Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Nov 4.
Published in final edited form as: Am J Prev Med. 2021 Oct 10;62(1):65–76. doi: 10.1016/j.amepre.2021.06.020

Longitudinal Analysis of Food Insufficiency and Cardiovascular Disease Risk Factors in the Coronary Artery Risk Development in Young Adults Study

Kelsey A Vercammen 1, Alyssa J Moran 2, Mercedes R Carnethon 3, Amanda C McClain 4, Lindsay R Pool 3, Catarina I Kiefe 5, April P Carson 6, Penny Gordon-Larsen 7, Lyn M Steffen 8, Matthew M Lee 9, Jessica G Young 1,10, Eric B Rimm 1,9,11
PMCID: PMC9635889  NIHMSID: NIHMS1749286  PMID: 34642058

Abstract

Introduction:

Most previous studies on food insecurity and cardiovascular disease risk factors are cross-sectional. Without longitudinal data, it is unclear whether food insecurity precedes poor health and how exposure timing impacts these relationships.

Methods:

Data from 2000 to 2001, 2005 to 2006, and 2010 to 2011 of the Coronary Artery Risk Development in Young Adults study were used. Food insufficiency—a screener measure related to food insecurity—was assessed in 2000–2001 and 2005–2006 using a single item. Cardiovascular disease risk factors were objectively assessed in 2010–2011. Impacts of food insufficiency patterns (food sufficient, food insufficient in 2000–2001 only, food insufficient in 2005–2006 only, food insufficient in both 2000–2001 and 2005–2006) on cardiovascular disease risk factors were estimated using inverse probability weighting of marginal structural models. Covariates that change over time were adjusted for using stabilized weights; baseline covariates were adjusted for in the marginal structural models. Analyses were conducted in 2020–2021.

Results:

The baseline sample included 2,596 participants (56% women, 47% White). In unadjusted analyses, all food insufficiency patterns were associated with higher BMI, waist circumference, and blood pressure than food sufficiency. After accounting for covariates, estimates were attenuated but still consistent with adverse effects of food insufficiency, particularly among women.

Conclusions:

After covariate adjustment, food insufficiency was associated with several cardiovascular disease risk factors. Findings from this study should be replicated in other settings and populations. If verified, this evidence could provide justification for intervening in food insecurity to reduce future cardiovascular disease risk.

INTRODUCTION

Food insecurity, defined by the U.S. Department of Agriculture as a “lack of consistent access to enough food for an active, healthy life,”1 is a leading public health issue around the world.2 In the U.S., the prevalence of food insecurity has been systematically monitored since 1995,3 and the federal government spends in excess of $95 billion each year on nutrition assistance programs aimed at improving food access for food-insecure households.4,5 Despite these efforts, 10.5% of U.S. households were food insecure at some point during 2019,6 with this number rising to an estimated 22% during the first few months of the coronavirus disease 2019 (COVID-19) pandemic in 2020.7,8

A growing body of evidence indicates that food insecurity may lead to poor health.2,9-13 Obesity has been examined frequently, with multiple studies reporting a harmful association among women but mixed findings for men and children.11,14 Emerging evidence suggests that food insecurity is associated with other cardiovascular disease (CVD) risk factors,15-19 including high cholesterol and blood pressure (BP).19 Households with insufficient financial resources to purchase food may compensate by increasing reliance on cheap, energy-dense, and nutrient-poor foods, which can lead to metabolic dysregulation and fat accumulation.12 Food insecurity may also affect cardiometabolic risk through nondietary pathways such as by activating a physiologic stress response, triggering harmful coping behaviors, and reducing the ability to manage chronic conditions.20 Given that obesity and CVD are leading causes of morbidity and mortality worldwide,21,22 a robust scientific evaluation of food insecurity and its impact on these outcomes is needed for informing interventions and policy development in this area.

Substantial gaps in knowledge regarding food insecurity and health still exist. Importantly, available research among adults is largely cross-sectional. Longitudinal data are needed to understand the temporal ordering of food insecurity and poor health as well as to distinguish whether adverse health impacts are the result of cumulative damage from years of experiencing food insecurity (persistent food insecurity) versus shorter-term adaptations to acute experiences of food insecurity (transient food insecurity). A few studies have attempted to estimate the cumulative impacts of food insecurity over time,23-25 but previous research is limited by methodologic challenges related to handling time-varying confounding and selection bias.26

In this study, longitudinal data from the Coronary Artery Risk Development in Young Adults (CARDIA) study are used to (1) examine the longitudinal relationships of food insecurity (as assessed by food insufficiency, a related screener measure27) with CVD risk factors and (2) determine whether experiencing persistent versus transient food insecurity has differing relationships with CVD risk factors.

METHODS

Study Population

The CARDIA prospective cohort study includes 5,115 Black and White adults aged 18–30 years at recruitment in 1985–1986.28 CARDIA’s goal is to examine the determinants of clinical and subclinical CVD and their risk factors through interviewer-administered questionnaires, anthropometric assessments, imaging, and biosample collections. Study recruitment was intended to be balanced on age, gender, race, and educational attainment across 4 urban field centers: Birmingham, AL; Chicago, IL; Minneapolis, MN; and Oakland, CA. Participants provided written informed consent at every examination, and IRB approval was obtained by each field center.

A total of 3 CARDIA examinations that assessed food insufficiency were used: Year 15 (2000–2001), Year 20 (2005–2006), and Year 25 (2010–2011). The analysis included participants with complete 2000–2001 data on exposure and covariates and who in 2000–2001 had no previous history of myocardial infarction or stroke and were not currently pregnant. The sample was further restricted to participants with an annual household income < $100,000 in 2000–2001 because there were few food insufficient participants with higher incomes. This baseline analytic sample (N=2,596; participant flowcharts are shown in Appendix 1, available online) was further restricted to those who reported fasting ≥8 hours in 2000–2001 when examining fasting glucose as an outcome (n=2,441) and to those who reported fasting ≥12 hours in 2000–2001 when examining low-density lipoprotein cholesterol and triglycerides (n=2,311).

Over follow-up, participants were censored in 2005–2006 or 2010–2011 if they were missing exposure, covariate, or outcome data (or reported an inadequate fast duration for fasting outcomes). Approximately three quarters of the overall sample (n=1,897) remained uncensored by the end of follow-up in 2010–2011.

Measures

The CARDIA study assessed food insufficiency, a validated single-item measure often used as a screener for food insecurity surveys.27 Food insufficiency is more limited in scope than food insecurity and tends to overestimate assessments of food insecurity.29-31 At each timepoint (2000–2001 and 2005–2006), food insufficiency was assessed by asking participants to choose the statement that best describes the food eaten in their household during the last year: (1) We have enough food to eat and the kinds of food we want. (2) We have enough food to eat, but NOT always the kinds of food we want to eat. (3) Sometimes we don’t have enough food to eat. (4) Often, we don’t have enough food to eat. In line with previous research,31,32 responses were dichotomized, with food sufficiency defined as having an adequate quantity and quality of food (Response Option 1) and food insufficiency defined as inadequate quantity or quality of food (Response Options 2–4). Food insufficiency assessments from 2000 to 2001 and from 2005 to 2006 were used to create 4 time-varying food insufficiency patterns: (1) food sufficiency (food sufficient in 2000–2001 and 2005–2006), (2) transient food insufficiency only in 2000–2001, (3) transient food insufficiency only in 2005–2006, and (4) persistent food insufficiency (food insufficient in 2000–2001 and 2005–2006).

All outcomes were measured by trained study staff in 2010–2011 using standardized techniques.33 BMI was calculated as weight in kilograms divided by height in meters squared. Waist circumference was measured in duplicate with a tape to the nearest 0.5 cm around the minimal abdominal girth. BP was measured 3 times after participants rested in a quiet room for 5 minutes and was calculated as the average of the last 2 measurements. Blood samples were collected from participants to assess total cholesterol, high-density lipoprotein (HDL) cholesterol, low-density lipoprotein cholesterol, triglycerides, and fasting glucose. Participants were instructed to fast overnight and avoid smoking and strenuous physical activity for ≥2 hours before blood collection.

Study covariates are described in detail in the footnote of Table 1.34 Baseline covariates assessed at CARDIA’s initial examination were gender, race, age, and recruitment center. Time-varying covariates assessed in 2000–2001 and 2005–2006 included household income, employment status, marital status, household size, cholesterol or BP medication use, self-reported diabetes, smoking status, and physical activity score. Previous BMI in 2000–2001 and 2005–2006 was also adjusted for when examining BMI as the outcome in 2010–2011 (with previous values of other outcomes adjusted for in an analogous manner).

Table 1.

Characteristics of Analytic Sample by Food Insufficiency Status at Baseline: CARDIA, 2000–2001

Characteristicsa Total
(N=2,596)		 Food insufficient
(n=464)		 Food sufficient
(n=2,132)		 p-value
Gender 0.002
 Women 1,466 (56) 293 (63) 1,173 (55)
 Men 1,130 (44) 171 (37) 959 (45)
Age, years 0.003
 Mean (SD) 40.0 (±3.7) 39.5 (±3.9) 40.1 (±3.7)
Race <0.001
 White 1,219 (47) 146 (31) 1,073 (50)
 Black 1,377 (53) 318 (69) 1,059 (50)
Employment status <0.001
 Full time 1,959 (75) 303 (65) 1,656 (78)
 Part-time 547 (21) 127 (27) 420 (20)
 Unemployed 90 (3) 34 (7) 56 (3)
Smoking status <0.001
 Current 634 (24) 168 (36) 466 (22)
 Former 459 (18) 72 (16) 387 (18)
 Never 1,503 (58) 224 (48) 1,279 (60)
Household income, $ <0.001
 <5,000 72 (3) 38 (8) 34 (2)
 5,000–11,999 123 (5) 56 (12) 67 (3)
 12,000–15,999 99 (4) 43 (9) 56 (3)
 16,000–24,999 226 (9) 62 (13) 164 (8)
 25,000–34,999 309 (12) 78 (17) 231 (11)
 35,000–49,000 532 (20) 93 (20) 439 (21)
 50,000–74,999 743 (29) 72 (16) 671 (31)
 75,000–99,999 492 (19) 22 (5) 470 (22)
Marital status <0.001
 No partner 1,218 (47) 270 (58) 948 (44)
 Partner 1,378 (53) 194 (42) 1,184 (56)
Household size <0.001
 1 person 444 (17) 77 (17) 367 (17)
 2–4 people 1,745 (67) 286 (62) 1,459 (68)
 ≥5 people 407 (16) 101 (22) 306 (14)
Diabetes 0.720
 Yes 123 (5) 20 (4) 103 (5)
 No 2,473 (95) 444 (96) 2,029 (95)
Cholesterol or BP medication 0.103
 Yes 247 (10) 54 (12) 193 (9)
 No 2,349 (90) 410 (88) 1,939 (91)
Physical activity tertile <0.001
 Low 935 (36) 206 (44) 729 (34)
 Moderate 847 (33) 151 (33) 696 (33)
 High 814 (31) 107 (23) 707 (33)
Recruitment center 0.032
 Birmingham 680 (26) 130 (28) 550 (26)
 Chicago 497 (19) 100 (22) 397 (19)
 Minneapolis 793 (31) 146 (31) 647 (30)
 Oakland 626 (24) 88 (19) 538 (25)
Open in a new tab

Note: Boldface indicates statistical significance (p<0.05). Table values are n (%) for categorical variables and mean (SD) for continuous variables. p-value for categorical variables is based on a chi-square test, whereas p-value for continuous variables is based on a t-test.

a

Employment status: participants were categorized as working full time if they answered affirmatively to the question Are you working full time? Otherwise, they were classified as part-time or keeping house if they responded affirmatively to questions about working part-time or keeping house full time or as unemployed if they responded affirmatively to questions about being unemployed, laid off, or currently looking for work. Smoking status: participants were classified as never smoker if they reported never smoking any tobacco products or never smoking cigarettes regularly for at least 3 months (regularly defined as ≥5 cigarettes per week almost every week), current smokers if they reported smoking regularly now, or former smokers if they reported smoking regularly at some point in their life but not now. Income: participants were asked to report their total combined family income from the past 12 months in 9 categories (note: ≥$100,000 was excluded for this analysis). Marital status: participants were categorized as having a partner if they reported being married or living with someone in a marriage-like relationship or as having no partner if they reported being widowed, divorced, separated, never married, or other. Household size: participants reported the total number of people currently living in their household, including themselves. Diabetes: participants were asked Has a doctor or nurse ever told you that you have diabetes (high sugar in blood or urine)? Thus, Type 1 and Type 2 diabetes were not distinguished. Cholesterol/BP medication: participants were categorized as Yes if they responded affirmatively to either Are you taking medications for high blood pressure? or Are you taking medications to lower your blood cholesterol? Physical activity: participants completed the Physical Activity History, a brief questionnaire developed by CARDIA that asks about the frequency, intensity, and duration of 13 categories of sports/exercise over the past 12 months.34 Responses were used to determine a total physical activity score (with scores ranging from 0 to 1,932 exercise units in 2000–2001). Participants were then divided into tertiles on the basis of their total physical activity scores.

BP, blood pressure; CARDIA, Coronary Artery Risk Development in Young Adults.

Statistical Analysis

The distribution of outcomes across individuals was compared across food insufficiency status before considering covariates. Coefficients were estimated using linear regression models, where the dependent variable was a continuous version of each outcome in 2010–2011, and covariates were indicators for food insufficiency in 2000–2001, food insufficiency in 2005–2006, and an interaction term between these 2 food insufficiency indicators. The parameters of this model were used to compare outcomes across the 4 food insufficiency patterns.

Next, inverse probability (IP) weighting of marginal structural models was used to compare outcome distributions, adjusted for baseline and time-varying covariates.35,36 IP weighting was chosen as the analytic approach because it allows for estimation of the effects of exposures that vary over time and may affect and be affected by covariates that also vary over time.37,38 IP weights were used to adjust for time-varying confounders, and baseline confounders were included directly in the marginal structural models. In a similar manner, weights were used to account for selection bias owing to censoring (i.e., bias induced by loss to follow-up or missing follow-up data). As in all observational analyses that attempt to make causal inferences, the validity of these findings is based on many untestable assumptions (a discussion of the approach and assumptions is provided in Appendix 2, available online).

To estimate the coefficients of the marginal structural models, weighted generalized linear regression models were fit among uncensored participants but with all participants in the overall baseline sample contributing to the estimation of the weights using data before their censoring time. In these models, the dependent variable was the outcome in 2010–2011, and covariates were an indicator for food insufficiency in 2000–2001, an indicator for food insufficiency in 2005–2006, an interaction term between the 2 food insufficiency indicators, and baseline confounders (with time-varying confounders accounted for using IP weights). Because previous research suggests that there may be gender differences in the impacts of food insufficiency on health outcomes,11,14 an interaction term between gender and each food insufficiency term was included. In secondary analyses, models with an interaction term instead between race and each food insufficiency term were fitted; this analysis was motivated by previous research suggesting that food insufficiency may have differential impacts by race owing to socially driven factors, including differences in coping strategies and diet quality.32,39,40 Owing to the small sample size, the analysis did not include interaction terms for effect modification by both race and gender.

Parameter estimates from the marginal structural models were used to estimate the differences in mean outcome values for each of the food insufficiency patterns compared with those of food sufficiency. The 95% CIs were constructed using nonparametric bootstrapping with 5,000 replications. All analyses were performed in 2020 using RStudio, version 1.3.959.

RESULTS

Table 1 shows the characteristics of the analytic sample in 2000–2001 (N=2,596). The sample was approximately half women and half White, with a mean age of 40 years. About 20% of participants reported food insufficiency at baseline. Appendix 3 (available online) reports the baseline characteristics of the uncensored sample by time-varying food insufficiency pattern.

In unadjusted analyses, all food insufficiency patterns were associated with higher BMI and waist circumference than food sufficiency (Figure 1). After accounting for covariates, point estimates were attenuated, but directions were generally still consistent with adverse effects of food insufficiency compared with those of food sufficiency on adiposity (although estimates were imprecise with CIs often overlapping the null value of 0). When generating estimates for men and women separately, the estimated associations between food insufficiency and adiposity were often stronger among women, although the differences in the effect estimates for women from those for men did not reach statistical significance at a 2-sided α=5% (Appendix 4, available online, Table 4.1).

Figure 1. Unadjusted and IP-weighted estimates of the differences in BMI and WC across food insufficiency patterns.

Figure 1.

Open in a new tab

Note: Panel A: BMI, kg/m2. Panel B: WC, cm. Unadjusted analyses were conducted among participants with complete exposure and outcome information (no requirement for non-missing covariate data, no additional inclusion/exclusion criteria). IP-weighted analyses were conducted among participants with complete 2000–2001 data on exposure and covariates who had not previously had a myocardial infarction or stroke, were not currently pregnant, and had an annual household income <$100,000. Covariates included gender, race, age, recruitment center, household income, employment status, marital status, household size, cholesterol or BP medication use, self-reported diabetes, smoking status, physical activity score, and previous BMI in 2000–2001 and 2005–2006 when examining BMI as the outcome in 2010–2011 (with previous values of other outcomes adjusted for in an analogous manner). Data from both censored and uncensored participants was used in the construction of IP weights (n=2,596). MSMs were fit among uncensored participants only (n=1,897; no food insufficiency, n=1,437; food insufficient 2000–2001, n=164; food insufficient 2005–2006, n=170; persistently food insufficient, n=126) using weighted data.

IP, inverse probability; MSM, marginal structural model; WC, waist circumference.

In unadjusted analyses, all food insufficiency patterns were associated with higher systolic and diastolic BP than food sufficiency (Table 2). In IP-weighted results, no statistically significant differences in systolic or diastolic BP were observed for any food insufficiency pattern, although point estimates for both transient food insufficiency patterns were consistent with higher BP than food sufficiency. Some gender-specific associations emerged: for example, women with transient food insufficiency in 2005–2006 had significantly higher diastolic BP than those with food sufficiency (an effect estimate that was significantly different from the effect observed among men).

Table 2.

Unadjusted and IP-Weighted Estimates of Differences in Other Outcomes Across Food Insufficiency Patterns

Outcomes Food sufficiency:
food sufficient in
2000–2001 and
2005–2006		 Food insufficient in 2000
–2001 only: food
insufficient in 2000–2001,
food sufficient in 2005
–2006, estimate (95% CI)		 Food insufficient in 2005
–2006 only: food
sufficient in 2000–2001,
food insufficient in 2005
–2006, estimate (95% CI)		 Persistent food
insufficiency: food
insufficient in 2000
–2001 and 2005–2006,
estimate (95% CI)
Overall sample
 Systolic BP, mmHg
  Unadjusted ref 4.08 (1.69, 6.56) 4.87 (2.57, 7.31) 4.24 (1.40, 7.07)
  IP weighted —overall ref 1.91 (−0.62, 4.23) 1.62 (−0.64, 3.86) −0.31 (−3.27, 2.40)
  IP weighted—men ref 3.31 (−0.22, 6.66) 0.53 (−1.80, 2.90) 1.45 (−3.65, 6.10)
  IP weighted—women ref 0.83 (−2.65, 4.14) 2.47 (−1.12, 6.08) −1.66 (−4.97, 1.43)
 Diastolic BP, mmHg
  Unadjusted ref 3.07 (1.38, 4.73) 3.33 (1.80, 4.95) 2.28 (0.45, 4.09)
  IP weighted—overall ref 1.55 (−0.21, 3.26) 1.24 (−0.23, 2.72) −0.37 (−2.54, 1.69)
  IP weighted—men ref 2.67 (0.02, 5.28) −0.81 (−2.85, 1.29) 1.19 (−2.96, 4.93)
  IP weighted—women ref 0.68 (−1.60, 2.92) 2.83 (0.74, 4.97) −1.57 (−3.73, 0.50)
 Total cholesterol, mg/dL
  Unadjusted ref −2.20 (−7.75, 3.45) 1.08 (−4.79, 7.06) −8.07 (−13.74, −2.31)
  IP weighted—overall ref −1.38 (−6.86, 3.95) −2.00 (−7.66, 3.78) −2.67 (−8.41, 2.95)
  IP weighted—men ref −1.89 (−10.91, 6.68) −5.60 (−14.61, 3.36) −1.07 (−11.17, 8.94)
  IP weighted—women ref −0.98 (−8.07, 6.04) 0.77 (−6.61, 8.38) −3.91 (−11.05, 2.90)
 HDL cholesterol, mg/dL
  Unadjusted ref 0.26 (−2.76, 3.59) −1.66 (−3.94, 0.71) −3.59 (−5.94, −1.26)
  IP weighted—overall ref 0.39 (−2.60, 3.53) −0.61 (−2.42, 1.25) −3.70 (−6.08, −1.28)
  IP weighted—men ref −1.98 (−5.14, 1.17) 3.40 (0.68, 6.53) −2.75 (−6.51, 1.13)
  IP weighted—women ref 2.22 (−2.39, 7.24) −3.71 (−6.14, −1.21) −4.44 (−7.59, −1.38)
Fasting sample (≥12 hours)
 Triglycerides, mg/dL
  Unadjusted ref 2.14 (−10.74, 16.34) 0.88 (−9.18, 11.77) 2.82 (−10.69, 19.56)
  IP weighted—overall ref −4.63 (−15.93, 8.10) 1.55 (−7.70, 11.11) −1.98 (−12.51, 9.64)
  IP weighted—men ref −1.40 (−20.90, 22.60) −11.82 (−26.65, 4.13) −11.82 (−26.65, 4.13)
  IP weighted—women ref −7.03 (−19.17, 5.85) 11.49 (−0.89, 23.83) 2.83 (−8.22, 13.88)
 LDL cholesterol, mg/dL
  Unadjusted ref −3.07 (−8.08, 2.05) 3.06 (−2.65, 9.01) −3.25 (−8.22, 1.71)
  IP weighted—overall ref −0.52 (−6.65, 5.56) 2.42 (−3.74, 8.89) −2.73 (−8.67, 3.16)
  IP weighted—men ref 0.10 (−7.75, 8.78) −0.22 (−10.01, 10.48) −3.18 (−12.27, 6.11)
  IP weighted—women ref −0.98 (−9.67, 7.49) 4.39 (−4.12, 12.50) −2.39 (−10.06, 5.09)
Fasting sample (≥8 hours)
 Fasting plasma glucose, mg/dL
  Unadjusted ref −1.11 (−4.64, 2.78) 3.39 (−1.21, 8.51) 5.39 (−0.25, 11.70)
  IP weighted—overall ref −3.87 (−7.25, −0.44) 0.43 (−4.27, 5.59) 2.14 (−3.34, 7.87)
  IP weighted—men ref −2.26 (−7.22, 2.99) −3.42 (−9.00, 2.76) 4.05 (−4.92, 14.64)
  IP weighted—women ref −5.09 (−9.64, −0.56) 3.37 (−3.64, 11.23) 0.69 (−5.99, 7.92)
Open in a new tab

Note: Unadjusted analyses in the overall sample were conducted among participants with complete exposure and outcome information (no requirement for non-missing covariate data, no additional inclusion/exclusion criteria). Unadjusted analyses in fasting samples were conducted among participants with complete exposure and outcome information who had fasted for the adequate time period in 2010–2011 (≥12 hours for triglycerides and LDL, ≥8 hours for plasma glucose) (no requirement for non-missing covariate data, no additional inclusion/exclusion criteria). IP-weighted analyses in the overall sample were conducted among participants with complete 2000–2001 data on exposure and covariates who had not previously had a myocardial infarction or stroke, were not currently pregnant, and had an annual household income <$100,000. IP-weighted analyses in fasting sample were conducted among participants with complete 2000–2001 data on exposure and covariates who had not previously had a myocardial infarction or stroke, were not currently pregnant, had an annual household income <$100,000, and had fasted for the adequate time period (≥12 hours for triglycerides and LDL, ≥8 hours for plasma glucose). Covariates included gender, race, age, recruitment center, household income, employment status, marital status, household size, cholesterol or blood pressure medication use, self-reported diabetes, smoking status, physical activity score, and previous BMI in 2000–2001 and 2005–2006 when examining BMI as the outcome in 2010–2011 (with previous values of other outcomes adjusted for in an analogous manner). Data from both censored and uncensored participants were used in the construction of IP weights (n=2,596 for systolic BP, diastolic BP, total cholesterol, and HDL; n=2,311 for LDL and triglycerides; n=2,441 for fasting plasma glucose). MSMs were fit among uncensored participants only. For systolic BP, diastolic BP, total cholesterol, and HDL, this included n=1,897 participants (no food insufficiency, n=1,437; food insufficient 2000–2001, n=164; food insufficient 2005–2006, n=170; persistently food insufficient, n=126). For LDL and triglycerides, this included n=1,379 participants (no food insufficiency, n=1,072; food insufficient 2000–2001, n=102; food insufficient 2005–2006, n=114; persistently food insufficient, n=91). For fasting plasma glucose, this included n=1,605 participants (no food insufficiency, n=1,232; food insufficient 2000–2001, n=128; food insufficient 2005–2006, n=141; persistently food insufficient, n=104).

BP, blood pressure; HDL, high-density lipoprotein; IP, inverse probability; LDL, low-density lipoprotein; MSM, marginal structural model.

Point estimates of the associations between food insufficiency and lipids were less consistent, with wide CIs and estimates that varied in direction and magnitude depending on the food insufficiency pattern and gender. In IP-weighted results, persistent food insufficiency was associated with lower HDL than food sufficiency. Both food insufficiency in 2005–2006 and persistent food insufficiency were associated with significantly lower HDL among women than food sufficiency, whereas food insufficiency in 2005–2006 was associated with significantly higher HDL among men.

In secondary analyses, where the effects of food insufficiency were estimated by race, some race-specific associations emerged: for example, Black participants with food insufficiency in 2005–2006 had significantly higher systolic and diastolic BP than those with food sufficiency (Table 4.2 in Appendix 4, available online). Although this impact of food insufficiency on BP in 2005–2006 versus that of food sufficiency was significantly higher among Black participants than among White participants, no other differences in effect estimates by race reached statistical significance (Table 4.3 in Appendix 4, available online).

DISCUSSION

This study analyzed longitudinal relationships between food insufficiency—a screener measure of food insecurity—and several CVD risk factors. Food insufficiency patterns were generally associated with higher BMI, waist circumference, and BP and lower HDL than food sufficiency, with some gender- and race-specific patterns. These longitudinal findings are unique to the literature because they were generated with an analytic approach that can be used to estimate the impacts of exposures in the presence of time-varying confounding and selection bias.35,36 The findings also have important health policy implications: now more than ever, research linking food insecurity to poor health outcomes is needed to guide nutrition policies and programs.

Consistent with previous cross-sectional studies,11 food insufficiency was associated with higher BMI and larger waist circumference than food sufficiency (particularly among women). Longitudinal studies examining this relationship are limited and have reported mixed results.23-25,41,42 The differences between this study’s findings and previous longitudinal studies that reported null results may be explained by different exposure assessments (other studies assessed food insecurity), distinct study populations (several only included pregnant women or young mothers), shorter follow-up periods (all were <5 years), and differing analytic approaches (none adjusted for time-varying confounding). Contrary to the study’s hypothesis, no clear evidence was found that persistent food insecurity is worse for health than transient food insecurity, a finding that could be a true effect (e.g., owing to the development of more effective coping mechanisms over time) or due to bias (e.g., residual confounding and selection bias).

Researchers have proposed a number of possible mechanisms by which food insecurity may increase adiposity and CVD risk. Food-insecure households often cycle between periods of food adequacy and scarcity,12 resulting in the development of compensatory strategies (e.g., overconsumption of calories when available or skipping meals when food is limited) and constrained food choices (e.g., a reliance on cheap, nutrient-poor foods).12 In particular, studies have found that food-insecure individuals have lower micronutrient intakes (e.g., iron),43-45 eat fewer fruits and vegetables,46-48 and consume more added sugars.49,50 These dietary behaviors may lead to metabolic dysregulation and adipose accumulation.20 Food insecurity may also act as a chronic stressor that can elevate CVD risk factors either directly or by triggering unhealthful coping behaviors (e.g., smoking or excessive drinking).20 More research is needed to examine which behavioral or physiologic responses to food insecurity are most related to disease risk.

In line with previous research,11,14 this study’s findings suggest that women may experience more adverse effects on adiposity (particularly waist circumference) from food insufficiency than from food sufficiency. Although CARDIA does not distinguish between biological sex and gender, it is possible that both are contributors to the observed differences in this study. For example, societal gender norms may mean that mothers feel pressure to put their children’s needs first, which may result in the adoption of unhealthy coping strategies to protect their family when the food supply is threatened (e.g., skipping meals).51,52 With respect to biological sex, the accumulation of fat as a physiologically regulated response to a reduced food supply may happen disproportionately among women because of the important role adiposity plays in reproduction and offspring survival.53 More research is needed to understand what mechanisms are driving effect modification by sex and gender to develop targeted strategies to reduce the disproportionate impact of food insecurity among female individuals/women.

This study’s findings also have health equity implications. First, this study contributes to mounting evidence on racial disparities in the burden of food insecurity,6,54 with 23% of Black participants reporting food insufficiency at baseline, compared with 12% of White participants. Second, this study reports some race-specific findings for the impacts of food insufficiency on CVD risk factors, with transient food insufficiency associated with higher BP than food sufficiency among Black participants. Some possible explanations for this variation in findings by race include the differences in neighborhood food environment and different coping strategies and diet quality during times of food insecurity.38,55,56 Moving forward, there is a need for adequately powered longitudinal studies to examine research questions around how structural racism may contribute to differences in the impacts of food insecurity on health.

Future research may also expand on our work to construct additional, policy-relevant assessments of the health impact of food insecurity that depend on time-evolving risk factors (e.g., income) or are based on realistic changes in food insecurity (e.g., a 30% reduction in food insecurity, consistent with an estimated impact of Supplemental Nutrition Assistance Program), as outlined in Appendix 5 (available online).

Limitations

This study has multiple limitations. First, CARDIA assessed food insufficiency, not food insecurity. Moreover, owing to small sample sizes, responses reporting a lack of sufficient quantity or quality of food were collapsed together, meaning that it was not possible to assess effects by food insufficiency severity. Next, this study’s primary analysis did not account for interim CVD events (e.g., stroke). However, in sensitivity analyses treating interim CVD events as a censoring criterion and a time-varying covariate (Appendix 6, available online), results did not vary meaningfully. In addition, because CARDIA does not ask participants about participation in nutrition assistance programs, it was not possible to incorporate these factors into the analysis. However, a sensitivity analysis treating dietary variables as baseline and time-varying covariates is included in Appendix 7 (available online). In addition, it was assumed that smoking and physical activity were confounders, but it is plausible that they are instead mediators.57,58 Finally, although this study’s interpretation of results did not focus on statistical significance, it examined a number of different but correlated outcomes, which may raise concerns about multiple comparisons

Despite these limitations, the study has many strengths. CARDIA’s study design allowed for examination of relationships prospectively over 10 years of follow-up, increasing confidence in the temporal ordering of exposure before outcome. In addition, an analytic method was used that can estimate the effects of exposures in the presence of time-varying confounding and selection bias.35,36

CONCLUSIONS

Food insufficiency—one measure of food insecurity—was associated with several CVD risk factors. Findings from this study should be replicated in other settings and populations. If verified, these associations could provide further justification for intervening in food insecurity.

Supplementary Material

Appendix

ACKNOWLEDGMENTS

The authors would like to thank the investigators and staff of the Coronary artery Risk Development in Young Adults Study (CARDIA) study. A full list of participating CARDIA investigators and institutions is available at www.cardia.dopm.uab.edu/. The authors would also like to thank Dr. Sara Bleich and Dr. Erica Kenney for their helpful comments. EBR and JY made equal contributions.

The CARDIA study is conducted and supported by the National Heart, Lung, and Blood Institute in collaboration with the University of Alabama at Birmingham (HHSN268201800005I and HHSN268201800007I), Northwestern University (HHSN268201800003I), University of Minnesota (HHSN268201800006I), and Kaiser Foundation Research Institute (HHSN268201800004I). This manuscript has been reviewed by CARDIA for scientific content.

KV was supported by a Canadian Institute of Health Research doctoral foreign study award (#0492002603). PGL is supported by R01HL143885. ACM was supported by a National Heart, Lung, and Blood Institute Mentored Research Scientist Development K01 Award (HL150406). APC has received investigator-initiated funding from Amgen, Inc. for unrelated work. No other financial disclosures were reported.

Footnotes

CREDIT AUTHOR STATEMENT

Kelsey Vercammen: Conceptualization, Formal analysis, Funding acquisition, Methodology, Software, Visualization, Writing-original draft. Alyssa Moran: Conceptualization, Methodology, Writing-review & editing. Mercedes Carnethon: Conceptualization, Data curation, Methodology, Resources, Writing-review & editing. Amanda C. McClain: Conceptualization, Methodology, Writing-review & editing. Lindsay R. Pool: Methodology, Writing-review & editing. Catarina I. Kiefe: Data Curation, Methodology, Resources, Writing-review & editing. April P. Carson: Data curation Methodology, Resources, Writing-review & editing. Penny Gordon-Larsen: Methodology, Writing-review & editing. Lyn M. Steffen: Data curation, Methodology, Resources, Writing-review & editing. Matthew Lee: Software, Validation, Writing-review & editing. Jessica Young: Conceptualization, Methodology, Software, Supervision, Writing-review & editing. Eric B. Rimm: Conceptualization, Funding acquisition, Methodology, Supervision, Writing-review & editing.

SUPPLEMENTAL MATERIAL

Supplemental materials associated with this article can be found in the online version at https://doi.org/10.1016/j.amepre.2021.06.020.

REFERENCES

  • 1.U.S. Department of Agriculture. Definitions of food security. Economic Research Service. https://www.ers.usda.gov/topics/food-nutrition-assistance/food-security-in-the-us/definitions-of-food-security.aspx. Updated September 9, 2020. Accessed January 30 2021. [Google Scholar]
  • 2.Food and Agriculture Organization, International Fund for Agricultural Development, United Nations Children’s Emergency Fund, World Food Programme, WHO. The state of food security and nutrition in the world 2017: building resilience for peace and food security. Rome, Italy: Food and Agriculture Organization; 2017. http://www.fao.org/3/a-I7695e.pdf. Published 2017. Accessed August 2, 2021. [Google Scholar]
  • 3.Andrews M, Bickel G, Carlson S. Household food security in the United States in 1995: results from the Food Security Measurement Project. Fam Econ Rev. 1998;11(1/2):17–28. https://www.fns.usda.gov/household-food-security-united-states-1995-summary-report-food-security-measurement-project. Accessed December 18, 2020. [Google Scholar]
  • 4.USDA nutrition assistance programs. National Agricultural Library, U.S. Department of Agriculture. https://www.nal.usda.gov/fnic/usda-nutrition-assistance-programs. Accessed December 18, 2020. [Google Scholar]
  • 5.U.S. Department of Agriculture, Food and Nutrition Service. Nutrition assistance programs Keydata release. Washington, DC: U.S. Department of Agriculture, Food and Nutrition Service; 2021. www.fns.usda.gov/data/april-2021-keydata-report. Published 2021. Accessed August 13, 2021. [Google Scholar]
  • 6.Coleman-Jensen A, Rabbitt MP, Gregory CA, Singh A. Household food security in the United States in 2019. Washington, DC: Economic Research Service. U.S. Department of Agriculture; 2020. https://www.ers.usda.gov/publications/pub-details/?pubid=99281. Published 2020. Accessed August 2, 2021. [Google Scholar]
  • 7.Schanzenbach D, Pitts A. Estimates of food insecurity during the COVID-19 crisis: results from the COVID impact survey, week 1 (April 20–26, 2020). Evanston, IL: Institute for Policy Research Rapid Research Report, Northwestern University; 2020. https://www.ipr.northwestern.edu/documents/reports/food-insecurity-covid_week1_report-13-may-2020.pdf. Published 2020. Accessed August 2, 2021. [Google Scholar]
  • 8.Schanzenbach D, Pitts A. Estimates of food insecurity during the COVID-19 crisis: results from the COVID impact survey, week 2 (May 4–10, 2020). Evanston, IL: Institute for Policy Research Rapid Research Report, Northwestern University; 2020. https://www.ipr.northwestern.edu/documents/reports/food-insecurity-covid_week2_-report-18-may-2020.pdf. Published 2020. Accessed August 2, 2021. [Google Scholar]
  • 9.Gregory CA, Coleman-Jensen A. Food insecurity, chronic disease, and health among working-age adults. Washington, DC: U.S. Department of Agriculture, Economic Research Service; 2017. https://www.ers.usda.gov/webdocs/publications/84467/err-235.pdf?v=0. Published 2017. Accessed August 2, 2021. [Google Scholar]
  • 10.Gundersen C, Ziliak JP. Food insecurity and health outcomes. Health Aff (Millwood). 2015;34(11):1830–1839. 10.1377/hlthaff.2015.0645. [DOI] [PubMed] [Google Scholar]
  • 11.Larson NI, Story MT. Food insecurity and weight status among U.S. children and families: a review of the literature. Am J Prev Med. 2011;40(2):166–173. 10.1016/j.amepre.2010.10.028. [DOI] [PubMed] [Google Scholar]
  • 12.Seligman HK, Schillinger D. Hunger and socioeconomic disparities in chronic disease. N Engl J Med. 2010;363(1):6–9. 10.1056/NEJMp1000072. [DOI] [PubMed] [Google Scholar]
  • 13.Seligman HK, Bindman AB, Vittinghoff E, Kanaya AM, Kushel MB. Food insecurity is associated with diabetes mellitus: results from the National Health Examination and Nutrition Examination Survey (NHANES) 1999-2002. J Gen Intern Med. 2007;22(7):1018–1023. 10.1007/s11606-007-0192-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Franklin B, Jones A, Love D, Puckett S, Macklin J, White-Means S. Exploring mediators of food insecurity and obesity: a review of recent literature. J Community Health. 2012;37(1):253–264. 10.1007/s10900-011-9420-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Berkowitz SA, Berkowitz TSZ, Meigs JB, Wexler DJ. Trends in food insecurity for adults with cardiometabolic disease in the United States: 2005-2012. PLoS One. 2017;12(6):e0179172. 10.1371/journal.pone.0179172. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Ford ES. Food security and cardiovascular disease risk among adults in the United States: findings from the National Health and Nutrition Examination Survey, 2003–2008. Prev Chronic Dis. 2013;10:E202. 10.5888/pcd10.130244. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Holben DH, Taylor CA. Food insecurity and its association with central obesity and other markers of metabolic syndrome among persons aged 12 to 18 years in the United States. J Am Osteopath Assoc. 2015;115(9):536–543. 10.7556/jaoa.2015.111. [DOI] [PubMed] [Google Scholar]
  • 18.Parker ED, Widome R, Nettleton JA, Pereira MA. Food security and metabolic syndrome in U.S. adults and adolescents: findings from the National Health and Nutrition Examination Survey, 1999–2006. Ann Epidemiol. 2010;20(5):364–370. 10.1016/j.annepidem.2010.02.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Vercammen KA, Moran AJ, McClain AC, Thorndike AN, Fulay AP, Rimm EB. Food security and 10-year cardiovascular disease risk among U.S. adults. Am J Prev Med. 2019;56(5):689–697. 10.1016/j.amepre.2018.11.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Laraia BA. Food insecurity and chronic disease. Adv Nutr. 2013;4(2):203–212. 10.3945/an.112.003277. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Roth GA, Johnson C, Abajobir A, et al. Global, regional, and national burden of cardiovascular diseases for 10 causes, 1990 to 2015. J Am Coll Cardiol. 2017;70(1):1–25. 10.1016/j.jacc.2017.04.052. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Wang YC, McPherson K, Marsh T, Gortmaker SL, Brown M. Health and economic burden of the projected obesity trends in the USA and the UK [published correction appears in Lancet. 2011;378(9805):1778]. Lancet. 2011;378(9793):815–825. 10.1016/S0140-6736(11)60814-3. [DOI] [PubMed] [Google Scholar]
  • 23.Jones SJ, Frongillo EA. The modifying effects of Food Stamp Program participation on the relation between food insecurity and weight change in women. J Nutr. 2006;136(4):1091–1094. 10.1093/jn/136.4.1091. [DOI] [PubMed] [Google Scholar]
  • 24.Whitaker RC, Sarin A. Change in food security status and change in weight are not associated in urban women with preschool children. J Nutr. 2007;137(9):2134–2139. 10.1093/jn/137.9.2134. [DOI] [PubMed] [Google Scholar]
  • 25.Olson CM, Strawderman MS. The relationship between food insecurity and obesity in rural childbearing women. J Rural Health. 2008;24(1):60–66. 10.1111/j.1748-0361.2008.00138.x. [DOI] [PubMed] [Google Scholar]
  • 26.Hernán MA, Robins JM. Treatment-confounder feedback what if. Boca Raton, FL: Chapman & Hall/CRC; 2020. https://cdn1.sph.harvard.edu/wp-content/uploads/sites/1268/2021/03/ciwhatif_hernanrobins_30mar21.pdf. Published 2020. Accessed August 2, 2021. [Google Scholar]
  • 27.Rose D, Oliveira V. Validation of a self-reported measure of household food insufficiency with nutrient intake data. Washington, DC: U.S. Department of Agriculture; 1997. https://naldc.nal.usda.gov/download/CAT10862471/PDF. Published 1997. Accessed August 2, 2021. [Google Scholar]
  • 28.CARDIA overview. Division of Preventive Medicine, NIH. https://www.cardia.dopm.uab.edu/cardia-overview/overview-more. Accessed July 31, 2020.
  • 29.Barrett CB. Measuring food insecurity. Science. 2010;327(5967):825–828. 10.1126/science.1182768. [DOI] [PubMed] [Google Scholar]
  • 30.Weaver LJ, Hadley C. Moving beyond hunger and nutrition: a systematic review of the evidence linking food insecurity and mental health in developing countries. Ecol Food Nutr. 2009;48(4):263–284. 10.1080/03670240903001167. [DOI] [PubMed] [Google Scholar]
  • 31.Schanzenbach D, Pitts A. Food insecurity in the census household pulse survey data tables. Northwestern Institute for Policy Research. June 1, 2020. https://www.ipr.northwestern.edu/news/2020/schanzenbach-household-pulse-survey-analysis.html. Accessed August 3, 2021. [Google Scholar]
  • 32.Pool LR, Kershaw KN, Gordon-Larsen P, et al. Racial differences in the associations between food insecurity and fibroblast growth factor 23 in the Coronary Artery Risk Development in Young Adults Study. J Ren Nutr. 2020;30(6):509–517. 10.1053/j.jrn.2020.01.020. [DOI] [PubMed] [Google Scholar]
  • 33.Year 25 (2010–2011) manual of operation. Division of Preventive Medicine, NIH. https://www.cardia.dopm.uab.edu/exam-materials2/manual-of-operations/year-25. Accessed August 3, 2021. [Google Scholar]
  • 34.Jacobs DR Jr, Hahn LP, Haskell WL, Pirie P, Sidney S. Validity and reliability of short physical activity history: CARDIA and the Minnesota Heart Health Program. J Cardiopulm Rehabil. 1989;9(11):448–459. 10.1097/00008483-198911000-00003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Cole SR, Hernán MA. Constructing inverse probability weights for marginal structural models. Am J Epidemiol. 2008;168(6):656–664. 10.1093/aje/kwn164. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Robins JM, Hernán MA, Brumback B. Marginal structural models and causal inference in epidemiology. Epidemiology. 2000;11(5):550–560. 10.1097/00001648-200009000-00011. [DOI] [PubMed] [Google Scholar]
  • 37.Kupzyk KA, Beal SJ. Advanced issues in propensity scores: longitudinal and missing data. J Early Adolesc. 2017;37(1):59–84. 10.1177/0272431616636229. [DOI] [Google Scholar]
  • 38.Robins JM, Hernán MA. Estimation of the causal effects of time-varying exposures. In: Fitzmaurice G, Davidian M, Verbeke G, Molenbergs G, eds. Longitudinal Data Analysis. Boca Raton, FL: CRC Press, 2009:553–599. https://cdn1.sph.harvard.edu/wp-content/uploads/sites/343/2013/03/abc.pdf. Accessed March 31, 2021. [Google Scholar]
  • 39.Kamdar N, Rozmus CL, Grimes DE, Meininger JC. Ethnic/racial comparisons in strategies parents use to cope with food insecurity: a systematic review of published research. J Immigr Minor Health. 2019;21(1):175–188. 10.1007/s10903-018-0720-y. [DOI] [PubMed] [Google Scholar]
  • 40.McClain AC, Xiao RS, Gao X, Tucker KL, Falcon LM, Mattei J. Food insecurity and odds of high allostatic load in Puerto Rican adults: the role of participation in the Supplemental Nutrition Assistance Program during 5 years of follow-up. Psychosom Med. 2018;80(8):733–741. 10.1097/PSY.0000000000000628. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Cheung HC, Shen A, Oo S, Tilahun H, Cohen MJ, Berkowitz SA. Food insecurity and body mass index: a longitudinal mixed methods study, Chelsea, Massachusetts, 2009–2013. Prev Chronic Dis. 2015;12:E125. 10.5888/pcd12.150001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Jones SJ, Frongillo EA. Food insecurity and subsequent weight gain in women. Public Health Nutr. 2007;10(2):145–151. 10.1017/S1368980007246737. [DOI] [PubMed] [Google Scholar]
  • 43.Rose D, Oliveira V. Nutrient intakes of individuals from food-insufficient households in the United States. Am J Public Health. 1997;87(12):1956–1961 10.2105/ajph.87.12.1956. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Dixon LB, Winkleby MA, Radimer KL. Dietary intakes and serum nutrients differ between adults from food-insufficient and food-sufficient families: third National Health and Nutrition Examination Survey, 1988–1994. J Nutr. 2001;131(4):1232–1246. 10.1093/jn/131.4.1232. [DOI] [PubMed] [Google Scholar]
  • 45.Lee JS, Frongillo EA Jr.. Nutritional and health consequences are associated with food insecurity among U.S. elderly persons. J Nutr. 2001;131(5):1503–1509. 10.1093/jn/131.5.1503. [DOI] [PubMed] [Google Scholar]
  • 46.Tarasuk VS. Household food insecurity with hunger is associated with women’s food intakes, health and household circumstances. J Nutr. 2001;131(10):2670–2676. 10.1093/jn/131.10.2670. [DOI] [PubMed] [Google Scholar]
  • 47.Robaina KA, Martin KS. Food insecurity, poor diet quality, and obesity among food pantry participants in Hartford, CT. J Nutr Educ Behav. 2013;45(2):159–164. 10.1016/j.jneb.2012.07.001. [DOI] [PubMed] [Google Scholar]
  • 48.Dharod JM, Croom JE, Sady CG. Food insecurity: its relationship to dietary intake and body weight among Somali refugee women in the United States. J Nutr Educ Behav. 2013;45(1):47–53. 10.1016/j.jneb.2012.03.006. [DOI] [PubMed] [Google Scholar]
  • 49.Nguyen BT, Shuval K, Bertmann F, Yaroch AL. The Supplemental Nutrition Assistance Program, food insecurity, dietary quality, and obesity among U.S. adults. Am J Public Health. 2015;105(7):1453–1459. 10.2105/AJPH.2015.302580. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Leung CW, Epel ES, Ritchie LD, Crawford PB, Laraia BA. Food insecurity is inversely associated with diet quality of lower-income adults. J Acad Nutr Diet. 2014;114(12):1943–1953 e2. 10.1016/j.jand.2014.06.353. [DOI] [PubMed] [Google Scholar]
  • 51.McIntyre L, Glanville NT, Raine KD, Dayle JB, Anderson B, Battaglia N. Do low-income lone mothers compromise their nutrition to feed their children? CMAJ. 2003;168(6):686–691. [PMC free article] [PubMed] [Google Scholar]
  • 52.Martin MA, Lippert AM. Feeding her children, but risking her health: the intersection of gender, household food insecurity and obesity. Soc Sci Med. 2012;74(11):1754–1764. 10.1016/j.socscimed.2011.11.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Dhurandhar EJ. The food-insecurity obesity paradox: a resource scarcity hypothesis. Physiol Behav. 2016;162:88–92. 10.1016/j.physbeh.2016.04.025. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54.Odoms-Young A, Bruce MA. Examining the impact of structural racism on food insecurity: implications for addressing racial/ethnic disparities. Fam Community Health. 2018;41(suppl 2):S3–S6. 10.1097/FCH.0000000000000183. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 55.Allen AJ, Kuczmarski MF, Evans MK, Zonderman AB, Waldstein SR. Race differences in diet quality of urban food-insecure blacks and whites reveals resiliency in blacks. J Racial Ethn Health Disparities. 2016;3(4):706–712. 10.1007/s40615-015-0189-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 56.Morland K, Wing S, Diez Roux A, Poole C. Neighborhood characteristics associated with the location of food stores and food service places. Am J Prev Med. 2002;22(1):23–29. 10.1016/s0749-3797(01)00403-2. [DOI] [PubMed] [Google Scholar]
  • 57.American Academy of Pediatrics. The relationship between food insecurity and physical inactivity. AAP Grand Rounds. 2015;33(1):9. 10.1542/gr.33-1-9. [DOI] [Google Scholar]
  • 58.Kim-Mozeleski JE, Pandey R. The intersection of food insecurity and tobacco use: a scoping review. Health Promot Pract. 2020;21(1)(suppl):124S–138S. 10.1177/1524839919874054. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Appendix
NIHMS1749286-supplement-Appendix.pdf (544.9KB, pdf)

RESOURCES