Date of Award


Document Type

Open Access Dissertation


Epidemiology and Biostatistics


The Norman J. Arnold School of Public Health

First Advisor

Angela Liese


Background: Obesity is a big public health concern in the US. Previous studies examined its association with food shopping practices measured by distance to the food store, shopping frequency, and type of store selected. However, not much is known about the actual food acquisition and shopping habits integrating multi-dimensional aspects. The purpose of this study was to identify distinct food acquisition and shopping patterns in populations primarily residing in food deserts in South Carolina (SC) and a general population in the US and characterize these patterns with respect to socioeconomic status (SES), nutritional knowledge, and perceptions and store selection reasons, and then examine the association between the identified patterns with body mass index (BMI).

Methods: Two datasets were employed, including a sample of 522 participants from two SC counties and 4826 households from a national representative survey. Food acquisition and shopping habits measures including travel distances between residential location and each of the used stores, shopping frequency, store type, transportation, and utilization of community food resources, such as food banks or pantries and church or social services were used. Latent class analysis was employed to explore the acquisition and shopping patterns. Multivariable linear regression was used to assess the association between the identified patterns and BMI adjusting for sociodemographic information.

Results: Three classes were identified among the SC low-income population, defined by distance, frequency, transportation and community resources utilization. Among the national population, three classes among urban households and two classes among rural households (with similar attributes as two classes that identified among urban households) were identified, which were defined by distance, travel time, transportation, and farmers’ market utilization. SES factors, nutritional knowledge, perception of food environment, and store selection reasons were associated with the identified patterns. No significant associations were found between the identified patterns and BMI.

Conclusions: Different patterns were identified among general and low-income populations, and among urban and rural populations. Future interventions on increasing healthy food access and intake should take into consideration the different food acquisition and shopping patterns and factors that impact those patterns.


© 2018, Xiaonan Ma

Included in

Epidemiology Commons