Features

Unlucky Vioxx, Unlucky World

Jerusalem, Israel June 29, 2006

On Sept. 24, 2004, Raymond Gilmartin, former chief executive of Merck & Co., got an alarming message. Peter Kim, Merck's research chief, told him that the panel overseeing the APPROVe clinical trial urged Merck to halt the trial and stop patients from taking Vioxx.

The reason: Patients on Vioxx were twice as likely to have heart attack or stroke as those on placebo. On September 30, Merck announced that it is withdrawing Vioxx from the world-wide market, with huge economical and legal consequences.

Typical to such senseless harsh effects, on April 21, 2006, a Texas jury found Merck liable for the death of 71-year-old Leonel Garza, a diabetic smoker, who suffered for more than 20 years from heart disease and a quadruple bypass in 1989. Garza's fatal heart attack occurred after taking Vioxx for a week. Merck was ordered to pay $32 million.

When it comes to Vioxx, all barriers of Reason, Science and Logic are broken - after Merck publicly admitted that Vioxx was lethal after 18 months, based on the APPROVe clinical study.

Merck's decision to withdraw Vioxx followed a report on APPROVe clinical study that applied 'gold standard' statistical methods designed early last century. Merck's decision will long be remembered as the worst blunder in recent history that occurred due to unfortunate application of statistical method and analysis.

Applying contemporary Data Mining and Knowledge Discovery methodologies would have convinced Merck's management that Vioxx was absolutely benign, before the fateful decision was reached.

Limitation of current statistical methods for discovery of effects that are on small subsets of population - became evident to our team when we conducted introductory analysis of de-identified Patient Longitudinal Data ("PLD"), following up the devastating recall of Baycol. Our Longitudinal Analytical Methodology improved since then measurably. When the Vioxx crisis erupted, we executed Patient Longitudinal Analysis on prescription history on millions patients. The analysis discovered amazing facts. The wrong perception about Vioxx resulted from the high level of concomitancy between Vioxx and CV related drugs. When interpreted improperly, concomitancy could easily be mistaken with 'causal' effect.

Our Analytical Methodology relied on three components:

• Database compaction: PLD, residing on standard Relational Database, were compacted losslessly by factor of 1:100, reducing the PLD from Terrabytes down to Gigabytes of interactively analyzable structure.
• Analytical Indexing Methodology: Provided new analytical insight on the relationship between Patients, Prescriptions, Regimens etc.
• Longitudinal Data Analysis methodology: Analyzes the trends and interventions for each individual patient over time, reporting and analyzing the summary of evolutions and intervention's effect.

The initial reaction to the report by famous and prestigious scientist was in line with earlier reactions, baseless by any statistical or scientific standard. Headlines stated: "New data show Vioxx remained dangerous at least one year after use" or "Merck Caught Misrepresenting Vioxx Risks Again" and "New Vioxx data imply earlier risk"

Contrary to such reports, the Follow up data positively confirms that Vioxx has no causal effect for cardiovascular events.

Here is a white paper discussing the APPROVe follow up report and its implications.

Avi Shatz is the founder CTO of EdenBase Ltd.

Avi authored white papers on the Vioxx recall.

Find earlier press releases:

• www.medicalnewstoday.com/medicalnews.php?newsid=16502
• www.kdnuggets.com/news/2004/n22/3i.html
• www.kdnuggets.com/news/2005/n09/7i.html

For Further Information contact: Avi Shatz, in Jerusalem, ISRAEL

US toll free:
Office 1-888-202 9801, Cellular 1-877-271-7330

or by email: avi.shatz@edenbase.com