Table 2.
Associations between fast-food outlet exposure and each of body mass index (also stratified by genetic risk score for BMI (BMI-GRS)) and risks of overweight and obesity in the Fenland Study (n = 10,798), Cambridgeshire, UK, estimated using linear and multinomial logistic regression models, respectively
| Per five additional neighbourhood fast-food outletsa | Quartiles (Q) of fast-food outlet exposureb | ||||
|---|---|---|---|---|---|
| Q1 (n = 4167) | Q2 (n = 1360) | Q3 (n = 3096) | Q4 (n = 2175) | ||
| Body mass index, β (95% CI)c | 0.14 (0.05, 0.23)* | REF | − 0.08 (− 0.36, 0.20) | 0.27 (0.04, 0.50)* | 0.65 (0.23, 1.06)* |
| BMI-GRS low (n = 5399)d | 0.24 (0.12, 0.36)* | REF | 0.08 (− 0.30, 0.46) | 0.30 (− 0.01, 0.61) | 1.10 (0.54, 1.66)* |
| BMI-GRS high (n = 5399)d | 0.06 (− 0.07, 0.19) | REF | − 0.26 (− 0.67, 0.16) | 0.28 (− 0.06, 0.62) | 0.18 (− 0.44, 0.79) |
| Overweight, 25 kg/m2 ≥ BMI < 30 kg/m2, RR (95% CI)c | 1.02 (0.98, 1.07) | REF | 0.96 (0.83, 1.11) | 1.07 (0.95, 1.21) | 1.04 (0.84, 1.30) |
| Obese, BMI ≥ 30 kg/m2, RR (95% CI)c | 1.08 (1.02, 1.14)* | REF | 0.94 (0.79, 1.13) | 1.19 (1.21, 2.05)* | 1.58 (1.21, 2.05)* |
*P < 0.05
Interaction (fast-food outlet exposure quartiles x genetic risk score for BMI (BMI-GRS) z-scores) tested using an F-test for BMI and a likelihood ratio test for overweight and obesity: BMI P = 0.09; overweight P = 0.51; obesity P = 0.27
For the overweight and obese outcomes, estimates of the association are from multinomial logistic regression models and hence are risk ratios (relative to normal weight)
aAssociations with BMI, overweight and obesity per five additional neighbourhood fast-food outlets are shown in addition to estimates per quartile, which is our primary operationalisation of fast-food outlet exposure
bHome neighbourhood fast-food outlet exposure, quartiles (Q): Q1 (least exposed) = 0–1 outlets; Q2 = 2; Q3 = 3–14; Q4 (most exposed) = 15–51. Quartiles are unequal in sample size due to the distribution of the underlying data
cAdjusted for age, sex, household income, highest educational attainment, car access, smoking status, physical activity energy expenditure, counts of supermarkets in home neighbourhoods, and BMI-GRS
dResults from a model stratified by sample median BMI-GRS; adjusted as for other models but omitting adjustment for BMI-GRS