A recent editorial in the normally stuffy Archives of General Medicine started with an imaginary dialogue between two Star Trek characters. Commander Spock's rational but pessimistic prediction was that the possibility of success for their project was less than 4.3 percent. But Captain Kirk's contradicted Spock with unshakeable optimism: "Spock, it will work."
Drs. Gramling and Epstein, the physician authors went on to make a declaration that was picked up in the media: "optimism is a powerful 'drug' that compares favorably with highly effective medical therapies." They propose that there is "compelling evidence" that optimistic outlooks extend life, so much so that physicians should promote optimistic expectations in their patients, maybe even withholding information that patients realistically have poor prognoses, slim chances of beating the odds.
Aside from the ethical issues posed by such a strategy, what is the "compelling evidence" that manipulating patient's expectations actually extends their lives? The editorial accompanied a study by John Barefoot and colleagues. The study found that heart patients' initial expectations of recovery predicted how well they were doing at one year and even their survival over the next 15 years, even when a full range of predictors were controlled statistically.
Provocative, but what can we conclude? Patients probably know something that is not captured in the investigators' identification and measurement of predictor variables. This is not surprising: in many studies, patients' ratings of their health predict how long they live, even after similar statistical controls.
But would anyone really conclude from such studies that artificially manipulating patients' perceptions of their health would actually prolong their lives? It is important to note that these studies, like Barefoot and colleagues', are correlational, not interventions, and so it is not warranted to infer any causation of health by belief or expectations. Predictions of causality from correlational studies often prove disappointing when tested in randomized controlled trials that manipulate the very variables expected to have causal power.
Gramling and Epstein also cite a pair of Gramling's own studies as the rest of the compelling evidence. But these are even weaker evidence than what Barefoot's study provided. Gramling had too few deaths to explain for the powerful statistical techniques he used, and so any positive findings are unreliable, they are suspect. And even then, his results did not hold for women, something that was not predicted ahead of time.
So, available evidence hardly suggests taking a leap from these correlational studies to a large randomized controlled trial in which patient expectations are manipulated by withholding or distorting information. There is no difficult ethical decision whether to deceive patients at risk for death in the hope of extending their survival, there is only bad science and bad medicine in the suggestion that it would work, given available evidence. Ah, but the idea got Gramling and Epstein mentioned in the media and inspired this blog post.