← Treatments

GLP-1s for Long COVID and ME

Treatment Experience Survey Results

120 respondents /  April 2026

of respondents

In 2023 and 2024, dozens of patient anecdotes began circulating on Reddit, Facebook, and other communities, describing dramatic symptom relief from their Long COVID or ME from GLP-1 agonists like Tirzepatide and Semaglutide.

But the myriad of scattered anecdotes revealed that while some patients experienced game-changing improvements, others worsened permanently. Tens of thousands of patients have been trying these drugs blind, without structured data on who responds and who doesn't.

To solve this problem, we posted on Twitter, Reddit, Facebook, and in private group chats for ME and Long COVID, advertising the first ever survey for respondents who had tried GLP-1s. We ended up collecting 120 responses, making this the largest structured patient datasets on the subject. What follows is an analysis of our most pertinent findings.

Note: due to our limited sample size, there are a number of limitations present in the data compared to a more rigorous clinical trial. These are detailed in the Limitations section at the bottom of the page.

Results at a Glance

Improvement or Worsening?

As expected, response to GLP-1s was extreme at both ends.

53% of respondents reported some degree of improvement in their symptoms, with a whopping 36% describing themselves as "much improved" or better. A small number reported near-complete recoveries, returning to work, exercise, and activities they hadn't managed in years.

A mere 20% noticed no change at all, while a concerning 28% reported worsening of some kind, with some patients reporting permanent baseline reductions even after discontinuing the drug.

Same Drug, Two Extremes

These two anecdotes represent the most improved and most worsened patients in our dataset.

Patient #45

Started running and strength training, no more weekends in bed

Drug
Tirzepatide
Condition
Long COVID + ME, 3 years
Baseline
Part-time work, housebound or bedbound on days off
Subtypes
PEM/fatigue, brain fog, GI symptoms
“I took the first dose of Mounjaro and within 24 hours realised that my energy levels had increased and brain fog had lessened. I took note over the following weeks and realised that inflammation was decreased and brain fog had dropped by more than 50%. In the 18 weeks since, I’ve started running twice a week and strength training once a week, as well as back to work more days and not spending weekends in bed.”
Patient #67

100% bedbound four months later

Drug
Tirzepatide
Condition
Long COVID + ME, 4.5 years
Baseline
Housebound, limited standing
Subtypes
PEM/fatigue, brain fog, POTS/dysautonomia
“About 3 weeks after starting the tirzepatide, I experienced a sudden decline. I woke up one morning feeling severely worsened fatigue, sensory hyper-sensitivity, muscle weakness, internal vibrations, and visual artifacts. My heart rate was >100 bpm laying flat in bed. I went from being homebound but able to complete [daily activities] independently, walk between rooms in my home, shower with a shower chair, socialize and watch TV for some time every day, to being 100% bedbound in a dark, quiet room 24/7, relying on caregivers for all [daily activities]. As of now (4+ months later), I am still 100% bedbound and have not recovered to my previous baseline.”

Anecdotes

Hover over a dot to read the corresponding anecdote 💬

Our survey had a free prose section to let patients describe their experience in their own words. We selected a few of these anecdotes to represent a typical response for each category.

How Quickly Did It Work?

Click a bar to read a patient anecdote 💬

Among respondents who improved and described a timeline, a third noticed same-day improvement, their brain fog lifting, post-meal crashes resolving, and inflammation quieting quickly after their first dose. Another large group noticed improvements within one to two weeks, often after a dose increase.

The inverse held just as clearly: patients who felt nothing by the 4–6 week mark almost universally ended up as non-responders.

For those who mentioned discontinuing, at least one reported returning to their prior baseline within months, and several described symptoms creeping back as each weekly dose wore off. On the worsening side, the story is more nuanced. Several patients reported their deterioration appeared permanent at time of survey, but others returned back to their previous baseline.

n=42 respondents who described a specific timeline in their free-text responses.

By Symptom Cluster

The clearest subgroup signal in the data is patients with POTS / dysautonomia: of the 78 respondents with autonomic symptoms, 46% improved, the lowest of any cluster, while 32% reported worsening. By contrast, respondents with MCAS and GI symptoms both saw 60% improvement rates.

So why do most clusters hover around the overall average of 53% improved, while POTS patients sit meaningfully below it? This may reflect the complex effects of GLP-1s on the autonomic nervous system, which can exacerbate heart rate and orthostatic symptoms, a pattern echoed in many of the anecdotes.

Long COVID vs. ME

While we didn't find as large of a difference between the groups here as we expected, a notable difference is among ME-only respondents: of the 12 patients in the survey whose ME was not linked to a COVID infection, none reported "Much worse" or "Very much worse" outcomes. The worst any ME respondent experienced was minimal worsening, compared to 18% of Long COVID patients who reported significant deterioration.

Note: these sample sizes, especially ME (n=12), are likely too small to draw firm conclusions, so treat these as directional signals only.

Tirzepatide vs. Semaglutide

3×
more "very much improved" outcomes on Tirzepatide
Tirzepatide 14%  /  Semaglutide 5%

While the overall improvement rates are nearly identical - 54% for Tirzepatide and 50% for Semaglutide - the distributions tell a different story: Tirzepatide produced dramatically more "Very much improved" outcomes (14% vs 5%), while Semaglutide responses clustered toward modest improvement and no change.

Again, this is worth treating carefully: the Semaglutide sample is small (n=20), so the gap may partly reflect noise. But it does align with reports that Tirzepatide's dual GIP/GLP-1 agonism produces stronger anti-inflammatory effects in some patients.

Which Symptoms Improved?

44%
Brain fog
37%
Fatigue
27%
Exercise tolerance
22%
PEM
21%
Inflammation
11%
MCAS symptoms
10%
Sleep quality
6%
Migraines
5%
POTS / dysautonomia

Tap 💬 items to read a patient anecdote

Brain fog was the most consistently reported improvement - and often the fastest to be noticed. Some patients described a cognitive clarity they hadn't experienced in years, sometimes within the first day of dosing. Fatigue and exercise tolerance frequently improved together, suggesting a shared underlying mechanism.

PEM reduction was reported by 12% of all respondents - several patients described being able to do light activity without triggering crashes for the first time. Inflammation, joint pain, and MCAS symptom reduction followed a similar pattern, consistent with GLP-1s' known anti-inflammatory effects.

n=63 respondents whose anecdote mentioned at least one symptom improvement

Side Effects

30%
Nausea
20%
Constipation
19%
Post-dose fatigue
16%
Insomnia
13%
Diarrhea
10%
Appetite suppression
10%
Heartburn / reflux
7%
Dizziness
5%
Depression / mood changes

Tap 💬 items to read a patient anecdote

Rare but serious
Gallstones2 patients
Suicidal ideation1 patient, at 5mg Tirzepatide
Sensorineural hearing loss1 patient

GI side effects dominated: nausea and constipation were by far the most commonly reported complaints, consistent with GLP-1 effects on gut motility. Most resolved within the first few weeks or with dose reduction.

Insomnia emerged as an early warning sign: several patients who eventually worsened reported sleep disruption in the first days before other effects became clear. Appetite suppression severe enough to affect protein intake was also noted, with a handful of patients describing muscle loss concerns over longer use.

The rare serious events, while low in absolute count, warrant attention given the small sample. The suicidal ideation case occurred at 5mg tirzepatide and resolved on stopping.

n=83 respondents whose anecdote mentioned at least one side effect

By Baseline Severity

Severe
35%
improved
25%
no change
40%
worsened
n=20
Moderate
55%
improved
22%
no change
24%
worsened
n=55
Mild
58%
improved
16%
no change
26%
worsened
n=43

Severity at baseline was one of the strongest predictors of outcome. Severely ill patients had the worst results - only 35% improved while 40% worsened, nearly double the survey average. Moderate and mild patients were nearly identical and both tracked close to the overall average.

The implication is significant: severely ill patients may lack the physiological reserves to respond to GLP-1s' anti-inflammatory effects, or may be more vulnerable to the side effects that push them further down. The data suggests that patients attempting GLP-1s during a period of relative stability rather than at their lowest point may meaningfully improve their odds.

n=118 respondents.

By Illness Duration

↑ worsened
improved ↓

Surprisingly, outcomes by illness duration were nearly the opposite of what we expected. The 6+ year groups had the best results - 67% improved and only 14% worsened - while the 2–4 year group fared worst, with the highest worsening rate at 38%.

The early group (<2 years) may reflect Long COVID patients still in an acute or subacute phase, where inflammatory pathways are more active but also more volatile. The 2–4 year dip is harder to explain and may reflect a window where disease processes are entrenched but the body hasn't adapted. We simply cannot explain the results of the 6+ year groups, other than being very happy that this is the case!

n=104 respondents who provided a parseable illness duration.

What Didn't Correlate?

Not everything we looked at produced a meaningful finding. A few things we tested that showed no clear correlation with outcome:

Past treatment successes. We asked patients to list which treatments they had success with in the past, thinking we'd find correlations. We didn't - prior treatment success didn't predict GLP-1 response in any direction, and our sample sizes were too small to draw meaningful conclusions.

Concurrent medications. We looked at LDN, antihistamines, Metformin, Abilify, and others being taken alongside GLP-1s. Most groups had n<6 or less, too small to make any conclusions, though patients concurrently taking LDN (n=33) appeared to underperform.

Dose amount. We could not identify a reliable population-level optimal dose. Some patients did best at 0.25mg; others at 7.5mg or higher. The right dose appears to be highly individual - consistent with community experience, and the smart thing to do seems to be to start as low as possible, but we weren't able to tease that out with our data.

Limitations

We realize that our sample is small, and we do not mean for this data to replace real clinical trials. It is simply meant to bridge the gap between individual anecdotes on social media and waiting a year for the Scripps trial to read out. Here are limitations we acknowledge in our data:

Self-selected sample. Patients who had a clear positive or negative experience were more likely to be motivated to write about it - patients who felt little or no effect may have simply had less to say. This likely biases the sample toward the extremes, and improvement or worsening rates should not be treated as representative of all LC/ME patients.

No control group. This is a pure observational survey without a placebo comparison. Some reported improvements may reflect natural disease course, seasonal variation, regression to the mean, or concurrent interventions, not GLP-1 effects.

Retrospective self-report. Patients described their experience after the fact, often months later, which could have led to memory bias and anchoring to the final outcome. For example, though we tried to control for this, the 7-point outcome scale is subjective - "minimally improved" may mean different things to different patients.

Survivorship bias. Patients who recovered and left the community, or patients who became so severe they were unable to use their phones, may be underrepresented.

Small subgroup sizes. Most subgroup analyses involve n=10–40 and should be considered directional signals only. We ran Fisher's exact tests and chi-square tests on all four subgroup comparisons (population, drug, baseline severity, illness duration) - none reached statistical significance, even without multiple comparison correction. The sample is simply too small to distinguish real subgroup differences from chance variation.

Raw Data

Want to do your own analysis? We've made the anonymized survey dataset publicly available.

View dataset

9 respondents opted out of data release and have been excluded.

What's Next?

We're looking to start the next treatment in our Treatment Experiences Survey series soon. Any ideas for which one we should do? Email us!

Looking to get notified when the next survey drops? Subscribe here: