Prevalence and predictors of recovery from chronic fatigue syndrome in a routine clinical practice

[alex3619]

Acceptable success rate is not the same as full recovery. Its partial recovery. While I will take 90% to full recovery over no recovery, we should not let them get away with claiming poor recovery or no recovery as full recovery.

 

[Dolphin]

It's a while since I read the comments so I may repeat some points:

One of the oddest things to me with this paper is that people could be recovered from the Oxford and CDC criteria without satisfying either at baseline! What patients included had to satisfy at baseline was the NICE criteria. 47.4% didn't satisfy the CDC criteria at baseline and 27.3% didn't satisfy the Oxford criteria at baseline. 16.8% satisfied neither at baseline.

A related odd thing that people could still satisfy the NICE criteria for CFS at the end and be counted as recovered.

So to summarise one enters the trial:
- Satisfying NICE criteria
- Not satisfying Oxford criteria
- Not satisfying CDC criteria

and one could at the end be
- Satisfying NICE criteria
- Not satisfying Oxford criteria
- Not satisfying CDC criteria

and be counted as recovered

 

[Dolphin]

It's a while since I read the comments so I may repeat some points:

One of the oddest things to me with this paper is that people could be recovered from the Oxford and CDC criteria without satisfying either at baseline! What patients included had to satisfy at baseline was the NICE criteria. 47.4% didn't satisfy the CDC criteria at baseline and 27.3% didn't satisfy the Oxford criteria at baseline. 16.8% satisfied neither at baseline.

A related odd thing that people could still satisfy the NICE criteria for CFS and be counted as recovered.

So to summarise one enters the trial:
- Satisfying NICE criteria
- Not satisfying Oxford criteria
- Not satisfying CDC criteria

and one could be
- Satisfying NICE criteria
- Not satisfying Oxford criteria
- Not satisfying CDC criteria
and be counted as recovered

Another example of the odd cohort they use relates to the other criteria:
One could have the same score on the SF-36 Physical Functioning Scale and the CFQ as baseline and be counted as recovered.

The only thing that has to change is that one said one felt "better" or "much better" (I think the wording was likely "much better" and "very much better" based on the description they gave for the scale).

 

[Dolphin]

The baseline data in Table 1 are interesting enough to me:
16.1% had CFQ of <=18 at baseline (this was called a normal level of fatigue in the PACE Trial).

28.9% had SF-36 physical functioning >=65 at baseline (in the PACE Trial, >=60 was considered normal physical functioning so at least 28.9% had normal physical functioning at baseline).

Slightly less important but 12.3% had SF-36 physical functioning >=83 at baseline (the original recovery criteria in the PACE Trial as >=85, which is basically the same score as scores go up in units of 5 unless there is missing data).

 

[Dolphin]

"Between 6 and 20 sessions were attended with a mean of 15.3 sessions (SD=3.7)". When NICE did its calculations on the costs of CBT it was based on 9 sessions.

I think non-pharmacological therapies are often costing more than what was estimated - spending isn't watched as closely as with drugs I think.

 

[Dolphin]

"The control variable (CFQ cut-off at pre-treatment) was not significantly related to recovery* at 6 months"

CFQ cut off was CFQ <18 (this may actually be <=18)

This would (almost certainly) mean that some people's scores must have deteriorated i.e. those who were within the cut-off at baseline. Because if all of them were recovered, a large chunk of those >18 wouldn't be recovered and this would show up with testing.

* recovery on the CFQ scale

 

[Dolphin]

Quite a few of the associations in Table 3 (and hence in Table 4) are likely just random noise, there by chance.
There are 76 tests so one would expect on average 7.6 with p<0.1 or lower. There are 24.
Similarly one would expect 3.8 with p<0.05 or lower. There are 13.

Ones that I think are particularly likely to be due to chance are ones where they are both there (i.e. statistically significant) but in opposite directions for different tests of "recovery":
CBRQ- Catastrophizing
CBRQ - Avoidance/resting
CBRQ - All-or-nothing

The ones I would be more convinced about would be those in more than one column, in the same direction, in Table 3.

 

[Dolphin]

Multivariate analyses showed that worse scores on the work and social adjustment scale, unhelpful beliefs about emotions, high levels of depression and older age were associated with reduced odds for recovery.

I find this to be quite misleading. They looked at recovery in four different ways but this isn't made clear to the reader. Each of these measures was only found relevant for one of the four measures of recovery in the multivariate analysis and not relevant for the other three.

 

[Dolphin]

In relation to the operational criteria for CFS about half of the patients in the
present study met the Oxford criteria at follow up.This rate is similar to previous studies
carried out in the UK (Deale et al., 2001 and Whiteet al., 2013). More than 60% of patients
no longer met the CDC criteria in this study, again, similar rates to those found in previous
studies (Knoop et al 2007; White et al., 2013). Recovery defined as not meeting the CDC
criteria yielded slightly higher recovery rates compared to the Oxford criteria.

I find the comparisons regarding the Oxford criteria misleading. In Deale et al. and White et al. 100% of patients satisfied the Oxford criteria at baseline, while in this study only 72.7% did.

The comparisons regarding the CDC criteria are also somewhat misleading: in the Knoop et al. study, 100% satisfied the CDC criteria at baseline. In White et al., 67% satisfied the CDC criteria at baseline. Only 52.6% did in this study.

 

[Dolphin]

No mention in this of the lower rate of 41% found for CBT in the PACE Trial - Trudie Chalder was one of the Principal Investigators for this trial and results from it had previously been mentioned in the preceeding paragraph with regard to satisfying Oxford or CDC criteria:

Using a self-rated global measure of improvement 60.8% of the patients in the present
study reported feeling better or much better. This percentage is very similar to the results of
Quarmby et al (2007) in which 57% of patients in routine care reported feeling better or much
better. Overall this rate is lower than the 70% global improvement rates in the context of a
randomised controlled trial (Deale et al., 2001). Self-rated improvement was also lower than
the 78% self-rated improvement reported in the context of a randomised trial of therapist
aided internet therapy for adolescents with CFS (Nijhof et al., 2012). This sample was
obviously younger and so is not directly comparablebut it does suggest that a younger cohort
may have better chances of recovery. Although the global measure does not pertain to
symptom reduction, the high prevalence of subjectively experienced improvement attests to
the benefit patients experienced when receiving CBTin the present study.

There's no mention here of the control group. In the PACE Trial, 25% (vs 41%) reported improvement in the group that didn't get individual therapy.

 

[Dolphin]

(I think this has been commented on before)

Knoop and colleagues (2007) found that 48 % were within 1 SD of the normal
population mean of fatigue, which is comparable to the prevalence of 48.6% found in the
present study. Although different fatigue scales were employed, recovery was based on
normal population data in both studies. Although Deale and colleagues (2001) found that
63% of the participants were within normal range of fatigue at 6 months (Deale, et al., 2001)
the larger study carried out by White et al (2013)found only 41% of those who received CBT
were within the normal range which suggests that this figure is more accurate within a larger
cohort.

Knoop et al. used a healthy group to derive a threshold, not a general population.

Deale et al. used <4 (bimodal scoring) vs <=18 (or made <18 - it's not clear) (likert scoring) in this study. White et al. used <=18.

 

[Dolphin]

Total Recovery was measured similarly in all previous studies. We found a total
recovery rate of 18.3%, compared to 23% (Knoop, et al., 2007a), 24% (Deale, et al., 2001),
and 22% (White et al., 2013) respectively. Factors such as patient selection and the use of
manualised protocols may have affected the minor differences in outcomes between the
present study and the randomised controlled trials. Nevertheless, all in all, the recovery rates
found in these studies were comparable.

Artificially so I think. White et al. used quite different criteria to Deale et al. and Knoop et al. (much laxer).

27.3% of the people in the current study included didn't satisfy the Oxford criteria at baseline.

 

[Dolphin]

Somewhat surprisingly, more avoidance/resting at baseline was positively related to
recovery as defined by physical functioning, while this association was negative in the
univariate analysis (i.e., related to reduced odds for recovery). The relationship with recovery
may indicate a difference in response to treatment among patients with this coping strategy at
pre-treatment. Interestingly, Cella and colleagues (2010) found that patients characterised by
avoidance and resting were less likely to show a reduction in level of fatigue symptoms
(Cella, et al., 2011a). There may be different mechanisms at play with regard to recovery in
physical function.

This last sentence is weak and very speculative.

 

[Dolphin]

Buried in the discussion:

There was a drop-out of 27.8% in the present study.

This would presumably artifically increase the success rate.

 

[Dolphin]

Different predictors were related to recovery in terms of official criteria, subjective
improvement, reduced fatigue symptomatology and reduced physical impairment. This
suggests that a broad therapeutic focus is necessary to improve all facets of CFS, and to
facilitate total recovery.

Alternatively, some or all of what was picked up was random chance. Also, total recovery wasn't proven.

CBT has this broad focus but the number of sessions may need to be
increased to facilitate a full recovery.

That's all the have to offer - more CBT. No other suggestions.

 

[Denise]

(Hope I am not duplicating something that has already been said)
A drop-out rate (27.8%) that is higher than recovery rate (18.3%)..... That doesn't look too good to me.

 

[Dolphin]

(Hope I am not duplicating something that has already been said)
A drop-out rate (27.8%) that is higher than recovery rate (18.3%)..... That doesn't look too good to me.

Not sure I'd agree, to be honest. If they were really getting a 18.3% recovery rate, one could probably live with that drop-out rate (if those people (i.e. the drop-outs) weren't long-term worse).

 

[Denise]

Not sure I'd agree, to be honest. If they were really getting a 18.3% recovery rate, one could probably live with that drop-out rate (if those people (i.e. the drop-outs) weren't long-term worse).

Could it be that a drop-out rate that is higher than the "recovery" rate in some way indicates participant dissatisfaction with the intervention?

 

[Dolphin]

Could it be that a drop-out rate that is higher than the "recovery" rate in some way indicates participant dissatisfaction with the intervention?

The drop-out rate of 27.8% could be telling us something interesting all right, and it would be interesting to ask them what happened.

I just don't like seeing the 18.3% recovery rate quoted as fact when its definition and the design of this study was so lax.

 

[Denise]

..

I just don't like seeing the 18.3% recovery rate quoted as fact when its definition and the design of this study was so lax.

I totally agree about the problems with definition and study design - sorry I did not note that - to me it was a given that those were of great concern.

Is there a move afoot to gather these questions and concerns and ask Chalder about them or is that a laughable idea?