(Reuters) – An analysis by U.S. government scientists found GlaxoSmithKline Plc’s controversial diabetes drug Avandia increased heart risks and deaths compared to a rival pill, according to an email from a Food and Drug Administration reviewer who worked on the study.

The reviewer, Dr. David Graham, known for his strong opposition to the now-withdrawn painkiller Vioxx, and other government staffers reviewed data on patients in the Medicare health insurance program for the elderly who used Avandia or Takeda Pharmaceutical Co Ltd’s competitor Actos.

The findings, which have not been published, come just weeks ahead of a highly anticipated public meeting next month on the safety risks of Avandia. A 2007 study first raised concerns about the drug’s cardiovascular effects.

Graham has argued that Avandia is more dangerous to the heart than Actos and should be removed from the market.

In the newest research, Graham and colleagues at the FDA and the Centers for Medicare & Medicaid Services looked at data on nearly 230,000 Medicare patients and found Avandia increased the risk of stroke by 27 percent, heart attack by 25 percent and death by as much as 17 percent.

Overall, there were 48,000 excess complications with Avandia use in older patients between 1999 and June 2009, Graham said in a May 28 email to FDA Commissioner Margaret Hamburg. But because Medicare patients are 65 and older, and more than 60 percent of Avandia users are under age 65, “the national impact is probably 100,000 or more,” he wrote. A copy of the email was obtained by Reuters.

An FDA official confirmed the email was authentic. Graham did not respond to requests for comment.

FDA spokeswoman Meghan Scott said agency officials reviewed Graham’s findings and told him he could submit them for publication with a disclaimer that the conclusions were not those of the agency. The FDA has not yet completed its review of Avandia’s safety issues, she said.

Another FDA spokeswoman, Karen Riley, said Graham’s findings would be included among the presentations at the July advisory panel meeting.

The results will likely amplify concern over Avandia’s risks, despite the fact that it is no longer a major product for the British drugmaker.

Sales of Avandia have declined sharply in the wake of the 2007 Cleveland Clinic study that initially raised the heart issue, and the drug loses its U.S. patent protection in 2012. Liability claims are still an issue.

Global sales of the drug topped $3 billion in 2006 but fell to $1.2 billion in 2009.

According to Graham’s email, the government scientists are seeking to publish their findings in the Journal of the American Medical Association (JAMA), a peer-reviewed medical publication. A spokeswoman for JAMA said the journal does not comment or confirm any submissions before publication.

Graham, an outspoken safety reviewer who is often at odds with official FDA positions, voiced concern that FDA officials were blocking the study from becoming public.

But the FDA’s Riley said “senior management did not seek to suppress this study.” It was reviewed “expeditiously” and has been submitted to JAMA, she said.

A known whistleblower, Graham has drawn attention to a number of other prescription drugs since the 2004 withdrawal of Merck & Co Inc’s Vioxx. Some of those medicines were later withdrawn or saw strict new warnings.

A study Graham had done looking at the heart risks of Vioxx in late 2005 using data from a private insurer was also repressed, he wrote, “so you can understand that I am very sensitive to anything that remotely smells like suppression or censorship.”

In 2007, Graham argued to an advisory panel of outside experts that Avandia sales should be stopped. The panel voted 22-1 to urge the FDA to keep the drug on the market.

Glaxo has defended Avandia, known generically as rosiglitazone. The company says scientific evidence does not establish that Avandia increases the risks of cardiovascular problems such as heart attacks. The current warning on Avandia about those risks says the overall data are “inconclusive.”

In an email on Friday, Glaxo said it was unclear if Graham’s newest study had been peer-reviewed, “and until then, it would be premature to comment.”

“However, it is important to note that there are inherent limitations with retrospective observational studies, including variables that were not considered nor corrected for when conducting the study, which can significantly impact the validity of the data,” the company said.

At the FDA’s public meeting in July, an advisory panel will weigh data on the drug and could offer recommendations ranging from keeping the drug on the market, to calling for further restrictions, to pulling the product altogether.

(Reporting by Susan Heavey and Lisa Richwine; editing by John Wallace)