Update, and it’s a big one: this paper has now been retracted by the authors, who say that they are “unable to vouch for the veracity of the primary data”. None of its conclusions can be regarded as valid. Read on for historical interest only! But note that this is not the only evidence against HCQ as a therapy for coronavirus.
At this point, it’s getting hard to see how the idea of a hydroxychloroquine (or hydroxychloroquine/azithromycin) therapy for coronavirus infection can be taken seriously. I reviewed some of the recent studies here, but missed a May 11 preprint from France that had claimed benefit for the combination. No matter, though: this was just withdrawn by the authors, who say that they are revising the manuscript.
This morning brings this paper from The Lancet. Update: this paper has set off a great deal of controversy – see here. It’s a retrospective look at registered patients across 671 hospitals around the world, and it covers four patient groups: treatment with chloroquine, chloroquine plus a macrolide antibiotic (azithromycin, doxycycline), hydroxychloroquine, or hydroxychloroquine with a macrolide. All of these patients were started on these treatment regimens within 48 hours of diagnosis. The study specifically excludes those patients whose treatment started later, anyone whose therapy was started while they were on mechanical ventilation, or anyone received remdesivir as well. Early treatment in less severe patients only, in other words.
96,032 patients were registered in these hospitals with the coronavirus during the study period (December 20, 2019 to April 14, 2020); this is a large data set. The mean age of the patients was just under 54 years, 54/46 male/female. 14,888 of them were in the treatment sets defined above: 1868 got straight chloroquine, 3783 got chloroquine with a macrolide, 3016 received hydroxychloroquine by itself, and another 6221 got HCQ with a macrolide). That leaves 81,144 patients as a control group getting other standard of care. Let’s note at the start that the authors controlled for a number of confounding factors (such as age, sex, race or ethnicity, body-mass index, cardiovascular disease and risk factors, diabetes, lung disease, smoking, immunosuppressed condition, and overall disease severity). How’d it go?
Judge for yourself. The mortality in the control group was 9.3%. The mortality in the chloroquine group was 16.4%. The mortality in the chloroquine plus macrolide group was 22.2%. The mortality in the hydroxychloroquine group was 18%. And the mortality in the hydroxychloroquine plus macrolide group was 23.8%.
Let’s look at cardiac arrhythmia. The 0.3% of the control group developed new arrhythmias during their hospitalization. But 4.3% of the chloroquine treatment group did. And 6.5% of the chloroquine plus macrolide group. As did 6.1% of the hydroxychloroquine group. And 8.1% of the hydroxychlorquine plus macrolide group.
There are other interesting things about this paper (for example, it confirms earlier reports that ACE-directed therapies are associated with a survival benefit in coronavirus patients). But I’m going to leave it at this. There was no evidence whatsoever of any benefit with any of these treatment regimes. There was significant evidence of harm. Here’s how it works: when something is real, you continue to see a real signal as you collect more and better data. When something is not real, it disappears. Tell me again why anyone should be advocating such treatments. But your reasons had better stand up to 14,888 patients versus 81,144 comparators. Make it good.