As usual, Dr. Fox provides what the field needs: a tough-minded assessment of the state of knowledge regarding the possible influence of psychological and social factors on cancer incidence and progression.
This kind of rigorous examination of data is especially important in an area that is either dismissed out of hand or supported with religious fervor. Dr. Fox sets a high standard for proof that a psychosocial variable influences the course of cancer, since he assumes that the possible existence of confounding variables means that the findings are suspect. For example, he observes that the suppression of emotion often observed among cancer patients at the time of diagnosis  may be better explained as a response to the cancer diagnosis. Patients who are experiencing more disease-related dysphoria may be more inclined to attempt to suppress those emotions.
Nonetheless, in holding studies to such a high standard, Dr. Fox does the field a service by taking it seriously. The stringent demand for proof is one way of acknowledging its possibility. The idea that stress and emotional suppression could increase or social support decrease the incidence or rate of progression of cancer is one that deserves empirical examination. It is worthy of study; it may be wrong, but it is not silly or implausible.
While, in general, Dr. Fox's critique is clear and dispassionate, he occasionally descends to the ad hominem, speculating that disciplinary narrowness keeps psychiatrists and psychologists from considering the biologic effects of the disease upon the psyche. There are, of course, other possibilities. The adverse effect of cancer on emotions is so much taken for granted that it is often used to dismiss or minimize the occurrence of depressive and anxiety disorders [2,3]. Indeed, many of these researchers have devoted their resources to swimming upstream, searching for any clue that the psychosocial influences the biologic, despite the obvious converse effect.
In attacking the issue of emotional suppression, especially anger, Dr. Fox makes a cogent point: Emotional suppression may result from disease-induced distress rather than cause it. The interesting study by Schwarz  does raise questions about the general observation, suggesting that it is bad news that leads to anger suppression, but it does not disprove entirely studies that assessed anger suppression prior to informing the patients of the diagnosis. It is always possible that those who turned out to have a malignancy were more anxious to begin with, or had been led to have more negative expectations, as Dr. Fox suggests. However, there is also evidence among patients already diagnosed with cancer. For example, less emotional suppression has been found to be associated with slower progression of such diseases as malignant melanoma . Dr. Fox could argue that those whose disease is progressing more rapidly become more dysphoric and therefore repress emotions more. That possibility could account for some of the findings, but that does not, per se, disprove the other possibility: the psychobiologic influence as opposed to the biopsychologic one.
Dr. Fox's review of the psychotherapeutic intervention literature is reasonable, but a few comments could be made. In his thoughtful discussion of my study, he points out that after randomization, 14 intervention and 12 control patients were lost to further follow-up. This does not mean that none of those intervention patients participated in treatment. It simply means that they were unavailable for the first follow-up point 4 months after randomization. Furthermore, and most important, the treatment/control comparison was based on intention to treat. There was a significant difference in survival time between treatment and control groups, even though patients who received little or no intervention were included. This is the only proper method of analysis of a randomized trial, but it means that the relationship between the intervention and extended survival time was sufficiently vigorous to produce an overall between-group difference, even when part of the intervention sample received little or no group treatment.
We did indeed find in post-hoc analysis that those patients who attended more groups lived longer than those who attended few or none. We did not include this finding in the original report , because it is confounded with the opposite effect: Those who died early could not attend treatment groups. As Dr. Fox notes, we are in the middle of conducting a full-scale replication trial at the moment, and clearly more data are necessary.
A study by Richardson and colleagues  is worth noting as well. This was a randomized trial of a home-visiting educational intervention for patients with lymphomas or leukemia. The researchers found that patients who received one of three educational interventions (n = 75) were more compliant with treatment that those who received routine care (n = 25). However, even when differences in compliance with treatment were controlled, they observed significantly longer survival in the intervention sample.
Thus, three of four randomized trials suggest that psychosocial intervention influences survival time. Dr. Fox is correct in noting some design problems in the Ilnyckyj et al study , including the variety of poorly defined psychotherapeutic interventions and the authors' inability to demonstrate any psychological benefit from the intervention. It is hard to imagine that an intervention that does not benefit patients psychologically will extend survival time.
Dr. Fox's critique of the clinical salience of psychoimmune system changes is apropos. However, it should be noted that in Fawzy's study , baseline natural killer cell cytotoxic activity was significantly associated with the rate of relapse, although not of mortality.
Conclusions Too Negative
Dr. Fox's conclusion seems somewhat more negative than the literature review he provides would warrant. He notes that Reynolds and Kaplan , for example, had conducted a careful study in a large sample demonstrating that social isolation among men was associated with shorter survival time, and that socially isolated women had higher cancer incidence. Furthermore, as Dr. Fox notes, among studies of the relationship between social support and cancer, three showed a clear relationship, four showed mixed or uncertain relationships, and one demonstrated none. This is clearly better than chance. Dr. Fox estimates a relatively low increase in relative risk due to social isolation. Nonetheless, the fact that there is any relationship at all is interesting.
Similarly, three of four randomized trials of psychosocial intervention show a survival advantage for patients randomly assigned to psychosocial support. Given the general improbability that an intangible intervention would have such a tangible result, Dr. Fox's conclusion seems a bit dour.
Dr. Fox contributes to the field by insisting on a high standard of proof for the assertion that psychosocial variables influence cancer incidence or progression. At the same time, there is a double standard, since many biologic factors are considered naturally associated with cancer on much weaker grounds. He is rightly scornful of those who believe that there must be a path by which the mind influences the body's resistance to cancer. At the same time, the evidence in regard to social support and psychotherapeutic intervention suggests some relationship. While the relationship for other variables may be weaker, it is not negligible, especially given the difficulty of demonstrating such relationships. Psychosocial variables are inherently difficult to assess, and error variance makes it harder, not easier, to show a relationship with other factors. The fact that anything at all emerges suggests that the relationships between social support, affective regulation, cancer incidence, and disease progression deserve more systematic exploration.
The Benefit of Social Support
Social support may positively affect cancer patients even though our understanding of its basis is wrong. In the last century, for example, miasma theory incorrectly held that all kinds of diseases could arise from a single miasm or infectious area. We now know that certain bacteria and viruses induce specific diseases. We have a far better notion now of the specific connection between infectious agents and disease. At the same time, it is worth noting that some of the most important public health measures ever taken to reduce the incidence of disease, which involved installing sanitary water and sewage systems in major cities, were based on the outmoded and incorrect miasma theory. People now live better and longer because of an incorrect theory of the etiology of infectious disease. The notion that there were infectious agents in the environment that led to disease was correct. The belief that they all were of one general type was incorrect.
In the same way, it may be that the general belief that social support can ameliorate the course of cancer, and stress or lack of emotional support can adversely influence its course may be right even if some of the specific variables examined are not, in fact, empirically related to disease progression. The awakening to this possibility may stimulate opportunities for helping patients better deal with their cancer.
Furthermore, such findings have tremendous potential to influence medical practice. Medicine has focused so much on attacking the tumor that it has tended to ignore the body coping with the tumor, and the social and psychological variables that influence the somatic response to tumor invasion. If there is any evidence at all that social support or affect regulation influences the rate of disease progression, the need for systematic study of possible mechanisms is obvious. But perhaps more important, if providing psychotherapeutic support ameliorates the course of cancer, even, as Dr. Fox notes, for some cancers in some people, then such findings cry out for application in clinical settings.
Biologic treatments that produce only marginal increases in survival are widely employed despite considerable risks and side effects. Many psychosocial interventions that are clearly helpful emotionally and carry with them very little in the way of risks, side effects, or expense are far less widely employed [11,12].
This domain, hampered though it be by conflicting results and methodologic problems, potentially has much to teach medicine about the importance of caring and support on the outcome of medical illness. If being a lonely man increases your risk of cancer mortality, and being a lonely woman increases your risk of cancer incidence, then loneliness is a cancer risk factor every bit as much as fat in the diet or exposure to known carcinogens. The relative risk ratios may differ, but, in both cases, the opportunity to intervene must not be ignored.
Dr. Fox calls on us not to go beyond the evidence, and he is correct. At the same time, I would urge medicine not to sink beneath the evidence either, but rather to consider the possibility that compassionate caring and social support may exert an independent influence not only on the well-being of the cancer patient but also on his or her survival with the disease. The null hypothesis has a well-earned central place in research. At the same time, there are real life costs to rejecting incompletely proven theories, just as there are to swallowing them whole. Living better may not mean living longer, but if it does and we ignore it, we are missing a major opportunity to enrich and perhaps even to extend life.