Figure 6. Metaregression of the proportion of patients with diabetes in a trial on the X-axis against the log Peto odds ratio on the Y-axis.

This shows that there is a significant trend to greater benefit with oral volume expansion in trials with a higher proportion of diabetic patients.