Estimating Prevalence, Demographics and Costs of ME/CFS Using Large Scale Medical Claims Data and Machine Learning, 2018, Valdez, Proskauer et al

I find these figures hard to believe. The prevalence is almost double of previous estimates. Perhaps these insurance files relied on GP diagnosis of CFS? Previous studies showed that 40% of such GP diagnoses turn out to be wrong when checked by a specialist. Looking forward to the full text.

 

(I am quite fuzzy-headed at the moment and hope this is somewhat understandable)
I agree that if these numbers rely on GP diagnoses, it is likely that they are inaccurate because of the poor level of knowledge of most GPs.
Could it also be that the very high estimated prevalence is because they are talking about the prevalence of 2 things - ME and cfs? And that if one separated the ME diagnoses from the cfs diagnoses, the numbers would be more in line with what we have been thinking prevalences are?

 

Using the current US population, a prevalence of 857/100,000 calculates to roughly 2.8 million. That's 7x the prevalence of MS. I do find that hard to believe.

In the US, many providers won't allow an ME diagnosis, muddying the waters even further. For example, Kaiser Permanente currently only allows their doctors to make a "CFS" diagnosis (ICD-10 R53.82, Chronic fatigue unspecified), regardless of which criteria are used.

 

I don't have an opinion on prevalence one way or another but when people point out that a number is rather high and that real cases of ME are probably overstated because swept up with other illness isn't it just possible that real ME is also misdiagnosed at times as depression or some other somatoform illness thus making the numbers lower than they might have been?

 

It's so hard to get accurate info on prevalence when there are still legitimate controversies about how CFS should be diagnosed.

The male/female divide is less stark here than in some other studies, which is a bit interesting.

 

Havent seen the full paper yet so it will be interesting to see what limitations are listed.

But US doctors across the country diagnose CFS and only seldom diagnose ME. As Webdog notes, the ICD-10-CM (the US version of the ICD-10) currently classifies CFS in the symptom chapter where is uses the same code as chronic fatigue.

So based on codes, this could be drawing conclusions based on a cohort that includes both CFS and diagnoses unspecified chronic fatigue which would bias the numbers high. Similarly, ME uses the same code as postviral fatigue syndrome which would produce a cohort that combines PVFS and ME diagnoses. Will have to see what they do to separate out these different conditions in the medical records.

 

These numbers suggest about 500,000 patients in the UK (twice as many as normal).

I was reading another study this morning (can't remember the link at the moment, but I'll dig it out) that put the number at 280,000 (close to what we normally assume) and a third (by Crawley) that gave a 1% figure (so 600,000) for Oxford criteria CF.

Given that ICC/CCC select between 50-70% of CDC cases and Oxford is even broader, I would assume that the smaller number of ~250,000 in the UK is right, and that we should only double this number if we're including everyone with idiopathic chronic fatigue.

I think the higher rates of 4% or so reported in some trials can only be possible if you include fibromyalgia using the newest criteria (as these don't rely on tender points any more and use a points-scoring system that include a much wider range of symptoms).

 

We really need accurate numbers. It informs everything from advocacy to where research money can be directed. I don’t like the idea that people often claim “there are millions more undiagnosed” etc, as well. In the UK the oft used 250,000 figure has been the same for decades, which would put it as double as many as have MS. Someone really needs to do some work on this, particularly if it can be cited in the new NICE guidelines.

 

This really is a core issue. It affects the bottom line of healthcare, as noted in the abstract:

It's my hope that, at least in the US, healthcare companies that diagnose and treat ME/CFS will want more accurate prevalence figures in order to estimate costs.

Private healthcare providers might succeed where governments are failing. I've been exploring that possibility recently. It's the cost data that interests me the most in the abstract, and I look forward to the full paper.

 

It also seems that in order to get accurate prevalence figures, the uninsured have to be taken into account (which the study cannot do). That's in addition to dx'ing via specific criteria that have required (intensity/frequency/duration of) symptoms.....
(Sigh. We have huge knowledge gaps, too little funding, etc ----)

 

That was the rational behind the CDC "research" that phoned people up at random and found that some people with CFS did not even know they were ill!

 

Ughh - that study. Only 17% of the patients were unemployed due to their illness and only 8% of patients maintained a CFS diagnosis two years in a row.

So what were they studying?

 

Biomarkers. We need biomarkers.

 

Doubtful since private health care providers do not bear any of those added costs. ME patients are essentially free for them, unlike with a public insurance system where the public provider bears the same fiscal and economic consequences. HCPs have even less of an incentive to see any change, aside from losing a couple million clients who can't afford private insurance anyway.

Governments bear most of those costs, and they barely see most of those as they just get lost in noise since almost no useful data is gathered to make reliable accounting. It will take large-scale public health efforts to count, categorize and follow-up for years to get a full picture of the costs. At the rate things are going, this is likely to happen post-hoc, when people try to figure out why millions of disabled were just left with no medical care or support for decades.

Current estimates are plenty sufficient to make a case that this needs to happen, there's just no will to suspend disbelief yet, too much sunk cost into past denial and systemic discrimination.

 

You are correct. But my hope is that is changing in the US. Paradigm shifts in medicine take time.

 

Full text now available (pdf has tables, figures etc):
Estimating Prevalence, Demographics, and Costs of ME/CFS Using Large Scale Medical Claims Data and Machine Learning (Valdez et al.)

https://www.frontiersin.org/articles/10.3389/fped.2018.00412/full.

 

This is an important paper, with lots of data. I tried to tweet a summary of it:

1) An important paper on ME/CFS was published today. The authors looked at a large medical insurance database of 50 million people. This gave insight into enormous data about ME/CFS.

The paper is open access and can be read here: https://www.frontiersin.org/articles/10.3389/fped.2018.00412/full

2) The authors claim that ME/CFS is much more prevalent (“it is not a rare disease, but in fact a relatively common one”) and that relatively more men have the disease than previous thought (“ME/CFS is not a women’s disease.”).

3) However: the diagnosis of CFS based on insurance codes is problematic as it has been defined as “chronic fatigue, unspecified”. So doctors could have used this code for a CFS-like illness or simply unexplained chronic fatigue.

4) More interesting is the ICD code for ME/postviral fatigue syndrome. Here the prevalence was around 0,12% which may not look like much but it’s a lot considering how little physicians know how to make that diagnosis.

5) In the records, the authors found more than 14.000 individuals with this diagnosis.

Based on this data and machine learning the study calculated the number of ME-patients that are undiagnosed. Total prevalence was estimated at 0,85%, more than double of current estimates.

6) The female/male ratio for the 10.000 ME-diagnoses was also interesting. It was around 60%, much lower than previous estimates. And curiously in the youngest patients (0-9 years) boys outnumbered girls.

7) Finally there was also data for the average annual medical costs paid by insurance and the patient. For ME patients this cost was around 30.000 dollar per year, 50% higher than for MS or lupus patients in the database.​

 

Thanks for posting, @Michiel Tack

There are a number of useful analyses here. There are also a number of important caveats that need to be kept in mind, particularly because of issues with the ICD-10-CM codes that could account for the higher than expected prevalence rates or otherwise influence the findings.

In a nutshell, the issue is that as in ICD-10, the ICD-10-CM uses the code G93.3 for both ME and postviral fatigue syndrome. Unlike the ICD-10, the US has always used a different code for CFS - one in the symptoms chapter. And since 2015, the ICD-10-CM has used the same code, R53.82, for both CFS and the symptom of unspecified chronic fatigue.

So when G93.3 and R53.82 are used to select records for analyses, the resulting cohorts also include the poorly defined PVFS (in the "ME" cohort) and the symptom of chronic fatigue (in the "CFS" cohort). This could result in inflated prevalence estimates. And as we know, even aside from the ICD issue, a "CFS" diagnosis itself has sometimes been used as a wastebin by the medical community which could further compound the issue of overestimating prevalence. Both issues affect not only prevalence conclusions but could potentially impact interpretation of other findings as well.

As the authors note, a proposal was made in September to move CFS back to the neurological chapter and to provide separate codes for ME, CFS, PVFS, and also SEID. We dont know the results of that but if done, that could begin to improve the ability to use medical records for these kinds of retrospective analyses in the future. But even if that's accepted, its important that researchers carefully consider this issue because it will influence the use of medical records for a long time.

IMO, the paper further underscores the need for proper epidemiological research (including prevalence, gender, disease duration, levels of severity, etc) and how issues like the ICD-10-CM coding can complicate moving research forward.

 

This article is part of a special issue being put together for Frontiers in Pediatrics. These articles are being published one at a time as they become available but the full issue will be made available in print when all the articles are finally published. You can see the list of articles published so far is here:

https://www.frontiersin.org/research-topics/7718/advances-in-mecfs-research-and-clinical-care