Patients with CFS remain severely disabled after treatment with graded exercise therapy in a specialist clinic in the UK Response to Smakowski et al.

* This article refers to: Graded exercise therapy for patients with chronic fatigue syndrome in secondary care – a benchmarking study (https://www.tandfonline.com/doi/full/10.1080/09638288.2021.1949049).

*

Dear Editor,

We read the study "Graded exercise therapy for patients with chronic fatigue syndrome in secondary care – a benchmarking study" [1] with interest. However, we feel the authors’ conclusions that "Graded exercise therapy for patients (GET) is a safe and efficacious treatment for patients with CFS/ME in a clinical specialist environment" misrepresents the data.

Firstly, the conclusion that GET is safe is unsupported as adverse events were not formally assessed, or if they were assessed, no information about how it was done nor supporting data are presented. This is an important omission in light of a recent survey by Oxford Brookes University carried out for the NICE review process in which over 80% of patients (n = 2274) [2] reported worsened symptoms after GET.

There are further important omissions which we believe calls into question the authors’ assertion that GET is efficacious. Firstly, only the subjective data (which is inherently open to sociability/desirability bias) are presented; the objective data (from the One-minute sit to stand test and two-minute Walk Test) are missing. This could be considered selective reporting, given that previous trials (including the PACE trial of which the corresponding author for this paper (Professor Chalder) was a co-author) have shown GET does not improve objective measures of activity [3].

Adherence data are also ‘generously` presented. The authors state that only 21 patients (i.e., 23%) did not complete the optimal number of therapy sessions, however examination of the data in the Supplementary material [1] shows that actually 40% (37/95) of patients attended less than the standard number of treatment sessions and attendance for 11% of patients was unaccounted for. It would be reasonable to assume that data were missing for these patients because they dropped out of treatment before assessment was complete. Thus, less than half of patients completed the treatment programme. Such poor adherence indicates that GET is poorly tolerated and unacceptable to patients. This would certainly be the conclusion if such data were reported in a research trial. The authors speculate that patients who did not complete the treatment sessions “may have done well with treatment and reached asymptomatic status faster”. However, they do not present evidence to support this assertion (although it would have been simple to do so). In view of the aforementioned majority of patients who deteriorated after GET, it is more likely that patients dropped out because their treatment was intolerable and/or their symptoms worsened. This explanation is supported by the outcome data (measured at the end of treatment or three month follow up appointment). At discharge/follow-up, the mean scores for the main outcomes were

• Physical functioning (SF-36) = 54.4 (standard deviation or SD 29.7) NB a score of 65 or less equates to severe disability [4]

• Fatigue (Chalder Fatigue Questionnaire) = 20.80 (SD 8.17)

• Work and Social Adjustment Scale (WSAS) = 21.48 (SD 11.23) NB A score >20 indicates moderately severe disability [5]

Putting these figures into perspective shows the following: After treatment most patients remained severely disabled, and in fact would have been sufficiently ill to enter the infamous PACE trial. Although some patients may have improved, few, if any, became asymptomatic and most remained severely disabled.

Another unacknowledged limitation of the study is the high degree of missing data; data were available for only one-third (36% for the measure of fatigue, 35% for the measure of physical functioning) of patients at the main outcome assessment (end of treatment). In a research trial, such a high degree of missing data (64–65%) would mean that any analysis should be treated with extreme caution, if attempted at all. The authors used the data from the 3-month follow-up appointment when it was missing at the primary end point (end of treatment) which gives the impression of more data (and thus acceptability to patients) than was actually the case. The authors also created more data for analysis by calculating a prorated score “using the mean of the remaining items … if a patient did not complete a discharge questionnaire”. Using a mean score would artificially inflate scores if a patient was deteriorating.

Finally, the authors are selective in the other evidence to which they refer to support their conclusion that GET can be an effective treatment for ME/CFS. For example, they do not put their findings in the context of the aforementioned survey showing that 81% of ME/CFS patients deteriorate after GET [2]. Furthermore, although reference is made to the 2019 amended Cochrane Review of exercise for ME/CFS by Larun et al. [6], they do not mention that this review was subsequently reanalysed. The independent reanalysis highlighted how the amended Cochrane Review had not addressed the problems identified. For example, just like Smakowski et al., analysis of objective data was unreported, and the reanalysis showed no improvement after GET, further strengthening the impression of selective reporting in both publications. Larun et al. also did not report the lack of evidence that GET is safe and not harmful [7]. In response to Larun et al.’s inability to address the shortcoming of the review, The Cochrane Library subsequently announced that the review is being overhauled [8]. Most importantly, no mention is made of the recently published NICE Guidance [9] which states that any programme “which uses fixed incremental increases in physical activity or exercise, for example, graded exercise therapy” should NOT be used.

The authors conclude that the study shows GET therapy is safe and efficacious for patients with ME/CFS. However, as the authors noted themselves, the lack of a controlled comparison group limits the ability to conclude that any improvement is down to the treatment under investigation. This is an important limitation because many included patients had been ill for less than 2 to 3 years, which is when the chances of spontaneous recovery are the highest [10]. Some patients had only been ill for four months (mean illness duration 3.4 years, SD 0.4-34.0). Thus, our conclusion is that the data presented indicate that adherence to GET is poor and dropout rates are high, indicating that GET is poorly tolerated, and that after treatment most patients remain moderately –severely impaired and functionally limited. However, the lack of a control group and the demographics of the sample recruited, mean that spontaneous recovery cannot be ruled out for those who did improve. These findings support recent NICE clinical guidance that GET should not be used for people with ME/CFS.

Disclosure statement

The authors report no conflicts of interest.

Notes

* This article refers to: Graded exercise therapy for patients with chronic fatigue syndrome in secondary care – a benchmarking study (https://www.tandfonline.com/doi/full/10.1080/09638288.2021.1949049).