Socioeconomic settings and food consumption patterns of 2–5-year-old children in developed countries: a scoping review

We identified relevant articles using three electronic databases including Medline, the Cumulative Index to Nursing and Allied Health Literature (CINHAL), and Public Health Database. These databases were selected after multiple consultations with a University of Saskatchewan librarian expert in scoping reviews.

The search terms included: feeding behavior or diet or dietary behavio* AND socioeconomic factors OR income OR educational status OR social determinants of health OR occupations OR employment AND urban/or rural population OR area of residence. The inclusion and exclusion criteria was modelled after Zarnowiecki et al.’s (2014) study that looked at differing SES in children aged 9–13 years old. The inclusion criteria were: (i) children aged 2–5 years old, although due to varying definitions of preschoolers, 6-year-olds were also included; (ii) investigate influencers of child dietary patterns in relation to at least one indicator of socioeconomic factors (education, income, employment status, social class, area-level SES); (iii) studies conducted on humans; (iv) full-text papers published in peer-reviewed journals in the English language, papers that were published between 1 January 2007 and 17 September 2019; (v) studies conducted in developed countries that are classified as high-income, using the World Bank Group (2016) definition: a country with a gross national income per capita of $12 236 USD or more. High-income country filters were applied in CINHAL and Public Health Database. Papers were limited to the indicated period in the interest of staying aligned with stage 5 of the nutrition transition. In developed countries, the food environment has transitioned through several patterns over time: (stage 1) a hunter-gather diet; (stage 2) early agriculture where famine was prevalent; (stage 3) end of famine; (stage 4) overeating foods high in energy but containing more fat, sugar and processed foods; and finally (stage 5) behaviour change to combat chronic conditions (Popkin 2006). This “nutrition transition” has been linked with the rise in noncommunicable diseases like obesity (Popkin and Gordon-Larsen 2004; Popkin 2017). The rise in overweight and obesity rates in most developed countries suggest that the early1990s was the shift into stage 4 (Popkin 2006). The chronic conditions that we are currently treating and attempting to prevent is the stage of present society. Stages in the nutrition transition prior to 2007 are not within the scope of this review as the aim of this review was to look at the most current literature relevant to present-day health complications. Exclusion criteria included: (i) no statistical analysis approaches reported; (ii) studies not involving children aged 2–6 year old; (iii) studies not conducted on humans; (iv) studies in languages other than English, with no English translation; (v) studies where diet was not the primary concern of the study; (vi) studies investigating dietary intake correlates in relation to eating disorders; (vii) studies involving special groups of participants, for instance, overweight/obese children, or ill individuals; (viii) papers in low-SES groups only, with no comparison among groups of different SES; and (ix) studies conducted in countries that are classified as low-income, lower-middle-income, or upper-middle-income using the 2016 World Bank definition (countries with a gross national income of less than $12 235 USD per capita). Covidence systematic software (Veritas Health Innovation, Melbourne, Australia, 2019) was the screening tool used for this review.

We analyzed selected articles by charting study characteristics and key points. Table 1 illustrates the extracted data including: the first author, origin country, sample size and population, study design, methods, outcomes, and key findings relevant to the research question.

After charting the data, key themes and knowledge gaps were identified using a narrative synthesis by two of the co-authors (SMA, EBT). Key themes such as high parental education, low parental education, parental income, location, and other socio-demographic factors were reported. A stakeholder consultation was not held due to the low yield and logistical constraints.

The characteristics of studies included in this analysis are presented in Table 1. Studies originated from Belgium (Huybrechts et al. 2011), Canada (Pabayo et al. 2012), France (Lioret et al. 2008; Camara et al. 2015), Norway (Kristiansen et al. 2013), Spain (Sotos-Prieto et al. 2015; Torres-Luque et al. 2018), the United Kingdom (Fogarty et al. 2007; Fisk et al. 2011; Mcgowan et al. 2012; Mak et al. 2013), and the United States (Kant and Graubard 2013). Four of the articles had a longitudinal study design (Fogarty et al. 2007; Kristiansen et al. 2013; Mak et al. 2013; Camara et al. 2015). The remaining nine studies were cross-sectional (Lioret et al. 2008; Fisk et al. 2011; Huybrechts et al. 2011; Mcgowan et al. 2012; Pabayo et al. 2012; Kant and Graubard 2013; Mak et al. 2013; Sotos-Prieto et al. 2015; Torres-Luque et al. 2018). In regards to the longitudinal studies, the follow-up periods were, two years (Fogarty et al. 2007; Mak et al. 2013), four years (Camara et al. 2015), and eight years (Kristiansen et al. 2013). The data collection for all studies occurred between 1971 and 2018.

All studies included preschool-aged children from high-income developed countries. Children were between the ages of 2 and 6 years. All the studies included a parental survey or interview about SES variables except for two that used postal codes (Fogarty et al. 2007; Pabayo et al. 2012), and one that considered participants’ residence in rural or urban areas. Parents either filled out Food Frequency Questionnaires (FFQs), food diaries, or questionnaires on diet quality for themselves and (or) their child(ren). Studies took place via preschools/schools (Fogarty et al. 2007; Huybrechts et al. 2011; Mcgowan et al. 2012; Rideout et al. 2015; Sotos-Prieto et al. 2015; Torres-Luque et al. 2018), hospitals (Camara et al. 2015), the parent’s home (Lioret et al. 2008; Fisk et al. 2011; Kant and Graubard 2013; Mak et al. 2013), public health units (Pabayo et al. 2012), or by mail (Kristiansen et al. 2013). Sample sizes ranged from 340 (Lioret et al. 2008) to 12 403 (Kant and Graubard 2013) with nearly equivalent total male (n = 13 051) and total female (n = 2767) participants. The sample size for males and females are only approximations, as one of the articles did not distinguish between males and females.

Dietary intake and dietary behaviour were collected using FFQs, food diaries, and surveys completed by parents/guardians or caregivers. One study examined child (3–5 years) self-report data in addition to parent-reported data (Sotos-Prieto et al. 2015). From the information collected, dietary data were assessed by either fruit and vegetable intake (Fogarty et al. 2007; Mcgowan et al. 2012; Mak et al. 2013), food intake (Lioret et al. 2008; Pabayo et al. 2012), dietary habits (Kant and Graubard 2013; Sotos-Prieto et al. 2015), dietary score (Fisk et al. 2011; Torres-Luque et al. 2018), micronutrient intake (Huybrechts et al. 2011), or dietary patterns derived from principal component analysis (Fisk et al. 2011; Kristiansen et al. 2013; Camara et al. 2015).

Parental education was the most common indicator of socioeconomic status in the selected articles. Two studies used parental occupation as the sole indicator of socioeconomic status (Lioret et al. 2008; Mak et al. 2013). One study used the UK National Statistics Socioeconomic Classification (NSSEC5) as an SES household indicator, which is a standardized measure in the United Kingdom (Mak et al. 2013). The NSSEC5 uses employment relations and conditions surrounding occupation to determine socioeconomic status (Great Britain Office for National Statistics 2005). One study assigned high SES to those with executive or professional jobs, middle SES to employees or technician jobs, and low SES to other jobs or the unemployed (Lioret et al. 2008). One study used a combination of parental education, occupation, employment, household income, presence of an older sibling, and childcare arrangements (Camara et al. 2015). Postal code was used by two studies to indicate SES (Fogarty et al. 2007; Pabayo et al. 2012). Lastly, family income as an indicator of SES was used in three of the studies, in combination with at least one other SES indicator such as parental educational level (Kant and Graubard 2013; Camara et al. 2015; Sotos-Prieto et al. 2015).

Our search yielded one article that explored the effect of urban and rural location to diet (Torres-Luque et al. 2018). The results of the study indicated that there were no differences in adherence to the Mediterranean diet between urban and rural areas in Spain. The small number of studies did not allow us to draw any conclusions regarding the potential associations between urban and rural location and dietary patterns of preschool-aged children. From the studies that focused on SES, there were no mentions of the distribution of participant residence in terms of rural or urban, and only one study suggested that their findings might be limited to urban settings (Sotos-Prieto et al. 2015).

A higher parental education was significantly associated with a healthier diet and dietary behaviours (Fisk et al. 2011; Huybrechts et al. 2011; Kristiansen et al. 2013; Mak et al. 2013; Camara et al. 2015; Sotos-Prieto et al. 2015), whereas a lower parental education was significantly associated with unhealthy dietary patterns (Mcgowan et al. 2012; Camara et al. 2015). Preschool-aged children of parents with the highest educational status had a lower energy intake from snacking, had a healthier eating score, and consumed more fruit and vegetables (Kant and Graubard 2013; Sotos-Prieto et al. 2015). Those who came from deprived neighbourhoods tended to consume excess calories (Pabayo et al. 2012; Kant and Graubard 2013).

The association between income and diet were less clear as the relevant studies used proxies to collect income data (Fogarty et al. 2007; Pabayo et al. 2012) and in other studies, income data were not collected. In the three studies that did collect income data, a higher income was associated with a healthier diet (Kant and Graubard 2013; Camara et al. 2015; Sotos-Prieto et al. 2015). Three studies used occupation as an indicator of SES. Two of the three found no significant association between occupation and diet (Lioret et al. 2008; Camara et al. 2015). The third study did find a relationship between high SES (indicated by occupation) and higher fruit and vegetable intake, suggestive of a healthier diet, in children (Mak et al. 2013).

Three studies included family size when collecting sociodemographic data. One study found that increased calcium was significantly associated with increased family size (Huybrechts et al. 2011). The Norwegian study found that there was a significant association between the unhealthy dietary pattern and parity (Kristiansen et al. 2013). The other study, while reporting family size, did not further investigate family size and diet quality (Sotos-Prieto et al. 2015). In census data collection, family size is used to calculate poverty, thus it could be an important factor to consider when looking at SES (American Psychological Association 2020). Other parental and family characteristics such as maternal age and childcare arrangements may influence dietary patterns in young children and were included in some of the studies we reviewed (Fisk et al. 2011; Huybrechts et al. 2011; Kristiansen et al. 2013; Sotos-Prieto et al. 2015) but were outside the scope of this review.

Nine out of the 12 articles were cross-sectional, indicating that there is a lack of high-quality study designs looking into the influence of SES on preschoolers’ dietary patterns. Causation cannot be derived from cross-sectional analysis; thus, more quasi-experimental or randomized controlled trials are needed as further evidence. Many of the studies we reviewed included interventions targeting low-income or deprived populations; however, there was no comparison or control groups included that would allow for a more meaningful quantification of the effect of low SES on preschoolers’ dietary quality. Many of the studies exploring SES in preschoolers are concerned with childhood obesity and only included diet as a covariate. There were very few studies that looked at dietary intake as an outcome of SES in preschoolers.

The results suggest that both high maternal education (Smithers et al. 2011) and paternal education (Camara et al. 2015) influence healthier eating patterns in preschoolers. In support of our findings, a Canadian study conducted with fifth graders found that parents that were socio-economically advantaged tended to encourage and support their children to eat healthy (and be physically active) more so than their socio-economically disadvantaged counterparts (Simen-Kapeu and Veugelers 2010). An unhealthy diet was associated with low maternal education, in addition to younger maternal age, household financial disadvantage, presence of older siblings, and being cared for by someone other than their mother at home (Camara et al. 2015). This is not surprising as similar findings were found in the NHANES national survey in the United States in which children with parents in the lowest educational category were more likely to consume a greater proportion of their energy intake from snacking and obtain an eating score associated with an unhealthy diet (Kant and Graubard 2013).

As such, nutrition and dietary education for parents with young children may be a key intervention strategy for targeting childhood obesity, particularly for parents with lower educational attainment or lower income. For example, a nutrition education randomized controlled trial conducted in low-income parents showed improved nutrition behaviour changes, including feeding their children, after this trial (Dollahite et al. 2014). Preschoolers’ dietary intake is likely fully dependent on what their parents provide them to eat, indicating that parental nutritional knowledge may be an important factor contributing to preschoolers’ dietary patterns.

The results were inconclusive regarding parental income and eating patterns, which was unexpected as a consistent relationship has been found in adults (Patrick and Nicklas 2005; Darmon and Drewnowski 2008). For example, two review studies found that higher quality diets are generally consumed by highly educated and more affluent people whereas a lower diet quality was more common in those with a limited education and income (Patrick and Nicklas 2005; Darmon and Drewnowski 2008). Another study in the United States, looking at purchasing patterns based on household income, found that people from lower income households purchased lower quality foods than those from high-income households (French et al. 2019). This suggests that those from low-income households consume a diet with limited nutritional quality, whereas high-income households consume more nutritional diets (French et al. 2019). There is a lack of evidence supporting or refuting the impact of family income on preschoolers’ diet. We suggest that this may be because most studies identified used a combination of income and education to indicate SES. As such, we were unable to parse the independent effect of income on preschoolers’ diet. This observation is the second gap identified in the literature and further research is needed to clarify the role of income on dietary patterns of preschoolers.

Other than increasing household income, it is incredibly challenging at the level of the individual to make changes that would greatly improve their diet quality. A strategy to address food insecurity due to limited resources is through policy and public programs. In the United States, there are programs such as the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) that have potentially decreased the prevalence of child food insecurity by approximately 20% (Kreider et al. 2016). However, for WIC specifically, program participation is low likely due to structural barriers (Liu and Liu 2016). Although such initiatives can improve diet, it can be difficult to foresee flaws of programs and policies; thus, it is important to conduct pilot projects and ensure that researchers and policymakers have strong partnerships.

In a systematic review conducted on children aged 9–13 years, the authors found that parents with low socioeconomic status were less likely to model healthy eating behaviours than parents with a high socioeconomic status (Zarnowiecki et al. 2014). However, these findings were based on children aged 9–13 years. While the relationships may be similar in preschoolers, there is a need for evidence-based interventions able to confirm this. The lack of studies in preschool children is particularly concerning because it has been demonstrated that the early years between 2 and 5 are pivotal for the development and establishment of lifelong health behaviours (Mikkila et al. 2005). For example, a prospective longitudinal study found that the strongest indicator of foods liked by 8-year-old children was based on the foods liked at 4 years of age (Skinner et al. 2002). It is evident that food preference for healthy foods is formed in the early years (Nicklaus et al. 2005).

In the context of developed countries, a vast number of young children attend childcare where they can consume up to two-thirds of their daily diet. A US-based study suggests that children tend to consume healthier diets in childcare than at home (Sisson et al. 2017). Additionally, data from the Healthy Start/Départ Santé project in Canada suggest that parental SES did not impact child uptake of nutrition education encouraging healthy eating behaviours (Anstruther and Vatanparast 2018). This may warrant implementation of funding programs for childcare enrolment and centres to ensure that children are consuming a healthy diet. However, more research examining if healthy eating behaviours learned in childcare could mitigate potential unhealthy behaviours learned in the home environment is needed.

This study is not without its limitations. Only English studies were included as the main reviewers (SMA and EBT) speak only English, which may have led to a selection bias. Our review looked only at developed countries as the diet and SES relationship is likely to be different from that found in developing countries. For example, a systematic review on dietary patterns and socioeconomic factors in children aged 2–19 years old living in high-, medium-, and low-income countries, found that there tended to be unhealthy dietary patterns in the more affluent groups in medium-income countries (not significant), whereas the results were inconclusive in low-income countries (Hinnig et al. 2018). Alternatively, in high-income countries, higher parental education was associated with a healthy dietary pattern and lower education with an unhealthy pattern (Hinnig et al. 2018) and that a higher income correlated with a lower adherence to the unhealthy dietary pattern (Hinnig et al. 2018). Thus, the context surrounding dietary patterns and SES in developed and developing countries differ considerably. Lastly, we did not proceed with the sixth step of Arksey and O’Malley’s (2005) framework due to the limited yield and logistical constraints.

A limitation of the studies included in this review was that assessment methods were inconsistent. There is not a standardized way to measure diet quality, so it is difficult to make a cohesive conclusion. Additionally, FFQs, food diaries, and surveys are prone to recall bias and subjectivity, but other methods would not have been as feasible given the large sample sizes in each of the studies. Similarly, varying indicators were used to measure SES across studies, meaning that a low SES in one study may be deemed a moderate SES in another study. As there is no gold standard for measuring SES, it is difficult to compare studies objectively, both within and between countries. Regarding dietary assessment methods, future studies should consider using a weighted food intake/food waste method as it has shown to be the most accurate measure in children aged 0.5–4 years (Burrows et al. 2010).