The great obstacle to discovery is not ignorance,  it is the illusion of knowledge.

                                                                    Daniel J. Boorstin

Efficacy of Psychotropic Medications

Treatment practices are often a product of research driven by sourcing the etiology of the problem.  An important issue relative to the biological causation worldview is that a review of the literature reveals that there is less evidence of effectiveness for biological treatment of mental health issues than one would suppose.  A number of meta-analytic reviews have revealed that evidence for effectiveness of medications is scarce, a surprising finding given the frequency with which medications are prescribed (Khan, Leventhal, Khan, & Brown, 2002; Kirsch & Sapirstein, 1998; Sommers-Flanagan & Sommers-Flanagan, 1996).

In a meta-analysis published in 1998, Kirsch and Sapirstein compared the mean effect size changes in symptoms of depression across 19 double-blind studies assessing the efficacy of antidepressant medications. Results demonstrated that placebos accounted for approximately 75% of the improvement found in the active drug. Furthermore, the authors asserted that the remaining 25% of improvement accounted for by the active drug was debatable, and could have been the result of an enhanced placebo response due to the side effects that patients experience when taking the active drugs, or other factors.

Other factors that may contribute to inaccurate analysis of data are (1) a paucity of double-blind methodologies, (2) failure to account for placebo effects, and (3) over-generalization of results.  For example, Kirsch and Weixel (1988) argued that placebo and drug effects obtained in double-blind conditions are not comparable to those obtained in clinical practice.  They asserted that because clinical administration procedures do not lead people to suspect that they might be receiving an inactive preparation, the expectations are greater for change.

The placebo effect, in this case a sugar pill, has long been acknowledged to have the capacity to rally healing processes (Harrington, 1999), and has been cited in many studies to equal, if not exceed, improvement rate when compared to drug effects  (Khan et al., 2002; Sommers-Flanagan et al., 1996; Kirsch & Sapirstein, 1998). For example, Khan et al., (2002), found that individuals who received a diagnosis of mild to moderate depression had a superior response (e.g., self report of improved mood) to placebo than to a drug. Although Khan argued that severe degrees of depression seemed to be more responsive to medication than placebo, they noted that the greater change among patients with severe scores on the initial depression scale scores may simply have reflected regression to the mean, rather than true drug effects.  One would expect exactly that from a mathematical point of view (Cronbach & Furby 1970; Rogosa & Willett, 1983; Willet, 1994).

The methodologies of drug studies also present some problems in interpretation.  First, over-generalization of treatment effects from research participants to the general population is frequently evident.  Research participants usually must meet stringent exclusion and inclusion criteria and are not representative of the population of individuals with specific disorders (Kahn et. al, 2002).  For example, Nierenberg, et al. (1995) assessed early non-response to the anti-depressant fluoxetine (Prozac) as a predictor of 8-week outcome.  They reported that patients were excluded from the study if they  had:  failed to respond to any antidepressant therapy during the current episode; had another Axis I disorder; experienced any serious medical illness; presented with significant lab tests (CBC, urea nitrogen, creatinine, electrolytes, plasma glucose, liver function, thyroid, etc); were receiving anticoagulants; had a positive drug screen; reported suicidal ideation,; “ever” received CBT or ECT; were pregnant or lactating women; were those of childbearing potential who were not using contraceptives or if they had responded to placebo during a 2-3 week “single-blind placebo run-in period”.  It is likely not generally known outside the drug research community that exclusion of potential subjects such as those is standard practice in psychotropic medication research studies.  Zimmerman (2004) suggests that there is much variability in the generalizability of antidepressant efficacy trials (AET).  Subjects treated in AET’s represent only a minority of patients treated for major depression disorder (MDD).  He further states that since the inclusion and exclusion criteria used to select subjects for participation in AET’s vary from study to study, it is unknown how much impact different sets of exclusion criteria have on the representativeness of subjects treated.

Of special interest for this research is that in most drug studies there is a practice called placebo run-in or “washout” group factored into the design (Leutcher, Cook, Witte, Morgan, & Abrams, 2002; Lydiard, Steiner, Burnham, & Gergel, 1998; Nierenberg et al, 1995; Londborg, Wolkow, Smith, DuBoff, England, Ferguson, Rosenthal, and Weise, 1998; Pohl, Wolkow, & Clary, 1998; Sommers-Flanagan & Sommers-Flanagan, 1996). A given rationale for this practice appears to come from a number of studies performed to determine how to separate placebo and true clinical responses to antidepressants. Results are reported to indicate that antidepressant placebo responses occur early and are of short duration, whereas true drug responses occur later and last longer (Laurie, 1996; Womack, Potthoff; & Udell, 2001;).  Using a washout period is certainly logical and intuitively compelling in efforts to minimize placebo effect.  However, this process impacts overall validity of the true drug effect by minimizing the generalizability of the sample size and therefore erroneously exaggerating efficacy results.

After enrollment in the study, all potential subjects are given a one to three week clinical trial of placebo (single-blind) treatment. Those subjects who respond favorably to the placebo treatment are then excluded from the remainder of the study.  Thus when the remaining subjects are divided into two groups (drug v. placebo) results suggesting a higher drug effect when compared to placebo is exaggerated.  Conclusions such as “sertraline was safe in reducing panic attacks” (Londborg et. al, 1998), or “sertraline (Zoloft) is an effective and well tolerated treatment for patients with panic disorder” (Pohl et al, 1998) or “paroxetine (Paxil) was significantly more efficacious than placebo” (Lydiard et al, 1998) are then often found in publications targeted at professionals.  More troubling is that the message of drug efficacy presented to the public by the pharmaceutical industry in direct-to-consumer advertising, is often accepted unwarily by the consumers of psychotropic medication.

Whether treatment is a pharmacologically inert pill or a chemically arranged Serotonin Selective Reuptake Inhibiter (SSRI), or some other psychotropic drug, the recognition of its symbolic value, expectancy effects and its power as a conditioned stimulus seem to be undervalued and regularly ignored (Khan, et al. 2002; Montgomery & Kirsh, 1997).  Read more commentary on the side effects of psychotropic medication.

Donna M. Midkiff, PsyD