Health Update: Push-Crash-Rebound. I was doing quite a bit better until I had to do some heavy physical work on the van about a week ago. I thought I got through it OK, but then the crash came, and I have been slowly rebounding again. This has become something of a theme! Five times since I exposed myself to too much heat, something has interrupted my recovery. That’s life on the road (sometimes…).
The STIMULATE-ICP project was the UK government’s big attempt to understand and treat long COVID. Began early in the long COVID saga (2021) it was notable for its cost 6.9 million pounds ($9.15 million US) and 30 organizations that participated in it.
The project ended in August, 2025 and has produced over 30 papers (!) including it’s centerpiece – the treatment study “Efficacy and safety of rivaroxaban, colchicine, and famotidine–loratadine with specialist supportive clinical care for fatigue in patients with post-COVID-19 condition in the UK: a multisite, open-label, randomised, controlled trial” that was recently published in none other than the Lancet Journal.
Drugs
The trial assessed three drugs, rivaroxaban, colchicine, and famotidine–loratadine against standard care. At 200 people per treatment arm, it was a nicely sized study.
They were the kind of drugs the UK health system was looking for: relatively inexpensive, repurposed drugs it could put to use without breaking the bank.
- Rivaroxaban – anticoagulant
- Colchicine – innate immune and inflammatory activation
- Famotidine plus loratidine
They might not have been the first choice of another group, but they were all plausible treatments.
Results
Three drugs – three failures
All the drugs failed to produce “clinically significant” results; i.e., results the patients would have clearly noticed.
In fact, it wasn’t close. The participants would have had to have improved by a minimum of 3 points on the FAS scale to achieve that. The closest they came was for colchicine and famotidine (1.49, 1.48). The anticoagulant rivaroxaban produced minimally negative effects on fatigue (-1.06)
Flawed Study
While the drugs were plausible choices, the way they were assessed made one wonder if this trial was doomed to failure. Ultimately, the trial appears to have prioritized size over precision – a dangerous decision in this very heterogeneous disease.
It was perhaps no surprise that back in 2021, a pernicious problem that exists today was present in spades then: vague (very vague!) criteria. The participants simply had to have ongoing symptoms (or a symptom) four weeks after the infection.
It was an open-label trial; i.e., everyone knew what treatment they were getting. Placebo effects could have crept in in the patients receiving the drugs and in patients not getting the drug (who might have had a negative reaction).
The two symptom report measures were used to assess treatment effectiveness: the Fatigue Assessment Scale (FAS), and the EQ-5D-5L visual analogue scale (EQ-VAS) didn’t begin to assess the wide range of symptoms found in long COVID.
The way the trial was put together suggested it was going to have a hard time succeeding.
The symptom assessment done at baseline to determine if a subgroup with, say, more mast cell symptoms responded to the mast cell drugs was similarly flawed. The assessments were apparently so broad that virtually everyone in the study met the criteria for belonging to all of the cardiorespiratory, neuropsychiatric, and mast cell subgroups.
The symptom assessments also appear to have been far cruder (yes/no to each symptom) than one would have hoped for at this point. Leonard Jason has long established that yes/no symptom queries tell us nothing about symptom severity in ME/CFS – and symptom severity is a crucial measure in diseases with so many symptoms.
His DePaul ME/CFS questionnaires – used in many studies – assess symptom frequency and severity. A frequently occurring, severe symptom would get the highest score. An infrequently occurring, mild symptom would get the lowest score. This study would not have differentiated between those two.
If the study had used something like the DePaul Questionnaire at baseline and in the 12- and 24-week assessments, it could have determined if people with more severe gut, skin, allergic-type symptoms benefited from the mast cell drugs.
The study also did not assess – and explicitly stated it would not assess – whether the participants met the criteria for ME/CFS. That seemed puzzling as including the criteria would have highlighted a more severely ill group – and might have been helpful for people with ME/CFS.
A ChatGPT analysis of the methodology and statistics suggested that while the study could have uncovered broad effects, it was not sufficiently powered to uncover smaller subsets that responded well.
The authors acknowledged the shortcomings of the trials: they noted that using a more long COVID-focused functional test could have helped, that the study lacked the statistical power to uncover subsets, and that the trial results could have been “confounded by placebo or expectation effects or mask smaller subgroups or endotypes that might or might not benefit from the interventions tested.”
The study analysis is not over, and the authors stated that they are trying to identify subgroups and describe “phenotypical and/or biochemical characteristics.”)
Summary
The STIMULATE-ICP trial was kind of a big, messy study! There were so many potential confounding issues that it’s a little hard to know what to make of it.
It does appear that none of these three drugs should be used to treat your run-of-the-mill long COVID patient, as even a possibly positive placebo result did not move the needle.
It’s possible that colchicine and famotidine/loratidine might be helpful for a subset of patients, but it will take further study to determine that. That’s not much bang for the buck for this large, undoubtedly expensive, trial.
In the end the drugs would have had to have a large and broad positive effect in order to be successful. Particularly with the type of drugs this trial focused on (relatively inexpensive) probably not going to happen with these diseases. The authors inability to subset their participants was huge.
Institutional Conservatism Prevails
An innate conservatism prevailed in the two early UK long COVID projects and the RECOVER project in the US.
Interestingly, the STIMULATE-ICP trial failed in some similar ways to the first round of the RECOVER treatment trials: it tested a few drugs in pretty large sample sizes, was not able to target or uncover subsets (at least, not yet), and did not assess biological outcomes.
An assessment of the principal study authors found that, like the RECOVER project, most of the authors came not from the ME/CFS, fibromyalgia, POTS fields but from cardiology, respiratory medicine, infectious disease, clinical trial management, etc.
It was not surprising, for instance, to see the authors choose to assess colchicine, which some of them had, no doubt, used to treat pericarditis, chest pain, and breathlessness. That ignored the fact, though, that the chest pain/breathlessness found in ME/CFS is probably caused by far different issues (dysautonomia, dysfunctional breathing patterns, microvascular problems).
The UK presented STIMULATE-ICP as its largest major long-COVID project, and it ended up producing dozens of studies. In the end, though, the project included one large, rather crude treatment trial, and despite the projects promise to explore multiomics, for instance, the studies published thus far have focused on the healthcare system and explored things like long-term trajectories, access to health care, demographics, electronic health risk factors, etc.
Long COVID was treated as a brand new disease, its connections to ME/CFS were discounted, and ME/CFS researchers were not apparently welcome. The studies found what we would expect given the UK’s experiences with ME: “only one-third of individuals under follow-up achieved satisfactory recovery”, “long Covid can double a patient’s cost to the health service”, “Long Covid can be a stigmatising condition”, “Long-COVID has a significant and long-lasting impact on individuals’ health and well-being”…
The £19.6 million 2021 NIHR funding round suffered from a similarly conservative, unimaginative, biology-lite approach. The largest study (LOCOMOTION) was intended to develop a “gold standard” NHS care model. (How do you develop a gold standard of care when you don’t know anything about the biology?). More “soft” science and even some shades of CBT era crept in in the brain fog study, which was intended to “develop and test neuropsychological rehabilitation”, a personalized self-management support study, a weight management program, how patients and families manage long COVID, and adaptive pacing. Of the 15 studies, only 4 (T-cell, 2 immune, and a lung MRI) dug into long COVID biology.
Like the RECOVER project, the UK government apparently punted; i.e., it reflexively turned to what it knew (epidemiology, assessing care, self-management) when it met up with long COVID, instead of diving in and trying to understand the complex biology present. Like RECOVER, its priority was not in breaking new ground in long COVID or ME/CFS pathophysiology. For that, we have to go to places like Polybio, Yale, the UCSF LIINC project, the OMF, Solve ME, MERUK, Invest in ME, the Germans, and others.
The UK has continued to fund long COVID research, but the two early large funding packages constituted its big shot. Forty million of the roughly 57 million pounds spent on long COVID were funded by those two packages.
About Face!
It seems that the NIHR does not want to repeat the mistakes of the past. In July, 2025, it opened a grant package on post-infectious diseases including long COVID and ME/CFS. The new grant package will conduct phase 2 clinical trials of “multiple different therapies simultaneously and efficiently”. These trials should be more targeted and flexible, investigate mechanisms, and may include Decode ME participants. Call these trials the antidote to the Stimulate ICP trial.
The process, though, is agonizingly slow. The applications – which were to design the treatment platform – closed in Dec. 2025. In March 2026, the committee gave a preliminary green light to two project teams. The two teams then had up to 18 months to design the platforms, which means that in the worst-case scenario the actual trials could conceivably not take place until 2028 (!). Let’s hope it gets going sooner. Funding for the actual trials is expected to be provided at a later date.
It’s good to see the UK and Germany produce post-infectious grant packages. (The US, embarrassingly, is still lumbering behind…)
Growing Pains in the Treatment Arena
No hits yet! Perhaps it’s not such a surprise.
Six years after long COVID showed up, I was able to find 14 largish (> 100 participants) long COVID clinical trials that have been completed. Only one drug (fluvoxamine) showed a modest improvement. The big (n=312) Wyller cognitive behavioral trial is the only other one to show benefit. (A blog will look into that trial.)
It’s a pretty poor showing thus far, but one has to ask if it’s really that much of a surprise. Metformin, Ivabradine, transcranial magnetic stimulation, and high-dose CoQ10 were kind of nice tries, but surely nobody expected dramatic results from them. Short-term Paxlovid (2 large trials), cognitive retraining, and computer puzzles probably never had much of a chance.
Two large studies were quite targeted. The Donezipil study attempted to repair a specific cholinergic deficit in the brain the Japanese believed was caused by HHV-6 reactivation. Likewise, temelimab targeted patients with HERV-W activation. Both, however, failed.
The field is learning. Instead of doing large clinical trials that simply assess treatment responses, clinical trials are now increasingly assessing the biological response to the drugs and attempting to biologically target patient subgroups. We didn’t know 6 years ago, or even a year ago, for instance, that 4 different types of microclots may exist in long COVID – each of which requires a different treatment regimen. There’s also an emphasis on doing smarter, platform trials that eat up fewer resources and allow more treatments to be assessed at once.
The next set of results will, hopefully, be better. RECOVER’s IVIG and stimulant trials should be winding up, and more powerful and targeted drugs are being assessed. Baricitinib, anti-inflammatory/JAK inhibitors, maraviroc plus atorvastatin, HBOT, a triple antiviral, two GLP-1 agonist trials, rapamycin, stellate ganglion block, long COVID LDN, Jarred Younger’s smaller but more comprehensive ME/CFS LDN trial, and the OMF’s LIFT pyridostigmine/LDN trial are some large or largish trials underway or getting underway. Of course, more small trials are underway as well.
