The empirical evidence highlights US veterans (hereafter, referred to as ‘veterans’) as an at-risk population for being overweight or obese( Reference Koepsell, Littman and Forsberg 1 , Reference Kress, Hartzel and Peterson 2 ). A recent study noted that approximately 37 and 33 % of women and men veterans are obese, respectively( Reference Das, Kinsinger and Yancy 3 ), while others demonstrated higher prevalence of overweight status( Reference Koepsell, Forsberg and Littman 4 ) and greater waist circumference among veterans( Reference Koepsell, Littman and Forsberg 1 ) as compared with the civilian population. Similarly, recent studies have demonstrated increased weight gain among veterans post military discharge( Reference Koepsell, Littman and Forsberg 1 ), thus increasing their long-term weight-related health complications( Reference Kopelman 5 ).
One potential driving factor for such prevalence of overweight and obesity among veterans may be their dietary practices. For instance, in a study among veterans with type 2 diabetes in Washington State, Nelson and co-workers demonstrated that 42 % of the study population reported consumption of a high-fat diet( Reference Nelson, McFarland and Reiber 6 ). Similarly, in another study utilizing the 2000 Behavioral Risk Factor Surveillance System, Wang et al. noted that 40 % of the study sample reported change to a low-energy and low-fat diet in order to lose weight( Reference Wang, Kinsinger and Kahwati 7 ), demonstrating a low adherence to healthy dietary practices in the veteran population.
In recent years, food security, defined by the US Department of Agriculture as ‘access by all people at all times to enough food for active, healthy life’( 8 ), has been highlighted as a significant contributor to population health, including diet. For example, a study among Inuit communities demonstrated that household food insecurity was related to lower scores on the Healthy Eating Index, lower consumption of fruits, vegetables, dairy products and grains, and higher intake of energy from high-sugar foods( Reference Huet, Rosol and Egeland 9 ). Similarly, food insecurity has been associated with lower nutrient intakes among women( Reference Tarasuk and Beaton 10 ) and increased consumption of fruit juices among low-income adults( Reference Mello, Gans and Risica 11 ).
Although current evidence on food insecurity and the putative impact on diet among veterans is limited, a recent study found 24 % of veterans in the Veterans Aging Cohort Study as food insecure( Reference Wang, McGinnis and Goulet 12 ), while another study among veterans who served in the Iraq and Afghanistan wars noted 25 % had low food security and 12 % had very low food security in the past year( Reference Widome, Jensen and Bangerter 13 ). In the present study we aimed to address such a gap in the literature and thus our objective was to utilize data from a population-based survey to determine the relationship between sociodemographic characteristics, food insecurity and dietary practices among veterans.
Methods
Data source
We utilized the 2009, 2011/2012 California Health Interview Survey (CHIS). The CHIS, a biennial survey with self-reported data, is considered the largest state health survey. The CHIS utilizes a random digit-dial system (including both landlines and cell phones), with further methodological details found elsewhere( 14 ). In the current study, respondents who reported serving on active duty in the US Armed Forces (and for those whom length of stay in the military could be ascertained) were defined as veterans.
Measures
The outcome variables in the present study were past week consumption of fast foods, soda, fruits and vegetables. The primary independent variable was food insecurity, defined by CHIS-provided variables: living at or above 200 % of the federal poverty level (FPL); living at less than 200 % of the FPL and food secure; and living at less than 200 % of the FPL and food insecure. The CHIS asked the following questions on food insecurity to those living below 200 % of the FPL or reporting ‘unknown’ to poverty level and created a food security variable, which was utilized in the present study: (i) ‘The food that (I/we) bought just didn’t last, and (I/we) didn’t have money to get more’; (ii) ‘(I/We) couldn’t afford to eat balanced meals’; (iii) ‘In the last 12 months, did you or other adults in your household ever cut the size of your meals or skip meals because there wasn’t enough money for food? How often did this happen? Almost every month, some months but not every month, or only in 1 or 2 months?’; (iv) ‘In the last 12 months, did you ever eat less than you felt you should because there wasn’t enough money to buy food?’; and finally (v) ‘In the last 12 months, were you ever hungry but didn’t eat because you couldn’t afford enough food?’
Covariates for the study included Supplemental Nutrition Assistance Program (SNAP) participation (yes, no), age (18–24 years, 25–44 years, 45–64 years, 65 years or older), sex (male, female), race/ethnicity (African American, Asian American/Pacific Islander, Hispanic, Non-Hispanic White, other), marital status (currently married, not currently married), education level (high school or less, some college/vocational/associate degree, bachelor’s or more), employment status (currently unemployed, currently employed), risk behaviours (smoke and binge drink, smoke or binge drink, none), weight status (overweight or obese, not overweight or obese), general health status (fair/poor, excellent/very good/good), geographic setting (rural, urban) and year (2009, 2011). Overweight or obese was based on a BMI of 25·0 kg/m2 ( 15 ). Risk behaviours were identified as being a current smoker and/or binge drinking in the past 12 months.
Data analysis
We utilized the statistical software package Stata version 14 for all statistical analyses. With the exception of non-parametric tests, all analyses were survey-weighted and α level less than 0·05 was used to denote significance. To determine the distribution of each dietary practice by each population characteristic, we conducted bivariate analyses using the non-parametric Kruskal–Wallis test for differences in medians with ties. If a significant difference was detected between three or more groups, a post hoc multiple comparisons test with Dunn’s pairwise comparisons using the Holm–Sidak adjustment was utilized. Cuzick’s non-parametric test for trend across ordered groups was conducted to assess for trend in dietary practices across poverty and food security categories. Finally, we used negative binomial regression to model the association between rate of dietary practices with poverty and food security after adjusting for control variables.
Results
Our study included a total sample size of 11 011 veterans, reflecting an extrapolated average population estimate of 2 234 186. In the study population, 4·89 % of veterans were below 200 % of the FPL and food insecure, and 2·35 % reported participating in SNAP. In addition, 71·13 % of the population was overweight or obese with additional characteristics displayed in Table 1.
Table 1 Characteristics of the population of US veterans (n 11 011; N 2 234 186), California Health Interview Survey (2009, 2011/2012)
FPL, federal poverty level; SNAP, Supplemental Nutrition Assistance Program.
As shown in Table 2, veterans living in poverty and with food insecurity reported the highest mean intakes of fast foods and soda while reporting the lowest mean intakes of fruits and vegetables. Significant differences in intake of such dietary outcomes by each poverty and food security level were further noted (Kruskal–Wallis test for differences in median, P<0·0001; P for trend <0·05).
Table 2 Mean intakes of fast foods, soda, fruits and vegetables (servings/week), by poverty and food security level, among the population of US veterans, California Health Interview Survey (2009, 2011/2012)
FPL, federal poverty level.
As displayed in Table 3, those who lived below 200 % of the FPL and were food insecure reported, on average, 24 and 142 % higher mean intake of fast foods and soda in the past week, respectively, compared with those living at or above 200 % of the FPL. Living in poverty and being food insecure, as shown in Table 4, was also significantly associated with reporting an average of 24 % lower mean fruit intake in the past week, in comparison to living at or above 200 % of the FPL. Living in poverty (below 200 % of the FPL) but being food secure did not yield a significant association with any of the dietary behaviours assessed.
Table 3 Rate ratio (RR) and 95 % confidence interval of fast food and soda intake in the past week among the population of US veterans, California Health Interview Survey (2009, 2011/2012)
FPL, federal poverty level; SNAP, Supplemental Nutrition Assistance Program; Ref., reference category.
*P<0·05, **P<0·005, ***P<0·0001.
Table 4 Rate ratio (RR) and 95 % confidence interval of fruit and vegetable intake in the past week among the population of US veterans, California Health Interview Survey (2009, 2011/2012)
FPL, federal poverty level; SNAP, Supplemental Nutrition Assistance Program; Ref., reference category.
*P<0·05, **P<0·005, ***P<0·0001.
Our results further showed several factors to be significantly associated with past week intakes of fast foods, soda, fruits and vegetables (Tables 3 and 4). For example, those who participated in SNAP had on average 31 % lower intake of fruits than those who did not. For both fast food and soda consumption, decreasing age was significantly associated with an increasing mean intake of such dietary items, with the highest increase noted among those aged 18–24 years. Similarly, males, as compared with females, reported on average higher intakes of fast foods and soda but lower intakes of fruits and vegetables. Decreasing education level was also associated with increasing average intake of both fast foods and soda, and decreasing intake of both fruits and vegetables, while being unemployed was associated only with a 16 % decrease in average consumption of fast foods.
When evaluating dietary practices by ethnic group, as compared with Whites, Latinos had on average 19 and 34 % lower consumption of fruits and vegetables, respectively, while Asian American/Pacific Islanders had a 14 % lower intake of vegetables. African Americans, as compared with Whites, further had 36 % higher and 9 % lower soda and vegetable intake, respectively.
Both fair/poor general health status and overweight/obesity status yielded significant results only for soda intake, although in the opposite direction. For instance, overweight/obese individuals, as compared with those who were not, reported on average 17 % lower consumption of soda. On the other hand, those with fair/poor health status on average had 32 % higher intake of soda v. those who reported good to excellent health status. A negative role of risk behaviours was also found, with those reporting being a current smoker and/or binge drinking in the past 12 months having lower fruit and vegetable consumption while having a higher intake of soda. Study participants residing in rural areas on average consumed 25 % less fast food per week.
Discussion
Despite the high prevalence of CVD and clinical risk factors, such as obesity, among veterans( Reference Koepsell, Littman and Forsberg 1 , Reference Kress, Hartzel and Peterson 2 ), few studies have accounted for barriers to healthy dietary practices in this population. In the present study we addressed this gap in the literature, with further emphasis on the role of food insecurity on fast food, soda, fruit and vegetable consumption among veterans in California utilizing a population-based survey.
In our study, a major finding was the significant inverse relationship of food insecurity with fruit intake and the positive association with fast food and soda intakes. Such results are comparable to studies among other vulnerable populations. For example, in a study among 1874 low-income adults, Mello and co-workers noted that food-insecure participants were more likely to have high fat intake, as well as juice intake( Reference Mello, Gans and Risica 11 ). Similarly, in a study among rural women, food insecurity was significantly related to lower consumption of fruits and vegetables( Reference Kendall, Olson and Frongillo 16 ), while among food pantry participants, food security was related to increased fruit, vegetable and fibre intakes( Reference Robaina and Martin 17 ). Additionally, economic analysis has demonstrated that the low cost of high-fat and high-energy-dense foods could further be driving vulnerable populations away from healthier items that usually are more expensive( Reference Drewnowski and Darmon 18 – Reference Jetter and Cassady 20 ).
While similar studies among veterans remain limited, upon analysis of interviews with sixty-four veterans Smith and colleagues noted that eating behaviours adapted during food insecurity at the time of military service persisted post service( Reference Smith, Klosterbuer and Levine 21 ), and such emergent themes could further explain our results. For example, the authors reported that eating anything available was common during times of food insecurity during military service, as well as eating fast, binge eating and hoarding; all demonstrating a trend towards poor eating behaviour. In addition, veterans reported a preference for specific food items, such as burgers and fries, considered to be status foods due to putatively low access during deployment, and such diets high in fat and carbohydrates during military service persisted post service( Reference Smith, Klosterbuer and Levine 21 ). As such, the negative dietary behaviours noted among food-insecure veterans in our study are comparable to Smith et al.’s qualitative analyses, although our study adds quantitative data to the literature on specific dietary practices of consumption of fast foods, soda, fruits and vegetables and the relationship to food insecurity among such a vulnerable population.
Cumulatively, the literature and our results on the association between food insecurity and negative dietary practices among veterans highlight the imperative need for health promotion measures focused on a healthy diet in this population, especially those with limited access to healthy food options. Empirical evidence has demonstrated the effectiveness of price reduction of healthy foods( Reference French 22 ), faith-placed interventions and point-of-purchase incentives( Reference Glanz and Yaroch 23 ), and educational( Reference Marcus, Heimendinger and Wolfe 24 ) and voucher options( Reference Anderson, Bybee and Brown 25 ) to increase purchase and consumption of such items. For example, a faith-placed intervention incorporating motivational interviewing techniques was shown to increase intakes of fruits and vegetables among African Americans( Reference Resnicow, Jackson and Wang 26 ), while peer education worksite interventions have been shown to be effective in increasing healthy dietary behaviours among low-income adults( Reference Buller, Morrill and Taren 27 ). Likewise, among women participating in government assistance programmes, vouchers for farmers’ markets have been shown to significantly increase and sustain intakes of fruits and vegetables( Reference Herman, Harrison and Afifi 28 ). Similar community-based interventions incorporating motivational interviews at veteran centres, dietary counselling lines, vouchers and/or coupons for veterans on healthy food items, and faith-placed initiatives promoting access to healthy food items and choices, may help mitigate the burden of poor diet among food-insecure veterans.
Several additional factors associated with poor dietary behaviour among veterans, as found in our study, warrant further discussion and highlight target groups for aforementioned health promotion measures. For instance, SNAP participation was associated with lower fruit intake, consistent with previous literature that noted SNAP participants were less likely to consume fruits( Reference Leung, Ding and Catalano 29 ). Such an association may be due to the higher cost of fruits and the financial burden of purchasing such items that participation in SNAP alone cannot alleviate. In an evaluation of the cost of fruits and vegetables, a report by the US Department of Agriculture demonstrated that the average retail price per pound (~450 g) of fresh fruits ranged from $US 0·26 (watermelon) to $US 7·29 (raspberries), while the average retail price per pound (~450 g) of vegetables ranged from $US 0·48 (potatoes) to $US 4·02 (sliced mushrooms)( Reference Stewart, Hyman and Buzby 30 ). Since SNAP participants are likely to be low income and given the comparably higher price of fresh fruits, the association between SNAP participation and lower fruit intake could be attributed to such price differences, although further analysis of the association between price differences and food purchases among SNAP participants is needed.
Studies have found that post-service veterans are at risk of increased weight gain. In our study, we noted approximately 71 % of veterans were overweight or obese, a prevalence comparable to or higher than that in other studies. While weight standards are required for the US military, much of the literature notes that post-discharge weight gain, and thus obesity prevalence among veterans, is comparable to or higher than that of their non-veteran counterparts( Reference Koepsell, Littman and Forsberg 1 , Reference Koepsell, Forsberg and Littman 4 , Reference Almond, Kahwati and Kinsinger 31 – Reference Teachman and Tedrow 34 ). For example, in a study of approximately 1·8 million veterans Das and colleagues found approximately 37 and 33 % of women and men veterans to be obese, respectively( Reference Das, Kinsinger and Yancy 3 ), while another study noted overweight status to be significantly higher among veterans than non-veterans( Reference Koepsell, Forsberg and Littman 4 ). Similarly, a higher prevalence of obesity has been documented among veterans utilizing Veterans Affairs health care, as compared with non-veterans( Reference Nelson 33 ). The higher prevalence of overweight/obese status in our study population could be attributable to a majority of the population being 45 years or older, with a higher proportion aged 65 years or more.
On the other hand, the reverse relationship between increasing age and poor dietary practices demonstrates that while older veterans were more likely to be overweight or obese, the younger groups in our study were the higher at-risk population for negative outcomes and thus susceptible to prolonged burden of chronic diseases including obesity. As such, significant efforts in health promotion should be targeted towards younger veterans to alleviate the abovementioned need for status foods and learned negative behaviours during service time, as highlighted by Smith et al.( Reference Smith, Klosterbuer and Levine 21 ).
While the relationship of low education and unemployment with poor dietary practices is not surprising and is consistent with previous established literature( Reference Ervin 35 , Reference Hiza, Casavale and Guenther 36 ), the association among veterans does highlight the need for ensuring adequate diet among such at-risk populations. Similarly, studies in the general population have documented poorer diet among males( Reference Hiza, Casavale and Guenther 36 ), potentially due to stronger beliefs and control of weight management among females( Reference Wardle, Haase and Steptoe 37 ), as was observed in our study among veterans.
Given that urban areas are more densely populated and thus more likely to have increased access to fast foods, the association between geographic location and fast food intake is not surprising; although the results do highlight the need for improved policies to implement stronger zoning laws to limit such establishments( 38 ).
The relationship between unhealthy diet and African American ethnicity may be attributed to the higher density of fast food places and lower access to healthy sources of foods in predominantly Black neighbourhoods( Reference Powell, Chaloupka and Bao 39 , Reference Franco, Diez Roux and Glass 40 ). Moreover, the lower intake of vegetables among Asian Americans could be attributable to the lack of Asian language-specific assessments of vegetables, such as okra or drumstick( Reference Ansari, Houlihan and Hussain 41 ). Moreover, studies among the general population have shown lower fruit and vegetable intakes among Asian Americans as compared with non-Hispanic Whites( Reference Maxwell, Crespi and Alano 42 ). As such, health promotion measures targeted at ethnic minority veterans are critical, similar to such measures among the general population. Likewise, given that veterans reporting health-risk behaviours of being a current smoker and binge drinking had poorer dietary practices, public health initiatives should aim to further promote positive health behaviours to counteract the already negative health effects of cigarette and alcohol use( Reference Thun, Carter and Feskanich 43 – Reference Mukamal, Maclure and Muller 45 ).
Two specific results in our study further warrant additional studies to comprehensively evaluate the underlying reasons for the association of obesity and perceived general health with soda intake. Our result on the inverse relationship between health status and soda intake could be attributable to continued public health efforts targeted at the negative health effects of soda; although lack of such an association with fast foods warrants further exploration in future studies. Similar to previous literature demonstrating the relationship between diet and self-perceived health( Reference Lu, Samuels and Huang 46 ), our study also demonstrated that those with low self-perceptions of health were more likely to report higher soda intake, although no other dietary outcome yielded significance. While such results could be due to limitations of our population or other underlying factors not assessed in these cross-sectional data, further longitudinal studies are needed to comprehensively understand the association between self-perception and diet.
Additionally, the cross-sectional data of our study limit our ability to draw causal or temporal relationships between dietary outcomes and food insecurity. Moreover, the self-reported data of the CHIS are susceptible to bias and the lack of assessment of hunger in our study presents a limitation that could be addressed in future studies. Notwithstanding such limitations, the results of our study have significant implications to help mitigate the burden of obesity and associated negative health behaviours among veterans. Given the relationship between food insecurity and low fruit/vegetable intake and high fast food/soda intake, even after accounting for SNAP participation, public health efforts are of imperative need. As discussed earlier, such best practices of providing vouchers, coupons and targeted health education initiatives that promote healthy dietary behaviour among low-food-secure veterans, with follow-up counselling, are urgently needed.
Acknowledgements
Financial support: This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors. Conflict of interest: None. Authorship: M.B.B. and B.J.B. were co-principal investigators of the study. M.B.B. developed the conceptual framework for the study and conducted preliminary analysis. B.J.B. conducted the data analysis. M.M.B., C.M.H. and B.J.B. contributed to data interpretation and final approval for the manuscript. Ethics of human subject participation: Only de-identified public-use secondary data were utilized in this study.
