Retrospective 1,108 hospitalized patients in New York showing significantly higher mortality with HCQ treatment.
Time based confounding is very likely because HCQ became increasingly controversial and less used over the time covered (Mar - Jun 2020), while overall treatment protocols during this period improved dramatically, i.e., more control patients likely come later in the period when treatment protocols were greatly improved. Authors note that for every week or month later that a person was admitted, their risk of death dropped by 16% and 49%, respectively, yet they do not consider time based confounding.
Sammartino et al., 5/10/2021, retrospective, propensity score matching, USA, North America, peer-reviewed, 7 authors.
risk of death, 240.0% higher, RR 3.40, p = 0.002, treatment 137, control 191, PSM, model 1a, RR approximated with OR.
This study is excluded in the after exclusion results of meta analysis: substantial time varying confounding likely due to declining usage over the early period when overall treatment protocols improved dramatically.
Effect extraction follows pre-specified rules
prioritizing more serious outcomes. For an individual study the most serious
outcome may have a smaller number of events and lower statistical signficance,
however this provides the strongest evidence for the most serious outcomes
when combining the results of many trials.