From the PROactive study:
Lancet 2005 Oct 8;366(9493):1279-89
There is no significant reduction in the primary endpoint, but a reported reduction in the secondary endpoint which is a composite of all-cause mortality, non-fatal myocardial infarction and stroke.
But look at table 4: there is in fact no difference in each individual endpoint of death, non-fatal myocardial infarction (including silent MI), and stroke. Why is this?
A careful review of the article shows why. For the secondary endpoint they chose non-fatal myocardial infarction excluding silent MI. When this is thrown in you get figure 3 with an HR of 0.84 and CI of 0.72-0.98.
Does this really indicate anything significant about a reduction in macrovascular disease events when the data is manipulated in this way?