> EMIS – odd they don't give HbA1c decrease for the other doses , although they do say the HbA1c showed a dose response.<
I question whether this study was genuinely blinded. If you recall, the original statistical analysis plan specified that HbA1c from the aggregated insulin arms would be compared to placebo (i.e. metformin alone). Then, after the study had been in progress for a while, they added patients and changed the analysis plan to compare the individual insulin arms to placebo. Then, as it turns, out, only one arm shows good results.