Psychosocial factors as mediators of food insecurity and weight status among middle school students

Appetite 103 (2016) 236e243

Contents lists available at ScienceDirect

Appetite journal homepage: www.elsevier.com/locate/appet

Psychosocial factors as mediators of food insecurity and weight status among middle school students Don E. Willis a, *, Kevin M. Fitzpatrick b a b

Department of Sociology, 312 Middlebush Hall, University of Missouri, Columbia, MO, 65211-6100, USA Department of Sociology & Criminal Justice, 211 Old Main, University of Arkansas, Fayetteville, AR 72701, USA

a r t i c l e i n f o

a b s t r a c t

Article history: Received 29 September 2015 Received in revised form 13 April 2016 Accepted 17 April 2016 Available online 21 April 2016

Research regarding the association between food insecurity and weight status among youth has produced mixed results. However, few studies on this topic have utilized data that includes survey responses from children themselves regarding their experience with food insecurity. This study was undertaken to examine the association between food insecurity and weight status among youth, as well as the potential mediation by psychosocial factors. A survey of 5th-7th grade students was administered to gather information on food insecurity, social and psychological resources, and health. The primary analysis includes OLS (Ordinary Least Squares) regression conducted using SPSS software and Sobel's test for mediation. Results suggest a positive association between food insecurity and weight status even when controlling for key demographic variables. In addition, we find that this association is mediated by psychosocial factorsdnamely, perceived social status and depression. Insights from this work highlight the need to consider non-nutritional pathways through which food insecurity impacts health as well the need to continue surveying youth directly when examining their experiences with food insecurity. © 2016 Elsevier Ltd. All rights reserved.

Keywords: Food insecurity Weight status Psychosocial factors Youth Depression Social status

1. Introduction Approximately 49 million Americans were living in food insecure households in 2013; among them were 15.8 million children (Coleman-Jensen, Gregory, & Singh, 2014). According to the definition of food security provided by the U.S. Department of Agriculture, these families did not have “access by all people at all times to enough food for an active, healthy life” (Coleman-Jensen et al., 2014). At the same time, another major health concern is prevalent in the United States; over a third of school-aged children are overweight and nearly one in five are obese (Ogden, Carroll, Kit, & Flegal, 2012). Both sets of statistics are alarming because of the documented associations between obesity with diabetes, hypertension, and heart disease (Freedman, Mei, Srinivasan, Berenson, & Dietz, 2007; Schwimmer, Burwinkle, & Varni, 2003) and the major health and developmental consequences scholars have shown to be linked with food insecurity (Alaimo, Olson, & Frongillo, 2001a,b; Jyoti, Frongillo, & Jones, 2005). These two phenomenadobesity and food insecuritydare often experienced simultaneously by the same households and individuals (Dinour, Bergen, & Yeh, 2007;

* Corresponding author. E-mail address: [email protected] (D.E. Willis). http://dx.doi.org/10.1016/j.appet.2016.04.022 0195-6663/© 2016 Elsevier Ltd. All rights reserved.

Franklin et al., 2012; Larson & Story, 2011). While the association between food insecurity and obesity has been fairly consistent for women, few studies find an association among men, and evidence remains inconsistent across groups of youth (Dinour et al., 2007; Franklin et al., 2012; Larson & Story, 2011). Using a cross-sectional data set with measures of child food insecurity that overcome limitations of previous workdnamely, their lack of direct responses from children themselves regarding food insecurity despite research showing that youth experience food insecurity differently than their parents (Fram et al., 2011; Harvey, 2016)dthis study addresses three central questions related to the link between food insecurity and weight status. First, is food insecurity associated with weight status among middle-school age children? Second, does the association remain significant when controlling for key demographic factors such as race, ethnicity, poverty, and sex? And third, is the relationship mediated by psychosocial factors such as depression, perceived social status, social capital, or self-esteem? 1.1. Health, food, and a psychosocial framework The psychosocial framework emphasizes exposure to both risks and resources that influence health and well-being over time.

D.E. Willis, K.M. Fitzpatrick / Appetite 103 (2016) 236e243

Everyday stressors as well as major life events shape how we experience our material circumstances (Pearlin, Menaghan, Lieberman, & Mullan, 1981, 2005). Psychosocial risks (e.g. depression) and resources (e.g. social capital, self-esteem, and social status) have profound influences on the health outcomes of a general population (Wilkinson, 2005). Chronic exposure to stressors can result in serious “wear and tear on the body and brain,” or “allostatic load” (McEwen, 1998). Both risks and resources potentially mediate external stressors such as food insecurity, or other forms of inequality as Wilkinson (2005) has persuasively argued, and thereby influence physiological responses such as weight gain. As Wilkinson (2005) explains, “psychosocial factors reflect material life because material life is a source of stress, whether in the form of unhappiness, depression, insecurity, anger, or anxiety.” Food insecurity is likely a state that carries with it both stress related to perceptions of social status and truly vulnerable material conditions in terms of access to healthy food. Thus, while it is possible that a good portion of the relation between food insecurity and obesity is due to the fact that food insecurity is not just about lack of access but lack of access to balanced, healthy meals on a regular basis, (Fram, Ritchie, Rosen, & Frongillo, 2015; Nackers & Appelhans, 2013), it also remains reasonable that the relation may be partially due to the adversity, anxiety, and stress (i.e. nonnutritional pathways) related to food insecurity (Chilton & Knowles, 2014; Chilton, Knowles, Rabinowich, & Arnold, 2015; Fram et al., 2011, 2013). Hunger and food insufficiency have been linked to depression and suicidal ideation among adolescents (Alaimo, Olson, & Frongillo, 2002; McIntyre, Williams, Lavorato, & Patten, 2013). Food insecurity, poverty, and social status are closely linked. Low social status can be a salient stressor in the everyday life experience. While status is often thought of in economic terms, there are other daily practices closely linked to status that may signify our social position. One such practice is eating. Not just the way one eats, but also the types of foods one eats, are among the many ways individuals distinguish class and status in everyday practices (Bourdieu, 1984). Thus, eating unhealthy foods, or attaining them in socially unacceptable ways, can signify lower social status and/or be stigmatizing. Food may be one of the primary ways through which status is signified among youth given that they typically have little control over any significant amount of money. 1.2. What about the Kids? Food insecurity disrupts the family dynamic in ways that may place members of the household at different levels of exposure to stress (Harvey, 2016; Martin & Lippert, 2012). Understanding how children experience this disruption is difficult because most research has relied on reports from adults, or focused on the ways in which household or parental stressors influence children's health outcomes (Bronte-Tinkew, Zaslow, Horowitz, & McNamara, 2007; Casey et al., 2006; Garasky, Stewart, Gundersen, Lohman, & Eisenmann, 2009; Lohman, Stewart, Gundersen, Garasky, & Eisenmann, 2009). For example, the “sacrifice theory” or “buffering hypothesis” suggests that parents alter their own eating practices with the intent to shield their children from food insecurity (Franklin et al., 2012). The sacrifice theory helps makes sense of evidence showing that children living in food insecure households are only insecure themselvesdbased on parent reportsdabout half the time (Nord, Andrews, & Carlson, 2009). Moreover, research suggests that this burden of sacrifice falls largely on mothers due to the gendered expectation that they are the primary caregivers in terms of feeding children (DeVault, 1994; Martin & Lippert, 2012). This research, while valuable, can tell us little

237

about the individual experiences of food insecurity for children within the household. In addition to the sacrifice theory, discrepancies between parent and child levels of food insecurity may be due to differences in food insecurity reporting between parents and children (Bernal, Frongillo, Herrera, & Rivera, 2014; Fram et al., 2013; Harvey, 2016). Possible explanations for this incongruity include both the social desirability of parents to report less or no food insecurity for their children, as this may threaten expectations of their role as parents/mothers and personal identities attached to this role, as well as the fact that parents may simply be unaware of children's experiences due to attempts by both children and parents to protect each other from knowing the full extent of their hardship (Harvey, 2016). Moreover, food insecurity is experienced and responded to differently across the life course (Fram et al., 2011; Nord, 2013; rez-Escamilla, & Kac, 2013). While the sacSchlüssel, Silva, da, Pe rifice theory offers a great deal of insight into how the association between food insecurity and weight status operates among adults, it conceptualizes children as passive, with little to no agency. Few studies have approached the question of how food insecurity might relate to overweight or obese weight status among children by either surveying or interviewing children directly; however, those that have taken this approach have found that youth are anything but passive when experiencing food insecurity (Bernal, Frongillo, Herrera, & Rivera, 2012; Connell, Lofton, Yadrick, & Rehner, 2005; Fram et al., 2011) and that their food practices are deeply tied to social relationships (Neely, Walton, & Stephens, 2014). Overall, this research highlights youth as having distinct experiences from adults (Bernal et al., 2012; Fram et al., 2011), underreporting of child food insecurity by adults (Bernal et al., 2014; Fram et al., 2013), as well as conscious efforts by youth to cope with material hardship (Bernal et al., 2012; Connell et al., 2005; Fram et al., 2011). Additional research suggests that the reports from children capture the adverse effects and correlates of child food insecurity better than reports from parents (Bernal, Frongillo, & Rivera, 2015; Choi, Frongillo, & Fram, 2013; Fitzpatrick & Willis, 2015; Fram et al., 2013). Altogether, four studies have examined the association between food insecurity and obesity among children/adolescent samples close to the age of the youth sample (ages 9e14; Mean ¼ 11.42) in this study. Three of them found no association; however, two of those three used the same sample, which was limited to low income adolescents (ages 10e15), had a surprisingly low rate of food insecurity, and did not survey the youth directly (Gundersen, Garasky, & Lohman, 2009; Gundersen, Lohman, Eisenmann, Garasky, & Stewart, 2008; Lohman et al., 2009). Moreover, the study that did find a significant relationship was limited to homeless adolescents (ages 9e18) in Minnesota (Smith & Richards, 2008). Olson, Bove, and Miller (2007) offer parallel evidence from adults who were asked questions about their childhood experiences with food. Households that were previously food-insecure currently tended to overeat during times when food was most available, in part due to the excitement of the influx of food (Olson et al., 2007). Their results suggest that food deprivation related to poverty in childhood has a lasting impact on eating practices well into adulthood, as well as deep emotional responses to the potential of experiencing food insecurity again. Kaur, Lamb, and Ogden (2015) recently analyzed nationally representative data from the National Health and Nutrition Examination Survey (NHANES) that included individual-level data for children age 2e11. They found no associations between aggregated household food insecurity and child obesity, but did find a significant association among children aged 6e11 using the individuallevel measure. While this solves the issues of food insecurity

238

D.E. Willis, K.M. Fitzpatrick / Appetite 103 (2016) 236e243

measures aggregated to the household level by gathering information for each individual child, the responses for this 5-item food insecurity measure came from adults or older siblings. Choi et al. (2013) also use data from the NHANES, but for children age 12e17 who were able to answer the questionnaire for themselves. The authors, while not reporting on BMI or weight status specifically, did find child-reported food insecurity to be a better predictor of nutritional health. However, the food insecurity measure in the NHANES data excludes non-nutritional dimensions of the concept such as worry and anxiety that other research has shown to be particularly relevant for youth (Fram et al., 2011, 2013; Harvey, 2016). The current study contributes to this extant literature by utilizing five of the nine food security items developed by Connell, Nord, Lofton, and Yadrick (2004) from the original USDA food security module to survey children directly, including an indicator related to the non-nutritional element of anxiety/worry. As mentioned earlier, reviews of the literature have found inconsistent results across groups, specifically among children (Dinour et al., 2007; Franklin et al., 2012). However, none of the studies in these reviews surveyed children directly despite evidence that youth experience food insecurity differently than their parents (Fram et al., 2011; Harvey, 2016), that parents often report less child food insecurity than children themselves (Fram et al. 2013; Harvey, 2016), and previous work that acknowledges this limitation in their analysis (Lohman et al., 2009; Gundersen et al., 2008, 2009). Moreover, the single study that has utilized the food insecurity reports from children to analyze the association between food insecurity and obesity does not include indicators of the nonnutritional elements of food insecurity (Choi et al., 2013). Focusing solely on nutritional components of food insecurity ignores a large piece of what previous research has shown to be central to youth experiences of food insecurity (Connell et al., 2005; Fram et al., 2011, 2013, 2015; Frongillo & Bernal, 2014; Harvey, 2016). By surveying children at the individual level to avoid conflating household food insecurity with child food insecurity, allowing them to report their experiences for themselves, and keeping the indicator of worry/anxiety in our measure of food insecurity to effectively capture what others have shown to be central to youth experiences, we offer a unique contribution to the existing literature that is consistent with multiple suggestions for future research given by previous scholars (Fram et al., 2011; Frongillo 2013; Frongillo & Bernal, 2014; Gundersen et al., 2009; Lohman et al., 2009). Given the empirical findings as well as theoretical insights of previous work, three hypotheses are put forward in this study. 1) Food insecurity and weight status will be positively associated. 2) The association will remain consistent even while controlling for sex, race, ethnicity, and poverty. 3) The association between weight status and food insecurity will be mediated by psychosocial factors. 2. Methods Cross-sectional data were collected from fifth through seventhgrade students in a middle school in Northwest Arkansas. The survey was administered in late September of 2012 with 361 students enrolled at the time. Across fifteen classrooms 334 students completed the surveyda 92 percent completion rate. All fifth, sixth, and seventh grade students were included in the sampling frame. Almost all of the students who did not complete the survey were either absent due to illness or had another form of excused absence. Three students who were present chose not to complete the survey.

The written survey was read aloud in English (to avoid any literacy bias) during the earliest morning period in all eligible classrooms by trained interviewers. A Spanish version of the survey was made available to students who raised their hands when asked if anyone preferred a Spanish copy. These students had their survey read to them in Spanish. Teachers were instructed to avoid helping students with the survey to maintain student privacy and to allow students to feel comfortable answering honestly. Teacher's aides were allowed to help a few special needs students but trained interviewers remained responsible for administering the survey. Students were informed that they were free to withdraw at any point. No personal identifiers were gathered. The University Institutional Review Board reviewed and approved this study, including an informed consent procedure for children under age 14. 2.1. Weight status Weight status was measured by calculating BMI based from students' self-reported height and weight. Self-reported height and weight provide accurate assessments for BMI calculations among student age populations (Goodman, 1999; Kubik, Lytle, & Story, 2005). Weight status categories were determined using these BMI scores and the Centers for Disease Control and Prevention (CDC) charts that provide estimated percentiles based on sex and age (CDC 2015). Students were categorized by the following national standards: (0) underweight (BMI < 5th percentile), (1) normal (BMI 5th-84th percentile), (2) overweight (BMI 85th-94th percentile), and (3) obese (BMI  95th percentile). Since 2007, all Arkansas students (Grades K, 2, 4, 6, 8, 10) have had their height and weight assessed by trained personnel. The state reports these findings in aggregate by grade only; students in grades 4 and 6 in 2011 in the surveyed school reported healthy BMI scores in 65% and 61%, of the students respectively. These same two cohorts would have been 5th and 7th graders in this school at the time of our survey in 2012. These percentages are very close to the percentage of healthy BMI scores reported in the current study (69%), which suggests further that self-reported height and weight among youth can be used to calculate reasonably accurate BMI scores. 2.2. Sociodemographic/control variables Often, there are important differences in life experience across demographic variables such as age, sex, class, race, and ethnicity. Therefore, these factors are generally controlled for in analyses of inequality. However, our data have a limited age variation because we sampled for a specific age group. Moreover, we found that race was not a significant factor in our preliminary analyses and early regression models. Thus, race and age have been excluded from the final analysis. The control variables we have included in our model are sex and ethnicity. Students were also asked, “Are you of Hispanic, Latino, or Spanish origin?” Because of a strong Hispanic/ Latino presence in the Northwest Arkansas region it was important to control for any possible differences between those with Hispanic/Latino origin and those outside of this ethnic group. Similar to race, visible physical differences and cultural practices makes people of certain ethnicity an easy target for discrimination and exclusion. 2.3. Food insecurity Connell et al. (2004) used cognitive interviewing methods to develop a module for assessing food insecurity through direct survey of children based on the original USDA module. Nine items from the original USDA module were deemed appropriate for

D.E. Willis, K.M. Fitzpatrick / Appetite 103 (2016) 236e243

administrating the survey to children. Five of those nine questions make up our measure of food insecuritydwe chose to shorten this due to the time constraints of surveying during a single class period. Additionally, we made a minor modification of the last item wherein we replaced the phrase “been cut” with “been smaller” based on some recommendations from children who we tested the survey on prior to administering it. Following the prompt, “Thinking about your experience with food over the past year,” children were asked; 1) Did you worry that food at home would run out before your family got money to buy more; 2) Did the food that your family bought run out and you didn't have money to get more; 3) How often were you not able to eat a balanced meal because your family didn't have enough money to buy food; 4) Did your meals include a few kinds of cheap foods because your family was running out of money to buy food; 5) Have your meals been smaller because your family didn't have enough money to buy food? Possible responses included “never,” “sometimes,” and “a lot.” These items were coded from 0 to 2 in the order listed, beginning with “never” coded as 0. From these 5 items, a composite food insecurity scale was computed ranging from 0 to 10 (Mean ¼ 1.91; SD ¼ 2.42). The scale was reliable (Cronbach's alpha ¼ 0.84). 2.4. Poverty Eligibility for free and reduced lunch requires a student's family income to fall within 185% poverty. Use of these programs can serve as a proxy for measuring poverty status. Students were asked, “How do you pay for your school lunch?” Possible responses included, “free,” “reduced price,” “parents pay,” “parents pack my lunch,” and “other.” Recoding was done so that all responses other than free or reduced price ¼ 0, and both free and reduced price ¼ 1. Students in our sample showed similar rates of poverty according to this measure as those reported for the school district; 57% and 52% respectively. 2.5. Depression Depressive Symptomatology is typically measured using a shortened version of the original 20-item Center for Epidemiological Studies for Depression (CES-D) Scale. This scale is designed to assess affective depressive symptoms within the two weeks prior to the survey. It has been used extensively in measuring depressive symptoms in both adolescents and adults (Radloff, 1977). The version used in the current study has eight items. Possible responses range from 0 (symptom experienced less than 1 day), 1 (1e2 days), 2 (3e4 days), and 3 (symptom experienced 5e7 days). These questions asked if and to what extent they felt sad, lonely, or fearful, as well as whether they had trouble sleeping, eating, getting along with others, etc., in the last 2 weeks. The properties of this measure have been demonstrated with very high coefficients of internal consistency when the scale is administered to school-age youth (Resnick et al., 1991; Piko & Fitzpatrick, 2003). The shortened CES-D was reliable in this sample of students (Cronbach's alpha ¼ 0.88). The CES-D scores were weighted by a factor of 2.5. This factor was calculated by dividing the number of original CES-D items by the number of items on the shortened version (20/8 ¼ 2.5). Weighting these scores allows for comparison with samples using the complete 20-question CES-D (Resnick et al., 1991; Piko & Fitzpatrick, 2003). 2.6. Social capital To capture students' social capital/resources they were asked, “how many close friends do you have?” (Mean ¼ 5; SD ¼ 3.6). Cues were provided with the question so that students could distinguish

239

between close friends and general friends or acquaintances. No coding was necessary for this measure since their response was numerical. 2.7. Self-esteem Rosenberg's self-esteem index (Rosenberg, 1986) was used as an indicator of psychological resources. This is a 10-item scale. The questions on this scale assess how students perceive themselves, including their sense of self relative to others. The answers range from 1 (strongly agree) to 4 (strongly disagree). Combined, these answers make up the score on the scale. A higher score indicates higher self-esteem. This self-esteem scale was reliable (Cronbach's alpha ¼ 0.82; Mean ¼ 22.4; SD ¼ 4.5). 2.8. Perceived social class/status Parent income, occupation, education, and so on are often difficult for youth to assess accurately. Therefore, a subjective measure of social class was used to gather some insight into how students perceive the economic circumstances of their family. Subjective assessment of SES has been shown to be a reasonable indicator of adolescent's beliefs of their relative status or social deprivation and its relationship to their health (Goodman et al., 2001; Piko & Fitzpatrick, 2001). Students were asked to respond to the prompt, “thinking about the money your family has and the things your family owns, would you think of your family as …” by selecting one of the following categories: upper class (7.5%), between upper and middle class (34%), middle class (37%), between lower and middle class (16.5%), and lower class (5%). Ranging from one to five, higher scores correspond to higher class/status. 2.9. Statistical analyses The analyses exclude cases with missing information for key variables (N ¼ 324). The primary analyses were conducted using SPSS. The focus is on relationships between weight status and what are often thought of as material conditions (food insecurity, poverty) and psychosocial factors (depression, perceived social status, social capital, self-esteem). Descriptive statistics give us some sense of the sample that is being examined, the correlation table indicates which variables in the analysis are significantly associated with each other, and finally, the regression analysis gives us an idea of how these variables relate to weight status while controlling for one another. Moreover, the regression analysis provides some sense of how important these variables are in explaining variation in weight status. Sobel's test (1982) was calculated using a website algorithm developed by Preacher and Leonardelli (2001) to investigate the possibility of mediation by psychosocial factors that were significantly related to weight status in earlier tests. Although we would typically control for standard sociodemographic characteristics (e.g. sex, age, race, and ethnicity) in an analysis like this, some of these characteristics and their relationship to weight status were examined and found to have no significant correlation with the dependent variable. Moreover, this sample is age-specific, restricting the ability of age to act as a meaningful variable in this analysis. Thus, only sex and ethnicity have been kept in the final analysis to offer some meaningful controls while not overloading the model with non-significant factors. Ethnicity was chosen over race because there is a significant Hispanic/Latino presence in this Northwest region of Arkansas. We also tested for a possible interaction of gender and food insecurity on obesity; the interaction was not significant.

240

D.E. Willis, K.M. Fitzpatrick / Appetite 103 (2016) 236e243

3. Results Table 1 presents the descriptive statistics for key variables of interest as well as sample demographics. The average age in our sample was 11 years old. Female and white students made up a slight majority. Around one in five students identified themselves as having Hispanic, Latino, or Spanish ethnicity. The average student in our sample would be considered to have a healthy weight status based on the percentile cutoffs for BMI suggested by the CDC. While over two-thirds of the sample was of a healthy weight, nearly a third would be considered overweight or obese according to the CDC standards. The average student reported a food insecurity score of 1.91 on scale ranging from zero to ten. The standard deviation was 2.42. Nearly three in five students reported receiving free or reduced lunch, meaning that at least this many students live in households with an income at 185% of poverty or lower. On average, students reported having about six close friendsdstandard deviation of 3.67. The weighted CES-D had a mean of 14.74 and a standard deviation of 13.60. Clinical caseness scores for the CES-D are 16; the scores in this sample appear elevated relative to the general youth/young adult population. We would expect an average CES-D score of 8e10 for non-clinical youth populations. The average self-esteem in this sample was 20.71 with a standard deviation of 5.18. Perceived social class had an average of 3.23; in other words, the average student saw their self as “middle class.” Table 2 is a correlations matrix providing insight into the strength, direction, and statistical significance of associations among each variable in the following regression model. The first column is of primary interest since it shows correlations for all controls and independent variables with the dependent variable, weight status. Neither of the control variables, sex and ethnicity, are significantly associated with weight status. Food insecurity, depression, and perceived social status all show significant associations with weight status. Food insecurity and depression both have a weak, positive association with weight status. As these two variables increase or decrease, so does weight status. Perceived social status has a weak, negative association with weight status; as one increases the other decreases. Poverty, social capital, and self-esteem are not significantly correlated with the outcome variable.

Additionally, there are some other significant associations among the independent variables themselvesdfood insecurity (Table 2, column 4) in particular has a number of significant associations with other independent variables. That said, in the final regression model most of the variance inflation factors (VIFs) were only slightly over 1.0, meaning that multicollinearity among the independent variables is not likely an issue within this model. Table 3 shows the results for the multiple regression models for weight status. Model one includes only the control variables. Neither control variable was significantly associated with weight status while controlling for the other nor was this first model significant in its collective influence. Model two introduces the food insecurity and poverty (free and reduced lunch) variables into the equation. This model is significant at the p < 0.05 level, suggesting that these variables have an important collective influence on weight status. This model also suggests that food insecurity has a significant, and positive, individual association with weight status even while controlling for sex, ethnicity, and poverty. Moreover, the addition of these two variables shifts R2 upward from the first model; model two accounts for more variation in weight status. Model three introduces our final set of independent variables, three psychosocial resources (social capital, self-esteem, and perceived social status) as well as one indicator of psychosocial risk (depression). Like model two, model three is significant at the p < 0.05 level. The most noteworthy shift in this final model is that food insecurity is no longer significant; only depression and perceived social status are significant in the presence of the other independent variables in this model. Furthermore, the unstandardized regression coefficient for food insecurity was reduced from 0.055 to 0.023 when the mediators were added, meaning that the mediating paths accounted for more than half of the magnitude of association of food insecurity with weight status. This supports the hypothesis that the food insecurity-obesity association is mediated by the psychosocial factors introduced in this final model. Additionally, R2 increased again in the final model. The final model explains 5.4% of the variation in weight status. The limited amount of variation explained is compensated for by both parsimony and elucidation of possible mediating variables. Sobel's test was calculated to further investigate the possibility of mediation of the relationship by the significant psychosocial factors (social status and depression). This test utilizes the

Table 1 Descriptive Statistics and Sample Demographic Characteristics (N ¼ 324). % Dependent Variable Weight Status (0e3) - Underweight (0) - Healthy Weight (1) - Overweight (2) - Obese (3) Demographic Characteristics Age Sex (1 ¼ Female) Race (1 ¼ White) Ethnicity (1 ¼ Hispanic)

Mean

S.D.

e 6.00% 64.50% 14.50% 15.00%

1.39

0.813

e 53.60% 52.10% 20.70%

11.42 e e e

0.917 e e e

Access and Poverty Food Insecurity (0e10) Poverty (1 ¼ Free or Reduced Lunch)

e 57.20%

1.91 e

2.42 e

Psychosocial Resources & Risk Social Capital (Close Friends) Depression (CES-D) Self-Esteem (1e40) Perceived Social Class (1e5)

e e e e

5.96 14.74 20.71 3.23

3.67 13.60 5.18 0.974

Original to this manuscript. Weight status calculations are based on BMI percentile cutoffs provided by the CDC.

D.E. Willis, K.M. Fitzpatrick / Appetite 103 (2016) 236e243

241

Table 2 Correlations among model variables (N ¼ 324).

1 2 3 4 5 6 7 8 9

Weight Status Sex (Female) Ethnicity (Hispanic) Food Insecurity Poverty Social Capital Self-esteem Depression Perceived Social Status

1

2

3

4

5

6

7

8

0.061 0.029 0.169*** 0.039 0.086 0.020 0.151** 0.158**

0.012 0.057 0.038 0.125* 0.029 0.121* 0.056

0.031 0.255*** 0.071 0.001 0.002 0.015

0.260*** 0.151** 0.143** 0.407*** 0.429***

0.095* 0.139** 0.050 0.174***

0.047 0.096* 0.062

0.172*** 0.127*

0.192***

Original to this manuscript. Weight status calculations are based on BMI percentile cutoffs provided by the CDC.

Table 3 Regressions for weight status among middle school students (N ¼ 324). Variables Controls Sex (1 ¼ Female) Ethnicity (1 ¼ Hispanic) Insecurity and Poverty Food Insecurity (1e10) Free and Reduced Lunch (1 ¼ Receiving) Psychosocial Resources and Risks Social Capital (Close Friends) Self-esteem (1e40) Depressive Symptoms (CES-D) Perceived Social Status (1e5) Constant Df R2

Model 1 b (b)

Model 2 b (b)

Model 3 b (b)

0.100 (0.061) 0.062 (0.030)

0.084 (0.052) 0.051 (0.025)

0.114 (0.070) 0.053 (0.026)

0.055**(0.165) 0.001 (0.000)

0.023 (0.068) 0.002 (0.001) 0.011 (0.049) 0.004 (0.023) 0.007 (0.109)* 0.095 (0.112)*

1.46 2 0.005

1.34 2 0.032*

1.77 4 0.054*

p < 0.05*; p < 0.01**; p < 0.001*** (Hierarchical F-test R2 Change). Original to this manuscript. Weight status calculations are based on BMI percentile cutoffs provided by the CDC.

unstandardized regression coefficient (a) and standard error (sa) for the association between the mediator and the independent variable, as well as the unstandardized regression coefficient (b) and standard error (sb) for the association between the mediator and the dependent variable, to calculate a Sobel statistic and a pvalue to indicate significance. The equation for this test is as follows: z-value ¼ a*b/SQRT(b2*sa2 þ a2*sb2) (Preacher & Leonardelli, 2001; Sobel, 1982). To test for mediation by social status we input (a) 0.170, (Sa) 0.020, (b) 0.094, and (Sb) 0.050 into the Sobel equation. This results in a Sobel statistic of 1.84 with a one-tailed significance of p < 0.05. Thus, social status significantly mediates the association between food insecurity and obesity. To test for mediation by depression (CES-D) we input (a) 0.901, (Sa) 0.113, (b) 0.015, and (Sb) 0.009 into the Sobel equation. This results in a Sobel statistic of 1.63 (one-tailed, p < 0.10). Although the one-tailed p-value (p ¼ 0.051) just barely misses the standard cutoff for significance, this remains a notable finding. Depression approaches significant mediation of the food insecurity-obesity association at the p < 0.05 level, and surpasses significance at the p < 0.10 level. 3.1. Summary of findings We find support for all three proposed hypotheses. First, there is a significant direct association between food insecurity and weight status among early adolescents in this sample; as food insecurity increases, so does weight status. Second, this direct association remains consistent even when controlling for sex, poverty, and ethnicity. Furthermore, additional analyses not reported here (available upon request) also suggest that the relationship remains

significant even when controlling for race. Third, due to the shift to non-significance by food insecurity in the final model and significant results from the Sobel tests, our analyses suggest that psychosocial factors such as perceived social status and depression are important mediators of the food insecurity-obesity association. In other words, food insecurity may be operating partially through non-nutritional, psychosocial factors (i.e. social status and depression) to impact weight status. 4. Discussion Overall, the findings of this study provide further evidence of a significant relationship between food insecurity and obesity among youthdspecifically middle-school age (9e14) youthdwhich is mediated by multiple psychosocial factors. This finding is similar to what has been found by Smith and Richards (2008) Casey et al. (2006) and Olson et al., (2007); yet, it also runs counter to the findings of Gundersen et al. (2009, 2008), and Lohman et al. (2009) who have concluded there is no relationship among children of a similar age. However, the present study is distinct from these studies due to the use of data collected from children themselves. Our findings also build on this work by further elaborating the role of psychosocial factors as mediators and taking the suggestion of previous scholars to include food insecurity reports by children themselvesda limitation recognized in almost all of the previous research. The findings complement previous research (e.g. Fram et al., 2011) that has framed food insecurity as more than a material condition with solely nutritional risks, but also a form of social status that can carry with it a number of psychosocial stressors that are often injurious to health. The influence of material living

242

D.E. Willis, K.M. Fitzpatrick / Appetite 103 (2016) 236e243

conditions must be understood to have an important psychosocial dimension; therefore, the task of service providers becomes more complicateddthey need not only provide food and basic resources, but they must do so in a way that does not injure the human dignity of those being served. The respect they give to those they serve may be just as important as the material resources they provide. Such approaches must be fundamental to the overall mission of advocates and service-providers. Providing just food may alleviate food insecurity or hunger but leave psychosocial risks to continue harming both the mental and physical health of those served. Thus, this may be an important area for mental health professionals and nutrition specialists or food pantries to work together for a more holistic approach that addresses both material and psychosocial dimensions of food insecurity. 4.1. Limitations and future research This study has taken up the task of explaining the association of food insecurity and weight status among children utilizing a psychosocial framework. While this framework contains a number of strengths for answering this particular question, it is also recognized that it is not without weaknesses. Psychosocial theory directs attention towards stress and gives considerably less attention to “who and what generates psychosocial insults and buffers” and “how their distributiondalong with that of ubiquitous or nonubiquitous pathogenic physical, chemical, or biological agentsdis shaped by social, political, and economic policies” (Krieger, 2001). While a number of scholars have already looked at some of the more upstream factors that determine access to foodda good deal of which is related to ongoing segregation and institutional racism (Alkon & Norgaard, 2009; Morland, Wing, Diez Roux, & Poole, 2002)dfuture research should connect these structural forces more closely to the psychosocial factors that previous work has linked closely to health outcomes (e.g. weight status) (Kennedy, Kawachi, Lecher, Jones, & Prothrow-Stith, 1996; Fitzpatrick, Willis, & O'Connor, 2014; Fitzpatrick & Willis, 2015) in order to provide a fuller view of the health equation as it relates to food insecurity. Clearly, there are additional factors which might be of importance to this association that are not captured in our data. For example, the prevalence of children of immigrants is unknown. Moreover, as our data are cross-sectional, this study is not equipped to make claims of causality. It is quite possible, for example, that children are more likely to experience depression due to their weight status. As Franklin et al. (2012) points out, most of the studies that have taken a longitudinal approach have not found confirmation of the notion that food insecurity increases weight status among adults or children (e.g. Jones & Frongillo, 2006; Olson & Strawderman, 2008; Whitaker & Sarin, 2007). However, each of these longitudinal studies were focused on adult women rather than children. Jyoti et al. (2005) analyzed longitudinal data of children and found a significant relationship between household food insecurity and weight gain among girls. Yet, the analysis is limited by its reliance upon household aggregates and parent reports of food insecurity. Thus, future research should endeavor to conduct both long-term analyses as well as collect measures of food insecurity from children themselves. Acknowledgements The authors would like to thank the Harvey and Bernice Jones Trust which provided funding for this research. References Alaimo, K., Olson, C. M., & Frongillo, E. A. (2001). Food insufficiency and American

school-aged children's cognitive, academic, and psychosocial development. Pediatrics, 108(1), 44e53. Alaimo, K., Olson, C. M., & Frongillo, E. A. (2001). Low family income and food insufficiency in relation to overweight in U.S. children: is there a paradox? Archives of Pediatrics & Adolescent Medicine, 155(10), 1161e1167. Alaimo, K., Olson, C. M., & Frongillo, E. A. (2002). Family food insufficiency, but not low family income, is positively associated with dysthymia and suicide symptoms in adolescents. The Journal of Nutrition, 132(4), 719e725. Alkon, A. H., & Norgaard, K. M. (2009). Breaking the food chains: an investigation of food justice activism. Sociological Inquiry, 79(3), 289e305. Bernal, J., Frongillo, E. A., Herrera, H., & Rivera, J. (2012). Children live, feel, and respond to experiences of food insecurity that compromise their development and weight status in peri-urban Venezuela. The Journal of Nutrition, 142(7), 1343e1349. http://dx.doi.org/10.3945/jn.112.158063. Bernal, J., Frongillo, E. A., Herrera, H. A., & Rivera, J. A. (2014). Food insecurity in children but not in their mothers is associated with altered activities, school absenteeism, and stunting. The Journal of Nutrition, 144(10), 1619e1626. http:// dx.doi.org/10.3945/jn.113.189985. Bernal, J., Frongillo, E. A., & Rivera, J. A. (2015). Food insecurity reported by children, but not by mothers, is associated with lower quality of diet and shifts in foods consumed. Maternal & Child Nutrition. http://dx.doi.org/10.1111/mcn.12206. n/ aen/a. Bourdieu, P. (1984). The forms of capital. In Readings in economic sociology. Bronte-Tinkew, J., Zaslow, M., C., R., Horowitz, A., & McNamara, M. (2007). Food insecurity works through depression, parenting, and infant feeding to influence overweight and health in toddlers. The Journal of Nutrition, 137(9), 2160e2165. Casey, P. H., Simpson, P. M., Gossett, J. M., Bogle, M. L., Champagne, C. M., Connell, C., et al. (2006). The association of child and household food insecurity with childhood overweight status. Pediatrics, 118(5), e1406ee1413. Chilton, M., & Knowles, M. (2014). Childhood stress: a qualitative analysis of the intergenerational circumstances of child hunger. University of Kentucky Center for Poverty Research Discussion Paper Series. Retrieved from http:// uknowledge.uky.edu/ukcpr_papers/40. Chilton, M., Knowles, M., Rabinowich, J., & Arnold, K. T. (2015). The relationship between childhood adversity and food insecurity: “It's like a bird nesting in your head.” Public Health Nutrition, 18(14), 2643e2653. http://dx.doi.org/ 10.1017/S1368980014003036. Choi, S. K., Frongillo, E. A., & Fram, M. S. (2013). Prediction of child health and nutrient intake by child versus parent reports of child food insecurity. The FASEB Journal, 27 (1_MeetingAbstracts), 369.4. Coleman-Jensen, A., Gregory, C., & Singh, A. (2014). Household food security in the United States in 2013. ERR-173. U.S. Department of Agriculture, Economic Research Service. Connell, C. L., Lofton, K. L., Yadrick, K., & Rehner, T. A. (2005). Children's experiences of food insecurity can assist in understanding its effect on their well-being. The Journal of Nutrition, 135(7), 1683e1690. Connell, C. L., Nord, M., Lofton, K. L., & Yadrick, K. (2004). Food security of older children can be assessed using a standardized survey instrument. The Journal of Nutrition, 134(10), 2566e2572. DeVault, M. L. (1994). Feeding the family: The social organization of caring as gendered work. University of Chicago Press. Dinour, L. M., Bergen, D., & Yeh, M. C. (2007). The food insecurityeobesity paradox: a review of the literature and the role food stamps may play. Journal of the American Dietetic Association, 107(11), 1952e1961. Fitzpatrick, K. M., & Willis, D. (2015). Parent/student risk and protective factors in understanding early adolescents' body mass index. The Journal of Early Adolescence. http://dx.doi.org/10.1177/0272431615570058, 0272431615570058. Fitzpatrick, K. M., Willis, D., & O'Connor, G. (2014). Circumstances, resources, and weight status outcomes among middle school students. The Journal of Early Adolescence, 34(8), 1058e1074. http://dx.doi.org/10.1177/0272431613518974. Fram, M. S., Frongillo, E. A., Draper, C. L., & Fishbein, E. M. (2013). Development and validation of a child report assessment of child food insecurity and comparison to parent report assessment. Journal of Hunger & Environmental Nutrition, 8(2), 128e145. http://dx.doi.org/10.1080/19320248.2013.790775. Fram, M. S., Frongillo, E. A., Jones, S. J., Williams, R. C., Burke, M. P., DeLoach, K. P., et al. (2011). Children are aware of food insecurity and take responsibility for managing food resources. The Journal of Nutrition, 141(6), 1114e1119. Fram, M. S., Ritchie, L. D., Rosen, N., & Frongillo, E. A. (2015). Child experience of food insecurity is associated with child diet and physical activity. The Journal of Nutrition, 145(3), 499e504. http://dx.doi.org/10.3945/jn.114.194365. Franklin, B., Jones, A., Love, D., Puckett, S., Macklin, J., & White-Means, S. (2012). Exploring mediators of food insecurity and obesity: a review of recent literature. Journal of Community Health, 37(1), 253e264. Freedman, D. S., Mei, Z., Srinivasan, S. R., Berenson, G. S., & Dietz, W. H. (2007). Cardiovascular risk factors and excess adiposity among overweight children and adolescents: the Bogalusa heart study. The Journal of Pediatrics, 150(1), 12e17. Frongillo, E. A. (2013). Confronting myths about household food insecurity and excess weight. Cadernos De Saúde Pública, 29(2), 229e230. Frongillo, E. A., & Bernal, J. (2014). Understanding the coexistence of food insecurity and obesity. Current Pediatrics Reports, 2(4), 284e290. http://dx.doi.org/ 10.1007/s40124-014-0056-6. Garasky, S., Stewart, S. D., Gundersen, C., Lohman, B. J., & Eisenmann, J. C. (2009). Family stressors and child obesity. Social Science Research, 38(4), 755e766. Goodman, E. (1999). The role of socioeconomic status gradients in explaining

D.E. Willis, K.M. Fitzpatrick / Appetite 103 (2016) 236e243 differences in US adolescents' health. American Journal of Public Health, 89(10), 1522e1528. http://dx.doi.org/10.2105/AJPH.89.10.1522. Goodman, E., Adler, N. E., Kawachi, I., Frazier, A. L., Huang, B., & Colditz, G. A. (2001). Adolescents' perceptions of social status: development and evaluation of a new indicator. Pediatrics, 108(2). http://dx.doi.org/10.1542/peds.108.2.e31. e31ee31. Gundersen, C., Garasky, S., & Lohman, B. J. (2009). Food insecurity is not associated with childhood obesity as assessed using multiple measures of obesity. The Journal of Nutrition, 139(6), 1173e1178. Gundersen, C., Lohman, B. J., Eisenmann, J. C., Garasky, S., & Stewart, S. D. (2008). Child-specific food insecurity and overweight are not associated in a sample of 10- to 15-year-old low-income youth. The Journal of Nutrition, 138(2), 371e378. Harvey, K. (2016). “When I go to bed hungry and sleep, I'm not hungry”: children and parents' experiences of food insecurity. Appetite. http://dx.doi.org/10.1016/ j.appet.2016.01.004. Jones, S. J., & Frongillo, E. A. (2006). The modifying effects of food stamp program participation on the relation between food insecurity and weight change in women. The Journal of Nutrition, 136(4), 1091e1094. Jyoti, D. F., Frongillo, E. A., & Jones, S. J. (2005). Food insecurity affects school children's academic performance, weight gain, and social skills. The Journal of Nutrition, 135(12), 2831e2839. Kaur, J., Lamb, M. M., & Ogden, C. L. (2015). The association between food insecurity and obesity in children-the National Health and Nutrition Examination Survey. Journal of the Academy of Nutrition and Dietetics, 115(5), 751e758. http:// dx.doi.org/10.1016/j.jand.2015.01.003. Kennedy, B. P., Kawachi, I., Lecher, K., Jones, C., & Prothrow-Stith, D. (1996). (Dis) Respect and black mortality. Ethnicity & Disease, 7(3), 207e214. Krieger, N. (2001). Theories for social epidemiology in the 21st century: an ecosocial perspective. International Journal of Epidemiology, 30(4), 668e677. http:// dx.doi.org/10.1093/ije/30.4.668. Kubik, M. Y., Lytle, L. A., & Story, M. (2005). Schoolwide food practices are associated with body mass index in middle school students. Archives of Pediatrics & Adolescent Medicine, 159(12), 1111e1114. http://dx.doi.org/10.1001/ archpedi.159.12.1111. Larson, N. I., & Story, M. T. (2011). Food insecurity and weight status among U.S. children and families: a review of the literature. American Journal of Preventive Medicine, 40(2), 166e173. http://dx.doi.org/10.1016/j.amepre.2010.10.028. Lohman, B. J., Stewart, S., Gundersen, C., Garasky, S., & Eisenmann, J. C. (2009). Adolescent overweight and obesity: links to food insecurity and individual, maternal, and family stressors. Journal of Adolescent Health, 45(3), 230e237. Martin, M. A., & Lippert, A. M. (2012). Feeding her children, but risking her health: the intersection of gender, household food insecurity and obesity. Social Science & Medicine, 74(11), 1754e1764. McEwen, B. S. (1998). Stress, adaptation, and disease: allostasis and allostatic load. Annals of the New York Academy of Sciences, 840(1), 33e44. McIntyre, L., Williams, J. V. A., Lavorato, D. H., & Patten, S. (2013). Depression and suicide ideation in late adolescence and early adulthood are an outcome of child hunger. Journal of Affective Disorders, 150(1), 123e129. http://dx.doi.org/ 10.1016/j.jad.2012.11.029. Morland, K., Wing, S., Diez Roux, A., & Poole, C. (2002). Neighborhood Characteristics associated with the location of food stores and food service places. American Journal of Preventive Medicine, 22(1), 23e29. http://dx.doi.org/10.1016/ S0749-3797(01)00403-2. Nackers, L. M., & Appelhans, B. M. (2013). Food insecurity is linked to a food environment promoting obesity in households with children. Journal of Nutrition Education and Behavior, 45(6), 780e784. Neely, E., Walton, M., & Stephens, C. (2014). Young people's food practices and social relationships. A thematic synthesis. Appetite, 82, 50e60. http://dx.doi.org/

243

10.1016/j.appet.2014.07.005. Nord, M. (2013). Youth are less likely to be food insecure than adults in the same household. Journal of Hunger & Environmental Nutrition, 8(2), 146e163. http:// dx.doi.org/10.1080/19320248.2013.786667. Nord, M., Andrews, M., & Carlson, S. (2009). Household food security in the US, 2008. Washington DC: US Department of Agriculture. Economic Research Service. Ogden, C. L., Carroll, M. D., Kit, B. K., & Flegal, K. M. (2012). Prevalence of obesity and trends in body mass index among U.S. children and adolescents, 1999-2010. Journal of the American Medical Association, 307(5), 483e490. http://dx.doi.org/ 10.1001/jama.2012.40. Olson, C. M., Bove, C. F., & Miller, E. O. (2007). Growing up poor: long-term implications for eating patterns and body weight. Appetite, 49(1), 198e207. Olson, C. M., & Strawderman, M. S. (2008). The relationship between food insecurity and obesity in rural childbearing women. The Journal of Rural Health, 24(1), 60e66. http://dx.doi.org/10.1111/j.1748-0361.2008.00138.x. Pearlin, L. I., Menaghan, E. G., Lieberman, M. A., & Mullan, J. T. (1981). The stress process. Journal of Health and Social Behavior, 22(4), 337e356. http://dx.doi.org/ 10.2307/2136676. Pearlin, L. I., Schieman, S., Fazio, E. M., & Meersman, S. C. (2005). Stress, health, and the life course: some conceptual perspectives. Journal of Health and Social Behavior, 46(2), 205e219. http://dx.doi.org/10.1177/002214650504600206. Piko, B., & Fitzpatrick, K. M. (2001). Does class matter? SES and psychosocial health among Hungarian adolescents. Social Science & Medicine, 53(6), 817e830. http://dx.doi.org/10.1016/S0277-9536(00)00379-8. Piko, B., & Fitzpatrick, K. M. (2003). Gender differences in depression among adolescents: a risk and protective factors approach. American Journal of Orthopsychiatry, 73, 44e54. Preacher, K. J., & Leonardelli, G. J. (2001). Calculation for the Sobel test: an interactive calculation tool for mediation tests. Retrieved July 2015 from http://quantpsy. org/sobel/sobel.htm. Radloff, L. S. (1977). The CES-D Scale: a self-report depression scale for research in the general population. Applied Psychological Measurement, 1(3), 385e401. http://dx.doi.org/10.1177/014662167700100306. Resnick, M. D., Bearman, P. S., Blum, R. W., Bauman, K. E., Harris, K. M., Jones, J., et al. (1991). Protecting adolescents from harm: findings from the national longitudinal study of adolescent health. The Journal of the American Medical Association, 278, 823e832. Rosenberg, M. (1986). Self-concept from middle childhood through adolescence. Psychological Perspectives on the Self, 3(1), 107e136. rez-Escamilla, R., & Kac, G. (2013). Household Schlüssel, M. M., Silva, A. A., da, M., Pe food insecurity and excess weight/obesity among Brazilian women and children: a life-course approach. Cadernos de Saúde Pública, 29(2), 219e226. http:// dx.doi.org/10.1590/S0102-311X2013000200003. Schwimmer, J. B., Burwinkle, T. M., & Varni, J. W. (2003). Health-related quality of life of severely obese children and adolescents. Journal of the American Medical Association, 289(14), 1813e1819. Smith, C., & Richards, R. (2008). Dietary intake, overweight status, and perceptions of food insecurity among homeless Minnesotan youth. American Journal of Human Biology, 20(5), 550e563. http://dx.doi.org/10.1002/ajhb.20780. Sobel, M. E. (1982). Asymptotic intervals for indirect effects in structural equations models. In S. Leinhart (Ed.), Sociological methodology (pp. 290e312). San Francisco: Jossey-Bass. Whitaker, R. C., & Sarin, A. (2007). Change in food security status and change in weight are not associated in urban women with preschool children. The Journal of Nutrition, 137(9), 2134e2139. Wilkinson, R. G. (2005). The impact of inequality: How to make sick societies healthier. The New Press.