A computer program to determine how likely the proposed name for a new drug is to be confused with that of one already on the market is now being developed in the University of Illinois at Chicago College of Pharmacy.
With a $1.7 million grant from the Agency for Healthcare Research and Quality, Bruce Lambert, associate professor of pharmacy administration, is embarking on a four-year project to develop an automated search and retrieval process for drug names to minimize the incidence of name-confusion errors.
Confusing soundalike words is something people do a lot, Lambert says. We hear a word pronounced correctly but perceive it as a different, more familiar, word with a similar sound. We hear what we expect to hear.
Many of the medication errors that contribute to thousands of deaths each year in the United States are attributable to drug names that look or sound alike. Lambert, who has completed a study on drug names that look alike, is now turning his attention to the soundalikes.
"By predicting the likelihood that one drug name will be confused with another, we can help the FDA and drug companies avoid adding new drug names to the market that will more than likely become confused with existing medications," Lambert said.
The software could also be used to screen new drugs before adding them to a pharmacy's formulary, giving front-line practitioners a heads-up that also will help reduce errors.
The software Lambert is developing to lessen the chance of word confusion will screen proposed drug names against databases of existing names. When a new name is entered, the software will return a list of existing names ranked in descending order of confusability. The result will be a confusability score.
How will a computer program predict confusability? Lambert says the score will be based on the target word's intelligibility and on words in the target word's "perceptual neighborhood."
The perceptual neighborhood includes all drugs relatively close to the target name in pronunciation, as well as how often they are prescribed. If an existing drug that sounds like the proposed name for a new drug is used infrequently, the confusability score will be lower.
Lambert will design the software using measurements developed in auditory perception studies he is conducting with 50 each of pharmacists, physicians, nurses, and lay people, who, wearing headphones, will hear a drug name and speak what they hear into a microphone. Background noises and changing bandwidths will simulate the way words are normally heard, such as on a telephone.
"If the neighborhood is dense, with a lot of high-frequency names, it will be very hard to correctly perceive the name," Lambert said. "But if it's a low-frequency neighborhood, it will be relatively easy to perceive the correct drug name.
"This will give a tool to people picking new drug names," he explained. "It will allow them to screen a name and pick a different one if it's going to be too confusing."
Lambert is conducting the project in collaboration with researchers from UIC's departments of computer science and psychiatry and with the University of Buffalo's psychology department.
For more information about UIC, visit http://www.uic.edu .