You're taking a new medication and have dry mouth and feel dizzy.
Want to know how many other people have reported those side effects—and how your drug compares with similar drugs?
The U.S. Food and Drug Administration has millions of such "adverse event" reports, ranging from fatigue to fatal heart attacks, for thousands of prescription drugs dating back to 1969. But the information hasn't been readily accessible—until now.
A start-up company, AdverseEvents Inc., has streamlined the FDA's often impenetrable database and made it easy to search the adverse-event reports for more than 4,500 drugs, free and online. Another start-up, Clarimed LLC, has done the same for reports filed with the FDA on 130,000 medical devices, a far more complex group that runs the gamut from syringes to stents to tanning beds and diagnostic machines that could impact tens of thousands of lives.
Both companies, which launched in September, see their services as empowering patients, many of whom now comb Internet discussion boards for medical information. "If your doctor tells you to take a drug and it's three times more likely to give you a heart attack than another drug, not having that information seems foolish," says AdverseEvents President and co-founder Brian Overstreet.
While basic searches will remain free, AdverseEvents plans to charge consumers $10 a month for access to full drug reports starting Wednesday, and will offer health-care professionals and businesses more detailed information for additional fees. Clarimed may follow suit. Both websites offer a way to file reports to the FDA, but few visitors have done so.
Nora Iluri, CEO and founder of Clarimed, likens these efforts to the advent of J.D. Power and Associates safety ratings for cars. "Suddenly, manufacturers started competing on quality," she says. "The best way to drive quality improvements is to make things crystal clear and transparent as possible."
The adverse-event reporting system for drugs (known as AERS) helps the FDA monitor side effects that crop up after a medication is approved and used in the real world. (The system for devices, called Maude, for Manufacturer and User Facility Device Experience, started in the 1990s.) AERS has received 6.4 million reports, and the volume has risen steadily. There were 759,000 for drugs and 238,000 for devices in 2010. But reporting is voluntary, and represents only a fraction of all the side effects and malfunctions, the FDA says.
Agency analysts mine the data for worrisome trends that prompt further investigation and sometimes stronger warning labels or even removal from the market. The cholesterol-lowering drug Baycol, for example, was withdrawn in 2001 after 52 deaths from rhabdomyolysis, a muscle and kidney disorder, turned up in the adverse-event files. People seeking AERS information can download raw quarterly data from the FDA's website, but it isn't cumulative and requires technical expertise to use. They can also file a Freedom of Information Act request for more specific data, but what they get back may be voluminous and impenetrable. That's what Mr. Overstreet, a data-marketing entrepreneur, found when a friend's wife suffered a severe drug reaction and he went looking for information. He and his fellow data experts found the FDA's AER files indecipherable. "That's when we said, 'Let's fix this,' " he says.
One problem is that the data are sometimes jumbled. Most reports come through drug and device manufacturers, but patients, physicians, family members—even lawyers—can send reports to the FDA, and they often contain errors and inconsistencies. "There are at least 440 different ways to spell Ambien and each has a separate file at the FDA," Mr. Overstreet says.
AdverseEvents Inc. developed an algorithm that filters out duplicates and combines spelling variations, reducing over 200,000 drug names to about 4,500. It also made the data easily searchable and comparable for thousands of conditions and side effects back to 2004.
Still, the FDA's data have other limitations that some critics say make it potentially misleading. For one, there is no way to determine whether a side effect is due to a drug or a coincidence. (Device malfunctions are even trickier, since operator error or surgical skill can affect how they perform.) For another, the reporting doesn't necessarily mirror the true incidence of problems. New drugs tend to generate more reports than older ones, and a negative news story about a drug or device can prompt a sudden spike in reported problems. Expectations matter, too. Chemotherapy drugs that cause severe side effects get far fewer reports than drugs for, say, heartburn. What's more, the FDA files don't indicate how widely a drug or device is being used, so there is no perspective. For example, Lipitor tops all statins for side effects, with 76,535 adverse-events reported from 2004 to 2011 (most often muscle pain, pain in extremities and muscle weakness), compared with 34,938 for rival Crestor, according to AdverseEvents.com. But users aren't told that Lipitor was prescribed far more frequently, so the proportion of problems was smaller. AdverseEvents plans to post prescription data for some drugs.
But sales information isn't available for many medical devices—even for costly implants like artificial joints, says Ms. Iluri. Without such numbers, she says, "Neither we nor the FDA can easily compare safety for devices across manufacturers."