Diets low in fruits, vegetables and whole grains, and high in saturated fat, salt and sugar, are the major contributors to the burden of chronic diseases globally( Reference Lim, Vos and Flaxman 1 ). Previous research, and studies in this issue of Public Health Nutrition, show that unhealthy diets are more commonly observed among socio-economically disadvantaged groups and are key contributors to their higher rates of chronic disease( Reference Davey Smith and Brunner 2 – Reference Martikainen, Brunner and Marmot 4 ).
Most research examining socio-economic inequalities in diet and body weight has been descriptive and has focused on identifying the nature, extent and direction of the inequalities. These types of studies are clearly necessary and important. We need, however, to move beyond description of the problem and focus much more on the question of why inequalities in diet and body weight exist. Furthering our understanding of this question will provide the necessary evidence base to develop effective interventions to reduce the inequalities. The challenge of tackling dietary inequalities, however, does not finish here: a maximally effective approach will also require equity-based policies that address the unequal population distribution of social and economic resources, which is the fundamental root-cause of dietary and body weight inequalities.
Current evidence on socio-economic inequalities in diet and body weight
Studies examining the association between socio-economic position (SEP) and diet have predominantly focused on working-aged adult populations in high-income countries. This research finds that low socio-economic groups are less likely to consume fruits and vegetables, fish, low-fat dairy products and whole grains, and more likely to consume red and processed meat and fast food( Reference De Irala-Estevez, Groth and Johansson 5 – Reference Maguire and Monsivais 11 ). Socio-economically disadvantaged adults in high-income countries also have lower intakes of important micronutrients including folate, vitamin C, vitamin D, Ca, Fe, Cu and Se( Reference Novakovic, Cavelaars and Geelen 12 ), although there is limited evidence that SEP is associated with total energy or macronutrient intake( Reference Darmon and Drewnowski 7 , Reference Giskes, Avendano and Brug 8 ). Studies describing associations between SEP and diet among children, adolescents and the elderly in high-income countries also show that individuals from disadvantaged backgrounds are more likely to have food and micronutrient intakes that are least consistent with dietary guidelines or reference intakes( Reference Novakovic, Cavelaars and Geelen 12 , Reference Hanson and Chen 13 ).
In this issue of Public Health Nutrition, four studies examine the association between SEP and diet in high-income countries. Ahmadi et al.( Reference Ahmadi, Black and Velazquez 14 ) found that during the school day, Canadian children (aged 10–15 years) with less educated parents were less likely to consume vegetables and more likely to consume sugar-sweetened beverages. Manios et al.( Reference Manios, Moschonis and Androutsos 15 ) show that European children (aged 10–12 years) from low-SEP families less frequently consumed breakfast daily. Khalaf et al.( Reference Khalaf, Westergren and Berggren 16 ) present a complex and mixed picture of the relationship between SEP and dietary habits among female university students (aged 18–25 years) in Saudi Arabia; low parental SEP was associated with both unhealthy and healthy dietary habits, depending on the food type and measure of SEP that was used. According to Dijkstra et al.( Reference Dijkstra, Neter and van Stralen 17 ), lower educated Dutch respondents (aged 55–85 years) were less likely to comply with the vegetable guidelines and respondents from lower-income households were less likely to adhere to the fruit and fish guidelines. Taken together, these four studies highlight that dietary inequalities occur at all points across the life course; hence intervention efforts designed to reduce the inequalities should be life-course tailored and targeted accordingly.
Reviews of the association between SEP and body weight in high-income countries( Reference McLaren 18 , Reference Rokholm, Baker and Sorensen 19 ) show that adults from disadvantaged backgrounds are more likely to be overweight or obese, with the relationship being consistently strong and graded for women and somewhat weaker and less consistent for men. Negative associations between SEP and body weight in high-income countries have also been observed in childhood( Reference Shrewsbury and Wardle 20 , Reference Wang and Lim 21 ). Moreover, during the next few decades inequalities in obesity are predicted to widen in some countries( Reference Backholer, Mannan and Magliano 22 ). In this issue of Public Health Nutrition, overweight/obesity was found to be significantly more likely among socio-economically disadvantaged schoolchildren in Europe (Manios et al.( Reference Manios, Moschonis and Androutsos 15 )) and among female university students from less advantaged households in Saudi Arabia (Khalaf et al.( Reference Khalaf, Westergren and Berggren 16 )).
The relationship between SEP and diet in low- and middle-income countries has been investigated in a relatively small number of studies. A recent systematic review of thirty-three studies from seventeen low- and middle-income countries concluded that, in general, persons of high SEP have healthier diets, as indicated by higher fruit and vegetable consumption, diet diversity and quality, and higher intakes of vitamin A, vitamin C and Fe( Reference Mayen, Marques-Vidal and Paccaud 23 ). Socio-economically advantaged groups, however, also have higher intakes of energy, cholesterol and saturated fats and lower intake of fibre. In the current issue, Landais et al.( Reference Landais, Bour and Gartner 24 ) show that Moroccan women of childbearing age from high socio-economic households were significantly more likely than their disadvantaged counterparts to consume fruit, and fruit and vegetables combined, and they ate more different types of fruit.
As countries develop economically and shift from being a low- to a middle-income society, they experience a nutrition transition( Reference Popkin and Gordon-Larsen 25 , Reference Popkin 26 ), characterised by a high consumption of fibre- and grain-rich diets in low-income countries (and higher relative levels of physical activity) moving to increased consumption of sugar, refined grains, animal fat and protein (and lower relative activity) as average wealth of the country grows( Reference Mayen, Marques-Vidal and Paccaud 23 , Reference Malik, Willett and Hu 27 ). The relationship between SEP and body weight in low- and middle-income countries tends to change in unison with a country’s stage in the nutrition transition. In low-income countries, where socio-economically advantaged groups have greater access to energy-dense foods, the association between SEP and body weight is positive for men, women and children. In middle-income countries, where advantaged groups have greater access to more expensive, low-energy-density, nutrient-rich foods (e.g. fruits, vegetables, unrefined grains), the association between SEP and body weight tends to be negative for women and mixed for men( Reference Dinsa, Goryakin and Fumagalli 28 ). In this issue, Christine et al.( Reference Christine, Diez Roux and Wing 29 ) provide a within-country case study of how the association between SEP and body weight changes concomitantly with increases in economic development. Against a backdrop of increasing BMI for both men and women in Argentina between 2005 and 2009, there was an overall inverse graded association between education and BMI for women, which was stronger in geographic regions at higher levels of per capita gross domestic product and weaker at lower levels.
Studies in this issue of Public Health Nutrition that examined associations between SEP, diet and body weight typically investigated these relationships using socio-economic indicators based on parents’ or respondents’ education, occupation/employment or income. Importantly, although moderately correlated, measures of SEP are not conceptually interchangeable( Reference Lallukka, Laaksonen and Rahkonen 6 , Reference Geyer, Hernstom and Peter 30 ) and reflect different pathways linking SEP with diet and body weight( Reference Galobardes, Lynch and Smith 31 , Reference Turrell, Hewitt and Patterson 32 ). Future dietary research should continue to investigate these relationships using a range of different socio-economic indicators within the same study( Reference Braveman, Cubbin and Egerter 33 ), as each indicator tells us something different about how and why SEP is related to diet and body weight, and the findings produced by each indicator point to different intervention options.
Explanations for socio-economic inequalities in diet and body weight
Understanding why socio-economic groups differ in their dietary behaviours and intakes is a necessary precursor to the development and implementation of appropriately designed and targeted interventions, and is integral to their long-term effectiveness. Explaining the existence, persistence and widening of socio-economic inequalities in diet is a complex and challenging task, in part because socio-economic groups differ on myriad interacting factors that influence their dietary behaviours, food choices and intakes. These factors include (but are not limited to): (i) access to, and capacity to afford, ‘healthy’ food; (ii) availability of unhealthy food; (iii) access to public and private transport; (iv) neighbourhood safety; (v) social support and peer networks; (vi) time; (vii) adequate income; (viii) knowledge and skills; (ix) beliefs, values, attitudes and motivations; (x) social norms, preferences and habits; (xi) customs, familiarity and tradition; and (xii) perceived capabilities (i.e. self-efficacy). In this issue of Public Health Nutrition, three studies add to this stock of knowledge by undertaking mediation analyses to elucidate the pathways by which SEP is related to diet and body weight. Dijkstra et al.( Reference Dijkstra, Neter and van Stralen 17 ) found that taste preference mediated the association between income and adherence to the Dutch dietary guidelines for fruit, while the perception that fish is expensive mediated the association between income and meeting guidelines for fish consumption. Ahmadi et al.( Reference Ahmadi, Black and Velazquez 14 ) observed that socio-economic differences in daily vegetable intake among Canadian school-aged children were mediated by parental norms; and Manios et al.( Reference Manios, Moschonis and Androutsos 15 ) found that the relationship between childhood SEP and overweight/obesity was mediated by daily breakfast consumption.
Mediation studies are generating an evidence base that will usefully inform policy and interventions designed to tackle dietary inequalities. To more fully understand and address these inequalities, however, we also require more complex, purpose-designed, multilevel and multidisciplinary dietary research programmes and studies that are conceptualised and conducted from a social determinants of health perspective. This work would be informed by ‘real world’ frameworks that articulate the complexity and relative contribution of social, economic and environmental factors and life-course processes that are fundamental in differentiating the dietary behaviours and intakes of different socio-economic groups. Ideally, these frameworks will be ‘context sensitive’ and acknowledge that many individual-level dietary influences (e.g. psychosocial factors) are shaped and circumscribed over time by the lived environments that we are exposed to during everyday life.
Interventions to reduce socio-economic inequalities in diet and body weight
At present, there is limited evidence to guide policy-making decisions about how to design and implement effective interventions to reduce socio-economic inequalities in diet and body weight. In the context of inequalities’ research, intervention effectiveness is usually defined in one of two ways: (i) the intervention results in a greater (healthful) improvement in the dietary outcomes of disadvantaged groups relative to their more advantaged counterparts; or (ii) the intervention produces a similar dietary improvement across all socio-economic groups without inadvertently widening the inequalities( Reference Kristjansson, Robinson and Petticrew 34 ). There are numerous challenges to intervention effectiveness given the large number of interacting factors (operating at multiple levels and across the life course) that differentially influence the diets of socio-economically advantaged and disadvantaged groups. Arguably, for maximum effectiveness the focus and design of dietary interventions should reflect, or at least be sensitive to, this complexity.
Despite a small intervention evidence base, and heterogeneity in intervention quality, duration and population representativeness (and under-representation of disadvantaged groups in particular), a consensus about the types of interventions likely to be effective in narrowing socio-economic inequalities in diet and obesity is emerging from a number of recent reviews( Reference Beauchamp, Backholer and Magliano 35 , Reference Backholer, Beauchamp and Ball 36 ). In brief, ‘agentic’ interventions – those which require individuals to make independent choices (e.g. social marketing, food labelling) – are assessed as being less likely to be effective among disadvantaged groups and hence to widen inequalities; by contrast ‘structural’ interventions – those which involve modifications to environments, contexts, settings or circumstances – are assessed as conferring equal or greater benefit to lower socio-economic groups and therefore unlikely to increase inequalities. Examples include food procurement policies, restriction of junk food advertising to children, taxes on unhealthy food products and healthy school food policies. To date, nutrition interventions have tended to be more agentic than structural( Reference Ni Mhurchu 37 ); hence the possibility exists that some well-intentioned efforts to improve dietary quality across the socio-economic spectrum may have inadvertently widened inequalities, although a number of reviews provide little compelling support for this( Reference Oldroyd, Burns and Lucas 38 – Reference Hillier-Brown, Bambra and Cairns 40 ).
In two studies in this issue, the authors call for fiscal measures (e.g. subsidies, vouchers, price discounts) as a way of increasing the consumption of fish (Dijkstra et al.( Reference Dijkstra, Neter and van Stralen 17 )) and fruit (Landais et al.( Reference Landais, Bour and Gartner 24 )) among low socio-economic groups. Targeted fiscal measures have been effective in increasing the consumption of nutrient-dense foods (e.g. fruits and vegetables) among low socio-economic groups( Reference Geliebter, Ang and Bernales-Korins 41 , Reference Waterlander, de Boer and Schuit 42 ) and therefore constitute a potentially effective public health strategy for reducing inequalities in dietary quality( Reference Ni Mhurchu 37 ).
Primavesi et al.( Reference Primavesi, Caccavelli and Ciliberto 43 ) in this issue propose a ‘nutrieconomic model’ as a method of identifying different food items (e.g. chicken, orange, anchovy, milk) within different food categories (e.g. meat, fruit and vegetables, fish, dairy) on the basis of their nutritional quality per serving and cost. When used to simulate weekly menus, the model was able to demonstrate a 30 % difference (i.e. 35 v. 48 €/person per week) in the cost of two equally healthy menus simply by varying the animal protein source. The authors posit that diets of high nutritional quality are not necessarily more expensive, rather the main issue is educating the population (and subgroups within it) about making nutritionally optimal choices. It remains an open question, however, whether the budgets of low-income families are sufficient to afford the cheapest healthy menu, even if nutritional decisions were optimised (which is itself a challenge given their lower levels of dietary knowledge).
Preventing socio-economic inequalities in diet and body weight
Dietary interventions – ranging from agentic to structural – might prove effective in (for example): (i) increasing levels of food and nutrition knowledge among low educated groups; (ii) promoting positive health-enhancing dietary attitudes, preferences and beliefs among unemployed single mothers; (iii) influencing fiscal pricing (taxes and subsidies) that result in low-income families being able to more comfortably afford healthy food; or (iv) changing school canteen policies that restrict access to unhealthy food. Clearly, these outcomes, and the interventions that generate them, are necessary components of any comprehensive (and ideally well-funded) approach to the reduction of socio-economic inequalities in diet and body weight. What these interventions cannot change, however, are the social and economic conditions that are the genesis of the inequalities that the dietary interventions are directed at addressing. So despite the effectiveness of the interventions, the least educated remain less educated, the unemployed continue to seek work, the poor remain poor and those living in disadvantaged neighbourhoods are still at the same address. In short, risk-factor-specific interventions are necessary but not sufficient in our attempts to reduce diet and body weight inequalities.
Fundamentally, social and economic inequalities cause dietary and body weight inequalities; hence, intervention efforts that focus exclusively on diet and body weight – irrespective of whether these efforts are agentic or structural – do not change the unequal socio-economic conditions that gave rise to the diet and body weight inequalities. Preventing and reducing socio-economic inequalities in diet and body weight will therefore require the simultaneous implementation of evidence-based intervention efforts targeted at these specific risk factors, in conjunction with progressive social and economic policies that result in a more equitable distribution of the fundamental determinants of health at the societal, neighbourhood, household and individual levels( Reference Loring and Robertson 44 ).
Acknowledgements
G.T. is funded by an Australian National Health and Medical Research Council Senior Research Fellowship (#1003710). S.V. is funded by a fellowship from the National Heart Foundation of New Zealand (#1614). Thanks go to Dr Belinda Hewitt at the Institute for Social Science Research and School of Social Science, The University of Queensland, St Lucia, Australia, and Dr Jerome Rachele at the School of Public Health and Social Work, Queensland University of Technology, Kelvin Grove, Australia, for useful comments on an earlier draft of this editorial.
