Oct. 7, 2005

GlaxoSmithKline Misled FDA, Doctors and Patients with Faulty Asthma
Drug
Study, Public Citizen Writes in Lancet Medical Journal

Drug Maker Included Data From Six Months After Trial Ended, Skewing
Results

WASHINGTON, D.C. - GlaxoSmithKline presented misleading results to the
Food and Drug Administration (FDA) from a study of the popular asthma
drug salmeterol (Serevent, known as Advair when combined with the
steroid fluticasone), Public Citizen writes in a letter in this week's
issue of The Lancet.

In 1996, the Salmeterol Multicenter Asthma Research Trial was initiated

to study tens of thousands of asthma patients who received either
salmeterol or a placebo. The study lasted 28 weeks and showed an
increased risk of asthma-related death for those taking salmeterol. The

results have never been published, although GlaxoSmithKline presented
the interim results to the FDA in July 2003, when the drug came up for
review before the FDA's Pulmonary-Allergy Drugs Advisory Committee. The

company submitted final study data to the FDA on August 29, 2003.

However, that data included adverse events that were reported six
months
after the trial ended and were not to be included, according to the
original study protocol. The inclusion of the post-study data reduced
the apparent dangers of salmeterol with respect to four critical study
outcomes, including asthma-related death.

GlaxoSmithKline did not clearly inform the FDA that the final study
included data from six months after the trial had concluded until the
FDA inquired about the results in April 2004. The FDA had presumed the
data were only from the 28-week trial itself, since that was the
"period
of interest," according to the FDA.

Since learning of the suspicious reporting of the study results, the
FDA's Pulmonary-Allergy Drugs Advisory Committee on July 13, 2005,
recommended strengthening the warning on the labels for both Servent
and
Advair, but the agency has yet to make a final decision. Public Citizen

learned of the misleading data presentation from materials provided to
the advisory committee.

"The behavior of GlaxoSmithKline in submitting these faulty data is
deplorable," said Dr. Peter Lurie, deputy director of Public Citizen's
Health Research Group and co-author of the letter. "Absent greater
transparency at the FDA, we will never know how often this kind of
self-serving data analysis occurs."

Public Citizen listed Serevent as a "Do Not Use" drug in its Worst
Pills, Best Pills newsletter (www.worstpills.org) in March 2003 because

of the interim study results.

Salmeterol was dispensed more than 2.1 million times in U.S. pharmacies

in 2004. The combination product, Advair, was dispensed more than 16.1
million times in U.S. pharmacies that year.

###