Questions raised about the manuscript

A common problem in assessing duration of an antibiotic therapy and outcome is that people who live longer are able to get more days of that antibiotic. However, in this situation, that was not a major issue at all. Durations of HCQ and AZM therapies were decided by the different ID consultants, not by who survived longer.

I received this comment on the MedRxiv website. https://www.medrxiv.org/content/10.1101/2021.05.28.21258012v1.article-metrics

Le Bon • 3 days ago

The idea of looking at patients who had at least a substantial cumulated dose is great, but there is an important immortal time bias, since the patients can't die the first 10 days in the HCQ group, as seen on the Kaplan-Meïer I suggest to calculate again the results with excluding the first 10 days death in the control group. Due to the small group size, you will probably loose significance, but it still will be usefull for pooled analysis.

Stephen Smith • 13 hours ago

Hello Le Bon and thank you for your comments.

First, you don't have to get to 10 days to receive >3 gm of HCQ and >1 gm off AZM. We gave 600-800 mg per day. In fact, 7 of the deaths in the >3gHCQ/>1gAZM group occurred within the first 10 days of admission and 1 of the survivors left before 10 days as well. In other words, of the 38 pts in this group, 8 were discharged by Day 10.

Second, since the HCQ iwas started so early, the immortal time bias is not particularly relevant. If you look at the details of the study, those who treated with HCQ were started on the drug very early, unlike toci or CP. Certainly, HCQ/AZM's benefit can occur much earlier that 10 days. Third, the cumulative doses of HCQ and AZM were determined not by length of stay but by the consulting ID physician. Fourth, I did a separate analysis not included inthe paper. I compared survival rates after removing all those in the "other" group who died in the first 5 days. The difference was still great (>25% absolute difference in survival) and still statistically significant.

Listen, I didn't expect to see these huge differences. But the fact that weight-adjusted HCQ dosing correlated with survival even stronger than cumulative dose is extremely strong evidence that HCQ/AZM were the cause of that difference. Younger pts were much heavier than older pts. Consequently, weight-adjusted HCQ cumulative dose shifted younger pts to a relatively lower dose and older pts to a relatively higher dose. That introduces a strong bias against weight-adjusted HCQ being associated with survival. Despite this bias, weight-adjusted HCQ cumulative dose was more strongly associated with survival than cumulative HCQ dose.

SMS