PACE Trial and PACE Trial Protocol

[Graham]

You are so right, Sean.

I have been playing around with reverting the Fatigue Scale back to the original bimodal scoring. If I have done the sums right, the baseline score of 28 would mean that the majority of patients would be at the bottom of the scale on 11. The SMC group improved to the continuous score of 24, which, depending on where the improvements were, could still be a score of 11 - say on the borderline of 10 and 11, and adding on GET would probably mean that the majority of scores were 10 and 11. No wonder they changed the measurement system. Just imagine the headlines - a year of GET+SMC and most patients may improve their fatigue score from 11 to 10, but equally, they may not.

 

[Dolphin]

You are so right, Sean.

I have been playing around with reverting the Fatigue Scale back to the original bimodal scoring. If I have done the sums right, the baseline score of 28 would mean that the majority of patients would be at the bottom of the scale on 11. The SMC group improved to the continuous score of 24, which, depending on where the improvements were, could still be a score of 11 - say on the borderline of 10 and 11, and adding on GET would probably mean that the majority of scores were 10 and 11. No wonder they changed the measurement system. Just imagine the headlines - a year of GET+SMC and most patients may improve their fatigue score from 11 to 10, but equally, they may not.

Good point. We won't get to see the numbers I guess but it doesn't seem right that they can change the scoring scheme like that after they have published the protocol, especially when the only validated definition of fatigue caseness (work done by Trudie Chalder amongst others) is in the bimodal scoring (>3) (which might have been one of the main things they were "running away from").

 

[Graham]

Hi all: I have been trying to find the reference to the statement that 1 in 7 or 1 in 8 showed clinically significant(?) improvements, but with 107 pages and so much repetition of these terms, I'm getting nowhere. Can anyone easily direct me to the right page please?

 

[oceanblue]

Hi all: I have been trying to find the reference to the statement that 1 in 7 or 1 in 8 showed clinically significant(?) improvements, but with 107 pages and so much repetition of these terms, I'm getting nowhere. Can anyone easily direct me to the right page please?

Surprised you couldnt find this given how I've seemingly banged on about it at every oppoortunity over the last couple of months , but for the record:

Just been reading the transcript of the ABC interview with Michael Sharpe (& Richard Horton) and was suprised by what he said about the trial results...

Nb this is based on the %age of improvers (improved by "clinically useful difference" in both fatigue and SF36). More info on calculating Number Needed to TReat, NNT.

 

[Graham]

Thanks! I couldn't find it because I forgot that it was in the transcript of the interview with Michael Sharpe. But that wouldn't come as a surprise to anyone who knows what my memory is like. I think I need CBT - Complete Brain Transplant, although perhaps GET, Generally Everything Transplanted, may be a safer option.

signed

Frank N Stein

(apologies for that one - but I'm an old man, so don't complain!)

 

[Dolphin]

I was just reading this paper:

Ciccone DS, Chandler HK, Natelson BH. Illness trajectories in the chronic fatigue syndrome: a longitudinal study of improvers versus non-improvers. J Nerv Ment Dis. 2010 Jul;198(7):486-93.

I thought some people might find the following of interest:

In agreement with previous longitudinal studies of CFS, the
majority of women in this tertiary sample reported improvement in
physical functioning over time (67%). The magnitude of the effect
was clinically meaningful (41.1% over baseline) but still far short of
recovery. According to SF-36 national norms compiled by Ware et
al. (1993), a typical woman aged between 35 and 44 might expect to
score 95 out of 100 on the PF scale (corresponding to the 50th
percentile). None of the present participants achieved this level of
physical function. However, 5 did achieve scores of 85 or higher on
PF corresponding to the 25th percentile (5.3% of the study sample).

Ware JE, Snow KK, Kosinski M, Gandek B (1993) SF-36 Health Survey Manual
and Interpretation Guide. Boston (MA): New England Medical Center, The
Health Institute.

The mean age at the end of the PACE Trial was 40 y.o. I doubt we'll ever get to see how many got to 95 on the PF SF-36 (people will recall the highest level we were given figures were was 60 although the future recovery paper may give a higher figure). The Knoop et al. (incl. Peter White) full recovery paper (1997) used a SF-36 threshold of >=80 for "no physical disability" and >=80 plus some other criteria for "full recovery".

------
In the other paper that I read tonight:

Johnson, S. K., Gil-Rivas, V. and Schmaling, K. B. (2008), Coping strategies in chronic fatigue syndrome: outcomes over time. Stress and Health, 24: 305312. doi: 10.1002/smi.1185

Functional status. The Medical Outcomes
Study Short Form-36 (SF-36) (Ware, 1993) was
used to evaluate functional status. This 36-item
questionnaire consists of eight scales that reflect
physical functioning, physical role functioning,
emotional role functioning, social functioning,
mental health, bodily pain, vitality and health
perceptions. Higher scores indicate better functioning (range, 0100), with the normal range
considered to be 80 and above. We retained three
scores for analysis: physical functioning, vitality,
and physical role functioning, because of these
scores demonstrated sensitivity to change over
time (Komaroff et al., 1996), and use in previous
studies (Ciccone et al., 2003; Schmaling et al.,
2003) examining predictors of outcome in CFS.

While in the PACE Trial, if you scored 60 or more, you were considered in the normal range.
[Aside: in Johnson et al, the scores on vitality and role physical were lower than the physical functioning scores]

 

[oceanblue]

Thanks, Dolphin. I'm sure this stuff will be very useful when the PACE 'recovery' paper finally comes out.

 

[Dolphin]

Gijs Bleijenberg's changing views of SF-36 PF scores of 60-65

For my sins, I have just been reading:

van't Leven M, Zielhuis GA, van der Meer JW, Verbeek AL, Bleijenberg G. Fatigue and chronic fatigue syndrome-like complaints in the general population. Eur J Public Health. 2010 Jun;20(3):251-7. Epub 2009 Aug 18.

Free full text at: http://eurpub.oxfordjournals.org/content/20/3/251.long or http://eurpub.oxfordjournals.org/content/20/3/251.full.pdf+html

Rand 36 Physical functioning (Cronbach's-?: 0.92):28,29 A 10-item subscale, transformed to a 0100 score, where higher scores mean better physical function. A cut-off of ?65 was considered to reflect severe problems with physical functioning.

Contrast with PACE Editorial:

Graded exercise therapy and cognitive behaviour
therapy might assume that recovery from chronic fatigue
syndrome is possible, but have patients recovered after
treatment? The answer depends on ones definition of
recovery.3 PACE used a strict criterion for recovery: a score
on both fatigue and physical function within the range
of the mean plus (or minus) one standard deviation of a
healthy persons score. (this was SF-36 PF of 60+ and Chalder Fatigue Questionnaire score of 18 or less)

Lancet. 2011 Mar 5;377(9768):786-8. Epub 2011 Feb 18.
Chronic fatigue syndrome: where to PACE from here?
Bleijenberg G, Knoop H.

 

[Esther12]

lol - nice find D.

 

[Andrew]

Dolphin, what are you trying to say with that information?

 

[Valentijn]

Dolphin, what are you trying to say with that information?

I'd guess the point is that one CFS researcher that has previously defined <= 65 as severely physically disabled, is now saying 60 is recovered in reference to the PACE trial. It's self-contradictory.

 

[oceanblue]

I'd guess the point is that one CFS researcher that has previously defined <= 65 as severely physically disabled, is now saying 60 is recovered in reference to the PACE trial. It's self-contradictory.

It is just possible that Bleijenberg & Knoop (Bloop?) made a mistake. They wrote:

PACE used a strict criterion for recovery: a score
on both fatigue and physical function within the range
of the mean plus (or minus) one standard deviation of a
healthy persons score.

This was the definition used in that Knoop 2007 recovery paper (with PDW...), nb they used SF36 data for a healthy population in their study (which is arguably a strict criterion) and came up with a threshold of 80 (vs 60 in PACE). Possibly they were merely being incompetent in the editorial and hadn't read PACE properly to discover the PACE approach was different to theirs. The editorial authors may have completely changed their own definition of recovery but I think incompetence is possible too (don't know how long they had to read the paper and write the editorial).

The PACE authors, however, can have no excuse for failing to point out the error in the editorial: they knew exactly what was going on.

 

[Snow Leopard]

Bleijenberg & Knoop are very familiar with the SF36 and knew full well that they were making misleading statements.

By the way, a certain letter also came up with a figure of ~80 based on the study cited in the PACE paper - the figure that excluded those reporting chronic illnesses.

 

[Dolphin]

I'd guess the point is that one CFS researcher that has previously defined <= 65 as severely physically disabled, is now saying 60 is recovered in reference to the PACE trial. It's self-contradictory.

Thanks Valentijn. And I think Bleijenberg & Knoop would know that scale well.

 

[oceanblue]

Bleijenberg & Knoop are very familiar with the SF36 and knew full well that they were making misleading statements.

By the way, a certain letter also came up with a figure of ~80 based on the study cited in the PACE paper - the figure that excluded those reporting chronic illnesses.

So when they wrote, erroneously

PACE used a strict criterion for recovery: a score on both fatigue and physical function within the range of the mean plus (or minus) one standard deviation of a healthy persons score.

do you think they were deliberately lying? PACE makes no mention of 'healthy'. I'm not so convinced B&K set out to deliiberately mislead though they clearly got it wrong.

 

[anciendaze]

If everything I just did got past various cognitive deficits, the normal distribution assumed in the earlier work has a mean of 95 and an SD of less than 15. This would apply to a healthy population where mean and mode are both around 95. It might be handy to plot that normal distribution against the histogram from Chalder and the normal distribution used in PACE.

 

[Dolphin]

PACE Trial: Freedom of Information response to MEA request for further information

Response to one FOI request:

http://www.meassociation.org.uk/?p=6668

PACE Trial: Freedom of Information response to MEA request for further information

On May 24th a joint letter was sent to the organisers of the PACE trial requesting further information on a number of aspects relating to how the trial was conducted, analysed and reported.

The text of the joint letter is here:

www.meassociation.org.uk/?p=6171

The attached Freedom of Information response has now been received from Queen Mary College. Please click HERE http://www.meassociation.org.uk/wp-content/uploads/2011/06/FOI+from+Queen+Mary.pdf to read it.

 

[Snow Leopard]

do you think they were deliberately lying? PACE makes no mention of 'healthy'. I'm not so convinced B&K set out to deliberately mislead though they clearly got it wrong.

The whole normal=60 idea was distinct twisting of the facts given that <65 = severe enough to participate and <70 fullfills CDC empiric criteria and the mean/sd figures they based it on explicitly included those with chronic diseases. B&K were fully aware of this.

But it is also likely that if the PACE trial reported recovery figures as originally specified, B&K probably wouldn't have been able to claim those PF>60 scores as recovered.

Maybe when they say "normal", they mean normal people with chronic fatigue (but not the syndrome).
Seriously, have a look at the mean PF scores of this Dutch study, page 15:
http://dare.uva.nl/document/203737

 

[anciendaze]

If anyone follows my suggestion above about plotting the latest normal distribution, and comparing it to the one used in PACE, there is another nice illustration to consider.

The distribution for the healthy population is so far from the middle of the total population distribution, and so different in size, you can get a fair estimate of what the unhealthy population distribution would look like by subtracting this from the tabulated values shown by the histogram. You can assume anything from zero overlap of healthy and unhealthy to complete overlap. You might try several models. I think models with almost no overlap work pretty well, which contradicts the fundamental assumption behind the psychosocial model used in PACE.

One inference from all this is that treating the total population as the sum of healthy and unhealthy works well. This implies a minimum of four parameters to define that distribution.

A second inference is that any particular estimate of a single standard deviation for the total distribution is unlikely to be reliable. If several substantially different values work about equally well, no single one can be trusted. This value works its way into almost every aspect of PACE.

 

[oceanblue]

The whole normal=60 idea was distinct twisting of the facts given that <65 = severe enough to participate and <70 fullfills CDC empiric criteria and the mean/sd figures they based it on explicitly included those with chronic diseases. B&K were fully aware of this.

But it is also likely that if the PACE trial reported recovery figures as originally specified, B&K probably wouldn't have been able to claim those PF>60 scores as recovered.

Maybe when they say "normal", they mean normal people with chronic fatigue (but not the syndrome).
Seriously, have a look at the mean PF scores of this Dutch study, page 15:
http://dare.uva.nl/document/203737

Hi Snow Leopard

I think we may be at cross purposes here. I know the 'normal' thresholds are crazy (my unpublished Lancet letter was about just this). My point was just that maybe K&B didn't check the PF thresholds (because they were in a hurry), saw 30% were normal and went with that as 'recovery'. the alternative is that they knew exactly what they were doing and just lied outright, but I don't think that's so likely because it's almost inevitably going to end with a correction, as happens here. I think they just erroneously assumed PACE had used the same critieria as the original Knoop study.

Nb the PACE authors went down a very different track. They went out of their way to mislead and were very careful never to use the 'recovery' word - but never actually lied (OK, apart from saying the SF36 was all adults). I'm just going for the cock-up over conspiracy interpretation for Knoop & Bleijenberg. The study went through review v quickly and they may not have had very long to write the editorial (maybe the PACE authors even implied to them that normal=recovery).

Anyway, we'll never know for sure and I guess it doesn't really matter.