Food Insecurity and Cardiometabolic Markers: Results From the Study of Latino Youth

Abstract

OBJECTIVES

Hispanic/Latino youth bear a disproportionate burden of food insecurity and poor metabolic outcomes, but research linking the two in this diverse population is lacking. We evaluated whether lower household and child food security (FS) were adversely associated with a metabolic syndrome (MetS) composite variable and clinically measured cardiometabolic markers: waist circumference, fasting plasma glucose, triglycerides, high-density lipoprotein cholesterol, and systolic and diastolic blood pressure.

METHODS

This cross-sectional study included 1325 Hispanic/Latino youth aged 8 to 16 years from the Hispanic Community Children’s Health Study/Study of Latino Youth, a study of offspring of adults enrolled in the Hispanic Community Health Survey/Study of Latinos. Multivariable regression analyses were used to assess relationships between household FS (high, marginal, low, very low) and child FS (high, marginal, low/very low) status, separately, and our dependent variables, adjusting for participant age, sex, site, parental education, and poverty-income ratio.

RESULTS

For both FS measures, youth in the lowest FS category had significantly lower high-density lipoprotein cholesterol than those with high FS (household FS: −3.17, 95% confidence interval [CI]: −5.65 to −0.70, child FS: −1.81, 95% CI: −3.54 to −0.09). Low/very low versus high child FS was associated with greater fasting plasma glucose (β = 1.37, 95% CI: 0.08 to 2.65), triglycerides (β = 8.68, 95% CI: 1.75 to 15.61), and MetS expected log counts (β = 2.12, 95% CI: 0.02 to 0.45).

CONCLUSIONS

Lower FS is associated with unfavorable MetS-relevant cardiometabolic markers in Hispanic/Latino youth. These findings also support the use of a child-level versus a household-level measure to capture the health implications of food insecurity in this population.

What’s Known on This Subject:

Early appearance of components of the metabolic syndrome and a high prevalence of low food security have been documented among US Hispanic/Latino youth. However, it is unknown if the two are associated in this population.

What This Study Adds:

This study is among the first to document adverse associations between household and child food security measures with cardiometabolic markers among Hispanic/Latino youth. It also supports using the child-level versus the household-level measure to capture the health implications of food insecurity in this population.

Food insecurity (FI) is characterized by limited or uncertain access to nutritionally adequate and safe foods or the ability to obtain such foods in socially acceptable ways.^1,2 FI has been associated with reduced quality of food, disrupted eating patterns,³ and high levels of stress,^4,5 all of which can adversely affect health.^4,6 Despite abundant food availability in higher-income countries like the United States, FI affects ∼14% of households with youth,⁷ a disproportionate number of whom are Hispanics/Latinos.^8–10 In fact, 2012 to 2014 data from the Hispanic Community Children’s Health Study/Study of Latino Youth (SOL Youth), a random sample of Hispanic/Latino youth from 4 large US cities, demonstrated 42% of Hispanic/Latino households with youth reported some level of FI in the previous year¹⁰; meanwhile, 11% reported the highest form of FI, which involves the uneasy/painful sensation of hunger.²

In adults, FI has been consistently associated with several cardiometabolic diseases, including type 2 diabetes,^11–17 obesity,^18–20 hypertension and hyperlipidemia,¹⁴ and other metabolic sequelae,²¹ but little is known about FI’s role on physical health in youth. Among Hispanic/Latino youth, the scarce FI literature is mixed and largely centers on overweight/obesity.^10,22–24 The few studies evaluating links between FI and other cardiometabolic markers (CMMs) in youth were mostly specific to non-Hispanic White individuals and revealed null findings.^16,25 Despite the concurrent disproportionate burden of FI,¹⁰ insulin resistance (IR), and the metabolic syndrome (MetS)^8,10,26–29 among Hispanic/Latino youth, no study, to our knowledge has evaluated FI’s role in MetS and MetS-relevant CMMs in this population.

The health implications of FI may also vary by parental place of birth. Although previous work suggests that healthier diets and strong social and familial ties among foreign-born Hispanic/Latino parents may help protect their children from adopting less healthy US behaviors,³⁰ FI may undermine the protection of these factors.³¹ Thus, we expect associations with MetS and MetS-relevant CMMs to be more pronounced among youth with foreign-born Hispanic/Latino parents or caregivers. Lastly, although food assistance (eg, Supplemental Assistance Nutrition Program [SNAP]) may serve as a potential buffer against poor diet quality and obesity in adults,³² research is needed to better understand whether receiving any food assistance helps minimize the adverse health implications of FI among Hispanic/Latino youth.

Taken together, FI may have health implications for Hispanic/Latino youth. Therefore, this study’s primary aim was to evaluate associations of household and child FI measures with MetS and MetS-relevant CMMs: waist circumference (WC), fasting plasma glucose (FG), high-density lipoprotein cholesterol (HDL-C), triglycerides (TGs), and systolic blood pressure (SBP) and diastolic blood pressure (DBP) measures. Our secondary aim was to investigate whether associations differed by parental place of birth and receipt of any food assistance in the previous year.

Methods

Study Population

Data came from SOL Youth, a study of Hispanic/Latino youth aged 8 to 16 years recruited between 2012 and 2014 from the Bronx, New York, Chicago, Illinois, Miami, Florida, and San Diego, California. Detailed information about the study design and objectives can be found elsewhere.^33,34 Briefly, SOL Youth is an ancillary study that enrolled children and adolescents living with the adults enrolled in the Hispanic Community Health Study/Study of Latinos (HCHS/SOL). HCHS/SOL is a prospective cohort study of US Hispanic/Latino adults of diverse origin recruited through a 2-stage area probability sampling design from the same 4 communities. Study design and procedures of HCHS/SOL have also been previously documented.^35,36 A total of 1466 youth (82% of eligible) were enrolled. Youth were instructed to fast for 10 hours before the baseline clinic visit (2012–2014), which included in-person interviews (in English or Spanish) of youth participants and their accompanying parent or caregiver and a blood draw. Institutions participating in SOL Youth received study approval from their respective institutional review boards. Written informed consent and assent were obtained from participating parents or caregivers and their youth, respectively.

Household and Child Food Insecurity

Most FI studies in youth have used a household versus a child FI measure, but the latter may better reflect FI in the youth as available food in food-insecure households may be prioritized for them. Thus, a child- versus household-level FI measure may show stronger associations.³⁷ Household and child FI was measured by using the United States Department of Agriculture 18-item Household Food Security Survey Module, a well-validated questionnaire which asks parents or caregivers about food security (FS) status over the past 12 months in the household (10 items) and for the youth (8 items).² On the basis of the number of affirmative responses for each set of questions, household and child FS measures were generated, each with 4 United States Department of Agriculture standard categories across the range of severity: household FS (high [0], marginal [1–2], low [3–7], and very low [8–18]) and child FS (high [0], marginal [1], low [2–4], and very low [3–8]). Lower FS categories are often collapsed into a single category to represent FI, but growing evidence suggests marginal, low, and very low FS are distinctly influential to diet and health.⁴ Because of very few observations in the lowest child FS category, however, we combined low and very low child FS categories.

MetS-Relevant CMMs and MetS Composite Score

FG, TGs, and HDL-C were assessed by using fasting (≥ 10-hour) blood samples taken from each participant and processed overnight at the University of Minnesota. FG was measured on a Modular P Chemistry Analyzer by using a hexokinase enzymatic method (Roche Diagnostics, Indianapolis, IN), serum insulin on an Elecsys 2010 Analyzer by using a sandwich immunoassay method (Rocha Diagnostics Corp.), serum TGs with a glycerol blanking enzymatic method (Roche Diagnostics), and HDL-C with a direct magnesium/dextran sulfate method.³⁸ After a 5-minute rest, 3 seated blood pressure measures were recorded from each participant by trained personnel using a standard sphygmomanometer and then averaged. WC was also measured 3 times and averaged per participant by using standard protocol.³⁹ On the basis of a modified child version of the adult-specific guidelines (Adult Treatment Panel III) on cholesterol management issued by the National Cholesterol Education Program,⁴⁰ we created a composite MetS score variable for CMMs constituting MetS.^41–43 For instance, youth received a score of 1 for each CMM criteria met: abdominal adiposity (WC ≥ 90th age- and sex-specific percentiles),⁴⁴ FG (≥ 100 mg/dL),⁴⁵ TGs (≥ 150 mg/dL),⁴⁵ HDL-C (≤ 40 mg/dL),⁴⁵ SBP or DBP (≥ 90th age-, sex-, and height-specific percentiles),⁴⁶ ranging between 0 and 5.

Covariates and Modifiers

Data collected from SOL Youth participants included sex (female or male) and age (years). Information collected from accompanying parents/caregivers included income (lowest and highest reported income categories were <$10 000 and >$100 000, respectively), parental education (less than high school, high school or equivalent, and more than high school), parental place of birth (US-born including born in Puerto Rico and foreign-born), site (Bronx, NY, Chicago, IL, Miami, FL, and San Diego, CA), and receipt of any food assistance in the previous 12 months (yes or no), including SNAP, Women, Infants, and Children (WIC), or National School Lunch Program for free or reduced school meals. We calculated the poverty-income ratio by determining the median income for a given reported income range and dividing by the state-specific federal poverty guidelines for the year in which data collection took place.^47,48

Statistical Analysis

Of the 1466 youth enrolled, 140 participants had missing data on at least 1 MetS-relevant CMM (n = 87) or relevant covariates (n = 53). Our final analytic sample included 1325 SOL Youth participants. All analyses were weighted and accounted for the study’s complex design by using Stata software, version 14.1 (StataCorp LLC, College Station, TX). Because of skewness in the CMM variables, we reported median and interquartile ranges in descriptive analyses and log-transformed all CMMs in regression analyses. We tested differences in sociodemographic and CMM variables across household and child FS categories by using t tests and Mann-Whitney U (Wilcoxon rank sum) tests for continuous variables and Pearson χ² tests for categorical variables. We used multivariable linear regression to estimate associations between household and child FS measures separately and each of the MetS-relevant CMMs (WC, FG, TGs, SBP, DBP, and HDL-C), adjusting for each participant’s sex, age, site, parental education, parental place of birth, and poverty-income ratio. To examine associations with the MetS score variable, we employed multivariable Poisson regression, adjusting for the same covariates. In a set of separate models, we also evaluated adjusted interactions between each of the FS measures and parental place of birth, testing the appropriate interaction terms (FS measures × parental place of birth). In these same models, we additionally adjusted for receipt of any food assistance, as participants may also be living with members of the household who are undocumented and, therefore, ineligible for government food assistance (eg, SNAP). In another set of models, we tested an adjusted interaction term between each FS measure and receipt of any food assistance in the previous year adjusting for the aforementioned covariates. All geometric mean estimates derived from linear regression analyses were back-transformed to their original scales to improve interpretation.⁴⁹ We considered statistical significance at P < .05 and for interactions at P < .10.

Results

Table 1 shows MetS-relevant CMMs and sociodemographic characteristics by household and child FS measures. In general, several CMMs were less favorable with lower FS. For instance, the lowest household and child FS categories had the highest median levels of WC (in cm) and TGs (in mg/dL), and the lowest median levels of HDL-C (in mg/dL). The proportions of youth meeting each of the MetS criteria also tended to be highest among individuals in the lowest categories of FS. In terms of sociodemographic characteristics, lower FS showed lower parental education, a greater proportion of youth receiving any food assistance in the previous year, and lower poverty income ratio, although there was little difference by age and sex.

TABLE 1.

CMMs and Sociodemographic Characteristics by Household and Child FS Status in the SOL Youth (n = 1325)

Characteristics	Household Food Security Status				Child Food Security Status
	High	Marginal	Low	Very Low	High	Marginal	Low/Very Low
	(n = 554)	(n = 220)	(n = 393)	(n = 158)	(n = 740)	(n = 143)	(n = 442)
MetS-relevant CMMsa
WC (cm)	75 (66–85)	72 (64–82)	76 (69–85)	78 (69–87)	74 (66–85)	73 (67–81)	77 (69–87)**
FG (mg/dL)	91 (87–95)	92 (88–96)	92 (88–96)	92 (89–96)	91 (87–95)	92 (87–96)	92 (89–96)
TGs (mg/dL)	66 (48–91)	68 (45–92)	71 (53–96)**	74 (53–103)**	67 (48–91)	68 (51–98)	72 (53–100)**
HDL-C (mg/dL)	52 (45–60)	50 (45–59)	50 (44–58)*	49 (42–55)**	52 (45–60)	52 (44–56)	49 (43–57)**
SBP (mm Hg)	105 (98–111)	104 (97–110)	105 (97–113)	106 (98–113)	105 (98–111)	105 (98–113)	106 (98–113)
DBP (mm Hg)	60 (56–66)	61 (56–65)	60 (56–66)	60 (56–65)	60 (56–65)	61 (56–66)	60 (56–65)
MetS score criteria,b n (%)
Abdominal obesity (≥ 90th percentile)	138 (25.8)	52 (21.1)	108 (23.1)	46 (30.2)	184 (25)	33 (17.3)	27 (26.9)
FG (≥ 100 mg/dL)	52 (8.7)	30 (10.9)	45 (10.2)	21 (12.7)	69 (8.7)	23 (10.8)	56 (12.4)
TGs (≥150 mg/dL)	33 (6.7)	13 (8.1)	29 (7.9)	19 (11.2)	44 (6.3)	7 (7.3)	43 (10.4)
HDL-C (≤40 mg/dL)	80 (11.8)	26 (11.4)	59 (13.5)	32 (20.9)*	101 (11.1)	18 (10.2)	78 (18.0)**
SBP or DBP (≥90th percentile)	25 (5.1)	3 (1.8)	25 (8.3)	9 (6.7)	28 (4.3)	7 (7.8)	27 (7.6)
MetS (score)a	1 (1–2)	1 (1–2)	1 (1–2)	1 (1–2)	1 (1–2)	1 (1–1)	1 (1–2)
Sociodemographics, n (%)
Child age, y	12.2 (2.6)	11.7 (2.7)	12.4 (2.5)	12.3 (2.6)	12.1 (2.6)	11.9 (2.5)	12.4 (2.6)
Female	271 (49)	104 (44.7)	207 (49.3)	79 (51)	365 (49.2)	70 (45.1)	226 (49.1)
Site, n (%)
Bronx	142 (31.2)	64 (35.5)	116 (40.1)	66 (51.9)**	204 (34)	50 (41.9)	134 (40)
Chicago	116 (13.4)	57 (17.5)	103 (14.2)	35 (14.6)	163 (14.1)	31 (12.6)	117 (15.6)
Miami	103 (14.2)	48 (16.6)	64 (11.2)	29 (15.1)	143 (14.7)	18 (7.1)*	83 (14.4)
San Diego	193 (41.2)	51 (30.4)	110 (34.6)	28 (18.5)**	230 (37.3)	44 (38.4)	108 (30)
Parental education, n (%)
< High school	175 (28.3)	69 (36.3)	180 (47.2)**	90 (58.2)**	230 (29.2)	58 (46.0)*	226 (52.2)**
High school or equivalent	152 (31.8)	73 (30.6)	118 (28.6)	30 (18.9)*	221 (32.4)	37 (22.7)	115 (26.1)
> High school	227 (39.9)	78 (33.1)	95 (24.2)**	38 (23.0)**	289 (38.4)	48 (31.3)	101 (21.7)**
US-born parent/caregiver, n (%)c	103 (17.6)	50 (24.9)	64 (20.5)	43 (31.6)*	146 (19.2)	48 (40.4)**	66 (17.6)
Food assistance in previous year, n (%)d	417 (76.6)	188 (87.7)*	346 (87.8)**	144 (88.3)**	577 (79.3)	123 (84.3)	395 (89.0)**
Poverty-income ratioa,e	1.3 (0.8–2.3)	0.8 (0.5–1.3)**	0.8 (0.6–1.2)**	0.6 (0.3–0.9)**	1.2 (0.6–1.8)	1.0 (0.7–1.4)*	0.8 (0.5–1.2)**

The SI conversion factors from mg/dL to mmol/L are as follows: multiply FG values by 0.0555, TG values by 0.0113, and HDL-C values by 0.0259 Values are means or percentages unless otherwise specified. All analyses were weighted for survey design, and. sample sizes are unweighted.

^aMedian and interquartile ranges reported for variables with nonnormal distributions. Mann-Whitney U (Wilcoxon rank sum) was used to test median differences.

^bMetS score for meeting each of the following criteria: abdominal adiposity (WC ≥ 90th age- and sex-specific percentiles), FG (≥ 100 mg/dL), TGs (≥ 150 mg/dL), HDL-C (≤ 40 mg/dL), SBP or DBP (≥ 90th age-, sex-, and height-specific percentiles).

^cUS-born includes Puerto Rico.

^dReceipt of food assistance in the previous 12 months includes SNAP, WIC, or National School Lunch Program for free or reduced school meals.

^eWe calculated poverty–income ratio by first determining the median income for a given reported income range and dividing by the state- and year-specific (based on recruitment year of the study) federal poverty guidelines specific to family size and issued by the US Department of Health and Human Services.

^*P < .05 and

^**P < .01 for comparisons between the indicated FS category and the high FS category of each FS measure.

Table 2 presents adjusted differences of MetS-relevant CMMs and MetS scores by household and child FS measures. For both FS measures, Hispanic/Latino youth in the lowest versus high FS category had significantly lower HDL-C (household: −3.17, 95% CI: −5.65 to −0.70; child: −1.81, 95% CI: −3.54 to −0.09). However, only the child FS measure was associated with other MetS-relevant CMMs. Compared with youth with high child FS, individuals with low/very low child FS had greater FG (1.37, 95% CI: 0.08 to 2.65), TGs (8.68, 95% CI: 1.75 to 15.61), and MetS expected log counts (2.12, 95% CI: 0.02 to 0.45).

TABLE 2.

Adjusted Back-Transformed Geometric Mean MetS CMM Differences and MetS Expected Log Count Differences by Household and Child FS Status in the SOL Youth (n = 1325)

Cardiometabolic Markers	Household Food Security Status			Child Food Security Status
	Marginal	Low	Very Low	Marginal	Low/Very Low
	(n = 220)	(n = 393)	(n = 158)	(n = 143)	(n = 442)
WC (cm)a	−2.28 (−4.89 to 0.32)	−0.59 (−2.57 to 1.38)	1.49 (−1.02 to 4.01)	−1.24 (−3.67 to 1.20)	0.63 (−1.12 to 2.37)
FG (mg/dL)a	0.47 (−1.05 to 1.99)	0.81 (−0.69 to 2.32)	1.22 (−0.42 to 2.85)	−0.13 (−2.17 to 1.91)	1.37 (0.08 to 2.65)
TGs (mg/dL)a	1.85 (−9.82 to 13.51)	6.35 (−0.99 to 13.68)	10.26 (−0.15 to 20.67)	2.10 (−8.60 to 12.81)	8.68 (1.75 to 15.61)
HDL-C (mg/dL)a	−0.93 (−3.16 to 1.31)	−0.93 (−3.02 to 1.16)	−3.17 (−5.65 to −0.70)	−0.63 (−3.07 to 1.80)	−1.81 (−3.54 to −0.09)
SBP (mm Hg)a	−1.06 (−2.78 to 0.67)	0.29 (−1.52 to 2.11)	2.06 (−0.03 to 4.16)	0.63 (−1.65 to 2.90)	0.92 (−0.74 to 2.58)
DBP (mm Hg)a	0.46 (−0.90 to 1.83)	0.42 (−0.92 to 1.76)	1.13 (−0.58 to 2.84)	1.23 (−0.73 to 3.19)	0.32 (−0.90 to 1.53)
MetS risk count (score)b	−0.42 (−0.47 to 0.30)	0.07 (−0.22 to 0.24)	1.88 (−0.01 to 0.59)	−0.35 (−0.33 to 0.23)	2.12 (0.02 to 0.45)

The SI conversion factors from mg/dL to mmol/L are as follows: multiply FG values by 0.0555, TG values by 0.0113, and HDL-C values by 0.0259. Estimates are back-transformed geometric mean differences (95% CIs) and expected log count differences (for MetS score) comparing lowest to high FS status categories by household and child measures. Data are from interviewer-administered questionnaires and clinical examinations. All models were weighted for survey design and adjusted for age (years), sex (male or female), site (Bronx, NY, Chicago, IL, San Diego, CA, and Miami, FL), parental place of birth (US-born, including Puerto Rico and foreign-born), parental education (less than high school, high school or equivalent, and more than high school), and poverty–income ratio (continuous).

^aGeometric mean estimates were back-transformed to their original scales and modeled by using linear regression.

^bMetS score for meeting the following criteria: abdominal adiposity (WC ≥ 90th age- and sex-specific percentiles), FG (≥ 100 mg/dL), TGs (≥ 150 mg/dL), HDL-C (≤ 40 mg/dL), SBP or DBP (≥ 90^th age-, sex-, and height-specific percentiles); modeled using Poisson regression.

We found statistically significant interactions between each of the 2 FS measures and parental place of birth for TGs only (P interactions: household = 0.05 and child = 0.008). To ease interpretation, Fig 1 shows multivariable TG mean differences by household and child FS stratified by parental place of birth (see Supplemental Table 3 for full findings). Among youth with foreign-born parents or caregivers, being in the lowest FS category was associated with significantly greater levels of TGs (household FS: 15.56, 95% CI: 1.63 to 29.48, child FS: 10.83 95% CI: 2.55 to 19.11) compared with the highest FS category. Findings were null among youth with US-born parents or caregivers. There were also no significant interactions for the remaining MetS-relevant CMMs.

FIGURE 1.

Adjusted-mean differences in TGs by (A) household and (B) child FS stratified by parental place of birth in the SOL Youth (n = 1325). Estimates are back-transformed adjusted-geometric-mean TG differences (95% CIs) from linear regression models comparing lower versus high household and child FS measures stratified by parental place of birth (US-born, including Puerto Rico and foreign-born). CIs not including zero indicate significant pairwise comparisons between lower FS and high FS (referent; not shown) at P < .05. Data are from interviewer-administered questionnaires and clinical examinations. Household FS: high (n = 554), marginal (n = 220), low (n = 393), very low (n = 158), child FS: high (n = 740), marginal (n = 143), low/very low (n = 442). All models were weighted for survey design and adjusted for participant’s age (years), sex (male or female), site (Bronx, NY, Chicago, IL, San Diego, CA, and Miami, FL), parental education (less than high school, high school or equivalent, and more than high school), parental place of birth (US-born, including Puerto Rico and foreign-born), poverty–income ratio (continuous), and receipt of any food assistance in the previous year (yes or no). P interactions: household FS = 0.06 and child FS = 0.010.

We also found statistically significant interactions between each of the 2 FS measures and receipt of any food assistance in the previous year in models of TG (P interactions: household FS = 0.03 and child FS = 0.005) and HDL-C (P interactions: household FS = 0.01 and child FS = 0.04) (Fig 2; see Supplemental Table 4 for full findings). Only among participants whose families did not receive any food assistance in the previous year, being in the lowest versus high FS categories was associated with significantly greater TGs (household FS: 36.12, 95% CI: 14.23 to 58.01; child FS: 18.79, 95% CI: 5.40 to 32.17), and significantly lower HDL-C (household FS: −10.17, 95% CI: −16.64 to −3.71; child FS: −6.99, 95% CI: −12.19 to −1.79). We also found statistically significant interactions for the child FS measure (but not household FS) in models of WC (P interaction = 0.008) and MetS score (P interaction = 0.009). Compared with youth with high child FS, those in the lowest child FS category had significantly higher WC (4.92, 95% CI: 0.15 to 9.69) and a higher MetS expected log count (0.75, 95% CI: 0.31 to 1.19), but only among youth whose families did not receive any food assistance in the previous year.

FIGURE 2.

Adjusted-mean differences in MetS-relevant CMMs by (A) household and (B) child FS stratified by receipt of any food assistance in the previous year in the SOLYouth (n = 1325). Estimates are back-transformed adjusted geometric mean unit differences (95% CIs) comparing lower-versus-high FS by household and child FS measures stratified by receipt of any food assistance (SNAP, WIC, or National School Lunch Program for free/reduced school meals) in the previous year. CIs not including zero indicate significant pairwise comparisons between lower FS and high FS (referent; not shown) at P < .05. Data are from interviewer-administered questionnaires and clinical examinations. Household FS: high (n = 554), marginal (n = 220), low (n = 393), very low (n = 158), child FS: high (n = 740), marginal (n = 143), low/very low (n = 442). All models were weighted for survey design and adjusted for participant’s age (years), sex (male or female), site (Bronx, NY, Chicago, IL, San Diego, CA, and Miami, FL), parental education (less than high school, high school or equivalent, and more than high school), parental place of birth (US-born, including Puerto Rico and foreign-born), and poverty–income ratio (continuous). TGs (in mg/dL), WC (in cm), and HDL-C (in mg/dL) were modeled by using linear regression whereas MetS score was modeled by using Poisson regression. MetS was calculated by summing the number of counts for each of the following criteria: abdominal adiposity (WC ≥ 90th age- and sex-specific percentiles), FG (≥ 100 mg/dL), TGs (≥ 150 mg/dL), HDL-C (≤ 40 mg/dL), SBP or DBP (≥ 90th age-, sex-, and height-specific percentiles). Household FS P interactions: TG = 0.04, HDL-C = 0.01, child FS P interactions: TG = 0.004, WC = 0.008, HDL-C = 0.03, MetS score = 0.009.

Discussion

Our study is among the first to document adverse associations between household and child FS measures with a MetS score variable and several MetS-relevant CMMs among US Hispanic/Latino youth. In addition, findings suggest that the health implications of FI may be greater among youth with foreign-born parents or caregivers and whose families did not receive any food assistance in the previous year.

Unlike the current study, 2 studies using nationally representative data among adolescents found no associations between FS measures and relevant CMMs (FG, TGs, and HDL-C).^16,25 One reason for these discrepancies may have to do with the populations under study. Although Hispanic/Latino youth (mostly of Mexican origin) were included in these studies, they were aggregated with other racial or ethnic groups, potentially attenuating associations. Moreover, interactions with race or ethnicity and other important sociodemographic characteristics were not evaluated in these studies.

Relatedly, our findings suggest greater health implications of FI among Hispanic/Latino youth with foreign-born parents or caregivers and among those who did not receive any food assistance in the past year. The differential patterns by parental place of birth, in particular, are consistent with previous work related to mental health in HCHS/SOL Youth.¹⁰ In these data, greater FI was associated with higher anxiety scores among youth with foreign-born parents or caregivers.¹⁰ Meanwhile, compared with those with US-born parents or caregivers, youth with foreign-born parents or caregivers have been shown to experience several acculturative (eg, adjusting to US norms or family dynamics) and economic (eg, low-income status or parental unemployment) chronic stressors (in addition to FI),^30,31 which contribute to poor mental and physical health outcomes. Another study found similar findings for self-reported health status among youth with foreign-born parents or caregivers, showing a lower probability of reporting very good health among children and adolescents of immigrants facing FI compared with their food-secure peers.³¹ The combination of these factors may amplify the adverse health consequences of FI for youth in these families. These chronic stressors may also undermine the health-protective role of sociocultural (eg, strong family and social support networks, optimism, and resilience) and behavioral (eg, healthy eating) factors^30,46 that may otherwise contribute to a health advantage among Hispanic/Latino youth with foreign-born parents/caregivers.

Youth with foreign-born parents or caregivers may also be living with members of the household who are undocumented and, therefore, ineligible for government food assistance (eg, SNAP), further worsening the impact of FI.^50,51 In our study, low levels of FS were adversely associated with MetS score, WC, TGs, and HDL-C to a greater extent for youth whose families did not receive any food assistance in the previous year than among those that did. However, even after adjusting for food assistance in interaction models with parental place of birth, the same adverse associations remained for youth with foreign-born parents or caregivers. Future research should more directly evaluate whether expansion and increased uptake of food assistance programs and other interventions among immigrant families can limit the adverse health consequences of FI in Hispanic/Latino youth.

Our study evaluated associations with 2 different FS measures and indicated that a child versus a household FS measure may better capture the role of FI in youth’s health. For instance, although household and child FS status detected associations with HDL-C, only the child FS measure also detected associations with FG, TGs, and the MetS score. This could be because a child measure may be more sensitive for detecting FI in this population compared with a broader household measure. In a context of low household-level FS, older family members may prioritize food resources for younger family members, protecting them against some of the negative health consequences.^2,37 In contrast, a child FS measure is more likely to capture food availability for these specific household members.

Taken together, these findings have potential implications for health care providers and nutrition-based policies. Although CMM differences by child FS status do not reflect disparities in clinical disease, this pattern may contribute to a disproportionate burden of cardiometabolic complications among the most food-insecure Hispanic/Latino youth as they transition to adulthood. Consequently, health care providers should consider early screening for FI using the child FS measure to identify youth who may benefit from additional resources. Relatedly, policies targeting Hispanic/Latino families that facilitate uptake into food assistance programs are needed. In our study, FI was adversely associated with CMMs particularly among youth from families that did not receive any food assistance. Improving access and expanding eligibility to such programs may help curtail the potential negative health consequences of FI in this population.

This study has some strengths and limitations. The diverse sample of US Hispanic/Latino youth in our study increases the generalizability of our findings to heritage groups beyond Hispanics/Latinos of Mexican origin, because most previous studies that included Hispanics/Latinos primarily focused on this population. Additionally, the large sample size allowed us to test effect modification to identify groups within this population that may be facing greater health implications from FI. Given the cross-sectional nature of our data, however, we could not infer causality, substantiating the need for longitudinal studies and better causal approaches to directly evaluate the health implications of low FS. Our findings may have also obscured important differences by country of origin among youth with foreign-born parents. Our power was limited to evaluate these differences, but future research should consider the role that these social and cultural factors may play in terms of the associations we reported. Lastly, we did not adjust for multiple comparisons, which could have inflated the likelihood of observing false positives (type I error). Most multiple comparison adjustment methods, however, assume independence among dependent variables.⁵² In our study, the dependent variables were biologically related. Employing an adjustment method that ignores these correlations may lead to overconservative adjustments, increasing the probability of false negative findings.⁵² Nevertheless, because findings were more consistent for the child-versus-household FS measure, the statistically significant estimate for household FS and HDL-C should be interpreted with caution.

Conclusions

Our results suggest that Hispanic/Latino youth with severe FI (low/very low FS) have worse cardiometabolic profiles compared with their food-secure counterparts. These findings argue for exploring interventions to address FI among Hispanic/Latino youth, a fast-growing segment of the US population at high risk of cardiometabolic complications. Given the increase in FI that resulted from the coronavirus disease 2019 pandemic, especially for Hispanic/Latino immigrant families,⁵³ these findings may also foreshadow concerning trends for the health and well-being of Hispanic/Latino youth.

Glossary

CI

confidence interval

CMMs

cardiometabolic markers

DBP

diastolic blood pressure

FG

fasting plasma glucose

FI

food insecurity

FS

food security

HCHS/SOL

Hispanic Community Health Survey/Study of Latinos

HDL-C

high-density lipoprotein cholesterol

IR

insulin resistance

MetS

metabolic syndrome

SBP

systolic blood pressure

SNAP

supplemental nutrition assistance program

SOL Youth

Hispanic Community Children’s Health Study/Study of Latino Youth

TGs

triglycerides

WC

waist circumference

WIC

Women, Infants, and Children