Maternal Stress Moderates the Relationship of Food Insufficiency with Body Mass Index Trajectories from Childhood to Early Adulthood among U.S. Rural Youth

Abstract

Background: Findings on the relationships between household food insufficiency (HFI), maternal stress, and youth body mass index (BMI) are mixed, possibly due to cross-sectional study designs and measurement issues. Furthermore, little is known about how childhood exposure to HFI and maternal stress influences BMI into young adulthood among rural youth. We aimed to determine the independent and moderating relationships of HFI and maternal perceived stress on youth BMI trajectories from age 9 to 24 years.

Methods: We used longitudinal data from rural New York youth (n = 341). At youth age 9 years, parents reported HFI using a reliable one-item measure, and mothers responded to the 10-item Perceived Stress Scale (PSS; range: 0–40). BMI was calculated (kg/m²) from objective measures of height and weight at 9, 13, 17, and 24 years. Multivariate random-intercept trajectory models estimated the relationships of HFI and PSS on BMI trajectories (p < 0.05 for main effects, p < 0.10 for interactions).

Results: At age 9 years, 16.4% experienced HFI and mean (standard deviation) BMI and PSS were 18.4 (3.6) kg/m² and 7.7 (2.9), respectively. HFI and PSS were not associated with BMI trajectories (p = 0.18, p = 0.64, respectively), but their interaction was significant (p < 0.01). Each one-unit increase in PSS was associated with 0.6 (0.2) kg/m² higher mean change in BMI trajectories for youth in food-insufficient, compared with food-sufficient, households.

Conclusions: Higher levels of maternal stress in food-insufficient households may lead to greater increases in BMI from childhood to young adulthood. Public health interventions should simultaneously address parental stress and quality food access among low-income rural households.

Introduction

Household food insecurity, or inconsistent and uncertain access of household members to sufficient, safe, and nutritious food for an active, healthy lifestyle, affects 10.5% of total U.S. households and 13.6% of U.S. households with children.¹ Household food insecurity prevalence is highest in principal cities (12.4%) and rural areas (12.1%), compared with suburban and other metropolitan areas outside of principal cities (8.3%).¹ Food insecurity is associated with poor behavioral and emotional health,² and lower diet quality³ among children and adolescents, with poor mental health² and diet quality,³ type 2 diabetes,⁴ and cardiovascular disease risk⁵ among adults, and obesity among women.⁶

The relationship between food insecurity and obesity among youth is less clear.^7,8 Approximately one-third of U.S. youth have obesity (35% of rural youth, 30% of urban youth).⁹ While some studies have demonstrated linear associations of food insecurity with youth obesity, other studies have shown U-shaped patterns or no association.⁶ Most studies have been cross-sectional, so longitudinal evidence is needed to better understand the potential long-term implications of food insecurity on youth body weight development. Of the longitudinal studies available, some studies found no significant association,^10–12 whereas other studies have shown that food insecurity may influence youth body weight differently depending on the sex of the child, transient versus persistent household food insecurity, proximal versus distal exposure to food insecurity, and birth weight.^7,13–17

Notably, two cross-sectional studies considered the interaction of food insecurity and maternal stress on obesity, an important contextual factor as psychosocial stressors in the family are linked to childhood obesity.¹⁸ One study found a negative association of the interaction of food insecurity and maternal stressors on obesity,¹⁹ whereas the other study found a positive association.²⁰ However, both studies assessed maternal stress as a cumulative index of external risk exposure that may cause stress rather than the mother's own perceptions of stress. The latter may better capture the degree to which the mother is coping with external stressors. Moreover, valid and reliable assessment tools for perceived stress are available and can be easily employed in clinical and community settings, further reflecting the value of assessing perceived stress.^21,22 Additional investigations using longitudinal data and a standard stress assessment tool are needed to understand the complexities of how food insecurity and maternal stress shape youth weight trajectories, particularly as youth enter young adulthood. Examining this longevity of childhood factors can inform more effective approaches to prevent unhealthy weight gain from childhood to adulthood.

To the best of our knowledge, none of the published longitudinal studies investigating both food insecurity and maternal stress have modeled body mass index (BMI) as a growth curve, or trajectory, from childhood to adulthood. Two previous studies have used a trajectory analysis to model the association between food insecurity and BMI in childhood and from childhood to adulthood, but did not incorporate maternal stress.^17,23 Likewise, several studies have used a trajectory analysis to model the association of parental stress with youth BMI over time, but did not incorporate food insecurity.^24,25 A trajectory analysis provides simultaneous estimates for both between- and within-person effects, accounting for the relationship between an intraindividual (e.g., household food security status, maternal perceived stress level) and an interindividual (e.g., BMI trajectory) variable over time.²⁶

Given the notable limitations of methodological approaches that have yielded mixed findings on food insecurity and obesity among youth, we sought to address these gaps by determining the relationship of household food insecurity and maternal stress at youth age 9 years, independently and as an interaction, on youth BMI trajectories from childhood (9 years) to young adulthood (24 years). We hypothesized that youth in food-insecure households at age 9 years would demonstrate significantly greater increases in BMI trajectories compared with youth in food-secure households at age 9 years, and that this relationship would be moderated by maternal perceived stress, such that youth in food-insecure households would have significantly greater increases in BMI trajectories with higher maternal stress. We selected child age 9 years, rather than later time points, as BMI trajectories are often established in early to middle childhood^27–29 and child age 9 years was the earliest data available for food insecurity and maternal stress. This approach can provide more rigorous evidence, compared with prior cross-sectional comparisons, as within-person changes in adiposity are examined and, thus, less subject to alternative explanatory variables.

Methods

Participants and Procedures

Participants were predominately (94%) white youth from rural upstate New York born between 1983 and 1990 and participating in a research study (n = 341) on rural poverty and child development.³⁰ Participants were recruited from Head Start, Section 8 federal housing, New York State Cooperative Extension, and public schools. Low-income families were oversampled to align with the goals of the research study. Participants and their mothers were interviewed when participants were ages 9, 13, 17, and 24 years. The University Institutional Review Board approved the study. Child assent and parental consent was provided at youth age 9, 13, and 17 years and informed consent was obtained at age 24 years.

Measures

Independent variables

Maternal stress was measured at youth age 9 years using the Perceived Stress Scale (PSS),^21,22 a 10-item scale assessing a respondent's perceived stress level in various aspects of her life (e.g., upset because of unexpected events, ability to control irritations) over the past year. Respondents rate items on a scale of 0 (never) to 4 (very often). Scale items are summed for a total score, ranging between 0 and 40, with higher scores indicating higher perceived stress. The PSS has undergone extensive psychometric development and is a valid and reliable instrument for the measurement of psychological stress.^21,22

Household food sufficiency status was assessed at youth age 9 years. A youth's parent responded “yes” or “no” to the statement, “Sometimes our family had little food to eat.” A response of “yes” was classified as household food insufficiency (HFI). At youth age 13 years, both the one-item HFI and the 18-item United States Department of Agriculture (USDA) food security survey module³¹ were employed to assess HFI and household food security status, respectively. Thus, to determine the reliability of the one-item HFI measure, we computed the Cronbach's Coefficient Alpha between the HFI measure at youth age 13 years and the USDA 18-item food security survey module at youth age 13 years. Reliability of the one-item measure was good (α = 0.76), as defined by Nunnally.³² We also ran one-factor analysis of variance (ANOVA) and logistic regression models to test the criterion validity of the one-item HFI measure in predicting the 18-item continuous score and the odds of being food insecure (category defined as three or more affirmative responses to the 18-item score), respectively. The one-item HFI significantly predicted higher scores on the 18-item module (F-value: 118.75, p < 0.0001) and was associated with higher odds of being classified as food insecure by the 18-item module [odds ratio (95% confidence intervals): 110 (13.3–908), p < 0.0001].

Dependent variable

Trained research assistants measured youth height and weight at each wave of the study, using a cloth measuring tape and calibrated digital scale. BMI was calculated as weight (kilograms) divided by height (meters) squared (kg/m²).

Covariates

Youth sex, youth birth weight (oz), and household income-to-needs were self-reported by the mother and were included as covariates. Income-to-needs is a ratio of the household income at youth age 9 years to the Federal poverty threshold for household size at that time. Ratios below 1.0 indicate income below the poverty level, while ratios above 1.0 indicate income above the poverty level.

Analyses

Our analytical sample was 341 youth. Missing data (HFI = 69; PSS = 2; BMI at age 9 years = 76; BMI at age 13 years = 137; BMI at age 17 years = 134; BMI at age 24 years = 120; and birth weight = 126) were imputed in SPSS version 25 using a Markov Chain Monte Carlo algorithm known as fully conditional specification, or chained equations. We tested bivariate relationships of youth, maternal, and household characteristics by food sufficiency status at age 9 years using t tests for continuous variables and chi-square (or Fisher exact test where appropriate) for categorical variables.

To build a longitudinal trajectory model, we first fit an unconditional model with a random intercept to quantify BMI variation across youth without regard to time. This unconditional model provided a standard by which subsequent models could be compared, and determined the overall variation in BMI by fitting an overall mean and variance across all youth and measurement occasions. We then calculated the intraclass correlation coefficient (ICC) to determine what proportion of variation could be attributed to between-persons effects, as data were longitudinal and clustering occurs at the within-person level, and to determine if clustering of BMI values differed significantly both within and between youth, or only within youth. The ICC equation is: ICC = between subject variance (random intercept estimate)/(between subject variance (random intercept estimate) + within subject variance (residual variance estimate)). The ICC for our data was: 13.8550/13.8550 + 21.6557 = 0.390 (39.0%), indicating a large proportion of the variation in BMI was occurring between youth and, thus, a multilevel model was appropriate for capturing BMI change.

Second, we added time to the unconditional model by including the wave (9, 13, 17, 24 years) of data collection (within-youth BMI change) to create our level-1 model. Next, we added our level-2 predictors, HFI and maternal PSS, to our level-1 model, which allowed us to test if these between-youth predictors significantly contributed to BMI change over time. We tested HFI and maternal PSS independently with their interactions with time. Then, we included them together in the model. Last, we tested an interaction effect of HFI and maternal PSS on BMI trajectories. Level-2 models were subsequently adjusted for youth sex, youth birth weight, and household income-to-needs ratio.

With maximum likelihood estimation in random intercept trajectory models, a sample size of 341 with four repeated measurement time points has sufficient (80%) theoretical and empirical power, even for comparison of groups with unbalanced group sizes.³³ All analyses were conducted in SAS 9.4 with significance set at p < 0.05 for main effects and p < 0.10 for interactions.

Results

Sample Characteristics

Mean [standard deviation (SD)] BMI at age 9 years was 18.4 (3.6), at age 13 years was 22.7 (5.0), at age 17 years was 25.6 (5.5), and at age 24 years was 26.7 (5.7). At age 9 years, 16.4% lived in a food-insufficient household, and mean (SD) maternal PSS was 7.7 (2.9). Compared with youth from food-sufficient households, youth from food-insufficient households lived in households with lower income-to-needs ratios and had higher BMI. Youth from food-insufficient households also tended to have mothers with higher PSS (Table 1).

Table 1.

Characteristics by Household Food Sufficiency Status among Rural New York Youth (2002–2014) at Age 9 Years (n = 341)

Characteristic	Food sufficient	Food insufficient	p
	n = 285	n = 56
Female	47.7	55.4	0.30
Household income-to-needs ratioa	1.8 (1.1)	1.2 (0.9)	0.02
Birth weight (oz)	118 (25.5)	113 (28.2)	0.19
BMI (kg/m2)	18.2 (3.3)	19.3 (4.5)	0.03
Maternal PSS scoreb	7.5 (2.7)	8.8 (3.3)	0.06

Data shown as either mean (SD) or %. Analyses included paired t tests for continuous variables [mean (SD)] and chi-square (or Fisher exact test when appropriate) for categorical variables (%).

^aRatio of household income to the appropriate poverty threshold for the household.

^bMeasure of perceived stress level in various aspects of life over the past year. Higher scores indicate higher perceived stress (range: 0–40).

BMI, body mass index; PSS, Perceived Stress Scale; SD, standard deviation.

Level-1 Model: Adding Data Collection Timepoint

When time was added to the unconditional model, BMI for each youth significantly increased from age 9 to 13 years, from age 9 to 17 years, and from age 9 to 24 years (Table 2).

Table 2.

Coefficients (Standard Error) from Mixed Effects Trajectory Models of BMI among Rural New York Youth (2002–2014) from Age 9 to 24 Years, by Household Food Sufficiency Status or Maternal Stress at Youth Age 9 Years (n = 341)

	Unconditional model	Model with time	Model with time and HFI	Model with time and HFI, adjusted	Model with time and maternal PSS	Model with time and maternal PSS, adjusted
Intercepta	23.4 (0.2)**	18.4 (0.3)**	18.2 (0.3)**	18.7 (1.1)**	18.0 (0.8)**	19.3 (1.4)**
Time (years)
Age 9		Ref.	Ref.	Ref.	Ref.	Ref.
Age 13		4.4 (0.2)**	4.2 (0.2)**	4.2 (0.2)**	4.3 (0.6)**	4.3 (0.6)**
Age 17		7.2 (0.2)**	7.2 (0.2)**	7.2 (0.2)**	7.1 (0.6)**	7.1 (0.6)**
Age 24		8.4 (0.2)**	8.4 (0.2)**	8.4 (0.2)**	8.0 (0.6)**	8.0 (0.6)**
Main effect
Food sufficient household			Ref.	Ref.	—	—
Food insufficient household			1.1 (0.7)	0.7 (0.7)	—	—
Main effect			—	—
Maternal PSS			—	—	0.1 (0.1)	−0.01 (0.1)
Covariates
Male				−0.1 (0.5)		−0.1 (0.5)
Birth weight (oz)				0.01 (0.01)		0.01 (0.01)
Income-to-needs ratio				−0.8 (0.2)*		−0.9 (0.2)**
Interaction
Food sufficiency status * Age 9 years			Ref.	Ref.
Food insufficiency * Age 13 years			0.8 (0.6)	0.8 (0.6)
Food insufficiency * Age 17 years			0.1 (0.6)	0.1 (0.6)
Food insufficiency * Age 24 years			−0.5 (0.6)	−0.5 (0.6)
Maternal PSS * Age 9 years					Ref.	Ref.
Maternal PSS * Age 13 years					0.004 (0.1)	0.004 (0.1)
Maternal PSS * Age 17 years					0.01 (0.1)	0.01 (0.1)
Maternal PSS * Age 24 years					0.01 (0.1)	0.1 (0.1)

HFI, household food insufficiency; PSS, Perceived Stress Scale.

^aRandom intercept model.

^*Denotes significance at p < 0.001.

^**Denotes significance at p < 0.0001.

Level-2 Models: Adding the Main Effects and Interaction

HFI

The main effect of HFI (p = 0.12) did not significantly predict BMI trajectories from age 9 to 24 years, before adjusting for covariates (Table 2). Including the interaction of HFI and time in the model was also not significant, before adjusting for covariates (p = 0.18). Adjusting for covariates did not change the relationship of HFI (p = 0.34) or the interaction of HFI and time (p = 0.18) with BMI trajectories.

Maternal PSS

The main effect of maternal PSS (p = 0.60) and the interaction of maternal PSS and time (p = 0.91) also did not significantly predict BMI trajectories from age 9 to 24 years, before adjusting for covariates (Table 2). After adjusting for covariates, maternal PSS (p = 0.64) and the interaction of maternal PSS and time (p = 0.91) remained nonsignificant predictors of BMI trajectories.

HFI and maternal PSS

Neither HFI nor maternal PSS were significant as main effects when both were included together in the unadjusted (HFI: p = 0.07; maternal PSS: p = 0.64) and adjusted (HFI: p = 0.18; maternal PSS: p = 0.60) models for BMI trajectories (Table 3).

Table 3.

Coefficients (Standard Error) from Mixed Effects Trajectory Models of BMI among Rural New York Youth (2002–2014) from Age 9 to 24 Years, by Household Food Sufficiency Status and Maternal Stress at Youth Age 9 Years (n = 341)

	Model with time, HFI, and maternal PSS, unadjusted	Model with time, HFI, and maternal PSS, adjusted	Model with time, HFI, and maternal PSS, adjusted+interaction
Intercepta	17.9 (0.7)****	19.0 (1.3)****	19.8 (1.3)****
Time (years)
Age 9	Ref.	Ref.	Ref.
Age 13	4.4 (0.2)****	4.4 (0.2)****	4.4 (0.2)****
Age 17	7.2 (0.2)****	7.2 (0.2)****	7.2 (0.2)****
Age 24	8.4 (0.2)****	8.4 (0.2)****	8.4 (0.2)****
Main effect
Food sufficient household	Ref.	Ref.	Ref.
Food insufficient household	1.2 (0.6)	0.9 (0.6)	−3.9 (1.8)*
Main effect
Maternal PSS	0.04 (0.1)	−0.04 (0.1)	−0.2 (0.1)
Covariates
Male		−0.1 (0.5)	−0.03 (0.5)
Birth weight (oz)		0.01 (0.01)	0.01 (0.01)
Income-to-needs ratio		−0.9 (0.2)***	−0.9 (0.2)****
Interaction
Food insufficient household * maternal stress			0.6 (0.2)**

HFI, household food insufficiency; PSS, Perceived Stress Scale.

^aRandom intercept model.

^*Denotes significance at p < 0.05.

^**Denotes significance at p < 0.01.

^***Denotes significance at p < 0.001.

^****Denotes significance at p < 0.0001.

HFI and maternal PSS interaction

The interaction of HFI and maternal PSS was significant (p = 0.0045; Table 3), indicating that the effect of age 9 years HFI on BMI trajectories depended on levels of maternal PSS at age 9 years. For every one-unit increase in the maternal PSS score for youth age 9 years in food-insufficient households, mean change in BMI trajectories from age 9 to 24 years was 0.6 kg/m² higher compared with their counterparts in food-sufficient households. The relationship of the interaction of HFI and maternal PSS on BMI at age 24 years is demonstrated in Figure 1.

Figure 1.

Mean (SE) youth BMI at age 24 years for food-sufficient versus food-insufficient households by maternal perceived stress at age 9 years. BMI, body mass index; SE, standard error.

Discussion

In this cohort of rural youth, HFI, compared with household food sufficiency, at age 9 years was associated with a greater increase in BMI from age 9 to 24 years among youth with mothers reporting higher perceived stress. HFI and maternal PSS at age 9 years were not independently associated with youth BMI trajectories from age 9 to 24 years. Of note, prevalence of HFI (16.4%) in this sample of rural youth at age 9 years (1992–1999) was higher than prevalence of food insecurity in rural areas of the U.S. between 1995 (10.8%) and 1998 (10.6%), the first years in which food insecurity were assessed nationwide.³⁴ The higher levels of HFI are probably due to our stratified sample with approximately half below the poverty line at the time of recruitment.

Despite higher prevalence of food insecurity¹ and obesity⁹ in rural areas, few studies to date have focused on the role of food insecurity in healthy weight development among rural youth.^23,35,36 In a mostly white sample of youth in rural Iowa, living in a food-insecure household and simultaneously experiencing harsh parenting at age 13 years were associated with higher odds of overweight or obesity at age 23 years among females, but not males.³⁶ Similar to our findings, the authors found that food insecurity alone was not associated with odds of overweight or obesity in emerging adulthood, but interacted with harsh parenting to influence weight status in adulthood.³⁶

Upon additional follow-up in the Iowa cohort and another Iowa cohort, Lohman et al. documented a strong association between higher levels of food insecurity at age 15 years and faster linear rates of increase in BMI from age 15 to 31 years, and a positive association between changes in food insecurity and changes in BMI over these 16 years among females, but not males.²³ However, these studies relied on self-reported height and weight. Compared with objective measures, self-reported height and weight have been shown to overestimate the association of food insecurity with BMI.³⁷ Furthermore, Lohman et al. did not examine maternal stress as a potential moderator of the impacts of early HFI on BMI trajectories. Our analyses demonstrated that higher maternal perceived stress early in childhood accentuates the link between HFI early in childhood and BMI from age 9 years through 24 years.

Previous studies among youth have demonstrated mixed results when testing the association of food insecurity with overweight or obesity. Our findings for an interaction of food insecurity with maternal stress contribute a new insight into previously inconsistent relationships. In addition, most of the prior longitudinal studies used data from the Early Childhood Longitudinal Study-Kindergarten Cohort,^7,16,17 a younger age group at a different developmental stage compared with our sample.

Comparable to our findings, food insecurity and maternal risk exposure were not independently associated with overweight or obesity in a cross-sectional sample of diverse adolescents (ages 10–15 years) from three major U.S. cities, but maternal risk exposure was associated with significantly higher odds of child overweight or obesity among the food insecure, and not the food secure.²⁰ Our similar findings in a longitudinal sample of rural youth underscore the importance of the food insecurity-maternal stress relationship in impacting youth BMI, regardless of geographic region and race/ethnicity. Additionally, our measures for food insecurity and stress differed from Lohman et al.; they assessed food insecurity using questions only related to the child, not the household, and they assessed maternal risk exposure using six constructs of potential external stressors (employment, disability, self-esteem, psychological distress, self-reported health, and support) and not the PSS which is a standardized, well-validated instrument.²⁰

Regardless of these measurement differences, the similar findings across the two studies suggest that even simple assessments of food insecurity, with a one-item screener, and maternal stress, with the commonly used PSS, could be useful in screening for risk of development of youth overweight or obesity in low-income, rural households. Although screening for risk may be one strategy to prevent or reduce the risk of unhealthy weight trajectories among rural youth, identifying feasible behavioral interventions to support low-income, rural households is necessary for progress, although more complicated.

The food insecurity/maternal stress relationship with BMI trajectories may be functioning through several potential mechanisms that could serve as behavioral targets. Parents in households with lower income-to-needs put more pressure on their child to eat, which may operate within the context of the type of food available in the home, namely high-fat, high-sugar foods.³⁸ Likewise, earlier and concurrent food insecurity have been associated with food-related parenting behaviors offering less structure for child eating (e.g., irregular meal times).³⁹ Low-income parents with higher parenting stress have also demonstrated the use of an uninvolved feeding style, which may contribute to a child's unhealthy weight gain.⁴⁰

Additionally, youth in households experiencing food insecurity and high maternal stress may develop unhealthy strategies to cope with stress, which may persist into adulthood. Previous research has suggested that food-insecure households use coping responses that can lead to weight gain, like substituting high-calorie, high-fat food options for lower-calorie, nutrient-dense food options and overconsuming food when it is readily available, leading to a “feast/famine” cycle.⁴¹ Furthermore, emerging evidence suggests that food insecurity may impact neuroendocrine and inflammation systems of the body, which mediate metabolic dysregulation,⁴² including adiposity.

The study has several notable strengths. First, the longitudinal design of this study allowed for testing exposure to HFI and maternal PSS in childhood with BMI trajectories throughout adolescence and into early adulthood, filling a gap in the literature, which has mostly documented cross-sectional relationships. Similarly, the longitudinal design permitted the use of random-intercept, multilevel models, which accounted for variance in BMI both within and between youth, and better represented our observational data. We also employed a widely used, standardized assessment tool for psychological stress, the PSS that can be replicated in other sample populations. Third, BMI at every wave of data collection was calculated from objective measures of height and weight, not self-reported. Last, although food insufficiency and obesity are unfortunately ubiquitous in rural America, we know little about their interplay among rural youth or adults and the present study contributes to addressing this gap.

One limitation of our study is the use of a sample of rural, mostly white participants. Although this was likely representative of upstate New York, the extent to which our findings would generalize to urban and more ethnically diverse populations is unknown. Likewise, mothers in our sample had relatively low levels of PSS [mean (SD): 7.7 (2.9)]. Previous work has demonstrated mean levels of PSS in representative U.S. samples to be higher [15.5 (7.4)],⁴³ suggesting that our findings in a sample with relatively low maternal stress may underestimate the moderation effect of maternal PSS on the HFI/BMI relationship. Additionally, our small sample size did now allow for testing interactions between HFI, maternal PSS, and youth sex, which would have contributed unique insights to growing evidence that the relationship between HFI and BMI differs by sex over the life course.^17,23,44

Another limitation was the lack of data on food assistance participation. Participation in the Supplemental Nutrition Assistance Program may provide some protective effect on child BMI in food-insecure households.⁴⁵ We also did not have the full 18-item or short 6-item household food security survey module to assess household food security at youth age 9 years, although our one-item measure had good reliability and validity when compared with the 18-item module at youth age 13 years. Likewise, our HFI measure ascertained parental perceptions of household food security, which may not accurately capture child experiences of HFI and, thus, miss food-insecure children in the household.⁴⁶

In our study, we tested the hypothesis that childhood experiences of HFI and maternal stress would shape BMI trajectories into adulthood. We did not assess the role of change in HFI over time on BMI; this would be a different but important analysis that could provide additional insight into the role of transient versus persistent food insufficiency on weight status. Last, maternal BMI was not collected so we could not control for it in our models. Maternal BMI has been found to be associated with youth BMI from infancy to adulthood, likely due to shared genetic and environmental factors.⁴⁷ However, child birth weight is incorporated into all analyses herein.

Our findings provide a valuable contribution to the literature on the HFI-BMI relationship among youth by demonstrating that, among youth living in food-insufficient households, increasing levels of maternal stress during childhood promote higher BMI trajectories from childhood to young adulthood, independent of household income-to-needs. These findings underscore the importance for public health approaches or interventions to simultaneously address parental psychological stress and quality food access among low-income households.² Future studies should explore the potential role of variable HFI on youth BMI trajectories.¹³ Households with more persistent HFI may develop protective coping strategies compared with those households that move in and out of food insecurity.¹³ More importantly, future studies should explore this HFI-maternal stress relationship with BMI trajectories in other marginalized populations experiencing a high prevalence of food insecurity and maternal stress, which may increase the risk of youth obesity and chronic disease and further contribute to health inequities across the life course.

Authors' Contributions

A.C.M. designed the research question, analyzed the data, interpreted the results, drafted the article, and had primary responsibility for the final content. G.W.E. and K.L.D. contributed to the analytical design, and to interpretation and presentation of results. All authors read and approved the final article.

Funding Information

The rural youth cohort and G.W.E. received support from the W.T. Grant Foundation, the John D. and Catherine T. MacArthur Foundation Network on Socioeconomic Status and Health, and a National Institutes of Health (NIH) National Institute for Minority Health and Health Disparities Grant (5RC2MD00467). A.C.M. received support from an NIH-National Heart, Lung, and Blood K01 Mentored Career Development Award (K01-HL150406).

Author Disclosure Statement

No competing financial interests exist.