Prescription for error

Anyone who doubts the National Academy of Science's recent estimate that 98,000 people may die each year from medical error should visit the doctor's office with a critical eye. Because of poor safety engineering, opportunities for mistakes are abundant - and readily observable.

In two out of three visits to the doctor, the patient will witness a familiarsight: the physician with pen poised over prescription pad.

So what's wrong with this picture? Deep into the information age, and after decades of human factors research, doctors are still trained to handwrite prescriptions from memory. Both are unacceptably fallible, numerous studies have concluded.

There are about 13,000 uniquely named prescription drug products, including generics, according to IMS Health, a consulting firm. A drug might have dozens of adverse effects and several dosage forms. This constellation of information is constantly changing. The Food and Drug Administration, for example, posts more than 20 safety-related changes to drug labeling every month.

When doctors rely on memory to dash off a prescription - as they typically do - those changes, those risks, are too easily overlooked. Unnecessary mistakes are too easily made.

Just look over the shoulder of Jean Dinwiddie, a drug specialist for the George Washington University Health Plan and also a clinical pharmacist. "I spent an hour in the pharmacy today,'' she said. ''I saw a prescription for a narcotic that was written as 0.5 milligrams instead of 5 milligrams and a prescription for Viagra for 10 milligrams instead of 100 milligrams."

Dinwiddie caught the errors because those two drugs were not available in those doses. But a memory lapse at another point in the prescribing range could go undetected. When small children are involved, a lethal overdose could easily be less than the maximum adult dose. Many a hospital patient has died from the dreaded 10-fold error - getting 10 times too much of a drug because the decimal placement was wrong.

Anyone who has gotten a prescription from the doctor might also have tried to read what was written and puzzled over the cryptic notations. Unfortunately so do pharmacists. ''I can remember distinctly on average of 10 to 12 times a day having to contact a physician's office to make sure I knew what drug was being prescribed,'' said Larry Sasich, a former practicing pharmacist who works now at Public Citizens Health Research Group.

When we use computers for everything from checking the price of radishes at the supermarket checkout to preventing commercial airliners from colliding, why are doctors among the last professionals still operating without modern computer decision support? Is it habit, hubris, or just unwillingness to change that keeps in place an antique pen-and-paper system from the last century?

Whatever the reason, it matters less than the consequences. And one of the most serious is the potential for harmful drug interactions. ''Physicians do not routinely screen for potential drug interactions even when medication history information is readily available,'' declares the National Academy of Sciences report.

It is a serious lapse. Consider two examples: Coumadin, a drug that helps prevent blood clots, is one of the most widely prescribed drugs for the elderly. The drug is so tricky to handle that about 3 percent of Coumadin patients are hospitalized for bleeding. Coumadin can interact with more than 60 other drugs, and any factor that can effect the dose places the patient in danger.

Another widely used drug among the elderly is the heart drug Lanoxin or digoxin. It, too, can interact with dozens of other drugs, and a toxic dose lies quite close to a therapeutic dose. One FDA study found 202,000 people were hospitalized over seven years from digoxin adverse events.

The risks posed by these drugs are just two examples of why it is unacceptable not to have a system to prevent drug interactions before a prescription is written, rather than hoping that a pharmacy might later catch and question it.

Another key safety gap in the doctor's office is the lack of a system to make sure that the most important FDA safety warnings about drugs are received and heeded. For example, Halcion - a best-selling sedative that was at the center of an intense debate about safety - has been linked to numerous cases of traveler's amnesia. Hours after the fact, people would have no memory of having arrived in a foreign country, or of having delivered a lecture to a large audience. Halcion was also a suspect in several celebrated cases of bizarre and even homicidal behavior.

In response to these concerns the British ordered Halcion withdrawn on grounds of an inadequate margin of safety. A more lenient FDA allowed its continued sale here, but warned doctors not to prescribe it for more than 7 to 10 consecutive days.

A few years later, an FDA task force concluded that 85 percent of prescriptions violated that recommended time limit.

A complete failure of the safety warning system occurred with Rezulin, an adult-onset diabetes drug. According to an FDA estimate, Rezulin has already been linked to approximately 400 cases of liver failure so severe that death or an emergency liver transplant were the likely outcomes.

In hopes of identifying early signs of liver damage, both the FDA and the manufacturer recommended liver function tests every month. Later, the FDA's post-market surveillance unit did a study to see how many patients were getting the recommended liver tests. It learned that 99.4 percent of Rezulin patients did not get the needed tests.

If FDA warnings are not being heard, to whom are the doctors listening? Studies have repeatedly demonstrated that the dominant influence on the prescribing pen is the drug company sales representative.

The sales hustle is now so intense that big companies such as Merck and Pfizer can contact 150,000 doctors every week to pitch a new drug. Often that means that, in just a week, a company can contact every single physician in the country likely to prescribe a particular drug

The impact on safety of such sales programs can be seen in the case of Duract, a painkiller withdrawn last year. Introduced in 1997, Duract was similar to ordinary ibuprofen in its ability to relieve pain, but so toxic to the liver that the FDA would not allow it to be prescribed for more than 10 days. Still, despite this big drawback, an effective sales effort prompted doctors to prescribe it to 2.5 million people in its 10 months on the market.

While some companies try hard to give a balanced picture of their drugs' risks and benefits, we can't rely on a system in which making the sale is obviously the priority.

It's time to reengineer the way drugs are prescribed in the doctor's office. Instead of relying on memory or a two-minute sound bite from a drug sales representative, doctors need good computer systems and accurate, unbiased, and up-to-date drug information. Drug prescriptions should be entered and double-checked with a computer, and electronically transmitted to the pharmacy, eliminating another opportunity for a mistake reading the prescription at the drugstore.

Doctors also need new tools to help them tell patients all about a newly prescribed drug during the increasingly brief office visit, and to help them tailor therapy to each individual. If you go to an electronics store to buy, say, a new battery, your name will probably be linked to a computer record with your previous purchases at any other store in the chain. When a doctor prescribes a new drug, why isn't this immediately linked to your previous drug prescriptions while a computer alerts the doctor to subtle dangers, not just the obvious life-threatening drug interactions?

We already have good doctors. Now we need to engineer a system that helps them do a good job prescribing drugs, rather than making that nearly impossible.