The Effect of Fluvoxamine and Metformin for Fatigue in Patients With Long COVID, 2026, Reis et al

[ME/CFS Science Blog]

A sensitivity analysis might have helped

Yes or a check to see if dropouts differed in their baseline measurements. They don't give much info on them.

if fluvoxamine did indeed have an effect, did it improve mood or fatigue?

They did try to check for this possibility by looking at an interaction between fatigue improvement and baseline anxiety/depression measured with the EQ-5D-5L. They found no significant interaction, suggesting the fatigue improvements weren't fully due to depression/anxiety.

In exploratory analyses examining whether baseline anxiety/depression modified the treatment effect on fatigue, we found no evidence of interaction(treatment-by-baseline anxiety/depression interaction term centered near zero, with the 95% CrI crossing zero).

They didn't have a depression questionnaire though to test this more thoroughly.

To me it seems like this was a reasonable trial and there might have been an effect. It might be too small to be of much importance though. Studies in MS or lupus estimate the minimal important difference the be higher than 0.45 (which was also the effect the trial authors used in their power calculation).

Sources:

Minimal clinically important difference for 7 measures of fatigue in patients with systemic lupus erythematosus - PubMed

Fatigue is a common and debilitating component of SLE. Estimates of MCID will help to interpret changes observed in a fatigue score and will be critical in estimating sample size requirements for clinical trials including fatigue as an outcome.

pubmed.ncbi.nlm.nih.gov

https://www.sciencedirect.com/science/article/abs/pii/S2211034819303190

 

[ME/CFS Science Blog]

Fluvoxamine has never been tested in ME/CFS as far as I know although I remember there have been a small number of trials of other SSRIs as well as TCAs, MAOIs and the SNRI duloxetine.

Looks like fluvoxamine is a serotonin reuptake Inhibitor but that it's a bit different than other antidepressants because of its high affinity for sigma-1 receptors.

Would be interesting to compare fluvoxamine to another SSRI and a placebo to see if it beats both.

 

[Mij]

Common antidepressant can treat long COVID fatigue symptoms study

"A common and low-cost antidepressant can successfully treat fatigue associated with long COVID, according to a new study.

Co-led by researchers from McMaster University in Ontario, the study found that the drug fluvoxamine, which is commonly sold under the name Luvox, significantly reduced fatigue symptoms in a clinical trial of 399 adults with long COVID. It is one of the first times a medication has been shown to effectively treat the condition.

“This is an important step forward for patients who have been desperate for evidence‑based options,” senior author and McMaster University professor Edward Mills said in a news release.

“Fluvoxamine showed consistent and meaningful benefits, and because it’s already widely used and well understood, it has clear potential for clinical use.”

​

 

[Jaybee00]

Would be interesting to compare fluvoxamine to another SSRI and a placebo to see if it beats both.

Probably more important to do a trial, even smallish, for Abilify/rexulti/vraylar (diff mechanism from fluv.) which should yield a higher positive effect than the effect fluvoxamine demonstrated here.

 

[rvallee]

Seems like the most important finding is how natural improvement is common, but we knew that already. The changes in the fluvoxamine are quite small, not nearly enough to overcome the flaws and noise, and the fact that no blinding can withstand the obvious side effects of SSRIs. It's so common for people who are desperate for any relief to take those effects as some sort of "I can feel it working", even if it makes no real difference.

This isn't the first fluvoxamine trial, it was hyped a few years ago. Given the many problems with doing clinical trials, it's very unlikely that this would hold up, and it's too small to merit trying again.

“Fluvoxamine showed consistent and meaningful benefits,

The problem with this is that meaningful to clinicians desperate to find anything that appears to slightly move some needle, any needle, is not the same as meaningful to patients. It's easy to find people who will report that acupuncture provides them with significant benefits, they just have to feel something, and SSRIs sure do.

The standards have to be significantly elevated here. This is way too little for this much hype. Stop hyping nothing, damnit.

 

[Utsikt]

Results:
Fluvoxamine showed a significant reduction in fatigue compared with placebo at day 60 (mean difference, −0.43 [95% credible interval {CrI}, −0.80 to −0.07]), with a sustained effect at day 90 (mean difference, −0.58 [CrI, −0.98 to −0.16]) [for the FSS scale].

They planned on having 500 participants in each group (but had stopping criteria in case of overwhelming evidence of efficacy or futility), and the trial was powered to detect an MCID of 0.45 for the FSS.

The protocol:

Sample Size

With the proposed sample size of 500 participants in each group, the study will have adequate statistical power to detect a clinically important treatment effect. A difference of 0.45 has been suggested as a minimally important difference (MID) of fatigue severity scale in multiple sclerosis patient population groups with chronic fatigue.2 Even with a 10% drop-out rate, our estimated power to detect a difference of 0.45 is above 90%.

Reference 2 is this paper on MS, where they found the MCID to be 0.45-0.88 for the FSS. The authors don’t explain why they used the lower bounds in this study. 0.58 (at 90 days) is only 32 % into the interval, so it’s at best «possibly clinically meaningful».

Which makes the decision by the independent data safety and monitoring committee (DSMC) to stop the trial due to the «superiority» of Fluvoxamine very difficult to understand:

We enrolled a total of 399 participants between 18 October 2023 and 10 February 2025 and randomly assigned them to fluvoxamine (n = 150), metformin (n = 111), or placebo (n = 138) (Figure 1). The DSMC recommended stopping the metformin group after the first interim analysis for futility on 18 September 2024 and stopping the fluvoxamine group for superiority on 19 February 2025.

Although it was based on predefined criteria, so the designers are to blame.

The stopping rules are supposed to be in the adaptive design report at the end of the supplementary file, but there must have been something wrong with their software because the formulas seem to be missing a lot of characters:

 

[EndME]

This looks a bit like a copy and past problem from someone copying stuff from LaTeX to word or similar.

The whole story is rather amusing: Bramante et al tested Metformin, Fluvoxamine (and some other drugs) in acute Covid and analysed this data for the prevalence Long-Covid (with the usual vague criteria). Because the other substances yielded null results they only trialed Metformin for Long-Covid and that trial yielded null results and those were reported a while ago. Did anybody check whether the Fluvoxamine placebo showed greater efficacy than the Metformin placebo?

I also wouldn't be suprised if this is just showing Fluvoxamine showing its usual effect that is also seen in trials of depression (which might have been excluded but depressive symptoms might still show relevant changes on the FSS ), especially because previous studies of this group with Fluvoxamine couldn't be corroborated.

 

[EndME]

The inclusions of people with diabetes who are already receiving metformin seems like a complication. I think we'll need to see how many people were in that group.

Yes, this could possibly create an effect, but I think one would have to see how large the group is. One should think a subset analysis should be possible. It's interesting that people already on Metformin were not excluded but people already on a different antidepressant were excluded. I think that is possibly quite valid as those are more likely to be suffering from depression or depressive symptoms.

 

[Hutan]

The Senior Author is Edward J Mills of

Department of Health Research Methods, Evidence, and Impact, McMaster University, Hamilton, Ontario, Canada, and VirX at Stanford, Stanford University, Stanford, California (K.T., E.J.M.)

McMaster University and specifically the Department of Health Research Methods, Evidence and Impact is a hotbed of psychosomatic thinking and Cochrane . Other faculty members are Jason Busse and Gordon Guyatt. The Department of Health Research Methods, Evidence and Impact seems to have been the driver of this study.


McMaster University helpfully has a feature article on Edward J Mills.

The People Behind the Impact with Edward Mills: A collective responsibility to look after each other​

Edward Mills, a part-time professor in the Department of Health Research Methods, Evidence, and Impact, is also a generous donor, guided by a strong sense of responsibility to care for one another.

The Department of Health Research Methods, Evidence, and Impact (HEI) can thrive thanks to the generous support of donors who share its commitment to advancing health research.

Among them is Edward Mills — “Call me Ed,” he says — a part-time HEI professor motivated by what he describes as “a collective responsibility to look after each other.”

Mills first connected with the department in 2002, when HEI was still known as the Department of Clinical Epidemiology and Biostatistics. He collaborated as a researcher with HEI distinguished professor Gordon Guyatt, who encouraged Mills to pursue a PhD focused on clinical epidemiology in the Health Research Methodology program.

“It was during this time that I met and developed friendships with other students who would later become brilliant faculty,” Mills shares. This includes current HEI faculty members P.J. Devereaux, Holger Schunemann, Elie Akl, and Jason Busse.

Mills describes his HEI learner experience as profound. “I was given the confidence to pursue an academic career and subsequently a business career in a closely aligned area,” he shares. “McMaster has always been a very collegial environment and so I worked closely with colleagues across the department and would subsequently also start businesses with them.”

Mills is referring to his friend and colleague Kristian Thorlund, a part-time HEI assistant professor. Both have previously created and exited companies including Redwood Outcomes and MTEK Sciences.

Redwood Outcomes appears to be a 'Systematic Health Research reviews for hire' company. They also do PR for companies wanting to promote health products.

More than ever, our clients in the pharmaceutical and biotech industries need to demonstrate from which of many competing treatments patients are most likely to benefit.​

​

Jason Busse and Gordon Guyatt are mentioned very often on this forum. They have close links with the health insurance industry and Cochrane. e.g.

Dr. Jason Busse has established multiple connections to the insurance industry through his clinical practice, research, and advisory roles:

Consultant to Prisma Health Canada: Dr. Busse serves as a consultant for Prisma Health Canada, a private company funded by employers and insurers that consults on disability claims. (PMC)

Clinical Director at Prisma Health Canada: In addition to his consultancy, Dr. Busse holds the position of Clinical Director at Prisma Health Canada, where he oversees the management of chronic disability claimants referred by private disability insurers. (http://orthohub.xyz)

Former Director at ATF Canada: From 2001 to 2010, Dr. Busse was the Director of the Complex Claims Division at ATF Canada, a private company that provided assessment and treatment for patients receiving long-term disability benefits. (http://orthohub.xyz)

Research in Insurance Medicine: Dr. Busse's research encompasses insurance medicine, focusing on areas such as disability benefits and workers' compensation. (Insurance Journal)

Collaborator with Cochrane Insurance Medicine: He is listed as a collaborator with Cochrane Insurance Medicine, an organization dedicated to evidence-based research in insurance medicine. (http://insuremed.cochrane.org)

Research on Impairment Ratings: Dr. Busse has conducted studies comparing editions of the American Medical Association's Guides to the Evaluation of Permanent Impairment, which are utilized in workers' compensation systems. (Insurance Journal)

These roles and research endeavors highlight Dr. Busse's active engagement with the insurance industry, particularly concerning disability assessment and management.

Call me cynical, but I think it's unlikely that this hyping of a rather doubtful (and, at best, inconsequential) benefit from an anti-depressant as a great treatment for Long COVID fatigue and the involvement of these people is a coincidence.

​

 

[Jonathan Edwards]

Call me cynical,

Let's all be cynicial together. Poor old Ed, only a 'part-time professor'; so busy with more financial issues, perhaps. And only at McSmarmster University at that, it seems. Backscratchers by name...

 

[Hutan]

Building lifelong connections

Mills first connected with the department in 2002, when HEI was still known as the Department of Clinical Epidemiology and Biostatistics. He collaborated as a researcher with HEI distinguished professor Gordon Guyatt, who encouraged Mills to pursue a PhD focused on clinical epidemiology in the Health Research Methodology program.

“It was during this time that I met and developed friendships with other students who would later become brilliant faculty,” Mills shares. This includes current HEI faculty members P.J. Devereaux, Holger Schunemann, Elie Akl, and Jason Busse.

Mills describes his HEI learner experience as profound. “I was given the confidence to pursue an academic career and subsequently a business career in a closely aligned area,” he shares. “McMaster has always been a very collegial environment and so I worked closely with colleagues across the department and would subsequently also start businesses with them.”

Mills is referring to his friend and colleague Kristian Thorlund, a part-time HEI assistant professor. Both have previously created and exited companies including Redwood Outcomes and MTEK Sciences.

Driven by a responsibility

As Mills puts it, “We are all connected and, to me, we all have a collective responsibility to look after each other.”

Ha, I've just noticed that Mill's seemingly altruistic quote about 'we are all connected' and 'looking after each other' immediately follows a section of the article about how Mills built lifelong connections with the likes of Guyatt and Busse. That juxtaposition creates substantial ambiguity about exactly who Mills has a 'collective responsibility to look after each other' with.

Of the two companies that Mills set up with his McMasters colleague, one produced systematic reviews and 'communications' for companies wanting to promote a health product. The other specialised in complex statistics and trial design to help generate evidence for companies developing heath products. In other words, those companies provided the infrastructure to help convince people that a drug or some other intervention should be used. I'm particularly concerned about the development and promotion of the highly sophisticated trial designs and analyses, the things like stopping trials early when they have the results they need. It becomes increasingly difficult for intelligent people to read a paper and be able to know whether the intervention truly worked or not.

It's easy to laugh about this stuff, the hype, the web of connections. But it seems likely that a whole lot of people are going to be given this anti-depressant (and probably other anti-depressants) on the basis of this flimsy evidence. Our thread on anti-depressants makes it clear that many of them can come with some nasty side effects, particularly when stopping. I don't know if fluvoxamine has those problems.

Those people prescribed anti-depressants will then have that on their medical records. That might have an impact on how they are viewed by medical professionals and how they are treated by health insurance. If there was a substantial treatment effect of course, those impacts would be dwarfed by the benefit. But, these are additional harms off-setting the uncertain and very small benefit this study reports.

 

[Evergreen]

They did try to check for this possibility by looking at an interaction between fatigue improvement and baseline anxiety/depression measured with the EQ-5D-5L. They found no significant interaction, suggesting the fatigue improvements weren't fully due to depression/anxiety.

What I'm thinking of is improvements in mood that change how optimistically you fill in a questionnaire. So those improvements could happen from any baseline level and affect the outcome of any questionnaire. I think you would need pre- and post- measures of depression for that.

It seems very odd to me not to include a pre- and post- measure of depression when testing an antidepressant regardless of what you're targeting with that drug. Am I missing something?

 

[forestglip]

It seems very odd to me not to include a pre- and post- measure of depression when testing an antidepressant regardless of what you're targeting with that drug. Am I missing something?

I think they did test that. It seems the results were similar for fluvoxamine whether they used the outcome of fatigue or depression.

When the EQ-5D-5L anxiety/depression dimension was modeled as an outcome, fluvoxamine showed high posterior probabilities of improvement at days 30, 60, and 90 (≥97.6% at all time points), whereas metformin did not show consistent benefit (Supplement Figure 1).