Mar 08, 2002
Patients may greatly overestimate the risk of drug side effects when given only vague information about risks through terms such as "common" and "rare," UK researchers report.
Based on their study findings, the investigators are questioning the European Union guidelines on how drug information should be translated to patients.
Professor Dianne Berry from the University of Reading, UK, and colleagues at Leeds University explain that the EU guidelines, which came into force in January 1999, recommend that drug information leaflets categorise the risks of side effects into five bands. These range from "very rare," affecting fewer than 0.01% of the population, to "very common," affecting more than 10%.
The research team carried out four studies, in which more than 750 adults recruited from public places were asked to estimate the probability of experiencing side effects when given qualitative or quantitative information on prescribed drugs. Qualitative information refers to terms like "rare" and "common," while quantitative information describes the risks in actual numbers.
In the largest of these studies, 360 adults were given qualitative descriptions of "common" and "rare" for the occurrence of side effects, which corresponded to actual risks of 2% and 0.02%, respectively.
Participants vastly overestimated the risk of side effects when given qualitative information, believing that the personal risk associated with the word "common" was 50%, while associating the word "rare" with a 21% risk, according to the report in the March 9th issue of The Lancet.
"To make informed decisions about taking medicines, people need accurate information on side effects," said study co-author Professor Theo Raynor. "While attempting to be helpful, the EU really put the cart before the horse because they issued the guidelines before there was any evidence that the words associated with risk would be meaningful.
"Not only did participants given qualitative information believe that the probability of side effects to be significantly higher than the true risk, but our study suggested that they were nearly two times less likely to comply with treatment."
When patients were given quantitative information on risk they still perceived the level of risk to be greater than the actual risk--although they were much closer this time. Those given a numerical value of 2% perceived their own actual risk to be 9.5%, on average. When given a value of 0.02%, they estimated their risk of side effects to be 7%.
"One possible reason is that people have a healthy scepticism on the health information they are provided with. They may feel that the numbers given are an underestimate," Raynor suggested.
The researchers suggest that terms like "rare" or "common" should not be used until research shows how people interpret them.
"We need to come up with something that does work, and my view is that we still need to use verbal descriptions," Raynor said. "For example, if the real risk of side effects is 10%, we need to find the word that is perceived by patients to match that figure most closely."
He noted that an average drug has a list of 20 to 30 potential side effects, so depicting the risk levels quantitatively for each drug would be confusing to patients.
Raynor and his colleagues are currently studying other aspects of drug labelling. "If someone feels they have a side effect listed on the medication, they need to know what they should do," Raynor pointed out. "Currently, the guidelines suggest that action should be taken 'immediately' or 'as soon as possible.' Our preliminary data suggest that patients do not distinguish between these options, so the distinction is effectively meaningless."
Raynor's team is currently working in partnership with academics in other EU states to see if their findings are applicable across the region.
But Noel Wathion from the European Medicines Evaluation Authority questioned the rationale behind the current study.
Although he said he had not had the opportunity to look at the report in depth, he said that the EU guidelines actually favour the use of numerical rather than qualitative information.
"We plan to review the outcome of this study in more detail, and in the context of the EU guidelines," he said in an interview.
SOURCE: The Lancet 2002;359:853-854.
Copyright © 2002 Reuters Limited