The controversy over COX-2 inhibitor drugs such as Vioxx and Celebrex served as a springboard for a lively two-day discussion in June on challenges in assuring the integrity of reporting and of patient safety in therapeutic trials. The editors of prestigious medical journals hashed over the issues with representatives of industry, academia, and government at the Department of Biostatistics' 2006 HSPH/Schering-Plough Workshop, held at The Conference Center at Harvard Medical School (HMS) on June 1 and at HSPH on June 2.
The workshop focused on the reporting of results in medical literature and on the interim evaluation of accumulating information from ongoing clinical trials. Topics included clinical trial registration, sharing safety information during a trial, the FDA's guidelines on independent monitoring of trials, and editorial strategies for ensuring the accuracy of research reports.
Perhaps most recognizable to the general public was the subject of COX-2 inhibitor drugs, which made headlines in 2004 after Merck & Co. announced that it was voluntarily withdrawing Vioxx from worldwide markets after data indicated that the drug was associated with cardiovascular risks. The FDA has since asked Pfizer to withdraw a drug called Bextra and to include a boxed warning about risks on the label of another COX-2 inhibitor drug, Celebrex.
In 2000, The New England Journal of Medicine (NEJM) published a study of Vioxx sponsored by Merck. The paper was later criticized for apparently not including information on three subjects who had experienced heart attacks while taking Vioxx. The study authors asserted that those cases were not included because they had occurred after a predetermined cutoff reporting date. For recent news on Vioxx, see box below.
NEJM has now changed how it handles submissions, said the journal's editor, Jeffrey Drazen to the workshop attendees. Drazen is also a professor at Harvard Medical School and in the HSPH Department of Environmental Health. For example, editors now ask authors for more information about adverse events data and about how faithfully the trial's execution follows the study protocol.
Another prestigious medical journal, The Journal of the American Medical Association (JAMA), has also changed how it manages submissions. The journal's editor, Catherine DeAngelis, explained at the workshop that JAMA now requires a statement for industry-sponsored studies signed by an author not employed by the company; the signator takes responsibility for the data's integrity and for the data analyses' accuracy. In addition, JAMA requires industry-sponsored studies to be subject to an independent statistical analysis performed at an academic institution.
A second issue discussed at the workshop was the need for a centralized clinical trial registry where newly diagnosed patients could find trials in which they might enroll. Thomas Haverty, Group Vice President of Global Clinical Operations at the Schering-Plough Research Institute, advocated one centralized registry for all trials. Until that becomes a reality, Schering-Plough is setting up its own registry listing its trials.
Whether such registries should include study findings is a subject of debate. Attaching data would make it easier for researchers who are conducting metaanalyses to get access to all of the available data on a subject, said Harold Sox, editor of Annals of Internal Medicine. However, he cautioned that such a practice would mean making publicly available results that haven't been filtered by the journal peer-review process.
Workshop attendees also discussed the FDA's recent approval of guidelines for Data Monitoring Committees. The committees, named by trial sponsors, track safety data during trials and can recommend that trials be stopped if necessary for safety reasons. At the moment, they are mandatory only for government-sponsored studies.
Janet Wittes, president of Statistics Collaborative, who has served on several DMCs, decried the fact that the FDA no longer refers to the panels as Data and Safety Monitoring Committees or DSMBs.
``I think the 'S' should still be in there,'' she said. ``The primary purpose is safety.''
But Gregory Campbell, director of the Division of Biostatistics at the Office of Surveillance and Biometrics, FDA, said that safety remains the agency's top concern. ``We're not ignoring safety at the expense of efficacy,'' he said.
—ML
Vioxx Study Error Corrected
In May 2006, Merck and Co., the makers of Vioxx, acknowledged that it had made an error when it reported that a statistical test showed Vioxx caused heart problems only after 18 months of use. NEJM posted a correction online on June 26, to a paper from the company published in the journal in 2005. The correction is accompanied by a Perspective piece by Stephen Lagakos, Henry Pickering Walcott Professor of Biostatistics at HSPH. Both pieces will appear in the July 13, 2006, printed issue of the journal.
Dept. of Biostatistics Confers Award for Distinguished Alumnus
Daniel Seigel accepts the award from Dianne Finkelstein of the HSPH Department of Biostatistics
Daniel Seigel, DS 61, formerly with the National Institutes of Health, received an award from the Department of Biostatistics at a ceremony on May 31 at HSPH. Seigel delivered a lecture called "Outcome Preferences and the Statistician."
Seigel played a central role in many national and international studies that changed health care practice. For example, he was involved in a study on risks of oral contraceptives that led to changes in recommended doses and in a study on the risks of amniocentesis that found this procedure to be safe. He also co-wrote fundamental papers on the epidemiology of several eye disorders, such as macular degeneration and retinal detachment.
—DF
Marvin Zelen Leadership Award in Statistical Science Given to NCI Researcher
Mitchell Gail
Mitchell Gail from the National Cancer Institute was named the winner of the 2006 Marvin Zelen Leadership Award in Statistical Science, named after a faculty member at FAS and at HSPH. Gail delivered a talk at HSPH on June 2 entitled "Absolute Risk: Clinical Applications and Controversies."
In his talk, Gail showed how absolute risk for a given patient can be estimated by multiplying relative risk estimates from observational studies by total risk estimates from health surveillance data.
He then discussed how absolute risk elements are often evaluated in terms of two concepts. "Calibration" is the accuracy of the estimate for various subpopulations. "Discrimination" is the difference in risk between those who do and do not develop the disease. Gail showed that while high discriminatory power may be necessary for screening, the tool may not be necessary for counseling patients about whether or not to use treatments that modify risk.
—DS
Copyright, 2009, President and Fellows of Harvard College
