February, 2002, Vol.16, issue 6, Page 9

How to increase trust in reporting pharmaceutical research findings

Mark Goodman [1] and Brian Hess [2]
[1] University of Medicine and Dentistry of New Jersey-School of Osteopathic Medicine (UMDNJSOM), West Orange, NJ, USA, [2] National Commission on Certification of Physician Assistants (NCCPA), Norcross, GA, 30092, USA

The recent debacle concerning allegations of intentionally withholding damaging longitudinal outcomes data by investigators supported with funds from the manufacturer of Celebrex [1-3] once again raises the issue of a temptation to compromise ethics in research. Researchers and authors supported by pharmaceutical manufacturers are in an unenviable position due to pressures from their ‘dual relationship’, wherein they are simultaneously functioning in two conflicting roles – to serve as doctoral level scientists, and to recognize and report findings satisfactory to the parent drug maker on whom the scientist may be dependent for current and future remuneration. Medical journals generating revenue from drug advertisements and the journal editors themselves also experience this dependent, symbiotic, dual relationship. ‘The journals are the major force for quality control in scientific work’ [4] and, coincidentally, the journals become the easiest point in the system to implement fail-safe mechanisms rapidly, in order to increase further the integrity of the manuscript’s findings.
The following are equitably and respectfully proposed:

(1) Within each medical journal’s editorial staff there should be a doctoral level biostatistician, without corporate funding affiliation, who will be supplied with the submitted manuscript’s summary, methods and results sections. The staff biostatistician must possess the methodlogical acumen to reconcile research design abnormalities that may skew reported results.

(2) At the time each item of dependent variable raw data is collected, the coded data are to be electronically and unerasably sent to an independent databank that is managed independently (e.g. by the Food and Drugs Administration in the USA) and supported from pharmaceutical company licensing fees. Individual databanks will be accessible by password supplied only to the medical journal staff biostatistician who is reviewing the submitted manuscript. This biostatistician will access raw data in the data bank for purposes of reanalyzing all collected data (ensuring the statistical tests fit assumptions of the data). This will ensure accuracy of ‘observed results’ as compared with the manuscript’s reported results.

The proposed extra step performed by the journal’s biostatistician on raw data, previously formatted, coded and entered by the researcher’s biostatistician, may consume some 60 minutes, but add immeasureable confidence to editors, reviewers and journals publishing pharmaceutical studies.

As baby boomers reach geriatric age, drugs will increasingly become the sine qua non for therapy and alternatives to surgery. With journal editors as quality control gatekeepers endowed with the responsibility for disseminating knowledge in an accurate, scholarly manner, the public is owed this additional protection.


1. Gottlieb S. Researchers deny any attempt to mislead the public over JAMA article on arthritis drug. Br Med J 2001; 323;301.
2. Lichtenstein DR, Wolfe MM. COX-2-Selective NSAIDS; new and improved? J Am Med Ass 2000; 284; 1297-1299.
3. Okie S. Missing data on Celebrex; full study altered picture of drug. The Washington Post 5 August 2001; A 11
4. Giantz SA. Biostatistics: How to detect, correct and prevent errors in the medical literature, Circulation 1980; 61:1-7.