Fruit and vegetable consumption in disadvantaged neighbourhoods

In developed countries, individuals of lower socioeconomic position (SEP; e.g. educational attainment, occupation, income), generally eat less fruit and vegetables than those of higher SEP¹. Beyond individual SEP, the local neighbourhood environment has the potential to influence the diet of residents. Residents of more disadvantaged neighbourhoods may be more likely to eat unhealthily if they cannot access stores in which healthy produce can be purchased. While there is some evidence to suggest that fruit and vegetable consumption is lower in residents of disadvantaged neighbourhoods^2,3, findings have been mixed⁴. These inconsistencies in findings between countries may reflect true differences. However, it is difficult to directly compare existing studies due to differences in the measurement of dietary outcomes and the potential confounding variables considered in analyses.

There exist few international comparisons of diet even though studies of this nature help to determine the generalisability of findings across nations. The aim of this study was to compare associations between neighbourhood SEP and fruit and vegetable consumption of adults across seven developed countries⁵. Collating the data for this study to control all aspects of the analysis, including the treatment of the variables and adjustment of confounders, enabled a closer comparison of associations than could be determined from comparing findings from separate studies.

Study design

This study involved secondary analysis of seven cross-sectional studies: the SocioEconomic Status and Activity in Women (SESAW) study from Australia, the Edmonton Population Health Survey in Canada, the Health and Living Conditions of the Population of Eindhoven study from the Netherlands, the New Zealand Health Survey, the National Health Survey for the Lisbon Metropolitan Area in Portugal, the Greater Glasgow Health Board Health and Wellbeing survey from Scotland, and the Healthy Environments Partnership study from the USA. Fruit and vegetable consumption was, where possible, grouped to ensure comparability across studies (fruit: <2/≥2 serves per day; vegetables: <3/≥3 serves per day). In six studies neighbourhood SEP was grouped as low, medium and high; in the Netherlands, neighbourhoods were selected from low and high SEP neighbourhoods to maximise contrasts meaning medium SEP could not be examined. All analyses of the association between neighbourhood SEP and fruit and vegetable consumption adjusted for age, gender (SESAW only considered women) and level of education.

Associations between fruit and vegetable consumption and neighbourhood SEP

Observed associations were generally in the expected direction, with increased consumption associated with higher neighbourhood SEP. In the Canadian, New Zealand and Scottish studies only, there was evidence of increased fruit consumption for those residing in more advantaged neighbourhoods. Neighbourhood SEP was positively associated with vegetable consumption in the Australian, Canadian, New Zealand and Portuguese studies.

Implications

This study suggests that associations between neighbourhood SEP and fruit and vegetable intake may be context specific, indicating that it may be inappropriate to generalise the findings from one country or study context to another. Inconsistencies, although potentially due to other methodological inconsistencies across datasets outside of our control (sampling or neighbourhood definitions), could reflect cross-country differences in the social, built, economic or regulatory factors that influence fruit and vegetable provision. For example, the availability of fresh food outlets is high in both low and high SEP neighbourhoods in the Netherlands⁶, while supermarkets have been found to be more common in low SEP neighbourhoods in Glasgow perhaps reflecting regulatory controls or land use costs⁷ which could explain null findings shown in these studies. Future studies designed with a priori matching on important characteristics are required to confirm the observation from this study that associations are context specific, with further investigation required into the drivers of these associations to identify potential policy actions to help redress socioeconomic inequalities in diet.