This is basically a case of garbage in, garbage out. In my epidemiological opinion. My masters course in critiquing studies like this was my favorite! So the authors looked back at patient records from various VA locations and compared COVID positive patients who got (1) HCQ, (2) HCQ+Z-pack, or (3) just “standard care” which they didn’t describe in any detail so no idea what that entails.
Next they simply compared rates of death, discharge, and intubation among patients in these 3 treatment groups. On the surface it looks like more of the group 1 patients died than group 3. However, were group 1 patients different somehow? Well, I would argue that they probably were, because they got the HCQ. They probably needed the meds more than others who only got standard care.
Because the researchers know this is probably an issue that will muddy the meaningfulness of the results, they use a fancy method called “propensity score adjustment” which is a statistical way to dust off the muddiness by adjusting the model to account for what they would deem differences might be between those who got treatment 1 vs 3. It’s used often and it can be helpful, but in this case it’s akin to washing your car with muddy water. The study also has not been peer reviewed or published. I would call it exploratory at best.
The more interesting result to me is the difference between groups 1 and 2. They were likely a cleaner comparison, and it appears the combo of HCQ and Z-pack worked better, which is consistent with other reports we have heard. Lastly the study group was all men, so the muddy results shouldn’t be generalized to women.