What are the limitations of observational registry studies? One major issue is confounding by indication. How physicians act is not the same as a Randomized Clinical Trial testing a therapy distributed equally on all comers. Docs will give experimental meds to sicker patients...

The Lancet registry adjusts for categorical qSOFA score 0 vs 1+ (which is normal mental status, Resp rate <22, BP >100) vs Not; categorical Oxygen sat (low <94% vs. 94%+). That is pretty simplistic adjustment for severity. CQ/HCQ groups were worse than controls, but how much?

40% received other antiviral therapy (lopinavir, ribavirin, oseltamivir), but interestingly not included in the mortality analysis (figure 2) but is included in arrhythmia analysis (figure 3). Another interesting point, 61% of deaths occur in obese (BMI >30) vs 27% survivors

Is this Lancet registry study a reason for institutional ethics boards to stop pre-exposure prophylaxis trials for safety? Absolutely not. That's ridiculous as it is a very different population.