Obesity is a growing public health concern in both developed and
developing countries, creating acute challenges in places with scant resources. In
Ghana, obesity rates have risen substantially in recent decades, a trend particularly
noted in urban areas. However, high levels of migration and urbanization indicate a
situation that is more complex than a simple urban/rural distinction may be able to
explain. Latent trajectory modeling (LTM) with eigenvector spatial filtering offers a
methodology to explore the spatial and temporal patterns of body mass index (BMI)
change by going beyond the urban/rural distinction and examining how different
environmental, social, and demographic variables contribute to BMI changes over
time. Using data from a regional LULC study and the Ghana Demographic Health
Survey (1993, 1998, 2003, 2008), the relationship between BMI and the amounts of
urban, agricultural, and natural land covers, household size, % of houses with
electricity, % houses with flush toilets, and % of houses with no toilets for 845
survey clusters is explored. Our findings suggest higher BMIs in the most urban
areas, yet larger BMI increases in peri-urban areas (and lower BMI changes in slums
and increasingly rural areas). The LTM modeling indicates a trajectory of BMI
growth in the study region, yet one that is slowing over time. Earlier, indicators of
higher socioeconomic status and larger households are associated with high BMIs,
but these indicators are not associated with rising BMI over the entire study period.
Areas with increases in urban land cover show consistent, significant relationships
with BMI growth.