Measuring fatigue in clinical and community settings.

[Dolphin]

Probably not the most exciting research for most people. But the Chalder fatigue scale is used in a lot of CFS research including as one of the two primary outcome measures in the PACE Trial.

Measuring fatigue in clinical and community settings.

J Psychosom Res. 2010 Jul;69(1):17-22. Epub 2009 Dec 11.

Cella M, Chalder T.

Institute of Psychiatry, King's College London, London, United Kingdom. matteo.cella@kcl.ac.uk

Abstract

OBJECTIVE: The Chalder Fatigue Scale (CFQ) is a widely used instrument to assess fatigue in both clinical and nonclinical settings. Psychometric properties of the scale and discriminative abilities were examined.

METHODS: A total of 361 patients with CFS and 1615 individuals in the community were assessed with the CFQ. Principal component analysis (PCA) was used to explore the structure of the scale. Receiver-operating characteristic curve (ROC) was used to investigate the discriminative properties.

RESULTS: Two components, physical and mental fatigue, were identified in the CFS patient group and in the general population samples. Area under the curve for ROC was .91. The fatigue scale effectively discriminates, at high scores, between CFS patients and the general population.

CONCLUSION: Physical and mental fatigue are clearly separable components of fatigue. The CFQ can discriminate reliably between clinical and nonclinical conditions.

 

[Dolphin]

The main reason I'm posting this is that I noticed something about this paper which I appeared to be a dubious claim. In the end, I was too busy to letter to the editor. Here's what I wrote to somebody at the time when I was wondering whether to write a letter or not. I used the 2.5% prevalence figure because of the following:

In order to be included in this study, patients had to be 18 or
older and meet either the Oxford [16] or the Fukuda et al.
[17] diagnostic criteria for CFS.

I dont know if it is worth writing a letter in about this.
Their claim that A score of 29 on the fatigue scale can
discriminate with more than 95% accuracy; that is, an
individual presenting with a score of 29 has less than a 5%
chance of not having CFS is incorrect.

Take the 96% figure they use.

In the first example, assume 2.5% of the population have CFS and the rest dont. Take a large sample like 1,000,000 so 25,000 have CFS and 975,000 dont.

Then the number of CFS cases that have a score of 29 and over is 25,000*0.202=5050.
The number of non-CFS cases that have a score of 29 and over is 975,000*0.004=3900.
So the probability that somebody who has a score of 29 and over has CFS is 5050/(5050+3900)=0.5642 or 56.42% of the cases.

However, one could say that not all the population are healthy. What proportion of the population would need to be healthy for this to be true.
Let x=the percentage of the population who are healthy.
So in 1 million people, the number who are healthy is 1,000,000*(x/100)=10,000x
The number of healthy people who have a score of 29 and greater is 10,000x*.004=40x
As calculated earlier, the number of CFS cases that have a score of 29 and over=5050.

Using the 96% figure, the number of CFS cases who have scores of 29 and greater is 24 times the number of healthy people.
So 5050=24(40x) or x=5.26.
So this would only work if 5.26% of the population are distributed like the control sample.

I just used 1,000,000 for illustrative purposes but one can let just use a constant, N, and the same result holds true.

 

[Enid]

It is unbelievable that measuring fatigue by psychiatrists sheds light on the pathology of ME or indeed aids.

 

[oceanblue]

I'm not sure that discriminating between healthy people and those with ME is a particularly impressive acheivement. How could a fatigue scale not discriminate between those with Chronic Fatigue Syndrome and healthy individuals?? Surely a better test would be the ability to distinguish between CFS and people with an illness that is only moderately fatiguing, such as arthritis. Lenny Jason's work here is interesting looking at the specifc nature of fatigue eg post-exertional fatigue, and the ability of such measures to discriminate between ME and other types of ill health.

 

[Dolphin]

This study was used to define normal fatigue for the PACE Trial paper in the Lancet (main one) and the recovery paper.
They used a sample of people who had attended their GP in the last year:

I happened to see the following on Twitter today:

Irish Dental Associ ‏@IrishDentists
#IDA2013 UK study of male adults shows only 15% visited their medical GP within past two years; how should dentists respond? #IDA2013

This would suggest the sample is likely not representative at all.

 

[Simon]

This study was used to define normal fatigue for the PACE Trial paper in the Lancet (main one) and the recovery paper.
They used a sample of people who had attended their GP in the last year:

I happened to see the following on Twitter today:

UK study of male adults shows only 15% visited their medical GP within past two years

This would suggest the sample is likely not representative at all.

I wonder what their source is? This huge study [p16] looking at GP practices with 1 million patients in 95/6, close to the time the Cella data was collected, gives an average of 2 consultations a year for men aged 18-40 (its much higher for older men). Unless those 15% of men are visiting their GP more than once a month the figures don't tally.

A 2009 poll of 3,000 men for a prostate cancer charity reported:

39 per cent of British men hadn’t been to the doctor at all in the last year which is almost twice the number of women at 22 per cent. The poll showed that the average number of visits to a doctor for a woman is 2.3 a year compared with 1.9 for men.

Those average number of visits are well short of those found by the much larger study - I'm guessing that most people's recall of the number of times they visited their surgery in the last year might not be that accurate.

I suspect a bigger bias is that, statistically, those who go to see their doctor many times a year (and so are likely to be sicker) were much more likely to be included the fatigue sample than those who only visit their doctor once a year.

 

[Dolphin]

Thanks for the extra info, Simon. It certainly makes me suspect the 15% figure.

I suspect a bigger bias is that, statistically, those who go to see their doctor many times a year (and so are likely to be sicker) were much more likely to be included the fatigue sample than those who only visit their doctor once a year.

Do we have reason to think this? I thought it was simply sent by post to people who had attended in the previous twelve months.

 

[Esther12]

The 15% figure seems low to me too... but it could be that the consensus figures (around 50%+?) are unduly high because healthy young men who move around and don't see their doctors are also the sort of people likely to be harder to poll.

Getting reliable figures for population norms seems really difficult.

 

[Simon]

Thanks for the extra info, Simon. It certainly makes me suspect the 15% figure.

I suspect a bigger bias is that, statistically, those who go to see their doctor many times a year (and so are likely to be sicker) were much more likely to be included the fatigue sample than those who only visit their doctor once a year.

Do we have reason to think this? I thought it was simply sent by post to people who had attended in the previous twelve months.

Yes:
-In stage 1 the survey was mailed to many thousands of registered patients
-In stage 2 they took 1,000 patients attending with a suspected viral illness, together with the next non-viral patient after each viral patient (another 1,000). Obviously, the more often you went to the doctor, the more likely you were to be in either cohort.
- the CFQ scores used in the study were the stage 1 survey scores of those who were included in stage 2

More here

 

[Dolphin]

Yes:
-In stage 1 the survey was mailed to many thousands of registered patients
-In stage 2 they took 1,000 patients attending with a suspected viral illness, together with the next non-viral patient after each viral patient (another 1,000). Obviously, the more often you went to the doctor, the more likely you were to be in either cohort.
- the CFQ scores used in the study were the stage 1 survey scores of those who were included in stage 2

More here

Thanks. I recalled the thread but not the details. I should probably have looked back at that before replying to the latest PACE Trial paper to help with the criticism of the use of <=18 as a threshold for recovery.