Because virtually every drug has side effects -- some merely annoying, some deadly -- an off-label prescription can mean that a patient is assuming the risk of harm with no assurance of benefit.

The authors of the study, which was released Monday in the prestigious Archives of Internal Medicine, said that the widespread use of off-label therapies needs to be examined further to see if the practice "compromises patient safety or represents wasteful medication use."

In all, the study estimated that 1 in 7 prescriptions were written without good medical evidence that they would work. The estimate was derived from prescriptions written by doctors in office-based medical practices.

Clear Risks, Uncertain Benefits

Senior author Randall Stafford said the study has implications for all drug consumers.

"We want to avoid the situation where a drug clearly has some risks associated with it but we are uncertain about its benefits," said Stafford, an associate professor of medicine at the Stanford Prevention Research Center.

Asked about the study's conclusions, Dr. Edward Langston, a member of the board of trustees of the American Medical Association, said that the nation's primary physician organization "certainly supports off-label use, but based on sound scientific data. Otherwise, we have to be very cautious."

The Food and Drug Administration approves drugs after extensive studies have found them safe and effective in treating specific ailments in specific populations. Once on the market, however, doctors can prescribe drugs as they see fit, even for unapproved, or off-label, uses. Physicians say they need that flexibility to practice the art of medicine.

Little Proof of Effectiveness

While some off-label uses are beneficial and based on good science, others have little proof of effectiveness behind them. The issue becomes one of risk: Because virtually every drug has side effects -- some merely annoying, some deadly -- an off-label prescription can mean that a patient is assuming the risk of harm with no assurance of benefit.

Stafford said that the risks of some off-label prescriptions could be higher than they are for on-label uses. "Just because the risk has been studied in one population doesn't mean that same level of risk will occur in other populations," he said. "It could be lower, it could be higher."

The study's conclusions mirror the findings of a Knight Ridder series, "Risky Rx," that ran in newspapers nationwide in November 2003. That series assessed the frequency of off-label drug use and the harm that patients sometimes suffer by taking drugs off-label.

The most recent study examined one of the nation's main drug-prescription databases for 160 drugs, determining whether each specific drug prescription met FDA or scientific standards. It was funded by a federal grant and was conducted by researchers from Stanford, Dartmouth Medical School and the Massachusetts Institute of Technology.

It found that 79 percent of the prescriptions were for FDA-approved uses, 6 percent were for off-label uses with "strong scientific support," and 15 percent were for off-label uses with "little or no scientific support."

Anecdotal Studies

Doctors frequently say that their off-label prescriptions are based on good science and that FDA approval hasn't caught up to current medical practices. But the study calls that belief into question by finding that nearly three-quarters of all off-label uses have little scientific support.

Doctors are often influenced by small anecdotal studies that don't have the rigor of large, well-designed clinical trials, Stafford said. Doctors also prescribe drugs after hearing that their colleagues are doing so or after being approached by drug company sales representatives. Patients sometimes request prescriptions, and other times doctors give patients drugs -- antibiotics in particular, even if not warranted -- to shorten the lengths of visits, Stafford said.

At times it's certainly legitimate to use drugs off-label if other therapies have failed, Stafford said. "But my real concern is when those off-label uses become so widespread that they are used for typical patients and become the first line of therapy."

The Knight Ridder series and the current Stanford study used different data sources and studied different years -- the Knight Ridder data were from 2003, the Stanford study from 2001 -- but many conclusions were similar.

In both cases, 21 percent of all prescriptions were off-label; further, in both studies, anticonvulsant drugs were widely used off-label, while diabetes drugs and cholesterol-lowering drugs were rarely used off-label. Psychiatric medications also had widespread off-label use.