Keith Laws
2 days ago
To return briefly to the underpinning ‘science’ for a moment… Dr Brurberg says that “A large number of trials have consistently shown that cognitive behaviour therapy (CBT) and graded exercise therapy (GET) may be supportive for patients with chronic fatigue syndrome/myalgic encephalomyelitis (CFS/ME) (Larun et al, 2016; Price et al, 2008).”
As the most recent – and therefore, influential – meta-analysis, I wanted to make a few brief comments on the Cochrane meta-analysis of CBT by Price et al …and Dr Brurberg’s reference to it
I was left asking myself… what is meant by the phrases: “large number” (of trials), ‘consistently’ and ‘supportive’?
Assuming that some answers might be found in Price et al., it is worth taking a look. The key headline message in the Price et al abstract states that CBT significantly reduces fatigue scores at post-treatment. The analysis consisted of 6 trials and it is the largest number of trials assessing any outcome presented by Price et al…so, I guess 6 is a ‘large number’
Are the findings consistent? Well, 2 trials found a significant CBT effect, but 4 were nonsignificant – are they consistent?….more consistently nonsignificant than significant?
In the broader context of the review, Price et al presented 44 separate (meta) analyses of various outcomes from 15 trials. Of those 44, only 9 analyses show better outcome for CBT than control (including 5 analyses that involved just a single study) – so, the majority of comparisons do not support CBT efficacy – In other words, approximately 1 in 5 of their analyses suggest CBT may be ‘supportive’
At the finer level of individual effect sizes, Price et al present 128 effect sizes – of which 37 are significant for CBT, while 91 are non-significant – so fewer than 30% of individual CBT effect sizes are significant
So, wherever we look – the majority of analyses, outcomes and effect sizes reported by Price et al do not obviously or clearly (in my view) point to CBT being supportive in CFS
Another issue concerns bias….of the 15 studies examined by Price et al, only 1/15 can be deemed ‘free of selective reporting’, only 3 are blind and none of the 15 are deemed free from ‘other bias’.
We might also note that the CBT for CFS trials assessed by Price et al are massively underpowered to detect the effects that they do propose as significant. For the main analysis on whether CBT reduces fatigue, the median power to detect true effects is remarkably low at .27 – although Price et al did not deem it necessary to remark upon poor study quality or the tiny sample sizes in their abstract or the Plain Language Summary (aimed at a wider audience) of their Cochrane review
Finally, Dr Brurberg raises the following interesting point
“It is worth noting that the effects of CBT and GET in CFS/ME are similar to those seen among patients with other serious diseases where fatigue is a prominent symptom, e.g. cancer (Furmaniak et al, 2016) and multiple sclerosis (Heine et al, 2015; van den Akker et al, 2016). The benefit of CBT and GET does not imply that we can conclude that cancer, multiple sclerosis or CFS/ME occur for psychological reasons. It is difficult to understand why the benefit of CBT and exercise in patients with cancer and multiple sclerosis seems widely accepted, whereas the usefulness of CBT and GET for patients with CFS/ME remains controversial.”
It seems quite commonplace amongst CBT advocates to make such a argument – it centres on what I would call ‘nominal analogy’ assumption
if CBT impacts symptom X (fatigue) in disorder A (Cancer), then it will also impact symptom X (fatigue) in disorder B (CFS/ME). This argument is based on the assumption that identifying symptom X in both ‘disorders’ means they are identifying the same ‘thing’ … the symptom is abstracted and decontextualised …and hence it often follows, that the same treatment is applicable and possibly equally efficacious
It only takes a moments reflection to see that argument holds no (logical) water….….For example, CBT may reduce the symptoms of depression in those diagnosed with depression; however, other evidence shows that CBT does not reduce the symptoms of depression in Bipolar Depression (see Jauhar, McKenna & Laws 2016). If we turn back to the current case – Dr Brurberg cites the Cochrane review as evidence of CBT efficacy- even here it fails to show that CBT reduces depressive symptoms in CFS/ME… calling something a dog does not make it bark