Table 3.

 Mean estimated meal calorie content among participants purchasing meals at fast food restaurant chains in four cities in New England, US, 2010 and 2011*

Variables	Parameter estimate (95% CI)	Exponentiated parameter estimate† (95% CI)
Adults
Intercept†	5.92 (5.84 to 6.00)	372 (345 to 402)
Log actual calorie content, relative change	0.57 (0.50 to 0.63)	1.77 (1.65 to 1.88)
Adolescents
Intercept†	5.59 (5.47 to 5.70)	268 (238 to 298)
Log actual calorie content, relative change	0.51 (0.42 to 0.59)	1.67 (1.53 to 1.81)
School age children
Intercept†	5.77 (5.58 to 5.95)	321 (265 to 384)
Log actual calorie content, relative change	0.39 (0.08 to 0.70)	1.48 (1.09 to 2.01)

*1658 adult, 1081 adolescent, and 254 school age participants had complete data on all covariates examined in multivariable regression models (shown in table 4); we used this sample for these models.

†Parameter estimates of intercept are arithmetic means of estimated calorie content on log scale. Because we log transformed outcome of estimated calories and centered predictor log actual calorie content on its mean, exponentiated intercepts are geometric means of estimated calorie content for those participants consuming meals of mean actual calorie content (836, 756, and 733 calories for adults, adolescents, and school age children). Geometric means of estimated calorie content are less than half of mean actual calorie content of meals.