How much can we trust a medical diagnosis of ME/CFS? (Analysis of Newton, 2010; Devasahayam, 2012; Collin, 2012; plus more)

[Simon M]

Intro and summary

If all we know is that someone has a medical ME/CFS diagnosis, what is the chance they do have this, or at least do not have an undetected alternative diagnosis that would explain the illness?

This is particularly valuable information for biobanks, the main source of large samples of biological data, such as UK Biobank and All of Us. I looked into this question for the recent ME/CFS biobank GWAS replication paper to inform the discussion of the possible influence of different cohorts on the findings.

I was surprised that the evidence is as good as it appears to be, with four converging sources of evidence, including the three papers cited in that study. This thread is a revised version of that analysis.

The evidence indicates that if someone has a GP diagnosis only, they have roughly a 50-50 chance of an ME/CFS diagnosis, as opposed to an alternative explanation for the illness. For those who also have a diagnosis confirmed at an NHS clinic, the chance goes up to about two thirds:



I'm posting material in three phases:

Phase 1 is the bedrock of the analysis: 2 detailed published studies showing diagnoses given to referrals at both the Newcastle and Barts CFS clinics. Both look good – including the Barts Peter White one. The studies cover a thousand referrals and the data converges on an estimate of about 50% not having ME/CFS.

Phase 2 looks at evidence of diagnostic rates in standard NHS ME/CFS clinics, which indicates a 77% diagnostic rate (~9,000 referrals, 4,100 with full data).

Phase 3 estimates the chances of having ME/CFS vs another medical cause for those with a medical diagnosis, and a more detailed look at the differences between UK Biobank, which recruited broadly, and DecodeME, which recruited mainly from the ME/CFS community. Those in the community are better informed and so it is more likely their diagnosis will be accurate.

I developed the analysis and conclusions in these posts using Claude as an analytical and editorial aid, including for checking the reasoning, copy-editing and presentation.
.

 

[Simon M]

Phase 1: Two large and detailed audits of patients at well resourced CFS clinics

Diagnostic difficulties

ME/CFS is defined by disabling fatigue lasting at least six months, post-exertional malaise, and, for an IOM diagnosis, cognitive problems and/or orthostatic intolerance, with more symptoms required for CCC. There is no diagnostic test.

The fatigue must be unexplained, which means excluding the many other conditions that cause it — and that exclusion is difficult because chronic fatigue has so many medical and psychiatric causes. Since the hallmark symptom of PEM is rarely assessed - at least not well - what matters most is the quality of the work done to rule out the alternatives before a diagnosis is given. Fukuda and Canadian criteria require a thorough medical and psychiatric assessment as part of diagnosis.

The following two studies provide well-documented and detailed assessments and diagnosis.

Newton, 2010. 60% of patients seen at Newcastle Clinic had CFS, maybe fewer​

Source: Newton et al., 2010. The Newcastle NHS Chronic Fatigue Syndrome Service: not all fatigue is the same. PMID 21132135

This is the first of two studies that looked directly at how many people referred to a specialist CFS clinic turn out to have something else. The Newcastle fatigue service was run by Julia Newton, a senior clinician and researcher specialising in the illness, with a well-resourced multidisciplinary team — a far better-equipped clinic than most.

Newcastle confirmed CFS in 60% of everyone it saw, giving the remaining 40% alternative diagnoses. The main ones were other chronic diseases, sleep disorders and psychiatric illness.

An earlier internal audit at the same clinic, using a random sample of case-notes, found a much lower rate: only 36% confirmed as CFS.

These are audits of routine clinical practice rather than one-off studies, so they say something about what a thorough assessment finds in ordinary referrals. An important point for the comparison with Barts (post 3): Newcastle had no psychiatrist on the team, and the paper describes no psychiatric screen or mental state examination. If that meant psychiatric cases were missed, the true CFS rate is lower than 60%.

Also, nobody was excluded for lacking post-exertional malaise or other symptoms, so 60% is the rate before any PEM screen.

Spoiler: Study detail

Patients and method. 260 patients were assessed between November 2008 and December 2009; non-attenders were excluded before counting. The team could refer on to other specialties. They cited the Fukuda criteria but gave no details of how they were applied, and the only exclusions were for alternative diagnoses.

Results. 157 (60%) were confirmed as CFS. As a percentage of all patients assessed, the main alternative diagnoses were: other chronic disease 19%, sleep disorders 8%, psychiatric 6%, idiopathic chronic fatigue 5%.

The earlier audit. For 2005–7 a random sample of 56 case-notes was drawn from 375 patients: 20 of the 56 (36%) were CFS cases, far below the 60% (157/260) recorded in 2008–09 (p<0.0001). The paper gives no confidence interval; AI’s exact binomial calculation gives a 95% CI of 23%–50%.

What was not assessed. There was no psychiatrist on the team, and the paper describes neither a psychiatric screen nor a mental state examination — both mental and physical health assessment are recommended in Fukuda and the other case criteria. No minimum fatigue or function threshold is reported, though this was probably applied subjectively, and strict thresholds are not needed in clinical practice.

 

[Simon M]

Devasahayam, 2012. 54% of patients assessed at Barts had CFS — but only 42% of referrals​

Source: Devasahayam et al, 2012. PMID 2229907. Alternative diagnoses to chronic fatigue syndrome in referrals to a specialist service.

This study comes from the Barts specialist CFS clinic, led by Peter White — like Newton in Newcastle, a senior clinician and researcher with a well-resourced multidisciplinary team.

Barts worked in two steps. Referral letters were screened first, and 37% of referrals were rejected at that stage, mostly because consultants decided there was evidence of another illness that would explain the fatigue or because the patient lived nearer another service. The remaining patients were then assessed at the clinic, where 54% were confirmed as CFS and almost all the rest were given an alternative diagnosis.

That 54% is not comparable with Newcastle's 60%, because Newcastle assessed everyone referred and Barts had already screened out a third. Across the whole Barts referral funnel, the CFS rate is 42%. Some letter rejections will have been wrong, and allowing for that the figure is likely in the range 42% to 48%, so mid-forties.

Together, Newcastle and Barts cover 1,053 patients seen or referred. They were thorough studies, and they broadly replicate the same finding — and good studies that replicate are rare in ME/CFS research.

Spoiler: Study detail and the 2-step funnel derivation

Design. The referral letters and the clinic assessments were for near-identical 18-month windows at the same clinic, spanning March 2007 to November 2008.

Step 1: letter screen. A consultant in infectious diseases and a consultant liaison psychiatrist — presumably Peter White, who ran the clinic and was senior author — screened referral letters, rejecting 154 of 418 (37%). 127 of the 154 rejection letters could be reviewed; the paper said the others were likely have been sent back asking for more information. Of the 127 letters reviewed, 77 (61%) were judged likely to have an alternative diagnosis; the rest were rejected mostly on administrative grounds, the commonest being that a nearer CFS clinic was available.

The authors acknowledge that some of the 77 may have had CFS, but that over 40% of accepted referrals did not get a CFS diagnosis indicates a conservative approach to rejection.

Step 2: clinic assessment. 250 patients were assessed: 54% (137) confirmed as CFS, 43% given a confirmed alterna tive diagnosis, and 2.4% (6 patients) outside both categories. So, as at Newcastle, almost all exclusions were due to other diagnoses. Every assessed patient had a mental state examination.

Overall rates. Combining the two stages gives a CFS rate of 42% (137 CFS diagnoses among the 327 patients whose status is known) and an alternative-diagnosis rate of 58%. The referrals sent to a nearer service and the 27 whose letters were unavailable are excluded from both numerator and denominator, since they were never assessed.

If a tenth of the letter rejections were wrong the figure is 44%, and a quarter gives 48% (an error rate of 50% would mean the screen was no better than chance). Two consultants covering both medical and psychiatric ground, applying a threshold loose enough that 46% of those they accepted proved not to have CFS, are unlikely to have been at the high end of that. The figure is probably in the low-to-mid forties.

(The paper itself reports a combined alternative-diagnosis rate of 49%. That includes in the denominator referrals rejected for largely administrative reasons, and assumes all letter rejections were correct.)

What was not assessed. As at Newcastle, no minimum fatigue or function threshold is reported, though this was probably applied subjectively, and nobody was excluded for not having post-exertional malaise or other symptoms.

 

[Simon M]

Newcastle vs Barts: similar findings, and about half of referrals have CFS​

Sources: Newton et al (2010) and Devasahayam et al (2012), see previous 2 posts.

Of those assessed, the two teams reached similar results: Newcastle found 60% had CFS, Barts 54%, with almost all non-cases due to alternative diagnoses. They also diagnosed similar types of illness — chronic disease, sleep disorders, cardiovascular problems. Comparing like with like across the whole referral funnel, medical diagnoses account for 26.3% of Barts' referrals against 28.5% of Newcastle's.

One category diverges sharply: Barts diagnosed psychiatric illness in 22% of those it assessed, Newcastle in 6%. That single category accounts for most of the difference in overall diagnosis rates.



Referral bias is unlikely to be a factor, as GPs have little choice of where to refer; 'nearer CFS clinic' was the top reason for Barts' letter rejections. The main factor is likely to be the quality of psychiatric assessment. Barts screened every referral with a consultant liaison psychiatrist, and every assessed patient had a mental state examination, though the paper does not say who carried it out. Newcastle had no psychiatrist on the team and describes neither. Both medical and psychiatric assessment are required by Fukuda and the other case criteria.

If that reading is right, Newcastle missed cases that Barts would have identified, so its 60% is higher than its true rate, which was probably closer to that of Barts. That would make Barts the better guide.

I was surprised that Barts excluded so many cases due to psychiatric diagnoses, since Peter White is a coauthor of the Oxford criteria, which are very permissive for psychiatric causes of fatigue. The PACE trial, for which Barts was a recruitment centre, also used Oxford criteria.



Conclusion for phase 1. Newcastle confirmed CFS in 60% of everyone it saw; across its whole referral funnel Barts confirmed 42%, or the low-to-mid forties allowing for screening error; the earlier Newcastle audit found 36%. These audits support a range rather than a single figure, and something around 50% misdiagnosis sits fairly within it — a fair, even conservative estimate of the rate of misdiagnosis among GP referrals.




And none of these figures involved screening for post-exertional malaise. The 50% is therefore the rate before applying PEM, so is likely too high: these are credible cases, not the unknown "true cases".

Notes:

Newcastle's idiopathic cases. Newcastle's idiopathic chronic fatigue rate — 13% of its non-CFS patients, 5% of all patients assessed — is also well above the Barts equivalent, and may be mostly undetected psychiatric cases.

Why 54% is not the comparable figure. Barts 54% is a rate among patients who had already survived the letter screen, whereas Newcastle saw everyone referred. Across the whole funnel Barts confirmed 42%, or 42%-48% allowing for screening error of up to 25% (post 3 gives the derivation). Those, rather than 54%, is the figure comparable with Newcastle's 60%.

PEM. Neither study excluded anyone for lacking PEM so the true rate is likely to be lower. However, it maybe that excluding those with alternative diagnoses effectively removes many of those without PEM

 

[Evergreen]

It always feels a bit like Christmas when you take a close look at something, Simon!

Careful and clear, as always.

By contrast, my post will be a bit rambly due to brain fog.

On this point:

Referral bias is unlikely to be a factor, as GPs have little choice of where to refer; 'nearer CFS clinic' was the top reason for Barts' letter rejections. The main factor is likely to be the quality of psychiatric assessment. Barts screened every referral with a consultant liaison psychiatrist, and every assessed patient had a mental state examination, though the paper does not say who carried it out. Newcastle had no psychiatrist on the team and describes neither. Both medical and psychiatric assessment are required by Fukuda and the other case criteria.

If that reading is right, Newcastle missed cases that Barts would have identified, so its 60% is higher than its true rate, which was probably closer to that of Barts. That would make Barts the better guide.

I think it's hard to disentangle referral bias from what we might call specialisation bias.

I agree that being seen by both an infectious diseases consultant and a psychiatrist rather than just a physician is more likely to identify those with a psychiatric explanation for their fatigue and other symptoms.

However, I suspect there is still likely to be significant referral bias, despite the geographic rejections. A GP could decide not to refer a patient with chronic fatigue to the local CFS clinic because they know the clinic is run by psychiatrists offering CBT/GET and the GP doesn't think that particular patient's fatigue is psychological or psychiatric in origin or doesn't think they'll benefit from CBT/GET. The same GP might refer another patient with chronic fatigue to the local clinic because they think they might have a bit more going on psych-wise and might benefit.

[Edit: This is incorrect: see Simon's reply below for the comparable percentages.] Newton et al. write:

The Newcastle CFS Service has close links with the local sleep service and cardiovascular testing.

• Newcastle diagnosed 20% with sleep disorders, compared to 6% at Bart's.
• Newcastle diagnosed 4% with cardiovascular disorders, which they clarify was vasovagal syncope, whereas Bart's diagnosed 2% with "cardiac disorders and infections".

So we could equally argue that Bart's probably missed sleep disorders and what we might call primary orthostatic intolerance.

I think at least some GPs would, over time, get to know who does what and exercise a bit of discretion in terms of whether they refer a patient or not.

At the letter stage, Bart's batted away 52 "on the grounds of a likely alternative psychiatric diagnosis", and then went on to diagnose another 54 with psychiatric diagnoses. So 24% of the people referred to them either already had or were given a psychiatric diagnosis. [Edit: The next line is incorrect: see Simon's reply below for the comparable percentages.] Newton et al. write: 24% psychiatric at Bart's vs 15% psychiatric at Newcastle is smaller than the difference in diagnosis of sleep disorders.

Hope it's OK to copy some posts from another thread where we talked about misdiagnosis after @Peter T brought up the studies:

at least one study reported in the region of 40% of false positives and another 40% of false negatives. There are enormous problems resulting from lack of a biomarker, multiple definitions (both clinical and research) and minimal general medical training on this topic, so I would not place any reliance on these figures, though to me 40% false positives does not seem not seem impossible if you looked at a historic contingent diagnosed as CFS, under such as the Oxford criteria by clinicians who do not reliably distinguish between the symptom of chronic fatigue and the condition chronic fatigue syndrome, then reassessed under say the NICE definition of ME/CFS which requires PEM. Obviously there could be reasonable discussion in this hypothetical cohort around what is misdiagnosis and what is a re diagnosis, however for many people this is just semantics and, though I hope, in the developed World at least, diagnostics are improving, there are still likely to be a lot of people out there who believe they have CFS and/or ME who would not meet any of the current criteria requiring PEM.

Though it might be a reasonable proposition that the number of false positives is declining for new cases, given the proliferation of new potential misdiagnoses such as MUS, FND and persistent somatic conditions it could even be that the number of false negatives is on the rise. This may be being exacerbated by confusions around definitions of Long Covid, Covid triggered ME/CFS and new cases of ME/CFS in people who have previously had a Covid infection.

Here's a little write-up by ME Research UK on the two (mis)diagnosis studies I posted about above.

They include their own anecdotal evidence:

This one includes a look at Walitt et al.'s cohort:

I think a lot of the misdiagnoses would work themselves out in time, because many conditions will reveal themselves with new symptoms and signs.

In pre-pandemic studies, patients were usually seen at specialist clinics after about 3 years (see Collin & Crawley 2017, PACE etc). A long time to go misdiagnosed, but some, if not many, conditions can be a bit nebulous in the early years.



Regarding Walitt et al.'s cohort, of 484 who were initially screened (this is all in the supplementary data, file 2, download where it says "Supplementary Data 1-24" here):

View attachment 28646


And in the next round of exclusion where medical records were reviewed:

View attachment 28647

If moderators want to copy these posts into a separate "Misdiagnosis" thread, I'm fine with that.

It's hard to tell what proportion of those screened for the Walitt et al. study had psychiatric diagnoses, as they could be included in the 78/484 with "Other diseases". Of those whose medical records were reviewed, 10/217 had psychiatric diagnoses. Of those who underwent medical evaluation, there's a group that was excluded "based on being adjudicated NOT post-infectious", which could include some with psychiatric diagnoses. [Sidebar: I noticed that IOM criteria were more permissive than Fukuda, with 17 meeting IOM and 14 meeting Fukuda. Relevant to my concern about the use of the IOM criteria in Jason's studies discussed elsewhere.]

And we have a member who features in the Bart's study:

WOW I am one of these. I’m one of the accepted but only depressed with Fibromyalgia, in mid 2007.


So very much proves my point; been assessed 3 times in recent years and clearly stated by a Dr and IDConsultant that I have ME and looking at my history back to diagnosis in 2007 by Mr P White, I have always had ME

Hi all

I was reviewing some ME history and PACE trial stuff last year when it suddenly hit me -That Peter White was the Peter White I saw at Bart’s in 2007.
I had mild ME at that time and was still working. He diagnosed me with Fibromyalgia. He asked a really leading question - which is worse, pain or fatigue? I said pain as I’d recently had to stop ibuprofen after a stomach bleed and was in a lot of pain taking paracetamol.
I had the non-specific cold/sore throat, bad sleep and PEM.

@MrMagoo's experience raises the possibility/probability that some of the misdiagnoses were themselves misdiagnoses.

 

[MrMagoo]

I know we are in n=1/territory with my story

-Yes, Peter White misdiagnosed me and I’m in his “did not have ME” stats of the Bart’s report. Happy to DM the evidence @Simon M if helpful.

Yes, I am also in DecodeME, and have several NHS diagnosis of ME/CFS and a 2011 letter from a Rheumatologist saying he doesn’t think it’s FMS he suspects ME

-my 2007 GP begged me not to be referred and said it wouldn’t help me (but I was put under pressure by work OH dr to seek a referral)

-my GP was young, the large GP surgery was just about within a distance of walking to Bart’s (for a well person) a few tube stops away

- White’s advice for treating the fibromyalgia and depression he diagnosed me with was…GET and CBT

(fortunately there wasn’t a proper GET service, just a slightly confused physio who gave me a daily stretches routine, next time asked how it went, told me to keep increasing the number of stretches then discharged me!) Sadly the CBT was well established.

As a side note I have also always been somewhat bothered by the fact that in my letter/assessment report by White, he reworded my background to erase/remove any LGBT reference, would we say say he “heterowashed” perhaps?

 

[MrMagoo]

“Neither study excluded anyone for lacking PEM” actually I suspect neither study considered PEM or took any account of it, Bart’s in particular; I don’t see any evidence it was even considered at any stage of my referral and assessment at Bart’s. I certainly hadn’t heard of it as a concept back then and wasn’t told of it until many years later.

 

[MrMagoo]

I also think it worth highlighting that the Bart’s/ East London Service was run by a Consultant Psychiatrist and Co-author of the Oxford Criteria

Peter White

me-pedia.org

 

[Simon M]

Thanks for taking the trouble to go through my posts.

However, I suspect there is still likely to be significant referral bias, despite the geographic rejections. A GP could decide not to refer a patient with chronic fatigue to the local CFS clinic because they know the clinic is run by psychiatrists offering CBT/GET and the GP doesn't think that particular patient's fatigue is psychological or psychiatric in origin or doesn't think they'll benefit from CBT/GET. The same GP might refer another patient with chronic fatigue to the local clinic because they think they might have a bit more going on psych-wise and might benefit.

That's a good point and could account for some of the difference. But how common are GPS like that? My experience over several decades is that they are extremely rare, which reflects what I recall of others with ME saying. A journal published survey with a good response rate from that time show a dismissive attitude to ME prevailing amongst GPs - Bowen et al., 2005.

Your post has a lot of striking comparisons, but they are not on a like for like basis. Confusingly, Newton quotes percentages of those with alternative diagnoses while the Barts figures are of those who were assessed. The picture looks considerably different using comparable numbers, which is what my analysis is based on (it is now, initially I went down the same route as you and reached similar conclusions):

Newton et al. write:

• Newcastle diagnosed 20% with sleep disorders, compared to 6% at Bart's.
• Newcastle diagnosed 4% with cardiovascular disorders, which they clarify was vasovagal syncope, whereas Bart's diagnosed 2% with "cardiac disorders and infections".

So we could equally argue that Bart's probably missed sleep disorders and what we might call primary orthostatic intolerance.

The 20% is the percentage of alternative diagnosis. The like for like figure for comparison is 8.5% of all cases assessed vs 6% for Barts, which isn't a huge difference, particularly given that Newcastle had links with a sleep clinic - see the graph from the comparison post above, repeated here, and the equivalent comparison figure for cardiovascular disorder is 1.5%

Spoiler: Alt diagnoses as % of those assessed

And I wouldn't expect identical figures for the different clinics, just similar.

24% psychiatric at Bart's vs 15% psychiatric at Newcastle is smaller than the difference in diagnosis of sleep disorders.

Similarly, the comparable figures are 21.6% at Barts vs 5.8% at Newcastle = +15.6%. For sleep and heart the differences (Barts-Newcastle) are -2.5% and -1.5% (=-4%).

>At the letter stage, Bart's batted away 52 "on the grounds of a likely alternative psychiatric diagnosis",
You might also could say they batted away 67 on the grounds of a likely alternative medical diagnosis. However, only 77 people in total were rejected for alternative diagnoses, 119 in total among 77 patients and we don't know how they split. (Those assessed only had a single diagnosis.). It's also worth noting that 46% of those they did assess were found not to have CFS, suggesting the two consultants were taking a conservative approach to letter rejections, only rejecting those they were confident did not have CFS. The paper acknowledges that some of these decisions were probably wrong - and the 42%-48% range I estimate assumes that between 0% and 25% of rejections were wrong.

Hope it's OK to copy some posts from another thread where we talked about misdiagnosis after @Peter T brought up the studies:

Absolutely. When I started the analysis I expected the usual feeling of deflation I get on careful reading of a study whose results I find interesting. But here I found both studies convincing because of their detail and the strength of the clinics. . Not perfect but good. No MUS diagnoses, for instance..

@MrMagoo's experience raises the possibility/probability that some of the misdiagnoses were themselves misdiagnoses.

I agree, and that would put up the true rate of CFS, though we don't know by how much.

Equally, as @MrMagoo also points out, there was no assessment of PEM and probably no understanding of it either, which, as its the hallmark symptom, could pull down the CFS rate much further.

@MrMagoo , if I am well enough I will reply to your thoughtful post tomorrow. I don't doubt what you say, but feel free to DM me if you think that would be useful.

 

[Evergreen]

Your post has a lot of striking comparisons, but they are not on a like for like basis.

Oh thank you for pointing that out. I miss things like that when brain not working well! Will look at it again tomorrow and edit my post so it won't confuse anyone.

 

[MrMagoo]

Thanks for taking the trouble to go through my posts.

That's a good point and could account for some of the difference. But how common are GPS like that? My experience over several decades is that they are extremely rare, which reflects what I recall of others with ME saying. A journal published survey with a good response rate from that time show a dismissive attitude to ME prevailing amongst GPs - Bowen et al., 2005.

Your post has a lot of striking comparisons, but they are not on a like for like basis. Confusingly, Newton quotes percentages of those with alternative diagnoses while the Barts figures are of those who were assessed. The picture looks considerably different using comparable numbers, which is what my analysis is based on (it is now, initially I went down the same route as you and reached similar conclusions):


The 20% is the percentage of alternative diagnosis. The like for like figure for comparison is 8.5% of all cases assessed vs 6% for Barts, which isn't a huge difference, particularly given that Newcastle had links with a sleep clinic - see the graph from the comparison post above, repeated here, and the equivalent comparison figure for cardiovascular disorder is 1.5%

Spoiler: Alt diagnoses as % of those assessed

View attachment 34474

And I wouldn't expect identical figures for the different clinics, just similar.


Similarly, the comparable figures are 21.6% at Barts vs 5.8% at Newcastle = +15.6%. For sleep and heart the differences (Barts-Newcastle) are -2.5% and -1.5% (=-4%).

>At the letter stage, Bart's batted away 52 "on the grounds of a likely alternative psychiatric diagnosis",
You might also could say they batted away 67 on the grounds of a likely alternative medical diagnosis. However, only 77 people in total were rejected for alternative diagnoses, 119 in total among 77 patients and we don't know how they split. (Those assessed only had a single diagnosis.). It's also worth noting that 46% of those they did assess were found not to have CFS, suggesting the two consultants were taking a conservative approach to letter rejections, only rejecting those they were confident did not have CFS. The paper acknowledges that some of these decisions were probably wrong - and the 42%-48% range I estimate assumes that between 0% and 25% of rejections were wrong.



Absolutely. When I started the analysis I expected the usual feeling of deflation I get on careful reading of a study whose results I find interesting. But here I found both studies convincing because of their detail and the strength of the clinics. . Not perfect but good. No MUS diagnoses, for instance..


I agree, and that would put up the true rate of CFS, though we don't know by how much.

Equally, as @MrMagoo also points out, there was no assessment of PEM and probably no understanding of it either, which, as its the hallmark symptom, could pull down the CFS rate much further.

@MrMagoo , if I am well enough I will reply to your thoughtful post tomorrow. I don't doubt what you say, but feel free to DM me if you think that would be useful.

I’m just adding some colour for you, up to you what you want to do with it. And happy to offer proof, if proof should be required.

I think my TLDR is that this is all “pre-PEM” it’s totally limited due to that, so it makes me uncomfortable they are seen as a useful paper/test/statistic. I don’t think it should be endorsed. It’s bad enough people still quote the “40%misdiagnosis rate” as a known fact (like the “250,000 pwME in the UK” statistic which existed, unchanged, for over 30 years)

I’d feel better if it was said “we used to think the misdiagnosis rate was X, but there haven’t been any UK studies since PEM was introduced by NICE so we really don’t know”.

That’s before we get into the “my personal experience” rabbit hole, and PD White marking his own homework.

But I’m not an academic or a scientist or researcher, so it’s just my personal opinion!

 

[Liii]

Very interesting information.

I suspect this is much of explanation for the stories we hear about recoveries after CBT/GET or similar cognitive-behavioural courses ("brain retraining").

 

[neurophusion]

Great work @Simon M.

Adding to what @MrMagoo reported:

* From the full unabridged PACE trial protocol on how the Oxford criteria for CFS was assessed: "Is your fatigue (or a synonym), the principal (main, primary) symptom (e.g. tiredness, lack of energy, weariness, exhaustion)?"

* From the 2011 publication in the Lancet, the results apply to patients "only if fatigue is their main symptom".

 

[MrMagoo]

Great work @Simon M.

Adding to what @MrMagoo reported:

* From the full unabridged PACE trial protocol on how the Oxford criteria for CFS was assessed: "Is your fatigue (or a synonym), the principal (main, primary) symptom (e.g. tiredness, lack of energy, weariness, exhaustion)?"

* From the 2011 publication in the Lancet, the results apply to patients "only if fatigue is their main symptom".

Yes, absolutely it says in my notes I identified pain as my main symptom. It’s crazy to make you choose! Or lead you to differentiate, whatever.

But in my defence I’d been taking so many painkillers I’d given myself a stomach bleed so I wasn’t taking anything (and I was working full time because I was mild, y’know). I had enough general knowledge to know they can’t treat fatigue but they can treat pain. And surely being in pain is what keeps you tired, disturbs your sleep etc?

Also they used to ask you to walk up a big old-fashioned staircase to the consulting room

 

[Trish]

Thanks for this interesting analysis, Simon. I am surprised there is no other study more recently. Surely just the sort of task to give to a PhD student. Though I guess there are so few if any specialist ME/CFS services in hospitals linked to academic departments that have PhD students.

I have many concerns, shared by you and others, that the data tells us very little about what is really going on with diagnosis and misdiagnosis in current ME/CFS services in the UK.

On the Peter White paper, it may be apparently well done in terms of providing data, but he used the Oxford criteria which required fatigue as the primary presenting symptom, and it is in their interests to only diagnose CFS for people who they think will be good candidates for GET, including, for example, people who are deconditioned, mildly depressed or anxious, with burnout, stress or other 'lifestyle' induced tiredness. As MrMagoo has reported, if someone actually had physical symptoms like muscle pain they wouild be likely to diagnose them with FM, or gut symptoms, IBS, or headaches, migraine and so on. I wouldn't trust their data at all.

 

[Hoopoe]

I don't think this exercises tells us much. There is referral bias. Clinics are biased to diagnose whatever it is they are funded to diagnose.

 

[richie]

Phase 1: Two large and detailed audits of patients at well resourced CFS clinics

Diagnostic difficulties

ME/CFS is defined by disabling fatigue lasting at least six months, post-exertional malaise, and, for an IOM diagnosis, cognitive problems and/or orthostatic intolerance, with more symptoms required for CCC. There is no diagnostic test.

The fatigue must be unexplained, which means excluding the many other conditions that cause it — and that exclusion is difficult because chronic fatigue has so many medical and psychiatric causes. Since the hallmark symptom of PEM is rarely assessed - at least not well - what matters most is the quality of the work done to rule out the alternatives before a diagnosis is given. Fukuda and Canadian criteria require a thorough medical and psychiatric assessment as part of diagnosis.

The following two studies provide well-documented and detailed assessments and diagnosis.

Newton, 2010. 60% of patients seen at Newcastle Clinic had CFS, maybe fewer​

Source: Newton et al., 2010. The Newcastle NHS Chronic Fatigue Syndrome Service: not all fatigue is the same. PMID 21132135

This is the first of two studies that looked directly at how many people referred to a specialist CFS clinic turn out to have something else. The Newcastle fatigue service was run by Julia Newton, a senior clinician and researcher specialising in the illness, with a well-resourced multidisciplinary team — a far better-equipped clinic than most.

Newcastle confirmed CFS in 60% of everyone it saw, giving the remaining 40% alternative diagnoses. The main ones were other chronic diseases, sleep disorders and psychiatric illness.

An earlier internal audit at the same clinic, using a random sample of case-notes, found a much lower rate: only 36% confirmed as CFS.

These are audits of routine clinical practice rather than one-off studies, so they say something about what a thorough assessment finds in ordinary referrals. An important point for the comparison with Barts (post 3): Newcastle had no psychiatrist on the team, and the paper describes no psychiatric screen or mental state examination. If that meant psychiatric cases were missed, the true CFS rate is lower than 60%.

Also, nobody was excluded for lacking post-exertional malaise or other symptoms, so 60% is the rate before any PEM screen.

Spoiler: Study detail

Patients and method. 260 patients were assessed between November 2008 and December 2009; non-attenders were excluded before counting. The team could refer on to other specialties. They cited the Fukuda criteria but gave no details of how they were applied, and the only exclusions were for alternative diagnoses.

Results. 157 (60%) were confirmed as CFS. As a percentage of all patients assessed, the main alternative diagnoses were: other chronic disease 19%, sleep disorders 8%, psychiatric 6%, idiopathic chronic fatigue 5%.

The earlier audit. For 2005–7 a random sample of 56 case-notes was drawn from 375 patients: 20 of the 56 (36%) were CFS cases, far below the 60% (157/260) recorded in 2008–09 (p<0.0001). The paper gives no confidence interval; AI’s exact binomial calculation gives a 95% CI of 23%–50%.

What was not assessed. There was no psychiatrist on the team, and the paper describes neither a psychiatric screen nor a mental state examination — both mental and physical health assessment are recommended in Fukuda and the other case criteria. No minimum fatigue or function threshold is reported, though this was probably applied subjectively, and strict thresholds are not needed in clinical practice.

Annoying that they contrast medical causes with psychiatric ones as if psychiatric conditions were not medical. An open door for "what a bunch of dualists" criticism form the usual quarters and as confusing to the uninitiated as competing usage of the term "inflammation".

Mental distress is very common in chronic ill health and immunological conditions are replete with patients with psychiatric comorbidities . This must be borne in mind so as not to exclude those with both/and, though lack of diagnostic biomarkers poses particular problems. Excluding psychiatric co-morbids from research co horts is another matter, I accept.

 

[Utsikt]

I would also be in the «did not have ME/CFS» category for at least two institutions. One because I complied with GET which for some reason meant that I did not have ME/CFS, and one because of a bad referral letter from my GP so they refused to see me because I didn’t fulfil the CCC (I did, the GP just didn’t mention all of my symptoms).

 

[Simon M]

I know we are in n=1/territory with my story

-Yes, Peter White misdiagnosed me and I’m in his “did not have ME” stats of the Bart’s report. Happy to DM the evidence @Simon M if helpful.

Yes, I am also in DecodeME, and have several NHS diagnosis of ME/CFS and a 2011 letter from a Rheumatologist saying he doesn’t think it’s FMS he suspects ME

I am sorry you had such a bad experience at Barts and with Peter White, and it is very relevant to these studies.

Re fatigue not being the main symptom as a reason for non-diagnosis:

* From the full unabridged PACE trial protocol on how the Oxford criteria for CFS was assessed: "Is your fatigue (or a synonym), the principal (main, primary) symptom (e.g. tiredness, lack of energy, weariness, exhaustion)?"

* From the 2011 publication in the Lancet, the results apply to patients "only if fatigue is their main symptom".

>Yes, absolutely it says in my notes I identified pain as my main symptom. It’s crazy to make you choose! Or lead you to differentiate, whatever.

But in my defence I’d been taking so many painkillers I’d given myself a stomach bleed so I wasn’t taking anything (and I was working full time because I was mild, y’know). I had enough general knowledge to know they can’t treat fatigue but they can treat pain. And surely being in pain is what keeps you tired, disturbs your sleep etc?

That's an interesting point, and might explain why they decided not to give you a diagnosis. It isn't a good reason.

I think all case definitions including IOM and CCC begin with fatigue and the need for it to be unexplained, but making fatigue the main symptom seems to be an extreme interpretation of the principle by the Pace trial, and presumably by Barts too. FWIW, my fatgiue was terrible at one point (I was bedbound) but my pain was my worst symptom despite opioids, though in my case I still got a CFS diagnosis, from another author of the Oxford crtieria.

“Neither study excluded anyone for lacking PEM” actually I suspect neither study considered PEM or took any account of it, Bart’s in particular; I don’t see any evidence it was even considered at any stage of my referral and assessment at Bart’s. I certainly hadn’t heard of it as a concept back then and wasn’t told of it until many years later.

I agree. It was around as a concept though not in most clinics. I'm not sure that has changed much in reality

I think my TLDR is that this is all “pre-PEM” it’s totally limited due to that, so it makes me uncomfortable they are seen as a useful paper/test/statistic. I don’t think it should be endorsed.

I’d feel better if it was said “we used to think the misdiagnosis rate was X, but there haven’t been any UK studies since PEM was introduced by NICE so we really don’t know”.

I am not sure we are in the PEM era now, at least not as far as NHS clinics go. PEM became a mandatory symptom in the NICE crtieria since 2021, but what does that mean in practice? Do staff in BACME clinics know what PEM is and apply it well? I would be very surprised.

Probably most research studies that people here like use weak PEM assessment, often via the DePaul Symptom Questionnaire whose "PEM" definition only describes exertion intolerance, which is common to man illnesses, including MS and depression. That definition has been widely criticised on S4ME for many years.

DecodeME is the first study that made a decent attempt at captruing PEM, but I think research studies that we can say robustly assessed PEM in their participants are thin on the ground.

It's also the case that at least all research definitions require a thorough medical and psychiatric diagnosis to exclude alternative causes because fatigue has so many, and ME/CFS is often a diagnosis of exclusion. Such assessments is seen as good medical practice by doctors regardless of how they view the illness, and very few clinics do this well - which is why these studies stand out.

If I have understood right, you are saying that at least some of these exclusions could be simply comorbidities for people who have PEM and so have ME?

That is certainly possible - you would be a prime example. How common your experience is, I don't know. The conplaint I have heard about diagnosis is that it takes years to seem a consultant to get one. Yours is the first experience I have heard, and it's good to have it discussed here.

- White’s advice for treating the fibromyalgia and depression he diagnosed me with was…GET and CBT

That probably applies to many other diagnoses too at that clinic. I guess at least you didn't miss out on the brilliant treatment available at Barts for CFS. Though a proper diagnosis was presumably your aim.

I also think it worth highlighting that the Bart’s/ East London Service was run by a Consultant Psychiatrist and Co-author of the Oxford Criteria

Peter White

me-pedia.org

I agree, and that's why I had highlighted it too:

I was surprised that Barts excluded so many cases due to psychiatric diagnoses, since Peter White is a coauthor of the Oxford criteria, which are very permissive for psychiatric causes of fatigue. The PACE trial, for which Barts was a recruitment centre, also used Oxford criteria.

Let me know if I have missed any of your points, I didn't want to make it too long.

 

[MrMagoo]

I am sorry you had such a bad experience at Barts and with Peter White, and it is very relevant to these studies.

Re fatigue not being the main symptom as a reason for non-diagnosis:

>Yes, absolutely it says in my notes I identified pain as my main symptom. It’s crazy to make you choose! Or lead you to differentiate, whatever.

That's an interesting point, and might explain why they decided not to give you a diagnosis. It isn't a good reason.

I think all case definitions including IOM and CCC begin with fatigue and the need for it to be unexplained, but making fatigue the main symptom seems to be an extreme interpretation of the principle by the Pace trial, and presumably by Barts too. FWIW, my fatgiue was terrible at one point (I was bedbound) but my pain was my worst symptom despite opioids, though in my case I still got a CFS diagnosis, from another author of the Oxford crtieria.


I agree. It was around as a concept though not in most clinics. I'm not sure that has changed much in reality



I am not sure we are in the PEM era now, at least not as far as NHS clinics go. PEM became a mandatory symptom in the NICE crtieria since 2021, but what does that mean in practice? Do staff in BACME clinics know what PEM is and apply it well? I would be very surprised.

Probably most research studies that people here like use weak PEM assessment, often via the DePaul Symptom Questionnaire whose "PEM" definition only describes exertion intolerance, which is common to man illnesses, including MS and depression. That definition has been widely criticised on S4ME for many years.

DecodeME is the first study that made a decent attempt at captruing PEM, but I think research studies that we can say robustly assessed PEM in their participants are thin on the ground.

It's also the case that at least all research definitions require a thorough medical and psychiatric diagnosis to exclude alternative causes because fatigue has so many, and ME/CFS is often a diagnosis of exclusion. Such assessments is seen as good medical practice by doctors regardless of how they view the illness, and very few clinics do this well - which is why these studies stand out.

If I have understood right, you are saying that at least some of these exclusions could be simply comorbidities for people who have PEM and so have ME?

That is certainly possible - you would be a prime example. How common your experience is, I don't know. The conplaint I have heard about diagnosis is that it takes years to seem a consultant to get one. Yours is the first experience I have heard, and it's good to have it discussed here.



That probably applies to many other diagnoses too at that clinic. I guess at least you didn't miss out on the brilliant treatment available at Barts for CFS. Though a proper diagnosis was presumably your aim.

I also think it worth highlighting that the Bart’s/ East London Service was run by a Consultant Psychiatrist and Co-author of the Oxford Criteria

I agree, and that's why I had highlighted it too:

Let me know if I have missed any of your points, I didn't want to make it too long.

Well, regarding PEM, I didn’t explain this fully. The notes from Peter White quite clearly describe PEM. I had the notes reviewed by the AfME Dr who wrote in my diagnosis report that it was clear from White’s notes I’d always had ME.

So I’m coming at it from the general “PEM wasn’t considered” angle but from a personal level, PEM wasn’t even identified despite evidence of it being recorded in White’s report, which I think is important.It’s not just that “fatigue” wasn’t number one so therefore it’s not CFS, the report shows a classic description of ME (interestingly not a description of FM - clearly states pain increases from work, walking or jogging) yet gets categorised as FM.

Depression was of course going to be blamed, I don’t recall if I was depressed at that point, possibly I was. I knew things had changed when a later dr said “well, who wouldn’t be depressed in those circumstances? Trying to keep your job, life, whilst feeling so ill, and not getting better!”.

We can argue back and forth about whether it was right that he took the approach of “fatigue as main symptom” but in my opinion actually he disregarded relevant information (if fatigue wasn’t the main issue).

Also I’m not sure where the discussion about GP referral standards fell, but certainly as someone in the catchment area 2 of the 3 surgeries I used over the years were staffed by young, newish qualified fairly switched on GPs (last one was only part-time as was a researcher rest of the time) and 2 of 3 were absolutely tried to talk me out of referrals, and asked me to reconsider whether it would be better to just have GP mgmt.
The 3rd surgery I used was just locums, and I left due to that.

I didn’t notice wher in the report it stated Bart’s was run by a Consultant Psychiatrist, just that Newcastle didn’t have a Psychiatrist, hence I raised it, but perhaps I missed it.