Skip to main content
. 2013 Jan 30;97(3):584–596. doi: 10.3945/ajcn.112.041467

TABLE 3.

Association of hypothesized determinants with the RR of T2D (n = 24 studies)1

Increment in RR2 P-trend
Increment in RR2 Unadjusted Adjusted3 R 2 I 2 P for I2
SEX (F > M)
 Single covariate model 0.24 (−0.01, 0.57) 0.065 17 43 0.02
 Bi-covariate model (model 3)4 0.23 (0.01, 0.47) 0.036 0.036 72 14 0.20
 Fully adjusted (model 5)5 0.22 (0.02, 0.46) 0.031 0.031 97 2 0.43
CORR (over the maximum range, 0.0–1.0)
 Single-covariate model 2.19 (0.5, 6.0) 0.004 53 30 0.09
 Bi-covariate model (model 3)4 2.15 (0.5, 5.5) 0.002 0.004 72 14 0.20
 Fully adjusted model (model 5)5 2.05 (0.6, 4.7) 0.0005 0.001 97 2 0.43
FUY (/10 y)
 Single-covariate model −0.01 (−0.28, 0.26) 0.92 −9 49 0.01
 Fully adjusted (model 5)5 0.00 (−0.19, 0.20) 0.96 >0.99 97 2 0.43
ETH (European American > other ethnicities combined)
 Single-covariate model 0.24 (0.01, 0.52) 0.041 18 43 0.016
 Fully adjusted model (model 5)5 0.22 (0.05, 0.41) 0.011 0.044 97 2 0.43
1

RR of T2D was for an energy-adjusted glycemic load 100 g higher than in the referent quantile. P values were calculated by using the z test for RR and the Q test for I2. CORR, energy-adjusted and deattenuated dietary instrument correlation for carbohydrate; ETH, proportion of participants of European American ethnicity; FUY, number of follow-up years; I2, heterogeneity that remained after adjustment for covariates; R2, heterogeneity accounted for by adjustments for covariates; SEX, proportion of male study participants; T2D, type 2 diabetes.

2

All values are trends; 95% CIs in parentheses. Trends and 95% CIs are shown untransformed. RR was analyzed as the ln.

3

Adjusted for 2 or more covariances according to Bonferroni (see footnote 1 of Table 2). The following are listed in order of priority: 1 × P for the first hypothesized determinant (SEX), 2 × P for the second hypothesized determinant (CORR), 3 × P for the third hypothesized determinant (FUY), and 4 × P for the fourth hypothesized determinant (ETH).

4

Inclusive of the first 2 hypothesized covariates (ie, SEX and CORR).

5

Inclusive of all 4 hypothesized covariates (ie, SEX, CORR, FUY, and ETH).