Table 3

Food items associated with largest estimated effect on mean waist- and hip-circumference in women.

Estimated effect on mean waist circumference

Estimated effect on mean hip circumference

Combined¹

Women

Difference in mean intake 1986–99

Model 2³

Model 2³

Model 1²

Model 2³

1²

2³

Food item³

/mo⁴

/'86 ⁵

mm⁶

p⁷

mm⁶

p⁷

mm⁶

p⁷

mm⁶

p⁷

mm⁶

mm⁶

beer, 4% alcohol

0.6

153%

-1.10

0.60

-0.92

0.83

-0.68

0.59

-1.24

0.36

0.42

-0.32

bread, crisp

-13.4

-27%

-0.31

0.21

-0.37

0.26

0.06

0.65

-0.09

0.64

0.37

0.28

cream/crème frβiche/sour cream

1.4

40%

-1.27

0.26

-0.94

0.32

1.38

0.01

0.99

0.09

2.65

1.93

fruit soups/fruit creams

-1.3

-36%

-3.24

0.01

-3.13

0.02

-0.07

0.58

-0.25

0.67

3.17

2.88

hamburger

0.8

73%

3.21

0.07

2.94

0.13

-0.66

0.47

-0.41

0.67

-3.87

-3.34

milk, 1.5% fat

18.0

331%

-1.11

0.02

-0.76

0.03

1.24

0.48

0.98

0.45

2.35

1.75

milk, 3% fat

-21.8

-84%

-1.47

0.25

-0.75

0.38

-0.05

0.44

-0.74

0.55

1.42

0.01

oil, cooking

8.7

693%

-6.18

<.01

-4.86

<.01

1.35

0.40

0.23

0.75

7.53

5.09

oil, dressing

3.3

210%

-2.54

0.07

-1.96

0.17

0.50

0.73

0.03

0.79

3.04

1.99

pasta

4.2

132%

-2.21

0.25

-1.47

0.45

2.14

0.09

1.61

0.15

4.35

3.09

potato chips/popcorn/salted nuts

1.5

188%

-0.78

0.44

-0.24

0.45

1.15

0.59

1.14

0.53

1.94

1.38

potatoes, French fried

0.9

95%

2.11

0.14

1.90

0.13

-0.42

0.05

0.04

0.13

-2.52

-1.86

soft drinks

1.9

93%

2.11

0.19

2.00

0.23

0.31

0.91

0.57

0.77

-1.81

-1.43

sugar/honey in tea/coffee

-10.4

-36%

-0.30

0.39

0.02

0.41

0.13

0.68

-0.05

0.83

0.44

-0.08

wine

0.7

65%

-1.03

0.05

-0.83

0.08

1.08

0.15

1.67

0.22

2.11

2.51

¹Estimate of total effect on diabetes risk: Differences in mean hip-circumferences (negatively associated with DM) – Differences in mean waist-circumferences (positively associated with DM)

²Model 1: regression of change in mean reported food frequency on waist/hip circumference adjusted for BMI, age and survey-year.

³ Model 2: all variables are adjusted for BMI, age, survey-year, smoking-status (never/ex-/current), level of reported physical activity (lower/higher), educational level (primary school/secondary school/college), level of reported alcohol (quartiles) intake as well as interaction between smoking and low physical activity, smoking and high alcohol intake, low physical activity and high alcohol intake.

⁴ number of reported intakes per month

⁵ relative changes in % of mean reported intake 1986

⁶ estimated net effects of trends in reported intakes on mean waist/hip circumference.

⁷ probabilities that partial regression coefficients for different levels of intake are equal to zero at 95% level

Krachler et al. Nutrition Journal 2006 5:34   doi:10.1186/1475-2891-5-34