Prevalence, Incidence, and Classification of Chronic Fatigue Syndrome in Olmsted County (CDC study)

[Dolphin]

Free full text: http://www.mayoclinicproceedings.org/article/S0025-6196(12)00923-8/fulltext


This study was part funded by the CDC CFS program and involves many of the team there.

Prevalence, Incidence, and Classification of Chronic Fatigue Syndrome in Olmsted County, Minnesota, as Estimated Using the Rochester Epidemiology Project

• Ann Vincent, MD
  ,
• Dana J. Brimmer, PhD, MPH
  ,
• Mary O. Whipple, BA
  ,
• James F. Jones, MD
  ,
• Roumiana Boneva, MD, PhD
  ,
• Brian D. Lahr, MS
  ,
• Elizabeth Maloney, DrPH, MS
  ,
• Jennifer L. St. Sauver, PhD
  ,
• William C. Reeves, MD, MS†

published online 12 November 2012.
Corrected Proof
Abstract

Objective

To estimate the prevalence and incidence of chronic fatigue syndrome in Olmsted County, Minnesota, using the 1994 case definition and describe exclusionary and comorbid conditions observed in patients who presented for evaluation of long-standing fatigue.

Patients and Methods

We conducted a retrospective medical record review of potential cases of chronic fatigue syndrome identified from January 1, 1998, through December 31, 2002, using the Rochester Epidemiology Project, a population-based database. Patients were classified as having chronic fatigue syndrome if the medical record review documented fatigue of 6 months' duration, at least 4 of 8 chronic fatigue syndrome–defining symptoms, and symptoms that interfered with daily work or activities. Patients not meeting all of the criteria were classified as having insufficient/idiopathic fatigue.

Results

We identified 686 potential patients with chronic fatigue, 2 of whom declined consent for medical record review. Of the remaining 684 patients, 151 (22%) met criteria for chronic fatigue syndrome or insufficient/idiopathic fatigue. The overall prevalence and incidence of chronic fatigue syndrome and insufficient/idiopathic fatigue were 71.34 per 100,000 persons and 13.16 per 100,000 person-years vs 73.70 per 100,000 persons and 13.58 per 100,000 person-years, respectively. The potential cases included 482 patients (70%) who had an exclusionary condition, and almost half the patients who met either criterion had at least one nonexclusionary comorbid condition.

Conclusion

The incidence and prevalence of chronic fatigue syndrome and insufficient/idiopathic fatigue are relatively low in Olmsted County. Careful clinical evaluation to identify whether fatigue could be attributed to exclusionary or comorbid conditions rather than chronic fatigue syndrome itself will ensure appropriate assessment for patients without chronic fatigue syndrome.
Abbreviations and Acronyms: CFS, chronic fatigue syndrome, H-ICDA, H-ICDA: Hospital Adaptation of ICDA, ICD-9, International Classification of Diseases, Ninth Revision, ISF, insufficient/idiopathic fatigue, REP, Rochester Epidemiology Project

 

[Dolphin]

I've various observations on the full text I'll add here.

The paper doesn't clearly indicate how different this prevalence figure to the last CDC prevalence study (Reeves et al., Georgia, 2007, 2.54%). The prevalence in this study is less than 3% of that figure.

It talks about different estimates for the prevalence but doesn't indicate that some of these used the Oxford criteria.
I'm not sure any Fukuda studies gave prevalences of over 1% (or perhaps there was some small sample, looking at GP attendees, which gave over 1% but that's an artificial subset). The prevalence of 2.54% from the Georgia study really stands out in terms of how big it is.
The Georgia study claimed to used the Fukuda criteria: the empiric criteria supposedly operationalise the Fukuda criteria - but it is done in a really weird way.
--------

The 4 most common symptoms in the CFS group were muscle pain (74 patients [97%]), nonrestorative sleep (69 patients [91%]), multijoint pain (69 patients [91%]), and headaches (58 patients [76%]). Postexertional malaise was only documented in the medical records of 27 patients (36%).

They do mention this % may be artificially low:

Additionally, the low percentage of patients with postexertional malaise in our study may reflect the lack of inquiry about this important symptom in routine clinical practice.

However, they don't explain that this might mean that some patients would not meet the threshold for 4 out of 8 symptoms, reducing the prevalence.

I'm not sure if they explicitly explain about the problems in general with this sort of retrospective analysis. Patients are not necessarily going to mention every symptom they have so again other CFS cases are probably being missed.

It is the case that the ISF group in this study isn't very big either so even if all of them were added in it would only bring the prevalence up to 145 per 100,000.
However, patients may not be explicitly mentioning fatigue - I was more focused on muscle pain when I was more mildly affected: that was the thing that was preventing me doing a lot of things I wanted to do.

-----

TABLE 4. Demographic Characteristics, Comorbid Conditions, and Clinical Findings in Patients With CFS and ISFa,b

I would have found this a lot more interesting if they had given data for matched controls, or even some sort of background data. Surely the more interesting point is how these relate to the general population?

For example, in the results section, they said the CFS group are:

well educated (mean [SD], 14.3 [2.3] years of education),

However, in the discussion they say:

Our demographic results regarding educational achievement are different from previous reports in which CFS was observed in patients with lower levels of education.10,18 Our results may reflect the higher educational attainment of Minnesota’s population in general.24 Future studies focused on samples of patients with varying educational and socioeconomic status may be useful in further clarifying this issue.

 

[Dolphin]

Further points:
A minor point is "Reasons for CFS Exclusion" - I'm very curious about this group

Other (eg, morbid obesity, polymyalgia rheumatica) 64 (13)

- it would be interesting to have a bigger list of missed diagnoses there are similar to CFS, perhaps containing more interesting/unusual diagnoses.

---

We used ICD-9 and H-ICDA codes from the REP in conjunction with medical record review to classify patients with CFS according to international research criteria. 1 The codes accurately identified 66% of patients with CFS, which reflects a high level of knowledge and awareness of CFS among health care professionals in Olmsted County

I find this statement strange - there are a variety of codes mentioned.

Also, this ignores any cases that doctors missed completely (i.e. there wasn't enough information to pick up some CFS cases from the records).

---

In addition, almost half the patients with CFS or ISF in our current study had at least one nonexclusionary comorbid condition. Particularly notable is the high percentage of somatic disorders, insomnia, and comorbid psychiatric conditions, which suggests physiologic dysregulations that may overlap in these disorders. We also found that psychiatric comorbid conditions (particularly depression and anxiety) were common in patients with both CFS and ISF. The presence of comorbid conditions in CFS and ISF is an important factor for clinicians to consider as they develop and revise management plans. For example, a patient meeting CFS criteria may also have a sleep disorder as a comorbid condition. The sleep disorder requires proper treatment and management wholly apart from the management of CFS and should be routinely evaluated.

Although I'm not 100% sure what they're saying in the underlined part, it looks like they might be saying that because CFS has shared symptoms with other conditions, the aetiology/pathophysiology may be related. If so, this may be a "bad" sign in terms of the authors views given the mention of psychiatric disorders.

---

The current study identifies questions that should be addressed, such as whether comorbid illnesses contribute to CFS or whether the symptoms and consequences of CFS and ISF simply allow identification of patients whose underlying disease is not yet manifest. Additionally, the sensitivity and specificity of the methodologies we used could have resulted in our prevalence estimates. Careful initial identification of CFS and ISF with use of uniform criteria is essential, and case identification could serve as an important stepping stone to understanding the origins of these syndromic illnesses as well as a possible beacon leading to underlying diagnoses.

It seems they may be suggesting that CFS is simply a collection of missed diagnoses rather than a thing by itself?

Clinical record review offers advantages over population surveys, which are considerably more costly and can be subject to both selection and participation bias. However, since population surveys can more accurately estimate the population burden of illnesses like CFS because a large proportion of persons with such illness do not seek medical care,26 this may be a limitation of our study.

This is at the very end. I don't know if it was a suggestion from a reviewer but the wording ("may be", etc.) is odd.

 

[Kati]

The other variable worth mentioning is that physicians do not necessarily record all the symptom complaints from the patients. After reading my own GP office notes, I was apalled at how little she has been writing all along despite me describing my symptoms each and every time I would go in.

Edit to add: Physicians also tend to structure their notes around the diagnosis they have in mind, sometimes leaving vague symptoms behind because it does not describe the diagnosis.

 

[WillowJ]

another point about anxiety and depression, is that this may represent physician bias about patients who are difficult to immediately classify and who have fatigue or pain or other symptoms (especially many symptoms), rather than actual anxiety and depression on the parts of all patients for whom that was charted.

clearly some of us will have those things.

in other cases it represents the doctor not knowing what is going on so they chart an equivalent of "not my problem" which is considered professionally acceptable,

or non-pathological emotions are misperceived (I cannot get any help and I may be frustrated or discouraged about that, or I may be worried whether I will be understood or helped, is often interpreted as pathological depression and/or anxiety rather than a realistic response to life circumstances) (I being, any patient)

 

[Dolphin]

another point about anxiety and depression, is that this may represent physician bias about patients who are difficult to immediately classify and who have fatigue or pain or other symptoms (especially many symptoms), rather than actual anxiety and depression on the parts of all patients for whom that was charted.

clearly some of us will have those things.

in other cases it represents the doctor not knowing what is going on so they chart an equivalent of "not my problem" which is considered professionally acceptable,

or non-pathological emotions are misperceived (I cannot get any help and I may be frustrated or discouraged about that, or I may be worried whether I will be understood or helped, is often interpreted as pathological depression and/or anxiety rather than a realistic response to life circumstances) (I being, any patient)

Yes, on a personal note, I remember before being diagnosed, crying when I was in with one of the college GP's surgery (you could get free appointments on campus so I tended to go to them than the family GP). I actually had a very stable mood on a day-to-day basis: I had read information on positive mental health and did things like doing an thing you enjoy thing every day, etc. I probably hadn't cried once anywhere in the 12 months before this (and wasn't on antidepressants or anything). Yet, crying once could get depression written on your notes (not sure if it happened in my case).

 

[Dolphin]

another point about anxiety and depression, is that this may represent physician bias about patients who are difficult to immediately classify and who have fatigue or pain or other symptoms (especially many symptoms), rather than actual anxiety and depression on the parts of all patients for whom that was charted.

clearly some of us will have those things.

in other cases it represents the doctor not knowing what is going on so they chart an equivalent of "not my problem" which is considered professionally acceptable,

or non-pathological emotions are misperceived (I cannot get any help and I may be frustrated or discouraged about that, or I may be worried whether I will be understood or helped, is often interpreted as pathological depression and/or anxiety rather than a realistic response to life circumstances) (I being, any patient)

Regarding this, they claim:

One explanation for the low incidence is identification of patients who fulfill diagnostic criteria for a syndrome but who have an underlying exclusionary disease process that is readily identifiable or that is evolving. Such an outcome may be attributed to the current study design in which patients have undergone thorough medical and psychiatric evaluations.

I may have missed it but I don't recall any evidence the patients had thorough psychiatric evaluations. In prevalence studies, each patient with got proper psychiatric assessments with long detailed interviews. I didn't see any evidence of this. This makes me suspicious of the psychiatric rates in the CFS cases quoted. Alternatively maybe those were the patients where GPs note symptoms like fatigue, etc while lots of other non-depressed cases don't make it on the radar.

 

[Snow Leopard]

They were pretty heavy handed on the exclusions, eg: Pregnancy within 12 mo before CFS diagnosis. Eg this removed 29 patients, which reduced the total number from 105 to 76. Along with Restless legs syndrome/periodic limb movement disorder (another 58!).

Including these alone would boost the prevalence (in this study) to 0.15%

It is noteworthy that 70% of potential patients were excluded.

 

[Dolphin]

What I would find interesting would be long-term data from this cohort: seeing how people did over 5/10/20 years. That's an area where there's a real lack of data in the ME/CFS field. There was no mention of such a study in the paper (they did use data from 1998-2002).

 

[Simon]

I'm mystified as to the point of this Reeves legacy study: they fail to follow Fukuda diagnosis procedures so how can they measure Fukuda prevalence? Specifically, they merely trawl through medical records while Fukuda requires eg full medical & psychiatric examination, and specific lab tests. Game over.

Even if it wasn't, if Olmsted county, like the rest of the US, doesn't have universal healthcare, then many CFS cases may be excluded simply because they can't afford access to doctors.

I'm hoping that CDC studies in the post-Reeves era will be more altogether useful than this one.