How to (re)create the illusion that psychotherapy extends the life of cancer patients

  • I provide a quick analysis of a story summarizing a peer-reviewed paper that did not encourage me to take a look at the paper.
  •  Life is too short, and there is just so much dubious stuff out there to devote much time pursue tracing claims that don’t pass a first screen.

smile or dieThe British Psychological Society Digest recently flagged one of its stories as the most accessed in some unknown period of time. I gave it a look. The story aroused skepticism and discouraged me from looking further at the target article which appeared in Psychology & Health. 

Oh, P., Shin, S., Ahn, H., & Kim, H. (2016). Meta-analysis of psychosocial interventions on survival time in patients with cancer Psychology & Health, 31 (4), 396-419 DOI: 10.1080/08870446.2015.1111370

P & H is not a journal where I would expect to find a high quality meta-analysis. Sorry, but I think that busy readers need to evaluate where articles are published and decide whether it is worth the trouble to access them, particularly when they are behind a paywall.

Here is my brief analysis of the BPS Digest story. I welcome readers to compare and contrast what is said in the Digest to what I have said in a previous blog post, as well asanother post discussing how well-designed negative studies get left out of reviews. The conclusions of my 2007 systematic review have not been contradicted by more recent studies.I recommend it as well. This not an area of research where new findings are expected to unseat what we conclude from older studies.

I would welcome comments about either the BPS story or the Psychology & Health article for the comparison/contrast with my blog posts and systematic review linked below.

I also invite speculation as to why the myth of psychotherapy and support groups extending the lives of cancer patients so resists being laid to rest.

A quick look at ‘Can psychosocial interventions extend the lives of cancer patients?’

Not only is the evidence for the benefits of psychosocial interventions extremely mixed, but some cancer patients and their relatives have understandably railed against the “cruel” suggestion that they might live longer if only they looked on the bright side and didn’t get so stressed.

Hmm, by emphasizing that  it is “cancer patients and their relatives” who object, does that mean we should take the objections less seriously and even ignore the many scientists who are objecting? And what of the suggestion that claims that psychotherapy extend the lives of cancer patients ultimately lead to blaming patients for succumbing to what is a debilitating and often fatal disease?

The researchers, led by P.J. Oh at Sahmyook University, found over 4,000 studies that looked promising, published between 1966 and 2014. However, once the researchers included only those papers that were randomised controlled trials and that included interventions delivered by professionals and had data on patient survival times, they were left with just 15 suitable studies conducted in five different countries and involving a total of 2940 participants with an average age of 52 years. The studies involved different types of intervention including psychoeducational programmes, CBT and supportive-expressive groups (a form of psychodynamic psychotherapy). The patients in the studies had a range of different cancers at different stages, including breast cancer, gastrointestinal cancer and melanoma.

Hmm, the problem in the small number of studies, is great heterogeneity in the interventions and patient populations suggesting that combination in meta-analysis is not appropriate because of a lack of clarity in which interventions for which populations the summary effect size would generalize we have a serious problem here. Furthermore, the authors selected many of the RCTs did not have “extension of life as a primary outcome,” and there is a strong bias in which studies follow up data was collected.

Looking at all the data from all 15 studies, there was no evidence that psychosocial interventions prolong the lives of cancer patients. However, because of the huge variation between the studies in terms of the interventions and the types of patient, the researchers also broke down the evidence into sub-categories and here the picture was more promising.

Hmm, a meta-analysis that finds no evidence for an effect on survival simply confirms past systematic reviews. Why continue post hoc, particularly when number of studies in these “subcategories” are of necessity so small?

For example, by excluding six studies that had exclusively involved patients with late-stage cancer, the researchers found that psychosocial interventions reduced the likelihood of patients dying during the course of the study (follow up times varied from one to 12 years) by 27 per cent, on average. “Stress reduction, if that is the causal mechanism, may have to occur earlier to achieve positive results,” the researchers said.

Hmm, the idea that “stress reduction”is a causal mechanism is a big leap, given this is a small post hoc analysis of a literature that previously has been evaluated in terms of no effect on survival. If there is  no effect size to explain, no explanation needed.

Other details to emerge included the finding that a positive benefit of psychosocial interventions was only apparent for studies involving patients with gastrointestinal cancer, although there was too little data to speculate as to whether this finding is meaningful or a chance result. Comparing the different types of intervention, the strongest positive evidence was for one-on-one programmes compared with groups, and for psychoeducational approaches delivered by medical doctors and nurses, as opposed to psychologists or other non-medical professionals.

Hmm, now we are really getting far out in focusing on results only for studies of patients with gastrointestinal cancer. Is there anything really here to explain that is not a post hoc rationalization? Him why would we expect results only for this particular kind of cancer?

Psychoeducational interventions involve health education, coping skills training, stress management and psychological support, and the researchers speculated their benefit might arise through a mixture of reducing patients’ distress, encouraging healthy behaviours and treatment compliance. “In addition, supportive social relationships might buffer the effects of cancer-related stress on immunity and thereby facilitate the recovery of immune mechanisms and may be important for cancer resistance,” they said.

Hmm, I think I recognize that study.

My colleagues and I thoroughly debunked its claims that did not show up in simple analysis, but emerged only on dubious multivariate analyses. The authors refused to respond to an invitation by the journal to respond to my criticism. I then requested the data from the authors to validate independently  their conclusions. The authors refused, and the funding agency, National Cancer Institute said it had no ability to enforce data sharing, but that’s a story for another blog post. I am further confident that there is no evidence that recovery of immune function through psychological intervention has ever been shown to increase survival. The claims depend on cherrypicking the most promising from a broad array of measures of immune functioning in a context where it is just as easy to obtain many as one. Particular positive results are not replicated across studies. There is no evidence linking changes in these immune measures to changes in survival. We are dealing with unvalidated claims of a surrogate outcome.

 Critics may question whether it is reasonable to combine results from such varied studies as was done in the current meta-analysis, and the researchers acknowledged that many of the studies were not as robust in their methodology as they should be. However, they end their review on a positive note: “a tentative conclusion can be reached,” they said, “that psychosocial interventions offered at early stage may provide enduring late benefits and possibly longer survival.”

Hmm, this sounds dismissive of the critics. Who are they?  What is their evidence?  Why should this review end on such a positive note, rather than giving up on a literature that has consistently generated null findings? Why doesn’t the author of this story provide the names of critics and allow the readers to evaluate the contrasting conclusion?