Wednesday, 21 September 2016

Key points from the analysis of the PACE trial raw data for “recovery”

Margaret Williams 21st September 2016

The following extracts are taken from:

“A preliminary analysis of ‘recovery’ from chronic fatigue syndrome in the PACE trial using individual participant data”.

(The article referred to includes: "The recovery rate is … 6.8% for cognitive behavioural therapy, 4.4% for graded exercise therapy, 1.9% for adaptive pacing therapy.")

by Alem Matthees (1), Tom Kindlon (2), Carly Maryhew (3), Philip Stark (4), Bruce Levin (5)

1. Perth, Australia.
2. Information Officer, Irish ME/CFS Association, Dublin, Ireland.
3. Amersfoort, Netherlands.
4. Associate Dean, Mathematical and Physical Sciences; Professor, Department of Statistics; University of California, Berkeley, California, USA.
5. Professor of Biostatistics and Past Chair, Department of Biostatistics, Mailman School of Public Health, Columbia University, New York, USA.

Key Points

1. There was no committee approval for the re-definition of “recovery”.

2. “Recovery” rates for CBT and GET were not statistically significant.

3. The PACE PIs originally reported “recovery” rates of 22% for CBT and GET.

4. The published “recovery” rates were based on thresholds that deviated substantially from the published protocol and were inflated by an average of four-fold.

5. In contrast to the published paper by the PIs, the recovery rates in the CBT and GET groups are not significantly higher than in the SMC (standard medical care) group alone.

6. APT (adaptive pacing therapy) was a highly modified version of “pacing” (preferred by patients).

7. 13% of participants at baseline simultaneously met the trial entry criteria for “significant disability” and the revised “recovery” criteria.

8. The Investigators excluded drop-outs, which is not recommended practice in clinical trials.

9. Logistic regression (used by the PIs) has been shown to be an inappropriate method of analysis in randomised trials.

10. The figures originally given by the PIs for the four groups were:

SMC 7% (but according to the protocol are 3%)
APT 8% (but according to the protocol are 2%)
CBT 22% (but according to the protocol are 7%)
GET 22% (but according to the protocol are 4%)

11. “Our findings therefore contradict the conclusion of White et al (2013) that CBT and GET were significantly more likely than the SMC group to be associated with ‘recovery’ at 52 weeks”.

12. “The multiple changes to the recovery criteria had inflated the estimates of recovery by approximately 2.3 to 5.1-fold, depending on the group, with an average inflation of 3.8-fold”.

13. When using the revised recovery criteria, 8% of the “recovered” participants still met trial eligibility criteria for “significant disability”.

14. “The changes made by the PACE investigators after the trial was well under way resulted in the recovery criteria becoming too lax to allow conclusions about the efficacy of CBT and GET as rehabilitative treatments for CFS”.

15. “This analysis, based on the published trial protocol, demonstrates that the major changes to the thresholds for recovery had inflated the estimates of recovery by an average of approximately four-fold”.

16. “It is clear from these results that the changes made to the protocol were not minor or insignificant, as they have produced major differences that warrant further consideration”.

17. “The PACE trial provides a good example of the problems that can occur when investigators are allowed to substantially deviate from the trial protocol without adequate justification or scrutiny”.

18. “It seems prudent that the published trial results should be treated as potentially unsound, as well as the medical texts, review articles, and public policies based on those results”.


Margaret Williams 14th September 2016

The PACE trial was instigated and carried out mostly by a group of psychiatrists well-known for teaching that ME/CFS does not exist other than as an aberrant belief: their assumption was that ME/CFS is a behavioural disorder that is amenable to behavioural interventions. The Investigators had no evidence for their assumption and despite abundant scientific evidence to the contrary, it remained their firmly-held belief. They favoured two interventions in particular: cognitive behavioural therapy (CBT), which was to “change the behavioural and cognitive factors assumed to be responsible for perpetuation of the participant’s symptoms and disability” and graded exercise therapy (GET), which was to correct the assumed deconditioning resulting from avoidance of activity.

The original (selective) results on the PACE trial were published in The Lancet in early 2011; they were accompanied by press releases from The Medical Research Council, King’s College London and Queen Mary University of London, all of which proclaimed: “Two effective treatments benefit up to 60 per cent of patients with CFS/ME”. Importantly, this figure was achievable only because the Investigators used a much less demanding definition of improvement than they had stated in their published protocol.

Following lengthy Freedom of Information (FOIA) requests, all of which were refused until the final one, the raw data from the PACE trial had to be released, following which the Investigators re-analysed their data according to their own published protocol.

Those results were different from what had been published in The Lancet to such loud acclaim (orchestrated by the Science Media Centre, of which Professor Sir Simon Wessely, one of the PACE team, was a founder member).

It revealed that the improvement figure was only 21% for the GET group and 20% for the CBT group versus 10% for those who received usual medical care alone, meaning that for every ten people treated with CBT or GET, only one person would show protocol-defined improvement. All participants received what was described as standardised “specialist” medical care (SSMC), but those receiving SSMC alone may have seen the Fatigue clinic doctor only three times for 30 minutes each time during their participation in the trial, a total of 90 minutes throughout the trial.

Hence the protocol-specified figures are that CBT and GET helped only an additional 10% of participants over usual medical care and not the widely reported 60%.

To read the rest of this article, please click –

No comments: