The Elephant in the Fluvoxamine Room

A fraught debate on repurposed drugs

“Weigh Scale” by haven’t the slightest is licensed with CC BY-ND 2.0

Update Sep 6th: The fluvoxamine Together Trial preprint came out Aug 23rd.

First, the good news in COVID-19 trialing. On Aug 11th, the repurposed drug fluvoxamine’s Phase III trial results for COVID-19 came out in the Together Trial. Through a 1,480 patient, randomized double-blind clinical trial it has been found to reduce risk of hospitalization with greater than 95% confidence (relevant slides start at 32:40), which is the clinical standard for reporting statistical significance. Caveating that the news is still so fresh that there is yet to be a preprint posted, we do already know:

  • This was a double-blinded study: the researchers had to generate their analysis process before being unblinded. This means they couldn’t do the equivalent of p-hacking, or modifying analytical parameters in order to achieve a statistically significant outcome.
  • Intervention and control arms were almost identically sized.
  • The first endpoint was defined as # of hospitalizations or ER observation stays > 6 hrs (I will call these ‘hospitalizations’ for short). There were 74 such hospitalizations in the intervention group. There were 107 such hospitalizations in the control group. This endpoint achieves the 95% confidence clinical standard. (Raucous applause!).
  • The 2nd endpoint was simple mortality. There were 17 fatalities in the intervention arm, and 24 fatalities in the control arm. However, despite that the Relative Risk of fatalities between the two arms was almost identical to the hospitalization endpoint (0.70 vs 0.68), the absolute number of fatalities in both arms is too scant for this endpoint to achieve the 95% confidence clinical standard.
  • That it failed to achieve significance or the mortality outcome is already becoming the message about this study,. However, the stats behind this endpoint should be viewed with further circumspection:
Image credit: by harvest316 licensed with CC BY-NC-SA 2.0.
  • If it is found that the fatalities all followed hospitalization, then we would need to recognize that the two endpoints aren’t independent of each other, and so need not be considered in isolation of each other. Think of it as a funnel with layers. If in looking at the data one can guarantee that a certain proportion of those hospitalized ultimately become a fatality, that would represent a dependency. In that case one should, and indeed must, write-off the need for a 95% confidence interval for the fatality endpoint given the hospitalization data. Whether this progression dependency is the case will probably be evident from the preprint once it’s posted.
    Update Sep 6th: the preprint came out but doesn’t shed further light on this item.

So overall, (and even taking the hospitalizations endpoint exclusively without consideration of mortalities), this is very good news on its own. As humanity we should celebrate these results and the people who brought them to us, especially the participants and early trial research investigators & funders.

Unfortunately there’s a fly in the ointment. Unsurprisingly, new lines of rhetorical scrimmage are forming in the repurposed drugs debate.

As that post’s medical author writes,

“No, an interim analysis of an ongoing study is not enough to change treatment guidelines”

The camps are divided between:

  1. The community of public health decision-makers who’ve maintained there hasn’t been sufficient evidence to implement a fluvoxamine Emergency Use Authorization (EUA) for “an unapproved use of an approved drug”, or for NIH to update treatment guidelines, and want to delay for ‘reasons’.
  2. The proponents of an FDA EUA / NIH guidelines update who argue the evidence is there now, and perhaps even since months earlier.

(EUA background: Essentially, physicians understandably don’t want to get sued for prescribing a drug for an off-label indication in case any particular patient’s condition deteriorates for any reason and it gets blamed on the prescription. If the FDA has provided an EUA and/or NIH has endorsed a repurposed drug for treatment in its covid guidelines, then the physician has the necessary cover.)

Now we come to the matter of the elephant in the room:

(image credit: kaffeebart | Unsplash)

The decision to approve-or-delay is rooted in health politics. This is because the earliest pandemic Ph.II studies that showed a high degree of efficacy for fluvoxamine came out starting nine months ago already:

And if nationwide awareness was needed, that was achieved in March this year:

Rather than explore an expedited emergency approval, (one that would generate reams of statistically-significant data), NIH responded to these data by incorporating fluvoxamine as one of six drugs in NIH’s own ACTIV-6 Phase-III trial.

And thence the elephant: How many victims of COVID-19 have we arguably lost at the altar of requiring prior Ph-III evidence? If health decision makers had decided to grant an Emergency Use Authorization immediately following any of these three publication events in Nov ’20, Feb ’21, or Mar ’21, these are the theoretical number of lives that could have been saved (as of this post’s writing, 13-Aug 2021):

Theoretical fatalities recoverable based on 29.2% recovery rate

It doesn’t matter whether I personally think these casualty figures can justifiably be laid at the feet of our health institutions’ decision-makers. The fact, is, this elephant is already foremost on their minds as they decide whether to approve or delay a fluvoxamine EUA today, in August 2021. Why? Read on:

Our elephant figures are already irrevocably baked-in to public perception of health authorities’ decision to update treatment guidelines or grant an EUA. They figure in in two ways.

  • Delay the treatment guidelines updating & EUA: If NIH/FDA delays, they’ll make the case that there isn’t enough evidence to change clinical practice. Their best political case is to rationalize delaying approval until they can approve either Merck (molnupiravir) or Pfizer (PF-07321332)’s antivirals (whose approvals will now face internal pressure to accelerate). Then they’ll approve as a cocktail protocol of therapeutics that includes fluvoxamine. Why? Because when the resulting epidemiological Case Fatality Rate (CFR) numbers come in at scale — tens of thousands of cases — the fatality reductions will be attributable only to the cocktail, and not fluvoxamine in isolation. As long as they can delay, there won’t be epidemiological results showing that the emperor’s been wearing no clothes the past 6 months.
  • Approve treatment guidelines updating & EUA: If NIH/FDA approves now, they open themselves up to political criticism that they could have acted sooner and saved lives that were lost. Because of the large numbers of case data to draw from, these CFR values won’t require a stats-trained academician to process the results like the TogetherTrial does; rather any motivated senator’s office will be able to see the CFR results jump off the Johns Hopkins pandemic tracker page to begin levying their accusation. If NIH/FDA approves, their retrospective lack of approval to date will be politicized.
    We will not be surprised when that criticism falls along party lines in the US. One party will say that the NIH-FDA could have approved fluvoxamine after any of Nov’20/Feb’21/Mar’21 (depending on which event met the critic’s standard of evidence). That’s a criticism that could not only cost our healthcare institutional decision makers their jobs, but more crucially their public image and legacy.

A wise institutional decision-maker would take the data, act on its merits, accept any digs against it that are due, and motivate internal reform that’s due from a retrospective look at the institution’s response to the pandemic. This more circumspect approach will be crucial for our ability to respond better as a society to future pandemics.

But rather than act on the data at hand exclusive of political considerations, however, I believe (and would feel very happy to be proven wrong!) that the politically-minded public health decision maker will play for time. Maybe their supporters will make the case that one therapeutic drug applied at scale could select for a drug-resistant virus variant. That conclusion would reflect poor reasoning ability however because we know from the same trial that fluvoxamine has no direct antiviral effect on viral load. If it doesn’t reduce viral replication, it has no selection pressure on the virus to mutate.

What will they wait for?

I believe NIH/FDA will wait for any of the newer drugs from Merck (molnupiravir) & Pfizer (PF-07321332) that are yet to complete trialing. When either of them eventually pass Phase III, likely the NIH’s own ACTIV-6 fluvoxamine trial will have just completed too, and the health institutional decision maker will be able to say, “Look at this great protocol we have, congratulations to us, we can now finally administer these lifesaving medicines to patients”. At current fatality rates, and assuming the hospitalization->fatality dependency applies, a decision to delay comes with a price of 116 foregone-saves per day in the US, and 2,600 foregone-saves per day worldwide (calculated from July-end US and world daily fatality rates).

I like to close with a solution rather than just highlighting a problem. I hope it’s clear that NIH and FDA left to their own devices are unlikely to make a decision based on the clinical data merits alone. But they do have influencers — high-status, accomplished individuals in the pharmaceutical community. Those influencers need to do what has worked before to influence controversial opinion: 10 to 20 of them need to write and co-sign a joint letter for publication in a high-profile journal recommending that an EUA be made. Further, (and the part that stings for proponents like me who felt this drug could have been provisionally approved on a trial therapeutic basis much earlier), they need to share in asserting that August 2021 was the earliest time that approval could have justifiably been given.

This will give cover to NIH and FDA decision makers that, in case of political blowback (or for that matter, if anything at all should go south with healthcare outcomes following the approval), that the co-signers have their reputational back. Absent this, don’t expect the EUA to be issued until antivirals are ready to come out alongside. And we can all estimate how much the pandemic will unnecessarily toll until then.

What do you think? Does the state of evidence justify NIH/FDA delaying EUA approval? Evidence or not, what do you think they will actually do? The comments section is just below, or tweet into my feed at @nairobih3 .

How does this apply to vaccines?

But, you say , if we have vaccines, then who cares about a repurposed drug? Aren’t repurposed drugs just a crutch for anti-vaxxers?

Photography credit: EMSKE Phytochem

Allow me to help you reframe, dear Western reader. You may have access to efficacious vaccines, but most of the rest of the world does not. Yet the world is still seeing over 500k cases per day with no outlook that that will change immediately. A thousand of those daily cases are in the East African country where I am located. Fluvoxamine is an off-patent, low-cost (typically around $10 per treatment), widely available drug (I took the above photo in Nairobi) whose distribution would outpace vaccine logistics in heartbeat. They can help buy time till vaccines get there.

— — — — — — — — — — — — — — — — — — — — — — — — — —


  • Because of attitudes like this I have to include the following (which in my opinion should be self-evident without having to specify): Covid vaccines are great, I am vaccinated and happily so, and I think everyone who can should try to get vaccinated with those vaccines that are approved in their respective jurisdiction. I am very pro-vax. I recognize however that not every one can avail of vaccines, especially in emerging markets. Emerging markets’ health ministries tend to follow Western health approvals, not lead ahead of them.
  • If the Together Trial interim analysis on fluvoxamine doesn’t end up bearing out a strong hospitalization-leading-to-fatality connection, I’ll update the article to highlight that.
  • Neither EMSKE Phytochem, its commercial partners, nor myself have a financial interest in fluvoxamine. We do in fact have a financial interest in rationality being applied to approval processes for repurposed drugs.



Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
EMSKE Phytochem

EMSKE Phytochem

Coming from a multidisciplinary technical background in Silicon Valley, SE Asia, & East Africa, the author builds awareness of plant medicines. Tw: @EMSKEPhyto