The association between loneliness or social isolation and food and eating behaviours: A scoping review

.


Introduction
Loneliness and social isolation are being described as public health problems and international epidemics with growing evidence of adverse health outcomes including earlier mortality, cardiovascular disease, dementia and depression (Holt-Lunstad, Smith, Baker, Harris, & Stephenson, 2015;Leigh-Hunt et al., 2017;Morina, Kip, Hoppen, Priebe, & Meyer, 2021;Naito et al., 2021;Qiao et al., 2022).There are multiple plausible pathways through which loneliness or social isolation, food, eating behaviours and health outcomes could be interrelated.The American Heart Association noted that the pathways through which loneliness or social isolation could increase the risk of some adverse health consequences are likely to be complex and multifactorial and developed a conceptual model examining the associations between loneliness, social isolation and cardiovascular and brain health (Cené et al., 2022).The model identified the potential role of multiple mediators comprising behavioral (including diet), psychological (such as depression) and physiological (such as cardiovascular risk factors and inflammation) (Cené et al., 2022).Conversely, loneliness or social isolation and food and eating behaviours could be independent factors contributing directly to health outcomes.The possibility of confounding associations involving these variables and health outcomes should also be considered.The need for further evidence on the relationship between loneliness or social isolation and health behaviours such as food and eating has been highlighted (Cené et al., 2022;Leigh-Hunt et al., 2017) and mapping existing research is an important step to enable better understanding.
Loneliness is typically defined as a subjective feeling or experience of the absence of meaningful connections with a social network or companion (Fried et al., 2020;Perlman & Peplau, 1981;Prohaska et al., 2020).In contrast, social isolation is defined as an objective lack of interactions with others or the wider community (Wenger, Davies, Shahtahmasebi, & Scott, 1996).Across a range of countries, estimates for the prevalence of loneliness in adults across age groups range from 1.8 to 24% and in the USA and Europe the prevalence ranges from 5 to 43% (Surkalim et al., 2022).The COVID-19 pandemic and concomitant social restrictions focused attention on loneliness and social isolation as integral contributors to ill-health and highlighted their seriousness (Smith & Lim, 2020).Research has explored the need to distinguish between social isolation and loneliness in the assessment of mortality (Holt-Lunstad et al., 2015) and also concluded they appear to be independent risk factors for worse cardiovascular and brain health (Cené et al., 2022) however few researchers have concurrently investigated loneliness and social isolation and knowledge of their relative or synergistic influence is limited (Holt-Lunstad et al., 2015).
The parallels between social isolation or loneliness and food and eating have been discussed across a range of fields.For example, neuroscience research has concluded that acute isolation leads to social craving in a manner similar to hunger as a response to not eating (Tomova et al., 2020).Some discourses have directly connected this analogous relationship with suggestions that people eat to alleviate loneliness or fill a void (Ljubičić et al., 2023;Troisi & Gabriel, 2011) and neuroscience experts have discussed how loneliness can influence eating behaviours (Cacioppo & Patrick, 2008).It is acknowledged that some of the numerous determinants of food choice include social pressures and cognitive-affective factors (Leng et al., 2017).More broadly, loneliness has been identified as a component of emotional eating, the "tendency to eat in response to negative emotions with the chosen foods being primarily energy-dense and palatable ones" (Konttinen, 2020;Konttinen, 2020, p. 283) (Konttinen, 2020;Ljubičić et al., 2023).However, the direction of the relationship cannot be assumed.An important consideration is commensality, the act of eating with others, which has deep social and cultural significance (Jönsson, Michaud, & Neuman, 2021) and has been shown to have multiple possible social and individual benefits including improved nutrient or food intake (Dunbar, 2017;Fischler, 2011;Jönsson et al., 2021;Kimura et al., 2012) although a lack of causal evidence has been noted (Jönsson et al., 2021).Food restriction also could increase loneliness due to reducing the ability to bond with others through consumption of similar food (Woolley, Fishbach, & Wang, 2020).
The reasons why social isolation could influence food and eating are likely to overlap with loneliness, yet could also be distinct.For example, people that are socially isolated are more likely to live alone which is associated with dietary behaviours such as lower diversity and intake of fruit, vegetables and fish (Hanna & Collins, 2015).Persons who are socially isolated could have other characteristics that impact food access, preparation and intake such as adequate transportation, proximity to food markets and functional impairment (Locher et al., 2005).This further highlights the numerous factors likely to influence or mediate the associations between loneliness, social isolation, food, eating and health.
While non-scholarly and scholarly publications indicate interest in the relationships between loneliness or social isolation and food and eating behaviours it is not clear if there is evidence to support these links.One systematic review reported strong evidence of no association between social isolation and loneliness and malnutrition in older people (van der Pols-Vijlbrief, Wijnhoven, Schaap, Terwee, & Visser, 2014).However, no other reviews have been located that synthesise the evidence exploring the relationship between these areas.Given the rising recognition of loneliness and social isolation as a determinant of health, it is important that health professionals, especially those in the fields of nutrition and/or mental health, have an awareness and understanding of loneliness and social isolation and the interaction with food and eating.Public attention on loneliness has also led to its being a higher priority in policy agendas (Prohaska et al., 2020) and a better understanding of loneliness (Prohaska et al., 2020) and the determinants of food choice (Leng et al., 2017) is needed to inform the policy agenda into the future.Due to the limited evidence in this area a scoping review was chosen to assess a broad question investigating currently available literature and knowledge gaps.Therefore, the aim of this scoping review was to explore, map and provide a summary of evidence investigating the association between loneliness or social isolation and food and eating behaviours in people aged 16 years and over in high income countries both outside of and during COVID-19 restrictions to inform future research and identify gaps in the knowledge base.

Protocol and registration
The protocol for this scoping review was informed by the Joanna Briggs Institute methodological guidance for scoping reviews (Peters et al., 2020) and adheres to the Preferred Reporting Items for Systematic Reviews and Meta-analysis extension for scoping reviews (PRISMA-ScR) (Tricco et al., 2018) (Supplementary Table 1).The participants, concept and context of the review were specified in the protocol (Peters et al., 2020).Participants were persons aged over 16 years with further details provided below in eligibility criteria.The concept centred around exploring associations between loneliness and/or social isolation and food and eating behaviours.The context of the research included restriction to high income countries and inclusion of studies conducted outside of and within COVID-19 restrictions.The protocol was registered with Open Science Framework on May 9th, 2021 (https://osf.io/kmy6c/).As the research is a review, it was exempt from approval by the Queensland University of Technology University Human Research Ethics Committee.

Eligibility criteria
Eligible studies were published in English from the year 2000 and conducted in high income countries (The World Bank).The population of interest was individual adults and adolescents (aged ≥16 years) living in the community.Articles were excluded if the focus was on families, shift workers, people living in institutions with meals provided to them or participants with medical conditions including social anxiety or eating disorders because these factors are likely to modify the association between loneliness or social isolation and food and eating behaviours Observational studies including cross-sectional, longitudinal or qualitative designs and systematic reviews of observational research published as peer-reviewed journal articles or Masters or PhD theses were eligible for inclusion.Interventions, and studies published only as abstracts, conference presentations or narrative reviews were excluded.
Included quantitative studies reported a measure of either social isolation or loneliness, and a measure related to food and eating.Studies that focused on single food items were not considered representative of eating behaviours and were therefore excluded.The scope of the review focused on behaviours more immediately related to eating so behaviours such as cooking and food shopping were excluded.Studies that only measured social support were excluded.Studies where obesity or malnutrition was the explanatory variable for loneliness or social isolation were outside of the scope of this review.

Search and information sources
Five databases (CINAHL, PsycINFO, PubMed, Scopus and Embase) were systematically searched for literature published up to August 9, 2023.ProQuest Dissertations and Theses Global was searched to identify eligible relevant dissertations up to August 10, 2023.A predetermined list of key words, along with relevant database library index terms was developed by two authors (AN and JC) and finalized in discussion with all authors and a university liaison librarian.Due to the different database terms, there were differences in the search syntax between databases.The Scopus search strategy was as follows: (TITLE-ABS-KEY ((loneliness OR "social isolation")) AND TITLE-ABS-KEY (diet OR eating OR "eating behaviour" OR "eating behaviour" OR appetite OR "feeding behaviour" OR "feeding behaviour" OR "food preferences" OR "food choice" OR meals OR "food intake" OR "diet quality" OR "dietary patterns" OR "food habits" OR "ultra-processed foods" OR discretionary OR K. Hanna et al. "energy dense, nutrient poor"))).The full search strategies for the other databases are in Supplementary Table 2.

Selection of sources of evidence
Articles from each database search were exported into EndNote X9 (The EndNote Team, 2013) where duplicates were removed, and remaining articles exported into Rayyan (Ouzzani, Hammady, Fedorowicz, & Elmagarmid, 2016).Title and abstracts were screened by two independent reviewers (AN, JC or KH) using the predetermined inclusion and exclusion criteria.Full text for all potentially relevant articles were obtained and screened by two independent reviewers (AN, JC or KH).Additional information for one study, required to determine eligibility, was obtained from the author.Conflicts were resolved by consensus and additional reviewers (DG).Backwards citation searches of all eligible papers were independently conducted by two reviewers (KH, AN or JC) and disagreements resolved by consensus.

Data charting process
A pilot data extraction was completed on four articles by two independent reviewers (AN, JC) to ensure the extraction table encompassed all relevant variables.The final data extraction form included study characteristics, target group characteristics (gender/sex is reported as it appears in the original paper), loneliness or social isolation measures, food and eating behaviour measures, and main results.Two reviewers (AN, JC or KH) independently completed data extraction and reviewed the data charting completed by the other persons which was further confirmed by an additional reviewer (DG).

Critical appraisal of sources of evidence and synthesis of results
An appraisal of the quality of the quantitative studies was conducted using criteria derived from the National Institutes of Health Quality Assessment Tool for Observational, Cohort and Cross-Sectional Studies (Supplementary Table 3) (National Heart Lung and Blood Institute., 2013).Qualitative studies were appraised using criteria derived from the Critical Appraisal Skills Programme CASP (Qualitative) Checklist K. Hanna et al. (Critical Appraisal Skills Programme, 2018).Whilst the quality of data was considered, no articles were excluded as a risk of bias and quality assessment is not required for a scoping review.A synthesis of extracted data was undertaken by grouping findings from each study into common themes based on the food or eating behaviours assessed.Studies conducted within the context of COVID-19 lockdowns are discussed separately.

Selection of sources of evidence
Searches of electronic databases identified 3962 studies, after deletion of duplicates.Three qualitative and 26 quantitative studies met the eligibility criteria for inclusion.
Searching of the theses database identified 118 dissertations, one of which met the inclusion criteria.A PRISMA flow diagram of the process for study selection is shown in Fig. 1.

Results of individual sources of evidence and synthesis of results
The qualitative studies are reported in Table 1 and the quantitative studies in Table 2.The studies have been summarised into four main categories based on the food and eating variable explored: fruit and vegetable or other food group intake; energy-dense-nutrient-poor food intake, motivation or preference; overall diet quality, healthy eating, food intake or diet awareness; and meal consumption (Table 2).Qualitative research is included in the meal consumption theme.Studies conducted during COVID 19 are a fifth category.Some studies have results that are relevant to multiple themes.

Fruit and vegetable or other food group intake
Ten studies investigated fruit and vegetable consumption in Europe (Cruz-Piedrahita et al., 2022;Cuesta-Lozano et al., 2020;Delerue Matos et al., 2021;Hammig, 2019;Kalousova, 2014;Kobayashi & Steptoe, 2018;Richard et al., 2017), Japan (Yamaguchi et al., 2020), Canada (Ingham) and the US (Sahyoun et al., 2005).Nine of the studies found lower intakes in persons with higher loneliness or social isolation.The study that found no relationship only investigated a subset of participants aged 30-44 years that were living with other people and did not use validated tools to assess intake (Cuesta-Lozano et al., 2020).Nine of the studies used cross-sectional and one used a longitudinal design (Kobayashi & Steptoe, 2018).Eight of the nine studies finding associations conducted multivariable analyses with all or some associations remaining significant in every study (Cruz-Piedrahita et al., 2022;Delerue Matos et al., 2021;Hammig, 2019;Ingham;Kalousova, 2014;Kobayashi & Steptoe, 2018;Richard et al., 2017;Sahyoun et al., 2005).However, only two of the studies used a validated tool to assess fruit and vegetable consumption (Ingham;Sahyoun et al., 2005) with other studies using a variety of methods to assess and categorise intake levels.An anomaly was the finding from Cruz-Piedrahita et al. (2022).This study used a multi-exposure confounding model and found that persons with high fruit and vegetable intake, who lived in areas with the highest quartile of green space and engaged in high physical activity levels, had higher odds for loneliness (Cruz-Piedrahita et al., 2022), although this was of borderline significance and requires further investigation.
One study also assessed other food groups, investigating the difference between intake of non-dairy protein foods, dairy foods and whole grain intake in people with a low, medium or high social environment group (Ingham).Social environment group comprised four subgroups including a social isolation index (Table 2).For high compared to low social environments only there was a difference in protein and dairy foods and there was no difference in whole grain intake between groups.

Energy-dense-nutrient-poor food intake, motivation or preference
Four studies investigated the associations between loneliness and energy-dense-nutrient-poor food intake (also known as junk food and beverages or discretionary foods) (Henriksen et al., 2014;Yamaguchi et al., 2020), motivation (Sirois & Biskas, 2023) or preference (Alalwan et al., 2019;Sirois & Biskas, 2023) using cross-sectional study designs.Henriksen et al. reported that in women in Norway loneliness was significantly associated with higher intake of sugar and artificially sweetened beverages (soft drink, juices) in both bivariate and multivariable analyses.However, the association between loneliness and artificially sweetened beverages was not significant after multivariable adjustment (Henriksen et al., 2014).Yamaguchi et al. found no associations between salty food intake frequency and loneliness in women and men in Japan (Yamaguchi et al., 2020).Sirois and Biskas found a correlation between loneliness and coping via eating palatable foods and palatable foods preference in UK participants (Sirois & Biskas, 2023).Alalwan et al. reported that 40% of undergraduate students in Bahrain preferred to eat high-energy salty foods when feeling lonely, 28% preferred high-energy sweet foods, 28% preferred low-energy sweet foods, and 4% preferred low-energy salty foods with no significant differences between male and female, single or married or weight (BMI) categories (Alalwan et al., 2019).Overall, three of the four studies found associations between loneliness and higher intake of foods usually considered harmful to health in excess, however only one of the tools used to measure the food/eating variables was validated (Sirois & Biskas, 2023).
Six studies assessed diet quality, variety or healthiness using a range of validated (Bizzozero-Peroni et al., 2022;Bloom et al., 2017;Luo & Hendryx, 2022;Sahyoun & Zhang, 2005;Yokoro et al., 2023) and unvalidated (Yamaguchi et al., 2020) tools.All five studies that used validated tools found associations where higher loneliness or social isolation were associated with poorer diet/eating.Three of the studies assessed whether the association remained after multivariable adjustment with all finding the association remained significant (Bloom et al., 2017;Sahyoun & Zhang, 2005;Yokoro et al., 2023).Bizzozero-Peroni et al. assessed diet using the dietary inflammatory index and found that persons in quartile 4 (proinflammatory) had higher loneliness scores than persons in quartile 1 (anti-inflammatory).However multivariable analyses in this study used depression as the outcome with diet the explanatory variable and loneliness one of multiple covariates (Bizzozero-Peroni et al., 2022).Luo and Hendryx (2022) combined diet with low physical activity and smoking into a risky behaviour index for their multivariable analyses showing that the association between social isolation and mortality risk was significantly mediated by the risky health behaviour index.
One study investigated the association between feeling lonely and overall food intake using the validated Emotional Eating Questionnaire (Alalwan et al., 2019).They found that amongst university students 22.4% of participants reported eating less food than usual and 32% more food than usual when feeling lonely with no significant differences between male and female, single or married or weight (BMI) categories regarding food intake when lonely.Richard et al. (2017) assessed diet awareness as a yes/no answer to "do you pay attention to specific aspects of your diet?" and found that lonely men indicated more awareness of their diets than men who never felt lonely however there were no significant associations overall or amongst women only.

Meal consumption
Five studies investigated meal consumption with two using quantitative and three using qualitative designs.Both quantitative studies used cross-sectional designs.One found no association between loneliness and breakfast consumption among people in Japan (Yamaguchi et al., 2020) and the second reported that social isolation was linked to a lower likelihood of eating three meals a day in the Czech Republic and Hungary, but not in Poland (Kalousova, 2014).Only Kalousova conducted multivariable analysis and neither used validated tools.
Two of the qualitative studies were conducted in the UK and one in Norway.All recruited older adults and used focus groups or CurroCus® group interviews and aimed to explore: food choice and dietary habits of community-dwelling older adults in locations with high levels of deprivation (Whitelock & Ensaff, 2018); the influences on diet in a group of community-dwelling older adults (Bloom, Lawrence, et al., 2017); or to focus on the ability of elderly people to use food and meals as strategies to cope with loneliness (Hansen, 2022).In the first two studies the influence of factors relevant to loneliness or social isolation on food were discussed.Bloom et al. reported that "discussion about social activities and isolation, community spirit and loneliness, indicated the importance of social engagement as an influence on diet" (Bloom et al., 2017(Bloom et al., , p. 2690)).Whitelock and Ensaff similarly reported that many discussions emphasized the influence of living alone and corresponding social isolation and loneliness (Whitelock & Ensaff, 2018).The negative aspects of being on their own were reported, especially when participants were eating alone.For example, one participant stated, "It's amazing what difference it makes i'nt it?"(Whitelock & Ensaff, 2018, p. 413).Bloom et al. reported that women were more likely to speak about the loneliness or difficulty of eating alone.In both studies the impact of not having somebody else to cook for was raised.Hansen had a contrasting focus exploring how food and meals are used to cope with loneliness.Using a grounded theory approach the findings fell into the two main categories of togetherness and life action.Togetherness was defined as "an expression for how older people try to connect with people in their lives to minimize the likelihood of loneliness" (Hansen, 2022, p. 416) and the use of food and meals as a tool to avoid loneliness was discussed.Within life actions relevant sub-categories of purchase-making, appetizing activities and food activities and their roles in ameliorating loneliness were considered.Gender differences were raised.For example, women reduced loneliness by visiting cafes with friends and men learned to cook after becoming widowers leading to dinners with family and new friends.The importance of living conditions that feature common areas that allow groups to gather for meals and the role of a social atmosphere in creating appetite also emerged (Hansen, 2022).
While methods varied, six studies investigated perceived diet quality or healthiness during lockdown.Four of the six studies found an association between increased loneliness or loneliness and social isolation and lower diet quality, healthy eating or Mediterranean diet adherence (Brown et al., 2023;García-Esquinas et al., 2021;Paltrinieri et al., 2021;Vasan et al., 2022) although this was not significant at the multivariable level in one (Brown et al., 2023).The fifth study found that loneliness did not predict a decline or increase in healthy eating during lockdown (Robinson et al., 2020).The sixth study framed the analyses as self-reported better, worse or no dietary changes being predictors for perceived social isolation (Ingram et al., 2022a).Results showed perceived isolation scores were lowest in participants reporting their diets were about the same and highest in participants that reported their diets were either a lot worse or a lot better during lockdown.Ability to compare results with studies investigating overall diet quality or healthy eating outside of lockdown is limited as most reported changes during lockdown.However, the two studies that assessed diet quality using validated tools found results consistent with those reported outside lockdown (García-Esquinas et al., 2021;Vasan et al., 2022).
One study found a significant association between loneliness and persistently eating more than usual during lockdown (Herle et al., 2021).There were no associations with persistently eating less than usual, decreasingly eating more than usual or increasingly eating more than usual (Herle et al., 2021).One study assessed snacking on foods high in sugar, salt or fat outside of meals in the week prior to and weeks following lockdown (Mata et al., 2021, p. 287).Snacking was correlated with higher levels of loneliness, consistent with most of the findings investigating loneliness and high-energy-nutrient-dense foods outside of lockdown although changes in snacking behaviour were not related to loneliness (Mata et al., 2021, p. 287).

Critical appraisal of sources of evidence
The overall quality of the studies included in this scoping review was inconsistent (Supplementary Tables 3 and 4).Of the 26 quantitative studies, 19 used multivariable analysis to assess whether the relationship between loneliness/social isolation and food/eating behaviours remained after accounting for possible confounding factors.Associations remained significant for some or all analyses in seventeen of these studies.However, the range of covariates assessed varied so it is possible that the associations found in the studies could be confounded or modified by other factors such as other emotions, mental health conditions, physical health and socioeconomic position.Seven of the quantitative studies specified using validated tools to assess both key variables with implications for accuracy and repeatability of the findings.Only five studies examined the key variables using a longitudinal design therefore the temporal relationship between exposure and outcome is not always known.Fourteen of the studies specified random selection of participants and response/participation rates varied so generalisability should not be assumed.
Qualitative methodologies were deemed appropriate for all three studies included and two of three provided clear justification for why the design was appropriate.For all three studies there was incongruence with the data collection methods in that the aims investigated individual behaviours, however all studies collected data in groups which may have influenced the discussion.All the papers showed consideration of ethical issues however the relationship between the researcher and participants was not adequately considered.

Summary of evidence
This scoping review provides the first summary of evidence that has investigated the association between loneliness or social isolation and food and eating behaviours.Twenty-six quantitative and three qualitative studies published between 2000 and 2023 met the eligibility criteria.Eight of the quantitative studies were conducted within the context of COVID-19 lockdowns.The studies were grouped into five categories: fruit and vegetable or other food group intake; energy-densenutrient-poor food intake, motivation or preference; overall diet quality, health eating, food intake or awareness; meal consumption and changes or intake within the COVID-19 lockdown context.
Overall, 24 of 26 quantitative studies found at least one association between loneliness or social isolation and one or more food/eating behaviours that would usually be considered harmful to health including lower fruit and vegetable intake, higher intake of energy-dense-nutrientpoor foods and lower overall diet quality.Quantitative evidence on loneliness or social isolation and meal consumption was scarce and mixed.However, all three of the qualitative studies featured the importance of the relationship between loneliness or social isolation and meal consumption.Although not the focus of this research, qualitative evidence provided insight into some of the possible underlying reasons.For example, by discussing the influence of eating alone on shopping, cooking and eating.The bidirectional nature of loneliness was illuminated to some extent in these discussions.Hansen in particular focused on the role of food and meals as strategies to minimize loneliness although the impact of eating alone on loneliness was also raised in other studies.These insights align with the research on the significance of commensality (Andersen & Brünner, 2020;Björnwall et al., 2023;Dunbar, 2017;Fischler, 2011;Jönsson et al., 2021).
Restrictions during the COVID-19 pandemic created altered circumstances that impacted loneliness and social isolation (Knox, Karantzas, Romano, Feeney, & Simpson, 2022;Lazzari & Rabottini, 2022).Overall, results of the eight studies suggest the association between loneliness or social isolation and food/eating behaviours remained when people were in lockdown circumstances with most finding a relationship involving lower diet quality, a reduction in diet quality or snacking or eating more than before.Only two of these studies used a validated tool to assess diet (García-Esquinas et al., 2021;Vasan et al., 2022), likely due to the time restrictions in relation to the purpose of the study.It is beyond the scope of this paper to discuss detailed policy and enforcement variations between the countries however broadly, Australia, Italy, Spain and the UK implemented more restrictive and legally enforced measures (Cascini et al., 2022;Xylogiannopoulos, Karampelas, & Alhajj, 2021), Germany implemented comparatively milder and shorter restrictive measures (Cascini et al., 2022) and Japan a less obtrusive, requested, mild lockdown (Ingram et al., 2022a).One study was designed to compare the contrasting containment measures used in Scotland and Japan however provided results on the relationship between perceived isolation and dietary changes overall, not showing if the relationship differed transnationally (Ingram et al., 2022a).The study conducted in Germany reported that snacking was correlated with higher loneliness, although changes in snacking behaviour before and following restrictive measures were not related to loneliness (Mata et al., 2021, p. 287).Comparison between this and the countries with stricter containment approaches needs to be cautious because they used overarching measures of changes in diet (Brown et al., 2023;Herle et al., 2021;Ingram et al., 2022a;Paltrinieri et al., 2021;Robinson et al., 2020), diet quality (Vasan et al., 2022) or Mediterranean diet adherence (García-Esquinas et al., 2021).The wider literature investigating changes in diet within COVID-19 lockdowns has similarly reported dietary changes that are mostly harmful, although research has also found some changes that could be beneficial to health such as a tendency towards decreased fast-food and ordered food consumption (Bakaloudi, Jeyakumar, Jayawardena, & Chourdakis, 2022).How these insights relate to loneliness and isolation in usual circumstances is uncertain.Findings therefore should not be generalised to outside lockdown, however awareness could be important in any future situations where containment measures are used for health reasons.
As considered in the introduction the relationship between the key variables is likely to be bidirectional and influenced by a complex range of factors.A strength therefore is that nineteen of the 26 quantitative studies did include multivariable analysis although whether they were able to fully assess the role of confounding or effect modifying variables is uncertain.The five longitudinal studies are important to explore evidence on the direction of the relationship between loneliness, social isolation, food and eating.Three of these were within the context of changes in diet following COVID-19 restrictions (García-Esquinas et al., 2021;Herle et al., 2021;Mata et al., 2021, p. 287).Garcia-Esquinas found that both loneliness and social isolation were associated with worsening Mediterranean diet adherence in Spain and Herle found that higher loneliness increased the odds of persistently eating more than usual in the UK.While Mata found loneliness was correlated with more snacking in Germany, it was not linked with changes in snacking from before to during lockdown.Two longitudinal studies were conducted in usual circumstances (Bloom et al., 2017;Kobayashi & Steptoe, 2018).Bloom found no association between baseline social isolation measures and changes in prudent diet score, whereas Kobayashi found that higher social isolation was associated with reduced likelihood of persistently eating five or more serves of fruit and vegetables.The study assessing loneliness found no associations with consistent intake of five or more serves of fruit and vegetables (Kobayashi & Steptoe, 2018).Overall longitudinal data is currently scarce, and results and study characteristics vary, hampering the ability to draw conclusions about temporal interactions between loneliness and food in normal or restrictive circumstances.However, the significant relationships observed do indicate the need for further investigation.
Caution in the interpretation of findings is also needed due to the range of measures used to assess both loneliness or social isolation and food and eating behaviours.An important limitation is the inconsistent use of validated tools to measure loneliness/social isolation and food/ eating behaviours with only seven studies specifying use of validated tools to assess both.Methods also did not consider different domains of loneliness such as emotional, social, intimate, relational and collective (Prohaska et al., 2020).
It is a strength that many of the studies recruited participants from wide age ranges, although all the qualitative and ten of the quantitative studies were restricted to older participants.Much of the focus on loneliness is in older people as they may be more at risk of loneliness or social isolation due to higher incidence of living alone, inadequate support, bereavement, retirement, declining functional capacity and disruptive life events (Donovan & Blazer, 2020).However, loneliness is also prevalent in younger adults aged 18-25 years (Full stop inclusion; Ending Loneliness Together, 2020).Further evidence of the importance of loneliness and isolation in younger adults comes from a large multinational study which found that loneliness decreased with age (Barreto et al., 2021) and a study located in Northern California found that adults aged 21-30 years reported twice as many days lonely and isolated than adults aged 50-70 years (Child & Lawton, 2019).
Only twelve of the 29 papers reported ethnic background of participants and of these, most participants identified as white.Research indicates that loneliness and its conceptualization are affected by culture, although understanding these relationships is complex (Barreto et al., 2021;Dykstra, 2009).Given that our experiences of all aspects of food and eating are also associated with ethnicity and culture (Bennett, Bardon, & Gibney, 2022;Enriquez & Archila-Godinez, 2022) research involving people from more diverse ethnic backgrounds and those living with socioeconomic disadvantage could provide better insight into human experiences of loneliness or social isolation and how this interacts with food and eating.
The lower proportions of male participants is another limitation as evidence indicates that loneliness or social isolation are influenced by gender, although results are mixed.As loneliness is a subjective measure, self-reporting measures of loneliness can result in social desirability bias (Caputo, 2017).Studies have reported differences depending on the tool used to assess loneliness (Pinquart & Sorensen, 2001).A study that did find higher loneliness in men commented that certain conditions may be needed for men to speak about loneliness that may have been provided through their online assessment setting (Barreto et al., 2021).Another commented that measures of loneliness requiring a person to self-label themselves tend to report higher numbers of females reporting being lonely than if they avoid having to self-describe themselves as lonely (Heinrich & Gullone, 2006).Overall, research does indicate the importance of including men in investigations of the determinants and consequences of loneliness and considering the impact of stigma in the methodology.The lack of sex and gender diverse people in the studies eligible for this review is also an important gap.

Relevance to key groups
The results found in this paper could be of relevance to multiple groups including policy makers, health professionals and consumers.There appears to be increasing recognition of the importance of loneliness and social isolation and their consequences at the political and population levels, for example as indicated by the establishment of: a Minister for Loneliness in the UK (HM Government, 2018); funds with the purpose of addressing loneliness (Queensland Government, 2022); and organisations with the purpose of reducing loneliness (Ending Loneliness Together., 2020).It has been identified that evidence on risk factors for loneliness is often conflicting and confusing for those developing policy and services (Prohaska et al., 2020).The need for better knowledge and understanding of food choice determinants to inform policy has also been acknowledged (Leng et al., 2017).This review therefore provides a summary of current evidence and gaps around the relationship between eating and loneliness.This could be especially applicable given the potential for inclusion of food-based programs as interventions to address loneliness (Middleton, Velardo, Patterson, & Coveney, 2022, pp. 1-23).
The findings of this review are also likely to be of relevance to multiple health professionals including dietitians, nutritionists, psychologists and others.Results could be of particular interest to dietitians and nutritionists who need to consider the potential impact of loneliness or social isolation on food choices when developing strategies for prevention or treatment at individual, group or population levels (Ending Loneliness Together., 2020).The bidirectional nature of these relationships is important to consider in interventions to address loneliness and social isolation given the important role of food and eating in social interactions (Dunbar, 2017;Fischler, 2011) and some interventions to address loneliness centre around food and eating (Flinders University, 2021; Middleton et al., 2022, pp. 1-23).The review is relevant to members of the public by highlighting the possible importance of loneliness and social isolation influencing food and eating behaviours and the use of food to ameliorate loneliness and may provide insight into personal choices.

Recommendations for research
The relative consistency of results identified despite limitations indicates the importance of further investigating the relationship between loneliness, social isolation, food and eating.As considered in the introduction the interactions between and pathways involving loneliness, social isolation, food, eating and health are complex and multifaceted.There are also plausible explanations for the relationships between these variables being multidirectional as well as mediated, modified or confounded by each other and additional factors.Findings from the longitudinal studies found in this review highlight the need for more research using longitudinal designs, such as cohort studies, that better enable investigation of temporal relationships.The design also should include multivariable analysis and use validated tools to assess loneliness and social isolation and food/eating variables.Evidence that uses consistent tools to measure the variables of interest would facilitate easier study comparison.Inclusion of both loneliness and social isolation would also allow comparison of whether these constructs are related to food and eating in contrasting ways.Considering different domains of loneliness could also provide further insight into relationships with food and eating.Another recommendation for future quantitative research is to include more diversity in the sex and gender, ethnicity and age of participants.
Further qualitative research is warranted to expand and deepen understanding of the quantitative evidence on the relationships between loneliness, social isolation, food and eating.Studies included in this review already give some insight into the possible reasons.Further evidence is needed that involves more diverse participants including age, gender, socioeconomic and ethnicity groups.Qualitative research could include opportunities for co-design with people with lived experience of loneliness or social isolation to better understand these problems and determine possible solutions.The qualitative studies identified in this review all used group interviews.Interviews with individuals exploring food and eating behaviours and loneliness or social isolation would also be valuable given that many of the concepts explored involve individual behaviours.Qualitative research approaches that could be appropriate for exploring these areas include ethnography, netnography and phenomenology.
Binge eating and other disordered eating behaviours were outside the scope of this review however they appeared multiple times in search results suggesting that this could be an important topic for future reviews.This could inform whether disordered eating behaviours are a part of the reasons for the associations observed in this review.

Limitations of the scoping review
This scoping review contains only literature published in English, reducing the number of articles included which may have been eligible.It also includes only studies that investigate high income countries due to the potential differences in social environments and their impact on social isolation and loneliness.Further investigation into the topics of loneliness, social isolation and behaviours relating to food and eating in low-and middle-income countries could be explored given the identification of loneliness or social isolation across cultures and socioeconomic levels.The prevalence of social isolation between high income, middle income and low-income countries is similar.However, the cause of the social isolation may be different depending on the economy of the area as education and employment may affect social isolation and connectedness (Naito, et al., 2021).Another limitation was around varying terminology used for social isolation.Some studies may have used different terms, such as social contact or networks, social inclusion/exclusion or social cohesion with reference to objective measures of contact with other people, that may not have been identified with the search terms used.Our study protocol also excluded studies focussing on social support which may have excluded studies with methods overlapping some studies that were included.The use of the term social isolation as a synonym for lock-down in post 2020 literature was another complication.The use of terms to capture ultra-processed or high-energy-nutrient-dense poor foods may also not have captured the wide range of different terms used formally and colloquially to discuss these food types.

Conclusions
The aim of this scoping review was to explore the evidence available that investigates the association between social isolation or loneliness and food and eating-related behaviours.This review identified broad consistency where associations between loneliness or social isolation and behaviours relating to food and eating were almost all harmful, yet some studies found no associations.Whilst study quality and methodology were inconsistent, the findings indicate the possibility that part of the observed relationship between loneliness or social isolation and health could be due to eating behaviours, justifying the need for further research to provide better evidence.The potential bidirectional nature of the relationship between food/eating and loneliness/social isolation needs to be further explored to understand if the focus should be dietary interventions to potentially impact loneliness, addressing loneliness to potentially improve food and eating behaviours or both.Overall, this review illuminates the interconnectedness of food and eating with social and emotional well-being and that addressing food behaviours needs to consider the complex matrix of factors influencing intake.Further research on this topic could allow for a clearer understanding on how to apply the findings from this scoping review in practice and the role of food and eating in the relationship between loneliness or social isolation and health outcomes.

Table 1
Study characteristics, methods and results of qualitative studies.
a Where stated in methods.bEthnicity of participants included when available.K.Hanna et al.

Table 2
Study characteristics, methods and results of quantitative studies.

Table 2
(continued ) BV Analysis: Persons that ate no daily servings of fruits or vegetables were more likely to be socially isolated than persons who ate daily servings in the Czech Republic (7% vs. 4%, p < 0.05) and Poland (11% vs. 3%, p < 0.001) NS in Hungary.MV Analysis: Social isolation associated with lower likelihood of daily fruits or vegetable intake OR (95% CI) in the Czech Republic (0.62 (0.38-1.02), p < 0.10) and Poland (0.35 (0.16-0.75), p < 0.01).NS in Hungary.(continued on next page) K.Hanna et al.

Table 2
(continued ) (continued on next page) K.Hanna et al.

Table 2
(continued ) (continued on next page) K.Hanna et al.

Table 2
(continued ) NP = not presented; NS = not significant; OR = odds ratio; CI = confidence interval; SD = standard deviation; BV = bivariate; MV = multivariable; SE = standard error; F = female; W = women; UCLA=University of California Los Angeles. 5a collected pre-confinement; t1 data collected during state of alarm.6Otheralsopresented in gender (0.8%).aWhere stated in methods.bEthnicity of participants included when available.cHigherscoresindicate higher loneliness or social isolation for all studies when measured as continuous variables.d Pr based on data from four cohort studies, three of which were eligible.Information presented for eligible cohorts only.K.Hanna et al.