If you trawl through that thread that we used to explore the stats you can see some red flags that we missed along the way, but the main problem was that Mesoblast powered the trial so aggressively my stats were undershooting the risks considerably. I was looking for a 10% decrease in mortality and they were looking for 40ish % which if you plug that into my model is a bit reckless (and was partially what caused me to sell). Here are some posts that seem quite illuminating in hindsight:
@bedger bringing up the change in SOC due to dexamethasone in early September -
https://hotcopper.com.au/threads/overwhelming-efficacy-the-stats-required.5579134/?post_id=47054817my reply where I explain what it might do to the trial and then subsequently convince myself that it probably won't be an issue -
https://hotcopper.com.au/threads/overwhelming-efficacy-the-stats-required.5579134/?post_id=47058153If you read through that thread the white papers said we should expect 35% mortality, but MSB management said they were still observing up to 70% mortality in the trial cohort population, so we didn't know quite who to believe when simulating the result. What was the actual placebo group mortality rate? 40%! So the issue is that the whole way along there were posters here reading white papers expecting the placebo cohort to have an increasingly larger change of survival, and when asked about it, management reiterated higher mortality rates.
This post here was unfortunately very accurate in predicting the risks that made the trial a failure:
https://hotcopper.com.au/threads/overwhelming-efficacy-the-stats-required.5579134It's good to go back and read over those posts, there were a few level heads that properly analysed the situation and did not let emotion and bias creep in when predicting what might happen. I think I did an ok job of data gathering and a poor job of interpreting the data and making a good investment decision. Everything is easier in hindsight, though - and I was able to move my MSB funds into other opportunities that have more than covered my losses so I'm not too salty. I think ARDS will eventually be approved, so long as these trial results hold up in an RCT.
(20min delay)