Skip to main content

Table 2 Adjusted differences in behaviors between intervention and combined control communities after 2-year intervention period

From: Changes in diet and physical activity resulting from the Shape Up Somerville community intervention

 

Baseline

After intervention

Pre–post change: adjusted difference1

Model properties

Behavior

Control group

Intervention group

Control group

Intervention group

 

n

Mean (SD)

n

Mean (SD)

n

Mean (SD)

n

Mean (SD)

Effect (95% CI)

p-value

additional covariates

R2

Fruit & vegetable (servings/day)

317

3.1 (1.5)

103

3.5 (1.6)

317

3.4 (1.6)

103

3.7 (1.8)

0.16 (-06,0.38)

0.09

parental foreign born status, # siblings

0.18

Sugar-sweetened beverages (ounces/day)

265

6.5 (6.0)

72

6.1 (6.3)

265

7.6 (7.0)

72

5.5 (6.7)

-2.00 (-3.76,-.25)

0.04

# rules

0.21

Sports (# per year)

343

2.9 (2.8)

111

3.6 (2.9)

343

3.4 (2.7)

111

4.0 (2.9)

0.20 (0.06,0.33)

0.02

 

0.21

Walk to/from school (# trips per week)

248

2.7 (4.0)

87

3.5 (4.1)

248

2.6 (3.9)

87

3.9 (4.2)

0.65 (-0.53,1.82)

0.14

parent marital status, maternal BMI

0.20

TV time (hrs/day)

325

2.2 (1.1)

104

1.6 (1.1)

325

2.2 (1.0)

104

1.7 (1.2)

-0.24 (-0.51,0.04)

0.06

# rules

0.27

Total screen time (hrs/day)

332

3.8 (1.8)

106

2.7 (1.6)

332

3.9 (1.9)

106

3.0 (2.2)

-0.24 (0.42,0.06)

0.03

parent marital status, # siblings, # rules

0.22

TV in bedroom2 (% yes)

250

50.4%

85

29.4%

250

54.8%

85

31.8%

0.39 ( 0.11,0.89)

0.13

child weight category, # siblings, maternal BMI

--4

Dinner with TV3 (% not very much/never)

337

61.4%

110

73.6%

337

62.0%

110

71.8%

OR 0.94 (0.88, 1.00)

0.06

 

--4

  1. 1All models adjusted for sex, age, ethnicity, primary language in the home, baseline value, and clustering by community. Additional covariates that were unbalanced between study groups, which demonstrated significant correlation with outcomes are also added to the model. These included parent marital status, parental foreign born status, number of siblings, maternal BMI, and rules.
  2. 2In the regression model to obtain the intervention effect, the dependent variable is a 3-level ordinal variable: -1 (acquired a TV in bedroom), 0 (no change), 1 (TV removed from bedroom).
  3. 3The outcome is dichotomous. Adjusted difference is reported as an odds ratio. OR < 1 indicates that those in the intervention were less likely to watch TV with dinner not very much/never at the end of 2 years.
  4. 4R-squared values were not calculated for the ordinal or dichotomous variables.