Here are some thoughts - no promises on them being deep or accurate, just reflective of my experiences reading/presenting the outcomes of studies to physicians over the years. (If anyone cares, I've been in medical sales - not pharmaceutical - since 2003, specifically in the cardiovascular space for the last 5 years.)
Mild to Moderate trial
- I know they exceeded their planned enrollment, and assume that was by accident (speed of enrollment across multiple sites - they beat their enrollment estimate date by months and never activated two sites). However, with such a small number of people overall there could potentially be some issues with the power of the study.
I've seen this happen where you get good results in terms of separation from the placebo/control group but because of being underpowered the study may not come back as statistically significant. Those are confusing results, especially in a world where even many physicians just look at the headline to help them make their decisions.
[*]However, I think they picked a decent primary endpoint which is just symptomatic improvement based on a composite score.
I can't find the actual protocol, so I'll assume there's some specifics around how they are scoring each of the 4 components across the scoring spectrum of 0-3. The better defined this is, the more people can easily agree with (not dispute) the results. Especially if the definitions are in line with industry norms.
- If these scores are more subjective then you can run into issues where different locations are scoring things differently and your results aren't as accurately represented - especially with lower enrollment (power) numbers.
[*]Assuming they hit the primary with good statistical significance (many studies say they want a p-value of <0.05, but still many academics want to see 0.01 or nowadays even 0.001) the secondary endpoints can help sell the lebronlimeade story.
They have 12 secondary endpoints listed which means they really want to extract a LOT from this study and are not expecting another bite at the apple.
- If the primary endpoint is not met (even if results are good, but not significant, it's considered "not met" and a "failed trial") most academic purists will say that none of these other endpoints matter.
- However, some folks really disagree with this perspective and will say you need to look at the totality of the data. Bottom line from a stock perspective you want them to really nail the primary outcome, then the secondary ones are gravy. If you miss the primary but crush half the secondary it can mean some positive things for the drug, but you have a longer road ahead of you to prove that's the case to the medical community.
- Some shorter who is looking to crap on CYDY could cherry pick some secondary endpoints that aren't met (even if the primary is met) and try to knock down the results. Don't let that sway you. For instance, I could see secondary endpoints 5, 9, 11, and 12 not making significance even if the primary is met. Then someone will say "hospitalizations, O2 usage, mortality, and returning to normal activity didn't improve - I don't care if symptoms got better based on some scoring system that NP concocted to make his product look good." Don't worry about that. Doctors will see it hit the primary endpoint and will use it for these cases.
Severe to Critical trial
- 390 patients seems low to me at first glance for total mortality, but I think that's because I'm used to trials with patients in the thousands because mortality on cardiac events occurs at a lower rate (percentage basis) than I presume does for severe COVID cases.
- However, these studies are powered based on estimates and data known at the time. So when they started this the death rate for severe COVID patients may have been higher than what we are seeing today. As doctors learn to somewhat treat these patients better, and we are seeing a younger population infected, we are seeing a decline in death rates (I think).
- If that's the case there is the possibility of seeing what I described before - an improvement versus control group but not statistical significance. Buuuuuttt....
- When they take a peek at the data they may see that this is occurring. Specifically, they can see a treatment effect PLUS the product being safe - and in that case they may recommend to the primary investigator to increase the enrollment, which would push back the estimated completion timeline. Probably not from the one listed on the study, but from estimates discussed.
- Again, if that were to happen shorters could jump on it and say that they are just trying to enroll more people to hope they get a different result. That isn't really true. The hope in doing that is to show that the result being seen is in fact statistically significant. AND this is only done if the safety of the treatment is already not in question. So if this happens this isn't bad for the drug, it just potentially extends the timeline.
- Speaking of enrollment, I think they have a good mix of places to get the people they need. Seeing NY/NJ/CT/MA on here is less "helpful" since those states are doing a better job controlling cases. But seeing NC/CA/TX does help with getting the patients into the trial.
- TOTAL MORTALITY
Just as an RCT (randomized controlled trial) is the gold standard for medical trials, Total Mortality is the gold standard for trials where the intervention is needed to save a life. This doesn't mean it's perfect - but if you get a good result for this metric as your primary endpoint the medical community will be very happy.
- The problem is that total mortality is just that - if they die from ANYTHING during the trial, it counts. So first you have to understand that if a person signs onto the trial and is randomized into the CYDY group then even if they die before receiving a treatment, that hurts the number. If they die for reasons unrelated to COVID or the drug therapy, that hurts the number. It cuts both ways, though, as it also applies to the control group.
Heck, even if a person signs on and is randomized and then says "nevermind, I'm out" that doesn't matter. They could transition to palliative care the next hour and die that day having never received the therapy and still count to the trial based on their randomization. If CYDY doesn't meet the total mortality significance threshold, expect to hear discussion about this. This sort of parsing the data is understandable, but docs will take a LONG time to "buy it."
[*]So to understand how they want to mitigate this, look at the exclusion criteria. They did a decent job of trying to make it so that people unlikely to stick with the trial or have problems with the therapy. What they weren't able to do is exclude people based on comorbidities such as diabetes or heart disease, because that would probably knock out too many of their actual severe patient population. This one could be tricky as a result. But if the result comes back positive and strongly significant, that's great.
[*]The secondary outcomes here (again) will mostly come into play only if the primary is hit. They are really counting on hitting the mortality overall, and then showing speed (and breadth) of effect. Some are in place to point to some sub-populations that could be successful even if total mortality is missed, but it will be hard to convince clinicians of this in any short-term timeframe.
Anyway, that's it for now. I should probably get back to actual work. Hope this helped some folks out. I'll try to answer questions if anyone has any but don't get mad if I give you an "I have no clue."