PACE Trial and PACE Trial Protocol

[Dolphin]

Changes made to PACE trial

(Haven't looked at these myself)

From Co-Cure:
https://listserv.nodak.edu/cgi-bin/wa.exe?A2=ind1107b&L=co-cure&T=0&P=58

An FOI was made requesting any changes between the PACE trial
protocol, Trial Identifier, etc. and the published results, as well as
the justifications given for these changes. I haven't had time to go
through all of the docs, so if anyone notices any changes that were
made to the trial but were not requested from and/or approved by the
ethics committee, such as the changing of reporting adverse events
from a drop of 20 points in one interview to a 20 point drop in two
consecutive interviews (I didn't see this one mentioned or explained
by the authors in the paper), please let me know. Also, from the file
'Appendix5_UMC_Guidance'-

"Treatments to be avoided
Whilst no treatment is prohibited from medical care we would ask those
providing this to restrict their care to that usually given and not to
either seek to provide or refer patients for treatment which is being
evaluated in another arm of the trial. As part of the trial we would
ask you not to prescribe any medicine that is regarded as a putative
specific treatment for CFS. Excluded medicines would involve hormones,
such as hydrocortisone, but would not include indicated endocrine
treatments, such as thyroxine in a patient with corroborated and
compensated hypothyroidism. Putative immune treatments for CFS (e.g.
Ampligen, immunoglobulin, inosine pranobex) would be excluded.
Putative energy enhancing treatments (e.g. NADH) would be excluded."

How nice! 'Specialist treatment' minus the treatment! Brilliant!

zip file of changes made to the PACE trial-
http://www.mediafire.com/?uihi8zzfzihviux

Saw the following posted elsewhere today - presume it's very similar:

File: FoI PACE amendments etc.zip

Size: 14.83 MB

Expires: Content will be available for download until July 18, 2011 15:22 PDT

Here's the link to this file:

http://www.yousendit.com/download/MFo0NHA2V3J6NEt4dnc9PQ

 

[Dolphin]

Perhaps just a technical point:

http://www.meassociation.org.uk/wp-content/uploads/2011/06/FOI+from+Queen+Mary.pdf

A symptom was counted as present if scored as "more often than not", or more frequently than that.

We are not given this meaning in the main text. So when people look at the reductions in the presentages with symptoms, readers who just had the main text (virtually everyone) might think that the symptom had disappeared.

The options were:

Not at all present
Present a little
Present more often than not
Present most of the time
Present all the time

 

[biophile]

I never understood why this particular editorial featured in the above YouTube video claimed that Hooper's criticism was "now available via Wikipedia". ...

I emailed an editor at The Lancet about this issue and received a reply. Basically, "Wikipedia" meant a "Wikipedia-style" document in the wiki section of the ME/CFS Forums:

http://www.mecfsforums.com/wiki/PACE_trial_official_complaint_to_the_Lancet

The Lancet are not going to issue a correction. It doesn't seem to matter that Wikipedia and the ME/CFS Forums have nothing to do with each other.

 

[Bob]

I've found a study with "7.6" as a Chalder score for healthy controls (Likert scale 0-33) in a paper co-authored by Wessely:

Urinary Cortisol and Cortisol Metabolite Excretion in Chronic Fatigue Syndrome
Walid K. Jerjes, BSc, Norman F. Taylor, PhD, MRCPath, Timothy J. Peters, PhD, FRCP, FRCPath, Simon Wessely, MD, FRCP, FRCPsych and Anthony J. Cleare, MRCPsych, PhD
2006
http://www.psychosomaticmedicine.org/content/68/4/578.full

"The CFS subjects reported high mean scores for fatigue on the Chalder fatigue scale compared with controls (24.4 2.9; 7.6 2.3, p < .0001)."

Control group: "Results were compared with those of 40 healthy volunteers (20 males and 20 females) well matched for age and body mass index."

 

[Bob]

I'm sure that this has been posted before, but here it is again, in case helpful.

Mean SF-36 Physical Function score for adults without long term health problem = 92.7

Mean SF-36 Physical Function score for adults of working age without long standing illness = 92.5

From paper cited in the PACE Trial paper (reference no. 33):

Short form 36 (SF-36) health survey questionnaire: which normative data should be used? Comparisons between the norms provided by the Omnibus Survey in Britain, The Health Survey for England and the Oxford Healthy Life Survey.
Bowling A, Bond M, Jenkinson C, Lamping DL.
J Public Health (1999) 21 (3): 255-270.
doi: 10.1093/pubmed/21.3.255
http://jpubhealth.oxfordjournals.org/content/21/3/255.abstract

 

[Snow Leopard]

I emailed an editor at The Lancet about this issue and received a reply.

I wondered about this too and came to the same conclusion. It is strange they are not going to correct the error though...

 

[Dolphin]

I'm sure that this has been posted before, but here it is again, in case helpful.

Mean SF-36 Physical Function score for adults without long term health problem = 92.7

Mean SF-36 Physical Function score for adults of working age without long standing illness = 92.5

From paper cited in the PACE Trial paper (reference no. 33):

Short form 36 (SF-36) health survey questionnaire: which normative data should be used? Comparisons between the norms provided by the Omnibus Survey in Britain, The Health Survey for England and the Oxford Healthy Life Survey.
Bowling A, Bond M, Jenkinson C, Lamping DL.
J Public Health (1999) 21 (3): 255-270.
doi: 10.1093/pubmed/21.3.255
http://jpubhealth.oxfordjournals.org/content/21/3/255.abstract

Good to be reminded about these figures.

However, I think the second figure is from:
Jenkinson C, Coulter A, L W: Short form 36 (SF-36) Health Survey questionnaire: normative data from a large random sample of working age adults. BMJ 1993, 306:1437-1440. Free full text: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1677870/pdf/bmj00022-0017.pdf
a paper that was referenced in the PACE Trial protocol paper when they were saying why they chose the cut-offs they used, but which they dropped in the final paper.

Another thing which probably keeps needing to be hammered home is that the median value is higher again for a left-skewed distribution like the SF-36 PF e.g. the mean for the SF-36 for the population tends to be around 85 but the median is 95. A few low scores can bring down the mean.

So for example a dataset of {10, 25, 40, 80, 80, 90, 95, 95, 95, 95, 95, 100, 100, 100, 100, 100, 100, 100, 100, 100} has a mean of 85 but a median of 95 (and a mode of 100 which is like the SF-36 PF) - just because a number is one SD from the mean doesn't mean it's anything like a typical score. One can see a distribution of SF-36 PF scores in the top left of Figure 1 in Bowling et al (the paper Bob pointed out was referenced in the Lancet PACE paper): http://jpubhealth.oxfordjournals.org/content/21/3/255.full.pdf+html

A median of 95 for the whole population basically means that half of more scored 95 or 100. Given that half the population aren't superfit, it basically tells one that 95-100 are pretty "normal" scores.

 

[Bob]

Good to be reminded about these figures.

However, I think the second figure is from:
Jenkinson C, Coulter A, L W: Short form 36 (SF-36) Health Survey questionnaire: normative data from a large random sample of working age adults. BMJ 1993, 306:1437-1440. Free full text: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1677870/pdf/bmj00022-0017.pdf
a paper that was referenced in the PACE Trial protocol paper when they were saying why they chose the cut-offs they used, but which they dropped in the final paper.

Thanks Dolphin.
I should have added some more info before I posted my previous post.

The Bowling et al paper, that was referenced in the final PACE Trial paper, discusses and compares three different data sets:

• Health Survey for England (HSE) 1996 (ages 16+)
• The Oxford Healthy Life Survey 1991-1992 (ages 18-64) (Central England survey)
• The Omnibus Survey in Britain - British ONS Survey 1992 (ages 16+)

The Bowling et al paper quotes the data from the British ONS Omnibus survey, and specifically includes the mean figure of 92.7 which I quoted in my previous post.

The Bowling et al paper does not specifically quote the '92.5' figure which I also quoted in my previous post, however if we look at the whole data sets for the other two surveys discussed in the Bowling et al paper then we can find the mean figures for healthy adults for both of them...

The data for the Health Survey for England 1996 is given here:

HEALTH SURVEY FOR ENGLAND 1996
http://www.archive.official-documents.co.uk/document/doh/survey96/ehtitle.htm
Specific data re mean scores for adults without long standing illness:
http://www.archive.official-documents.co.uk/document/doh/survey96/tab5-34.htm
SF-36 Physical Function mean score for adults 16 and over, without long standing illness = 91​

And the data for The Oxford Healthy Life Survey 1991-1992 is given in this paper:

Short form 36 (SF36) health survey questionnaire: normative data for adults of working age.
C Jenkinson, A Coulter, and L Wright
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1677870/?page=3
Mean SF-36 Physical Function score for working age adults not reporting long standing illness = 92.5
​

The PACE Trial incorrectly gives a mean figure of 84 for the "UK working age population" and the authors have since corrected this to be for the "English adult population". The only reference the authors give for the '84' figure is for the Bowling et al paper, but '84' isn't quoted anywhere in that paper and I can't work out which of the two English surveys the authors got '84' from. I can't see it quoted anywhere.

The Bowling et al paper gives the 'total sample means' for each of the three studies as: 81, 88.4, and 89.6. I'm surprised that the PACE Trial didn't use the lowest mean of '81'.

If we look at the entire data sets, then we can find the means for healthy adults from each of the data sets, as I have above.

So, to clarify, the PACE Trial references the Bowling et al paper, which compares three data sets:
Health Survey for England (HSE) 1996 (ages 16+)
The Oxford Healthy Life Survey 1991-1992 (ages 18-64) (Central England survey)
The Omnibus Survey in Britain - British ONS Survey 1992 (ages 16+)

The means are as follows:
Health Survey for England (HSE) 1996 (ages 16+)
Total sample mean = 81
SF-36 Physical Function mean score for adults 16 and over, without long standing illness = 91

The Oxford Healthy Life Survey 1991-1992 (ages 18-64) (Central England survey)
Total sample mean = 88.4
Mean SF-36 Physical Function score for working age adults not reporting long standing illness = 92.5

The Omnibus Survey in Britain - British ONS Survey 1992 (ages 16+)
Total sample mean = 89.6
Mean SF-36 Physical Function score for adults without long term health problem = 92.7


Paper referenced in PACE Trial:
Short form 36 (SF-36) health survey questionnaire: which normative data should be used? Comparisons between the norms provided by the Omnibus Survey in Britain, The Health Survey for England and the Oxford Healthy Life Survey.
Bowling A, Bond M, Jenkinson C, Lamping DL.
J Public Health (1999) 21 (3): 255-270.
doi: 10.1093/pubmed/21.3.255
http://jpubhealth.oxfordjournals.org/content/21/3/255.abstract

Data:
Health Survey for England (HSE) 1996 (ages 16+)
http://www.archive.official-documents.co.uk/document/doh/survey96/ehtitle.htm
Specific data re mean scores for adults witout long standing illness:
http://www.archive.official-documents.co.uk/document/doh/survey96/tab5-34.htm

The Oxford Healthy Life Survey 1991-1992 (ages 18-64) (Central England survey)
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1677870/pdf/bmj00022-0017.pdf

The Omnibus Survey in Britain - British ONS Survey 1992 (ages 16+)
http://jpubhealth.oxfordjournals.org/content/21/3/255.abstract

 

[Graham]

Thanks Bob - downloaded and stored for future reading. Off the cuff though I noticed that (as is often the case) women's scores are different from men's, and the PACE trial had a much bigger ratio of women. When I was looking at the 6 min walking test, putting together men's and women's scores in the correct ratio gave me a value that wasn't directly from the tables - could that be what is happening here?

 

[Bob]

Thanks Bob - downloaded and stored for future reading. Off the cuff though I noticed that (as is often the case) women's scores are different from men's, and the PACE trial had a much bigger ratio of women. When I was looking at the 6 min walking test, putting together men's and women's scores in the correct ratio gave me a value that wasn't directly from the tables - could that be what is happening here?

That's a good point. So the mean scores should be even higher, if the gender ratios are matched.
And Dolphin's point that it is more appropriate to use median scores would take the average scores very close to 100.

ETA: In the next post, Ocean points out that average scores for women are lower than for men, so matching the gender ratios would actually lower the average scores.

 

[oceanblue]

Another thing which probably keeps needing to be hammered home is that the median value is higher again for a left-skewed distribution like the SF-36 PF e.g. the mean for the SF-36 for the population tends to be around 85 but the median is 95. A few low scores can bring down the mean.

So for example a dataset of {10, 25, 40, 80, 80, 90, 95, 95, 95, 95, 95, 100, 100, 100, 100, 100, 100, 100, 100, 100} has a mean of 85 but a median of 95 (and a mode of 100 which is like the SF-36 PF) - just because a number is one SD from the mean doesn't mean it's anything like a typical score.

Thanks. As Graham posted elsewhere, we almost all have more fingers than (mean)
average...

Bob, thanks for all those links, esp the archives. As there were more women in the PACE trial and women have lower scores in general that would lower the means slightly.

Worth noting that as well as giving a higher mean, looking at those without chronic health also gives smallers SDs, resulting in a much higher 'mean-1SD' threshold.

 

[Dolphin]

The PACE Trial incorrectly gives a mean figure of 84 for the "UK working age population" and the authors have since corrected this to be for the "English adult population". The only reference the authors give for the '84' figure is for the Bowling et al paper, but '84' isn't quoted anywhere in that paper and I can't work out which of the two English surveys the authors got '84' from. I can't see it quoted anywhere.

They probably got the figures from Table 3:
http://jpubhealth.oxfordjournals.org/content/21/3/255.full.pdf+html
e.g. an average of
Male: 86.3 (22.5)
Female: 81.8 (25.7)

One could use those figures to get a weighted mean similar to the PACE Trial participants. One (presumably big) reason that the female score is quite a bit lower is that the percentage of women who are old incl. very old in society is higher than the equivalent percentage for men.

 

[Dolphin]

Thanks Dolphin.
I should have added some more info before I posted my previous post.

The Bowling et al paper, that was referenced in the final PACE Trial paper, discusses and compares three different data sets:

• Health Survey for England (HSE) 1996 (ages 16+)
• The Oxford Healthy Life Survey 1991-1992 (ages 18-64) (Central England survey)
• The Omnibus Survey in Britain - British ONS Survey 1992 (ages 16+)

The Bowling et al paper quotes the data from the British ONS Omnibus survey, and specifically includes the mean figure of 92.7 which I quoted in my previous post.

The Bowling et al paper does not specifically quote the '92.5' figure which I also quoted in my previous post, however if we look at the whole data sets for the other two surveys discussed in the Bowling et al paper then we can find the mean figures for healthy adults for both of them...

The data for the Health Survey for England 1996 is given here:

HEALTH SURVEY FOR ENGLAND 1996
http://www.archive.official-documents.co.uk/document/doh/survey96/ehtitle.htm
Specific data re mean scores for adults without long standing illness:
http://www.archive.official-documents.co.uk/document/doh/survey96/tab5-34.htm
SF-36 Physical Function mean score for adults 16 and over, without long standing illness = 91​

And the data for The Oxford Healthy Life Survey 1991-1992 is given in this paper:

Short form 36 (SF36) health survey questionnaire: normative data for adults of working age.
C Jenkinson, A Coulter, and L Wright
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1677870/?page=3
Mean SF-36 Physical Function score for working age adults not reporting long standing illness = 92.5
​

The PACE Trial incorrectly gives a mean figure of 84 for the "UK working age population" and the authors have since corrected this to be for the "English adult population". The only reference the authors give for the '84' figure is for the Bowling et al paper, but '84' isn't quoted anywhere in that paper and I can't work out which of the two English surveys the authors got '84' from. I can't see it quoted anywhere.

The Bowling et al paper gives the 'total sample means' for each of the three studies as: 81, 88.4, and 89.6. I'm surprised that the PACE Trial didn't use the lowest mean of '81'.

If we look at the entire data sets, then we can find the means for healthy adults from each of the data sets, as I have above.

So, to clarify, the PACE Trial references the Bowling et al paper, which compares three data sets:
Health Survey for England (HSE) 1996 (ages 16+)
The Oxford Healthy Life Survey 1991-1992 (ages 18-64) (Central England survey)
The Omnibus Survey in Britain - British ONS Survey 1992 (ages 16+)

The means are as follows:
Health Survey for England (HSE) 1996 (ages 16+)
Total sample mean = 81
SF-36 Physical Function mean score for adults 16 and over, without long standing illness = 91

The Oxford Healthy Life Survey 1991-1992 (ages 18-64) (Central England survey)
Total sample mean = 88.4
Mean SF-36 Physical Function score for working age adults not reporting long standing illness = 92.5

The Omnibus Survey in Britain - British ONS Survey 1992 (ages 16+)
Total sample mean = 89.6
Mean SF-36 Physical Function score for adults without long term health problem = 92.7


Paper referenced in PACE Trial:
Short form 36 (SF-36) health survey questionnaire: which normative data should be used? Comparisons between the norms provided by the Omnibus Survey in Britain, The Health Survey for England and the Oxford Healthy Life Survey.
Bowling A, Bond M, Jenkinson C, Lamping DL.
J Public Health (1999) 21 (3): 255-270.
doi: 10.1093/pubmed/21.3.255
http://jpubhealth.oxfordjournals.org/content/21/3/255.abstract

Data:
Health Survey for England (HSE) 1996 (ages 16+)
http://www.archive.official-documents.co.uk/document/doh/survey96/ehtitle.htm
Specific data re mean scores for adults witout long standing illness:
http://www.archive.official-documents.co.uk/document/doh/survey96/tab5-34.htm

The Oxford Healthy Life Survey 1991-1992 (ages 18-64) (Central England survey)
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1677870/pdf/bmj00022-0017.pdf

The Omnibus Survey in Britain - British ONS Survey 1992 (ages 16+)
http://jpubhealth.oxfordjournals.org/content/21/3/255.abstract

Well done for finding the Health Survey for England 1996 - don't think we had that before.

 

[biophile]

PACE spinning into a deeper hole

So in their Lancet published "authors' reply" White et al admitted that they didn't use a "working age population" for normative data on PF/SF36 scores as claimed but used a general population instead (which included a significant proportion of people who were elderly and/or diseased). However, rather than recalculating the figures for people returning to "normal" they dig a deeper hole for themselves by implying that a general population is more relevant than a working age population (which was already questionable for including people with diseases that are exclusionary for a CFS diagnosis):

"We determined the normal range by use of the conventional mean plus or minus 1 SD from what we regarded as the most relevant general population data. For physical function, this was a demographically representative sample (in our paper we stated that this was a UK working-age population, whereas more accurately this should have been an English adult population).[3] For fatigue, this was a population sample of patients registered with a general practitioner in the southeast of England, who had consulted for a health problem at some time in the year after completion of the fatigue measure (ie, they were not consulting at the time). It is important to clarify that our paper did not report on recovery; we will address this in a future publication."

http://www.thelancet.com/journals/lancet/article/PIIS0140-6736(11)60651-X/fulltext

Just in case anyone detects possible ambiguity in the statement "what we regarded as the most relevant general population data" which could be interpreted or backpedaled as "the most relevant GENERAL population data" rather than "the most relevant population data, period", this latter absurdity is confirmed/repeated in the response to the FOI request sent by the ME Association:

"As explained in the published response to correspondence received by the Lancet (including your own published letter making a very similar point), the authors used a different population study to that mentioned in the protocol to derive the normal range scores for the SF36 physical function scale as they believed this to be the most representative study for the trial sample (Bowling et al, 1999)."

http://www.meassociation.org.uk/wp-content/uploads/2011/06/FOI+from+Queen+Mary.pdf

What they are claiming as the "most relevant" normative dataset is actually the least relevant possible aside from other dedicated disease populations. IIRC on this thread we have established from various datasets that the "mean minus SD" rule should give a lower threshold of 80/100 not 60/100. Ironically, Bleijenberg & Knoop (PACE editorial) and White (lead author of PACE) co-authored a study, "Is a Full Recovery Possible after Cognitive Behavioural Therapy for Chronic Fatigue Syndrome?" where they stated:

"Physical disabilities were measured with the physical functioning subscale of the Medical Outcomes Survey Short Form-36 (SF-36) [14, 15] . The scores range from 0 (maximum physical limitations) to 100 (ability to do vigorous activity). Healthy adults without a chronic condition [16] were used as a norm group, with a mean score of 93.1 (SD 11.7). A patient had to score 80 or higher to be considered as recovered."

http://www.cfids-cab.org/rc/Knoop-1.pdf

Not to mention the related blunder in the accompanying editorial, which Hooper says a correction will be issued for. In a 2010 paper on "Fatigue and chronic fatigue syndrome-like complaints in the general population" (thanks Dolphin?), Bleijenberg co-writes: "A cut-off of ?65 was considered to reflect severe problems with physical functioning." (http://eurpub.oxfordjournals.org/content/20/3/251.full.pdf). Yet in the PACE trial editorial Bleijenberg implies an even poorer score of 60 is the lower cut-off point for "recovery" to healthy norms.

For reasons we have already discussed in this thread, it is totally inappropriate to use a general population as these authors have done in the PACE trial. By doing so they contradicted their own work on other papers. Even when applying the "mean minus SD" rule, 60/100 simply isn't "normal" for the average person of similar age to PACE trial participants, and no amount of spin can change that.

These researchers know exactly what they are writing, there is simply no excuse for these blunders, and when confronted with the error they just dug a deeper hole. The whole incident seems rather dodgy to me, and I've lost more respect for these people as "scientists".

 

[Sean]

Nice work, biophile.

I agree, they know exactly what they are doing, and how dodgy it is.

Well, it is all on the formal record, they can't say they didn't know or didn't do it.

 

[Snow Leopard]

In addition, An SF-36 PF score of less than 70 is one of the thresholds for the CDC empirical criteria. It is clear that the specificity of PF>60 is highly questionable.

 

[oceanblue]

I think that when PACE publishes its 'recovery' paper will be the time to really nail down these sleights of hand.

 

[Dolphin]

In addition, An SF-36 PF score of less than 70 is one of the thresholds for the CDC empirical criteria. It is clear that the specificity of PF>60 is highly questionable.

Good point. Just to be exact, the PACE Trial authors used: PF>=60 (i.e. 60 or more).

I think that when PACE publishes its 'recovery' paper will be the time to really nail down these sleights of hand.

Good point. There are a few papers in the pipeline => more opportunities to challenge.

One I'm keeping an eye out for at the journal site is:

Cella, M and Chalder, T and White, P Does the heterogeneity of chronic fatigue syndrome moderate the response to cognitive behaviour therapy?
An exploratory study. Psychotherapy and Psychosomatics (In press).

 

[biophile]

'Normal' was defined AFTER the outcome data was unblinded

As we know, PACE replaced a stricter definition of "recovery" with a rather dubious definition of "normal", which allowed trial participants to enter the trial classed as having CFS with substantial disability, make no improvement after 52-weeks, then be classed as "normal" ie a success story at the end of the trial. Participants averaging 38(SD=11) years old could also be classed as "normal" after 52-weeks despite having the physical functioning of the average 80 year old.

Changes were made to the original published protocol, which just happened to coincide with making the paper look better than it otherwise would have, possibly in response to the failed FINE trial or a similar warning sign. White et al state that these changes were approved before examining the outcome data:

"Changes to the original published protocol were made to improve either recruitment or interpretability, such as changing the proposed composite primary outcomes to single continuous scores. The analysis was guided by a Statistical Analysis Strategy (which we intend to publish), which was completed before analysis of outcome data, and which was much more detailed than the plan in the protocol; this is now conventional in the conduct of clinical trials. The eight secondary outcomes presented in our paper were selected for clinical relevance. All these decisions and plans were approved by the Trial Steering Committee, were fully reported in our paper, and were made before examining outcome data to avoid outcome reporting bias."

http://www.thelancet.com/journals/lancet/article/PIIS0140-6736(11)60651-X/fulltext

However, according to White's (unpublished?) response to Hooper via the Lancet:

"The primary analysis compared the mean differences in the primary outcome scores across treatment arms, which are in the paper. The normal range analysis was plainly stated as post hoc, given in response to a reviewers request. We give the results of the proportions with both primary outcomes within normal ranges, described a priori, using population derived anchors."

http://www.meactionuk.org.uk/whitereply.htm

Why did a peer reviewer need to make such a request when there was already an OK definition in the protocol? Assuming the "reviewer" was for the Lancet paper and not the changes to the protocol, this suggests that the authors either submitted a paper without any definition of recovery whatsoever or less likely the peer reviewer requested the definition of "recovery" to be replaced with one for "normal".

While changes to the original published protocol were approved before examining the data, it appears that they were still allowed to omit whatever measurement they wanted at their own discretion. As Hooper states in his reply to White, "if the outcome results had been resoundingly successful, it is nave to think that the recovery statistics would not have been at the forefront of the published paper" (http://www.meactionuk.org.uk/Comments-on-PDW-letter-re-PACE.htm).

A convenient probable scenario for PACE: drop an important outcome measure they either couldn't change and/or didn't like after examining the disappointing data, have a "peer reviewer" who is sympathetic to their cause "request" a watered-down version they did like after re-examining this data with it, maintain the overall impression that "changes" were done before being unblinded to the data, parade the inflated figures of patients "getting back to normal" at a press conference without mention of the word "recovery" or the adequate definition for it, avoid responsibility for "recovery" and "normal" being confused in the news media, and have the journal that published their paper portray critics as emotional and irrational for questioning this.

 

[oceanblue]

The primary analysis compared the mean differences in the primary outcome scores across treatment arms, which are in the paper. The normal range analysis was plainly stated as post hoc, given in response to a reviewers request."


Why did a peer reviewer need to make such a request when there was already an OK definition in the protocol?

I'm pretty sure this just means that the reviewer insisted that the 'normal range' was labelled as post-hoc, to make it clear the researchers were taking libertys. Which means that PACE had originally not owned up to this.