Aug 03 2012
My science writing covers diverse topics but increasingly concerns two intertwined themes in cancer and psychology. First, I bring evidence to bear against an exaggerated role for psychological factors in cancer, as well as against claims that the cancer experience is a mental health issue for which many patients require specialty mental health interventions. Second, I explore unnoticed social and organizational influences and publishing practices, which limit evaluation of the best evidence for theories and practices claiming to be evidence based, especially those recommended (and even mandated) by professional organizations and accrediting bodies.
I benefit from a great set of international collaborators, and my colleagues and I have repeatedly debunked claims that psychological interventions increase the survival time of cancer patients by improving their immune systems. Wally Sampson and Bernie Fox provided important inspiration for these efforts. A key source of such claims is the classic Lancet study by David Spiegel, which I will dissect in a later post for ScienceBasedMedicine.org (for now, see our published critique of Spiegel).
When specifically asked, I recommend support groups for cancer patients who want to participate. Emotional support and validation are reasonable goals, but patients should not be misled into believing that they are extending their lives. Sadly though, many breast cancer patients do go to support groups based on the belief that they are bolstering their immune systems and thereby improving their chances of survival. Unfortunately, they can find lots of encouragement for this belief in media coverage of the claims of prominent psychiatrists and psychologists.
My colleagues and I have also questioned exaggerated estimates of the psychiatric disorder associated with cancer — comorbidity with mental health problems is no higher than for other chronic and acute physical illness — and the wisdom of professional guidelines recommending (and mandating) that all cancer patients receive routine screening for psychological distress and depression, if improvement in patient well-being and quality life is the goal. Oncologists who have been ignoring such guidelines — and even the most resourced cancer centers have not consistently implemented them — will soon be unable to close the electronic record for a particular patient until they have screened the patient for psychological distress and reported what interventions they provided if the patient is distressed. Pfizer gave a psychologist $10 million to develop a set of quality indicators that can be used to monitor electronic records and the American College of Surgeons and other groups are mandating documentation of screening as a condition for accreditation.
The evidence for screening and other aspects of psycho-oncology remains weak. Guidelines and mandates that are being promulgated are consensus-based, not evidence-based. My colleagues and I continue to point to the lack of evidence that mandated routine screening will improve patient well-being, and we continued to warn that implementing screening may even disrupt existing efforts to provide cancer patients with the support and clinical and practical information that they need to cope. At some cancer centers, after being identified as “distressed” by a screen, a patient may have to pay up to $200 for an interview with a mental health professional, and yet no psychiatric disorder is found in most cases. Most patients decline services. Why not let interested patients directly access the support and services they want without this expensive intervening step?
Our efforts to introduce evidence into discussions about practices that are branded as “evidence-based” have met with considerable resistance from individuals and professional organizations with a vested interest in promoting these practices. A number of professionals have refused to debate me in print or in symposia, and instead have written letters of complaint to my university’s administration or attacked me on the internet. At the same time, I’ve noticed that “evidence” consistent with the favored views — that psychotherapy improves survival, that psychiatric disorders are rampant among cancer patients, and that screening improves patient outcomes — is all met with uncritical acceptance, privileged access to publication in high impact journals without peer review, and protection from criticism, all no matter how weak the evidence may be. Sometimes the “evidence” has even been demonstrated to be fraudulent, as with the Eysenck/Grossarth-Maticek data denying a link between smoking and lung cancer because personality explains both smoking and lung cancer. In Eysenck and Grossarth-Maticek’s view, any apparent relationship between smoking and developing lung cancer is spurious. These data keep getting cited, demonstrating a persistent reluctance to exclude the fake data from consideration when they are impressively supportive of the view that mind matters in cancer. As my grandmother used to say in Italian, “Se non è vero, è bene trovato”, which she claimed to be translated as “If it is not true, it ought to be.”
Difficulty publishing embarrassing evidence or getting commentaries published about flawed reasoning and bad data has sensitized me to my second theme: the need to explore social and political influences on the availability, suppression or distortion, and interpretation of “evidence.” What is available in the literature and cited in practice guidelines and other official statements as “evidence” is continually artificially edited, constrained, and manipulated. We need to recognize that the claim of being “evidence-based” is an important marketing tool, and strong vested interests mobilize to protect that labeling of their practices from encounters with embarrassing evidence.
I cover these topics in my peer-reviewed papers, but also in my Skeptical Sleuth blog, which is designed to disseminate critical skills for interpreting both the scientific journals and media reports and to encourage a general skepticism towards these sources. If people believe everything they read in the journals, they shouldn’t be reading.
For my inaugural post here, I will touch on both of my major themes in discussing how the Journal of Clinical Oncology is not a reliable source for evidence about the relationship between cancer and the mind. JCO has a high journal impact factor of 18.97 for 2010 and a high rate of rejection, but this does not guarantee that what is published there is best evidence, particularly in the case of articles espousing a role for psychological factors in the incidence, progression and outcome of cancer. From time to time, I have challenged especially outrageous JCO articles, which are often accompanied by laudatory editorials. It is rather quixotic to do so and I’ve recently run afoul of unwritten editorial policies of JCO, whereby post-publication peer commentary can be squelched by authors whose articles are criticized. No matter how fundamental the flaw being identified, the author of an article is the final judge of whether the critique will be published.
Let’s start with the article for which JCO would not publish my commentary because the authors refused to reply. I co-authored the commentary with an eminent expert on posttraumatic stress disorder, Naomi Breslau and I have drawn on it with her permission and enthusiasm for my writing this blog.
Rampant Post Traumatic Stress Among Cancer Survivors? Not really
Smith SK, Zimmerman S, Williams, CS, et al: Post-Traumatic Stress Symptoms in Long-Term Non-Hodgkin’s Lymphoma Survivors: Does Time Heal? JCO, Dec 1;29(34):4526-33
This article claimed that more than a third of long-term survivors of non-Hodgkin’s lymphoma experience persistent symptoms of posttraumatic stress disorder. Clinicians need to be aware of this risk and screen survivors with standardized measures and provide treatments to alter survivors’ perceptions of the cancer experience in order to improve their long-term outcomes. The authors drew parallels between NHL survivors and victims of disaster and violence. They claimed their results provide “strong messages” and called for psychosocial treatments, including cognitive behavior therapy, as well as recognition that NHL patients from deprived backgrounds are at highest risk. They can concede, however, that “there were no known psychosocial interventions developed specifically for NHL survivors,” but suggest that other evidence-based treatments can be tailored to meet the unique needs of NHL survivors.
Critique. This article displayed some typical flaws found in psycho-oncology studies providing inflated estimates of psychiatric diagnoses from checklists administered to patients without involving a trained interviewer to explain to them what is being asked of them or probe their responses. Also typical is that diagnosis was made on the basis of complaints that are nonspecific and common across health conditions which give patients good reason to worry about their future. The PTSD checklist used in this study was validated with combat veterans who had been exposed to a war zone, but who had left. The stressor was securely in the past for these vets, unlike for cancer patients facing a high likelihood of recurrence or actual recurrences. A third of this sample of cancer patients had a recurrence before completing the questionnaire. Yet, the only complaints that were endorsed by at least a 1/6 of them as being moderately to extremely bothersome were: trouble falling or staying asleep, having difficulty concentrating, loss of interest in activities previously enjoyed, feelings of your future will somehow be cut short, being more watchful or on guard. These are not specific to PTSD, but are reasonable and expectable responses to actual or anticipated recurrence of cancer. More generally, these complaints become symptoms of PTSD only when they co-occur as part of a syndrome, not just in isolation, and when they can be linked to exposure to a past psychological threat. The authors did not establish a syndrome or a link.
If these cancer patients were actually suffering posttraumatic stress, the most empirically supported cognitive behavioral treatment would be exposure therapy, in which a patient is exposed to the traumatic situation with evidence that danger is no longer present, a luxury that current and former NHL patients do not have. I am skeptical that this treatment would benefit many NHL patients and that they would take up the offer of therapy until the threat of the disease was assuredly in the past.
It’s a reasonable recommendation that clinicians making a special effort to ensure that low-income patients have access to supportive services and that clinicians more generally make an effort to reduce the negative impact of the treatment and diagnosis of NHL on appearance, body image, and everyday functioning. But this recommendation is not derived from the data presented in this article and making a psychiatric issue of the patients’ predicament would not seem to help.
I am disappointed that JCO let these authors decide that our critique could not be published and I remain quite interested in how they would have responded.
Not Demonstrating That Talk Therapy Improves 10 Year Survival of Cancer Patients
Kuchler, T., B. Bestmann, et al. (2007). “Impact of psychotherapeutic support for patients with gastrointestinal cancer undergoing surgery: 10-year survival results of a randomized trial. JCO,25(19): 2702-8.
This article claimed that an average of just 222 minutes of psychotherapy produced a 10-year survival benefit among a mixed group of patients with gastrointestinal cancer. “Lots of bang for a few bucks” proclaimed an accompanying editorial. The editorial went on to concede the methodological problems of studies claiming effects of therapy on survival and that this study was no exception, but when “all is said and done and the appropriate methodologic mea culpas have been muttered, 10-year survival in the treatment arm was more than twice that of the control arm (21% v 9%).” The editorial ended with a call for more such studies, particularly given the low cost of a potentially life extending intervention.
Critique. Nonsense. Steve Palmer and I pointed out that preliminary reports of this clinical trial had designated quality of life as the primary outcome, but that no effect on quality of life was ever reported. The authors conceded that no effect on survival had been hypothesized. There was no manual guiding this therapy, but patients received an eclectic mix of supportive talk therapy, crisis intervention, and a small amount of relaxation training, with considerable time taken up with strategizing and rehearsing interactions with oncology physicians and nurses. The oncology treatment team was aware of which patients were in the intervention group and the therapists spent as much time meeting with the surgical and nursing teams as with the patients. As a result, the patients assigned to psychotherapy got considerably more medical treatment: almost twice as many days of intensive care, and they were twice as likely to receive postoperative chemotherapy, five times as likely to receive radiotherapy, three times more likely to receive alternative treatments, and four times more likely to receive a combination of three treatments in the post-treatment period. This co-treatment confound can’t be dismissed in claiming the intervention’s effect are due to some occult psychological process. Furthermore, our inspection of the survival curves for the two groups indicated that an advantage for the group getting psychotherapy was observed immediately and lasted for a few months with no additional advantage accumulating in years 2 through 6.
Better medical surveillance and treatment for patients getting talk therapy in this study are the most likely source of differences between groups, particularly since the investigators could not identify a possible mechanism by which talk therapy should influence survival. As for a larger literature demonstrating that psychotherapy improves survival, the studies claiming an effect do not withstand scrutiny.
JCO is a prestigious journal that frequently provides results of clinical trials of biomedical interventions. I seriously doubt that the biomedical trials enjoy such relaxed standards for reporting or interpreting results or that a biomedical trial with such co-treatment confounds or increased surveillance given to the intervention group would go unchallenged.
Talk Therapy as Pro-Biotic Yogurt: Linking “Depression,” the Immunity System and Survival
Steel, J. L., D. A. Geller, et al. (2007). Depression, immunity, and survival in patients with hepatobiliary carcinoma. JCO, 25(17): 2397-2405.
This study found that over a third (37%) of hepatobiliary carcinoma patients reported elevated depressive symptoms on a standardized questionnaire. The subsample that had elevated symptoms (23 patients) had lower NK cell numbers and a “robust” statistical trend linked NK cell number to survival and a mediational model “preliminarily” supported a link between depressive symptoms and survival via NK cell count number. The authors conclude “If the tenets of this research are correct, then reduction of depressive symptoms through psychotherapy and/or pharmacologic intervention in patients diagnosed with HBC may enhance immune system functioning and possibly survival.”
An accompanying editorial conceded that the link between depressive symptoms and cancer progression remains “elusive,” but cited purported evidence that psychosocial intervention with patients with malignant melanoma decrease their depressive symptoms while increasing NK cell activity. The editorial called for research evaluating of interventions to reduce depressive symptoms that incorporated designs by which the effects on cancer progression and survival could be examined.
Critique. This editorial refers to post hoc claims by Fawzy that his stress reduction intervention increased NK cell activity and improved survival in malignant melanoma patients to which. My colleagues and I previously dispensed with this claim in a detailed critical analysis.
While 37% sounds like a high prevalence of depressive symptoms in these cancer patients, it is the same as what is found among patients recruited from a primary care waiting room and the mean depression score for these cancer patients was only 2.1 points higher than the general population. And it is not clear what is represented by these “depressive symptoms.” About a third of elevated scores represent treatable clinical depression, but, as with checklist PTSD, symptoms are quite nonspecific and variously represent general malaise, sickness burden, demoralization, and social economic deprivation. Psychologist Paul Meehl has coined the term crud factor to indicate that negative affect variables like depression have so many antecedent, concurrent, and subsequent correlates that prospects for identifying tight causal models are bleak. Notoriously, “causal associations” between depressive symptoms and physical health outcomes including mortality come and go with introduction of statistical controls.
If anything, what is surprising that is that cancer patients do not have higher rates of depressive symptoms.
The efforts of these authors to show causality are thwarted by many factors, but most basically the small number of patients with depressive symptoms. The simple unadjusted statistically significant relationship between depressive symptoms and survival would be undone by reclassification of one patient. And the use of multivariate statistics is simply inappropriate and prone to uncovering spurious findings. The authors’ “mediational model” is not merely preliminary, it is voodoo statistics. First, a mediational model would require demonstration of a simple statistical association between the mediator, NK cell activity, and survival, and there is none there. Second, with only 23 participants, the authors included at least 11 variables in their regression equation attempting to link depression to survival via NK cell activity. It is difficult to believe that a statistician was involved in the peer review of this article.
Overall, this article is typical of psychoneuroimmunological studies of links between “depression,” the immune system, and progression and survival of cancer. There has yet to be a demonstration that reducing clinical depression or depressive symptoms will affect the course of cancer. With inconsistent evidence that “depression” predicts progression and survival, and inconsistent evidence as to what an elevated score on a depression questionnaire represents, tooth fairy science is required to explain how an association happens when the association has not been established in the first place.
Studies claiming to demonstrate that association typically include a large battery of measures of immune function and a cherry picking of any findings that are significant and post hoc explanation why this particular measure proved significant. What is missing is any demonstration of clinical significance of findings and the complex pathways between available measures of immune function and cancer progression and outcome are yet to be established.
An early criticism of psychoneuroimmunological studies remains valid:
Each of the measures used in this literature represents only a small facet of a complex, highly redundant system. It would therefore be inappropriate to conclude that intervention-related changes in any specific immune parameter signal a state of “immune enhancement” or altered susceptibility to immune-mediated disease. The normal functioning range for most immune measures is very broad, and psychological interventions typically do not induce changes of sufficient magnitude to move people outside of these boundaries (p. 48).
Unfortunately, cancer patients desperate to take control of their condition are vulnerable to advice that by improving their immune system, they can “fight” cancer. The claims of psychoneuroimmunology are no more grounded in science than the claims for probiotic yogurt.
Fitting the Evidence to Preconceived Notions
This has not been not a systematic review, but can be seen as a black swan review that calls attention to the fact that not all psycho oncology studies in JCO are white swan exemplars of good science and evidence-based recommendations. Psycho-oncology studies do not appear in every issue of JCO, but when they do appear, they are usually of poor quality and yet accompanied by laudatory editorials. There is a strong confirmatory bias, often manufactured by putting a positive spin on findings that actually show psychotherapy does not improve survival, psychiatric disorders are no more common among cancer patients than other medically ill populations, or that screening improves patient outcomes over simply giving them the opportunity to discuss their problems. It is best to first read the results section of JCO psycho-oncology articles before the introduction and discussion sections, and to skip the laudatory editorials.
Psycho-oncology once had more modest aspirations to which it should return. Wikipedia defines it in reasonable and attainable terms:
a field of interdisciplinary study and practice at the intersection of lifestyle, psychology and oncology. It is concerned with aspects of cancer that go beyond medical treatment and include lifestyle, psychological and social aspects of cancer.
Being able to extend the life of cancer patients is a much stronger claim than merely being able to improve the quality of their lives. The prestige that would go with demonstrating that the mind influences cancer is difficult to resist and a recent commentary in the International Society of Psycho-Oncology journal Psycho-Oncology offered a redefinition — “The very name ‘psycho-oncology’ implies interaction between brain and body” — and went on to make discredited claims of psychotherapy improving survival without any citation of the numerous critiques of this literature that have appeared.
The majority [of intervention studies] show a survival advantage for patients randomized to psychologically effective interventions for individuals with a variety of cancers, including breast, melanoma, gastrointestinal, lymphoma, and lung cancers. Importantly, for breast and other cancers, when aggressive anti-tumor treatments are less effective, supportive approaches appear to become more useful.
My colleagues and I carefully reviewed the same studies as this commentary, and concluded
No randomized clinical trial designed with survival as a primary endpoint and in which psychotherapy was not confounded with medical care has yielded a positive effect.
With its unsubstantiated claims about psychological factors influencing cancer incidence, progression, and outcome by way of the immune system, psycho oncology is approaching the woo of complementary and alternative medicine. Recently, David Gorski, Christoffer Johansen, and I published a critical commentary on an article in the American Cancer Society journal Cancer that claimed psychoneuroimmunological mechanisms for energy field therapy. David also elaborated upon it here on SBM. At about the same time, the prestigious journal Brain, Behavior, and Immunity published an article about a very similar therapeutic touch therapy for cancer patients and also explained its effects in psychoneuroimmunological terms. It is getting increasingly difficult to differentiate psycho-oncology from woo science. Articles in JCO covering psycho-oncology warrant careful scrutiny and readers should be forewarned to approach them with a healthy skepticism.
About the Author
James C. Coyne, Ph.D. is Professor of Psychology in the Department of Psychiatry, Director, Behavioral Oncology Research of the Abramson Cancer Center, and a Senior Fellow at the Leonard Davis Institute for Health Economics, all at the Perelman Medical School of University of Pennsylvania. Additionally, he is Professor of Health Psychology, University of Groningen, the Netherlands. Previously, he served on the faculties of University of California, Berkeley and University of Michigan School of Medicine. Dr. Coyne has been elected a Fellow of the American Psychological Association, Society of Behavioral Medicine, and Academy of Behavioral Medicine. His critical commentaries have challenged whether psychosocial intervention extends the survival of cancer patients, whether recommended and mandated depression programs improve patient outcomes, and whether meta analyses of behavioral medicine commissioned by professional organizations are valid and credible. A 2008 systematic review and meta-analysis in JAMA of screening for depression among cardiovascular patients was designated by BMJ as one of the eight top papers of the year. He is known for presenting and defending controversial positions and for promoting reform of the clinical and health psychology journals. He is the co-author or editor of a number of books including the 2009 Screening for Depression in Clinical Settings: An Evidence-Based Review (Oxford University Press) with Alex Mitchell.
23 Responses to “The Mind in Cancer: Low Quality Evidence from a High-Impact Journal”