Multiple linear regression model for the post-methionine load increase in homocysteine. Data on subjects with diabetes and controls combined. Top, subjects in the highest tertile of betaine excretion and bottom in the lowest tertile of betaine excretion. Dependent variable: log(rise in plasma homocysteine). Independent variables: log(baseline fasting homocysteine), log(dimethylglycine excretion as mmol dimethylglycine per mole creatinine) and age (years). Bars show the significance of the factors as t-values, negative for a negative relationship and positive for a positive relationship; dashed lines p = 0.05; dotted lines p = 0.001. Pie graphs at the right show the proportion of variance (dark segment) explained by regression: unexplained variance in grey.