by Kenneth W. Krause.
Kenneth W. Krause is a contributing editor and “Science Watch” columnist for the Skeptical Inquirer. Formerly a contributing editor and books columnist for the Humanist, Kenneth contributes regularly to Skeptic as well. He may be contacted at firstname.lastname@example.org.
Dissatisfied with outrageous fees and insensitive bureaucracies and effectively victimized by a pandering, uncritical mainstream media, an increasingly sentimental and scientifically illiterate American public has embraced faith as a supplement to—if not as a substitute for—empirically based health care. In Blind Faith, Columbia University professor of behavioral medicine, Richard Sloan, investigates the credibility of alleged links between religion and medicine and opines as to how the joining of faith and science have affected the integrity of each. Sloan’s theological assessments are unhelpful. His forays into medical ethics are relevant, but infected with an acute and probably chronic case of professional arrogance. More impressive are the professor’s presentations and critiques of the pertinent studies.
Sloan scrutinizes the two compilations most frequently cited by supporters of religiously infused medicine. In 2000, Frederic Luskin reviewed twelve papers on cardiovascular disease in the Journal of Cardiopulmonary Medicine, claiming that “evidence continues to mount that demonstrates the positive value of spiritual and religious factors in maintaining health.” Of those 12 papers, two pertained to denominational distinctions, one merely summarized a project, and another simply reviewed other studies. The first four, in other words, were either irrelevant to the stated issue or inadequately detailed.
Two of the remaining eight papers failed to correct for what researchers pejoratively refer to as “multiple comparisons.” In order to distinguish between a statistically significant or “real” finding and a “chance” finding, scientists have established the five percent standard. A result is deemed real, in other words, if it would occur by chance less than five percent of the time. For example, if a coin were tossed four times and came up heads in each instance, we would conclude that it did so by chance because the statistical probability of that outcome is 6.25 percent. But if a coin were tossed five times instead, and came up heads each time, we would reject chance as an explanation because that result’s probability is only 3.13 percent. We would conclude that the second coin was fixed. If one were to continue flipping perfectly legitimate coins, however, eventually she would get five heads in a row because multiple tests lower the scientific standard and increase the likelihood of mistaking a chance result for a real finding.
Sloan identifies a paper written by Harold Koenig and colleagues as one of the two that violated the multiple comparisons standard. Purportedly finding differences in blood pressures between religiously involved and uninvolved persons, Koenig’s team made at least 126 different medical comparisons, only a few of which were ultimately (and erroneously) deemed significant. Koenig could have corrected for the breach simply by multiplying the five percent standard by the number of comparisons his team made, but he didn’t. Though likely the products of chance alone, Sloan concludes, it is not at all surprising that a few among the 126 results at least appeared significant to persons caught unaware of the research standard and its purpose.
Luskin reviewed two more studies relating to the effect of religious activity upon blood pressures. One paper reported comparatively low diastolic but not systolic blood pressures among subjects declaring greater frequencies of attendance at religious services. The second study found reduced systolic but not diastolic pressures among members of a similar group. Obviously, both reports can’t be right. And neither research group controlled for the subjects’ prior health. If they had been ill before the experiment, for example, they would have been more likely to have high blood pressure and less likely to be able to attend church.
Of the remaining four, one paper examined religiosity’s effect on patient recovery time following coronary artery bypass surgery. The problem here was that the researchers failed to consider even the most obvious potential confounding factors, including other health indicators such as smoking and high blood pressure, or relevant social variables like marital status. A similar study on bypass patients attempted to associate lack of religiosity with mortality. Its strength was that it considered at least some potential confounders. Its fatal weakness, however, was that it focused upon a single religiosity factor, reported “comfort from religion,” among a total of five studied. Once again, a supposedly scientific inquiry failed to correct for multiple comparisons. A third paper, this time involving heart transplant patients, reported just one significant finding among forty-two statistical tests. The final study included in Luskin’s twelve suggested that religious coping was protective. But according to Sloan, there was nothing at all about protection in the paper.
Next, the author critiques 89 cardiovascular health reports contained in the second popular compilation, Koenig’s 2001 Handbook of Religion and Medicine. Sloan dismisses many of these studies for familiar reasons. 33 based conclusions on denominational differences, 14 simply abstracted, reviewed, or described other projects, and eight drew no connections whatever between religious activity and health. Others made uncorrected multiple comparisons and some involved inadequate or nonexistent control groups. One weight-control program had nothing to do with religion, but was apparently included because it was conducted in a church. Three studies admitted to unearthing not one statistically significant effect, and another paper—written by Koenig himself—contained findings directly opposed to the predicted result. Although its authors continue to represent the Handbook as a definitive and affirmative examination of the relevant literature, according to Sloan, what the book in fact accomplishes “is to show us definitively how incredibly weak the evidence actually is.”
For an evaluation of claims regarding religion’s general health benefits, Sloan briefly hands the reigns over to Dr. Lynda Powell and colleagues who reviewed the pertinent literature and, in 2003, published their findings in the American Psychologist. After weeding out methodological disasters of the aforementioned variety, Powell examined the remaining evidence relevant to nine different hypotheses. Two results predominated. First, of the nine claims, Powell found four to be “consistent failures.” She discerned no reason to believe that religion or spirituality protected against death or disability, slowed the progression of cancer, or improved recoveries from acute illnesses. Second, however, Powell found “persuasive” evidence (meaning that there existed at least three supportive studies without serious methodological flaws or five supportive studies that were, at a minimum, generally sound) suggesting that consistent religious service attendance protected against mortality.
But Sloan remains skeptical for three compelling reasons. First, the association between church attendance and relatively elevated survival rates might not be causal. Projects based on religious beliefs and practices, unfortunately, must rely on observational studies where researchers cannot select participants and randomly assign them to groups according to the variable of interest. Proper clinical studies, involving randomly selected treatment and control groups, simply cannot be accomplished in these contexts. The trouble here is that subjects who self-select make it difficult for researchers to identify confounders. In other words, religious attendees might well differ from their non-attending counterparts in very important ways. For example, Sloan suggests, people who are depressed, live alone, or are already ill might tend to stay home. Although researchers can always attempt to control for such factors, they can never be sure that they have anticipated them all.
Second, because such studies are mostly conducted by interview, they are susceptible to self-presentation biases. People tend to present themselves in the best possible light, and at least some will embellish their observance of socially accepted norms, especially during impersonal telephone interviews. Studies devoted specifically to such interview methods estimate attendance reports to be over-represented by as much as 33 percent. Third, Sloan raises the issue of construct validity, arguing that church attendance is often unreflective of actual religious devotion. That the vast majority of churchgoers know little about their traditions’ histories or theologies seems to cut deeply in Sloan’s favor here. The author appears to contend, quite convincingly, that no matter the result, any study attempting to link health to religiosity is doomed from the start.
Somewhat more credible are controlled evaluations of long-distance intercessory prayer (IP), where numerous patients are randomly and unwittingly assigned to one or more treatment or control groups. Sometimes subjects are treated in a variety of ways, as occurred in the MANTRA (Monitoring and Actualization of Noetic Training) and the MANTRA II studies, published by Mitchell Krucoff and others in 2001 and 2005. In the pilot program, 150 acute coronary patients undergoing angioplasty were selected to receive one of five treatments: imagery, stress relaxation, touch therapy, IP, or standard care. The results of MANTRA yielded not one statistically significant difference. Nevertheless, Krucoff initiated a larger evaluation. MANTRA II, designed as the definitive study on IP and released in The Lancet, randomized 748 angioplasty or cardiac catheterization patients to one of four treatments: standard care, IP, MIT (a combination of music, guided imagery, and touch), and IP plus MIT. Again, the outcomes demonstrated that IP had no effect. Finally, in 2006, Harvard’s Herbert Benson published the $2.4 million Study of the Therapeutic Effects of Intercessory Prayer (STEP) in The American Heart Journal. 1802 patients recovering from coronary artery bypass surgery were randomized into one of three groups: those who unknowingly received prayer (from three mainstream religious groups), those who unknowingly did not, and those who knew they would receive prayer. Either way, the prayers had no effect.
But what if the results had been different, somehow indicating religiosity’s positive affect on health outcomes? Should physicians inform and advise their patients accordingly? Not if they intend to practice medicine ethically, warns Sloan. Armed with such dangerous facts, the author vies, some patients would harm themselves, concluding that religion alone would protect them from disease. And because regular folks are incapable of seeing doctors as anything but intimidating authority figures, patients might well feel coerced when making personal decisions. Better they should remain ignorant. So why write a science book for these poor, wretched creatures? In the end, Sloan robs his readers of the rationality and responsibility he initially presumed them to possess—as if their megalomaniacal physicians hadn’t already performed that valuable (and expensive) service.