A couple of months ago my colleague handed me an article for the blog investigating the use of the anti-depressant Fluvoxamine (Luvox) to treat COVID-19. I briefly reviewed the paper. It was a randomized control trial performed here in the US published in JAMA under the research letter section. The trial only had 80 participants in each arm. There were 6 events total (all in the placebo group). Since the trial was so small I didn’t think Fluvoxamine was ready for the big time, so I forgot about it.
Since then, fluvoxamine has gained some momentum. Published this week in Lancet researchers in Brazil published a much larger trial investigating the effect of Luvox on symptomatic COVID-19. They found a significantly positive result. The article has received surprisingly little coverage in the media and has yet to make the rounds on social medial. Maybe we have become desensitized to COVID. Or it could be a fool me once shame on me situation ("plaquenil"- retracted, "ivermectin" - retracted, "azithromycin" - ineffective). Given the positive results I thought I would dive a little deeper into the article.
The study was a randomized, placebo controlled, multi-center trial performed in Brazil. The centers used for the study were unique. These were not tertiary centers or major urban centers. These were mostly smaller hospitalizes and clinics located outside of major cities. The hospitals ranged in size from 50-80 beds (fairly small with limited resources).
To be enrolled in the study you had to have typical symptoms of covid and a positive covid test within 7 days of enrollment. You also had to have one of many risk factors for severe disease. Examples would be: over 50 yo, diabetes, hypertension, chronic kidney disease, asthma, etc. The researchers targeted unhealthy individuals which will be reflected in their event rate.
The primary endpoint was a composite of observation period in the emergency room for > 6 hr, or referral to a larger care center within 28 days. This composite endpoint was used due to lack of available hospital beds during the pandemic. Secondary endpoints were viral load at the end of 7 days, time to hospitalization, death due to covid, and all cause mortality. They monitored for safety as well.
For their power analysis they determined minimal clinical utility to be a 37.5% relative risk reduction. This would require 681 participants in each arm. They recorded outcomes 24 hours after the initiation of treatment and planned on an intention to treat analysis.
They were able to enroll about 750 participants in each arm. The groups were well balanced. They lost no patients to follow-up. 84 participants did not complete the fluvoxamine protocol. Oddly, 64 participants stopped their placebo.
They found a relative risk reduction for the composite endpoint of 0.69 (0.53-0.90 CI) in their modified intention to treat analysis. They modified the analysis to include only events that occurred 24 hours after the first dose of the medication.
This corresponded with a number needed to treat of 22.
They presented their Bayesian analysis graph (very nice presentation) which showed a 99.7% chance of superiority over placebo. Which you can see below.
None of the secondary endpoints were statistically significant based on their intention to treat analysis. However, mortality was statistically significant in their per protocol analysis. More on this in a minute.
These results show that fluvoxamine has some promise. It showed a significant treatment benefit in a randomized placebo-controlled trial in which most of the participants tolerated the medication. However, the treatment setting was quite unique. It was in Brazil. In less populated areas with limited resources. It was during the peak of the pandemic when available beds were in short supply, so their > 6 hour observation period may be inflated. The individuals were unvaccinated as well.
Also, their target population was unhealthy individuals over 50 and likely lower class Brazilians. These participants likely have poor access to healthcare and limited understanding of underlying health conditions. It is highly unlikely they had access to any of the modern treatments we have here in the US, such as mono-clonal antibody infusions, home monitoring, high flow nasal cannula, BiPAP machines, and other advanced ventilation techniques.
I say all this to say that it is difficult to extrapolate the results from this setting to the US. For example, they found an NNT of 22 for fluvoxamine, which would be inferior to the NNT of around 10-12 for the antibody cocktail. Also, it will likely be inferior to Merck’s new covid pill Molnupiravir, which is likely going to carry an NNT similar to the antibody infusion once it is approved.
Another issue they had was the gap in results between their intention to treat (IIT) analysis and their per protocol analysis. In an intention to treat analysis the participants are analyzed in the group to which they are randomized no matter which group they end up in at the end of the trial. This preserves randomization. In this case, for the per protocol analysis they excluded all of the individuals that stopped the treatment early. Per protocol analysis break randomization, so they can’t really be trusted. Also, the participants likely stopped because they were getting better and didn’t see the need to continue taking the experimental drug. In this case, the per protocol result is exaggerated and shouldn’t be used as corroborating evidence.
Despite all of this I was quite impressed by their results. They speculate that the underlying mechanism for counteracting the effects of covid is due to the activating effects of fluvoxamine on an endoplasmic reticulum chaperone protein that mitigates the inflammatory response. This is a plausible bench to body thesis.
The cost of fluvoxamine is only $4 US. It is a pill that can be taken at home and can be safely given to a high risk cohort. I would consider its use in individuals that are skeptical about the antibody cocktail. Its utility may lessen once Merck’s covid treatment is made available by the FDA at an affordable price.
Fluvoxamine should be given strong consideration in the developing world where access to vaccines and antibody cocktails is limited. This relatively safe, low cost, easily disseminated medication can be procured at little expense to cash strapped governments for their citizens suffering from COVID.