PMC

Literature search and review flow. CHD indicates coronary heart disease; GI, glycemic index.*The Hardy et al study, for which data were unavailable, only reported a rate of change in risk of CHD per 5 and 30 units of GL.

Test for publication bias in the overall pooled analysis of CHD risk estimates associated with the highest GI quantiles; Grau et al report on men was identified outside the 95% pseudo–confidence limits. Neither Begg's test (P>0.837) nor Egger's test (P=0.621) revealed evidence of publication bias.^, CHD indicates coronary heart disease; GI, glycemic index.

Test for publication bias in the overall pooled analysis of CHD risk estimates associated with highest GL quantiles; Grau et al report on women was identified outside the 95% pseudo–confidence limits. Begg's (P>0.115) and Egger's (P=0.134) tests approached significance for evidence of publication bias.^, CHD indicates coronary heart disease; GL, glycemic load.

Pooled risk estimate of all prospective cohorts investigating the association of highest GI exposure with CHD events (including death and myocardial infarctions) relative to the reference exposure (ΔGI between mean of highest exposure and mean of reference=12.1±1.1 SE). The figure is stratified by sex-specific subgroups with subtotal boxes in 2.1.1 and 2.1.2 summarizing the pooled analysis for women^,,,,, and for men,^,,,, respectively. The total analysis box represents the overall pooled analysis for both men and women. P values in circles are based on generic inverse variance (IV) methods in random-effects models and represent the significance for association of high-GI diets with CHD. The P value in a rectangle depicts the significance of differences between the subgroups. Interstudy heterogeneity was tested by Cochrane's Q (χ²) at a significance level of P<0.10 and quantified by I². CHD indicates coronary heart disease; GI, glycemic index.

Pooled risk estimate of all prospective cohorts investigating the association of highest GL exposure with CHD events (including death and myocardial infarctions) relative to the reference exposure (ΔGL between mean of highest exposure and mean of reference=89.4±9.5 SE). The figure is stratified by sex-specific subgroups with subtotal boxes in 3.1.1 and 3.1.2 summarizing the pooled analysis for women^,,,,, and for men,^,,,, respectively. The total analysis box represents the overall pooled analysis for both men and women. P values in circles are based on generic inverse variance (IV) methods in random-effects models and represent the significance for association of high-GL diets with CHD. The P value in a rectangle depicts the significance of differences between the subgroups. Interstudy heterogeneity was tested by Cochrane's Q (χ²) at a significance level of P<0.10 and quantified by I². CHD indicates coronary heart disease; GI, glycemic index.