Psychosocial interventions for fatigue during cancer treatment with palliative intent (Cochrane Review), 2017

Psychosocial interventions for fatigue during cancer treatment with palliative intent (Cochrane Review), 2017

Hanneke Poort, Marlies Peters, Gijs Bleijenberg, Marieke FM Gielissen, Martine Margaretha Goedendorp, Paul Jacobsen, Stans Verhagen, Hans Knoop

https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD012030.pub2/full

Abstract

Background
Fatigue is a prevalent and burdensome symptom for patients with incurable cancer receiving cancer treatment with palliative intent and is associated with reduced quality of life. Psychosocial interventions seem promising for management of fatigue among cancer patients.

Objectives
To assess the effects of psychosocial interventions for fatigue in adult patients with incurable cancer receiving cancer treatment with palliative intent.

Search methods
We searched the following databases: CENTRAL, MEDLINE, Embase, CINAHL, PsycINFO, and seven clinical trial registries; we also searched the reference lists of articles. The date of our most recent search was 29 November 2016.

Selection criteria
We included randomised controlled trials that compared psychosocial interventions in adults aged 18 years or over undergoing cancer treatment with palliative intent for incurable cancer versus usual care or other controls. Psychosocial interventions were defined as various kinds of interventions provided to influence or change cognitions, emotions, behaviours, social interactions, or a combination of these.

Psychosocial interventions of interest to this review had to involve at least two interactions between the patient and the care provider in which the care provider gave the patient personal feedback concerning changes sought by these interventions. We included trials that reported fatigue as an outcome of interest.

Data collection and analysis
We used standard methodological procedures expected by Cochrane. Two review authors independently considered trials for inclusion in the review, assessed risk of bias, and extracted data, including information on adverse events.

We assessed the quality of evidence using GRADE (Grading of Recommendations Assessment, Development, and Evaluation) and created a 'Summary of findings’ table.

Main results
We identified 14 studies (16 reports) that met inclusion criteria for this review and involved 3077 randomised participants in total. Most of these studies included a mixed sample of participants; we obtained data for the subset of interest for this review (diagnosis of incurable cancer and receiving cancer treatment) from the study investigators of 12 studies, for which we included 535 participants in the subset meta‐analysis for fatigue post intervention.

Researchers investigated a broad range of psychosocial interventions with different intervention aims and durations. We identified sources of potential bias, including lack of description of methods of blinding and allocation concealment and inclusion of small study populations.

Findings from our meta‐analysis do not support the effectiveness of psychosocial interventions for reducing fatigue post intervention (standardised mean difference (SMD) ‐0.25, 95% confidence interval (CI) ‐0.50 to 0.00; not significant; 535 participants, 12 studies; very low‐quality evidence).

First follow‐up findings on fatigue suggested benefit for participants assigned to the psychosocial intervention compared with control (SMD ‐0.66, 95% CI ‐1.00 to ‐0.32; 147 participants, four studies; very low‐quality evidence), which was not sustained at second follow‐up (SMD ‐0.41, 95% CI ‐1.12 to 0.30; not significant; very low‐quality evidence).

Results for our secondary outcomes revealed very low‐quality evidence for the efficacy of psychosocial interventions in improving physical functioning post intervention (SMD 0.32, 95% CI 0.01 to 0.63; 307 participants, seven studies).

These findings were not sustained at first follow‐up (SMD 0.37, 95% CI ‐0.20 to 0.94; not significant; 122 participants, two studies; very low‐quality evidence). Findings do not support the effectiveness of psychosocial interventions for improving social functioning (mean difference (MD) 4.16, 95% CI ‐11.20 to 19.53; not significant; 141 participants, four studies), role functioning (MD 3.49, 95% CI ‐12.78 to 19.76; not significant; 143 participants, four studies), emotional functioning (SMD ‐0.11, 95% CI ‐0.56 to 0.35; not significant; 115 participants, three studies), or cognitive functioning (MD ‐2.23, 95% CI ‐12.52 to 8.06; not significant; 86 participants, two studies) post intervention.

Only three studies evaluated adverse events. These studies found no difference between the number of adverse events among participants in the intervention versus control group.

Using GRADE, we considered the overall quality of evidence for our primary and secondary outcomes to be very low. Therefore, we have very little confidence in the effect estimate, and the true effect is likely to be substantially different from the estimate of effect.

Limitations in study quality and imprecision due to sparse data resulted in downgrading of the quality of data. Additionally, most studies were at high risk of bias owing to their small sample size for the subset of patients with incurable cancer (fewer than 50 participants per arm), leading to uncertainty about effect estimates.

Authors' conclusions
We found little evidence around the benefits of psychosocial interventions provided to reduce fatigue in adult patients with incurable cancer receiving cancer treatment with palliative intent.

Additional studies with larger samples are required to assess whether psychosocial interventions are beneficial for addressing fatigue in patients with incurable cancer.

 

From the section "How the intervention might work":

"Although research has provided empirical support for the efficacy of psychosocial interventions for fatigue (irrespective of the presence of a medical condition), knowledge about the therapeutic mechanisms of these interventions is lacking.

"Limited work has been done in the field of cognitive‐behavioural therapy (CBT) for patients with medically unexplained fatigue (i.e. chronic fatigue syndrome (CFS)).

"Mediation analysis of CBT for CFS has revealed that changes in both beliefs and behaviours can mediate the effects of CBT (Chalder 2015; Wiborg 2011; Wiborg 2012).

"Mediation analysis of CBT for patients with multiple sclerosis shows that the decrease in fatigue may be explained by a change in beliefs about fatigue (Knoop 2012).

"The scarcity of knowledge about therapeutic mechanisms is even more evident for interventions that reduce fatigue in patients with cancer. Although CBT was found to be effective for reducing post‐cancer fatigue (Gielissen 2006) and is now recommended in ASCO clinical practice guidelines for cancer‐related fatigue (Bower 2014), the mechanisms of change remain unknown, and effects on fatigue have not been shown to be mediated by an increase in objective physical activity or fitness (Gielissen 2012; Prinsen 2013).

"To permit unequivocal conclusions about the therapeutic mechanisms of psychosocial interventions that may produce a reduction in fatigue, further research is needed."

 

Don't know when I will have the capacities to write more why I think this Cochrane Review, even though from 2017, deserved a comment.

Here's point No. 1:

I think this review is a good example of how authors who state that "research has provided empirical support for the efficacy of psychosocial interventions for fatigue (irrespective of the presence of a medical condition)" rely on studies that have been heavily disputed or earned criticism for making conclusions that aren't conclusive with what they themselves wrote about the limitations of their studies at other parts of their papers.

Also, often they mainly reference their own work.

In the part quoted in the post above, the authors reference their work more often than it seems at a first glance -- Knoop is also a co-author on Wiborg 2011&2012, and Blijenberg is co-author of Prinsen 2013, and Paul Jacobsen co-author of Bower 2014.

So from the eight references in the quote only one reference is a paper not co-authored by one of the review authors.

Edited: Referenced papers' co-authors.

 

I guess they would reply you should look at all the research that supports the potential effectiveness -- research that they've done themselves or by colleagues they are strongly affiliated with.

It's problematic that Cochrane invites authors to do a systematic reviews in a research area where the same authors are among the main contributors to the evidence base.[*] So that would be reason enough to criticize the review.

But that seems what these authors' research mainly is about anyway -- reviewing their own work, skipping the criticism, so they always will either find more evidence in support of their claims or reasons to do more of the same badly designed research -- and thereby produce more and more references allegedly justifying their research.

If you check the references, they often don't actually support what the authors suggest.

[*] Edited to add: The review authors weren't involved in any of the studies considered for the actual meta-analysis. (They only co-wrote some other papers with some of the studies' authors.) However, they justify the need of further research in this area and their suggestion that properly powered studies would support the evidence for a benefit from 'psychosocial' intervention by referencing their own work while they already work on exactly this kind of research.

 

The review authors mention two relevant ongoing studies:

"We identified two studies that have not been completed (Poort; Serfaty)"

See Characteristics of ongoing studies table.

Five of the review authors (among them Knoop and Bleijenberg ) were investigators of one of these two ongoing studies. (*) This information is added in the authors' "declaration of interest" where they at the same time state "none known".

The other ongoing study they identified as relevant (CanTalk) -- with no apparent involvement of the Cochrane review authors -- was on depression, not explicitly on fatigue.

Both studies have been published by now.

The CanTalk study authors (not involved in the Cochrane review) conclude:

- "Although it is feasible to deliver CBT through IAPT proficiently to people with advanced cancer, this is not clinically effective".

See @dave30th 's take on it here, forum thread here.

And what do you guess did the authors of the "TIRED study" by Poort, Bleijenberg, Knoop et al find out?

Haven't read their paper yet so can't assess what they actually found out, but that's what they conclude:

"Among advanced cancer patients with severe fatigue during treatment, a CBT intervention was more effective than usual care for reducing fatigue. Following GET, patients reported lower fatigue, but results were not significant, probably due to a smaller sample size and lower adherence than anticipated."

Study paper:

Poort H, Peters MEWJ, van der Graaf WTA, Nieuwkerk PT, van de Wouw AJ, Nijhuis-van der Sanden MWG, Bleijenberg G, Verhagen CAHHVM, Knoop H. Cognitive behavioral therapy or graded exercise therapy compared with usual care for severe fatigue in patients with advanced cancer during treatment: a randomized controlled trial. Ann Oncol. 2020 Jan;31(1):115-122. doi: 10.1016/j.annonc.2019.09.002. PMID: 31912784.

https://www.annalsofoncology.org/article/S0923-7534(19)35403-1/fulltext

 

That's what I thought, too.

Added to my previous post:
The review authors weren't involved in any of the studies considered for the actual meta-analysis. (They only co-wrote some other papers with some of the studies' authors.) However, they justify the need of further research in this area and their suggestion that properly powered studies would support the evidence for a benefit from 'psychosocial' intervention by referencing their own work while they already work on exactly this kind of research.

Sadly, the references they quote (mostly their own work and that of Sharpe, Chalder and Moss-Morris) seem to be accepted as established knowledge by the majority of people in the research area of fatigue, both in ME and as a symptom or (complex of symptoms) in non-disputed disease.

 

2.) The review authors acknowledge that the evidence from the current state of research is too uncertain to conclude anything about the effectiveness of "psychosocial" interventions in incurable cancer. It's unclear however, why they did a meta-analysis at all.

The reasons they give for downgrading all outcomes to "very low" quality of evidence must have been obvious from reading the papers.

• "The main limitation of this review involves the relative lack of data in this field. Six studies consisted of a homogenous sample of patients with incurable cancer. The remaining eight studies comprised a mixed sample of potentially curable and incurable patients. As a result, interventions from these eight studies were not specifically tailored to patients with incurable cancer. Yet, tailoring of psychosocial interventions could be important in achieving intervention effects [...] Although investigators from these mixed‐sample studies were able to provide data for the subset of incurable cancer patients, the sample sizes of these subsets were quite small. This is likely to result in lack of power to detect treatment effects that may arise from the psychosocial interventions."
• 
  
• "We identified a limited number of studies (six) investigating interventions specifically aimed at addressing fatigue. However, only two of these six studies reported that the presence of some level of fatigue was an entry criterion for trial participation. This may lead to floor effects, restricting the potential range of fatigue scores and resulting in less room for improvement."

• "Furthermore, a specific feature of data available from identified studies was the heterogeneity of intervention and follow‐up durations. In addition, only four of the 12 studies contributed to findings on follow‐up effects."

• "Among the four studies included in the primary analysis for follow‐up effects, we noted considerable attrition between post‐intervention and first follow‐up assessments (attrition rate 18.5%). [...] the attrition is unlikely to be random. In fact, attrition may be associated with deterioration of health or death of the participant. This has implications for interpretation of follow‐up findings; thus these results should be interpreted with caution."

• "Finally, in two studies (Sharpe 2014; Walker 2014), the investigated population had received a diagnosis of major depressive disorder comorbid with cancer. Fatigue outcomes in these participants may have been associated with this depression, making it difficult to distinguish fatigue as a symptom of depression from cancer‐related fatigue."

These I think are all important points in assessing the evidence base. But to me it seems it would have been sufficient to write a literature review on the state of the research, and on that basis decide that a meta-analyzis won't make sense.

 

3.) More a question:

That’s how the review authors deal with the impossibility to blind patients and treatment providers to the intervention:

"Blinding of outcome assessment (checking for detection bias):


"This is usually assessed by looking at the methods used to blind study participants and personnel from knowledge of which intervention a participant received.


"However, in RCTs investigating effects of psychosocial interventions, it is impossible to blind care providers to the intervention they are giving to participants. It is also nearly impossible to blind participants to the intervention to which they were assigned.


"We judged risk of bias in blinding of outcome assessment on whether measures were administered and collected by an assessor who was blind to treatment allocation.


"We assessed methods as having low risk of bias (study states that outcome assessment was blinded and describes the method used to achieve blinding); unclear risk of bias (study states that it was blinded but does not provide an adequate description of how this was achieved); or high risk of bias (studies states that outcome assessors were not blinded)".

The review authors only assessed subjective outcomes (I’m too brain-fogged now to check whether the studies also uses objective outcomes). If the trialists collecting and assessing the outcomes where blinded, that was sufficient to grade the quality of evidence from these open label studies as having a low risk of bias – no matter whether the studies used objective outcomes or not.

Is it common practice in assessing the evidence quality in this kind of intervention studies to only check “whether measures were administered and collected by an assessor who was blind to treatment allocation“, but not check whether objective outcomes were used?

 

One last addition to the first point -- how the review authors use references to justify their own past and future research.

It appears to be a kind of double cherry-picking. They do not only cherry-pick references but the references themselves are made of cherry-picking and a failure to properly respond to criticism.

Apologies I won't back up this with more than one example -- I now don't remember which references I checked in addition to the papers by Chalder et al. I just remember that so many people with ME, as scientists or citizen scientists (among them the late Graham and the late Robert Courtney) never got proper replies to their valid criticism on the PACE authors' publications. Even if the authors replied to some criticism, they mostly failed to actually reflect on their work and provide appropriate answers.

That failure to genuinely engage with criticism on the PACE papers and their authors' allies continues until today. But it's becoming harder and harder to miss their failure thanks to quite a few people documenting it in blogs, open letters, Tweets, and also peer-reviewed publications.[*]

(Continued in the following post.)

[*] If you are new to this topic -- the first address to go to is David Tuller's 'Trial by Error' series on Virology Blog: https://www.virology.ws/mecfs/

There are many more blogs and also Twitter accounts that I'd like to recommend with this regard -- should I start a new thread where we could collect them?

 

So, only one example for this double cherry-picking for now.

The review authors state: "Mediation analysis of CBT for CFS has revealed that changes in both beliefs and behaviours can mediate the effects of CBT."

The references for that statement are Chalder 2015; Wiborg 2011; Wiborg 2012.

I leave it to others to check Wiborg 2011+2012.

Chalder 2015 is: Chalder T Goldsmith KA White PD Sharpe M Pickles AR (2015), Rehabilitative therapies for chronic fatigue syndrome: a secondary mediation analysis of the PACE trial.[1]

This analysis by the PACE trial investigators received much criticism.

The Lancet published the very balanced response by @Simon M : 'Omission of data weakens the case for causal mediation in the PACE Trial' [2]

Simon made diverse detailed points why the authors' conclusions "might be premature". He acknowledges that the study showed an "association between putative mediators and outcomes" but points out that it didn't demonstrate "the causal link central to establishing mediation."

In their reply, the authors deal with this critique in just two lines:

"McGrath reiterates the limitations of this mediator analysis, which we have already acknowledged in the Article.2
We intend to further explore multiple mediator effects and the potential effects of outcomes on mediators in future papers."

Other criticism addressed the problem of using only objective outcomes in unblinded trials.

The authors reply:

"Some correspondents suggest that objective outcomes are more important than self-report measures. In view of the subjective nature of fatigue and the overriding importance of the patients own judgement of their functioning, we believe that self-rated outcomes remain the most important ones." [3]

They add:

"Although it is entirely understandable that some patients with chronic fatigue syndrome are cautious about engaging in activity and exercise, increasing those activities gradually offers both improved functioning and reduced fatigue in some patients. Our findings also suggest that better targeting of fearful beliefs in treatment may result in better outcomes."[3]

[1] Chalder T Goldsmith KA White PD Sharpe M Pickles AR (2015), Rehabilitative therapies for chronic fatigue syndrome: a secondary mediation analysis of the PACE trial, Lancet Psychiatry. 2015; 2: 141-152, https://doi.org/10.1016/S2215-0366(14)00069-8

[2] McGrath, Simon (2015), Omission of data weakens the case for causal mediation in the PACE Trial, The Lancet Psychiatry, Volume 2, Issue 4, e7 - e8 , https://doi.org/10.1016/S2215-0366(15)00085-1

[3] Chalder, Trudie et al. 2015, Methods and outcome reporting in the PACE trial–Author's reply, The Lancet Psychiatry, Volume 2, Issue 4, e10 - e11, https://www.thelancet.com/journals/lanpsy/article/PIIS2215-0366(15)00114-5/fulltext