JMIR Publications

ABSTRACT

Background:

Contemporary neighborhoods resulting from rapid urbanization processes are often saturated with local eateries for local communities. These neighborhood eateries are susceptible to expansion of exposure to unhealthy foods, potentially creating diabetogenic residential habitats.

Objective:

We aimed to determine the proximity of commercial food outlets’ access to local neighborhood residences of Type 2 Diabetes (T2D) cases, understand how local T2D rates vary by locations, and to propose a policy-driven metrics monitoring how much a particular food outlet density in town needs to be restructured to control local T2D rates.

Methods:

This cross-sectional, ecological study included 11,354 active T2D cases aged ≥ 20 years geocoded from approximate neighborhood residence, aggregated to areal level rates and counts by subdistricts (mukims) in Pulau Pinang, Northern Malaysia. We executed point-level proximity analysis and density-based area level analysis through multi-model aspatial and spatial regression methods by leveraging Geographic Information Systems (GIS) to establish connections between neighborhood T2D burden and the food-environment.

Results:

Our hierarchical linear regression reveals how much variation occurs in associations between T2D prevalence and local food environments across neighbourhoods, with R-squared values ranging from 1.1% in step 1 to approximately 6.5% in step 4 when different food outlets were populated in each step in the model, adjusted for age. In the Multiscale Geographical Weighted Regression (MGWR) model, the adjusted R-squared values were ranging from 0.15 to 0.62, with lower values visualized across the mainland. The MGWR beta coefficients showed how different food outlet density varied by space indicating areas that should be prioritized for local public health interventions. Finally, we established population attributable fractions (PAF) metric that suggests if local neighbourhoods undergo township restructuring to reduce, remove, or relocate food complexes, hawker markets, or kopitiams, an estimated reduction of 0.21%, 0.27%, and 0.09% in the risk of T2D cases in Pulau Pinang would be anticipated, however, if the local neighbourhoods undergo township restructured to hawker complexes, nasi kandar restaurants, fast-food outlets, 24/7 convenience stores, public markets, and/or supermarkets, an estimated reduction of between 0.07% to 0.64% in the number of residents with risk of T2D were estimated.

Conclusions:

The reported variations give local insight into the associations between high neighbourhood T2D rates and a range of different food outlets’ density. We observe that these associations vary by place giving insight into potential monitoring and policy considerations for targeted interventions.