Publication Bias and the Pharmaceutical Industry: The Case of Lamotrigine in Bipolar Disorder

S. Nassir Ghaemi, MD, MPH; Arshia A. Shirzadi, DO; Megan Filkowski, BA


Medscape J Med. 2008;10(9):211 

In This Article


We searched the Web sites of the following companies with FDA-approved drugs for the treatment of bipolar disorder patients: Abbott Laboratories (Depakote brand of divalproex sodium), GlaxoSmithKline (Lamictal brand of lamotrigine), Pfizer Inc. (Geodon brand of ziprasidone), Eli Lilly and Co. (Zyprexa, Symbiaxbrands of olanzapine and olanzapine-fluoxetine combination), Janssen Pharmaceutica (Risperdal and Risperdal Consta brand of risperidone), Bristol Myers Squibb (Abilify brand of aripiprazole), and AstraZeneca (Seroquel brand of quetiapine). We supplemented searches of the primary company Web sites with searches of the Web sites for each drug. If no studies were identified, or if links were provided, we accessed the Web site established by the Pharmaceutical Research Manufacturers Association (PHRMA) for clinical trial results ( and that of the National Institutes of Health (NIH) for current clinical trials ( Table 1 describes the studies found on their Web sites, whether they were positive or negative, and whether they were previously published. Published studies of the same drugs were assessed by a computerized MEDLINE search of the National Library of Medicine (NLM) database (1966–2006).

Negative vs Failed Trials

Our use of the term "negative" to describe the outcomes of studies is defined as follows: In the strictest definition, this refers to a study in which the experimental drug is the same as placebo, and in which a proven active control is more effective than placebo.[18] If an active control fails to separate from placebo, as well as the experimental drug, then this study would be seen as a "failed" study, meaning that no conclusion could be drawn for or against efficacy, since an already proven treatment also failed to show benefit. Usually in that case, peculiar characteristics of the sample (such as a low severity of illness leading to high placebo response) can obscure a real treatment effect.[18] In the absence of an active control group, which is often the case for FDA regulation-oriented pharmaceutically sponsored studies, one cannot be definitively certain that equivalence of drug to placebo represents a "negative" outcome.[18] However, if the study is adequately powered to demonstrate a modest effect size, then one can at least conclude that the study is most suggestive of lack of moderate or greater benefit.[7] In that sense, we will define studies as negative in this report if drug is equivalent to placebo in the absence of an active control group. Further, all definitions of difference between drug and placebo will be limited to the primary outcome measures, since secondary outcome measures are associated with inflated chance of statistically significant findings (P value below .05).[19] This limitation does not mean that secondary outcomes are always false or unhelpful, but rather one should look for multiple measures showing robust effects, rather than isolated borderline P value-based analyses.[20]

Where the comparison is of a drug to an active control without a placebo group, usually the purpose of the study is to demonstrate "noninferiority" or "equivalence" to a proven treatment.[7] If the experimental drug is equivalent to or better than the proven treatment, the study is considered positive. If the experimental drug is less effective then the proven treatment, the study is considered negative.[7]


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.