Perceived neighbourhood food environment applied in urban neighbourhoods of Malaysia: mediation effects

Abstract

Studies engaging the role of place in predicting health outcome in Malaysia are still very much in their infancy. A particular gap is testing the association of physical environments, particularly the food environment of different socio-economic neighbourhoods in predicting body mass index (BMI). Thus, the aim of this study was to examine the mediating effects of perceived neighbourhood food environment variables (PNFE) towards the relationship between neighbourhood socio-economic status (NSES) and BMI status, while controlling for socio-demographic variables. The four main indicators of PNFE identified from a systematic review were perceived food store type availability, perceived geographical food store accessibility, perceived healthy food availability and perceived food affordability. The BMI status of 60 subjects from three different NSES were collected. Their perceptions toward the neighbourhood food environment in which they live in were measured using a culturally and linguistically appropriate questionnaire (ICC = 0.59 - 0.82) that was adapted to Malaysians living in urban areas. Multiple regression models were run, and mediation tests were conducted. Overall, both NSES and PNFE explained a low variability proportion of BMI status (R2 = 0.36), whereby perceived food store type availability was positively associated with BMI status (Beta = 0.200 (95% CI = 0.021 to 0.380), p = 0.030), perceived geographical food store accessibility was positively associated with BMI status (Beta = 0.644 (95% CI = 0.316 to 0.971), p = <0.000) and perceived healthy food availability (Beta = 0.315 (95% CI = 0.041 to 0.589), p = 0.025; but not perceived food affordability. Through the addition of the PNFE variables, NSES was found to be significant and positively associated with BMI status (Beta = 0.200 (95% CI = 0.021 to 0.380), p = 0.030). It is worth noting that although a model containing NSES and PNFE variables might be a poor predictor for BMI status, this does not negate the fact that those variables are statistically significant as predictors ofBMI status. It is recommended that more environmental factors, such as physical building environment and social environment, as well as behavioural factors, such as diet quality and physical activity level, be included in future studies to better predict BMI.