Published Conclusions Must Accurately Reflect Research Results
Dr. Vineet Gupta details several important ideas in his rapid
response to Kamran Abbasi’s “Editor’s Choice” article from 15 January
2005. (1) I am responding specifically to Dr. Gupta’s remarks suggesting
that, “uniform guidelines should be released for declaration of the
results and not left to the companies to define the registration.” (2)
My understanding of the above comment is this: It is essential that
the declaration of any results---those summaries that are provided in
clinical research papers that form the basis of the abstract and
conclusions---be guided by certain accepted, mandated and rigorously
applied publication standards. They must accurately represent the findings
of the clinical research.
The reason is straightforward. Neither voluntary nor mandated
registration of clinical trials at their inception or the publication of
all clinical trial results requires that the conclusions of those research
findings accurately reflect the content of the research.
Here is a case-in-point from a paper that I recently reviewed from
previously published neuropsychiatric research. The results of the
research were published in The Journal of the American Medical Association
in an article entitled, “Gabapentin for the Treatment of Postherpetic
Neuralgia.” Two of the authors of the study, Leslie Magnus-Miller, M.D. and Ms Paula Bernstein, MS, were paid employees of Parke-
Davis. The pharmaceutical company, subsequently acquired by Pfizer, also
paid the compensation for study expenses related to this research effort.
The conclusion of the article summarizes the author’s findings as
follows: “Gabapentin is effective in the treatment of pain and sleep
interference associated with postherpetic neuralgia. Mood and quality of
life also improve with gabapentin therapy.”
There is a problem with those robust summary statements, however.
They do not accurately reflect the data found within the body of the
results section of the same research paper. Certain findings within the body of the published ‘results’ and ‘safety’ sections do not corroborate the authors’ stated conclusions about mood and quality of life.
Specifically, the clinical research results demonstrated that the
persons receiving gabapentin suffered emotionally and were ‘less well’
when compared to those receiving placebo. That finding, numerically
depicted in the ‘Role-emotional’ section of the Quality of Life Domains as
judged by the authors’ use of Short Form-36 Quality of Life Questionnaire,
was omitted from mention in the abstract’s Results and/or Conclusion
sections. Likewise, in the same Short-Form 36, under item ‘General
Health,’ individuals receiving placebo improved more and were judged
healthier than the comparison gabapentin group. That research finding was
also ignored in the Results and/or Conclusion sections of the abstract.
Finally, the gabapentin ‘treatment’ group had a higher drop-out rate and
suffered significantly more adverse events than the dissimilar placebo
group to which it was compared---findings that did not make it into the
The significance of these omissions: Inaccurate conclusions from
pharmaceutically-sponsored research are published in a prestigious peer-
reviewed journal. The authors specifically misrepresent their own findings
in their otherwise robustly positive summary statements by failing to
include data that contradicts their published conclusions.
And, the problem with that? Pharmaceutical firms base much of their
marketing efforts, including physician detailing, on the summary
statements and the conclusions of work that they choose to support and
publish. If those conclusions misrepresent their findings, then the
subsequent advertising and the marketing of their pharmaceuticals based
upon that misrepresentation is, at best, inaccurate and, at worst,
purposefully and deceptively misleading.
1. Abbasi, K. Trial results: The next battle. BMJ: 2005; 330 (15
2. Gupta, V. Rapid Response to No. 1(above)
3. Rowbotham, M et al. Gabapentin for the Treatment of Postherpetic
Neuralgia. JAMA. 1998; 280(21): 1837-42.
Competing interests: No competing interests