A mixed‐method systematic review and meta‐analysis of the influences of food environments and food insecurity on obesity in high‐income countries

Abstract Obesity remains a serious public health concern in rich countries and the current obesogenic food environments and food insecurity are predictors of this disease. The impact of these variables on rising obesity trends is, however, mixed and inconsistent, due to measurement issues and cross‐sectional study designs. To further the work in this area, this review aimed to summarize quantitative and qualitative data on the relationship between these variables, among adults and children across high‐income countries. A mixed‐method systematic review was conducted using 13 electronic databases, up to August 2021. Two authors independently extracted data and evaluated quality of publications. Random‐effects meta‐analysis was used to estimate the odds ratio (OR) for the association between food insecurity and obesity. Where statistical pooling for extracted statistics related to food environments was not possible due to heterogeneity, a narrative synthesis was performed. Meta‐analysis of 36,113 adults and children showed statistically significant associations between food insecurity and obesity (OR: 1.503, 95% confidence interval: 1.432–1.577, p < .05). Narrative synthesis showed association between different types of food environments and obesity. Findings from qualitative studies regarding a reliance on energy‐dense, nutrient‐poor foods owing to their affordability and accessibility aligned with findings from quantitative studies. Results from both qualitative and quantitative studies regarding the potential links between increased body weight and participation in food assistance programs such as food banks were supportive of weight gain. To address obesity among individuals experiencing food insecurity, wide‐reaching approaches are required, especially among those surrounded by unhealthy food environments which could potentially influence food choice.


| 3691
ESKANDARI Et Al. offered by single-method reviews to answer complex applied health and public health questions (Stern, Lizarondo, Carrier, Godfrey, et al., 2020). Therefore, the current mixed-method review substantially differs from two recent systematic reviews and meta-analyses which only included quantitative study designs to investigate links between food insecurity and weight status regardless of considering the impact of food environments and qualitative study designs in their studies that were not limited to high-income countries (Moradi et al., 2019;Pourmotabbed et al., 2020). Building on these, we reviewed literature for both quantitative and qualitative studies restricted to high-income countries up to August 2021, with a strong focus on considering the impact of both different food environments (objective or perceived measures) and food insecurity status on obesity.

| Search strategy and selection of studies
Thirteen electronic databases were searched including PubMed, CINAHL, EMBASE, MEDLINE, PsycINFO, ERIC, Scopus, Web of Science, EThoS, Cochrane Library, JBI Library, PROSPERO, and Google Scholar. Search strategy and full search example can be found in Tables S1 and S2. The reference lists of selected articles for critical appraisal were also checked for additional relevant studies. The searches were undertaken from January 2019 to 31 August 2021. The first author (FE) performed the searches and imported citations into Endnote version x9 for the screening process and for removing duplicates. The titles and abstracts were assessed separately by two investigators; all titles and abstracts were screened by the first author (FE) in the first screening stage. Twenty percent of titles and abstracts were then independently double screened by a second reviewer (KR). A third reviewer (CO) resolved any discrepancies by consensus or provided clarification. In the next stage, fulltext articles were evaluated against the eligibility and study quality criteria. The screening for full-text papers was performed by (FE and KR). The PRISMA flow diagram shows the number of articles at each stage ( Figure 1).

| Eligibility criteria
Eligible study designs included observational studies such as cross-sectional studies, cohort studies, and case-control studies, aimed at exploring the association between the food environment, food insecurity, and obesity in adults or children conducted in high-income countries (as defined by the World Bank; The World Bank Group, 2021) published from 1992 onwards (as major studies related to the food environment and food insecurity began from this date). Furthermore, qualitative studies that investigated the perceptions and experiences of obesity arisen from food poverty/insecurity and unhealthy food environments were included. Systematic reviews were excluded, and studies published in English were included. Studies without scientific credibility or non-peer-reviewed were excluded. Animal studies and those investigating obesity grade 3 or more (BMI > 40 kg/m 2 ) were also excluded as the aim of this study was not to look at severe type of obesity.

| Data extraction
Two reviewers (FE, KR) independently extracted data from quantitative studies using the data extraction tool from JBI-MAStARI (Aromataris & Munn, 2020). Study characteristics including specific details of included studies, population demographics, methods and outcomes of interest to the review questions were extracted from each study (Aromataris & Munn, 2020). The qualitative data were extracted independently by both reviewers (FE, KR) using the standardized data extraction tool form JBI-QARI (The Joanna Briggs Institute, 2020).

| Assessment of methodological quality
Included studies were evaluated independently by two reviewers (FE, KR) for methodological validity and risk of bias using the standardized critical appraisal tools from JBI-MAStARI (Aromataris & Munn, 2020), which were specific for each article's study design.
The quality scores were based on the possibility of risk of bias in the methodology, conduct, and analysis in which "Yes" represented a quality score of 1 (Oldroyd et al., 2022). The following maximum scores showed the highest quality: cross-sectional, 8; cohort studies, 11; and qualitative, 10. For this review, methodological quality is reported; however, this did not influence inclusion or exclusion of studies.

| Quantitative data synthesis
To examine the heterogeneity and suitability of quantitative data for meta-analysis, a statistician (AB) was consulted. The main statistics extracted from each study were the mean BMI for individuals experiencing food security and the mean BMI for individuals with food insecurity, together with odds ratio (OR) and 95% confidence intervals (CI). When an OR was not reported, it was estimated from other data based on the methods outlined in the Cochrane handbook (Higgins et al., 2021). The Comprehensive Meta-analysis software version 3 was used to pool effect sizes in a random-effects meta-analysis. A random-effects model was applied to quantify pooled effect sizes and 95% CI. Using Tau statistics, heterogeneity was calculated.
Where statistical pooling for extracted statistics (variables related to the food environments) was not possible, the results were presented in a narrative form according to the Synthesis Without Meta-Analysis (SWiM) guidelines (Campbell et al., 2020). F I G U R E 1 PRISMA flow diagram of the search and screening process for the current mixed-method systematic review.

| Qualitative data synthesis
An approach outlined by Thomas and Harden (2008) was used to develop thematic synthesis for qualitative data. To this end, data were firstly open-coded using line-by-line coding technique (FE, CO).
Then, based on similarities identified within the data, descriptive themes were developed. Finally, analytical themes were developed and were reviewed and agreed (FE, CO).

| Data synthesis for mixed-methods synthesis
The findings of quantitative and qualitative data were aggregated according to a convergent segregated approach outlined in the JBI Reviewers' Manual for JBI Mixed Methods Systematic Reviews (Stern, Lizarondo, Carrier, et al., 2020). This included a configurative analysis approach to generate the links between the findings that represented aggregation. The finding themes from quantitative and qualitative synthesis are presented in narrative description (Campbell et al., 2020;Stern, Lizarondo, Carrier, et al., 2020).

| Main characteristics of the studies
After removal of duplicates, a total of 6307 citations were found.

| Methodological quality
Quantitative studies: Thirty-five cross-sectional studies were included in this review. Overall quality scores ranged from six to eight out of 8 (Table 1 and Table S3). One longitudinal study was also included, having a maximum overall quality score of 11 (Table 1 and   Table S4). Thus, the studies were deemed to be of very good quality, with the risk of selection bias remaining low.
Qualitative studies: Eleven studies were critically appraised and considered of very good quality ( Table 2 and Table S5). Overall quality scores for these studies ranged from 7 to 10 out of 10, indicating a high level of data integrity and congruity between methodology and the research aims, data collecting methods, and analysis.

| Type of food environment characteristics
In this review, food environments were defined as objective (e.g., geographic information system) and/or perceived aspects of the physical and economic food environment inside and outside the home.
Included studies were diverse in their measures and in their results for the food environment component.

| Home food environment
Two studies used home food availability (HFA) scales; in a crosssectional study of 817 individuals, healthy and obesogenic HFA scales were used to assess how frequently particularly foods were available at the home (Poulsen et al., 2019). In another study of 432 parents or caregivers of kindergarten-age children, HFA was assessed using yes or no questions asking about availability of different types of fruits and vegetables (FV), energy-dense foods, and beverages in their homes (Bauer et al., 2012).
In a study of 4589 middle and high school students, household food availability was ascertained via two scales (Widome et al., 2009), measuring both the availability of healthy and unhealthy foods in their homes. Fast food intake was also determined by asking how often they ate something from a fast-food restaurant during past week.
In a study of 2095 parents, participants reported on home food environment using different items. Four items assessed perceptions of access to fruit and vegetables, addressing quality, variety, and cost of produce (Bruening et al., 2012). The types of food consumed at family meals were measured using six items and one item reported on family meal frequency. Fast-food consumption was also assessed.
A cross-sectional study of 152 females participating in Supplemental Nutrition Assistance Program (SNAP) program (aged 18-50 years) used a multi-dimensional home environmental scale (MHES). This scale was created to measure home environment from the perspective of adolescent children and their mothers (Sanjeevi et al., 2018). The environmental influence was measured by questions related to availability of specific healthy and unhealthy food items at home (Sanjeevi et al., 2018).
In a study of 124 largely Hispanic and fifth-grade children (aged 9-13 years), their mothers provided reports of household food supplies (Matheson et al., 2002). Mothers completed a 40-item household inventory of food supplies.
In a cross-sectional secondary study, 50 mothers of 8-to 10-year-old children completed different questionnaires (Kral et al., 2017). Using these, meal and snack patterns of children, Other variables: 1. Dietary intake data that were collected using Intake24 (an open source, on-line, self-completed dietary recall system) Outcome 1: There was no significant association between food insecurity status and BMI in the UK data (this study compared patterns found in the UK sample with those from the USA National Health and Nutrition Examination Survey [NHANES 2013[NHANES -2014), even considering an expected interaction with gender (p = .91). Therefore, the authors did not examine mediation of the food insecurity-BMI association by food consumption variables Outcome 2: Adults with food insecurity consumed significantly more carbohydrate but less protein than those with food security. However, there was no significant difference in total energy intakes or relative fat and fiber intakes by food security status Outcome 3: Adults with food insecurity had larger and more variable time gaps between consumption events. However, they had a significantly smaller and less variable number of foods per consumption events . A question asked, "What would currently help you improve your food situation?" Outcome 1: Most of students (63%) reported they struggle with some level of food insecurity. Near 7% of them reported they sometimes/often did not have enough to eat. Another 56% indicated they had enough food, but not always the kinds of food they wanted Outcome 2: A significant but weak positive association between food insecurity status and BMI was found (r s = .1026, p = .05) Outcome 3: Obesogenic behaviors were more prevalent among those with food insecurity. This group was 13% more likely to purchase cheap processed food (such as frozen pizza or Ramen noodles), 17% more likely to eat more than normal when food was plentiful, and 24% more likely to eat less healthy meals (to eat larger quantities of food) Outcome 4: Only 9% of those with food insecurity and 5% with food security status reported obtaining food from a foodbank (14% of respondents rated access to an on-campus food pantry as a helpful form of support) (p < .05) Outcome 5: The largest proportion of students ranked employment as a helpful form of support, followed by "learn how to eat healthy," "learn how to make a budget," "more financial aid at school," and "learn to cook" (p < .05)

6
Walch and Holland ( Outcome 2: After controlling for socio-demographic factors associated with food insecurity, NEMS-S score in a 2-mile radius was significantly associated with food insecurity (where those with food insecurity had on average 0.39 higher NEMS-S scores). This suggested that those with food insecurity did not live further away from healthier grocery stores, and this was not also modified by ecological measures of vehicle access. It was concluded that those with food insecurity in Detroit are likely to be limited by contextual factors (and not by their neighborhood or physical access to healthy grocery stores) Outcomes 3: Those with food insecurity were younger at screening (p < .001) and were less likely to be married (p < .001) Weight status: Self-reported height and weight Other variables: 1. Dutch Healthy Diet Food Frequency Questionnaire (DHD-FFQ) was used to assess dietary intakes and to construct diet quality scores 2. Information on food bank use was also collected Outcome 1: Prevalence of food insecurity was 26% (of which 18.2% experienced low food security and 7.8% experienced very low food security) Outcome 2: Obesity prevalence was higher for those with very low food security (58%) compared to those with high food security (24%). In total, 43% of those with food insecurity were obese while 25% of those with food security were obese Outcome 3: Compared to those with food security, those living with food insecurity reported more often to have an income below the basic needs budget, to have a lower educational level, and reported less often currently employed Outcome 4: Findings from this study suggested an association between food insecurity and obesity. In the unadjusted model, food insecurity was associated with obesity (OR = 2.49, 95% CI = 1.16, 5.33), but not with overweight (OR = Outcome 3: Those with food insecurity were more likely than those with food security to be above the guidelines for total fat (51% vs. 40%; OR 1.58, 95% CI: 1.07-2.33) Outcome 4: Those with food insecurity were also less likely than those with food security to meet the guidelines for whole grains (8% vs. 15%) Outcome 5: The mean diet quality score was 49 (out of 100). Diet quality for those experiencing food security was significantly higher than those with food insecurity (p = .03) Outcome 6: More than 82% of participants lived in a urban food desert (census tracts of low income and low access to supermarkets. However, the results of the linear regression models indicated that food desert residence was not related to diet quality Outcome 2: Regarding the neighborhood fresh produce environment, women with severe food insecurity perceived significantly lower neighborhood availability (72.5%) and affordability (46.5%) of fresh produce in comparison to women who did not experience severe food insecurity Outcome 3: Males with severe food insecurity reported a lower percentage of neighborhood availability (75.9%) and affordability (40.9%) in comparison to males who had food security Outcome 4: Neighborhood affordability of fresh produce was statistically associated with overweight/obesity. If women with severe food insecurity were able to afford fresh produce in their neighborhood, then they had lower odds of obesity (OR = 0.70, 95% CI Weight status: Self-reported height and weight (BMI was then calculated) Other variables: Home food environment: To assess perceptions of FV access, addressing cost, variety, and quality of produce Outcome 1: Near 39% of participants had food insecurity, and about 13% experienced very low food security. Outcome 2: Parents with food insecurity reported significantly poorer eating behaviors than did parents with food security. Parents with food insecurity consumed breakfast less frequently but ate more serving of SSB Weight status: Self-reported height and weight Other variables: Participation in 3 government-sponsored nutrition programs Outcome 1: 51% of respondents experienced overweight, and 25% were affected by obesity Outcome 2: 30% of households reported having food insecurity during the year prior to the survey and nearly 45% of them had severe food insecurity with hunger Outcome 3: Cases who categorized as food insecure or who had food insecurity with hunger experienced significantly higher BMI than those classified as food secure (p < .01) Outcome 4: Respondents who reported their food supplies did not last, those who were not able to afford balanced meals, those who cut their meal sizes and consumed less than their required need had significantly higher BMI than those who never having those experiences (p < .01) Outcome 5: Respondents who participated in the FSP, WIC and free/reduced-price school meals over the 12 months before to the survey had significantly higher BMI than those who did not receive benefits from government (p < .01) Outcome 6: Those who acquired food from charitable sources such as food banks or soup kitchens had significantly higher BMI (p < .01). Shopping at convenience stores (p = .04) and higher consumption of fast foods was associated with higher BMI (p < .01 Other variables: 1. Home food availability was measured 2. Family food practice: Parents were asked about the frequency of fast-food trips per week 3. Barriers to healthful food at home were reported Outcome 1: Almost 40% of parents and their households had food insecurity over the last year. More than 10% of households had very low food security. Families with a lower total household income and those unemployed had more proportion of food insecurity Outcome 2: No significant differences were observed in either children's or parents' weight status by food security status Outcome 3: Youths from families who had very low food security ate hot food or readymade food from a convenience store or gas station more than twice as did youths from households who were food secure (p = .002). Youths who had food insecurity reported to consume pizza and fried chicken more often than did youths with food security Outcome 4: No differences were observed in households' home food availability or frequency of families' fast-food trips according to food security status Outcome 5: Parents who had food insecurity were most likely to report there was little variety of fruits and vegetables where they buy groceries (p = .003). They were more likely to agree that where they buy groceries the fruits and vegetables had poor condition (p = .03) 6 1. Self-efficacy for food preparation data. Two self-efficacy scores were used to calculate healthy food preparation and food preparation in general 2. Household food supplies, grocery shopping practices, access to traditional food, and perceptions of fruit and vegetable supplies in local stores were also reported Outcome 1: The average BMI was high (mean BMI: 33.7, SD: 6.7) Outcome 2: Approximately 50% of households experienced food insecurity; 39% reported moderate food insecurity and 9% experienced severe food insecurity Outcome 3: Most households shopped at a supermarket located 145 km or more from their homes. Participants perceived negatively about their local grocery stores as they did not usually carry fresh FV, and it was used mainly as a backup Outcome 4: Among BMI categories, only women with severe obesity had less confidence in their healthy food preparation skills (beta coefficient: −0.23, p = .03) Outcome 5: The association between household food insecurity and the general food preparation score was not significant. However, severe household food insecurity and both self-efficacy scores were inversely associated (beta coefficient: −0.25, p = .01) Outcome 6: No association was found between self-efficacy scores and grocery shopping practices and access to traditional food. Lack of availability was reported as a reason for not buying fruits and vegetables locally and this was positively associated with self-efficacy in healthy food preparation (beta coefficient: 0.29, Other variables: 1. Dietary intake: One single 24-h recall was taken to collect information about type and quantity of food consumption, food preparation style, food distribution throughout the day, and where the food is eaten. 2. Coping strategies to alleviate hungers Outcome 1: 45% of boys and 50% of girls were at risk-for-overweight or were affected by overweight Outcome 2: More than 50% of children had no enough food in the house and 25% reported they went to bed hungry Outcome 3: Excessive servings of fats, oil, and sweet group were consumed by youths (18.6-22.7 servings) Outcome 4: Strategies to cope with food insecurity included overeating, eating anything, eating disliked foods, and eating at the homes of family and friends Outcome 5: Overeating was associated with eating at the home of family or friends and eating anything when really hungry (p < .05) with high rates of obesity, poverty, and food insecurity Outcome 1: Over half of participants reported running out of food before the end of the month. Almost all participants received SNAP benefits Outcome 2: Major themes were: store choice (price was most important deciding factor for most of participants), outshopping (having to leave the district to find lower prices and better quality foods), methods of acquiring foods other than the grocery store, and food insecurity (very few participants reported they go hungry, but indicated they sometimes struggle to feed their families) Outcome 3: Concerns regarding price, quality, and transportation were identified as factors negatively impacting food security. Participants perceived cost as a barrier to provide healthy foods for their families Participants reported they usually only had access to one grocery store in their districts, but they do not shop there because of perceived high food prices and poor food quality Outcome 4: In terms of potential solutions, providing job opportunities with reliable hours and a liveable wage were reported by participants as the major way to improve their food environment to examine the root causes of increased obesity risk among individuals with food insecurity Outcome 1: Several themes indicated the link between obesity inequities and structural characteristics of the food banking system. These themes included (i) inadequate access to nutrient-dense food resulted from access to unhealthy foods offered by donors, (ii) gaps in food supply vs. client needs; (iii) geographic disparity in access to health-promoting resources; (iv) state-level emergency food policy and programming; and (v) federal emergency food policy and programming Outcome 2: Themes that addressed the causes of ongoing risk of obesity inequality among individuals with very low food security were found to be (i) access to information; (ii) lack of inclusion/representation among leaders of food banks; (iii) mistrust in communities of color; and (iv) media representation and stereotypes about food pantry users Outcome 3: Many of stakeholders stated that improving the characteristics of the food banking system can alleviate obesity and hunger risks over time in the United States. However, two of them stated that other social issues such as poverty, racism and/or sexism as more important factors that should be addressed instead of improving the characteristics of food banking system alone as they believed that clients rely more on other types of food outlets than on pantries Outcome 2: Food security, measured using 6-item USDA HFSSM, was a major concern among participants with 50% of them reported low or very low food security status Outcome 3: Many participants highlighted participation in both federal nutrition assistance programs (51.9% participated in SNAP, and 27.8% used WIC, and etc) and community food assistance programs as important component of their community food environment. Some participants linked the history of such programs to their poor health outcomes (they stated a mistrust of modern food systems such as processing and largescale growing practices) Outcome 4: In terms of food access sites, many participants reported that groceries were mainly purchased, however, transportation was an issue for most of them because of long commute times and car sharing Outcome 5: Participants also reported reliance in food away from home due to time constraints reported household participation in SNAP Outcome 2: Even those participants employed reported using SNAP, and/or charitable food assistance programs as they struggle to secure a consistent, healthy diet Outcome 3: Most participants stated that they want to eat healthfully, however, they find healthy eating too expensive (they described fruits and vegetables as unaffordable). In all groups, the most noticeable barrier to this was related to the challenge of inadequate income. They discussed about financial constraints because of inadequate income, high costs of housing, transportation, utilities, food, and other expenses. Outcome 4: Other key findings were related to the poor neighborhood food environments, and efforts to stretch food resources. Participants described relatively affordable unhealthy foods (junk foods at schools and corner stores, processed ready-to-heat microwaveable foods, etc.) as too easy to access. They stated that fast food restaurants were prevalent in their neighborhoods and their pricing strategies were reported to encourage unhealthy choices. They also noted about placement of cheap unhealthy foods in grocery stores as a barrier to healthy eating, especially those foods on sale. In fact, participants reported making purchases not based on their preferences but as tradeoffs between their food budgets and their food environment Outcome 5: In addition to a lack of money, participants reported to struggle with inadequate time. Low-wage employment was associated with combined struggle of inadequate time and income. Participants stated that lack of time cause them to consume and feed their families junk food, fast food and highly processed prepared foods (such as microwaveable snacks) Outcome 6: Different food acquisition strategies were used by participants such as buying staple and shelfstable foods, buying foods on sale, searching food donations and food assistance benefits, and shopping at multiple stores to get the best deals 9 TA B L E 2 (Continued)

| Food source destinations (neighborhood food access)
Twelve studies examined different types of neighborhood food access. In a large cross-sectional study of 3748 children (2-18 years old), neighborhood retail food access was measured using numbers of food outlets such as stores, and restaurants located within 1 mile of youths' home (Gorski Findling et al., 2018). Access was reported for specific food outlet types such as fast-food and non-fast-food restaurants, convenience stores, supermarkets, grocery stores, and other stores. Alternative neighborhood food access measures included store type of closest the SNAP retailer (Gorski Findling et al., 2018). Household food purchases and acquisitions were evaluated. To measure spending on unhealthy food items, the mean of sugary beverage spending by each child's household was also examined (Gorski Findling et al., 2018). Another cross-sectional survey of adults (n = 298) measured the sources of foods that were purchased from over the past month (Vedovato et al., 2016). The food sources included fast-food restaurants, convenience stores, bar or pub, food pantry, family and friends, church or community centre, street food vendor, etc. (Vedovato et al., 2016). A study of parents or caregivers of kindergarten-aged children asked parents to report the frequency of family fast-food visits per week (n = 432; Bauer et al., 2012).
In a study of adults aged 18 years and over who lived on lowincome neighborhoods (n = 435), participants were asked to answer questions regarding use of supermarkets or other store types for food purchases and use of free or low-cost food from charitable sources (i.e., food banks, soup kitchens, church or community outreach programs, shelters, friends, and/or family; Webb et al., 2008).
A study of 212 food pantry users asked the participants about frequency of going to food pantries, and/or to soup kitchens (Robaina & Martin, 2013 a Scores for quality assessment (QA).

TA B L E 2 (Continued)
money, shopping at specific stores due to a sale, buying food in bulk, and using a shopping list) in the previous 30 days (McCurdy et al., 2015). Respondents also answered questions about using emergency food parcels from soup kitchens, food banks, community cupboards, or churches (McCurdy et al., 2015).
In another study of 153 women on a low income, a selfadministrated survey asked women about their perceived food access (Watt et al., 2013). Seventy-five percent of women reported that they experienced limited food access. Finally, a cross-sectional study of 107 women responsible for household food supplies measured grocery shopping practices and access to traditional foods (Mercille et al., 2012). Most families made their grocery shopping from a supermarket within 145 km or more from their homes.

| Food source destinations -nutrition assistance programs
Nine studies assessed participation in nutrition assistance programs; the interactions between SNAP participation, food insecurity, and BMI were examined in a study of 2003-2010 US NHANES (n = 8333; Nguyen et al., 2015). SNAP is previously recognized as food stamps and is the largest federal program in the United States that offers support for the purchase of foods to low-income US households to alleviate food insecurity (Sachdev et al., 2019).  (Webb et al., 2008). In a study of 212 food pantry users, participants were asked whether they receive SNAP or WIC (Robaina & Martin, 2013). Over half ( income, data on participation in WIC and/or SNAP were obtained from the self-administrated survey (Watt et al., 2013). Sixty-four percent of participants used the benefits obtained from WIC and half of them benefited from food stamps (SNAP) over the previous year.

| Healthy food beliefs and attitudes
Five studies measured this aspect of food environments; in a crosssectional survey of 298 adults, four subscales were developed to indicate different aspects of beliefs and opinions about healthy foods: affordability, convenience, importance, and taste (Vedovato et al., 2016). In a study of 432 parents or caregivers of kindergartenaged children, barriers to healthy foods at the home were assessed (Bauer et al., 2012). A cross-sectional study of 107 females who were responsible for household food supplies measured self-efficacy for food preparation using the calculation of two self-efficacy scores.
One scale measured food preparation in general and another one measured healthy food preparation (Mercille et al., 2012). Selfefficacy in food preparation was described as individuals' confidence in their ability to make dishes and balanced meals using store-bought food. Women were fairly confident about their capability to prepare store-bought food. However, the average score for self-efficacy in healthy food was slightly lower, suggesting more difficulty in this regard (Mercille et al., 2012). They also reported on their perceptions of FV supply in local stores. The local grocery store was perceived negatively, as it did not usually bring fresh FV and it was used mainly as a backup.

| Dietary intake (diet quality)
This determinant was reported as intake of specific foods or food groups are associated with obesity and such diets may relate to different aspects of neighborhood food environments.
Thirteen studies measured dietary intakes of participants; in a cross-sectional analysis of 7741 Adult California Health Interview Survey, dietary information was collected by asking about the frequency of eating of fruits (excluding fruit juice); vegetables (excluding fried potatoes); soda (excluding diet soda); French fries, and fast food (Leung & Villamor, 2011). A study of 432 parents or caregivers of kindergarten-aged children, parents were asked about the frequency of fast-food visits per week. They were also asked about the frequency of their child food consumption from hot or readymade food from a convenience store or gas station over the past 30 days (Bauer et al., 2012). Another study of 212 food pantry users, diet quality of participants was measured using the Block Food Frequency Screener. These users reported about their usual consumption of fruit, vegetables, and fiber (Robaina & Martin, 2013).
A cross-sectional study of 202 young people (9-18 years) who were homeless and were living in two of the largest family shelters in the USA assessed dietary intake by completing a single 24-h recall to provide information about the type and quantity of food consumed, preparation style, where food is eaten, and how it is spread over the day (Smith & Richards, 2008). Dairy, fruits, and vegetables were consumed less than recommended levels (below the estimated average requirements) by both males and females of all ages. All youths ate excessive servings of sweet groups, fats, and oil (18.6-22.7 servings). Another cross-sectional study of 195 Somali refugee women in the United States estimated their regular dietary intakes by completing a short food frequency questionnaire. Questions were asked to estimate how often specific food items were consumed such as eggs, meats, beans/lentils, grains, dairy, fruits, and vegetables (Dharod et al., 2013). In a study of 153 women on a low income, mother's diet was measured using an 8-item index from Starting the Conversation (Watt et al., 2013). It included questions about intakes of FV, sugar-sweetened beverages (SSB), high-fat foods, and desserts. Infant's diet was evaluated as breastfeeding initiation and consumption of particular foods such as fruits and French fries. The majority of women did not meet dietary guidelines and nearly 64% of them reported weekly intakes of fast-food. Drinking SSB at a daily basis was reported by 44% of women. Most of the women breastfed after delivery. Usual feeding practices were that 39% of women reported they gave their infants high-sugar fruit/vegetable juice daily and 24% of them reported feeding their infants sweets on a weekly basis (Watt et al., 2013).

| Coping strategies to alleviate hunger
One cross-sectional study of 202 young people (9-18 years) evaluated coping strategies used by youths who were homeless and were living in two of the largest family shelters in the USA (Smith & Richards, 2008). Coping strategies to alleviate food insecurity included overeating, eating at the homes of family and friends, eating disliked foods, and eating anything.

| Quantitative component 3.3.1.1 | Associations between food insecurity and BMI
To analyze the association between food insecurity and obesity,  Webb et al., 2008;Widome et al., 2009;Wirth et al., 2020), involving both adults and children, was pooled together for the metaanalysis. These studies used a cross-sectional approach in addition to one cohort study. As shown in Figure 2, meta-analysis of these studies showed an overall small but statistically significant association between food insecurity and obesity (OR: 1.503, 95% CI: 1.432-1.577, p-value = .000) when all ORs were combined with the random-effects model. This means food insecurity increased the risk of obesity among adults and children. Therefore, individuals experiencing food insecurity were more likely to be affected by obesity. Another study of parents or caregivers of kindergarten-aged children (n = 432) found no differences in families' HFA by food security status (Bauer et al., 2012).

| Associations between food environments and BMI
A study of middle and high school students (n = 4589) demonstrated that youths experiencing food insecurity had several eatingrelated risk factors for overweight/obesity (Widome et al., 2009).
Adolescents who "often" did not have enough to consume or that they suffered from hunger "some months" reported eating more fast food than did those who had food security ( This study found that youths with food insecurity were less likely to eat family meals than peers of a higher socioeconomic status (p < .001). The study suggested that this might be because of limited or irregular food availability that might be less likely to establish a regular family meal routine (Widome et al., 2009).
The population-based study of 2095 parents showed that the home food environment in households living with food insecurity was poorer than in households living in food security (Bruening et al., 2012). Parents experiencing food insecurity reported having more fast food at family meals (95% CI = 0.1, 0.2, p < .01) and more serving of SSB (95% CI = 2.5, 10.3, p < .01) at family meals than parents living in food security. They also reported serving frequently less healthy food items such as green salad, vegetables, and fruits (p < .05). The study explained that the higher BMI and poorer eating patterns of parents with food insecurity may be due to the fact that more obstacles in accessing healthy foods such as FV were reported by this group than parents experiencing food security. Great differences in perceived access to FV were reported between parents living in food security and parents living with food insecurity. For example, near 40% of parents experiencing food insecurity compared to near 14% of parents experiencing food security perceived that fruits were too expensive to purchase (95% CI = 21.5, 30.6, p < .01). Parents affected by food insecurity were 3-4 times more likely to find FV to be too expensive. These parents believed that the quality and variety of available FV were poor (p < .01).
Another study of 152 participants in SNAP program, availability of unhealthy food was significantly associated with BMI (beta coefficient = −0.227, p-value = .02; Sanjeevi et al., 2018). Also, a significant difference between groups living in food security and groups living with food insecurity was reported for availability of unhealthy foods at home. Groups experiencing food security scored higher than groups with food insecurity for the availability of unhealthy foods at home by almost 14% (p < .01). This study indicated that the relationship between food insecurity and obesity was partially mediated by home food environment (beta coefficient: 0.19, 95% CI: 0.01-0.42, p < .05). Thus, home food environment could play a vital role in mediating this relationship in this population.
The study of Hispanic and fifth-grade children (n = 124) demonstrated a significant association between household food supplies and household food security (p < .01). Household food supplies were significantly correlated with youth's consumption of fruit, meat, sweets, and snacks at home (p < .05; Matheson et al., 2002).
A cross-sectional secondary study of 50 mothers of 8-to 10-year-old children (n = 100) found large percentage of youths from households living in food security reported eating 3-4 snacks per day (46% vs. 15.4%), while a greater percentage of youths from households living with food insecurity reported eating 5 or more snacks per day (15.4% vs. 0%) (p = .02; Kral et al., 2017). Mothers from households experiencing food insecurity reported significantly more concern regarding their child's weight and subsequently limited access to food by their children at a greater extent than mothers from households living in food security (p < .03). Children from households with food insecurity had significantly more external eating, both past satiety and in the absence of hunger (p < .03). overall and those participated in SNAP (Gorski Findling et al., 2018).
Odds of childhood overweight/obesity were higher with greater access to combination grocery/other stores overall (OR: 1.10, 95% CI: 1.03-1.17, p < .05) and for children in SNAP (OR: 1.14, 95% CI: 1. 05-1.24, p < .05;Gorski Findling et al., 2018). In this study, alternative access measures of food exposure were not associated with child overweight/obesity (Gorski Findling et al., 2018). The average child lived in a household in which 6.3% of their total spending at food outlets was on sugary beverages (Gorski Findling et al., 2018). Compared to non-SNAP households, the average youths from households participated in SNAP also spent a higher percentage of their budget on sugary beverages (p < .05;Gorski Findling et al., 2018).
A cross-sectional survey of 298 households found no significant associations between food source use patterns (such as shopping from a convenience or grocery store) and excess body weight (Vedovato et al., 2016).
A study of 432 parents and caregivers of kindergarten-aged children found no differences in frequency of families' fast-food visits by food security status (Bauer et al., 2012).
In a study of 435 adults who were residents of low-income neighborhoods, BMI was significantly higher for those cases who acquired their foods from charitable sources such as food banks or soup kitchens (p < .01). Also, those who reported shopping at convenience stores (p = .04) and those who consumed fast-foods in the month prior to the survey had significantly higher BMI (p < .01; Webb et al., 2008). In this study, no association was found between BMI and different use of supermarkets, ethnic grocery stores, or use of farmers' markets.
A study of mothers of young children on a low income (n = 166) found no significant association between maternal BMI and number of weekly shopping visits to supermarkets (McCurdy et al., 2015). However, maternal BMI was significantly associated with variables related to food resources. Use of community food programs (p < .05) and more frequent use of food shopping practices to stretch food dollars (p = .04) were positively associated with maternal BMI.
In a study of women who were responsible for household food supplies (n = 107), no association was found between self-efficacy scores, grocery shopping practices, and access to traditional food (Mercille et al., 2012).

| Food source destinations -nutrition assistance
programs. Analysis of NHANES 2003-2010 showed that, while those experiencing food insecurity or SNAP participants had a higher BMI and greater possibility of obesity (p < .05), the combined association of food insecurity and SNAP participation indicated a decrease in BMI across all three groups of food insecurity (p < .05) and reduced the chance of obesity among those who had marginal food security (p < .05; Nguyen et al., 2015).

A cross-sectional analysis of 7741 Adult California Health
Interview Survey demonstrated that the incidence of obesity was 30% higher among those who participated in SNAP than among the nonparticipants (p = .01; Leung & Villamor, 2011). This association was more evident among males than females. Participation in SSI programs was positively associated to an adjusted 50% higher incidence of obesity compared to those who did not participate.  Vedovato et al., 2016).
A study of 432 parents or caregivers of kindergarten-aged children found that food security status changed parents' experience of barriers to having healthful food in their homes (Bauer et al., 2012). Parents experiencing food insecurity were most likely to report that there was little variety of FV where they buy groceries (p = .003) and were more likely to agree that where they buy groceries the fruits and vegetables were in poor condition (p = .03). These parents were also more likely to report that their family does not like FV (p = .01).  (Leung & Villamor, 2011).
The findings from another study of 432 parents or caregivers of kindergarten-aged children demonstrated that children from families who experienced very low food security had higher intakes of hot or ready-made foods bought from a convenience store or gas station than those youths from families living in food security (p = .002). Children living with food insecurity also ate pizza and fried chicken more often than children living in food security (p < .05; Bauer et al., 2012). Based on the results from a study of 212 food pantry users, participants living in food security were twice as likely to eat fruit, vegetables, and fiber than those who had food insecurity (OR = 2.3, 95% CI: 1.1, 5.2, p = .05; Robaina & Martin, 2013).
A cross-sectional study of 202 young people who were homeless found significant associations with overeating (as a coping strategy) and higher food intakes of fat (p = .037), protein (p = .010), and the meat food group (p = .014) among females 9-13 years (Smith & Richards, 2008). Among youths' males 9-13 years, overeating was associated with increased intakes of calories (p = .012), carbohydrates (p = .030), fat (p = .011), protein (p = .015), bread (p = .028), and vegetables (p = .029). Eating at the homes of family and friends, as the coping strategies, was also associated with overeating (p = .017; Smith & Richards, 2008). These results suggested that these youths have used coping strategies for dealing with a food insecure environment by overeating and consuming high-fat foods when tasty food was available. In this study, the major calorific snacks that were identified to be commonly consumed by youth were salty snacks, candy, soft drinks, fruit drinks, French fries, cheeseburgers, and pizza.
These types of foods are largely offered by fast-food restaurants and convenience stores, which are common stores in downtown urban neighborhoods (Smith & Richards, 2008). Another crosssectional study of 195 Somali refugee women in the United States demonstrated an association between BMI scores and daily intake from different food groups. A significant difference was noted in the fruits, vegetables, and beans groups (Dharod et al., 2013). Intake from all of these food groups at least once a day was less common among participants who were affected by overweight/obesity than individuals with normal weight (p ≤ .05

| Qualitative component
Two analytical themes were developed from 19 study findings extracted from included studies. The study findings and relations between descriptive themes are presented in Table S6 and Figure 3.
The synthesized analytical findings are as below.
1. A reliance on energy-dense, nutrient-poor foods due to their affordability, accessibility, and extended shelf life must be acknowledged. Policy efforts are needed to focus on affordability and availability of neighborhood fresh produce as well as to consider the importance of the food environment in mediating the relationship between food poverty/insecurity and BMI among low-income individuals.
2. Food banks and other food support networks, used as a coping strategy for food insecurity, have the potential to affect their users' health and body weight. Therefore, increasing the nutritional quality of food provided by them is essential.

| Mixed-methods aggregation of qualitative and quantitative synthesized findings
Mixed-methods syntheses were conducted to answer the following questions: "Are the results from each synthesis supportive or contradictory?", "Do the qualitative data support to explain variations in the direction and size of correlations within the included quantitative studies?". To facilitate the final aggregation of the individual syntheses, a convergent segregated approach was applied (Stern, Lizarondo, Carrier, et al., 2020). The synthesized results from the qualitative components were combined with textual descriptions translated from quantitative findings ( Table S7). The mixedmethod synthesis: To reduce the prevalence of overweight/obesity, holistic interventional approaches are required to be implemented to remediate both food insecurity and unhealthy individuals' dietary behaviors that are influenced by different types of food environments. These efforts should emphasize affordability and availability of neighborhood fresh produce as well as to consider other components of food environments such as unhealthy obesogenic food environments in mediating the relationships between food insecurity and overweight/obesity, especially among low-income families. It is essential that the nutritional quality of food provided by nutrition assistance programs is improved.

| DISCUSS ION
The results of our meta-analysis (n = 36,113) showed an overall small, but statistically significant, association between food insecurity and obesity. These results demonstrated that food insecurity increased risk of obesity among adults and children. Therefore, individuals experiencing food insecurity were more likely to be affected by obesity. These findings are important given the context of the "cost of living crisis" and rising health inequalities (Limb, 2022).
There were also positive associations between different types of food environments and overweight and obesity. A study of female participants in SNAP program demonstrated that availability of unhealthy foods at home was significantly associated with BMI (Sanjeevi et al., 2018). This study indicated that the relationship between food insecurity and obesity was partially mediated by home food environment. A large study revealed that women with severe food insecurity had lower odds of obesity if they were usually or always able to afford fresh fruits and vegetables in their neighborhood (Ro & Osborn, 2018). Therefore, neighborhood affordability of fresh produce was accounted as the driving factor that reduced the statistical association between food insecurity and overweight/obesity.
Another large study of children aged 2-8 years demonstrated that the average child and adolescents in SNAP household spent a higher percentage of their budget on sugary beverages than on non-SNAP households (Gorski Findling et al., 2018). In another study of adult residents of low-income neighborhoods, those who obtained food from charitable sources such as food bank or soup kitchens had significantly higher BMI than those who shopped at convenience stores and those who ate fast-foods in the month before the survey (Webb et al., 2008).

F I G U R E 3 Relations between descriptive themes
In the absence of data on the direct associations between food environment exposure and BMI outcome, the links between food insecurity (as outcome) and different types of food environments (as exposure) were also examined. This was performed to better understand the mechanisms behind the association between food environments and weight status as the links between food insecurity and overweight/obesity was established through our meta-analysis.
For example, a study of 4589 middle and high school students did not provide the association between BMI and home food environments directly (Widome et al., 2009). However, it showed that youths with food insecurity had several eating-related risk factors for overweight/obesity. It demonstrated that these youths, who had the greater percentage of obesity, reported eating more fast-food than did those who experienced food security. This group also had less both healthy and unhealthy food available in their home. They were also less likely to eat family meals than their counterparts experiencing food security. The study suggested that this might be because of limited or irregular food availability, leading to less instilling a regular family meal routine (Widome et al., 2009). In another population-based study, the food environment in households living with food insecurity was poorer than in households living in food security (Bruening et al., 2012). Large differences in perceived access to FV were available between parents experiencing food security and parents who experienced food insecurity. Parents who experienced food insecurity reported that the quality and variety of available FV were poor. More importantly, this group of parents perceived that fruits were too expensive to purchase compared to parents experiencing food security (Bruening et al., 2012). Another study of parents and caregivers of kindergarten-aged children found food security status influenced the parents' experience of obstacles to having healthful food in their home (Bauer et al., 2012). Parents with food insecurity were most likely to report that there was little variety of fruit and vegetables in poor condition where they buy groceries. It also found youths from families who experienced very low food security reported eating more than twice ready-made food or hot food from a convenience store or compared to children whose families experienced food security. In comparison to children living in food security, youths living with food insecurity also consumed fried chicken and pizza more often (Bauer et al., 2012). The findings from qualitative studies (n = 409 participants) regarding a reliance on energy-dense, nutrient-poor foods due to their affordability and accessibility aligned with quantitative studies. A study of children's lived experience found that participants mentioned healthy food is more expensive and cost was a barrier to purchasing fresh fruit (Genuis et al., 2015). They also raised the issue that accessibility and transportation play a key role as to reach the closest grocery store that sells a full range of healthy market choices, and that a vehicle is required. Also, findings from qualitative and quantitative studies regarding the potential links between increased body weight and participation in food assistance programs such as food banks, used as a coping strategy for food insecurity, were supportive. A qualitative study of food bank users revealed that although relying on food bank parcels meant that they could afford to pay bills, however, it also meant sacrificing fresh food that exacerbated their weight gain (Thompson et al., 2018).
In the present study, our findings from the aggregation of qualitative and quantitative analyses recommend that holistic approaches (including policy) are required to remediate food insecurity and unhealthy individuals' dietary behaviors that are influenced by different types of food environments in order to reduce the prevalence of overweight/obesity.

| Strengths and limitations
This review is particularly comprehensive by the inclusion of both quantitative and qualitative studies. However, the study is not without its limitations. The focus on the impact of both food insecurity status and the food environment on high BMI meant that articles that only measured one of these factors were excluded from this review. This could potentially limit the inclusion of relevant evidence.
For example, those studies that considered the important role of smartphone technology but not the role of food insecurity on obesity risk were excluded, although food environments particularly expand into online settings that shape consumers' food choices (Vadiveloo et al., 2021). Most included quantitative studies were cross-sectional in their designs. Thus, it was not possible to identify causality or direction among key variables (Ro & Osborn, 2018). Moreover, it was not possible to assess whether food insecurity status and/or the food environment variables temporarily caused different behaviors and perceptions or if all these factors shared a common cause (Widome et al., 2009). Therefore, further longitudinal studies are warranted to acknowledge the possible associations between these variables. For those studies that used self-reported measures to evaluate anthropometric indices, they are subject to misclassification of subjects (Ro & Osborn, 2018). For example, if overweight/obesity were underestimated in those studies, these might make a more cautious interpretation of our results. Regarding food environment exposures, some studies used personal perceptions of neighborhoods. For instance, in a cross-sectional study of 5957 individuals, neighborhood measures were personal perceptions of neighborhoods. As such, this may not represent objective neighborhood characteristics such as food (Ro & Osborn, 2018). Small sample size of qualitative studies (n = 11) could be also considered a limitation. These studies took place in specific context and communities which might limit transferability of findings from such small sample size. However, they provide some insights into the lived experience of individuals suffering from overweight or obesity and food insecurity. Finally, it is difficult to make comparisons between studies in relation to measurement tools used, as these differed from study to study.

| Implications for policy and practice
This systematic review highlights that obesogenic food environments and food insecurity significantly contribute to obesity. This supports the evidence concerning reliance on cheap energy-dense foods in favor of nutrient-dense foods such as fruits and vegetables. For instance, this review indicates that those living with food insecurity have higher fruit and sugary beverage intakes compared to those living in food security. Since these beverages are cheaper than the equivalent whole fruits, this might be preferred by these individuals under economic constraint (Yau et al., 2020). Since this review indicated that BMI is significantly higher for those who acquire their foods form charitable sources such as food banks, implementing policies and efforts to improve the nutritional quality of food parcels is essential to help food bank users to meet their individual dietary needs. A recent mixed-method systematic review has explored the nutritional quality of food parcels provided by food banks and the effectiveness of food banks at reducing food insecurity in developed countries (Oldroyd et al., 2022). The results of this study revealed that pre-packaged food parcels provided by food banks were inconsistent at meeting nutritional requirements of their users and often failed to meet individual needs, including cultural and health preferences. Use of food banks improved food security and dietary quality of users, allowing otherwise unachievable access to food. Nevertheless, food insecurity remained, and was explained by limited food variety, quality, and choice (Oldroyd et al., 2022).
These mixed-method findings encourage interventions to ensure consistent, adequate nutrition, and improved nutritional quality of food parcels at food banks to meet nutritional needs of those requiring food banks.
These findings emphasize the importance of structural and policy changes to the food and economic environment. There need to be societal changes to reduce inequalities to facilitate national and international goals of reducing overweight and/or obesity. This also provides scope for halting rising trends in food insecurity as well as eradicating food insecurity. A suggested approach to tackling such issues might be to address the high and rising cost of food, especially healthy foods (Yau et al., 2020) particularly within the context of the global cost of living crisis. It is reported that in high-income countries like the UK, even people who work full-time on the National Living Wage cannot necessarily achieve the Minimum Income Standard (i.e., the income needed to reach a minimum socially acceptable standard of living; Yau et al., 2020). Therefore, combined with the rising economic crises related to recent COVID-19 pandemic and world events, addressing wage-related policies to ensure sufficient income for adequate standards of living is critical to address health inequalities.

| Implications for research
Further longitudinal studies investigating the impact of obesogenic food environments and food insecurity on obesity among general populations, rather than minority-specific, and in countries beyond the USA, will strengthen the evidence base. Since this review indicated that BMI is significantly higher for those who acquire their foods from charitable sources such as food banks, further updated reviews for high-income countries investigating the nutritional quality of food parcels and whether using foodbanks reduces the food insecurity and improves their users' diets will strengthen the evidence base.

| CON CLUS IONS AND IMPLIC ATIONS OF THIS RE VIE W
Drawing on evidence from research across high-income countries, the present systematic review and meta-analysis showed that food insecurity and some types of food environments are a risk factor for obesity. Wide-reaching approaches (including policy changes) are required to address overweight/obesity among individuals experiencing food insecurity, especially among those whose food choices are influenced by unhealthy food environments. Our results suggest that these efforts should focus on affordability and availability of neighborhood fresh produce as well as to consider other components of food environments such as unhealthy obesogenic food environments in mediating the relationships between food insecurity and overweight/obesity, especially among low-income families. It is also essential that the nutritional quality of food offered by nutrition assistance programs is improved.

ACK N OWLED G M ENTS
This research was supported by Teesside University. Guidance with literature searches from Mrs Carol Dell Price, Teesside University, and Prof Alan Batterham, Teesside University for his advice with data analysis are gratefully acknowledged.

CO N FLI C T O F I NTE R E S T
No conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.

E TH I C A L S TATEM ENT
This review used only published sources of data. Ethical review by a Research Ethics Committee was not required.