Article Text

This article has a correction. Please see:

Download PDFPDF
Socioeconomic variations in women’s diets: what is the role of perceptions of the local food environment?
  1. V Inglis,
  2. K Ball,
  3. D Crawford
  1. Centre for Physical Activity and Nutrition Research, Burwood, Victoria, Australia
  1. Victoria Inglis, Food Standards Australia New Zealand, PO Box 7186, Canberra BC ACT 2610, Australia; vinglis{at}deakin.edu.au

Abstract

Objectives: To test the contribution of perceived environmental factors (food availability, accessibility and affordability) to mediating socioeconomic variations in women’s fruit, vegetable and fast food consumption.

Methods: A community sample of 1580 women from 45 neighbourhoods provided survey data on their socioeconomic position (SEP) (education and income); diet (fruit, vegetable and fast food consumption); and the perceived availability of, access to and cost of healthy food in their local area.

Results: Once perceived environmental variables were considered, the associations between SEP and diet were weak and non-significant, suggesting that socioeconomic differences in diet were almost wholly explained by perceptions of food availability, accessibility and affordability.

Conclusions: Strategies to decrease socioeconomic inequalities in diet could involve promoting inexpensive ways to increase fruit and vegetable consumption, and ensuring that people of low SEP are aware that many healthy foods are available at relatively low cost. Future research should also confirm if perceptions match objective measures of food availability, accessibility and affordability, in order to address the real and/or perceived lack of healthy options in low SEP neighbourhoods.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Compared with those of high socioeconomic position (SEP), individuals of low SEP have poorer diets according to a range of dietary indicators.16 For example, lower SEP individuals are more likely to consume diets high in fat, low in micronutrient density, and to have lower intakes of fruit and vegetables.710 As a result, studies repeatedly find that people of low SEP possess nutrient intakes and dietary patterns that increase risk of diet-related diseases.11 12

Even though SEP differences in diet are relatively well documented, the underlying causes of these differences are not well understood. Studies on the correlates of dietary behaviour have identified a range of individual, social and environmental influences on eating. However, the environmental influences (particularly food availability, accessibility and affordability) are under-researched, and existing studies show contradictory findings. Some studies have found that neighbourhood socioeconomic disadvantage is an independent risk factor for low fruit and vegetable intake or high fast food consumption.1315 However, other studies show that neighbourhood SEP is not associated with residents’ diets.16 17 When considering the environmental influences as mediators of SEP variations in diet, some studies show low SEP areas lack healthy eating options,18 19 whereas others show no difference between affluent and disadvantaged areas in availability, accessibility and cost.20 21 Australian research has found minimal socioeconomic differences in terms of access to or availability of healthy affordable foods22 and shopping infrastructure.23 Consistent with these findings, a recent qualitative study found that women from neighbourhoods across a range of SEP appeared generally satisfied with their local food environment and availability of healthy foods.24

The effects of socioeconomic disadvantage on diet could be mediated by socioeconomic differences in environmental factors such as food availability, accessibility and affordability. Few studies have directly tested this proposition. In one recent study that did, food environmental factors were not strong mediators of SEP variations in diet. However, the environmental factors assessed in that study were limited to two fairly crude indices: the number of large supermarkets and fruit and vegetable stores locally.25

This study aimed to investigate the relation between SEP, the perceived physical environment and women’s diets (in particular fruit, vegetable and fast food consumption). Specifically, aspects of the physical environment that may mediate the relation between SEP and diet were investigated. Women were the focus of this study since the determinants of women’s diets are likely to be different from those of men,26 and women remain largely responsible for domestic duties, including the provision of food.27

METHODS

Sample

The sample consisted of 1580 women involved in the Socioeconomic Status & Activity in Women (SESAW) study. Full details of the methods are described elsewhere.25 28 SESAW was restricted to a geographical area within approximately 25 km of the Melbourne central business district. Based on 2001 census data, the Australian Bureau of Statistics has assigned a SEIFA (Socioeconomic Index for Areas) score based on relative disadvantage to each suburb. All suburbs within the study area were ranked according to SEIFA score, and 45 suburbs (15 each from low, mid and high SEIFA areas) were randomly selected. An initial sample of 2400 women was randomly selected from the Australian electoral roll from all women of working age (18–65 years) within the 45 suburbs targeted.

Given discrepancy in response rates by SEP groups observed in health surveys,29 30 low and mid SEP suburbs were slightly oversampled relative to the high suburbs, by a ratio of 1.5:1.2:1. Hence the final sample drawn to receive the women’s diet survey consisted of 645 women from high SEP, 780 from mid SEP and 975 from low SEP suburbs. A total of 1136 women responded to the survey: 354 from high, 407 from mid and 375 from low SEP neighbourhoods. This response (50% overall, excluding from the denominator 127 women who had moved/were ineligible) is similar to those obtained in other recent mail-based surveys targeting women.31 The SESAW study focused on physical activity as well as diet. All participants completing a separate physical activity survey (which did not collect diet data) were asked if they were willing to complete a second survey, and those agreeing were posted the diet survey. This second phase of the study resulted in an additional 444 diet surveys (42% of those completing the original physical activity survey). Hence the final sample size for the present analyses was 1580. Owing to incomplete data, 252 women were excluded from analyses in table 2 (n = 1328); 204 women were excluded from analyses in table 3 (n = 1376); and 218 women were excluded from analyses in table 4 (n = 1362).

Table 2 Effects of adjusting for perceived environmental mediators on associations between women’s education level and likelihood of being a high fruit consumer in logistic regression models (n = 1328)
Table 3 Effects of adjusting for perceived environmental mediators on associations between women’s education level and likelihood of being a high vegetable consumer in logistic regression models (n = 1376)
Table 4 Effects of adjusting for perceived environmental mediators on associations between women’s income level and likelihood of being a frequent fast food consumer in logistic regression models (n = 1362)

Procedures

Approval to conduct the research was obtained from the Deakin University research ethics committee. Eating surveys were posted to 2400 women and physical activity surveys to a separate sample of 2400 women. A reminder protocol32 was used by which those women who did not respond received a postcard after three weeks, and a second reminder with a replacement survey package after a further three weeks. As small incentives, all women received a $1 lottery ticket with their initial survey package and all were advised that they would be entered into a drawer to receive one of five shopping vouchers or movie passes.

Measures

Predictor variable: socioeconomic position

There is little consensus about the appropriate means of conceptualising and measuring SEP,33 particularly among women.34 Using only one measure of SEP may underestimate true dietary differences because only one component of SEP is considered,35 and different SEP indicators may measure different causal mechanisms.36 Therefore two measures were used in this study; education and income. Women’s self-reported highest education qualification was categorised as: less than year 12; year 12, trade or certificate; degree or higher degree. Women also reported their own average gross (before tax) weekly income (including wages, salary, pensions and allowances). The response categories were; less than $500 per week; $500–$999 per week; $1000 or more per week; don’t know/don’t want to answer/missing.

Outcome variables

Because of their established associations with a range of health outcomes,37 consumption of fruit, vegetable and fast food were used as the dietary measures in this study.

Participants were asked in two separate questions: how many servings of fruit, and how many servings of vegetables they usually consumed per day. These questions have been validated in the Australian National Nutrition Survey38 in which they were shown to adequately discriminate between SEP groups with different fruit and vegetable intakes. The response categories were none; one serving; two servings; three to four servings; five servings or more. Women were then classed into two groups for each question. Those that met the Australian dietary guidelines37 for fruit consumption (two or more servings per day) were classed as high fruit consumers, and those that failed to meet the dietary guidelines (less than two servings per day) were classed as low fruit consumers. Initially a similar classification was applied for vegetable consumption, using a cut-off point of five servings or more, consistent with the Australian dietary guidelines.37 However only 5% of the women met this guideline, leaving insufficient numbers to enable detailed analyses. Consequently, the categories of vegetable consumption were reassessed. The next consecutive response option to the question on vegetable consumption was three to four servings per day (60–80% of the level of consumption recommended). A larger proportion (34% of women) reported consuming at least three to four servings of vegetables per day. Therefore, this cut-off point was used to categorise women as being low or high vegetable consumers.

Fast food consumption was assessed with the questions: “How many times per week, including breakfast, lunch and dinner, do you eat meals that are from fast food restaurants (for example, pizza, McDonald's) eaten in the fast food restaurant?” and, “How many times per week do you eat meals that are from fast food restaurants eaten as ‘fast food’ at home/work/study (including home delivery)?”. Responses for both questions were: never; less than one meal/week; about one meal/week; two to three meals/week; four to five meals/week; six to seven meals/week or more. These two questions were summed to calculate total fast food consumption. Women were categorised as infrequent fast food consumers (one fast food meal per week or less) or frequent fast food consumers (more than one fast food meal per week). Although there are currently no well established validated measures of fast food consumption, this cut-off point has been shown to distinguish women at risk of weight gain.39

Environmental mediators

Questions about women’s local neighbourhood assessed three domains: food availability, accessibility and affordability. These questions were developed for the present study.

Food availability: Perceived food availability was assessed with four items assessing agreement (1: strongly agree to 5: strongly disagree) with the statements: “I can do most of my food shopping at stores in my local neighbourhood; at the shop where I buy fruits and vegetables, the variety of fresh fruits and vegetables is limited; the fresh produce in my area is usually of a high quality; there are lots of healthy options for eating out in my local neighbourhood”.

Food accessibility: 11 potential perceived food accessibility mediators were assessed. Respondents were asked if they had difficulties with transport to or from their usual place of food shopping; access to a motor vehicle for private use; and access to childcare if they needed to go shopping without their children. Women were also asked to report how long it took them to get to their main shopping location (for example, supermarket) from home, with four response options, ranging from less than 10 minutes to more than 30 minutes. Respondents were also asked if a supermarket; a fruit and vegetable store; a small grocery store; a fresh food market; a fast food restaurant; a non-fast food restaurant; and a café were within walking distance of the woman’s home.

Food affordability: The women responded to five questions which assessed their perceptions of the cost of food. On a scale from 1: not at all to 4: very much, women were asked in two separate questions how much they considered the cost of food, and “specials/discounts/sales”, when deciding what food or groceries to buy when food shopping. Perceived cost of fruit and vegetables was measured with two items assessing agreement (1: strongly agree to 5: strongly disagree) with the statements: “I do not buy many fruits because they cost too much; I do not buy many vegetables because they cost too much”. Respondents were also asked (yes/no) if they had ever run out of food in the last 12 months and been unable to afford to buy more.

Statistical analyses

Analyses were conducted using SPSS version 11.0. Descriptive analyses were performed to describe the distributions of SEP, age, marital status and dietary variables. SEP and the dietary variables were then entered into separate logistic regression models, predicting the likelihood of being either a high fruit consumer, a high vegetable consumer or a frequent fast food consumer (model 1).

Baron and Kenny’s approach to testing mediation was used.40 Individual logistic regression analyses were used, as appropriate, to assess which potential mediating variables met the initial conditions of mediation: (1) the independent variable (education or income) being significantly associated with the potential mediator and (2) the potential mediator being significantly associated with the dependent variable (dietary behaviours).40 For these initial criteria an inclusive significance level of p<0.05 was set. The final condition of mediation was that inclusion of the potential mediator reduced the association between the independent and dependent variables.40 To test this, separate logistic regression analyses were conducted in which those environmental mediators found to satisfy criteria 1 and 2 above were added to model 1 (model 2). Based on its established association with diet,41 42 women’s marital status was also controlled for in analyses. Age was not controlled for as preliminary χ2 analyses showed it was not significantly associated with the dietary outcome variables.

RESULTS

Table 1 presents the participant characteristics and distributions of fruit, vegetable and fast food intakes by socioeconomic variables. Women who were high fruit and vegetable consumers had higher levels of education; and were married or living in de facto relationships. Frequent fast food consumers were more likely to be on a low income; and either separated, widowed, divorced or never married. Education was not associated with fast food intake, and income was not associated with either fruit or vegetable intake; therefore mediating models were not conducted for these variables.

Table 1 Participant characteristics and distributions of fruit, vegetable and fast food consumption with socioeconomic and bivariate variables

Table 2 shows the effects of adjusting for environmental variables on associations between women’s education level and the likelihood of being a high fruit consumer. Model 1 shows that, controlling for marital status, highly educated women were 67% more likely to be a high fruit consumer compared to low educated women. Model 2 added those mediating variables found to be significantly bivariately associated with both the predictor (education) and outcome (fruit consumption) variables; in total there were two availability variables, two accessibility variables and five affordability variables. The odds of being a high fruit consumer for high SEP women were reduced (to OR = 1.36) and became non-significant once the environmental mediators were added to model 2. Both availability variables—availability of quality fresh produce and healthy options to eat out locally; and two affordability variables—not buying fruits and not buying vegetables because they cost too much—remained predictive of the odds of fruit consumption in these models.

Table 3 shows that education level was significantly related to vegetable consumption, with those women who were highly educated 1.66 times more likely to be high vegetable consumers than those women with less than 12 years education. Those mediating variables found to be significantly bivariately associated with both the predictor (education) and outcome (vegetable consumption) variables were added in model 2. As a result, two availability variables, three accessibility variables and four affordability variables were included. The mediating analyses showed that the odds ratio for the association between high education and vegetable consumption was reduced and became non-significant when the environmental variables were added in model 2. Both availability variables—availability of quality fresh produce and healthy options to eat out locally, and one affordability variable—not being able to afford healthy foods, were significantly predictive of vegetable consumption in the final model.

Table 4 shows the effects of adjusting for environmental mediators on associations between women’s income level and likelihood of being a frequent fast food consumer. High income women were almost 50% less likely to be frequent fast food consumers than low income women (model 1). Again, mediating variables found to be significantly bivariately associated with both the predictor (income) and outcome (fast food consumption) variables were added in model 2; a total of two availability variables, two accessibility variables and four affordability variables were included. The odds for frequent fast food consumption among high SEP women became non-significant once the environmental mediators were added to the model. Availability of healthy options to eat out remained the only predictor of fast food consumption in this model, with those women who reported plenty of healthy options to eat out in their local neighbourhood less likely to be frequent fast food consumers.

DISCUSSION

This study is one of the first to examine the contribution of perceptions of food availability, accessibility and affordability to mediating socioeconomic differences in fruit, vegetable and fast food consumption among women. While separate studies have reported that SEP is inversely related to fruit and vegetable consumption710 and positively associated with fast food consumption,14 43 44 the present findings advance those of previous studies by showing that perceived environmental factors such as food availability, accessibility and affordability may mediate the association between SEP and diet.

Consistent with previous research, we found SEP variations in the outcomes examined, with those in the lowest educational group less likely to be high fruit and vegetable consumers and those in the lowest income group more likely to be high fast food consumers. Once all three sets of environmental variables were considered in each of the tables, the associations between SEP and diet became weak and non-significant, suggesting that socioeconomic differences in diet were almost wholly explained by the perceived food availability, accessibility and affordability variables examined. The contribution of perceived environmental mediators appeared to be the strongest for vegetable consumption, where the odds of being a high vegetable consumer were reduced by 24% (from 1.66 to 1.26) with the inclusion of environmental mediators.

The present findings suggest that focusing on perceived food availability, accessibility and affordability may be important in reducing socioeconomic differences in women’s diets. The findings also suggest that different perceived environmental factors played a role in mediating associations of SEP with different dietary outcomes. For example, the same two availability variables (quality of fresh produce and healthy options to eat out locally) were predictive of all three dietary outcomes; whereas other availability variables (ability to shop at food stores locally and variety of fresh fruit and vegetables available locally) were not predictive of any of the three dietary outcomes. It could be that there is minimal SEP difference in the food shopping that can be done in local stores as previous Australian evidence has shown2224; or that low SEP individuals may perceive that they lack quality but not variety of fresh produce in their local stores. Finally, more cost variables compared to the other variables were included as mediators, suggesting that food affordability may be a particularly critical mediator of the SEP differences in women’s diet. This is not surprising considering the current structure of food prices is such that sweet and high-fat foods provide dietary energy at a low cost, whereas the energy cost of lean meats, fish, vegetables, and fruit is likely to be higher.45

The findings of this study need to be confirmed since it is not clear how closely women’s perceptions of food availability, accessibility and affordability match the objective environments in which they live. The present findings suggest that women’s income and education were associated with both dietary outcomes and with a number of perceived environmental variables. However, other Australian research has found that neighbourhood SEP is not associated with food price or availability.22 23 These discordant findings may be attributable to the use of subjective as opposed to objective measures of food environmental variables, but this requires further investigation. The present findings are also based on cross-sectional data, and it is not possible to determine whether the associations are causal. Other limitations include the use of self-report measures, in particular the possibility of dietary measurement error resulting from the use of self-report food frequency questionnaire, and the restriction of study area to one city. This study received a modest response rate, although the response is similar to that achieved in other health surveys.46 Strengths of the study include the relatively large sample design and the incorporation of multiple potential mediators.

Acknowledging these limitations, findings of the present study have important implications for nutrition promotion among women, particularly those from disadvantaged groups. Strategies to decrease socioeconomic differences in diet could involve promoting inexpensive ways to increase consumption of fruit and vegetables, and to ensure that people of low SEP are aware that many healthy foods are available at relatively low cost. Food Cent$, an Australian health promotion programme targeting low income families, developed a proportional spending model which demonstrated that it is possible to achieve a healthy diet within a limited budget.47 Future studies, however, are also required to assess the relative importance of environmental perceptions alongside other potential mediators such as values, knowledge, cooking skills, social norms and support.

What is already known on this subject

It is well documented that individuals of low socioeconomic position (SEP) have poorer diets according to a range of dietary indicators. However, the mechanisms underlying these socioeconomic differentials are not known.

What this study adds

The findings advance those of previous studies by showing that environmental factors such as perceived food availability, accessibility and affordability mediate the association between SEP and diet.

Policy/programme implications

Public health strategies aimed at decreasing socioeconomic differences could involve promoting inexpensive ways to increase healthy foods such as fruit and vegetable consumption, and to ensure that people of low SEP are aware that many healthy foods are available at relatively low cost. Policies addressing poorer availability, accessibility and affordability of healthy foods in disadvantaged areas may also be required.

Acknowledgments

VI is supported by a scholarship from the Victorian Health Promotion Foundation. KB and DC are each supported by an Australian National Health and Medical Research Council/National Heart Foundation Career Development Award. This project was partly funded by a grant from the National Heart Foundation, Reference number G 02M 0658, and by the Australian Research Council.

REFERENCES

Footnotes

  • Competing interests: None.

Linked Articles

  • In this issue
    Carlos Alvarez-Dardet, Joint Editors John R Ashton, Joint Editors
  • Correction
    BMJ Publishing Group Ltd