Read on Twitter
Spent some time today going through late-phase protocols and reported data for major COVID-vaccine programs trying to understand primary/secondary endpoint definitions. Summary: great consistency in definitions of mild/moderate/severe but....
Some disparity w/ Janssen vaccine. It's the only one that made primary endpoint be moderate/severe and excluded mild COVID cases in primary. Also, the only one to measure twice (starting 14 days after shot and 28 days after shot).
Clearly trying to maximize shots on goal. Thought question - what if they miss lower statistical bound on one of the two co-primaries: I assume FDA will still authorize given massive public health need.
Also the only one to put post-hoc analysis in press release (no severe disease in cases starting at 49 days) - 49 days was not a defined endpoint in protocol.
Also, they did not report cases starting at 14 days in PR other than oblique mention (that was the co-primary efficacy endpoint for god's sake), and they only reported cases > 28.
Makes me think that when full Janssen briefing packet is published and then full data-set is out in peer-review, the actual protocol defined efficacy endpoints will be a worse picture that what was showed via PR/IR yesterday.
Also makes me wonder why they excluded mild disease from their co-primary endpoints unlike Moderna and Pfizer. My guess: lower protective antibodies in early phase human and non-human primate data pushed them to enrich odds of success by narrowing case definition for primary.
So I'm with @EricTopol - don't cherry pick some primary endpoints and not others. Show us all primary endpoints now. Here's my grid - open to suggestions on how to improve this. https://docs.google.com/spreadsheets/d/1mJaH6Us9Ooy82mBkTtEpsT9bz_XNXf6dU5lIKkDBpts/edit?usp=sharing
