security (enough to eat of desired kind), whereas 4,552 reported food insecurity (4,254 [enough to eat not always of desired kind], 230 [sometimes did not have enough food], and 68 [often did not have any food to eat]), and 1,530 had missing data on this item. Irrespective of food security, 3,131 reported reliance on any food assistance [United States Department of Agriculture (USDA) or government commodity foods, Meals on Wheels, Food Stamps, and Food Bank, among others], 74,696 reported non-reliance on food assistance, and 2,058 had missing data on this item. The recall period for these items was the prior year. Study-eligible women in the emulated target trial were randomly assigned to one of two treatment strategies examined separately at analytic baseline, namely, ‘food security achievement’ (Yes vs. No) and ‘reliance on food assistance’ (Yes vs. No). In this retrospective cohort study, women who met pre-specified eligibility criteria were non-randomly assigned at analytic baseline to one of two treatment strategies. We also examined ‘reliance on food assistance’ (Yes vs. No) among women who responded ‘No’ to ‘food security achievement’. To address confounding, we used demographic, socioeconomic, lifestyle, and health characteristics described below to perform propensity scoring, i.e. inverse probability of treatment weighting (IPTW), and subsequently compared women selected into exposed vs. non-exposed groups at analytic baseline on outcomes described below.
Outcomes:
This study examined three physician-adjudicated outcomes19: 1) CVD morbidity and mortality, 2) all-cause and 3) chronic disease-specific mortality risks (Appendix A). Fatal and non-fatal CVD outcomes included coronary heart disease, angina, coronary artery bypass graft, percutaneous transluminal coronary angioplasty, myocardial infarction (clinical, definite silent, possible silent, total), coronary revascularization, transient ischemic attack, ischemic stroke, hemorrhagic stroke, congestive heart failure, carotid artery disease, or CVD death. Deaths were ascertained via semi-annual or annual follow-up telephone calls by trained staff with family, friends, and medical care providers, National Death Index and obituaries20–22. The underlying cause of death was used to define CVD-specific and chronic disease-specific mortality, as described in Appendix A20–22.
Covariates:
Potential confounders included age [in years] at analytic baseline, region of residence [Midwest, Northeast, South, West], self-reported race [White, Black, Asian, Other, Unknown/Not reported], self-reported ethnicity [Hispanic, non-Hispanic, Unknown/Not reported], and marital status [Married/Partnered, Single, Divorced, Widowed]) at WHI enrollment. Socioeconomic position at WHI enrollment was calculated based on education [less than high school, high school, some college, completed college or higher level], occupation [managerial/professional, technical/sales/administrative, service/labor, homemaker/not working/retired/disabled], household income [< $20,000, $20,000-$49,999, $50,000-$99,999, ≥$100,000], and neighborhood socioeconomic status index, using z-scoring, proration, and principal component analyses23. As described elsewhere, the combination of individual-level and neighborhood-level socioeconomic characteristics resulted in a continuous variable that was defined in tertiles (low, medium, and high) of socioeconomic position23. Lifestyle characteristics at WHI enrollment included smoking status [Never Smoker, Past Smoker, Current Smoker], alcohol consumption [Non-Drinker, Former Drinker, < 1 drink/week, ≥ 1 drink/week], diet quality measured by the 2020 Healthy Eating Index (HEI-2020)24–26, and recreational physical activity [Metabolic equivalent-hours/week] measured by the International Physical Activity Questionnaire. Health characteristics at WHI enrollment included body mass index [BMI] (kg/m2), number of comorbid conditions27 (0, 1–2, 3+), depressive symptoms score28 29, and self-rated health [Excellent/Very Good/Good, Fair/Poor]. Self-reported change in alcohol consumption (Yes, No) at analytic baseline, change in smoking status (Non-Smoker to Smoker, Smoker to non-Smoker, No change) between WHI enrollment and analytic baseline, as well as changes in physical activity (Metabolic equivalent-hours/week) and weight (kg) between WHI enrollment and analytic baseline, were also examined as potential confounders (Appendix A). Trained staff collected anthropometric data, including weight and height measurements at WHI enrollment, whereas weight was self-reported at analytic baseline30 31. Besides smoking status, alcohol consumption, physical activity, and weight, background characteristics used to define covariates were assumed not to vary between WHI enrollment and analytic baseline.
Time:
Study-eligible women were followed from analytic baseline until first occurrence of an outcome, loss to follow-up, death, or the end of follow-up (February 15, 2025). Furthermore, all covariates used for propensity scoring, i.e. demographic, socioeconomic, lifestyle, and health characteristics, were from analytic baseline to avoid immortal time bias18.
Propensity score model:
Propensity scoring helps ensure comparability between groups in observational studies, facilitating causal inference by accounting for prognostic characteristics that associate with outcomes of interest32 33, and can be applied in the context of IPTW. Using the analytic sample without IPTW, we confirmed that background characteristics were confounders of hypothesized relationships by being associated with key exposure (Table 2) and outcome (Appendix B; Tables S.1–S.2.) variables. Second, we explored traditional and machine learning techniques [using forward, backward, and stepwise logistic regression (SAS HPLOGISTIC procedure) and Least Absolute Shrinkage and Selection Operator (LASSO) (SAS HPGENSELECT procedure)] to calculate propensity scores as predicted probabilities of ‘food security achievement’ and ‘reliance on food assistance’ for each woman. Subsequently, we defined these exposure variables as the outcome variable in multiple logistic regression models, whereby a subset of demographic, socioeconomic, lifestyle, and health characteristics were selected for entry as predictors based on Akaike Information Criteria minimization (Appendix B; Table S.3.). Third, propensity scores were used to perform IPTW analyses within Cox regression models, after verifying that an overlap in propensity scores existed between treatment groups in the non-weighted analytic sample (Appendix B; Table S.4.).
Table 2.
Demographic, socioeconomic, lifestyle, and health characteristics before or at analytic baseline by food security and food assistance in the non-weighted analytic sample
| Food security (n=41227) | Food assistance (n=41227) | |||
|---|---|---|---|---|
| Food security [Yes] (n=39366) | Food security [No] (n=1861) | Food assistance [Yes] (n=1203) | Food assistance [No] (n=40024) | |
| WHI-OS 6-year follow-up visit: | ||||
| Age (years) at analytic baseline: | ||||
| P=0.052 | P<0.0001 | |||
| Mean ± SD | 68.60 ± 7.32 | 68.24 ± 7.76 | 70.90±7.64 | 68.52±7.32 |
| P=0.0001 | P<0.0001 | |||
| 50–59 | 5050 (12.83) | 309 (16.60) | 92 (7.65) | 5267 (13.16) |
| 60–64 | 8212 (20.86) | 361 (19.40) | 186 (15.46) | 8387 (20.95) |
| 65–69 | 8747 (22.22) | 395 (21.23) | 260 (21.61) | 8882 (22.19) |
| 70–74 | 8207 (20.85) | 371 (19.94) | 250 (20.7) | 8328 (20.81) |
| ≥ 75 | 9150 (23.24) | 425 (22.84) | 415 (34.50) | 9160 (22.89) |
| WHI enrollment (1993—1998): | ||||
| Region of residence: | P<0.0001 | P=0.008 | ||
| Midwest | 9318 (23.67) | 390 (20.96) | 274 (22.78) | 9434 (23.57) |
| Northeast | 9681 (24.59) | 577 (31.00) | 255 (21.20) | 10003 (24.99) |
| South | 8770 (22.28) | 398 (21.39) | 290 (24.11) | 8878 (22.18) |
| West | 11597 (29.46) | 496 (26.65) | 384 (31.92) | 11709 (29.25) |
| Race a: | P<0.0001 | P<0.0001 | ||
| White | 35469 (90.10) | 1384 (74.37) | 878 (72.98) | 35975 (89.88) |
| Black | 1965 (4.99) | 281 (15.10) | 222 (18.45) | 2024 (5.06) |
| Asian | 1091 (2.77) | 59 (3.17) | 35 (2.91) | 1115 (2.79) |
| Other | 481 (1.22) | 73 (3.92) | 49 (4.07) | 505 (1.26) |
| Unknown/Not reported | 360 (0.91) | 64 (3.44) | 19 (1.58) | 405 (1.01) |
| Ethnicity: | P<0.0001 | P<0.0001 | ||
| Hispanic | 1157 (2.94) | 171 (9.19) | 64 (5.32) | 1264 (3.16) |
| Non-Hispanic | 38096 (96.77) | 1672 (89.84) | 1124 (93.43) | 38644 (96.55) |
| Unknown/Not reported | 113 (0.29) | 18 (0.97) | 15 (1.25) | 116 (0.29) |
| Marital status: | P<0.0001 | P<0.0001 | ||
| Married/Partnered | 26293 (66.79) | 780 (41.91) | 442 (36.74) | 26631 (66.54) |
| Single | 1821 (4.63) | 128 (6.88) | 80 (6.65) | 1869 (4.67) |
| Divorced | 5586 (14.19) | 541 (29.07) | 329 (27.35) | 5798 (14.49) |
| Widowed | 5666 (14.39) | 412 (22.14) | 352 (29.26) | 5726 (14.31) |
| Socioeconomic position: | P<0.0001 | P<0.0001 | ||
| Low | 6511 (16.54) | 952 (51.16) | 644 (53.53) | 6819 (17.04) |
| Medium | 20379 (51.77) | 774 (41.59) | 471 (39.15) | 20682 (51.67) |
| High | 12476 (31.69) | 135 (7.25) | 88 (7.32) | 12523 (31.29) |
| Smoking status: | P<0.0001 | P<0.0001 | ||
| Never Smoker | 20282 (51.52) | 1034 (55.56) | 638 (53.03) | 20678 (51.66) |
| Past Smoker | 17062 (43.34) | 651 (34.98) | 449 (37.32) | 17264 (43.13) |
| Current Smoker | 2022 (5.14) | 176 (9.46) | 116 (9.64) | 2082 (5.20) |
| Alcohol consumption: | P<0.0001 | P<0.0001 | ||
| Non-Drinker | 3549 (9.02) | 297 (15.96) | 190 (15.79) | 3656 (9.13) |
| Former Drinker | 6036 (15.33) | 532 (28.59) | 340 (28.26) | 6228 (15.56) |
| < 1 drink/week | 12676 (32.20) | 628 (33.75) | 382 (31.75) | 12922 (32.29) |
| ≥ 1 drink/week | 17105 (43.45) | 404 (21.71) | 291 (24.19) | 17218 (43.02) |
| Healthy Eating Index – 2020: | P<0.0001 | P<0.0001 | ||
| Mean ± SD | 67.69±10.05 | 63.52±10.97 | 64.84±10.86 | 67.58±10.09 |
| Physical activity [Metabolic equivalent-hours/week]: | P<0.0001 | P<0.0001 | ||
| Mean ± SD | 14.94±14.76 | 10.90±14.52 | 11.43±13.63 | 14.86±14.79 |
| Body mass index (kg/m2): | ||||
| P<0.0001 | P<0.0001 | |||
| Mean ± SD | 26.56±5.38 | 29.20±6.69 | 29.24±6.49 | 26.60±5.42 |
| P<0.0001 | P<0.0001 | |||
| < 25 | 17864 (45.38) | 550 (29.55) | 341 (28.35) | 18073 (45.16) |
| 25–<30 | 13394 (34.02) | 590 (31.70) | 397 (33.00) | 13587 (33.95) |
| ≥ 30 | 8108 (20.60) | 721 (38.74) | 465 (38.65) | 8364 (20.90) |
| Number of comorbid conditions: | P<0.0001 | P<0.0001 | ||
| 0 | 19445 (49.40) | 742 (39.87) | 432 (35.91) | 19755 (49.36) |
| 1–2 | 18964 (48.17) | 1005 (54.00) | 707 (58.77) | 19262 (48.13) |
| 3+ | 957 (2.43) | 114 (6.13) | 64 (5.32) | 1007 (2.52) |
| Self-rated health: | P<0.0001 | P<0.0001 | ||
| Excellent/Very Good/Good | 37686 (95.73) | 1535 (82.48) | 1028 (85.45) | 38193 (95.43) |
| Fair/Poor | 1680 (4.27) | 326 (17.52) | 175 (14.55) | 1831 (4.57) |
| Depressive symptoms: | P<0.0001 | P<0.0001 | ||
| Mean ± SD | 1.32±1.94 | 2.63±2.83 | 2.09±2.60 | 1.35±1.99 |
| P<0.0001 | P<0.0001 | |||
| > 0.06 | 19012 (48.30) | 1314 (70.61) | 720 (59.85) | 19606 (48.99) |
| WHI enrollment (1993—1998) + WHI-OS 6-year follow-up visit b: | ||||
| Change in smoking status: | P<0.0001 | P=0.0013 | ||
| Non-smoker to Smoker | 823 (2.09) | 52 (2.79) | 38 (3.16) | 837 (2.09) |
| Smoker to non-Smoker | 179 (0.45) | 21 (1.13) | 12 (1.00) | 188 (0.47) |
| No change | 38364 (97.45) | 1788 (96.08) | 1153 (95.84) | 38999 (97.44) |
| Change in alcohol consumption c: | P=0.16 | P=0.0007 | ||
| Yes | 6040 (15.34) | 308 (16.55) | 227 (18.87) | 6121 (15.29) |
| No | 33326 (84.66) | 1553 (83.45) | 976 (81.13) | 33903 (84.71) |
| Change in physical activity: | P=0.18 | P=0.12 | ||
| Mean ± SD | −0.76±13.45 | −1.22±14.39 | −1.36±13.99 | −0.76±13.47 |
| Change in weight (kg): | P=0.72 | P=0.040 | ||
| Mean ± SD | −1.34±15.02 | −1.49±17.44 | −2.39±18.11 | −1.32±15.04 |
Other race group includes American Indian/Alaska Native, Native Hawaiian/Other Pacific Islanders, and more than one race
Besides smoking status, alcohol consumption, physical activity, and weight, background characteristics are assumed not to vary between enrollment (1993—1998) and 6-year WHI-OS follow-up visits
Self-reported at 6-year follow-up visit.
Abbreviations: SD = Standard deviation; WHI = Women’s Health Initiative; WHI-OS = Women’s Health Initiative Observational Study.
Statistical analysis:
All statistical analyses were conducted using SAS Version 9.4 (SAS Institute, Cary, NC). Bivariate relationships of background characteristics with treatment groups were evaluated using independent samples t-tests and χ2 tests. Furthermore, logistic regression models were constructed to estimate odds ratios (OR) with their 95% confidence intervals (CI) in analyses involving food security or food assistance as the outcome variable. After estimating incidence rates of the selected outcomes with their 95% CI using the exact Poisson method, we applied IPTW within Cox regression models to estimate HR with their 95% CI, when examining the association of food security, food assistance, and food assistance among women who reported food insecurity. Sensitivity analyses were performed by adding food security to IPTW Cox regression models involving food assistance as the exposure variable, and food assistance was added to IPTW Cox regression models involving food security as the exposure variable, examining an interaction variable that combines food security and food assistance in a risk-adjusted model, and excluding women with food insecurity and no food assistance when examining the relationship of food assistance with the selected outcomes. Methodological details are described in Appendix C. Two-tailed statistical tests were assessed at α=0.05.
RESULTS:
Of 93,676 WHI-OS participants, 41,227 (44.0%) women (mean [± standard deviation] age at analytic baseline: 68.6 [±7.3] years) met all pre-specified eligibility criteria (Figure 1). Furthermore, 39,366 (95.6%) women reported food security and 1,203 (2.9%) reported food assistance at analytic baseline. Among 1,861 women who did not achieve food security, 308 (16.6%) relied on food assistance. Similarly, among 39,366 who achieved food security, 895 (2.3%) relied on food assistance.
Figure 1.
Study Flowchart
Over a maximum follow-up time of ~24 years, 6,037 fatal and non-fatal CVD cases, 15,868 all-cause deaths, and 11,641 chronic disease-specific deaths were recorded. The median follow-up time for these outcomes was less than 15.0 years. Table 3 displays the estimated incidence rates (95% CI) [per 100,000 person-years] of selected outcomes by exposure groups in the unweighted analytic sample, suggesting that, prior to propensity scoring, these outcomes were consistently lower among women with vs. without food security achievement and higher among women who relied vs. did not rely on food assistance.
Table 3.
Rates (95% confidence interval) [per 100,000 person-years] of selected outcomes by food security and food assistance statuses in the unweighted analytic sample
| Food security / assistance | |||
|---|---|---|---|
| Total | Yes | No | |
| Food security: | N=41227 | N=39366 | N=1861 |
| CVD | 1027.71 (1001.95, 1053.97) | 1015.57 (989.48, 1042.18) | 1346.59 (1196.05, 1510.84) |
| All-cause death | 2617.29 (2576.73, 2658.34) | 2599.08 (2557.90, 2640.77) | 3092.41 (2866.26, 3331.65) |
| Chronic-disease specific death | 1920.08 (1885.36, 1955.29) | 1905.13 (1869.89, 1940.87) | 2310.37 (2115.48, 2518.38) |
| Food assistance: | N=41227 | N=1203 | N=40024 |
| CVD | 1027.71 (1001.95, 1053.97) | 1874.01 (1644.40, 2126.71) | 1008.85 (983.04, 1035.16) |
| All-cause death | 2617.29 (2576.73, 2658.34) | 3960.08 (3631.28, 4310.65) | 2586.75 (2628.02, 2545.96) |
| Chronic-disease specific death | 1920.08 (1885.36, 1955.29) | 2951.52 (2668.64, 3256.23) | 1896.62 (1861.72, 1932.01) |
Abbreviations: CVD = Cardiovascular disease.
Table S.5. of Appendix B presents IPTW Cox regression models for relationships of food security achievement and reliance on food assistance with the primary outcome of interest. In separate models, food security (HR=0.65, 95% CI: 0.47, 0.89) and food assistance (HR=0.47, 95% CI: 0.46, 0.48) were inversely related to CVD morbidity and mortality risks. Similar results were obtained in IPTW Cox regression models examining all-cause and chronic-disease specific mortality risks (Table 4), with food security and reliance on food assistance being inversely related to all-cause and chronic disease-specific mortality risks.
Table 4.
Cox regression models for the relationship of food security and food assistance with CVD outcomes in the weighted analytic sample
| Fatal or Non-Fatal CVD | |
|---|---|
| HR (95% CI) | |
| Food security (Yes vs. No) a | 0.65 (0.47, 0.89) |
| Food assistance (Yes vs. No) b | 0.47 (0.46, 0.48) |
Abbreviations: CI = Confidence interval; CVD = Cardiovascular disease; HEI = Healthy Eating Index; HR = Hazard ratio. Note: As shown in Table S.3., inverse probability of treatment weighting included covariates selected based on backward elimination, and these covariates differed between models involving food security and food assistance. In food security models, the selected covariates were age (continuous), ethnicity, marital status, socioeconomic position, alcohol consumption, BMI (categorical), number of comorbidities, self-rated health, depression score (categorical), HEI-2020 score, and alcohol consumption change. In food assistance models, the selected covariates were region, age (continuous), race, marital status, socioeconomic position, smoking status, alcohol consumption, BMI (categorical), self-rated health, and alcohol consumption change.
Out of 41,096 participants, there was a total of 6,025 cases of fatal or non-fatal CVD.
Out of 41,227 participants, there was a total of 6,037 cases of fatal or non-fatal CVD.
Similar results were obtained when propensity scoring for IPTW Cox regression models took into consideration food security when examining food assistance, and food assistance when examining food security in relation to the primary and secondary outcomes of interest (Appendix B; Table S.6.). As shown in Appendix B (Tables S.7–S.8), results of IPTW Cox regression models were mostly comparable to those of unadjusted and risk-adjusted models when the exposure was ‘food security achievement’, whereas ‘reliance on food assistance’ was negatively associated with selected outcomes in IPTW models but positively associated with these outcomes in unadjusted and risk-adjusted models. Furthermore, risk-adjusted Cox regression models examining the interaction variable comprising food security and assistance and its relationship with the selected outcomes found that women who screened positive for food security and food assistance were high-risk for CVD, all-cause mortality, and chronic disease-specific mortality, whereas those who screened negative for food security (regardless of food assistance) were high-risk for all-cause and chronic disease-specific mortality compared to women who were food secure and had no food assistance (Tables S.7–S.8).
Among women who did not achieve food security, IPTW Cox regression models suggested that reliance on food assistance was positively related to CVD morbidity and mortality risks (HR=1.15, 95% CI: 1.01, 1.32) and chronic disease-specific mortality risks (HR=1.12, 95% CI: 1.01, 1.24), although it was negatively related to all-cause mortality risk (HR=0.86, 95% CI: 0.79, 0.94) (Appendix B; Table S.9.). After excluding women with food insecurity and no food assistance in a sensitivity analysis, IPTW Cox regression models suggested that reliance on food assistance was negatively associated with all three outcomes of interest (Appendix B; Table S.10.).
DISCUSSION:
In this target trial emulation involving 41,227 postmenopausal women, a significant association was found between achieving food security and lower CVD morbidity and mortality, and reliance on food assistance was generally related to lower CVD risks. Among women who reported food insecurity, reliance on food assistance was linked to modestly higher CVD and chronic-disease specific mortality risks, but lower all-cause mortality risks, and these nuanced and mixed findings necessitate further investigation.
Overall, the study findings for postmenopausal women are consistent with previous observational studies that focused on adverse health consequences, including CVD morbidity and mortality risks, and their relationships with food security and assistance. A 2011—2018 National Health Interview Survey (NHIS) study involving 218,136 participants, aged 20—74 years at baseline, compared three food insecurity instruments (10-item, 6-item, and 2-item) for their association with premature all-cause and CVD deaths by December 2019. HRs for all-cause mortality showed minimal variation among food insecurity instruments (10-item, 1.22; 6-item, 1.23; 2-item, 1.23) with identical c-statistics (0.823)3. HRs for CVD deaths were similar (10-item, 1.38; 6-item, 1.27; 2-item, 1.41), with c-statistics ranging from 0.852 to 0.853. In a replication using data on 37,027 participants from the 2004—2018 NHANES, associations weakened and lost statistical significance when adjusted for food assistance program participation, diabetes, diet quality, sleep patterns, and depression3. In another study, researchers retrospectively analyzed household food security status in relation to modified ‘ideal’ cardiovascular health (mICVH) among 157,001 adult participants from the 2014—2018/2020 NHIS data, whereby food security was categorized as very low, low, marginal, or high and assessment of mICVH utilized a score based on four health behaviors and three clinical factors, suggesting that the inverse association between very low food security and mICVH was stronger in women than in men1.
Aging women receiving food assistance may have long-standing risk factors for chronic disease development, including the propensity for poor dietary habits and greater chronic strain manifesting in higher scores of AL2, leading to increased CVD. Furthermore, addressing food insecurity through food assistance is a complex issue, given that individuals who are likely to need and receive food assistance may face additional barriers preventing them from achieving food security, as described elsewhere34. Longitudinal studies with larger samples and repeated measurements (e.g. multicohort studies involving WHI and other datasets) are needed to disentangle the causal effects of food security and food assistance, by examining the putative role of food assistance among participants with various degrees of food insecurity, using the target trial emulation framework.
The finding that suggests achieving food security and reliance on food assistance may reduce CVD morbidity and mortality risks among postmenopausal women is also consistent with the idea that food insecurity among these women may correlate with reduced diet quality, which in turn has been linked to adverse outcomes in volunteer participants from the WHI cohort. For instance, a study involving 63,805 postmenopausal women from the WHI-OS (1993—2010) examined associations between four diet quality indices (Healthy Eating Index-2010 (HEI), AHEI, Alternate Mediterranean Diet (aMED), and Dietary Approaches to Stop Hypertension (DASH)]) and all-cause, CVD-specific, and cancer-specific mortality, with results indicating that improved diet quality based on HEI, AHEI, aMED, and DASH scores, were associated with 18%−26% reduction in all-cause and CVD-specific mortality35. Also, higher HEI, aMED, and DASH scores (but not AHEI) were linked to 20–23% decrease in cancer-specific mortality35. Another study involving 59,388 WHI-OS (1993—2017) participants analyzed relationships between HEI-2020 scores and mortality from any cause, CVD, cancer, Alzheimer disease, and dementia not otherwise specified (NOS), suggesting that those with higher HEI-2020 scores exhibited 18% lower risk of all-cause mortality and 21% lower risk of cancer mortality compared to those with lower scores, but that HEI-2020 scores were not significantly related to mortality risks from CVD, Alzheimer disease, or dementia NOS36.
Emerging evidence from various high-income countries supports the link between food insecurity and CVD, similarly to findings initially observed in the United States. Systematic reviews across North America, Europe, and Australia confirm that food insecurity correlates with poorer dietary quality, obesity, hypertension, diabetes, and increased cardiovascular morbidity and mortality risks37–39. This relationship underscores food insecurity as a social determinant of CVD risk, with its negative impacts on cardiometabolic health being consistent across differing socioeconomic and healthcare landscapes, despite variability in national policies and food assistance programs37–39.
Our study findings are limited by several factors. First, potential selection bias arises from using a subset of WHI participants, which may not represent the broader population of postmenopausal women in the U.S. In particular, the prevalence of food insecurity in this study was estimated at 4–5%, which is substantially less than the USDA surveillance data estimate that approximately 10–12% of U.S. households experienced food insecurity during the late 1990s, possibly due to healthy cohort bias or exclusion of WHI participants who did not meet eligibility criteria and these might be more disadvantaged than others40. Second, reliance on self-reported assessment at one point in time introduces the risk of information bias, particularly since food security and assistance are likely chronic exposures that may vary over time. Third, despite using IPTW and the target trial emulation framework, residual confounding from unmeasured or inadequately measured variables is a concern due to the observational nature of this study. Also, the use of propensity scoring may have masked important socioeconomic differences among treatment groups that are also correlated with the exposure variables of interest. Fourth, women with food security either with or without reliance on food assistance can have unhealthy diets, complicating the interpretation of results linking socioeconomic position to CVD through food security, food assistance, and diet quality. Furthermore, we did not evaluate dietary patterns as potential mediators of hypothesized relationships in this study. Whereas food assistance covers a wide range of services, recipients of SNAP may also receive healthcare services through their Medicaid coverage, potentially increasing healthcare access and improving CVD outcomes. Fifth, random variation might influence outcomes given the small sample sizes pertaining to the exposure variables in some analyses. Finally, this study focused on postmenopausal women, limiting generalizability of results to other populations in the U.S. Therefore, caution is needed when interpreting these study findings.
In conclusion, food security and assistance are generally associated with lower CVD risks among postmenopausal women. Future studies should address gaps in knowledge regarding food safety and adequacy of assistance among food insecure individuals. Study findings have implications for enhancing food assistance programs, implementing targeted nutritional counseling, development of community-based interventions, and provision of holistic care among aging women in the U.S.
Supplementary Material
KEY MESSAGES:
What is already known on this topic:
Prior research suggests a negative link between food insecurity and cardiometabolic health, but few studies have specifically examined food security and assistance among postmenopausal women.
What this study adds:
This retrospective cohort study of 41,227 postmenopausal women from the Women’s Health Initiative utilized the target trial emulation framework and found that food security and assistance were inversely related to cardiovascular disease (CVD) morbidity and mortality risks.
How this study might affect research, practice or policy:
Study findings have implications for food assistance programs, targeted nutritional counseling, community interventions, and holistic care for aging women in the United States.
Acknowledgements:
This study was supported in part by the Intramural Research Program of the NIH, National Institute on Aging, NIA/NIH/IRP. The authors thank the WHI investigators and staff for their dedication, and the study participants for making the program possible. A listing of WHI investigators can be found at https://www-whi-org.s3.us-west-2.amazonaws.com/wp-content/uploads/WHI-Investigator-Short-List.pdf.
Funding Statement:
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 75N92021D00001, 75N92021D00002, 75N92021D00003, 75N92021D00004, 75N92021D00005. Dr. Nudy is supported by the National Center for Advancing Translational Sciences, Grant KL2 TR002015 and Grant UL1 TR002014. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. The sponsors did not play a role in the study design, collection, analysis, and interpretation of data, in the writing of the report or in the decision to submit the article for publication.
Footnotes
Competing Interests Statement: No competing interest.
Ethical Approval Statement: Not applicable. The WHI program involves human participants and received approval with informed consent for all participants through Institutional Review Boards of clinical centers, the Coordinating Center, and the National Institute of Health, in compliance with the Helsinki Declaration as revised in 1983. However, the current study did not involve human subjects since it analyzed de-identified data from the WHI study under a data use agreement. Accordingly, ethical approval for the current study was not required.
Disclaimer: The views expressed in this paper are those of the authors and do not reflect the official policy or position of the U.S. Department of Veterans Affairs, the National Institutes of Health, or the U.S. Government.
Data availability statement:
WHI data is available through application to the WHI data coordinating center (WHI Data Sharing Statement). Statistical code for this project can be made available from the corresponding author upon request.
References:
- 1.Murkey JA, Gaston SA, Alhasan DM, et al. Food security status and cardiometabolic health by sex/gender and race/ethnicity among adults in the United States. BMC Public Health 2024;24(1):1220. doi: 10.1186/s12889-024-18655-y [published Online First: 20240503] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Saelee R, Alexander DS, Onufrak S, et al. Household Food Security Status and Allostatic Load among United States Adults: National Health and Nutrition Examination Survey 2015–2020. J Nutr 2024;154(2):785–93. doi: 10.1016/j.tjnut.2023.12.041 [published Online First: 20231228] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Tian L, Jaeger BC, Marshall AN, et al. Comparison of Associations of Food Security Instruments and Mediators With Premature All-Cause and Cardiovascular Disease Death in US Adults. Circ Cardiovasc Qual Outcomes 2025;18(2):e011209. doi: 10.1161/CIRCOUTCOMES.124.011209 [published Online First: 20250109] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Goldberg SL, Mawn BE. Predictors of Food Insecurity among Older Adults in the United States. Public Health Nurs 2015;32(5):397–407. doi: 10.1111/phn.12173 [published Online First: 20141223] [DOI] [PubMed] [Google Scholar]
- 5.Morales ME, Epstein MH, Marable DE, et al. Food Insecurity and Cardiovascular Health in Pregnant Women: Results From the Food for Families Program, Chelsea, Massachusetts, 2013–2015. Prev Chronic Dis 2016;13:E152. doi: 10.5888/pcd13.160212 [published Online First: 20161103] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Mehta JM, Manson JE. The menopausal transition period and cardiovascular risk. Nat Rev Cardiol 2024;21(3):203–11. doi: 10.1038/s41569-023-00926-7 [published Online First: 20230926] [DOI] [PubMed] [Google Scholar]
- 7.Yousefzai S, Amin Z, Faizan H, et al. Cardiovascular Health During Menopause Transition: The Role of Traditional and Nontraditional Risk Factors. Methodist Debakey Cardiovasc J 2025;21(4):121–28. doi: 10.14797/mdcvj.1619 [published Online First: 20250812] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Liu Y, Eicher-Miller HA. Food Insecurity and Cardiovascular Disease Risk. Curr Atheroscler Rep 2021;23(6):24. doi: 10.1007/s11883-021-00923-6 [published Online First: 20210327] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Royer MF, Rosas LG, King AC. Food insecurity and cardiovascular disease risk factors among U.S. adults. BMC Public Health 2025;25(1):817. doi: 10.1186/s12889-025-22031-9 [published Online First: 20250228] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Huang M, Lo K, Li J, et al. Pasta meal intake in relation to risks of type 2 diabetes and atherosclerotic cardiovascular disease in postmenopausal women : findings from the Women’s Health Initiative. BMJ Nutr Prev Health 2021;4(1):195–205. doi: 10.1136/bmjnph-2020-000198 [published Online First: 20210430] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Neuhouser ML, Pettinger M, Tinker LF, et al. Associations of Biomarker-Calibrated Healthy Eating Index-2010 Scores with Chronic Disease Risk and Their Dependency on Energy Intake and Body Mass Index in Postmenopausal Women. J Nutr 2023;152(12):2808–17. doi: 10.1093/jn/nxac199 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Allison MA, Aragaki AK, Ray RM, et al. A Randomized Trial of a Low-Fat Diet Intervention on Blood Pressure and Hypertension: Tertiary Analysis of the WHI Dietary Modification Trial. Am J Hypertens 2016;29(8):959–68. doi: 10.1093/ajh/hpv196 [published Online First: 20151226] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Belin RJ, Greenland P, Allison M, et al. Diet quality and the risk of cardiovascular disease: the Women’s Health Initiative (WHI). Am J Clin Nutr 2011;94(1):49–57. doi: 10.3945/ajcn.110.011221 [published Online First: 20110525] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Reeves KW, Santana MD, Manson JE, et al. Predictors of urinary phthalate biomarker concentrations in postmenopausal women. Environ Res 2019;169:122–30. doi: 10.1016/j.envres.2018.10.024 [published Online First: 20181030] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Rissling MB, Gray KE, Ulmer CS, et al. Sleep Disturbance, Diabetes, and Cardiovascular Disease in Postmenopausal Veteran Women. Gerontologist 2016;56 Suppl 1(Suppl 1):S54–66. doi: 10.1093/geront/gnv668 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Cauley JA, Hovey KM, Stone KL, et al. Characteristics of Self-Reported Sleep and the Risk of Falls and Fractures: The Women’s Health Initiative (WHI). Journal of bone and mineral research : the official journal of the American Society for Bone and Mineral Research 2019;34(3):464–74. doi: 10.1002/jbmr.3619 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Grieshober L, Wactawski-Wende J, Hageman Blair R, et al. A Cross-Sectional Analysis of Telomere Length and Sleep in the Women’s Health Initiative. American journal of epidemiology 2019;188(9):1616–26. doi: 10.1093/aje/kwz134 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Hernan MA, Wang W, Leaf DE. Target Trial Emulation: A Framework for Causal Inference From Observational Data. JAMA 2022;328(24):2446–47. doi: 10.1001/jama.2022.21383 [DOI] [PubMed] [Google Scholar]
- 19.Curb JD, McTiernan A, Heckbert SR, et al. Outcomes ascertainment and adjudication methods in the Women’s Health Initiative. Ann Epidemiol 2003;13(9 Suppl):S122–8. doi: 10.1016/s1047-2797(03)00048-6 [DOI] [PubMed] [Google Scholar]
- 20.Liu Z, Chen BH, Assimes TL, et al. The role of epigenetic aging in education and racial/ethnic mortality disparities among older U.S. Women. Psychoneuroendocrinology 2019;104:18–24. doi: 10.1016/j.psyneuen.2019.01.028 [published Online First: 20190206] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Birru Talabi M, Mackey RH, Kuller LH, et al. Human Leukocyte Antigen Shared Epitope and Inflammation, Cardiovascular Disease, Cancer, and Mortality Among Postmenopausal Women in the Women’s Health Initiative Rheumatoid Arthritis Study. Am J Epidemiol 2017;186(2):245–54. doi: 10.1093/aje/kwx087 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Kuller LH, Mackey RH, Walitt BT, et al. Determinants of mortality among postmenopausal women in the women’s health initiative who report rheumatoid arthritis. Arthritis Rheumatol 2014;66(3):497–507. doi: 10.1002/art.38268 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Beydoun HA, Beydoun MA, Tsai J, et al. Triglyceride-glucose index and cardiovascular disease by cardiovascular-kidney-metabolic syndrome and socioeconomic status among postmenopausal women. Atherosclerosis 2026;414:120645. doi: 10.1016/j.atherosclerosis.2026.120645 [published Online First: 20260117] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Panizza CE, Shvetsov YB, Harmon BE, et al. Testing the Predictive Validity of the Healthy Eating Index-2015 in the Multiethnic Cohort: Is the Score Associated with a Reduced Risk of All-Cause and Cause-Specific Mortality? Nutrients 2018;10(4) doi: 10.3390/nu10040452 [published Online First: 20180405] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Reedy J, Lerman JL, Krebs-Smith SM, et al. Evaluation of the Healthy Eating Index-2015. J Acad Nutr Diet 2018;118(9):1622–33. doi: 10.1016/j.jand.2018.05.019 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Murakami K, Livingstone MBE, Fujiwara A, et al. Reproducibility and Relative Validity of the Healthy Eating Index-2015 and Nutrient-Rich Food Index 9.3 Estimated by Comprehensive and Brief Diet History Questionnaires in Japanese Adults. Nutrients 2019;11(10) doi: 10.3390/nu11102540 [published Online First: 20191021] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Cavanaugh AM, Rauh MJ, Thompson CA, et al. Racial and ethnic disparities in utilization of total knee arthroplasty among older women. Osteoarthritis Cartilage 2019;27(12):1746–54. doi: 10.1016/j.joca.2019.07.015 [published Online First: 20190809] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Kling JM, Manson JE, Naughton MJ, et al. Association of sleep disturbance and sexual function in postmenopausal women. Menopause 2017;24(6):604–12. doi: 10.1097/GME.0000000000000824 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Danhauer SC, Brenes GA, Levine BJ, et al. Variability in sleep disturbance, physical activity and quality of life by level of depressive symptoms in women with Type 2 diabetes. Diabetic medicine : a journal of the British Diabetic Association 2019;36(9):1149–57. doi: 10.1111/dme.13878 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Jung SY, Ho G, Rohan T, et al. Interaction of insulin-like growth factor-I and insulin resistance-related genetic variants with lifestyle factors on postmenopausal breast cancer risk. Breast cancer research and treatment 2017;164(2):475–95. doi: 10.1007/s10549-017-4272-y [DOI] [PubMed] [Google Scholar]
- 31.Beydoun HA, Beydoun MA, Wassertheil-Smoller S, et al. Depressive symptoms and antidepressant use in relation to white blood cell count among postmenopausal women from the Women’s Health Initiative. Transl Psychiatry 2024;14(1):157. doi: 10.1038/s41398-024-02872-5 [published Online First: 20240321] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.McMurry TL, Hu Y, Blackstone EH, et al. Propensity scores: Methods, considerations, and applications in the Journal of Thoracic and Cardiovascular Surgery. The Journal of thoracic and cardiovascular surgery 2015;150(1):14–9. doi: 10.1016/j.jtcvs.2015.03.057 [DOI] [PubMed] [Google Scholar]
- 33.Williamson EJ, Forbes A. Introduction to propensity scores. Respirology 2014;19(5):625–35. doi: 10.1111/resp.12312 [DOI] [PubMed] [Google Scholar]
- 34.Hung A, Durant RW, Corbie G. Addressing Food Insecurity Is Complicated but Needed. JAMA Intern Med 2026. doi: 10.1001/jamainternmed.2025.7846 [published Online First: 20260216] [DOI] [PubMed] [Google Scholar]
- 35.George SM, Ballard-Barbash R, Manson JE, et al. Comparing indices of diet quality with chronic disease mortality risk in postmenopausal women in the Women’s Health Initiative Observational Study: evidence to inform national dietary guidance. Am J Epidemiol 2014;180(6):616–25. doi: 10.1093/aje/kwu173 [published Online First: 20140717] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.George SM, Reedy J, Cespedes Feliciano EM, et al. Alignment of Dietary Patterns With the Dietary Guidelines for Americans 2015–2020 and Risk of All-Cause and Cause-Specific Mortality in the Women’s Health Initiative Observational Study. Am J Epidemiol 2021;190(5):886–92. doi: 10.1093/aje/kwaa268 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Chang R, Javed Z, Taha M, et al. Food insecurity and cardiovascular disease: Current trends and future directions. Am J Prev Cardiol 2022;9:100303. doi: 10.1016/j.ajpc.2021.100303 [published Online First: 20211210] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Teshale AB, Htun HL, Owen A, et al. The Role of Social Determinants of Health in Cardiovascular Diseases: An Umbrella Review. J Am Heart Assoc 2023;12(13):e029765. doi: 10.1161/JAHA.123.029765 [published Online First: 20230622] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Salinas-Roca B, Rubio-Pique L, Carrillo-Alvarez E, et al. Impact of Health and Social Factors on the Cardiometabolic Risk in People with Food Insecurity: A Systematic Review. Int J Environ Res Public Health 2022;19(21) doi: 10.3390/ijerph192114447 [published Online First: 20221104] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Household food security in the United States, 195–1998: United States Department of Agriculture, Economic Research Service, 2000. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
WHI data is available through application to the WHI data coordinating center (WHI Data Sharing Statement). Statistical code for this project can be made available from the corresponding author upon request.