ME/CFS: How Did We Get Into This Mess and Where Do We Go From Here?
- How did we get here?
It may seem that the recent furore over the new ME/CFS NICE guidelines is a recent development, but it has been a long time in the making. Back in 2002 I sent a letter to Simon Wessely outlining my concerns about research concentrating on neurosis and illness beliefs:
I feel that the simplistic explanations for CFS involving either neurosis or cognitive behavioural responses do not adequately explain the condition. Research clearly shows that psychological factors are important in perpetuating CFS, but the standard psychiatric explanations for the condition are clearly lacking. I feel that it is important to identify clearly the psychological factors which cause CFS, as labelling CFS sufferers as neurotic simply alienates them from seeing any psychological reason for their illness, and thereby results in continued illness. Such attitudes may even be self-reinforcing, due to the fact that a small minority of CFS sufferers get angry at this simplistic diagnosis, resulting in the continued perception by the medical community that CFS is caused by neurosis. This then further polarises the medical community into the mutually exclusive psychosomatic and organic dogmas, which does nobody any good.
Fast forward to 2011, when the PACE trial is published, based on the theories of deconditioning and fear avoidance perpetuating CFS. While deconditioning can certainly affect some patients, particularly those who are bedbound, even back in 2001 there was evidence showing that deconditioning is not a perpetuating factor. And while there is evidence showing that fear of movement is a factor for some patients, research shows that patients deteriorate if they attempt to do too much. The definition of the illness requires that post-exertional malaise or deterioration after normal levels of activity, so clearly some fear of overdoing things is rational.
The PACE trial did include other elements, such as addressing sleep issues, which is an important factor in the illness. However, the focus on deconditioning and fear avoidance seems ill advised, as it implicitly encourages healthcare providers to encourage patients to go beyond their limits and ignore symptoms.
Even prior to the PACE trial, patient surveys showed that CBT and GET tended to make patients worse rather than better, with GET causing 56.5% of patients to get worse according to an ME association survey from 2010. (This is in contrast to the RCTs, almost all of which showed benefits for both CBT and GET). After the PACE trial, the situation seems to have gotten even worse, with 74% of patients reporting worsening of symptoms after GET according to a 2015 survey by the ME association.
The Action for ME patient charity also bears responsibility for this predicament. Originally they were involved in the design of the PACE trial and collaborated with the researchers, including developing the protocol for the APT (adaptive pacing therapy) used in the trial. They originally supported the trial as they believed it would show that pacing (APT) would be superior to CBT and GET, and that the trial would flag up adverse effects in CBT and GET. However, they did not challenge the deconditioning and fear avoidance models being used for the CBT and GET arms of the trial, perhaps assuming that the trial would fail.
When the trial did show that CBT and GET were superior to APT, and after the intense criticism of the trial, Action for ME released a statement essentially disavowing themselves from the PACE trial.
2. Pushback
The result was perhaps inevitable: a huge outcry from patients and their advocates, a lot of bad feeling, and harassment of researchers. A lot of effort went into trying to prove that the PACE trial was fraudulent, or at least shoddy. A number of changes were made to the trial between the initial publication of the PACE trial protocol, and the trial itself. This isn’t all that unusual for clinical trials, but it is a problem when the primary outcome is changed during the trial, as that can be a sneaky way of getting a better outcome from a trial that looks like it is going to produce a negative result.
One of the changes made to the PACE trial was in the measurement of one of the primary outcomes. Instead of using a binary measurement (0,1) they used a Likert score (0,1,2,3). This was done to improve the sensitivity of the scale, so that instead of patients simply saying “better” or “worse”, they could say “slightly better”, “much better”, etc. This simply allowed more subtle changes to be picked up, so isn’t really a problem.
Changes were made to the criteria for entering the trial and being classed as recovered (the recovery analysis was published in a separate paper). Originally a physical function score below 60 was required to enter the trial, but this was changed to 65 to increase recruitment. At the same time, the “normal range” for physical function in the recovery criteria was reduced from 85 in the published protocol, to just 60 in the published trial. The reasoning for this was because half the working population would fall outside the normal range, although that may have been based on faulty reasoning. The end result is that patients could get worse in physical function from the start to the end of the trial and still be considered within the “normal range” for this measure. It should, however, be noted that this was just one part of the recovery criteria: patients also had to be within the normal range for fatigue, not meet the Oxford case definition of CFS any more, and have a self-rated Clinical Global Impression score of “better” or “much better”, so no patients could have been classed as recovered at the beginning of the trial. The authors of the PACE trial have confirmed that although 3 out of 640 patients (<1%) were within the normal range for both fatigue and physical function at trial entry, no participants met the full criteria for recovery at trial entry.
It should also be noted that there is nothing inherently wrong with having a lower threshold on a single measure for recovery than for trial entry. The threshold for trial entry is an upper limit, while for recovery is a lower limit. If both thresholds were the same then all patients would have needed to be below that threshold on trial entry, and above it at the end of the trial. As physical function varies with age, this doesn’t really make sense. The thresholds are there just to prevent anyone too healthy entering the trial, and too sick from being classed as recovered. The recovery criteria itself is arguably somewhat weak, and perhaps more of a measure of improvement rather than recovery per se.
A re-analysis of the PACE trial using the original criteria found, unsurprisingly, rates of improvement and recovery much lower than the published trial. The results were still statistically significant for the primary outcome, although this significance disappeared after making Bonferroni corrections for the number of comparisons done. It should be noted that Bonferroni is likely too conservative, although it is the correction that was listed in the original PACE trial protocol.
The upshot of all of this is that the PACE trial did show a significant reduction in fatigue and improvement in physical function due to CBT and GET using the more sensitive Likert scale.
The long-term follow-up from the PACE trial found that although improvement from CBT and GET were maintained at 134 weeks, there was little difference between the groups at 134 weeks, and all groups gradually improved over the duration of the trial and the follow-up period. A similar result was found in the FITNET trial of CBT for adolescents with CFS. The authors explain this as the illness gradually improving naturally, with CBT and GET speeding up the improvement. Critics say it means the treatment doesn’t work and that benefits are not maintained at follow-up; however, if that was the case then there should have been a deterioration in the CBT and GET groups after treatment, which did not happen.
3. Subjective outcomes
Another objection to the PACE trial (and CBT and GET in general) is that they rely on unblinded, subjective measures, and are therefore prone to various types of bias. The PACE trial did attempt to control for bias by asking patients for their impressions of the treatment beforehand. Patients actually reported being more confident in APT rather than CBT or GET, so bias does not seem to have been a problem (at least at the start of the trial). Also, patients were given the same number of sessions in all 3 treatment groups, and the newsletters given to patients during the trial were careful to mention all 3 treatment groups equally.
It’s certainly possible that patients felt pressured into reporting greater improvement than they actually experienced. However, if that was the case it seems unlikely it would result in a persistent gradual improvement over 134 weeks for all for treatment groups, long after the treatment itself has ended.
So, the bottom line is that PACE did seem to show improvement with CBT and GET, even though patient surveys show the exact opposite. The same is true for the many other RCTs into CBT and GET. How is that possible?
The two mutually exclusive explanations are:
[1] Patient surveys are unreliable, and we need to give the RCTs a greater weight.
[2] The PACE trial (and all CBT/GET trials) are shoddy and biased.
Neither of these makes sense. While patient surveys can certainly have methodological issues, we can’t simply reject the experiences of patients when they say they are getting worse, or the results of studies showing that excessive exercise can make patients worse.
The trials into CBT and GET don’t all seem to have serious enough methodological issues to explain the difference between the RCT results and patient experiences. So what the heck is going on here?
Western medicine likes to narrow things down to “do this, you’ll feel better”. CBT and GET are very traditional types of treatments that are used for various health conditions. Some ME/CFS patients are deconditioned, although that clearly isn’t a cause of the illness, and it doesn’t affect all patients. Still, deconditioning may cause symptoms for these patients, and GET may also benefit patients in other ways by, for example, reversing the effects of inactivity on the HPA axis and autonomic nervous system, or by increasing adenosine (which will improve sleep).
Some ME/CFS patients are afraid of exercise and activity. While some caution is warranted, this caution can sometimes be excessive, and fear avoidance CBT may help these patients. Perhaps education and gentle physiotherapy rehabilitation would be more appropriate for these patients, however.
Certainly, it is not ideal to rely on subjective outcomes, as has been done for the CBT and GET trials. Ideally a 2-day CPET test would be done at the start and end of the trial to compare objective improvement after treatment. The 2-day CPET test is currently the only biomarker for ME/CFS that shows promise in multiple trials. While the 2-day CPET can result in deterioration of ME/CFS patients for up to 2 weeks, this could be mitigated by having a washout period, to allow patients to recover from the CPET tests. Obviously it would be limited to mild and moderate patients, but currently CBT and GET trials do not have severe patients anyway.
4. Updated 2021 ME/CFS NICE guidelines
The UK National Institute for Health and Care Excellence updated their guidelines for ME/CFS in October 2021 after a 4-year consultation process. Previously the guidelines had recommend CBT and GET, and there was significant pressure to remove CBT and GET.
The new guidelines did indeed remove CBT and GET as primary treatments. CBT is now only recommended as support for patients managing their symptoms. This change was made to the guidelines due to the experiences of patients, who found that CBT is useful when used to support them and help reduce their distress, but can cause harm when used inappropriately, e.g. by encouraging them to ignore symptoms, or by assuming they have “abnormal” illness beliefs.
Graded exercise with fixed increments is now prohibited by the new guidelines, and patients must be warned about the potential benefits and risks of incorporating exercise into the management of their illness. This change was made due to harms reported by patients after undergoing GET programmes.
The criteria for diagnosing ME/CFS also changed: the new NICE guidelines now require cognitive dysfunction (“brain fog”), disturbed sleep and PEM in addition to fatigue.
While the new NICE guidelines are helpful in some ways, they are problematic in others. It is good that the treatments of graded exercise and CBT based on the problematic theories of fear avoidance and deconditioning have been removed. However, this essentially leaves no treatment at all, simply recommending that health practitioners develop a personalised programme for each patient to support and manage their illness, but emphasizing that it is not a cure. While it is certainly unhelpful to inflate the chances of recovery, it is also unhelpful to tell patients that there is no treatment and the management tools are not a cure.
The new guidelines specifically say that there is no cure for ME/CFS, that CBT is not curative, and that management is all that can be done. Back when the 2007 NICE guidelines had just been released one patient reported that:
During the relapse, I was referred by my doctor to the newly established CFS specialist team and was devastated to be told by specialists “you cannot recover, especially as you have had a relapse”. Every professional insisted I had to accept the condition and learn to manage it. They told me if I tried to fight it I would become depressed. Yet I became depressed because I was told I wouldn’t recover.
and:
Patients referred to the team were only allowed to accept group support on the NHS if we accepted “you will never recover.” I had to have a number of 1:1 sessions, before I finally capitulated and said the words “I accept I will never recover” and the relief on the practitioner’s face was visible
Giving unrealistic hopes for recovery is problematic, but telling patients they can’t recover is just as bad. This patient had previously recovered by herself, and was dismayed that the specialists told her she couldn’t recover again. Luckily she got good support from her GP (who was just as shocked), and eventually she did fully recover on her own, and she has been symptom free for the past 14 years.
Another issue with the new NICE guidelines is the requirement for brain fog and sleep dysfunction, which may arbitrarily rule out some ME/CFS patients from getting properly diagnosed. Some patients report that they did not experience brain fog during the initial stages of the illness, or when they were mild, but only started to suffer it later. Sleep studies show no objective differences between ME/CFS patients and controls. The issue of requiring cognitive dysfunction for diagnosis was pointed out in comments by s4me, ME Action UK and Hope 4 ME and Fibro Northern Ireland, all of whom pointed out that some patients do not experience this symptom, or only experience it later in the illness. The response from NICE was that 7 out of 9 other guidelines mention cognitive dysfunction. However, those other guidelines don’t necessarily require it, so this doesn’t seem entirely logical. Another reason NICE gives is that studies show around 90% of patients report cognitive dysfunction. However, testing shows that the magnitude of cognitive dysfunction in ME/CFS patients is quite small: 0.5–1.0 S.D. below that of healthy persons, and it tends to only be certain types of cognitive functions that are affected: initial learning, information processing speed and tasks requiring working memory over a sustained period of time.
5. Is recovery or improvement possible?
CBT is used in other health conditions, such as MS and cancer survivors. When used to treat fatigued cancer survivors, CBT is highly effective, resulting in 71% of patients being non-fatigued and similar to healthy controls at long-term follow-up. Patients who did not accept CBT were significant more fatigued at long-term follow-up.
Note that this is not cancer fatigue, but fatigue in cancer survivors who do not have any current cancer. The CBT involves addressing multiple psychosocial factors such as insufficient coping mechanisms, fear of recurrence, sleep dysfunction, dysfunctional cognitions concerning fatigue, activity dysregulation, and low social support. The CBT is specifically designed for cancer survivors, and while it does address activity dysfunction and dysfunctional cognitions, these are not central to the treatment, and there is no programme of graded increasing activity.
In terms of recovery from ME/CFS, there is reason to be hopeful. An RCT comparing multi-disciplinary rehabilitation (MRT) to CBT found that while both treatment were effective in reducing fatigue and other symptoms, MRT resulted in a greater reduction in fatigue (from 51.47 before treatment, to 33.4 after 26 weeks), and this was maintained at 52 week follow-up.
After treatment, 32.1% of patients in the MRT group and 35.2% in the CBT group felt they did not have CFS any more.
The large placebo-controlled trial into rituximab for ME/CFS also gave interesting results about recovery. Even though it found that there was no significant difference between the rituximab treatment and placebo, it found that the “response rate” was 35.1% in the placebo group. A separate, smaller open-label (not placebo controlled) trial by the same authors looked more closely at improvement and long-term remission. This study found that 55% of patients in the trial were classed as “responders” (a fatigue score greater than 4.5, where 6.0 is completely healthy). 68% of those responders (or 37.5% of all patients in the trial) were still in “remission” at 4 year follow-up. The SF-36 physical function of the patients in the trial increased from a mean of 33.0 at baseline to 51.5 at 18 months.
Given that rituximab was found to have no effect over and above placebo, clearly these results show that patients improved due to some combination of natural course and the placebo effect.
6. Where do we go from here?
Clearly there have been mis-steps from the medical establishment in how ME/CFS patients are treated or managed, particularly from practitioners of psychosocial and behavioural treatments. The unfortunate consequence of this has been for ME/CFS patients and patient organizations to outright reject any kind of biopsychosocial treatment that is not purely supportive in nature, and to downplay the ability of patients to improve or recover. Patients are naturally upset at being given inappropriate treatments, and the medical establishment has responded to that by limiting treatment to symptom management and support, downplaying the chances of recovery.
The PACE trial was a well-designed RCT that gave positive results, but due to a flawed premise and incorrect etiology it resulted in many patients deteriorating, and the subsequent backlash has widened the unscientific divide between the mind and body. The lack of objective outcomes is also a problem.
Perhaps it is time for everyone to step back, take a fresh look at the science and patient experiences (both those who recover and those who deteriorate), and come up with a more sensible etiology of the illness which takes into account all of the evidence.
A number of studies over the years have linked life stress and childhood trauma to the development ME/CFS. Typically patients report viral infection or life stress at onset, or a combination of both. A 2007 review noted that 5 cohort studies have confirmed a post-infectious condition after mononucleosis (EBV) infection, as well as after other infections such as Q-fever and Ross River virus, with a 10–12% risk of CFS 6 months after infectious onset.
We know that there is interaction between stress and the immune system. This has been known since the discovery of the “stress” hormones such as cortisol and adrenaline, both of which are significant modulators of the immune system. However, the links go much further than this. We now know that stress and infections act synergistically. Studies on animals have shown that injection of interleukin-1 (IL-1) results in activation of the HPA axis (the body’s stress system), resulting in increased cortisol and modulation of the immune response. This may have evolved to protect against a lethal overactivation of inflammatory responses by the immune system, as well as to provide energy to maintain survival during the infection (cortisol acts to increase glucose in the blood).
Interestingly, animal experiments have shown that injection of IL-1 can cross-sensitize the HPA axis to other types of stressors at a later date, such as foot shock and social defeat. Injection of IL-1 to rats can increase HPA axis response to a different type of stressor one to two weeks later.
A 2021 study has found that brain cells can retain a memory of experimentally-induced inflammation in the abdomen, and stimulating those brain cells can induce similar inflammation later.
The brain does not have a neat divide between psychological and physical, and it doesn’t make sense for us to maintain that false divide. Perhaps it is better to think of ME/CFS as either classical conditioning response to stress and infection, or as a maladaptive sickness response which gets reactivated after emotional, physical or mental stressor as a protection mechanism gone awry.
Medical science likes to have a hammer and nail approach: “try exercise, see what happens”, “try removing fear of activity, see what happens”, but these simplistic approaches don’t work for ME/CFS. What is needed is a treatment based on a plausible etiology, with little to no risk of adverse effects, tailored to the patient.
7. Future trials
Ideally we would have a trial of “appropriate” rehabilitation, with an objective 2-day CPET before and after, and a waiting list control to compare treatment to natural course. Patients can take up to 2 weeks to recover from a 2-day CPET, so a 3-week or longer recovery period would need to be factored in, and severe patients would need to be excluded. Rehabilitation would consist of elements such as: coaching, psychotherapy (e.g. CBT or ACT), sleep advice, physiotherapy and activity management.
Psychotherapy would be aimed at reducing stress, increasing uplifts, and addressing any trauma or psychological barriers to recovery, rather than trying to correct “abnormal” illness beliefs (which can be done with education).
Exercise would only be offered if the patient has low activity levels and can tolerate small and gradual increases in activity. Some patients do have an excessive fear of PEM, and they reduce their activity to an abnormally low level, which inhibits recovery. Many recovered patients say that addressing this fear was critical to recovering. Education and coaching from someone knowledgable about the illness who has an understanding of PEM would seem to be a better option than CBT at addressing this issue.
Severe patients could be included in a trial using smart-watch activity monitoring as an objective end-point, with sessions by phone or zoom, or with a therapist visiting at home. Previous trials using step counts haven’t shown any significant increase in steps, mainly because mild patients don’t tend to have any significant reduction in daily steps. However, a clinical trial using daily step counts with bedbound or housebound patients would give an accurate and objective measurement of improvement, and would avoid having to subject patients to the 2-day CPET.
Why hasn’t a trial such as this been done? There seems to be some resistance from patient organisations against any trial involving exercise or CBT. While this is understandable, it does not help us in our understanding of what is the best treatment or rehabilitation for patients. Ideally patients should lobby researchers, charities and governments to fund a good rehabilitation trial to settle the issue.
