Evidence-Based Medicine and Modernism: Still Better Than the Alternatives

Thomas, Bracken, and Timimi (2012) make an important contribution in critiquing the extent to which the profession of psychiatry can be so bureaucratic that patients are treated as problems to be solved in an ‘efficient’ assembly line fashion rather than as individual persons. The trouble with bureaucracies is that they promote a cold and impersonal accounting approach in which critical reflection on purposes is circumvented by decision-making algorithms (Zachar and Bartlett 2009). Psychotherapy treatment manuals definitely satisfy the bureaucratic instinct, and the fifteen-minute medication management session even more so (Harris 2011). Ideally, evidence-based medicine (EBM) should be used to promote the goals of practitioners, not insurance companies, and the role of researchers in EBM should be not to dictate the standards of practice, but to critically evaluate the evidence and the generalizations drawn from it.

The author’s postmodern-inspired suggestion that the narrow technical approach to mental health practice is the natural outcome of modern philosophy and scientific reasoning is another matter. Horkheimer and Adorno (1944) argued that, in the twentieth century, the Enlightenment approach to knowledge had become the opposite of Enlightenment. In the same way, one could argue that that the consequences of a postmodern approach will be the opposite of what was intended at its inception. One of the more influential criticisms of postmodernism (and probably postpsychiatry) is that it is conceived by its proponents as a progressive movement, but is closer to the anti-modern romanticism of the nineteenth century (Gross et al. 1996). Bruno Latour (2004) himself has seemingly developed doubts along these lines.

Although this is a worry more than a firm conclusion, in this commentary I explore two disagreements with Thomas, Bracken, and Timimi that highlight areas where their postmodernism flirts with anti-modernism in the negative sense of that term. The first is their criticism of how randomized, controlled trials (RCTs) are weighted in EBM and the second is their exchanging the primacy of technical expertise for the primacy of patient preferences.

RCTs and Different Grades of Evidence

Thomas and his colleagues are critical of EBM’s ranking of evidence, particularly the first place ranking of evidence from meta-analyses of RCTs. To explore this issue, we need to consider the purpose of an RCT. The question being asked by a comprehensive RCT is this: can improvement following a specific treatment be attributed to that treatment and not to something else? How does this work? In an RCT, a representative sample of people with a particular condition such as major depressive disorder are randomly assigned to a treatment group, a placebo control group, or a no-treatment control group. The treatment can be anything—a drug, psychotherapy, a community support program, and so on. How is it evaluated? To make things less complicated, let us assume that after two months, those receiving the treatment no longer meet diagnostic criteria for depression (and they had returned to their previous level of functioning) but those in the placebo and the no-treatment control groups have become more depressed. A modern scientist would say that in this case the evidence should compel acceptance of the comparative efficacy of the treatment.

We know that reality is not so uncomplicated. For example, how does one define improvement? Is symptom reduction adequate? I have seen people so medicated that they were no longer depressed, but not very functional either. Defining improvement requires a qualitative evaluation and consideration of values. Furthermore, all the effects of the treatment should be evaluated, not only the target effects.

Another complication is sample selection. One important flaw of outcome research conducted using treatment manuals is that patients with complicated presentations (in the current categorical system, these would be termed comorbid presentations) are not typically included in the sample. This means that the results of such studies cannot be readily generalized to a large number of persons who are depressed. Other restrictions in selection (such as age and gender restrictions) introduce similar problems.

Inadequate sample same size is another complication. Because the number of people in each of...