Why the recent obsession with NNT? It's hard to follow this discussion when it is so vague.

Keep in mind that comparing NNT across drugs and their respective trials is only appropriate if the patient population, endpoints and absolute event rates are taken into account. Otherwise, this is a pointless exercise. Very few trials have tested their drugs in a BETonMACE-like population: type 2 diabetic, low HDL, ACS event within 90 days. And apabetalone has not been tested in other populations. 

For the pre-specified 3-point MACE, the placebo event rate was 12.4% and the apabetalone event rate was 10.3% at the median follow up of 26 months.. The difference between these absolute event rates is 2.1%, or 0.021. NNT is 1/.021, which is 47.6. But the caveats above apply and if you start changing the endpoints or sub-populations, then you need to use those event rates for your NNT calculation.

BDAZ