Table 9. Univariate and multivariate meta-regression analyses of eligible cross-sectional studies for odds ratio for type 2 diabetes mellitus (T2DM) for 1 increment in risk alleles carried in relation to diabetes-associated single nucleotide polymorphisms (SNPs) by characteristics of study design^a.

Variable	Univariateb N = 40			Multivariate 1 N = 39c			Multivariate 2 N = 39c
	Coefficientd	SE	P	Coefficient	SE	P	Coefficient	SE	P
No. SNPs ≥10	−0.094	0.029	0.002	−0.077	0.033	0.03	−0.084	0.033	0.02
Western country	−0.062	0.025	0.02	−0.060	0.025	0.02	e
Asian ethnicity	0.065	0.025	0.01	e			0.081	0.031	0.02
Mean age ≥55 years	−0.002	0.028	0.95	−0.003	0.021	0.90	−0.019	0.023	0.42
Women-dominantc	−0.061	0.028	0.04	−0.065	0.022	0.005	−0.064	0.022	0.006
BMI ≥27 kg/m2	−0.057	0.025	0.03	−0.005	0.025	0.84	0.018	0.027	0.56
Adjustment for BMI	0.009	0.033	0.79	0.024	0.028	0.40	0.024	0.028	0.39
Using OGTT to diagnose DM	0.039	0.028	0.17	−0.004	0.026	0.88	−0.017	0.027	0.53
Excluding IGT from noncases	0.115	0.035	0.002	0.074	0.046	0.12	0.074	0.046	0.12
Excluding IFG from noncases	0.023	0.028	0.42	−0.019	0.023	0.42	−0.010	0.023	0.66
R-squared					64.4%			61.4%
F-test					F(9,29) = 4.60	P < 0.001		F(9,29) = 4.70	P < 0.001

Abbreviations: Same as in Table 8.

^aLogarithm of odds ratio for T2DM was a dependent variable, and each study characteristic was entered as an explanatory variable.

^bFundamentally, results were consistent with the stratified analysis.

^cN = 39 because one study⁵⁴ in which the proportion of men was not available was excluded.

^dPositive value of the coefficient means that the OR is higher when the answer to each variable is “Yes” compared with a “No” answer. The same interpretation of the results is applied to Table 10, Table 12, and Table 13.

^eWestern country and Asian ethnicity was not entered simultaneously because of collinearlity.