14

endpoint. The only difference here is that there are more events. Instead of a total of 11

15

events, there are now 40 events. So generally the confidence limits are tighter for the

16

individual study odds ratios and the overall odds ratio. Again we see that study 13

17

produces an estimate greater than one. Now because of all the discussion around the

18

TZDs and CV events, we thought we should look at those events by background TZD.

19

This table shows you the custom events by treatment group for both doses studied and by

20

background TZD and you can see there is nothing here that differentiates the TZDs for

21

these few events.

6

the one we get when we do an exact test which is 0.2 and this is not surprising, because

7

when you use a continuity correction the estimates tends to shift towards the not. Notice

8

there is only one study that produces an estimate greater than one. That's study 13.

9

Study 13 is the study where patients treated at time of enrolment with TZD, either

10

Pioglitazone or Rosiglitazone, are randomized to add-on therapy of Saxagliptin or

11

placebo. So overall we see the plot suggests no significant heterogeneity across the trails

12

and test for heterogeneity support that.

23

MACE endpoint for the short-term plus long-term data. We did these analyses because

24

we wanted some reassurance that the results seeing in an overall population of rather low

Scribes, LLC Toll Free 1-800-675-8846 www.scribesllc.com

13

Now let's look at the short-term plus long-term data for the custom MACE

22

Lastly, I would like to show you some subgroup results using the custom

1

used. There is only one study with events in both treatment arms, and that study is 39,

2

it has the largest symbol here, which is the study of initial combination of Saxagliptin

3

with Metformin. There are five events in this one study, and there are only a total of 11

4

events in all eight studies.

84

5

Now the estimate using the continuity correction is 0.3 which differs from