A New Look at an Old Study: How Do We Stop Data Spinning?

Nassir Ghaemi, MD, MPH


October 20, 2015

How to Address "Spinning" Results

Readers may wonder how data could be presented in such a way that the main planned statistical analysis is not reported as originally planned.

Many clinicians do not know that the database gathered by a pharmaceutical company when it conducts a research study is private property, protected by patent law. Pharmaceutical companies are not obligated to provide those data to anyone except the federal government if they seek FDA indication. Even then, the FDA is obligated to protect the actual data (as opposed to analyses of those data) from public exposure; they remain private property. Not even the academic authors have a right to see those data. (I know: I once asked AstraZeneca for access to their data when they asked me to be an author on a randomized trial; they turned me down. They were later subject to a class-action lawsuit for not fully disclosing results in their studies, in which I was not involved).

The prestigious journal that published the 2001 paroxetine study may not have known all of the above. It took a class-action lawsuit for the process that led to making those data available to the researchers who published the BMJ reanalysis. The BMJ study authors weren't paid or sponsored for their analysis, but at least one of them was involved in the class-action lawsuit related to this drug and this company and this study about a decade ago. (This is reported in the acknowledgements as follows: “JJ [coauthor Jon Jureidini] has been paid by Baum, Hedlund, Aristei and Goldman, Los Angeles, CA, to provide expert analysis and opinion about documents obtained from GlaxoSmithKline in a class action over Study 329.”). Thus, the reanalysis wasn't conducted for purely altruistic purposes, which shouldn't be surprising, as something isn't given for nothing. (The authors estimated putting over 100,000 hours of work into the reanalysis, which involved reviewing thousands of pages of case documents from patients, as well as extensive email communication with the manufacturer).

Have things gotten better since 2001? The focus in the past has been on publishing negative studies. The problem suggested by this reanalysis is not that negative studies are not published, but rather that they can be spinned as positive studies. The issue is not whether a study is published in this case, but how it is published.

The creation of the registry, which requires transparency regarding planned statistical analyses before a study is conducted, should help in future publication of pharmaceutical industry-sponsored clinical trials by providing at least some knowledge of what had been planned for statistical analysis.

But we still continue to have the problem of not having access to the databases of pharmaceutical industry-conducted studies. Patent law will not change any time soon, and those data are treated as private property rather than public domain. There is a conflict here between scientific standards, which require open access to data, and private enterprise practices, which require patent protection.

A compromise should be considered, like requiring independent analysis of randomized clinical trial data obtained by a pharmaceutical company and intended for regulatory approval. The independent analysis could be conducted with confidentiality, to protect patent rights, while giving the larger scientific public assurance that data interpretations were not solely reliant on the pharmaceutical companies' approval.[3]

To date, there is no such requirement for major medical journals. For now, there only seems to be litigation around this problem, with lawyers fighting over opposite perspectives. This approach is unlikely to lead to a solution. Drug companies that make billions of dollars yearly on a drug can afford to lose a lawsuit every few years for a few hundred million dollars. The companies and the lawyers get rich; the public health suffers in the meantime.

It is disappointing that the scientific community cannot achieve a consensus on how best to monitor its relations with private enterprise and government agencies so as to meet the demand of the ultimate scientific ethic of seeking truth.

(This post is revised in accord with a few of the points made in the comments below by one of the BMJ study authors.)


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.