• Welcome to Phoenix Rising!

    Created in 2008, Phoenix Rising is the largest and oldest forum dedicated to furthering the understanding of and finding treatments for complex chronic illnesses such as chronic fatigue syndrome (ME/CFS), fibromyalgia (FM), long COVID, postural orthostatic tachycardia syndrome (POTS), mast cell activation syndrome (MCAS), and allied diseases.

    To become a member, simply click the Register button at the top right.

Replication of an empirical approach to delineate the heterogeneity of chronic unexplained fatigue

Tom Kindlon

Senior Member
I'm not a fan of this paper but wanted to post the comments in message #2 (from 2009) somewhere

Free full text: http://www.pophealthmetrics.com/content/7//17

Replication of an empirical approach to delineate the heterogeneity of chronic unexplained fatigue.

Popul Health Metr. 2009 Oct 5;7:17. doi: 10.1186/1478-7954-7-17.

Aslakson E1, Vollmer-Conna U, Reeves WC, White PD.

Author information



Chronic fatigue syndrome (CFS) is defined by self-reported symptoms.

There are no diagnostic signs or laboratory markers, and the pathophysiology remains inchoate.

In part, difficulties identifying and replicating biomarkers and elucidating the pathophysiology reflect the heterogeneous nature of the syndromic illness CFS.

We conducted this analysis of people from defined metropolitan, urban, and rural populations to replicate our earlier empirical delineation of medically unexplained chronic fatigue and CFS into discrete endophenotypes. Both the earlier and current analyses utilized quantitative measures of functional impairment and symptoms as well as laboratory data.

This study and the earlier one enrolled participants from defined populations and measured the internal milieu, which differentiates them from studies of clinic referrals that examine only clinical phenotypes.


This analysis evaluated 386 women identified in a population-based survey of chronic fatigue and unwellness in metropolitan, urban, and rural populations of the state of Georgia, USA.

We used variables previously demonstrated to effectively delineate endophenotypes in an attempt to replicate identification of these endophenotypes.

Latent class analyses were used to derive the classes, and these were compared and contrasted to those described in the previous study based in Wichita, Kansas.


We identified five classes in the best fit analysis.

Participants in Class 1 (25%) were polysymptomatic, with sleep problems and depressed mood.

Class 2 (24%) was also polysymptomatic, with insomnia and depression, but participants were also obese with associated metabolic strain.

Class 3 (20%) had more selective symptoms but was equally obese with metabolic strain.

Class 4 (20%) and Class 5 (11%) consisted of nonfatigued, less symptomatic individuals, Class 4 being older and Class 5 younger.

The classes were generally validated by independent variables.

People with CFS fell equally into Classes 1 and 2.

Similarities to the Wichita findings included the same four main defining variables of obesity, sleep problems, depression, and the multiplicity of symptoms.

Four out of five classes were similar across both studies.


These data support the hypothesis that chronic medically unexplained fatigue is heterogeneous and can be delineated into discrete endophenotypes that can be replicated.

The data do not support the current perception that CFS represents a unique homogeneous disease and suggests broader criteria may be more explanatory.

This replication suggests that delineation of endophenotypes of CFS and associated ill health may be necessary in order to better understand etiology and provide more patient-focused treatments.

PMID: 19804639 [PubMed] PMCID: PMC2761845

Tom Kindlon

Senior Member
Some comments on this (originally written in 2009)
I thought I'd say that if one only reads some papers on CFS, I'd put this one into the bundle of ones of low priority to read!

They analysed the data of the women of the Georgia cohort including the healthy people and unwell people who didn't have CFS [so only 92 (23.8%) have CFS out of 386 and we are talking empirical definition (Reeves, 2005) so perhaps only 7-25 (out of 386!) "really" have CFS].

The main analysis came up with five classes (they did another analysis which came up with four classes and we are really not given much information on these four classes even though they are discussed now and then).

Nearly all of the CFS patients are in classes 1 (44%) or 2 (51%) (in the five class solution). 3% are in class 3, 1% are in class 4 and 0% are in class 5.

So if interested in heterogeneity with regard to CFS, you are interested in the differences between classes 1 and 2.

It is of course not an ideal way to look at heterogeneity with regard to CFS as the CFS patients only make up 41/97 (42.3%) of class 1 and 47/92 (51.1%) of class 2.

Here are the descriptions for classes 1 and 2 (remember that not all have CFS and this is the empirical definition) (from Table 7):

Class 1: polysymptomatic, depressed, insomnia, not obese
Class 2: polysymptomatic, depressed, insomnia, Obese, metabolic strain

Anyone notice the similarities and differences?

Similarities: polysymptomatic, depressed, insomnia

Differences: "not obese" vs "Obese, metabolic strain"

Here is another set of descriptions from the text: "Class 1 (25%) captured ill subjects with many symptoms, prominent fatigue, sleep problems, and depression, but no aberrant biological markers (including body mass index) characterized this group. Class 2 (24%) similarly captured ill subjects who reported prominent, widespread symptoms, insomnia, and depression. However, these subjects had an associated metabolic syndrome (elevated insulin and inflammatory markers) and were obese."

One can see from the pre-publication comments that in the original draft the authors said: "The replication of heterogeneity found in an independent population derived sample provides the strongest support yet published of the heterogeneous nature of CFS."

Not surprisingly, a reviewer told them to tone this down.

I might as well make one point here even though I also intend to make it as a comment:

They keep saying that it shows that the research shows that depression needs to be used to stratify CFS patients. But the two classes with virtually all the CFS patients (96%) had comparable levels of depression!

The authors make all sorts of other comments about what their study shows about CFS. But really it doesn't show very much as there are virtually no CFS patients in Classes 3-5 and around half the people in classes 1 and 2 don't have CFS. And as I say, Classes 1 and 2 aren't that different.

And of course CFS is defined using the empirical definition so lots wouldn't normally be said to have CFS at all!


Places I'd rather be.
At first it didn't make sense that there is no class for polysymptomatic, insomnia, not obese, not depressed. The SCID is mentioned, but it is not clear how they defined "depressed mood", it is possible that physical symptoms counted towards "depressed mood". However, it made perfect sense once realizing that it used the CDC's "empirical" definition of CFS, which was notorious for increasing the prevalence estimates by 10-fold, including or conflating with primary major depression rather than primary CFS, and allowing emotional problems to fulfill the limitations requirement without physical problems?
Last edited: