When NovoNordisk writes an article analyzing safety data from the CONTROL trial, you get a skewed perspective on the data. Specifically, if you only read the abstract, you’re going to think that it’s safer in some ways(ARDS was less), and there was no difference in adverse events (except for all that investigator-reported AMI/NSTEMI). So, that sounds favorable.
But, the real reason there’s no significant differences in outcomes – and the reason why they terminated the trial early – is because the interim data is underpowered to detect a difference. As you see, the 30-day mortality is 12% vs 11% in favor of placebo – and that wasn’t helping NovoNordisk, so they quit before they could reach sufficient statistical power to prove their product was unhelpful. However, they can now benefit from that same lack of power to detect differences by applying it to the safety aspect, and trumpeting its equivalency in terms of AEs.
When taken in the context of the original trial, this is just a flawed piece of pharmaceutical propaganda to try and prevent the building crackdown on off-label Factor VII use.
http://www.ncbi.nlm.nih.gov/pubmed/21610529