Sage Journals: Discover world-class research

Abstract

Objective:

Cardiometabolic disease is the leading cause of mortality in the United States. Cardiometabolic function during adolescence predicts future cardiometabolic disease, yet few studies have examined early determinants of cardiometabolic function. Informed by evidence of sex differences in the prevalence and severity of cardiometabolic disorders and evidence of sexual dimorphism in the stress response, we examined sex differences in the association between living in poverty and cardiometabolic function during adolescence, a precursor of later cardiometabolic disorders.

Methods:

We linked data from 10 415 adolescents aged 12-19 in the National Health and Nutrition Examination Survey (1999-2012) with US Census–tract data on area-level poverty (percentage of the population living in poverty, grouped into quartiles). We parameterized cardiometabolic dysfunction by summing the z scores of 6 cardiometabolic biomarkers, grouped into quintiles. Hierarchical ordinal models estimated associations.

Results:

Compared with residents in low-poverty areas, residents in high-poverty areas had elevated odds of cardiometabolic dysfunction (highest quartile of poverty odds ratio [OR] = 1.27; 95% confidence interval [CI], 1.08-1.50). This association was more pronounced among boys than girls (highest quartile of poverty for boys: OR = 1.36; 95% CI, 1.10-1.70; highest quartile of poverty for girls: OR = 1.17; 95% CI, 0.94-1.47).

Conclusion:

Our study supports the existence of sex-specific associations. These results highlight the potential for community-based programs, such as housing assistance, to improve population health.

Keywords

poverty adolescent health cardiometabolic risk factors socioeconomic status environment

Cardiometabolic disease, the leading cause of mortality in the United Sates,¹ has pronounced sex differences. Men have earlier onset of, higher prevalence of, and higher mortality from cardiovascular disease than women.¹ Sex differences in precursors of cardiometabolic disease, however, are underexamined.² This scarcity of research led the American Heart Association to call for an investigation of sex differences in precursors of cardiometabolic disease, with a focus on early experiences of adversity.²

Although the literature on physiologic sequalae of adverse childhood events provides compelling evidence of an association between exposure to adversity and cardiometabolic function,³ adverse childhood events focus on individual-level experiences (ie, sexual abuse, parental divorce)⁴ and not on contextual adversity. Residence in economically deprived areas, a well-recognized contextual source of adversity,² may be linked with cardiometabolic function in adolescence. Exposure to economic deprivation during childhood⁵ may be associated with the physiologic stress response, which in turn may be associated with increased cardiometabolic risk,⁶ but this hypothesis is under investigation.

To our knowledge, only 5 studies have examined the link between residence in economically deprived areas and cardiometabolic function in adolescence.^7

-11 These studies found that residents of economically deprived areas are more likely than residents of economically disadvantaged areas to have adiposity,^7
-9 high allostatic load (a construct related to cardiometabolic function),¹⁰ or cardiometabolic dysfunction.¹¹ Four of these studies examined differences by sex. Among participants aged 12-19 in the National Longitudinal Study of Adolescent Health (waves 2 and 3 [1996-2002], n = 7788)⁷ and among Canadian adolescents aged 11-16 in the Health Behavior in School-Aged Children Survey (2001-2002, n = 6684), the studies found no differences by sex.⁸ In contrast, among participants aged 2-18 in the National Longitudinal Survey of Youth (1986-2010, n = 11 504)⁹ and participants aged 12-20 in the National Health and Nutrition Examination Survey (NHANES; 1999-2006, n = 11 886),¹⁰ the association between neighborhood-level deprivation and obesity⁹ and allostatic load¹⁰ was higher among girls than boys. Both studies with positive findings had large samples and measured their outcome variables according to objective criteria. In contrast, both studies with negative findings^7,8 had self-reported outcomes. Random measurement error likely contributed to these null findings.

Another reason to suspect that the association between adversity and cardiometabolic function is sex specific is that the human stress response is sexually dimorphic. Important sex differences exist in the functioning of the brain’s stress circuitry (ie, the hypothalamic-pituitary-adrenal [HPA] axis).^12,13 This dimorphic HPA-regulated stress response is primarily mediated by changes in hormones released during critical periods of development.¹⁴ Sex differences in placental and embryonic responses to stressors are evident during gestation and become more pronounced during adolescence.^12
-14 Connectivity and functionality of the HPA axis is modified by a rise in gonadal hormones during adolescence.¹⁴ For example, the growth rate of certain regions of the adrenal cortex diverges among boys and girls: in 1 study, the growth rate significantly decreased among boys, and boys also had a diminished HPA stress response compared with girls.¹³ These sex-specific changes are predominantly mediated by increased testosterone production during adolescence.¹² Despite the relevance of sex in the stress response, only 4 studies have investigated sex differences in early precursors of cardiometabolic disease.^7

-10

We examined sex differences in the association between residence in economically deprived areas and cardiometabolic dysfunction among US adolescents to better understand cardiometabolic consequences of sexually dimorphic stress responses in the face of contextual-level adversity. Cardiometabolic dysfunction during adolescence is a marker for onset of cardiometabolic disease during adulthood.^15,16 Although adolescents rarely meet diagnostic criteria for cardiometabolic disease, those with cardiometabolic functioning deviating from population-level norms are at elevated risk of developing cardiometabolic disease during adulthood.^16
-18 We hypothesized that the cardiometabolic consequences of residence in an economically deprived area differ by sex.

Methods

Sample

We examined data from the 1999-2012 NHANES, a cross-sectional survey that provides vital and health statistics from a nationally representative sample of the noninstitutionalized US population conducted continuously in 2-year cycles.¹⁹ We selected the analytic sample from 13 343 adolescents aged 12-19. We excluded respondents who reported current pregnancy (n = 181) or a diagnosis of hypertension or diabetes or using medication for hypertension or diabetes (n = 83). We excluded participants with missing data on any of the cardiometabolic function variables (n = 2664), leaving an analytic sample of 10 415 adolescents aged 12-19.

We drew area-level data from the 2000 US decennial census²⁰ and 5-year estimates (2005-2009; 2009-2013) from the American Community Survey.²¹ We linked individual- and area-level data with contemporary census tract identifiers.

Cardiometabolic Dysfunction Score

As detailed elsewhere,¹¹ we created an index of cardiometabolic dysfunction because composite measures of cardiometabolic dysfunction during adolescence predict later cardiometabolic health better than any individual biomarker alone.¹⁷ Adolescents with cardiometabolic functioning deviating from population-level norms are at elevated risk of developing cardiometabolic disease during adulthood.^16,17 Our choice of biomarkers was informed by the literature on their predictive utility and the degree of missingness for the biomarkers available in NHANES.^15,22,23 We selected 6 biomarkers, all of which were obtained during the mobile examination center component of NHANES¹⁹: glycosylated hemoglobin A_1c (HbA_1c; 3-month average blood glucose level, %), mean systolic and diastolic blood pressure (mm Hg) when ≥2 measurements were available (n = 10 266; 98%), high-density lipoprotein (HDL) cholesterol (mg/dL), total cholesterol (mg/dL), and waist circumference (cm). We chose total cholesterol as a measure of lipid metabolism because it tracks over time and allows for sufficient sample size compared with other measures of lipid metabolism (ie, triglycerides), which were available for only a fasting subsample of NHANES participants.²² We chose HbA_1c as a measure of glucose metabolism because it tracks well over time and is less influenced than fasting glucose by recent events.²³

Sex-specific physiologic differences during adolescence contribute to sex-specific cardiometabolic profiles. For example, compared with boys, girls typically enter adolescence earlier, achieve their adult height at an earlier age, and develop a greater percentage of adipose tissue,²⁴ which is associated with lower HDL cholesterol levels among boys than among girls.²⁵ Compared with girls, boys have greater growth velocity and develop more lean muscle, a more dense skeleton, and greater blood mass, which is related to higher blood pressure levels.²⁴ Thus, we standardized cardiometabolic biomarkers by age and sex, except for blood pressure, which we standardized by age, sex, and height, consistent with guidelines for measuring hypertension in adolescents.²⁶ We summed z scores for each biomarker to create an overall cardiometabolic dysfunction index. This method of calculating the cardiometabolic dysfunction index can result in negative numbers, because the 2 blood pressure variables are standardized on an external reference category.²⁶ Higher cardiometabolic dysfunction scores indicate greater dysfunction (HDL was multiplied by –1 for directionality). We parametrized quintiles of cardiometabolic dysfunction, with the lowest quintile as the reference group.

This index is not a validated clinical predictor of disease and was designed to promote a better understanding of preclinical levels of cardiometabolic dysfunction that may increase the risk for future cardiometabolic disease. The biomarkers included in the index are informed by biomarkers used to diagnose metabolic syndrome.¹⁷ Each biomarker included in the index is predictive of future disease, and adolescents with metabolic syndrome are at increased risk for cardiometabolic disease in adulthood.^15,27

-30

Area-Level Poverty

We used the proportion of residents in a census tract living below the federal poverty level as a measure of area-level poverty. Area-level poverty is consistently measured over time, easily interpretable, and relates well to other area-level measures of socioeconomic status (eg, housing quality, crowded living spaces).³¹ We measured area-level poverty at the census-tract level, because census tracts are considered more economically homogenous and more socially relevant and produce more consistent observations than other small geographies.³¹ We parameterized area-level poverty into quartiles (coded from 1-4, with 1 = low poverty).

Covariates included a 2-year survey cycle, race/ethnicity (non-Hispanic white, non-Hispanic black, Mexican American, and other), family poverty-to-income ratio (ie, ratio of family income to federal poverty level based on family size; continuous), head of household education level (<high school diploma, high school diploma, >high school diploma), serum cotinine levels (continuous), physical activity (with responses to the question, “Do you do any vigorous-intensity sports, fitness, or recreational activities that cause large increases in breathing or heart rate, like running or basketball, for at least 10 minutes continuously?” coded as 1 for “yes” and 0 for “no”), and prevalence of non-Hispanic black population at the census-tract level (continuous). We did not include age and sex in the models because cardiometabolic dysfunction was standardized by age and sex.

Statistical Analysis

We obtained descriptive statistics by using survey commands accounting for complex survey design. We did not transform HDL cholesterol levels for directionality for descriptive statistics, so that higher levels of HDL cholesterol indicate better cardiometabolic function. Second, we estimated the association between area-level poverty and cardiometabolic dysfunction with a series of hierarchical ordinal regression models. We made this choice for ease of interpretation of results. Progression from one unit to another on a continuous scale reflects a small change in the risk of cardiometabolic dysfunction; in contrast, progression from one category to the next category arguably corresponds to a more biologically meaningful change in risk, because each ordinal category includes multiple units in a continuous scale.

Moreover, an ordinal variable may better predict the risk of later development of disease than a dichotomous variable.¹⁷ Each level of cardiometabolic dysfunction serves as a threshold sequentially, yielding a series of binary estimates in which the odds of being in categories at or above that threshold is compared with the odds of being in categories below that threshold (Box, equations 1-3). The cumulative odds ratio (OR) reflects the combined odds of worse cardiometabolic dysfunction across all possible thresholds, interpreted as the odds of being in any category of worse cardiometabolic dysfunction (Box, equation 4). We obtained ORs and 95% confidence intervals (CIs) from hierarchical ordinal regression models fit by using PROC GLIMMIX in SAS version 9.3,³² using the lowest cardiometabolic dysfunction category as the reference category.

Box

Ordinal logistic regression model specification for the association between area-level poverty and cardiometabolic dysfunction among adolescents aged 12-19, United States, 1999-2012

Equations 1-3

L_{1} = log (\frac{π_{1}}{π_{2} + π_{3} + \dots + π_{r}})

L_{2} = log (\frac{π_{1} + π_{2}}{π_{3} + π_{4} + \dots + π_{r}})

L_{n} = log (\frac{π_{1} + π_{2} + π_{n}}{π_{n + 1} + π_{n + 2} + \dots + π_{r}})

Equation 4

P (Y \leq j) = π_{1} + \dots + π_{J}

log (\frac{P (Y \leq j)}{P (Y > j)}) = log (\frac{π_{1} + \dots + π_{j}}{π_{j + 1} + \dots + π_{j}})

Components

π₁: Probability of being in category 1 of cardiometabolic dysfunction

π₂: Probability of being in category 2 of cardiometabolic dysfunction

π₃: Probability of being in category 3 of cardiometabolic dysfunction

π₄: Probability of being in category 4 of cardiometabolic dysfunction

π_r: Probability of being in a higher-ordered category of cardiometabolic dysfunction

π_j: Threshold of interest (cardiometabolic dysfunction category 1)

We conducted overall and sex-specific analyses given previous evidence suggesting sex-specific physiology during adolescence^12,14 and that the experience of residing in a high-poverty area may differ by sex. First, we fit a model that included area-level poverty and indicator variables for survey cycle. Second, we included race/ethnicity, family poverty-to-income ratio, head of household education level, serum cotinine level, and physical activity. Third, we added the prevalence of the non-Hispanic black population to the model.

We treated responses of “don’t know,” “refused,” and ”missing” as missing. We imputed missing values for family poverty-to-income ratio, serum cotinine level, physical activity, and head of household education by using SAS PROC MI in SAS version 9.3.³² Because NHANES uses a complex, stratified, multistage sample design, all analyses used mobile examination center survey weights. We analyzed restricted-use NHANES data in the Research Data Center at the National Center for Health Statistics. The National Center for Health Statistics reviewed all data, and the University of Maryland Institutional Review Board considered this project exempt from review.

Results

We found a linear association between area-level poverty and cardiometabolic dysfunction (Table 1). The cardiometabolic dysfunction score was on average 20% higher among residents of the highest-poverty areas (ie, quartile 4) than among residents of the lowest-poverty areas (ie, quartile 1; mean scores: –0.69; 95% CI, –0.77 to –0.61 [quartile 4] vs –0.86; 95% CI, –0.94 to –0.79 [quartile 1]).

Table 1.

Weighted mean cardiometabolic function scores and cardiometabolic biomarkers, by demographic characteristics and area-level variables, National Health and Nutrition Examination Survey, 1999-2012^a

Variable	No. (%)	Cardiometabolic Dysfunction Index,^b Mean (95% CI)	Systolic Blood Pressure, mm Hg,^b Mean (95% CI)	Diastolic Blood Pressure, mm Hg,^b Mean (95% CI)
Total	10 415 (100.0)	–0.79 (–0.83 to 0.75)	109.62 (109.34 to 109.90)	60.72 (60.41 to 61.04)
Sex
Female	5015 (48.2)	–0.63 (–0.69 to –0.57)	106.83 (106.48 to 107.18)	61.74 (61.32 to 62.16)
Male	5400 (51.8)	–0.94 (–0.99 to 0.88)	112.23 (111.82 to 112.64)	59.78 (59.31 to 60.24)
Family poverty-to-income ratio
<1.0	3074 (29.5)	–0.72 (–0.79 to –0.65)	109.99 (109.49 to 110.49)	60.72 (60.12 to 61.32)
1.0-2.9	4005 (38.5)	–0.76 (–0.83 to –0.70)	109.71 (109.27 to 110.15)	60.53 (60.02 to 61.05)
3.0-4.9	1606 (15.4)	–0.82 (–0.90 to –0.73)	109.61 (108.95 to 110.26)	60.85 (60.15 to 61.56)
≥5.0	967 (9.3)	–0.90 (–1.02 to –0.78)	108.92 (108.07 to 109.78)	61.12 (60.21 to 62.03)
Missing	763 (7.3)	–0.79 (–0.97 to –0.62)	109.65 (108.52 to 110.77)	60.39 (58.90 to 61.88)
Age, y
12-14	3981 (38.2)	–0.77 (–0.83 to –0.70)	106.86 (106.45 to 107.28)	51.98 (57.44 to 58.52)
15-17	3964 (38.1)	–0.75 (–0.81 to –0.69)	110.49 (110.06 to 110.93)	61.65 (61.17 to 62.12)
18-19	2470 (23.7)	–0.89 (–0.98 to –0.80)	112.60 (111.99 to 113.21)	63.60 (62.97 to 64.22)
Race/ethnicity^c
Non-Hispanic white	2756 (26.5)	–0.81 (–0.87 to –0.75)	109.42 (109.01 to 109.83)	61.27 (60.80 to 61.74)
Non-Hispanic black	3052 (29.3)	–0.61 (–0.67 to –0.55)	111.83 (111.40 to 112.25)	60.71 (60.21 to 61.21)
Mexican American	3342 (32.1)	–0.85 (–0.92 to –0.79)	109.34 (108.90 to 109.79)	58.33 (57.79 to 58.87)
Area-level poverty^d
Quartile 1	2599 (25.0)	–0.86 (–0.94 to –0.79)	109.06 (108.55 to 109.58)	60.68 (60.10 to 61.26)
Quartile 2	2615 (25.1)	–0.82 (–0.89 to –0.75)	109.57 (109.04 to 110.10)	60.70 (60.11 to 61.29)
Quartile 3	2584 (24.8)	–0.69 (–0.78 to –0.60)	109.98 (109.39 to 110.58)	60.86 (60.16 to 61.55)
Quartile 4	2595 (24.9)	–0.69 (–0.77 to –0.61)	110.56 (110.00 to 111.12)	60.69 (60.04 to 61.33)
Prevalence of non-Hispanic black population at the census-tract level^e
Quartile 1	2599 (25.0)	–0.79 (–0.87 to –0.72)	109.14 (108.59 to 109.68)	61.30 (60.71 to 61.88)
Quartile 2	2604 (25.0)	–0.78 (–0.86 to –0.70)	109.59 (109.04 to 110.15)	60.88 (60.23 to 61.53)
Quartile 3	2605 (25.0)	–0.83 (–0.91 to –0.76)	109.65 (109.13 to 110.16)	60.06 (59.50 to 60.62)
Quartile 4	2599 (25.0)	–0.72 (–0.80 to –0.64)	110.73 (110.22 to 111.24)	60.22 (59.93 to 60.91)

^a Data source: National Center for Health Statistics.¹⁹

^b Higher levels of this variable should be interpreted as an increase in cardiometabolic dysfunction.

^c Not all racial/ethnic groups are reported in the table as advised by the National Health and Nutrition Examination Survey.

^d Quartiles based on population-based distribution of area-level poverty. Each quartile represents 25% of the sample. Quartile 1 = low levels of poverty. Twenty-two had missing values, which were not reported because of small size.

^e Quartiles based on population-based distribution of prevalence of non-Hispanic black population. Each quartile represents 25% of the sample. Quartile 1 = low prevalence of non-Hispanic black population.

We observed differences in mean cardiometabolic dysfunction by sex. Boys had lower mean cardiometabolic dysfunction scores (mean: –0.942; 95% CI, 0.998 to –0.887) than girls (mean: –0.634; 95% CI, –0.691 to –0.577). Boys also had higher mean systolic blood pressure, waist circumference, and HbA_1c than girls, whereas girls had higher mean diastolic blood pressure, total cholesterol, and HDL cholesterol than boys (Tables 1 and 2).

Table 2.

Weighted mean cardiometabolic function scores and cardiometabolic biomarkers, by demographic characteristics and area-level variables, National Health and Nutrition Examination Survey, 1999-2012^a

Variable	No. (%)	Glycosylated Hemoglobin (HbA_1c), %,^b Mean (95% CI)	Waist Circumference, cm,^b Mean (95% CI)	Total Cholesterol, mg/dL,^b Mean (95% CI)	High-Density Lipoprotein Cholesterol, mg/dL,^c Mean (95% CI)
Total	10 415 (100.0)	5.15 (5.15-5.16)	81.42 (81.04-81.81)	159.83 (159.00-160.65)	50.69 (50.37-51.01)
Sex
Female	5015 (48.2)	5.14 (5.13-5.15)	80.81 (80.28-81.35)	162.54 (161.36-163.72)	53.14 (52.67-53.62)
Male	5400 (51.9)	5.17 (5.16-5.18)	81.99 (81.45-82.54)	157.29 (156.15-158.43)	48.39 (47.97-48.82)
Family poverty-to-income ratio^d
<1.0	3074 (29.5)	5.18 (5.16-5.19)	82.84 (82.08-83.16)	159.58 (158.03-161.12)	50.37 (49.75-50.99)
1.0-2.9	4005 (38.5)	5.15 (5.14-5.17)	81.86 (81.22-82.05)	160.26 (158.92-161.60)	50.19 (49.68-50.70)
3.0-4.9	1606 (15.4)	5.15 (5.13-5.17)	80.54 (79.66-81.41)	158.71 (156.77-160.65)	50.54 (49.77-51.30)
≥5.0	967 (9.3)	5.12 (5.10-5.14)	79.62 (78.62-80.63)	160.49 (158.08-162.90)	52.12 (51.20-53.04)
Missing	763 (7.3)	5.18 (5.15-5.21)	81.65 (80.18-83.12)	160.41 (157.59-163.24)	51.62 (50.42-52.82)
Age, y
12-14	3981 (38.2)	5.18 (5.17-5.19)	77.65 (77.07-78.24)	158.59 (157.33-159.85)	51.70 (51.18-52.22)
15-17	3964 (38.1)	5.15 (5.13-5.16)	82.35 (81.75-82.95)	158.03 (156.68-159.38)	49.99 (49.48-50.51)
18-19	2470 (23.7)	5.13 (5.11-5.14)	85.94 (85.13-86.75)	164.82 (163.03-166.61)	50.21 (49.54-50.89)
Race/ethnicity^e
Non-Hispanic white	2756 (26.5)	5.12 (5.10-5.13)	81.54 (80.97-82.12)	160.17 (158.93-161.42)	49.82 (49.34-50.30)
Non-Hispanic black	3052 (29.3)	5.26 (5.25-5.28)	80.66 (80.03-81.29)	160.26 (159.09-161.42)	54.25 (53.75-54.75)
Mexican American	3342 (32.1)	5.17 (5.16-5.19)	83.44 (82.83-84.06)	158.78 (157.60-159.96)	50.36 (49.83-50.89)
Area-level poverty^f
Quartile 1	2599 (25.0)	5.13 (5.11-5.14)	79.97 (79.31-80.62)	160.14 (158.61-161.66)	50.81 (50.21-51.41)
Quartile 2	2615 (25.1)	5.14 (5.13-5.16)	81.57 (80.85-82.29)	159.76 (158.19-161.33)	50.38 (49.78-50.98)
Quartile 3	2584 (24.8)	5.18 (5.16-5.19)	82.72 (81.82-83.61)	159.63 (157.85-161.40)	50.62 (49.94-51.29)
Quartile 4	2595 (24.9)	5.20 (5.19-5.22)	82.83 (82.00-83.66)	159.40 (157.87-160.94)	51.05 (50.38-51.72)
Prevalence of non-Hispanic black population at the census-tract level^g
Quartile 1	2599 (25.0)	5.14 (5.12-5.15)	81.54 (80.86-82.23)	159.91 (158.38-161.44)	50.08 (49.49-50.66)
Quartile 2	2604 (25.0)	5.14 (5.12-5.15)	81.54 (80.74-82.34)	159.34 (157.65-161.03)	50.48 (49.84-51.12)
Quartile 3	2605 (25.0)	5.16 (5.14-5.17)	81.44 (80.71-82.17)	159.92 (158.36-161.48)	50.93 (50.28-51.59)
Quartile 4	2599 (25.0)	5.21 (5.20-5.23)	80.89 (80.11-81.67)	160.36 (158.80-161.93)	52.02 (51.38-52.66)

^a Data source: National Center for Health Statistics.¹⁹

^b Higher levels of this variable should be interpreted as an increase in cardiometabolic dysfunction.

^c Higher levels of this variable should be interpreted as a decrease in cardiometabolic dysfunction.

^d Ratio of family income to federal poverty level based on family size. Higher value indicates higher income.

^e Not all racial/ethnic groups are reported in the table as advised by the National Health and Nutrition Examination Survey.

^f Quartiles based on population-based distribution of area-level poverty. Each quartile represents 25% of the sample. Quartile 1 = low level of poverty.

^g Quartiles based on population-based distribution of prevalence of non-Hispanic black population. Each quartile represents 25% of the sample. Quartile 1 = low prevalence of non-Hispanic black population.

Overall, residents of highest-poverty areas had 27% higher odds of cardiometabolic dysfunction (quartile 4: OR = 1.27; 95% CI, 1.08-1.50) than residents of lowest-poverty areas, independent of individual- and area-level covariates (Table 3). In the adjusted sex-specific model (Table 3, model 3), boys residing in the highest-poverty areas had 36% higher odds of cardiometabolic dysfunction (quartile 4: OR = 1.36; 95% CI, 1.10-1.70) than boys residing in the lowest-poverty areas, and girls residing in the highest-poverty areas had 17% higher odds of cardiometabolic dysfunction than girls residing in the lowest-poverty areas (quartile 4: OR = 1.17; 95% CI, 0.94-1.47), but the difference was not significant (Table 3).

Table 3.

Hierarchical models estimating the association between area-level socioeconomic status and cardiometabolic function, overall and by sex, National Health and Nutrition Examination Survey (NHANES), 1999-2012^a

Quartile^b	Overall, OR (95% CI), n = 10 415	Boys, OR (95% CI), n = 5400	Girls, OR (95% CI), n = 5015
Model 1 area-level poverty^c
Quartile 1	1 [Reference]	1 [Reference]	1 [Reference]
Quartile 2	1.05 (0.92-1.20)	1.08 (0.91-1.28)	1.02 (0.84-1.22)
Quartile 3	1.27 (1.11-1.45)	1.28 (1.07-1.52)	1.25 (1.04-1.51)
Quartile 4	1.25 (1.25-1.43)	1.27 (1.07-1.51)	1.22 (1.01-1.46)
Model 2 area-level poverty^d
Quartile 1	1 [Reference]	1 [Reference]	1 [Reference]
Quartile 2	1.06 (0.92-1.22)	1.11 (0.92-1.33)	1.01 (0.83-1.22)
Quartile 3	1.30 (1.12-1.51)	1.36 (1.11-1.66)	1.23 (1.00-1.51)
Quartile 4	1.24 (1.06-1.45)	1.32 (1.06-1.63)	1.17 (0.94-1.46)
Model 3 area-level poverty^e
Quartile 1	1 [Reference]	1 [Reference]	1 [Reference]
Quartile 2	1.06 (0.92-1.22)	1.12 (0.94-1.35)	1.01 (0.83-1.22)
Quartile 3	1.32 (1.13-1.53)	1.39 (1.14-1.71)	1.24 (1.01-1.52)
Quartile 4	1.27 (1.08-1.50)	1.36 (1.10-1.70)	1.17 (0.94-1.47)

Abbreviation: OR, odds ratio.

^a Data source: National Center for Health Statistics.¹⁹

^b Quartile 1 is a reference category and represents low area-level poverty.

^c Model 1 includes NHANES survey cycle.

^d Model 2 includes head of household education, race/ethnicity, family poverty-to-income ratio, physical activity, serum cotinine level, and NHANES survey cycle.

^e Model 3 includes all variables in model 2 plus prevalence of non-Hispanic black population at the census-tract level.

The inclusion of individual- and area-level covariates differentially influenced the association between area-level poverty and cardiometabolic dysfunction by sex (Table 3). In the unadjusted sex-specific analysis (model 1), the association between residence in high-poverty areas (quartiles 3 and 4) and cardiometabolic dysfunction was similar among boys and girls. Among girls, the inclusion of individual-level covariates reduced the ORs for all quartiles of area-level poverty by up to 5% (ie, model 1 quartile 4, OR = 1.22 [95% CI, 1.01 -1.46] vs model 2 quartile 4, OR = 1.17 [95% CI, 0.94-1.46]), and the OR for quartile 4 became nonsignificant. In contrast, among boys, with the inclusion of individual-level covariates, the ORs for all quartiles of area-level poverty were slightly higher (ie, model 1 quartile 4, OR = 1.27 [95% CI, 1.07 -1.51] vs model 2 quartile 4, OR = 1.32 [95% CI, 1.06-1.63]), and the ORs for quartiles 3 and 4 were significant. The inclusion of percentage of non-Hispanic black residents (model 3) resulted in no change among girls (model 3 quartile 4, OR = 1.17 [95% CI, 0.94-1.47]), yet the ORs for quartiles 3 and 4 were higher by up to 3% among boys (model 3 quartile 4, OR = 1.36 [95% CI, 1.10-1.70]).

Discussion

The greater cardiometabolic dysfunction among girls than among boys in our study aligns with evidence of a higher prevalence of metabolic syndrome among girls than among boys in the literature.³³ Also in line with other studies,^7

-11 we observed a positive association between residence in economically deprived areas and cardiometabolic dysfunction. Although girls had greater cardiometabolic dysfunction than boys, the association between residence in high-poverty areas and cardiometabolic dysfunction was more pronounced among boys. Our findings also align with clinical evidence that men are more prone to develop stress-induced metabolic syndrome than women.³⁴ Although sexual dimorphism in the stress response generally presents as heightened vulnerability to stress-induced mood and autoimmune disorders among females,³⁵ males exposed to stress, particularly chronic stress, are more susceptible to cardiometabolic effects of stress, including visceral adiposity, hyperglycemia, and hyperinsulinemia, than females.³⁶

These findings are noteworthy for at least 2 reasons. First, our findings advance our understanding of the link between exposure to contextual adversity and cardiometabolic function during adolescence by examining sex differences in this association for the first time. Second, our findings contribute to the sparse literature of population-based studies on precursors of cardiometabolic disease. Of the 5 extant population-based studies in this area, 4 examined sex differences, and 2 reported sex differences in the association between stress and cardiometabolic function.^9,10 Aligned with the 2 positive studies,^9,10 our study benefited from a large sample of adolescents with objectively measured cardiometabolic variables rather than self-reported outcomes. Our findings support the notion that low statistical power and random measurement error likely contributed to the null findings^7,8 and tip the balance toward the existence of sex-specific associations; however, this finding is not yet conclusive, and more research is warranted.

Boys and girls residing in economically deprived areas are subjected to distinct social experiences and influences.^37
-39 For example, boys generally have more unsupervised time than girls^37
-39 and, thus, may be more exposed to stressors in these areas. Indeed, the association between parental supervision and unhealthy behaviors, including risky sexual behaviors, smoking, and violence, is stronger among boys than among girls in such areas.^38,39 In addition, preliminary evidence suggests that gender role socialization (whereby boys and girls are socialized to be seen as or act more masculine or feminine based on their biological sex) may influence sex differences in the stress response by influencing the perception of a situation as threatening or by influencing the ameliorating effect of social support.⁴⁰

Limitations

This study had several limitations. First, the associations observed in this study are likely to be explained, in part, by societal processes, such as disinvestments in areas of concentrated disadvantage,⁴¹ and related sex-specific health behaviors, such as physical activity and diet.⁴² We lacked data on the full complement of health behaviors relevant to cardiometabolic dysfunction, and the available variables (eg, physical activity) were subject to errors inherent in large epidemiologic surveys. However, these errors likely occurred at random; thus, our estimates were likely to be conservative. Second, inclusion in the model of area-level percentage of non-Hispanic black residents, a widely used proxy for societal disinvestments (eg, access to health care, education),⁴³ strengthened the effect among boys and did not influence the null effect among girls. The apparently sex-specific effects of societal disinvestment were unexpected and were not anticipated by theory. Further investigation of this interaction is warranted. Third, examinations of area-level determinants of health can produce various results depending on how areas of residence are defined. Thus, our findings may diverge from analyses using different geographic units of analysis. However, census tracts produce more consistent observations than other small geographies.³¹ Fourth, persons residing in high-poverty areas may be at risk for greater exposure to adverse childhood experiences than those residing in low-poverty areas,⁴ and we did not account for adverse childhood experiences in our analysis.

Fifth, our cardiometabolic dysfunction score is not a validated predictor of future disease, and we used empirical quartile cut points that do not represent a clinically meaningful increase in risk. Our cardiometabolic dysfunction score reflects levels of 6 biomarkers of metabolic syndrome,¹⁷ and these biomarkers are predictive of future disease. Some persons with hypercholesterolemia may have been included in our analyses, because a clinical cut point for hypercholesterolemia among adolescents does not exist. Finally, inclusion of obese persons likely increased the prevalence of cardiometabolic dysfunction. Obese persons were not excluded because doing so would have reduced our sample by approximately 20%, thereby lowering statistical power.

Conclusions

The accumulating evidence linking area deprivation with cardiometabolic dysfunction^7

-11 emphasizes the potential of community-based health promotion programs targeting adolescents⁴⁴ as a way to improve health. Poverty is linked with housing instability.^45
-47 Of relevance are interventions that help to mitigate housing instability by linking at-risk families with social workers who facilitate eligible families’ access to public housing⁴⁸ and programs that subsidize heating fuel, control rent to improve accessibility, and provide tax incentives to promote home maintenance. These social benefits could have long-term health benefits through their influence on early precursors of cardiometabolic health, in addition to their well-documented more immediate health benefits.⁴⁹ More structural interventions include national policies, such as progressive tax codes, that are intended to mute the effects of relative poverty linked to income inequality.^50,51 Finally, broader neighborhood investments in safety,⁵² parks and other recreational facilities, walkability, and ready access to public transportation can improve public health regardless of individual income levels.⁴⁷

Through a rigorous epidemiologic investigation, we found more pronounced associations between residence in high-poverty areas and cardiometabolic dysfunction among adolescent boys than among adolescent girls in the United States. Funding of evidence-based community health programs is warranted to improve cardiometabolic health among adolescents, which may have lifelong benefits.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Edmond D. Shenassa, ScD, MA

References

Mozaffarian

Benjamin

, et al. Heart disease and stroke statistics—2016 update: a report from the American Heart Association. Circulation. 2016;133(4):e38–360. doi:10.1161/CIR.0000000000000350

Suglia

Koenen

Boynton-Jarrett

, et al. Childhood and adolescent adversity and cardiometabolic outcomes: a scientific statement from the American Heart Association. Circulation. 2018;137(5):e15–e28. doi:10.1161/CIR.0000000000000536

Elsenburg

van Wijk

KJE

Liefbroer

Smidt

. Accumulation of adverse childhood events and overweight in children: a systematic review and meta-analysis. Obesity. 2017;25(5):820–832. doi:10.1002/oby.21797

Baglivio

Wolff

Epps

Nelson

. Predicting adverse childhood experiences: the importance of neighborhood context in youth trauma among delinquent youth. Crime Delinq. 2017;63(2):166–188. doi:10.1177/0011128715570628

Finegood

Rarick

JRD

Blair

; Family Life Project Investigators. Exploring longitudinal associations between neighborhood disadvantage and cortisol levels in early childhood. Dev Psychopathol. 2017;29(5):1649–1662. doi:10.1017/S0954579417001304

Cozma

Dima-Cozma

Ghiciuc

Pasquali

Saponaro

Patacchioli

. Salivary cortisol and a-amylase: subclinical indicators of stress as cardiometabolic risk. Braz J Med Biol Res. 2017;50(2):e5577. doi:10.1590/1414-431X20165577

Merten

Williams

Shriver

. Breakfast consumption in adolescence and young adulthood: parental presence, community context, and obesity. J Am Diet Assoc. 2009;109(8):1384–1391. doi:10.1016/j.jada.2009.05.008

Janssen

Boyce

Simpson

Pickett

. Influence of individual- and area-level measures of socioeconomic status on obesity, unhealthy eating, and physical inactivity in Canadian adolescents. Am J Clin Nutr. 2006;83(1):139–145.

Alvarado

. Neighborhood disadvantage and obesity across childhood and adolescence: evidence from the NLSY children and young adults cohort (1986-2010). Soc Sci Res. 2016;57:80–98. doi:10.1016/j.ssresearch.2016.01.008

10.

Theall

Drury

Shirtcliff

. Cumulative neighborhood risk of psychosocial stress and allostatic load in adolescents. Am J Epidemiol. 2012;176(suppl 7):S164–S174. doi:10.1093/aje/kws185

11.

Williams

Shenassa

Slopen

Rossen

. Cardiometabolic dysfunction among U.S. adolescents and area-level poverty: race/ethnicity-specific associations. J Adolesc Health. 2018;63(5):546–553. doi:10.1016/j.jadohealth.2018.07.003

12.

Bale

Epperson

. Sex differences and stress across the lifespan. Nat Neurosci. 2015;18(10):1413–1420. doi:10.1038/nn.4112

13.

Romeo

. The metamorphosis of adolescent hormonal stress reactivity: a focus on animal models. Front Neuroendocrinol. 2018;49:43–51. doi:10.1016/j.yfrne.2017.12.003

14.

McCormick

Mathews

. HPA function in adolescence: role of sex hormones in its regulation and the enduring consequences of exposure to stressors. Pharmacol Biochem Behav. 2007;86(2):220–233. doi:10.1016/j.pbb.2006.07.012

15.

Nelson

Ragland

Syme

. Longitudinal prediction of adult blood pressure from juvenile blood pressure levels. Am J Epidemiol. 1992;136(6):633–645. doi:10.1093/exfordjournals.aje.a116543

16.

Magnussen

Cheriyan

Sabin

, et al. Continuous and dichotomous metabolic syndrome definitions in youth predict adult type 2 diabetes and carotid artery intima media thickness: the Cardiovascular Risk in Young Finns Study. J Pediatr. 2016;171:97–103.e1-3. doi:10.1016/j.jpeds.2015.10.093

17.

Kelly

Steinberger

Jacobs

Hong

Moran

Sinaiko

. Predicting cardiovascular risk in young adulthood from the metabolic syndrome, its component risk factors, and a cluster score in childhood. Int J Pediatr Obes. 2011;6(2-2):e283–289. doi:10.3109/17477166.2010.528765

18.

Chen

Srinivasan

, et al. Childhood cardiovascular risk factors and carotid vascular changes in adulthood: the Bogalusa Heart Study [published correction appears in JAMA. 2003;290(22):2943]. JAMA. 2003;290(17):2271–2276. doi:10.1001/jama.290.17.2271

19.

National Center for Health Statistics. National Health and Nutrition Examination Survey: questionnaires, datasets, and related documentation. https://wwwn.cdc.gov/nchs/nhanes/default.aspx. Published 2019. Accessed August 15, 2019.

20.

US Census Bureau. Introduction to Census 2000 data products. http://www.census.gov/prod/2001pubs/mso-01icdp.pdf. Published 2001. Accessed September 27, 2015.

21.

US Census Bureau. American Community Survey (ACS) design and methodology report. http://census.gov/programs-surveys/acs/methodology/design-and-methodology.html. Published 2014. Accessed September 27, 2015.

22.

Webber

Srinivasan

Wattigney

Berenson

. Tracking of serum lipids and lipoproteins from childhood to adulthood. The Bogalusa Heart Study. Am J Epidemiol. 1991;133(9):884–899. doi:10.1093/oxfordjournals.aje.a115968

23.

Siu

Yuen

. Supplementary use of HbA_1c as hyperglycemic criterion to detect metabolic syndrome. Diabetol Metab Syndr. 2014;6(1):119. doi:10.1186/1758-5996-6-119

24.

Spear

. Adolescent growth and development. J Am Diet Assoc. 2002;102(3 suppl):S23–S29.

25.

Baumgartner

Siervogel

Chumlea

Roche

. Associations between plasma lipoprotein cholesterols, adiposity and adipose tissue distribution during adolescence. Int J Obes. 1989;13(1):31–41.

26.

US Department of Health and Human Services, National Institutes of Health, National Heart, Lung, and Blood Institute. The Fourth Reports on the Diagnosis, Evaluation, and Treatment of High Blood Pressure in Children and Adolescents. NIH Publication No. 05-05267. Washington, DC: US Department of Health and Human Services; 1996. http://www.nhlbi.nih.gov/files/docs/resources/heart/hbp_ped.pdf. Published 2005. Updated May 2005. Accessed January 1, 2015.

27.

Morrison

Friedman

Wang

Glueck

. Metabolic syndrome in childhood predicts adult metabolic syndrome and type 2 diabetes mellitus 25 to 30 years later. J Pediatr. 2008;152(2):201–206. doi:10.1016/j.jpeds.2007.09.010

28.

van’t Riet

Rijkelijkhuizen

Alssema

, et al. HbA_1c is an independent predictor of non-fatal cardiovascular disease in a Caucasian population without diabetes: a 10-year follow-up of the Hoorn Study. Eur J Prev Cardiol. 2012;19(1):23–31. doi:10.1097/HJR.0b013e32833b0932

29.

Kaikkonen

Kresanov

Ahotupa

, et al. Longitudinal study of circulating oxidized LDL and HDL and fatty liver: the Cardiovascular Risk in Young Finns Study. Free Radic Res. 2016;50(4):396–404. doi:10.3109/10715762.2015.1133906

30.

Dietz

. Health consequences of obesity in youth: childhood predictors of adult disease. Pediatrics. 1998;101(3 pt 2):518–525.

31.

Chen

Rehkopf

Waterman

, et al. Mapping and measuring social disparities in premature mortality: the impact of census tract poverty within and across Boston neighborhoods, 1999-2001. J Urban Health. 2006;83(6):1063–1084. doi:10.1007/s11524-006-9089-7

32.

SAS [software]. Version 9.3. Cary, NC: SAS Institute, Inc; 2011.

33.

Ahrens

Moreno

Marild

, et al. Metabolic syndrome in young children: definitions and results of the IDEFICS study. Int J Obes (Lond). 2014;38(suppl 2):S4–14. doi:10.1038/ijo.2014.130

34.

Vitaliano

Scanlan

Zhang

Savage

Hirsch

Siegler

. A path model of chronic stress, the metabolic syndrome, and coronary heart disease. Psychosom Med. 2002;64(3):418–435.

35.

Bourke

Neigh

. Behavioral effects of chronic adolescent stress are sustained and sexually dimorphic. Horm Behav. 2011;60(1):112–120. doi:10.1016/j.yhbeh.2011.03.011

36.

Bjorntorp

Rosmond

. The metabolic syndrome—a neuroendocrine disorder? Br J Nutr. 2000;83(suppl 1):S49–S57. doi:10.1017/s0007114500000957

37.

Borawski

Levels-Landis

Lovegreen

Trapl

. Parental monitoring, negotiated unsupervised time, and parental trust: the role of perceived parenting practices in adolescent health risk behaviors. J Adolesc Health. 2003;33(2):60–70. doi:10.1016/S1054-139X(03)00100-9

38.

Romer

Stanton

Galbraith

Feigelman

Black

. Parental influence on adolescent sexual behavior in high-poverty settings. Arch Pediatr Adolesc Med. 1999;153(10):1055–1062.

39.

Feigelman

Stanton

. Perceived parental monitoring and health risk behaviors among urban low-income African-American children and adolescents. J Adolesc Health. 2000;27(1):43–48.

40.

Dedovic

Wadiwalla

Engert

Pruessner

. The role of sex and gender socialization in stress reactivity. Dev Psychol. 2009;45(1):45–55. doi:10.1037/a0014433

41.

Shenassa

Rossen

Cohen

Morello-Frosch

Payne-Sturges

. Income inequality and US children’s secondhand smoke exposure: distinct associations by race-ethnicity. Nicotine Tob Res. 2017;19(11):1292–1299. doi:10.1093/ntr/ntw293

42.

Lee

Tsenkova

Boylan

Ryff

. Gender differences in the pathways from childhood disadvantage to metabolic syndrome in adulthood: an examination of health lifestyles. SSM Popul Health. 2018;4:216–224. doi:10.1016/j.ssmph.2018.01.003

43.

Deaton

Lubotsky

. Mortality, inequality and race in American cities and states. Soc Sci Med. 2003;56(6):1139–1153. doi:10.1016/s0277-9536(02)00115-6

44.

Task Force on Community Preventive Services. Recommendations to increase physical activity in communities. Am J Prev Med. 2002;22(4 suppl):67–72.

45.

March

Ettinger de Cuba

Cook

, et al. Behind Closed Doors: The Hidden Health Impact of Being Behind on Rent. Boston, MA: Children’s Health Watch; 2011. http://childrenshealthwatch.org/behind-closed-doors-the-hidden-health-impact-of-being-behind-on-rent. Accessed August 15, 2019.

46.

Ellaway

Macdonald

Kearns

. Are housing tenure and car access still associated with health? A repeat cross-sectional study of UK adults over a 13-year period. BMJ Open. 2016;6(11):e012268. doi:10.1136/bmjopen-2016-012268

47.

Sharpe

Taylor

Fleming

Morrissey

Morris

Wigglesworth

. Making the case for “whole system” approaches: integrating public health and housing. Int J Environ Res Public Health. 2018;15(11):pii:E2345. doi:10.3390/ijerph15112345

48.

Tebb

Pica

Twietmeyer

Diaz

Brindis

. Innovative approaches to address social determinants of health among adolescents and young adults. Health Equity. 2018;2(1):321–328. doi:10.1089/heq.2018.0011

49.

Mari-Dell’Olmo

Novoa

Camprubi

, et al. Housing policies and health inequalities. Int J Health Serv. 2017;47(2):207–232. doi:10.1177/0020731416684292

50.

Ross

Lynch

. Commentary: the contingencies of income inequality and health: reflections on the Canadian experience. Int J Epidemiol. 2004;33(2):318–319. doi:10.1093/ije/dyh108

51.

Shmueli

. Population health and income inequality: new evidence from Israeli time-series analysis. Int J Epidemiol. 2004;33(2):311–317. doi:10.1093/ije/dyh035

52.

Shenassa

Liebhaber

Ezeamama

. Perceived safety of area of residence and exercise: a pan-European study. Am J Epidemiol. 2006;163(11):1012–1017. doi:10.1093/aje/kwj142

Sex-Specific Associations Between Area-Level Poverty and Cardiometabolic Dysfunction Among US Adolescents

Abstract

Objective:

Methods:

Results:

Conclusion:

Keywords

Methods

Sample

Cardiometabolic Dysfunction Score

Area-Level Poverty

Statistical Analysis

Results

Discussion

Limitations

Conclusions

Footnotes

Declaration of Conflicting Interests

Funding

ORCID iD

References