I wonder why the Yale professor has a different view than the systematic review BC referenced. I'm sure he knew of that systematic review (published online July 11).
I looked at the systematic review. As I read it, of the studies involving HCQ for prophylaxis, they discounted all the studies
but one. While they gave their judgement of the studies (risk of bias, low quality, etc.), I didn't see the reason for those judgements.
So I looked at that one study (published June 3), which I'm also sure the Yale professor was aware of. 400+ subjects in each arm (avg age about 40, mostly health care workers), with medication or placebo starting within 4 days of apparent exposure. They randomized the subjects so there obviously isn't any intentional biases. I did wonder why they had a binary categorization of "high risk exposure" (10 minutes within 6' of positive cases without mask or shield). [The rest of us can take away the fact that only about 1 out 8 total exposures were judged as getting the disease.] Were there no subjects with much higher exposures (1 hour; 24 hour)? Were all these indoor, small room exposures, or were some outdoor? Were the subjects exposed to high risk situations (intubation), or to coughing individuals, or just to individuals that were mildly symptomatic? Where in the progress of the progression of COVID were the exposures (pre-symptomatic/asymptomatic, early symptomatic, late symptomatic)?
What shocked me was that only 20 of the 113 subjects with probable COVID-19 had positive PCR tests. Why didn't the study conduct those tests themselves? "We assumed that health care workers would have access to Covid-19 testing if symptomatic; however, access to testing was limited throughout the trial period [mid-March - early May]" AFAICT, the report does not indicate how many, if any, got negative PCR results. Since they don't mention that, I guess any such results were for those that had a negative clinical diagnosis. Three tested that positive during the 14 day period had no symptoms within 14 days but developed them later. One tested positive during the 14 day period but never reported symptoms.
To me, this failure to test is almost enough to discount this study. How many of the 700+ subjects that had no symptoms had an asymptomatic case? The common belief is that the asymptomatic (or very mildly symptomatic) cases are on the order of the number of symptomatic cases. If one hypothesizes that the total number of cases was 2x the detected cases, that would have made the percentages: Placebo 28.6% vs HCQ 23.6%. Would that have been a significant reduction in cases?
Otherwise, I felt the study seemed pretty good (but this is not my area of expertise).