Pfizer pioneered the market for dog painkillers when it introduced Rimadyl in 1997. Nearly 15 million dogs have taken it, many for pain from degenerative joint disease, or osteoarthritis.

Rimadyl now has four competitors with more likely. The market for dog arthritis pain medications tops $130 million a year and is growing about 13% a year, consulting firm Wood Mackenzie says.

But 3,200 dogs have died or been put down after taking the drugs, Food and Drug Administration records show. Almost 19,000 dogs have had bad reactions to them.

The FDA says the drugs are safe, if properly used. Drugmakers say fewer than 1% of dogs have bad reactions to them. If not for the drugs, many dogs might be put to sleep sooner, they say.

"In theory, the drugs are safer for dogs than aspirin," says Steve Marks, head of small-animal medicine at the University of Illinois.

Yet the deaths and other adverse events are a reminder that all drugs pose risks, including those for animals. They also underscore that a drug's risks may not be fully known until the drug is widely sold, despite the FDA's stamp of approval.

Several million dogs received Rimadyl before its warning label was updated to add mention of death in rare cases. No. 2 pain reliever Deramaxx was marketed for a year before its label was so changed.

Despite stronger warnings and other safeguards, the drugs continue to create controversy. Some dog owners and veterinarians say the drugs are being overprescribed by vets who don't always give risk information to owners and who, like doctors for humans, are often educated on drugs by pharmaceutical companies.

"There are no safe drugs. There are only safe doctors," says Robert Rogers, founder of the Critter Fixer Pet Hospital in Spring, Texas. "The large number of adverse events occur with these drugs because veterinarians don't know how to use them." Drugmakers share the blame, he says. Their sales representatives often fail to present adequate risk information.

Adverse-event reports

Through November 2004, the FDA received almost 13,000 adverse-event reports about Rimadyl, far more than for any other dog pain reliever.

Pfizer's database includes almost 20,000 adverse-event reports. The FDA's data include those "possibly" or "probably" linked to the drug. Adverse events for all drugs are believed to be under-reported.

Pfizer won't release Rimadyl's revenue, but it says Rimadyl has the most reports because it's the oldest and biggest dog pain reliever.

Deramaxx has been used by about 1 million dogs since its 2002 launch, owner Novartis says. The FDA's data include 2,813 adverse-event reports for Deramaxx, including 630 dogs who died or were put down.

Rimadyl, with a generic name of carprofen, originated at Roche Laboratories as an anti-inflammatory drug for people. Deramaxx was developed by Novartis for dogs but was discovered by G.D. Searle, where Celebrex originated.

Both Rimadyl and Deramaxx are non-steroidal anti-inflammatory drugs, or NSAIDs, as is ibuprofen and aspirin for humans. Deramaxx is also a COX-2 inhibitor, like Celebrex and Vioxx, the pain reliever for people withdrawn from the market in September because of heart attack and stroke risk.

NSAIDs cause many of the same problems in dogs that they do in humans: Vomiting, diarrhea and ulcers are the most common. Reports of heart problems associated with dog NSAIDs are rare, the FDA says.

Before getting FDA approval, Rimadyl was tested in 549 dogs; Deramaxx in about 700. That's far fewer than the number of subjects in clinical drug trials for human drugs. Erectile dysfunction drug Cialis, for instance, was evaluated in more than 4,000 patients before it got FDA approval.

When it hit the market, Rimadyl was heralded as a "wonder drug" for freeing dogs from pain. Happy, bounding dogs were shown on TV commercials. "No drug in the history of veterinarian medicine has been met with such instant success," says Steve Dale, host of radio show Steve Dale's Pet World.

Rimadyl's fast start had a dark side: Reports of dogs suffering liver damage and other effects began pouring in. One of those dogs was George, a 12-year-old chocolate Labrador retriever owned by Jean Townsend of Johns Island, S.C.

She saw the Rimadyl ads and asked her veterinarian whether Rimadyl could help George, who, while seemingly healthy, was moving slower. Rimadyl seemed to work, but after several weeks, George sometimes couldn't move. He started vomiting and had blood in his stool. Within 30 days of starting Rimadyl, George was put down. An autopsy showed liver damage, ulceration and bleeding.