Spun headlines: CBT and GET are the most cost-effective and should become standard offerings.
Frank assessment: CBT and GET are essentially useless for reducing overall service costs and employment losses, fail to prevent the increased use of welfare benefits and other financial payments.
About time, it took nearly 18 months after the 1st publication! At
BACME 2012, White highlighted the supposed irony in patient organizations' comments about how the PACE findings that CBT/GET are safe and effective was "bad news for patients". Similarly, although I can understand why people are disheartened about how the latest PACE results are being presented, this new paper is partly "good news for patients" for the following reason:
CBT or GET do very little to prevent loss of employment hours and do not significantly help get people off welfare and insurance benefits etc, or prevent people from getting these in the first place. Coincidently, out of the
BBC article, the
MRC press release, the
Independent article, and numerous other articles parroting the same content, suspiciously none of them mention these failed outcomes, so it makes me wonder why not. Despite the spin in the above mentioned news articles about the success and cost-effectiveness of CBT and GET, I doubt the relevant authorities will be motivated much to invest in a major roll-out of CBT/GET just because of cost-effectiveness analyses based on small improvements in self-reported symptoms and quality of life which are largely contradicted by objective measures. Since when did these authorities give a flying toss about our quality of life anyway? Their main interest was getting people off disability/insurance payments and back into work, and this isn't really happening. So they cannot look to the PACE Trial for an excuse to start disingenuously "cracking down" on payments or forcing people into CBT/GET before getting payments. If the PACE Trial demonstrated clear benefits for employment/disability outcomes, which it frankly did not, the implications for patients on such payments would have been much different and this would have been featured in the headlines instead.
The data from Table 3 and Table 4 and Table 5 is sobering, total societal costs remain high and are barely reduced by CBT and GET vs SMC before even accounting for the extra cost of CBT and GET infrastructure and sessions. Overall there is no obvious advantage for CBT and GET in total service costs. Perhaps they emphasized the QOL/symptom based analyses to distract away from these otherwise disappointing results. These therapies are only "cost-effective" because, although they don't do much to reduce overall costs, they are relatively very cheap compared to the costs being incurred due to CFS. Consider the following crude analogy:
For treating CFS, there is no such thing as high quality candy (reasonably effective treatment). We do however have very cheap candy (CBT/GET) which a small proportion of patients find tastes slightly more pleasant (small improvement in subjective measures) than plain chalk bread (SMC). Therefore, enough very cheap candy for enough people will generate the same amount of taste goodness as high quality candy and at a reasonable cost too, therefore, very cheap candy is "cost-effective" at producing pleasant taste even if most patients do not like the taste better than plain chalk bread. Studies by candy proponents invested in promoting or selling candy report that patients do not dislike the taste (adverse effects), although external surveys of candy eaters suggest that many indeed dislike the taste. Candy proponents claim that there is nothing wrong with the candy and the people that don't like it aren't eating it properly.
Despite the spin this is overall bad news for PACE, they had a "soft" win (cost-effectiveness based on QOL) but a "hard" loss (poor employment/disability/insurance outcomes). These results are unsurprising, but I was sort of expecting a small effect, either because of a genuine but small effect on Oxford criteria patients, or because the milder affected or borderline-disability cases in the trial could at least be encouraged to delay their application for benefits or temporarily reduce them at least during the course of the trial. Good work to those already questioning how the cost-effectiveness was calculated and how such measures could be biased.
Another point to consider is that despite the increasing welfare/insurance payments to CBT participants over the course of the trial, the [Work and social adjustment scale] in the 2011 Lancet paper was the only secondary measure where CBT reached the authors definition of clinically useful difference (0.5SD of baseline scores). This scale is described in the 2007 protocol as a "comprehensive measure of participation in occupational and domestic activities", and IIRC Chalder has even published a paper validating its use for CFS. So one wonders what is actually being measured there.
In one of his presentations, White seemingly made out that the Ross et al 2004 paper showed CBT/GET was effective at "restoring the ability to work", although the actual data behind the statement was weak upon further investigation. They are going to have trouble spinning the latest results in their favour, and probably would have done so already if possible, that explains why employment/disability data was buried in the paper as lesser importance (they were obliged to report it eventually). IIRC they did not even factor these payouts into the cost analysis! Now we just need the data on recovery, other secondary outcomes, and mediating factors, so we can see how much the house of cards has fallen.
It is no surprise that only several days ago, the MRC announced additional funding for other research, in addition to the announcement late last year after much drought. This change and the PACE Trial itself may mark the beginning of the end for the cognitive behavioural "model" of CFS, although previous data already made it look unstable.