What to avoid

Treatments and programmes

• CBT framed as a cure. Supportive counselling for coping with a serious illness can help. Programmes that claim ME/CFS is maintained by “unhelpful beliefs”, and that changing them will cure you, are not supported. A 2026 review of 15 CBT trials reported small improvements in self-reported fatigue, but the trials couldn’t be blinded and relied on questionnaires, and reviewers have raised concerns about how some trials were counted. It doesn’t show that CBT treats the illness itself.³
• High-dose steroids. The PoCoVIT trial was stopped early after five serious blood clotting events, with no benefit.^4,5
• Blood thinners or blood-filtering (apheresis) for “microclots”. A careful UK study of 117 people with Long COVID who were never in hospital found no clotting abnormality and no extra fibrin “microaggregates” compared with people who had recovered from COVID. The authors say blood thinners aren’t justified outside clinical trials unless you have another reason to take them.⁶
• Immunoadsorption outside a trial. It showed no benefit in unselected patients (Charité, 2026).^7,5 It may help some people with specific autoantibodies,⁸ but only in a research setting for now.
• Expensive “protocols” and clinics that promise recovery, especially with upfront packages.

The debate about graded exercise therapy

Some Australian clinics still describe graded exercise as well supported by evidence. They aren’t making that up, but the evidence is weaker and more disputed than they may suggest. Here are both sides.

Why some clinics still recommend it

• The Cochrane review of 8 trials (1,518 people) concluded exercise therapy “probably” reduces fatigue at the end of treatment, rated moderate certainty: roughly 3 points on a 33-point fatigue scale. It is still published. Cochrane abandoned a planned update in 2024.^9,10 RCT
• The PACE trial (641 people, 2011) found GET modestly improved fatigue and physical function, and people walked about 35 metres further in six minutes than with usual care.¹¹ RCT
• GETSET (2017, 211 people) required post-exertional fatigue or malaise to enrol and still found a benefit on fatigue at 12 weeks.¹² RCT
• Trial safety data pooled from 10 GET trials found no excess of people becoming “much worse” (4% with GET vs 8% of controls), though the certainty is low and the analysis was written by a PACE author.¹³
• Several UK medical bodies objected to NICE. The Royal Colleges of Physicians, GPs and Psychiatrists said NICE’s definition of GET doesn’t reflect the “personalised paced exercise programmes” used in practice, while agreeing the term GET should be dropped.¹⁴ A 51-author critique, including the director of the UNSW Fatigue Clinic, argues NICE misapplied its evidence-grading method.¹⁵
• Australia’s only existing ME/CFS guideline (RACP, 2002) says graded exercise “may be effective”. It hasn’t been updated since 2002; the NHMRC’s replacement is due in 2027–28.^16,17

Why most guidelines have moved away from it

• Self-reported outcomes in unblinded trials. Patients knew which treatment they had and rated their own fatigue, which is especially prone to bias. NICE’s systematic review rated every GET outcome low or very low quality for this reason.¹⁸
• Objective measures barely moved. In PACE, apart from the walking test, there was no difference in fitness, days off work or benefit claims.^19,20 Using the trial’s original definition, recovery rates were 4% with GET versus 3% with usual care; the PACE authors dispute that reanalysis.^19,21
• Most trials didn’t require post-exertional malaise. In PACE only about half met a definition requiring it. The US AHRQ found that without the broad Oxford-criteria trials there wasn’t enough evidence to judge GET at all.^11,22
• Benefits faded. Longer-term, the groups in PACE and GETSET no longer differed.^19,23 A 2024 trial in 241 children found GET no better than activity management.²⁴
• Harm wasn’t well measured. In PACE, serious adverse events were twice as common with GET (8% vs 4%), though judged unrelated to treatment.¹¹ Patient surveys report worsening in 54–81% of people after GET. Surveys can over-represent bad experiences, but the findings are consistent across many surveys.^25,26
• Independent reviewers were cautious too. Germany’s IQWiG (2023) could not weigh GET’s benefits against its harms, and withdrew the “hint of benefit” in its draft because serious harm couldn’t be ruled out.²⁷

Where guidelines stand now

The bottom line: almost no one now defends rigid, push-through exercise. Whether a slow, symptom-led increase helps people with PEM is genuinely unresolved. If you’re offered a programme, use the questions on our pacing page to check which kind it is.

Tests

• VO₂ max or other maximal exercise testing is a known PEM trigger. The exception is a formal two-day CPET done specifically to document ME/CFS, and only if you decide the trade-off is worth it.
• Private “GPCR autoantibody” tests used to diagnose ME/CFS or Long COVID or to choose a treatment. In a 2026 German study of 206 people after COVID, these antibody levels were no different in people with post-COVID fatigue.³⁴ They’re a research tool, not a diagnostic test.
• Private panels sold as an ME/CFS workup. They usually miss the tests that matter. See Tests.

Apps and devices

• Fitness-oriented coaching, e.g. “strain” scores, step goals, “you’re recovered, push harder”. See Wearables.

Supplements

• Stimulant-containing blends, such as flavoured “3-in-1” creatine with caffeine, or NMN blends with green-tea extract
• Starting lots of things at once. You’ll never know what helped (or harmed).
• Unverified internet cures. Check them against the evidence first.

Habits

• Spending your good days. A good day is a sign pacing is working, not a sign you can do more. See Pacing.
• Waiting until you’re bedbound to rest. Cut back early during a decline.
• Visiting when unwell (for family and friends). New infections are the most common relapse trigger.

How to spot hype

Ask these questions about any new treatment claim:

1. Was there a placebo group? In ME/CFS, 40% or more of people can feel better on a sugar pill. Without a placebo group, a high success rate can mean very little.
2. How many people? Results from 10 or 20 people often don’t hold up.
3. Was it in people, or in cells in a dish? Lab results are a starting point, not proof.
4. Was it ME/CFS, or a related condition? Long COVID and fibromyalgia results may not transfer.
5. Who’s selling it? Be wary when the person making the claim profits from the treatment.
6. Has anyone else replicated it?

Our evidence labels are designed to help with exactly this.