CCBYNC Open access

Restoring Study 329: efficacy and harms of paroxetine and imipramine in treatment of major depression in adolescence

BMJ 2015; 351 doi: (Published 16 September 2015) Cite this as: BMJ 2015;351:h4320

De-coding serious adverse events in Study 329: The work of the RIAT team

Great is the power of steady misrepresentation - but the history of science shows how, fortunately, this power does not endure long.
— Charles Darwin 1902 [1]

The RIAT project and this article in particular make important contributions to the world wide movements towards greater transparency in biomedical research and in particular commercial influences in the conduct and ultimate impact of clinical trials on the health and safety of patients. In addition, the authors also contribute to the fields of social science and history of medicine by their meticulous recovery and analysis of primary data found in Case Report Forms (CRF) and Clinical Study Reports (CSR).

Their approach takes us to the crucial moment when the decision to use certain words or phrases construct explanatory threads and assigns coded meanings to the experiences of trial participants. Meaning is made by word choice embedded in decisions about the use of a particular coding system dictionary (ADECS). The choice to use certain words or phrases, over others, becomes the basis for analysis and ultimately becomes the statistical underpinning for assertions about the safety and efficacy of drugs.

In Box 2, LeNoury and colleagues outline the potential barriers to accurate reporting of harms. It is worthwhile to look at this more closely by reading Appendix 3 to get a feel for what the RIAT team has accomplished.

Appendix 3 of Le Noury and colleagues ( offers a detailed examination of primary data, in an almost ethnographic sense, from both the CSR and perhaps even more importantly by gaining hard-won access to the rarely glimpsed CRF documents. It is at this interface that the crucial narratives about suicidal behaviours among adolescent (ages 12 to 18) trial participants are noted and transformed into what will become the published face of the clinical trial.

In Appendix 3 and especially Table C, “Cases of suicidal & injurious behaviour in Study 329” (see above link to Appendix 3), Le Noury et al. usher us into a decision-making system with data from a variety of narrative sources. We see verbatim terms describing adverse events, Clinical Study Reports, and Case Report Forms enabling us to interpret how adverse event narratives have been misrepresented. We urge colleagues to review this Table in order to understand how the deployment of euphemistic terminology in coding transforms and minimizes the harrowing experiences of trial participants. The Table provides insight into a decision-making trail of revisions and edits which transform serious adverse events (e.g. the intentional swallowing of 80 Tylenol tablets) into the bland category of “emotional lability”.

The RIAT authors have challenged the oft-repeated marketing message of the remarkable efficacy and safety profile of Paxil (see 2012 U.S. Department of Justice;[2] see also McHenry and Jureidini 2008 [3]) and have brought much needed attention to a full array of data. They have taken us into the terra incognita of coding systems, language choices and investigator decisions and enabled those data to speak.

We hope the medical profession is listening.

1. Darwin C. XV. Recapitulation and Conclusion. The Origin of Species. New York: P. F. Collier & Son Company; 1909. p. 499–530.
2. United States District Court for the District of Massachusetts. United States et al. v. GlaxoSmithKline PLC et al. 2012. p. C.A. No. 11–10398 – RWZ.
3. McHenry LB, Jureidini JN. Industry-sponsored ghostwriting in clinical trial reporting: A case study. Account Res Policies Qual Assur. 2008;15(3):152–67.

Competing interests: No competing interests

27 September 2015
Harriet G. Rosenberg
Professor Emerita
Adrienne Shnier, Ph.D. Candidate, York University, Toronto, Canada
York University
4700 Keele Street, Toronto, Canada