The sample size is 62 and the mortality rates are between 0% and 7% depending primarily on age and co-morbidities. I don't think we can draw any meaningful conclusions from this other than what we already know: seems worth looking into HCQ more.
How does the 2 cases of collateral effects compare to the normal consequences of the drug? It is very likely that we can draw conclusions about those.
Luckily, long-term use of the drug isn't necessary here because the immune system successfully responds with antibodies to fight of the virus, so retinopathy isn't an issue. Also, as the paper states, "Fortunately, deciding on individual treatment plans scientifically, monitoring adverse reactions timely, to avoid overdose, short-term application of HCQ is relatively safe."