WASHINGTON — The widely used diabetes drug Avandia should not be taken off the market, government health advisers overwhelmingly recommended yesterday, saying evidence of an increased risk of heart attack doesn’t merit removal.

The nonbinding recommendation to the Food and Drug Administration came on a 22–1 vote by the panel.

“We’re being asked today to take a very draconian action based on studies that have very significant weaknesses and are inadequate for us to make that kind of decision,” said Rebecca Killion, a diabetic from Bowie, Md., and the panel’s patient representative.

However, in an earlier 20–3 vote, the panelists said available data show the drug does increase heart risks.

Panelists said the drug’s label should include a so-called “blackbox” warning, the most severe the FDA can require, to flag that risk. Some suggested the label caution against using the drug together with insulin because doing so may elevate heart risks. That joint use is currently FDA-approved. The experts also asked that the drug be studied further.

The FDA isn’t required to follow the advice of its advisory committees but usually does.

The manufacturer, GlaxoSmithKline PLC, earlier recommended continuing long-term studies of the drug and updating the label to inform doctors and patients of what’s known so far about any heart risks. FDA scientist Dr. David Graham said waiting for more results could subject as many as 2,200 people a month to serious side effects from the drug.

Dr. Graham also told the joint panel of experts that the drug’s heart risks, combined with its lack of unique short-term benefits in helping diabetics control blood sugar, meant continued sales were not justified.

But Glaxo contended there is no increased risk, citing its own analyses of studies of Avandia, also called rosiglitazone.

“The number of myocardial infarctions is small, the data are inconsistent and there is no overall evidence rosiglitazone is different from any other oral antidiabetes agents,” said Dr. Ronald Krall, the company’s senior vice president and chief medical officer.

Previously, the FDA had said information from dozens of studies pointed to an increased risk of heart attack.

That conclusion swayed the panel but apparently did not rise to the level of requiring any regulatory action more dire than beefed-up warnings and continued scrutiny.

“It’s suggestive but by no means conclusive,” said Dr. Thomas Pickering, an assistant professor of medicine at Columbia University Medical Center.

The lone dissenting panel member on the main vote, Arthur Levin, said there was a strong suggestion of a safety signal. That, along with widely shared doubts that further study would settle the issue and the enormity of the potential risk to the public health, moved him to vote “no.”