Emotional State Doesn’t Affect Cancer Survival Researchers found no difference between upbeat, less positive patients
By Madeline Vann, HealthDay Reporter
MONDAY, Oct. 22 (HealthDay News)—Neither positive nor negative emotional states predict how long a person with cancer will survive, a new study finds.
University of Pennsylvania researchers say that among head and neck cancer patients, emotional health—good or bad—is not an independent factor affecting prognosis.
“We anticipated finding that emotional well-being would predict the outcome of cancer. We exhaustively looked for it, and we concluded there is no effect for emotional well-being on cancer outcome,” said study author and University of Pennsylvania psychologist James Coyne. “I think [cancer survival] is basically biological. Cancer patients shouldn’t blame themselves—too often we think if cancer were beatable, you should beat it. You can’t control your cancer. For some, this news may lead to some level of acceptance.” ...
Well, I certainly agree with most of the generalizations in the blog about extrapolating from placebo effects on subjective variables like pain perception to more objective variables. I hate to be picky, but I think the conclusion that the placebo effect has no effect on cancer survival is a bit of a stretch from the studies he links to. Part of my problem with a lot of research on mental states and disease is the heavy reliance on questionaires and subjective grading of the mental states. It’s so hard to find a meaningfully objective measure of the the variables you are trying to show do or do not affects the disease process. I wouldn’t like to see the “good guys” be guilty of the same kinds of self-confirming research as the AM crowd. Unfortunately, that means we have to be twice as careful about conclusions that go beyond the data.
Could you elaborate on the design flaws you see in the studies Novello references? Just to be clear, Novello is an academic clinical neurologist at Yale University School of Medicine. This doesn’t mean he can’t get things wrong, of course, but at least I’d be likely to accept prima facie that he has some idea what he’s talking about.
Well, in this case I don’t think his qualifications really have any bearing. He undoubtedly knows a lot about scientific reszearch methodology, but so did Lnus Pauling and look at the nonsense he believed. If you have an a priori belief about something, it is likely to color your interpretation of the evidence, and this is true for everybody. I tend to share this prior belief, but that doesn’t mean I think the particular study is any kind of definitive nail in the coffin of the idea that attitude/psychological well0being can influence disease progression. To present it as such is to overinterpret the evidence, which weakens the overall case.
In general I am skeptical of the precision of survey-based, self-reported measures of well-being/mood, etc. I realize there is little objective data we can gather on such things (though in non-human animals catecholamine levels, cortisol levels, observed behaviors, blood pressure, and so on are used perforce, and these are at least a trifle more objective). So when you gather subjects into groups based on a such a categorization scheme, and then calculate statistical differences in outcome, I have to ask whether the lack of a difference is related to the lack of meaning to the original grouping scheme. Also, as he acknowledges in his blog, the findings for a particular form of cancer, in a clinical trial (which creates a non-representative sample of patients compared to the overall population with cancer) may or may not be evidence of a generalizable rule. There is also no discussion I can see about whether the investigators were blinded to the pateint categories, which is always important in guarding against bias.
In general, while I suspect his conclusion is true, I see little methodological difference between this study and those I see reported that argue in favor of a pyschological influence of the disease process. What I prefer all scientists to do is to interpret the results of particular studies in very limited way, and then let time and subsequent meta-analyses ferret out the underlying patterns. This would dampen the oscillating opinions (yes fiber reduces colon cancer, no it doesn’t; yes HRT reduces heart disease, no it doesn’t, etc) that the media creates by generalizing each new study as confirmation or disconfirmation of general principles.
Well, Brennen, you raise a few concerns about methodology that we don’t really know yet. I mean, I expect that you don’t have access to the original paper, which should make clear their experimental design. But as far as mood goes, do you think there is a better way of tracking it than self reporting?
At this point I am going to assume (with a relatively low degree of probability, but over 50%) that Novello knows enough about the study to know that the simple problems you outline were overcome to some degree. But certainly you are right to be skeptical about the design until you’ve had a chance to look over the paper itself.
And yes, the problem with Linus Pauling is that he was so brilliant and yet so stupid about Vitamin C in particular. It was like an obsession or cult for him. I doubt that the same is true in this respect with Novello, but it’s always possible.
And what documentation do you have, Doug, that larger amounts of vitamin C than the RDA is of no value?
A grad student at UCLA decided to check out whether the excess was dumped in one’s urine. He had been taking it for three months and checking the content of ascorbic acid in his urine daily. He had just about decided that the amount above four times the RDA was wasted. Then he got a mild cold. What he found interesting was that the amount in his urine dropped precipitously for about four days. He couldn’t draw the conclusion from this that the C helped, but it certainly was behaving differently during his cold.
I’ve been taking 3.5 gm of vitamin C daily for the last forty years. It’s cheap, and I’ve seen no physical damage during that time, I’ve been quite healthy so far, so I don’t feel like reducing it.
I have to say you seem really reluctant to admit any flaws in the study, which I have to think is related to the fact that it comes to a conclusion you support. As I’ve said, I don’t think Novello’s credentials are an issue here. Don’t you believe smat, educated people can interpret data in a skewed way in light of their own preconceptions? To me it seems common and obvious that they do.
As for specifics, I’ve already given them. I don’t accept the objectivity of a questionaire on emotional well-being. I listed a number of more objective alternatives in my previous response. And even if there are no other tools we can use to assess emotional state, if self-reporting is unreliable than doing statistics on outcomes for groups organized by self-reporting is a waste of time and doesn’t lead to valid conclusions. I also pointed out that the patient population is not representative of cancer patients generally since they have a particularly aggressive, disfiguring cancer and they are enrolled in a Phase III clinical trial (which means they are among the sickest, and also that they are among the best informed with the best access to advanced health care). It’s true I don’t know about the blinding issue (the actual article is published in the Dec issue of the journal, so I can only see the summary Novello and others provide). But even if the methodology were ironclad, at most I think it is reasonable to say that mood, as reported in response to a specific questionaire, does not appear to significantly affect overall survival for patients with advanced head and neck cancer enrolled in a Phase III clinical trial. Any further generalizations are speculation.
I think non-scientists tend to overestimate the strenght and underestimate the weakness of individual research reports. THIS is a great editorial on epidemiological studies which illustrates some of the dangers of drawing conclusions about health effects from anything other than a large-scale, multi-center, double-blinded, placebo-controlled studies, even if you have good methodlogy and lots of subjects. And frankly, even the best study can only prove what itr is narrowly tailored to prove. While the study Novello refers to is reasonably well-designed, my point is that overgeneralization is the cardinal sin of science. It is a big part of what differentiaties pseudoscience from real science. Unfrotunatly, even scientists are naturally inclined to seize upon a piece of evidence that supports what they already think and give it more weight than it really deserves. Overgeneralization also lead to wide, irrational oscillations in the public and media-reported understanding of things like healthcare, which then undemrines the public’s confidence in science generally.
I don’t necessarily think, as I’ve said before, the general conclusion is wrong. I just think that saying something so broad like “the placebo effect has no effect on cancer” based on a study of self-reported mood and survival in pateints in a very specific kind of cancer circumstance is excessive generalization. I think it is motivated by a desire to see validated a conclusion one already has drawn, and whether that conclusion is valid or not doesn’t change the limiktations of such small single studies in demonstarting it.
Well, I certainly do agree with your take on these studies, Brennen—that no one study should be taken too seriously, and that all studies should be taken against the background literature. The reason, however, that I am somewhat reluctant to admit flaws in this particular study is just that I haven’t heard anything that seems like a flaw yet from you. What you’ve presented are well-founded questions as to the methodology, and I agree that the study may be flawed. But it’s one thing to say it may be flawed and another to say it is flawed.
Also, I do believe the background literature on this subject supports the general conclusions of this study—that is, that the placebo effect generally occurs with diseases that are largely psychological in character, such as those associated with pain or stress. This is why my prior probability of this conclusion being correct is something higher than 50%, to be sure. Although IIRC Dr. Novello also says that state of mind can have a positive effect on some sorts of heart disease, which means that he at least is open to the possibility of it occuring with cancer. And for what it’s worth, had the study come out otherwise and been supported by a prominent doctor with a skeptical bent like Novello, I’d be certainly prone to at least accept it as prima facie good evidence that state-of-mind can effect (some) cancers.
... understood, of course, that no one study in medicine is totally probative.
It would be nice to get Dr. Novello here as well; I’m assuming he knows something more about the methodology than we do.
Wouldn’t you consider generalizing from an atypical subset of cancer patients to all cancer patients a flaw? Or doing post-hoc analyses of the data to look for something other than what the study was designed to investigate? I consider these both common flaws in research design.
Anyway, since we basically agree that the idea of mental attitude affecting disease progression is one that is unlikely based on general principles and certainly is unproven, we don’t necessarily have to argue the details. All I’m saying is I see a similarity in overgeneralizing from single studies to general conclusions on both sides of the issue and I expect more rigor from my own side.
Similarly, I find Barrett’s article on Pauling rife with ad hominum comments, bias, and conclusions he stretches beyond the limits of the data. In the dim past when I was much younger and more scientifically demanding and precise, I went through Pauling’s data and the data from other studies. I decided that the evidence was decent enough that I should take more than the RDA of vitamin C. However, Doug, if you have faith in Barrett, then far be it from me to attack anyone’s theology - oops, I meant ontology.
Brennen, yes, if the study is done on head and neck cancers alone then you are right to point out the limitation of the experimental design, to that extent.
Occam, Dr. Barrett is following the best research here. I don’t know of any reputable public sources (Consumer Reports, UW Berkeley Health Letter, Mayo Clinic, etc.) that support Pauling’s claims on vitamin C. He was seen as a crank in his day by the medical establishment, and meanwhile we’ve had decades of further studies reinforcing that conclusion.
Claims that vitamin C is effective for preventing senility and the common cold, and for treating asthma, some mental problems, cancer, hardening of the arteries, allergies, eye ulcers, blood clots, gum disease, and pressure sores have not been proven. Although vitamin C is being used to reduce the risk of cardiovascular disease and certain types of cancer, there is not enough information to show that these uses are effective.
That said, AFAIK there are no serious side effects to taking relatively large doses of vitamin C, so if it makes you happy, go for the placebo!
It’s only anecdotal, and it may be the placebo effect, but at 77 I’m surprised to see a number of friends beginning to have minor memory problems. So far, people keep remarking that I have an excellent memory for any age, not just for an old fud. So, I’ll stick with my placebo. I’ll remind you of this conversation in ten years, Doug, if you’ve forgotten it.