'Recovery' from chronic fatigue syndrome after treatments given in the PACE trial

[Sea]

Was there anything on severity of symptoms? I don't know anyone who would answer "no" to the question "Have you been fatigued in the last 2 weeks?" Now if you asked things like how many times in the last 2 weeks did you not shower? or clean your teeth? or eat? ...because of fatigue, well then there might be some comparison

 

[biophile]

Was there anything on severity of symptoms? I don't know anyone who would answer "no" to the question "Have you been fatigued in the last 2 weeks?" Now if you asked things like how many times in the last 2 weeks did you not shower? or clean your teeth? or eat? ...because of fatigue, well then there might be some comparison

Good point.

There is a table with the data on severity :
http://europepmc.org/articles/PMC3020067?table=tbl7/

Also given are adjusted odds ratios for reported 2-week prevalence of 20 symptoms by participant characteristics:
http://europepmc.org/articles/PMC3020067?table=tbl6/

For the 3 symptoms White et al listed (but it is also interesting to look at the other symptoms):

No chronic health condition vs chronic health condition:
Feeling tired/run down: 2.00.
Headaches: 1.47.
Joint pain: 1.59.

Cannot work due to illness vs full-time employment:
Feeling tired/run down: 1.54.
Headaches: 1.12.
Joint pain: 3.01.

 

[biophile]

22% [recovery] may indeed be setting the bar high but...

A net response rate of 15% (NNT=7) does not sound like setting the bar high when the trial is non-blinded with different levels of encouragement/optimism between groups and the thresholds for (mostly self-rated) improvements are so close to where some participants were already hovering at trial entry that is was possible to be classified as recovered without making a minimal clinically useful difference (which itself was already low). Of course, as you already said, it would not be fair to compare the recovery rates of other studies to 22% if those studies use stricter methodology.

 

[Valentijn]

Was there anything on severity of symptoms? I don't know anyone who would answer "no" to the question "Have you been fatigued in the last 2 weeks?" Now if you asked things like how many times in the last 2 weeks did you not shower? or clean your teeth? or eat? ...because of fatigue, well then there might be some comparison

The CGI (Clinical Global Impression) scale was designed to be assessed by physicians based on specific symptoms relevant to whatever the disease is, in line with your activity-based suggestions. In PACE, it was used to say if you think you feel better generally, and self-reported in some cases and doctor-reported in others.

 

[Sam Carter]

The several criteria for ill health and recovery are described in the text but the results are in Table 1:

...
3) A CGI (clinical global impression) score of 1 i.e. "very much better" or 2 i.e. "much better". Adding this onto the previous two criteria for recovery does significantly lower the recovery rates, which suggests that a significant proportion of the "recovered" participants only felt slightly better from baseline. Also note that in the original definition of recovery, only 1 or "very much better" was regarded as the CGI criterion for recovery.

...

(emphasis added)

Following on from Biophile's analysis it's interesting to note that adding the CGI criterion reduced the numbers of "recovered" APT / SMC participants by ~50% but for CBT / GET the numbers fell by only ~20%.

This discrepancy surely indicates something but without access to the raw data it's only possible to guess what that something is -- it might reflect the fact that CBT / GET really were more effective, but equally it might indicate that the CGI rating acted only as a proxy measure of the nocebo / placebo effect evoked by the different treatments.

From Table 1 b
............................................APT................CBT................GET................SMC
And Oxford criteria not met 15 (23/149).....28 (40/143).....28 (41/144).....14 (21/150)
And CGI 1 or 2 .....................8 (12/149).....22 (32/143).....22 (32/143).......7 (11/150)

 

[Valentijn]

Following on from Biophile's analysis it's interesting to note that adding the CGI criterion reduced the numbers of "recovered" APT / SMC participants by ~50% but for CBT / GET the numbers fell by only ~20%.

This discrepancy surely indicates something but without access to the raw data it's only possible to guess what that something is -- it might reflect the fact that CBT / GET really were more effective, but equally it might indicate that the CGI rating acted only as a proxy measure of the nocebo / placebo effect evoked by the different treatments.

22 of the patients had their CGI scores determined by the trial's SMC doctors in the results that are given. There's every indication that the SMC doctors knew which groups the patients were in, and it's certainly possible that the non-self-rating CBT/GET participants were deliberately given high scores and the non-self-rating APT/SMC deliberately given low CGI scores. But that's only 22 out of 607 patients, so I think the heavy CBT/GET treatment emphasis on denying illness had a bigger impart.

 

[Simon]

How useful is a clinically useful difference?

For each treatment group significantly more people improved by a clinically useful difference (2 points for fatigue and at least 8 points for physical function) than rated themselves "very much better" or "much better" on the CGI scale. This suggests that it was possible to improve by a CUD and either not notice any improvement or only rate yourself "a little better" which the protocol defines as "non-improvement"[*].

If I remember correctly the economic calculations in McCrone were predicated on a CUD actually representing a genuine and meaningful improvement.

Lancet 2011: improved by a clinically useful difference

........%
APT 42 64/153 participants
CBT 59 87/148 participants
GET 61 94/154 participants
SMC 45 68/152 participants

Psychological Medicine 2013: Overall health change CGI 1 or 2

This is a really good point.

Can't remember where I saw it, but there is some research on validating eg a CUD by comparing CUD data with other outcome data, such as CGI, to ensure that statistically-derived thresholds for improvement reflect what patients say makes a difference. Since we are talking self-reported measures here, for fatigue, physical function etc, what the patients says is a real difference is the most important thing.

 

[Firestormm]

Just to say that the Lord's private debate did feature this paper on recovery: http://www.parliamentlive.tv/Main/Player.aspx?meetingId=12482 from 17.25

Have listened. Some good points made. Will get transcript.

 

[user9876]

(emphasis added)

Following on from Biophile's analysis it's interesting to note that adding the CGI criterion reduced the numbers of "recovered" APT / SMC participants by ~50% but for CBT / GET the numbers fell by only ~20%.

This discrepancy surely indicates something but without access to the raw data it's only possible to guess what that something is -- it might reflect the fact that CBT / GET really were more effective, but equally it might indicate that the CGI rating acted only as a proxy measure of the nocebo / placebo effect evoked by the different treatments.

From Table 1 b
............................................APT................CBT................GET................SMC
And Oxford criteria not met 15 (23/149).....28 (40/143).....28 (41/144).....14 (21/150)
And CGI 1 or 2 .....................8 (12/149).....22 (32/143).....22 (32/143).......7 (11/150)

It should worry them that patients who they claim are meeting their basic recovery criteria are not even saying they feel better. If 30% of the patients that you claim have recovered and no longer meet the diagnosis criteria yet do not even get classed as much better something would seem to be wrong with the recovered criteria.

 

[Dolphin]

(emphasis added)

Following on from Biophile's analysis it's interesting to note that adding the CGI criterion reduced the numbers of "recovered" APT / SMC participants by ~50% but for CBT / GET the numbers fell by only ~20%.

This discrepancy surely indicates something but without access to the raw data it's only possible to guess what that something is -- it might reflect the fact that CBT / GET really were more effective, but equally it might indicate that the CGI rating acted only as a proxy measure of the nocebo / placebo effect evoked by the different treatments.

From Table 1 b
............................................APT................CBT................GET................SMC
And Oxford criteria not met 15 (23/149).....28 (40/143).....28 (41/144).....14 (21/150)
And CGI 1 or 2 .....................8 (12/149).....22 (32/143).....22 (32/143).......7 (11/150)

It should worry them that patients who they claim are meeting their basic recovery criteria are not even saying they feel better. If 30% of the patients that you claim have recovered and no longer meet the diagnosis criteria yet do not even get classed as much better something would seem to be wrong with the recovered criteria.

I think this would be a good point to make in a letter for anyone so inclined.

 

[Simon]

Was there anything on severity of symptoms? I don't know anyone who would answer "no" to the question "Have you been fatigued in the last 2 weeks?" Now if you asked things like how many times in the last 2 weeks did you not shower? or clean your teeth? or eat? ...because of fatigue, well then there might be some comparison

Think it's worth looking at this in more detail:

... it is important to note that recovery does not mean being free of all symptoms; population studies show that the average person in the UK reports a mean of four symptoms in any 2-week period. The three most common symptoms reported were fatigue, headache and joint pain; symptoms consistent with CFS (McAteer et al. 2011).

Which sounds a bit like "everyone's got problems, stop aiming for perfection and accept life as it is".
As Biophile noted, the McAteer study did not have a represenative sample: only 33% of questionnaires were returned correctly completed (67% didn't manage of this) and perhaps unsurprisingly 44% of these had a chronic illness. Presumably those with health problems are more likely to reply to questionnaires about their symptoms than those that are healthy.

So let's focus on the 56% who were healthy (or not chronically ill) since the symptom level of chronically ill patients are irrelevant when considering what counts as recovered. The mean number of symptoms for those without chronic illness is about 2.8.

Here are the prevalence rates for those 'CFS-conistent' symptomsreported for the previous 14 days for 'healty' people:

Feeling tired/run down = 34%
Headaches = 36%
Joint pain = 24%

Note that's 'feeling tired/run down', which is broader than just 'fatigue'. Even so, it's a high level of these symptoms- so how bad are they (as sea asked)? There is a 'severe' category but a larger category than that is 'highly interfering'. The tired/run down figure falls from to 9% using that definition (falls to 2% for time off work). But that figure includes the chronically ill, who are nearly half the sample. What about the healthy? It's likely that severe symptoms are more common among the chronically ill, but separate data isn't give.So, assuming the same rate of severity among the healthy as the ill, the rates of symptoms that are highly interfering for the healthy are:

Feeling tired/run down = 7%
Headaches = 3%
Joint pain = 4%

Which is a long way from implying that most people experience 4 symptoms in any 14 day period.

edit: thanks to Biophile for all the earlier work and links.

 

[Dolphin]

Think it's worth looking at this in more detail:
Which sounds a bit like "everyone's got problems, stop aiming for perfection and accept life as it is".
As Biophile noted, the McAteer study did not have a represenative sample: only 33% of questionnaires were returned correctly completed (67% didn't manage of this) and perhaps unsurprisingly 44% of these had a chronic illness. Presumably those with health problems are more likely to reply to questionnaires about their symptoms than those that are healthy.

So let's focus on the 56% who were healthy (or not chronically ill) since the symptom level of chronically ill patients are irrelevant when considering what counts as recovered. The mean number of symptoms for those without chronic illness is about 2.8.

Here are the prevalence rates for those 'CFS-conistent' symptomsreported for the previous 14 days for 'healty' people:

Feeling tired/run down = 34%
Headaches = 36%
Joint pain = 24%

Note that's 'feeling tired/run down', which is broader than just 'fatigue'. Even so, it's a high level of these symptoms- so how bad are they (as sea asked)? There is a 'severe' category but a larger category than that is 'highly interfering'. The tired/run down figure falls from to 9% using that definition (falls to 2% for time off work). But that figure includes the chronically ill, who are nearly half the sample. What about the healthy? It's likely that severe symptoms are more common among the chronically ill, but separate data isn't give.So, assuming the same rate of severity among the healthy as the ill, the rates of symptoms that are highly interfering for the healthy are:

Feeling tired/run down = 7%
Headaches = 3%
Joint pain = 4%

Which is a long way from implying that most people experience 4 symptoms in any 14 day period.

edit: thanks to Biophile for all the earlier work and links.

Can anyone recall what threshold was used in PACE Trial for having symptom e.g. with CDC criteria, etc?
Too tired to look, atm. I know they had to tick a severity.
Full protocol is here: http://evaluatingpace.phoenixrising.me/PACE_Protocol.pdf

but I think the authors said at some stage what the threshold was they used.

 

[biophile]

"In table 6, the maximal CDC associated symptom count is eight, the 'ninth' symptom being fatigue. A symptom was counted as present if scored as 'more often than not', or more frequently than that."

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

Presumably over 6 months? On p156 of the PACE protocol, it asks if patients experience these symptoms in the last week. It is unclear if this was also used to diagnose CDC criteria at baseline. If so, it is a violation of the CDC definition*. For purposes of "recovery", symptoms were only counted over 1 week too.

* "Accompanying symptoms must have persisted or recurred during 6 or more consecutive months of illness and cannot have predated the fatigue." http://www.ncbi.nlm.nih.gov/pmc/articles/PMC317472 (cited by PACE for CDC definition)

 

[Dolphin]

I think I've just realised something important.
I had thought that people had to score >65 on the SF-36 physical functioning scale AND <6 on the Chalder Fatigue Scale for the Oxford criteria not to be met.
However, if that was the case, one would have expected the scores to drop more from the first line to the second line of Table 1 (b).

However, if either of those situations occurs, then the person wouldn't have been eligible for the trial. That means a person could still have a SF-36 PF score of 60 or 65 and be counted as recovered/not satisfying the Oxford criteria/not satisfying the entry criteria, once their Chalder Fatigue Scale was <6. Similarly, somebody's Chalder Fatigue Scale score could be 6 or more and still be counted as recovered/ not satisfying the Oxford criteria/not satisfying the entry criteria, once their SF-36 PF score was more than 65 (although their Chalder Fatigue score would also need to be 18 or less, still possible at baseline.

 

[Dolphin]

I think I've just realised something important.
I had thought that people had to score >65 on the SF-36 physical functioning scale AND <6 on the Chalder Fatigue Scale for the Oxford criteria not met.
However, if that was the case, one would have expected the scores to drop more from the first line to the second line of Table 1 (b).

However, if either of those conditions are not met, then the person wouldn't have been eligible for the trial. That means a person could still have a SF-36 PF score of 60 or 65 and be counted as recovered/not satisfying the Oxford criteria/not satisfying the entry criteria, once their Chalder Fatigue Scale was <6. Similarly, somebody's Chalder Fatigue Scale score could be 6 or more and still be counted as recovered/ not satisfying the Oxford criteria/not satisfying the entry criteria, once their SF-36 PF score was more than 65.

---------
What this means is in effect, that to be counted as recovered, if one started with a SF-36 PF score of 60 or 65, one would only need improvements on two scales to be counted of recovered:

Example 1:
(i) that your Chalder Fatigue Scale (bimodal) score was less than 6 (entry criteria >=6) and your Chalder Fatigue Scale (likert) score was <=18 (it could have been like that at baseline)

and
(ii) you put one of the top two answers for the Global Change Score:

​

Overall, how much do you feel your health has changed since the start of the study? Please tick the one box below that most closely corresponds to how you feel now?​

​

Very much better <-- Original recovery criteria requirement​

​

Much better <-- added in for new recovery requirement​

​

A little better​

​

No Change​

​

A little worse​

​

Much worse​

​

Very much worse​

e.g.

baseline: SF-36 PF 65, Chalder Fatigue Scale: (bimodal) 6; Likert Score: 19

12 months: SF-36 PF 65, Chalder Fatigue Scale: (bimodal) 6; Likert Score: 18. Rates themselves as "much better".
This person would be seen as meeting all the recovery criteria

​

​

​

Example #2:​

Similarly, if one started with a Chalder Fatigue Score of <=18 (theoretically possible).​

​

Then
once one's SF-36 PF score was >65 (it could have started at 65).
and

(ii) you put one of the top two answers for the Global Change Score:

​

Overall, how much do you feel your health has changed since the start of the study? Please tick the one box below that most closely corresponds to how you feel now?​

​

Very much better <-- Original recovery criteria requirement​

​

Much better <-- added in for new recovery requirement​

​

A little better​

​

No Change​

​

A little worse​

​

Much worse​

​

Very much worse​

​

​

You would be counted as recovered.​

e.g.​

baseline: SF-36 PF 65, Chalder Fatigue Scale: (bimodal) 7; Likert Score: 18

12 months: SF-36 PF 70, Chalder Fatigue Scale: (bimodal) 7; Likert Score: 18. Rates themselves as "much better".
This person would be seen as meeting all the recovery criteria

 

[Esther12]

Dolphin: That's how I interpreted their criteria too. So sometimes in their paper they count a score of 60 as normal, sometimes as abnormal.

 

[Dolphin]

Dolphin: That's how I interpreted their criteria too. So sometimes in their paper they count a score of 60 as normal, sometimes as abnormal.

However, a lot of people seem to think being 60 or 65 was sufficient (for all the recovery requirements). However, if the Chalder Fatigue Scale (bimodal) score doesn't drop below 6, 60 or 65 isn't sufficient - in that scenario, the SF-36 PF score has to be >65 (basically 70+) or you won't be counted as recovered (because otherwise you would still meet the entry requirements).

 

[Esther12]

I guess it's a difficult criteria to explain concisely, so it's possible people were just not being clear in detailing the full criteria.

It's true to say that a patient scoring only 60 would be able to be classed as recovered, but it's not the case that this new criteria allows a patient to fulfil their criteria for severe and disabling fatigue and recovery at the same time.

I posted this last night, which summarises my thoughts on the politics and intent of the criteria:

http://forums.phoenixrising.me/inde...n-recovery-from-pace.20243/page-4#post-331876

It really depends if they use CFQ also. If they do, I think the SF-36 PF threshold would be easier again. I'm wondering whether not satisfying both of the entry criteria is what they are counting?

Yeah - that's what I'd guess too. Maybe they realised that having papers defining 'recovery' in a way that overlapped with their own definition for "severe and disabling fatigue" was too transparently absurd.

I think that we were pretty close here, and they they did realise having an actual overlap between 'recovery' and 'severe and disabling' fatigue looked ridiculous (even if only patients cared about it), and but still wanted as weak a criteria for recovery as was possible.

It didn't occur to me that they could leave the thresholds for fatigue and disability overlapping with their criteria for 'severe and disabling' fatigue, but then just add as a proviso that patients could not fulfil their entry criteria.

That their 'recovery' rate dropped so much shows how many patients classed as 'back to normal' still fulfilled their criteria for CFS. [edit - Dolphin pointed out that table 1b seems to indicate that the CGI self-rating was more of a limiting factor than the exclusion of those fulfilling the Oxford criteria]

 

[Dolphin]

Esther12, just to say I've gone back and added "(for all the recovery requirements)" to the previous post.

I'm not sure your point in your second last sentence, "that their 'recovery' .... ?
Agree otherwise.

 

[Dolphin]

Dolphin: So sometimes in their paper they count a score of 60 as normal, sometimes as abnormal.

This would be a good phrase or point to make in a letter.