Thread by @keithgeraghty

Why data from PACE trial author and London GET clinic is dubious and does not reflect real ME/CFS. 2021 paper attempts to show GET benefits = gains made on fatigue scale happen in 1st 4 hours/sessions then dont continue to normal levels. tandfonline.com/doi/full/10.10…

92 patients data since 2002 (almost 20 years) including 20 drop-outs still used as data. To take an ME/CFS patient and have them drop fatigue by 4th session may be an act of changing perception. Why then do patients not seem to be returning to normal fatigue levels by the end?

This paper can only be judged as an attempt by a PACE author to sure up evidence for GET use in the face of the impending dropping of GET by NICE, by clubbing together data from as many patients as could be mustered from their clinic over 20 years, no treatment uniformity at all.

I will explain this, you take ambulatory patients, not sicker ME/CFS patients, you take a broad criteria to ME/CFS< only 70% in this cohort met the loose Oxford Criteria; you tell patient to change how they see their fatigue session 1-4 then measure! No comparison group at all.

Some patients had 2 session, some had 15 sessions of GET - making this paper redundant other than as a historical record of how GET is applied to patients in clinic.

Where is the sit to stand test results? again, like in the PACE trial itself, these athors pick out the data they want you to see whilst keeping other test results from us - why perform a sit to stand test and not include the results?

patients were offered between 12 and up to 18-20 sessions depending on if they took extras - so why then do many only make less than 12, some only 2 sessions (why did circa 35 patients not reach 12+ sessions out of the 92)?

SF-36 physical function scale - we see between GET session 1 and 7 (most of treatment) mean stays the same! only between 7 and end of treatment 12-20 sessions does it pop up a bit, but this may be the remainers, the weller patients, pre-GET session 12 almost 1/3 stop as per last

do we know how many patients actually made up the end of treatment or follow up data out of 92n - I cant see it, authors dont appear to include numbers of patients in these slides

so if I am right, even if we accept a rough figure of 60 patients were used for data on SF-36 to derive this follow up data, avg scores went from 45 to 55s, still far below normal - so where are the 22% of patients who should fully recover according to the PACE trial run by same

Found it - wait for it, follow-up data, whether pure of combined, was only available for 18 patients for SF-36 scores at 3 months follow up - in a clinic of a PACE trial author, which claims to have seen patients since 2002 treating them with GET. 18 patients data !

where is dose effect: take ME/CFS patients who are hypothesied tobe deconditioned and afraid of exercise, give 4 hours of GET and see benefit but not a dose effect with more GET over a year of thearpy? does not add up - other than initial treatment effect and other biases

Here is how the authors pondered the fact most gains happen between session/hour 1-4: they say this mirrors CBT, so whether its CBT or GET 1/4 of patients return to normal levels of fatigue within 4 hours - really? what sort of ME patients are these - 1-4 hours to cure MEfatigue?

To make my point even clearer lets look back at the nurse-led CBT-GET trial that also failed to cure ME/CFS-FINE Trial 2010. Note the same pattern, all 3 treatments, blue-line CBT-GET also the green usual GP care all 3 improved fatigue: no care improves fatigue? = treatment bias

Final statement: strong treatment effect biases; no controls; researchers with strong treatment bias; physical test data omitted; missing data manipulated statistically; high drop out rates; data on small numbers of patients; no external replication; an uncertain patient cohort.

Authors conclusion: we cant get the same big recovery results using GET in our clinic as we got in our trial eg PACE, we cant explain why; we still think GET is effective but very modest gains, we only have a small number of patients' data at 3 month follow-up n50 after 15 years.