Graded exercise therapy

1. What it was

Version followed: the PACE-trial GET arm (UK, recruitment 2005–2008), as set out in its therapist manual (Bavinton, Darbishire & White, Graded Exercise Therapy for CFS/ME: therapist manual, final trial version 7, MREC version 2, dated 16 November 2004; online copy), the published protocol (White et al. 2007) and the main results paper (White et al. 2011). Earlier UK trial versions and later named variants, including the GETSET self-help version, are covered in 2.9. The current UK standard used for comparison is NICE NG206 (2021).

Source note. Sources include the PACE GET therapist manual (version 7, pages 1–63), the GETSET therapist manual, Wilshire et al. 2018 (PACE reanalysis), the FINE trial abstract (Wearden et al. 2010), the Forward ME / Oxford Brookes 2019 survey summary, Action for ME's 2025 summary of NICE NG206 rec 1.11.14, Retraction Watch (2025) on the Cochrane review, and King's College London's 2016 report of Roberts et al. Claims drawn from these carry [source: …]. Anything tagged [uncertain] is not confirmed. Per-arm adverse-event counts rest on Wilshire et al.'s summary of the PACE Lancet results paper.

This tab is a harm account for recognition and avoidance. It is not a how-to. NICE NG206 says not to offer people with ME/CFS any programme that uses "fixed incremental increases in physical activity or exercise", or one based on deconditioning and exercise-avoidance theories. PACE-era GET falls inside both definitions. [source: NICE NG206 rec 1.11.14, via ME Association reproduction]

What it was. Graded exercise therapy (GET) was a structured, physiotherapist-led programme for chronic fatigue syndrome (CFS, later CFS/ME). The person agreed a baseline of physical activity, usually walking, and then increased the time spent exercising in planned steps, and after that the intensity. The stated goal was to rebuild physical capacity, reduce fatigue and return the person to normal function, possibly to full recovery. [source: White et al. 2007; White et al. 2011, Procedures]

The graded-increase logic was not unique to GET. The same logic ran through the "graded activity" component of the CBT protocols for CFS of that era and through later "rehabilitation" programmes. This tab covers GET itself and the variants named in 2.9. [source: White et al. 2011 on the PACE CBT model] [synthesis on the shared logic]

Who developed and used it, where and when.

UK research groups in the 1990s, mainly in London (Barts / Queen Mary University of London, King's College London), Oxford/Edinburgh and Manchester. The best-known early trial is Fulcher & White (1997, BMJ). [source: Wikipedia GET page; MJA 2005 article listing the trials] [uncertain: Fulcher & White 1997]
The PACE trial (principal investigators Peter White, Michael Sharpe and Trudie Chalder) tested GET against adaptive pacing therapy (APT), CBT and specialist medical care (SMC) alone in six NHS specialist CFS clinics, with 641 participants. [source: White et al. 2011; Wikipedia PACE page]
NICE CG53 (2007) recommended GET, alongside CBT, for mild to moderate CFS/ME. From then until NICE NG206 in 2021 it was a standard offer in NHS specialist CFS services. [source: Wikipedia GET page, citing NICE 2007]
It was also recommended or used in the Netherlands, Belgium, Australia and New Zealand, and by the US CDC until 2017. Insurers and occupational-rehabilitation providers used it too. [source: Wikipedia GET page; MJA 2005] [uncertain on the extent of insurer use: not documented in the sources]

Stated rationale. The model held that CFS might be triggered by an infection or by stress, but was kept going by reversible factors:

physical deconditioning, meaning reduced cardiovascular fitness and reduced muscle strength;
an altered perception of effort, so that ordinary exertion felt harder than it was;
"exercise intolerance" and avoidance that grew out of relative inactivity.

On this view, symptoms after exertion meant an unfit body adapting to load, not ongoing disease, and gradual re-exposure to exercise would reverse the deconditioning and the intolerance. The manual also names sleep, mood and cognitive factors as likely perpetuating factors, but says GET does not target them directly. That is how it set GET apart from CBT, which did target illness beliefs. It does not name symptom focusing as a mechanism. [source: PACE GET therapist manual v7, pp. 21–23; White et al. 2011, Procedures]

How it was presented to patients and referrers.

To referrers, as the evidence-based, NICE-recommended first-line treatment. It was backed by randomised trials and a Cochrane review, and was described as safe. [source: Wikipedia GET page; NICE CG53 via Wikipedia]
To patients, as a hopeful, active route back to normal life, with recovery possible. A 2013 PACE paper (White et al. 2013) reported that 22% of people in each of the GET and CBT arms had "recovered", against 7% in the comparison group. A 2018 reanalysis using the protocol's own definition found 4% for GET, 7% for CBT and 3% for the comparison group, differences that were not significant (see 2.7). The 2011 results were widely reported in the media as showing that exercise and positive thinking could help or cure CFS. [source: Wilshire et al. 2018, citing White et al. 2013; Wikipedia PACE page for the media coverage]
Within the programme, as collaborative, flexible and mutually agreed. The protocol stressed negotiation, and the participant was described as owning the plan. Goals were to be behavioural and SMART, recorded on a jointly negotiated sheet. [source: White et al. 2007 protocol; PACE GET therapist manual v7, pp. 43–45]
Within the trial, a 2008 participant newsletter praised the therapies being tested while the trial was still running. Critics argue this could have biased participants' self-reports. [source: Wikipedia PACE page]

2. What was actually done

This describes a typical course in the PACE era, as a patient would have experienced it. Named variants are in 2.9. Where details come only from secondary accounts of the manuals, they are tagged.

2.1 Assessment and selection

Referral. People were referred by a GP to a specialist CFS clinic. In PACE, participants had to meet the Oxford criteria: six or more months of disabling fatigue affecting both physical and mental functioning, with other medical explanations excluded. [source: White et al. 2011, Study design]
Case definition. The Oxford criteria do not require post-exertional malaise (PEM). People whose defining problem was a delayed worsening after exertion were therefore entered into an exercise programme without being screened for that response. People with a wider range of fatigue states, including fatigue with a primary depressive picture, could also qualify. [source: White et al. 2011; Wikipedia GET page on the criticism of the criteria] [synthesis on the implication]
Severity. PACE took outpatients who could attend clinic. In practice this meant mild to moderate illness. Housebound and bedbound people were mostly outside the trial, but outside trials GET-style programmes were sometimes applied more broadly (see 2.9). [source: White et al. 2011] [uncertain on how widely it was used in severe illness]
First assessment. An initial session of about 90 minutes covered activity history, current activity, sleep and the person's illness beliefs, plus an assessment of physical capacity. Trial measures included a step test, a six-minute walk and the Borg exertion rating. [source: White et al. 2007; White et al. 2011]

2.2 How consent was obtained or pressured

In PACE. Participants gave written informed consent to random allocation. The information they received described all four arms as plausible treatments. [source: White et al. 2007 protocol] Critics later argued that the consent information did not disclose the investigators' links with insurers. [source: Wikipedia PACE page] [uncertain: not checked against primary records]
In routine NHS care, 2007–2021. GET (or CBT with graded activity) was often the only treatment a specialist CFS service offered. Declining it could mean discharge from the service. Patient surveys describe people feeling they had no real choice. [source: Kindlon 2011; Geraghty et al. 2019, as summarised] [synthesis on discharge practice; uncertain on how often it happened]
Benefits and insurance. Engagement with recommended treatment could be used as evidence in disability-benefit, insurance and employment decisions. Refusing GET risked being recorded as non-compliance with treatment. [uncertain: widely reported by patient organisations; no primary source is drawn on]
Children and young people. Families who declined graded programmes for a child with ME/CFS have reported safeguarding referrals, made on the grounds that refusing "recommended treatment" was itself a concern. [uncertain: reported by patient charities, not verified here]

2.3 Setting and staff

Setting. Outpatient clinic sessions, plus a home exercise programme. A session could be held by telephone if it could not be rearranged within five working days. [source: White et al. 2007; PACE GET therapist manual v7, p. 29]
Staff. Physiotherapists, and in PACE also one exercise physiologist supervised by a physiotherapist. Therapists were trained by a therapy leader until judged competent, then supervised individually each month and in groups every three months. [source: White et al. 2011, Procedures]

2.4 Session structure and frequency

In PACE. Up to 15 individual sessions: sessions 1–4 weekly, sessions 5–14 fortnightly, and session 15 a booster. The first session lasted up to 90 minutes and later ones up to 50 minutes. The manual is internally inconsistent on timing: its text puts the last fortnightly session in week 23 and the booster in week 39, while its summary table lists week 24 and week 36. [source: PACE GET therapist manual v7, pp. 29–31; White et al. 2011]
Typical review session.
- Go through the diary or exercise record for the past period: session times, exertion ratings, heart rate and symptoms.
- Discuss problems and beliefs.
- Agree the next step up.
- Set the plan until the next session. [source: White et al. 2007 protocol, "ongoing review, jointly planning the next fortnight"; PACE GET therapist manual v7, pp. 48–50]

2.5 Component techniques, in the order they were used

Education in the deconditioning model. The physiotherapist explained the illness as kept going by deconditioning, reduced strength and an altered perception of effort, with exercise intolerance growing out of inactivity. Participants were told to expect "a mild increase in fatigue or muscle stiffness/soreness" as a normal response to exercise. Symptoms were not supposed to disrupt sleep or basic function. [source: PACE GET therapist manual v7, pp. 21–23, 49]
Goal setting. The person chose long-term goals, such as returning to work, a sport or a valued activity, written as behavioural, SMART goals on a jointly negotiated sheet. The plan was then built towards them. [source: White et al. 2007; PACE GET therapist manual v7, pp. 43–45]
Activity diary and baseline. For the first two weeks the person kept an activity diary. The baseline was a level of aerobic exercise, usually walking, that the person could manage on 5 or 6 days a week, including on "bad days". It was negotiated, and renegotiated if the diary showed it was wrong. Each session was timed, with a Borg exertion rating and average heart rate recorded. "Boom-and-bust" swings were discouraged. In GET the baseline was the launch point for increases. [source: PACE GET therapist manual v7, pp. 42, 48–49; White et al. 2011]
Confidence check. The person rated from 1 to 10 how likely they were to complete the plan. A rating of 6 or below prompted the therapist to adjust the plan until the rating reached 7 or more. [source: PACE GET therapist manual v7, p. 47]
Agreeing a home exercise programme. Walking was most common. Swimming, cycling or exercise machines were also used. The therapist and the person agreed the type, the daily duration and, where used, a target heart-rate range. [source: White et al. 2007; White et al. 2011; PACE GET therapist manual v7, pp. 48–50, 58]
Heart-rate ceilings. At first the target heart rate was a rate not to exceed, not a goal. If the person went over it, they were to slow down or stop briefly, then carry on within the zone. Exercise above about 80–85% of maximum heart rate was not generally recommended. [source: PACE GET therapist manual v7, pp. 49, 60] A ceiling during exercise does nothing to detect PEM, which appears 12–48 hours later. [source: NICE NG206 definitions] [synthesis]
Planned increases in duration. Exercise time was increased in steps at reviews, by no more than about 20% at a time and without raising intensity, towards about 30 minutes on 5 or 6 days out of 7. Vink et al. (2025) argue from the same manual that the increases were in practice fixed and pre-planned rather than symptom-led. NICE NG206 defines GET by exactly such fixed incremental increases. [source: PACE GET therapist manual v7, p. 50; NICE NG206 definitions] [Vink et al. 2025: title only]
Planned increases in intensity. Only once 30 minutes was manageable was intensity raised, in stages of about 10–20%, for example by walking faster or adding hills. [source: PACE GET therapist manual v7, p. 50; White et al. 2011]
Setback management. A setback was defined as an exacerbation of symptoms. Therapists were to explain in advance, with a written setback plan, why exercise should be kept up, and during a setback the instruction was to "MAINTAIN exercise as much as possible". Cutting back was described as continuing the boom/bust cycle. The therapist was to check whether any part of the programme had contributed and adapt the plan promptly. A participant who resisted was to be encouraged to keep up as much as they could, with "additional support" if needed. The setback section uses the phrase "hurt does not equal harm". [source: PACE GET therapist manual v7, pp. 51–52]
Booster session. A single later session (week 36 or 39, see 2.4) to consolidate gains and plan further progression. [source: PACE GET therapist manual v7, pp. 29–31; White et al. 2011]

2.6 Homework

Daily exercise at the agreed duration and intensity, on 5 or 6 days a week.
An activity diary for the first two weeks, then an exercise record with session time, Borg rating and heart rate.
Applying the agreed step up at the start of each new period.
A written setback plan, to be followed by keeping exercise up during a setback.

[source: PACE GET therapist manual v7, pp. 42, 48–52]

What a person did between sessions was the main "dose" of the treatment. [source: White et al. 2007; White et al. 2011]

2.7 How "progress" was judged

Clinically: by adherence to the planned increases and by the person's own report of fatigue and function. [source: White et al. 2007] [synthesis]
In PACE: the primary outcomes were self-reported: the Chalder Fatigue Scale and the SF-36 physical function subscale. Both were changed from the original protocol before analysis. The post-hoc "recovery" threshold for physical function (60/100) was lower than the score that counted as disabled enough to enter the trial (65/100), and close to the mean score of people with class II heart failure. Objective measures added little. In the GET arm the six-minute walk improved by about 10–12% relative to the comparison group. Step-test fitness did not differ between arms, nor did employment loss or benefit receipt (benefit receipt rose in all arms). Actigraphy was collected only at baseline and never reported. [source: Wilshire et al. 2018]
Reanalysis. Trial data were released under Freedom of Information legislation in 2016 after a tribunal ruling. [source: Wikipedia PACE page for the tribunal] Wilshire et al. (2018) reanalysed them using the protocol's own definitions. Overall improvement fell from the published 61% (GET) and 59% (CBT), against 45% in the comparison group, to 21%, 20% and 10%. Recovery was 4% (GET), 7% (CBT) and 3% (comparison), with no significant difference between arms. [source: Wilshire et al. 2018]
Long-term follow-up. At about 2.5 years there were no differences between the arms. Wilshire et al. conclude that the treatment effects "did not endure beyond two years". [source: Wilshire et al. 2018; Wikipedia PACE page, citing Sharpe et al. 2015]

2.8 How non-response, refusal and worsening were handled

Non-response or slow progress. The manual's tools were to look for practical barriers, such as weather or childcare, and swap in indoor exercise; to adjust the plan when the person's confidence rating was 6 or lower; and to offer "additional support". Beyond this, slow progress was discussed in the model's own terms: boom-and-bust activity, low adherence or worry about symptoms. [source: PACE GET therapist manual v7, pp. 47, 49, 52] [uncertain on how this was handled in routine clinics]
Worsening. Mild post-exercise symptoms were framed as normal, and setbacks were to be met by keeping exercise up as much as possible. A delayed crash could therefore be read as a normal part of reconditioning rather than a reason to stop. The manual pages read for this check (1–63) contain no stopping rule for symptom worsening and no mention of post-exertional malaise or of symptoms delayed by 24–48 hours. The only stop instructions are acute ones, such as slowing or stopping when over the heart-rate ceiling, and adverse effects are attributed to inappropriately planned programmes. The GETSET manual likewise lists stop rules only for acute events such as chest pain, uncontrolled breathlessness, faintness, injury or fever. [source: PACE GET therapist manual v7, pp. 25, 49, 51–52; GETSET therapist manual] [uncertain on manual pages 64–189, not read]
Refusal or dropout. In PACE, participants who withdrew were analysed as allocated. In clinics, declining or not completing GET could lead to discharge and could be recorded as disengagement, which in turn affected benefits, insurance and how later clinicians saw the person. [source: White et al. 2011 on analysis] [synthesis and uncertain on clinic practice]
Adverse events. In PACE these were recorded, and serious adverse reactions were judged by reviewers who were blind to the arm. Routine NHS services had no equivalent systematic harm monitoring. [source: White et al. 2011, Table 4] [synthesis on routine care]

2.9 Named variants

Fulcher & White (1997). Supervised, heart-rate-targeted aerobic exercise (mostly walking or cycling) over about 12 weeks, compared with flexibility and relaxation training. [source: Wikipedia GET page] [uncertain]
Wearden et al. (1998). Home-based graded exercise tested in a factorial trial with an antidepressant. [source: MJA 2005 list of trials] [uncertain]
Powell et al. (2001). Education-based approach: the person was given a physiological explanation of symptoms and encouraged to self-direct graded exercise, with telephone follow-up. [source: MJA 2005 list] [uncertain on detail]
Wallman et al. (2004, Australia). A version guided by Borg exertion ratings, adding short higher-effort periods once sessions reached about 30 minutes. [source: MJA 2005]
FINE trial "pragmatic rehabilitation" (Wearden et al. 2010). A programme of gradually increasing activity designed jointly by patient and therapist. It was given at home in 10 sessions over 18 weeks by one of three nurses with four months' training, to 296 primary-care patients meeting the Oxford criteria in north-west England (2005–2007). It gave a small fatigue benefit at 20 weeks that was not sustained at 70 weeks, and no effect on physical function. [source: Wearden et al. 2010, BMJ abstract, University of Manchester repository]
GETSET (Clark et al. 2017). Guided graded exercise self-help: a GET booklet and activity diary, plus four physiotherapist contacts over about eight to nine weeks. The first contact lasted up to 30 minutes, ideally face to face, and the rest up to 20 minutes by telephone or video call. It used the PACE rules: a baseline held on 5–6 days a week; increases of no more than 20%, applied to every session that week, up to 30 minutes on 5 days out of 7; then intensity. A mild, transient rise in symptoms was framed as normal, and setbacks were met by keeping activity up, again with "hurt does not equal harm". [source: GETSET therapist manual (Clark, Tims & White)]
Graded activity inside CBT for CFS. Used in UK, Dutch and Belgian services, including internet-delivered programmes for adolescents. It applied the same stepwise increase, sometimes to all activity rather than only exercise. [uncertain]
Australia, 2024. The RACGP Handbook of Non-Drug Interventions (HANDI) carried pages on GET and "incremental physical activity" for CFS/ME. Critics have called the latter a rebranding. [source: RACGP HANDI pages and Trial By Error post] [uncertain]
Long COVID, 2020–2021. Some early rehabilitation programmes for long COVID used graded exercise until professional guidance warned against it. [source: World Physiotherapy Briefing Paper 9, 2021]

3. How it caused harm

3.1 Mechanisms

Repeatedly triggering post-exertional malaise (PEM).
- What PEM is. NICE defines PEM as the core feature of ME/CFS: a worsening of symptoms after physical, cognitive or emotional effort that is out of proportion to the effort. It is usually delayed by 12–48 hours and can last days, weeks or longer. [source: NICE NG206 definitions]
- How GET collided with it. A programme that sets the next step by plan rather than by delayed response pushes a person with PEM past their energy limit again and again. Each push can deepen and lengthen the crash, and repeated crashes can shift the person to a lower level of function. NICE distinguishes a flare-up, which resolves within days, from a relapse, which is sustained and can permanently lower energy limits. [source: NICE NG206 definitions and rec 1.11.14] [synthesis on the cumulative mechanism]
- Physiological evidence. Two-day cardiopulmonary exercise testing studies report that many people with ME/CFS cannot reproduce their first-day performance on the second day. [uncertain: e.g. Keller et al. 2014] In long COVID with PEM, muscle biopsies taken after an exercise challenge showed worsening mitochondrial and tissue abnormalities. [uncertain: Appelman et al. 2024, Nature Communications] Both findings conflict directly with the idea that post-exercise symptoms are harmless reconditioning. [synthesis]
Teaching people to override warning signals. If symptoms are presented as an expected and harmless part of reconditioning, people learn to push through the very signals that should make them stop. The heart-rate ceiling gave false reassurance: it limited effort during the session but could not detect the delayed crash. [source: NICE NG206 definitions on PEM timing] [uncertain on manual framing; synthesis on mechanism]
Case-definition mismatch. The Oxford criteria let people without PEM into trials. That could make GET look safe and effective on average, while it was then applied in clinics to people for whom exertion is the trigger. [source: White et al. 2011; Wikipedia GET page] [synthesis]
Attributing failure to the person. The model located the cause of non-recovery in deconditioning, avoidance and beliefs. A person who got worse could therefore be seen, by clinicians and by themselves, as not trying hard enough or as fearful. The results were self-blame, shame and being disbelieved. [source: NICE NG206 rec 1.11.14 on deconditioning/avoidance models; Geraghty et al. 2019, as summarised] [synthesis]
Structural coercion. Where benefits, insurance, employment or school accommodations depended on taking part, people kept exercising through worsening, or lost support when they stopped. [uncertain: patient-organisation reports, not primary sources]
Delayed or missed diagnosis. A broad fatigue label followed by exercise therapy could delay the recognition of other conditions, or of comorbidities such as orthostatic intolerance, which NICE asks clinicians to assess for. [uncertain on NICE rec numbering; synthesis]

3.2 Documented harms

Trial data.

What PACE reported. Serious adverse events affected about 8% of the GET arm against 4% of the SMC-alone arm. Wilshire et al. note this was the only significant between-group difference in harms. Serious adverse reactions, meaning events judged by blinded reviewers to be treatment-related, were rare in all arms and did not differ significantly. Non-serious adverse events were near-universal in all arms. [source: Wilshire et al. 2018; White et al. 2011, Table 4] [uncertain on exact per-arm counts]
Why critics say trials undercount harm. Trials used broad criteria, excluded more severely ill people, and lacked systematic follow-up of people who dropped out. [source: Wikipedia PACE and GET pages] [uncertain on detail]

Patient surveys. These are self-selected, so they cannot give true rates, but they are consistent across countries and years.

Kindlon (2011) pooled eight surveys with 4,338 respondents. On average 51% said GET had made them worse (range 28–82%), against 20% for CBT. [source: ME Association reproduction of Kindlon 2011]
Geraghty et al. (2019) analysed a primary survey (n = 1,428) and secondary surveys (n = 16,665). They found negative responses to GET of 54–74%, with pacing the most favoured approach. [source: PubMed abstract, Geraghty et al. 2019]
Wilshire et al. (2018), summarising patient surveys, give a range of 33–79% reporting worsened health after GET, with a weighted average of 54%, and note the risk of self-selection. [source: Wilshire et al. 2018]
The ME Association has stated that 30–50% of people get worse with GET. [source: Wikipedia GET page, citing the MEA]
Forward ME and Oxford Brookes University (2019; analysis led by Helen Dawes, funded by the ME Association and ME Research UK). Of 2,274 respondents, 542 had been offered GET and 79% of them started it. 81% said their symptoms got worse because of it, and 37% developed new symptoms. The share describing their illness as severe rose from 13% before treatment to 35% after. Fatigue and pain were the symptoms most often worsened. The authors acknowledged selection bias in an online, charity-promoted survey that people with severe illness may have been unable to complete. [source: ME Association reproduction of the Forward ME executive summary, 3 April 2019]

What patients describe.

A crash after the step-ups that did not resolve.
Moving from working or studying to housebound, or from housebound to bedbound.
New or worse symptoms: pain, cognitive impairment, sensory sensitivity and orthostatic intolerance.
Losing jobs or education.
Losing trust in clinicians.

[source: Kindlon 2011; Geraghty et al. 2019, as summarised] [uncertain on individual narratives]

Deaths. No source drawn on here attributes a death directly to GET. Deaths in very severe ME/CFS have been reported in the UK, and inquests have examined failures of care, but they are not established as GET injuries. Do not cite them as such. [uncertain]

Psychological sequelae reported by clinicians and patient groups:

self-blame and a sense of moral failure;
distress resembling medical trauma, centred on health settings and on exercise prompts;
hypervigilance about activity, which is adaptive caution, not phobia;
grief over lost function;
depression;
withdrawal from healthcare.

People with CFS have a raised risk of suicide. In a South London and Maudsley / King's College London cohort of over 2,000 people with CFS (2007–2013), five suicides were recorded against fewer than one expected, about six times the general-population rate; all-cause mortality was not raised. The authors stressed that the absolute risk was small and that the design could not show cause. [source: King's College London report of Roberts et al. 2016, Lancet] [synthesis on how GET-related sequelae may add to this risk]

3.3 Official recognition of harm

NICE NG206 (2021) reversed its 2007 recommendation. Its evidence review found 12 randomised trials of GET, all rated low or very low quality, and rec 1.11.14 lists among things not to offer "any therapy based on physical activity or exercise as a cure", generalised exercise programmes, programmes based on deconditioning and exercise-avoidance theories, and any programme using "fixed incremental increases in physical activity or exercise". Publication was paused in August 2021 after objections from some medical royal colleges, and the guideline was then published in October 2021 without change to this recommendation. [source: NICE NG206 rec 1.11.14 and evidence review G, as reproduced by Action for ME, March 2025] [uncertain on the pause details]
CDC removed GET and CBT from its recommendations in 2017. [source: Wikipedia GET page]
Other health authorities. New York State and Victoria (Australia) health departments describe GET as ineffective and potentially harmful. The Health Council of the Netherlands (2018) advised against GET as a curative treatment. [source: Wikipedia GET page] [uncertain on the exact Dutch wording]
World Physiotherapy (2021), on long COVID: "Graded exercise therapy or fixed activity prescription should not be used." [source: World Physiotherapy Briefing Paper 9]
Cochrane. The amended 2019 Cochrane review of exercise therapy for CFS (Larun et al.) remains published. Cochrane launched an update with an independent advisory group in October 2019, then announced in December 2024 that it was abandoning it, citing COVID-19 delays, loss of UK funding and insufficient new research. It said requests to withdraw the review did not meet its retraction criteria. A petition for withdrawal, started in September 2023, had more than 14,000 signatures by January 2025. [source: Retraction Watch, 23 January 2025]

3.4 Who was most at risk

People with PEM, meaning those who meet stricter definitions such as the Canadian Consensus Criteria, the Institute of Medicine criteria or NICE NG206. [source: NICE NG206 definitions] [synthesis]
People with moderate or severe illness, because each crash costs more when there is less margin. [synthesis]
Highly conscientious people and "pushers", who adhered strictly to the plan and overrode their symptoms. [synthesis]
People whose income, insurance, schooling or custody depended on taking part. [uncertain]
Children and adolescents, who had less control over participation. [uncertain]
People with long COVID and post-exertional symptom exacerbation (PESE) who entered generic rehabilitation in 2020–2021. [source: World Physiotherapy 2021]

4. Recognising it today

This section is mostly [synthesis] from NICE NG206's exclusion criteria and the features described above. Specific rebrandings are tagged where a source was located.

4.1 Core features (any one should prompt a closer look)

A planned upward schedule. Activity or exercise rises by a set amount or percentage every week or fortnight, whatever the person's delayed response. This is NICE's "fixed incremental increases". The PACE and GETSET manuals' rule of increases of up to about 20% is the template to recognise. [source: NICE NG206 rec 1.11.14; PACE GET therapist manual v7, p. 50; GETSET therapist manual]
Baseline as a launch point. The baseline is treated as the place increases start from, rather than as a level to drop below and hold for a long time. [source: NICE NG206 rec 1.11.13 for the contrast]
A deconditioning or avoidance explanation. The person is told their illness is maintained by unfitness, fear of activity, "kinesiophobia", illness beliefs or "symptom focusing". [source: NICE NG206 rec 1.11.14]
Post-exertional worsening framed as benign. Phrases include "a normal response to exercise", "hurt does not equal harm" (the PACE and GETSET manuals' own phrase), "keep exercise up through a setback", "part of getting fitter" or "push through and it will settle". [source for "hurt does not equal harm" and maintaining exercise through setbacks: PACE GET therapist manual v7, pp. 49, 52; GETSET therapist manual] [synthesis for the others]
Time-contingent, quota-based activity. Activity is fixed by the clock or a quota, not by symptoms. This approach is borrowed from chronic-pain operant programmes. [synthesis]
Exercise offered as a cure or route to recovery, or generic gym or fitness programmes designed for healthy people or other conditions. [source: NICE NG206 recs 1.11.9, 1.11.14]
No screening for PEM or PESE before activity advice is given. [source: World Physiotherapy 2021]
Heart-rate ceilings during exercise as the only safety check, with no review of how the person is 12–48 hours later. [synthesis]
Non-progress treated as non-adherence or resistance. The service discharges or penalises people who decline or stop. [synthesis]

4.2 Names and rebrandings to look behind

The label is not the test. Check the content against 4.1. Labels under which graded logic may appear:

"graded activity" and "activity management" with step-ups;
"reconditioning" or "progressive rehabilitation";
"functional restoration";
"incremental physical activity" [source: RACGP HANDI page title];
"graded exposure to activity";
"return-to-work conditioning";
"guided graded exercise self-help" [source: GETSET manual title, located by search];
the graded-activity component of "CBT for CFS" or of persistent-physical-symptom pathways;
generic long-COVID or fatigue rehabilitation courses;
app-based step-count escalation;
some self-help and coaching programmes that promise recovery through stepwise exercise or through "retraining" the brain's response to symptoms.

[synthesis] [uncertain on any individual product]

4.3 Phrases in referrals, reports and self-help that should raise a flag

These are illustrative phrasings to listen for, not quotations from any manual:

"Symptoms are a normal response to increased activity."
"Your body has become deconditioned."
"Avoidance is maintaining your fatigue."
"Increase by a set amount each week."
"Don't let symptoms dictate what you do."
"Patient is non-compliant with graded exercise."
"Failed to engage with rehabilitation."

[synthesis]

4.4 Red flags in a colleague's or service's practice

Activity or exercise targets for a person with ME/CFS or long COVID that are set outside a specialist ME/CFS physiotherapy context. [source: NICE NG206 rec 1.11.9]
A therapist (including a psychotherapist) interpreting crashes after sessions or activity as avoidance, catastrophising or secondary gain. [synthesis]
No written plan for flare-ups and relapse. [source: NICE NG206 rec 1.11.13 requires one]
Outcome reporting that relies only on self-report questionnaires with no tracking of harm. [synthesis]
Insurer, employer or school documents that make support conditional on a graded programme. [synthesis]

4.5 How legitimate approaches differ

NICE personalised physical activity programme. Offered only if the person feels ready to do more than their everyday activities, or asks for it. It must be overseen by a physiotherapist in a specialist ME/CFS team. It:

starts below the person's current activity level and holds there for a sustained period;
adjusts up or down within energy limits, guided by the person's response including delayed response;
includes a flare-up and relapse plan;
is never presented as a cure.

[source: NICE NG206 recs 1.11.10–1.11.14]

Other legitimate approaches.

Pacing and energy management aim to stay inside the energy envelope and to stabilise, with any increase following stability, not a schedule. [source: NICE NG206 definitions]
Heart-rate monitoring used for pacing, as a cue to stop and rest, is legitimate. Heart-rate ceilings used to permit progression are not. [source: World Physiotherapy 2021]
Graded exercise in other conditions (cardiac rehabilitation, COPD, depression, deconditioning after surgery) is not GET-for-ME/CFS. Those conditions do not feature PEM. Screen for PEM or PESE before carrying such programmes across to a fatigue presentation. [synthesis]
Graded exposure for anxiety or phobia targets feared stimuli with learning-based outcomes. It must not be applied to exertion in someone with PEM on the premise that their caution is a phobia. [synthesis]

5. What to do instead and how to help those harmed

5.1 Current recommended alternatives

Energy management (NICE NG206).
- What it is: a self-management approach, supported by a healthcare professional, for staying within the person's energy limit across all activity. "Activity" includes physical, cognitive, emotional and social effort, and these add up.
- What it involves: finding a sustainable level, planning rest, and adjusting for flare-ups.
- Who leads it: refer to a specialist ME/CFS service. The psychotherapist supports the approach but does not design activity plans.
[source: NICE NG206 definitions] [synthesis on role]
Pacing. The patient-community approach of staying within the energy envelope. Activity diaries are used to find personal triggers, and heart-rate or step monitoring is used as a cue to stop. [source: Geraghty et al. 2019, where pacing was the most favoured approach; World Physiotherapy 2021 on pacing tools]
NICE personalised physical activity programme. Only under the conditions in 4.5. [source: NICE NG206 recs 1.11.10–1.11.13]
Long COVID with PESE.
- Screen for PESE before giving any activity advice. A validated option is the DePaul Symptom Questionnaire PEM items. This is a licensed instrument, and its items are not reproduced here.
- Use "Stop. Rest. Pace." and energy-envelope planning.
- Rule out cardiac, oxygen-saturation and autonomic problems before any exercise.
[source: World Physiotherapy Briefing Paper 9]
Symptom management. Address sleep, pain, orthostatic intolerance and cognitive difficulties through the medical team. [uncertain on NICE rec numbers]
Psychological support. NICE NG206 allows CBT only to support people in managing symptoms and the distress of living with a chronic illness. It is not a treatment or cure for ME/CFS. Do not use CBT built on the PACE deconditioning or fear-avoidance model. [uncertain on exact rec numbers and wording] [source for model exclusion: NICE NG206 rec 1.11.14] [synthesis]

5.2 Assessing a client who went through GET

[synthesis throughout, grounded in NICE NG206 definitions]

Illness course. Map function before GET, the trajectory during it, and function now: the trigger point, any change in severity level (mild, moderate, severe, very severe), and new symptoms. Note the PEM pattern, including what triggers it, its delay and its duration.
Meaning made of it. Ask about self-blame ("I failed"), the beliefs absorbed from the programme ("symptoms are just unfitness"), and anger or betrayal towards clinicians.
Trauma responses. Look for responses tied to health settings, physiotherapy or exercise cues. Distinguish them from adaptive, experience-based caution about exertion. Do not label that caution as phobia, catastrophising or avoidance.
Losses. Work, study, identity, relationships, finances and benefits.
Risk. Use standard structured risk assessment. ME/CFS carries an elevated suicide risk (see 3.2), and isolation, disbelief and severity raise it. Build a collaborative safety plan, with lethal-means counselling (safe storage or removal) where indicated. Involve the GP or ME/CFS team.
Severity and energy cost of the assessment itself. Spread it across sessions where needed.

5.3 Treatment considerations

[synthesis throughout]

Name the iatrogenic harm plainly. For example (illustrative): "Current UK guidance no longer recommends the programme you were given, and getting worse after it is one of the most commonly reported experiences. What happened to you is not evidence that you didn't try."
Address self-blame and shame. Use compassion-focused or narrative work on the "failure" story, and grief work for lost function and identity.
Process trauma. For trauma-type responses to medical care, use trauma-focused methods that pace emotional load. Shorten or split processing, and watch for PEM after sessions. Exposure-based work should target distress about healthcare encounters, never exertion tolerance. Get explicit consent and stop if PEM appears.
Rebuild agency around pacing. Support the client in setting their own limits and in saying no to unsafe programmes. Help them prepare for medical appointments, for example with a one-page summary of their PEM pattern.
Adapt delivery.
- Ask about symptoms one to two days after sessions.
- Offer shorter, remote or audio-only sessions, rest breaks and a low-stimulus setting.
- Allow cancellation without penalty.
- For severe or very severe illness, consider very brief contacts, message-based therapy or a carer present, and coordinate with the care team.
- Never interpret a post-session crash as resistance.
Clinician beliefs. If you were trained in the deconditioning or fear-avoidance model, review NICE NG206 recs 1.11.9–1.11.14 and take the issue to supervision. Treat PEM reports as clinical data about the illness.

5.4 Ethical, reporting and legal duties

[synthesis; check your own jurisdiction and professional code]

Duty to avoid harm and to work within competence. Do not set exercise or activity targets for ME/CFS or long COVID. Do not endorse programmes that meet the 4.1 criteria.
A client currently in a programme that matches NICE's exclusions. With consent, support them to raise it with the provider, to ask for a specialist ME/CFS physiotherapy review, and to document worsening.
- In England, routes include the provider's complaints process and the Patient Advice and Liaison Service (PALS). NHS staff can also record a patient-safety event through the NHS incident-reporting system.
- The provider holds the statutory duty of candour.
[uncertain on system names]
Concerns about a registered professional's practice, such as continuing to deliver fixed-increment GET to someone with ME/CFS against the person's reports of worsening. Raise them first through supervision and the service. Escalate to the regulator if needed: in the UK, the Health and Care Professions Council for physiotherapists, or your own professional body.
Insurance, benefits and employment. With consent, write factual letters. Describe the client's reported PEM pattern and functional limits, and refer to NICE NG206 rec 1.11.14 and, for long COVID, World Physiotherapy 2021. Stay within your competence: describe what you observe and what the client reports, not a diagnosis.
Children and young people. Apply standard safeguarding duties. Note that declining a graded programme that NICE does not recommend is not, in itself, a safeguarding concern. [synthesis] [uncertain on local guidance]
Records. Document reported harms accurately, without pejorative terms such as "non-compliant" or "poorly motivated", which can follow the client into later care.

Sources

Sources:

Graded exercise therapy – Wikipedia
PACE trial – Wikipedia
NICE NG206 Recommendations
ME Association: NICE Guideline – Incorporating physical activity and exercise (recs 1.11.9–1.11.14)
White et al. 2007, PACE trial protocol, BMC Neurology
White et al. 2011, PACE main results, Lancet (PDF copy)
Exercise prescription for individuals with chronic fatigue syndrome, MJA 2005
Geraghty et al. 2019, J Health Psychol abstract (PubMed)
ME Association: Kindlon 2011, Reporting of harms associated with GET and CBT
Action for ME: Risks of graded exercise therapy (2025)
World Physiotherapy Briefing Paper 9: Safe rehabilitation approaches for people living with Long COVID (2021)
CDC: Treatment of ME/CFS

Sources:

Further reading (not drawn on; used only for titles and existence):

Further reading (not drawn on unless noted):

Bavinton J, Darbishire L, White PD (2004). Graded Exercise Therapy for CFS/ME: therapist manual (final trial version 7). PACE trial management group. (Online copy, pages 1–63.)
Clark LV, Tims E, White PD. Guided Graded Exercise Self-help for CFS/ME: therapist manual (GETSET trial).
Fulcher KY, White PD (1997). Randomised controlled trial of graded exercise in patients with the chronic fatigue syndrome. BMJ 314:1647.
Wearden AJ et al. (1998). Randomised, double-blind, placebo-controlled treatment trial of fluoxetine and graded exercise for chronic fatigue syndrome. Br J Psychiatry.
Powell P et al. (2001). Randomised controlled trial of patient education to encourage graded exercise in chronic fatigue syndrome. BMJ.
Wearden AJ et al. (2010). Nurse led, home based self help treatment for patients in primary care with chronic fatigue syndrome: randomised controlled trial (FINE). BMJ 340:c1777.
Clark LV et al. (2017). Guided graded exercise self-help for CFS (GETSET). Lancet.
White PD et al. (2013). Recovery from chronic fatigue syndrome after treatments given in the PACE trial. Psychol Med. (Figures via Wilshire et al. 2018.)
Sharpe M et al. (2015). Long-term follow-up of PACE. Lancet Psychiatry.
Wilshire CE, Kindlon T, Courtney R, Matthees A, Tuller D, Geraghty K, Levin B (2018). Rethinking the treatment of chronic fatigue syndrome: a reanalysis and evaluation of findings from a recent major trial of graded exercise and CBT. BMC Psychology 6:6.
Larun L et al. (2019). Exercise therapy for chronic fatigue syndrome. Cochrane Database Syst Rev CD003200.
Kindlon T (2011). Reporting of harms associated with graded exercise therapy and CBT in ME/CFS. Bull IACFS/ME.
Geraghty K et al. (2019). Myalgic encephalomyelitis/chronic fatigue syndrome patients' reports of symptom changes following CBT, GET and pacing treatments. J Health Psychol.
Keller BA et al. (2014). Inability of ME/CFS patients to reproduce VO2peak on two-day CPET. J Transl Med.
Appelman B et al. (2024). Muscle abnormalities worsen after post-exertional malaise in long COVID. Nat Commun.
Roberts E et al. (2016). Mortality of people with chronic fatigue syndrome: a retrospective cohort study in England and Wales. Lancet. (Reported via King's College London news.)
Bateman L et al. (2021). ME/CFS: essentials of diagnosis and management. Mayo Clin Proc.
NICE (2007). CG53. NICE (2021). NG206.
Health Council of the Netherlands (2018). ME/CFS advisory report.
Vink M, Vink-Niese A (2020). Graded exercise therapy for ME/CFS is not effective and unsafe. Work.

← Previous · II-052Covert sensitisation and covert conditioning Next · II-054 →Social skills training