Thank you Chad and Liane for your perspective.

There has been a lot of talk about the test for superiority in the Affinity 2 trial. From the conference call: “Over the 24-week treatment period, mean A1c levels decreased by 0.82% in the AFREZZA group compared to 0.42% in the oral therapy group. The between-group difference in the mean change in A1c was statistically significant at the P less than 0.0001 level, thereby establishing superiority of AFREZZA when added to their existing oral agents over the comparator oral therapy treatment alone. Some of you may recall that in designing this study, we made certain assumptions about the statistical power necessary to detect a between-group A1c difference of 0.5% with a one-sided a less than 0.025.

We use that statistical model to determine the number of subjects to be enrolled into each treatment group in order to perform the superiority analysis. The actual between-group A1c difference was 0.4%, but with a P value of less than 0.0001 there can be no doubt that the study was adequately powered to make an unequivocal superiority conclusion in keeping with the statistical analysis plan that was agreed to by the FDA.”

The original parameters that were used to set up the test are as follows:

The null hypothesis, H0: Afrezza is no more effective than a placebo in controlling blood glucose as measured by the difference in the average A1c levels of the two groups.

The alternative hypothesis, H1: Afrezza is superior to a placebo.

Or in more mathematical terms:

H0: the difference in the average A1clevels < 0.5%

H1: the difference in the average A1c levels >= 0.5%

We want to reject the null hypothesis and claim that Afrezza is superior.

The p value for this test was 2.5%, which means that the probability of committing a type I error is 2.5%. This means that there is only a 2.5% chance of us rejecting the null hypothesis that Afrezza is not more effective than a placebo when it really is not more effective.

The test that they performed to claim superiority is the following:

H0: the difference in the average A1c levels < 0.4%

H1: the difference in the average A1c levels >= 0.4%

The p value is 0.01%.

So, what they are saying is that Afrezza is better than a placebo, but not by as much as the original experiment would have indicated, but we are much more certain that we are not making a type I error.