Prevalence of ME/CFS in 3 Regions of England - new study

[jace]

This new study took a sample of 143,000 people. They defined 'CFS' using Fukuda, the CCC and ECD definitions, though it's not clear if they used them all on each patient.

"Results
The estimated minimum prevalence rate of ME/CFS was 0.2% for cases meeting any of the study case definitions, 0.19% for the CDC-1994 definition, 0.11% for the Canadian definition and 0.03% for the ECD. The overall estimated minimal yearly incidence was 0.015%. The highest rates were found in London and the lowest in East Yorkshire. All cases conforming to the Canadian criteria also met the CDC-1994 criteria but presented higher prevalence and severity of symptoms."

I've never heard of the ECD, so I went looking for it. I guess it's this one

Osoba T, Pheby D, Gray S, Nacul L: The Development of an Epidemiological Definition for Myalgic Encephalomyelitis/Chronic Fatigue Syndrome. Journal of Chronic Fatigue Syndrome 2008, 14(4):61-84.

I can't find it online.

We used to be told prevalence was 0.4%. Reeves 2005 diagnoses 'CFS' in some 2.5%. Creepy Crawley reckons about 5% of children in the UK have it - yeah right! Yet this large study gives 0.2% - half of the standard figure.

Using the figure of 0.2% and extrapolating to the whole population of the UK, that gives a total number of us as 130,000, as opposed to the commonly quoted 250,000 figure.

http://www.biomedcentral.com/content/pdf/1741-7015-9-91.pdf

 

[Enid]

It's still not recognised (all the mind) diagnosed, nor treated amongst so many GPs - this must affect any estimate of the entire UK population afflicated. I suggest much is hidden as ME sufferers cannot go beyond "first base" - their GPs.

 

[Firestormm]

28 July 2011: http://www.biomedcentral.com/1741-7015/9/91/abstract

An interesting study though I confess to not having read the whole paper that Jace has linked to above. This seems particularly opportune:

'We suggest combining the use of both the CDC-1994 and Canadian criteria for ascertainment of ME/CFS cases, alongside careful clinical phenotyping of study participants. This combination if used systematically will enable international comparisons, minimisation of bias, and the identification and investigation of distinct sub-groups of patients with possibly distinct aetiologies and pathophysiologies, standing a better chance of translation into effective specific treatments.'

At least by combining existing criteria one might save on the expense of a new one. As for the 0.2% - I will have to check - but here in the UK they were using a figure of 1 in 200 (based on US data) up until now. So very interesting to read of this development and will need to study the full paper.

Also thought it interesting to note - although no figures were mentioned - that 'chronic fatigue' would have generated far higher results than before either criteria were applied. So they are to some extent effectively streamlining the patients into some form of diagnosis...

 

[Bob]

That's interesting, thanks Jace.

I've never heard of the ECD either.

It's a shame that the new ICC weren't published in time for this study.

 

[Tony Mach]

Case ascertainment involved: 1) electronic search for chronic fatigue cases; 2) direct questioning of general practitioners (GPs) on cases not previously identified by the search; and 3) clinical review of identified cases according to CDC-1994, Canadian and Epidemiological Case (ECD) Definitions. This enabled the identification of cases with high validity.

They were looking for "definitive" cases.

Knowing how reluctant doctors are just making the diagnose of CFS (regardless of criteria), and knowing that about 50% of people with ME/CFS just don't bother anymore to seek help from the medical system, I would guess the numbers are too low. I would guess the actual incidences have to be multiplied by at least a factor of two, if not three.

There is something about this kind of research (and how it is done, and how not) in "Osler's Web", but I can not find it now. Was it Leonard Jason? Or Martin Lerner?

 

[Firestormm]

Good point Tony. The authors looked at 'clinical presentation' and that would mean (I am supposing) that these patients had a) been diagnosed and b) were registered. So in terms of true prevalence the 0.2% might not be representative of the population at large.

Indeed when I asked for the figures from my own PCT with a Freedom of Information request (these data are just not readily available), the 0.5% prevalence used resulted in approximately 0.25% actual number of patients registered with either General Practitioner and/or CFS/ME Service.

I think I had better read the whole paper to better understand where they are coming from.

 

[oceanblue]

What makes this study so impressive is its size - covering 143,000 people registered at doctors' practices and yielding 270 confirmed cases of ME/CFS, way bigger than any previous prevalence study. It also gives us the first reliable data on what proportion of people with CFS (defined by the common CDC-94/Fukuda criteria) that also meet the Canadian criteria: just over half.

However, as the study relies on GPs (General Physician, in the US, I think) diagnosis, they may not pick up all cases, though the study was "careful to choose practices with GPs with experience in diagnosing ME/CFS".

Comparison with other prevalence studies: all using CDC-94 criteria

Nacul 2011: sample*=143,000; 258 CFS cases, 0.19% prevalence. UK, mixed centres
Reyes 2003: sample=56,146; 43 cases, 0.24%. Urban, USA
Jason 1999: sample=18,675; 32 cases, 0.42%. Urban, USA
Wessely 1995: 1,985; 36 cases, 2.60%. UK, mixed
*Screened population: some contact with all of these but not all will have completed the assessment process

So the sample size in this new study is much bigger than in earlier ones. The prevalence of 0.19% for CDC-94 CFS cases is lower than the Reyes study (but not significantly different, statistically speaking) and a substantially lower than the Jason study. The Wessely study is just a bit of a laugh (see post #10).

Study Methodology
To understand the strenghts and weaknesses of this new study it helps to see how they diagnosed patients:

1. GP diagnosis

• Searched the GPs' practice database for patients with a recorded diagnosis likely to be ME/CFS e.g. CFS, ME, Fibromyalgia, Fatigue Syndrome.
• Asked GPs to review these cases (on paper, not in person) and confirm the ME/CFS cases using study case definitions.
• GPs were also asked to recall any cases that had been missed by the electronic search

573 cases were initially identified with 63 (11%) subsequently excluded by GPs on review.

2. Questionnaire comfirmation
The 510 GP-diagnosed cases were mailed a screening questionnaire to confirm symptoms matched any of the studies case definitions: 265 (69%) were confirmed as cases (by any study definition), 122 as non-cases.

This leads to a prevalence of 0.19% for CDC-94 CFS and 0.11% for Canadian Criteria ME/CFS.

As the authors (and posters here) note, the population prevalence rate may be higher than that, as some people may not consult their GPs, or do not receive a diagnosis.

On the other hand, these diagnoses were not independently confirmed by a CFS specialist, and it's not clear if all the lab tests or the psychiatric evaluation required by the CDC-94 criteria have been applied. This may have lead to some non-cases being classified as ME/CFS, leading to overstating of prevalence. That said, I suspect that overall the study under-reports the true cases of ME/CFS so the 0.19% prevalence rate is likely to be a lower limit for CDC-94 CFS.

Canadian vs CDC-94 criteria
Comparing Canadian criteria patients with CDC-only (ie met CDC-94 criteria but not Canadian ones), Canadian criteria patients had worse pain and were more disabled. They also had more symptoms, but as the Canadian criteria require more symptoms than CDC-94 by definition, that doesn't necessarily tell us anything.

What is surprising, and perhaps significant, is that the authors found no difference between Canadian and Fukuda in either the nature of triggering factors or in the proportion with Fibromyalgia. This is some - limited - evidence that the definitions might not be selecting very different types of patients.

The authors comment:

It is unclear whether cases conforming to the Canadian criteria represent the severe end of a disease spectrum, or if they represent a distinct clinical entity with distinct aetiology.

Like firestormm, I like the authors suggestion of always using the Canadian criteria alongside the CDC-94 ones in research studies.

 

[Firestormm]

Excellent post their Ocean. Sorry I dropped the ball on this thread. I have done some partial work on the full paper - just need to get back to it. I agree though it is probably the best prevalence study I have seen to date.

Previous studies though - off the top of my head - in the UK were based on a figure of 1 in 200 (from the US). This would equate to 0.5%. I confirmed this was being used when I applied under the Freedom of Information Act for data relating to my county.

As a rule - very annoyingly - the NHS don't carry such figures nationally, but they could confirm the number of people registered with a GP practice in the county with 'CFS/ME' as their diagnosis. Still it is very 'hit and miss'.

Will bookmark this discussion and return

Actually:

In the 'Neuro Numbers' 2003 publication (download from Neurological Alliance website) - published in association with the Association of British Neurologists - the prevalence figures were 0.3-0.5% (from memory).

This was itself based on a research paper (I think) submitted to parliament in 1999. Incidentally, Myalgic Encephalomyelitis was the label/name used at that time.

Of further interest might be that ME was included in Neuro Numbers as a neurological conditions AND this publication was used primarily in the National Service Framework for Neurological Conditions published in 2005. The NSF runs for 10 years and is still very much valid by the way.

I should know

 

[Bob]

Ocean, thanks for doing the work on this paper, and sharing. Much appreciated.

 

[Graham]

Very useful! Thanks Jace and Ocean. It backs up the idea that perhaps only a third (at most) of the PACE trial patients would actually have conformed to the Canadian criteria.

 

[oceanblue]

Thanks for that, folks

Firestormm, what was the sample size for your FOI request, and did they say if any CFS clinic cases were excluded from the GP data? The 1 in 200 would presumably be using the Jason data if based on US figures, or possibly the Wessely figure which came up with 0.5% prevalence if excluding patients with psychological morbidity (though I'm not sure if such a definition has any meaning in a chronic illness).

Graham: what isn't clear yet is what the significance of Canadian vs Fukuda-only patients is, and whether or not they would respond differently to PACE treatments.

 

[Graham]

Thanks for that, folks

Graham: what isn't clear yet is what the significance of Canadian vs Fukuda-only patients is, and whether or not they would respond differently to PACE treatments.

True! But what does seem likely is that those who conform to the Canadian criteria are likely to be more severely affected, so a disproportionate number of the worst affected are likely to be "Canadians". That might be relevant when we (if?) get some idea of individual improvements. If the psychological model for the PACE patients fails (which is a fair assumption given such poor results), then a new model will be needed to explain the complex pattern of results. My tentative playing about with the data splits the patients into three roughly equal groups, so if the Canadian set is about a third, that may be relevant to my musings. We only need a model because the PACE/Oxford group is such a mixed one.

 

[oceanblue]

True! But what does seem likely is that those who conform to the Canadian criteria are likely to be more severely affected, so a disproportionate number of the worst affected are likely to be "Canadians". That might be relevant when we (if?) get some idea of individual improvements. If the psychological model for the PACE patients fails (which is a fair assumption given such poor results), then a new model will be needed to explain the complex pattern of results. My tentative playing about with the data splits the patients into three roughly equal groups, so if the , Canadian set is about a third, that may be relevant to my musings. We only need a model because the PACE/Oxford group is such a mixed one.

Must be torture for you not to have the raw data! I think they will be reporting on the effect of initial severity on outcome which might throw some light on this - but it seems the next paper out will be on 'recovery'.

 

[Firestormm]

Thanks for that, folks

Firestormm, what was the sample size for your FOI request, and did they say if any CFS clinic cases were excluded from the GP data? The 1 in 200 would presumably be using the Jason data if based on US figures, or possibly the Wessely figure which came up with 0.5% prevalence if excluding patients with psychological morbidity (though I'm not sure if such a definition has any meaning in a chronic illness).

Funnily enough I was listening to the BMJ Learning Module: Neurological Disorders: Chronic Fatigue Syndrome and Prof Wessely was asked about prevalence. This is what he said (dated March 2010):

'Simon: It all comes down to definition. Weve been doing epidemiological studies over the years and if you use a broad definition, which includes for example also people who have comorbid you know concurrent depression anxiety, it may be up to 1% - 2% of the population. If you go for a narrow definition it comes down to about 0.2%-0.3%. So its a bit like blood pressure, it all depends on where you draw the line.'

Interesting I thought. Anyway...

I am a moderator over on Foggy Friends. This question surrounding the 250,000 statistic in the UK has been a wee thorny in my side for a while. Rather than repost all our findings might I post a link: http://www.foggyfriends.org/forum/s...-ME-CFS-Statistic&highlight=250,000+statistic

There remains some doubt as to where that figure originates specifically but from what has been said to me it harps back to the 1980s. I was referred to this thread on PR: http://forums.phoenixrising.me/showthread.php?6575-quot-cfs-affects-17-million-worldwide-quot-source which is about the worldwide projections.

I did have this from Prof Pinching - my consultant here in sunny Cornwall - but again no source for this unfortunately:

'* 1 in 250 or 0.4% of the population have ME/cfs
* there are higher numbers of people with the illness in rural areas [a less hostile environment ?]
* females outnumber males 3:1
* ages 5 to 95 are affected
* 50% of those with the illness are seeking medical attention
* ME/cfs is not a disease of modern life
* there is a high incidence of ME/cfs in Brazil, India and Nigeria i.e. this is not a 'Western World' illness

Taking the population of Cornwall to be 531,100 [2009 gov. figs.] then over 2000 having ME/cfs is an estimate but the figures are probably higher, due to Cornwall being a rural area.'

It must all come from somewhere

'NICE. Chronic fatigue syndrome /myalgic encephalomyelitis: NICE guideline. http:// guidance.nice.org.uk/CG53/NICEGuidance/doc/English.'

Apparently that contains the 0.2-0.4% prevalence range - maybe it quotes a source? Will check when I get round to it

 

[Dolphin]

Interesting to see 71% of patients with a primary FMS diagnosis have CFS-Fukuda

Interesting to see 71% of English patients with a primary Fibromyalgia diagnosis have Fukuda CFS (many have Canadian ME/CFS)

See Table 2 in http://bit.ly/oLLawP

See also:

"However, we have not found a higher frequency of GP diagnosis of fibromyalgia in Canadian positive cases (data not shown, P = 0.9), compared to CDC 1994."

A p value of 0.9 suggests proportions are similar i.e. approx 11/20. So approx (11/20) * 71.1% =39.105% of those with a primary Fibromyalgia diagnosis have Canadian ME/CFS.

I think I will write to the authors to try to get the exact percentage.

 

[Dolphin]

Study found 77% cases are female

Assuming equal numbers of men and women in 18-65 age group, study found a Female:Male ratio of 77:23 (30:9* - see http://bit.ly/qoEiTw) i.e. 77% cases are female.

Note, there wasn't simply 39 (=30+9) individuals; this is from 30/10000 vs 9/10000.

 

[Dolphin]

Differential diagnoses

Always interesting to see diagnoses people initially diagnosed with an ME/CFS-type illness* go on to get:

From the 573 patients identified as having any primary diagnoses (in their medical records), 63 (11%) were excluded as they no longer had chronic fatigue at the time of the search, or had been diagnosed with other conditions explaining their symptoms, such as hypothyroidism, hyperparathyroidism, hydrocephaly, cancer, psychiatric conditions, Crohn's disease and rheumatoid arthritis.

* In this study, the records of patients who had been seen as having Fibromyalgia were also examined

 

[Guido den Broeder]

The CCC definition still includes some non-ME patients, so the prevalence of ME patients will be about 0,10% (that is, not including patients that overcome the disease within 6 months).

These figures would make the prevalence of ME dead equal to the prevalence of MS, which somehow makes a lot of sense to me.

 

[Dolphin]

The CCC definition still includes some non-ME patients, so the prevalence of ME patients will be about 0,10% (that is, not including patients that overcome the disease within 6 months).

These figures would make the prevalence of ME dead equal to the prevalence of MS, which somehow makes a lot of sense to me.

I think there may be some M.E. patients who don't satisfy the CCC. I don't believe I would have satisfied the CCC criteria in the first few years of my illness (when I was mildly affected and avoiding strenuous activity (i.e. I might have if somebody had pushed me for a week or two e.g. army training or something)). Which one mean one can't necessarily use this study to give a prevalence of 0.10% for M.E.

 

[Dolphin]

I learn something new every day.

Lots of people may know this but I thought there was an error in the following:

Annual incident risks also varied among and within the three regions, with an overall risk of 0.15.

I thought was a typo and it was saying 0.15% which was incorrect.
So I wrote to the authors about this (and a couple of other points). It turns out is per thousand.