If PP means "primary prevention", it's the hard MACE that looks better for REDUCE-IT -- RRR of 19% hard vs only 12% soft.
In fact, there was a 19 event advantage on hard, and only 17 on soft -- that means there were 2 more revascs or angina events for placebo than for Vascepa.
No - You are getting it backwards. The lack of significant interaction (p>.15) for hard MACE can be interpreted statistically as indicating that BOTH primary and secondary prevention groups got the same benefit. That is the problem - you can come to a different interpretation depending on whether you look at the interaction or just do analyses within subgroups. I was trained in statistics to test the interaction, and ONLY if it significant, do you proceed to look at the effects within single subgroups as being meaningful.