One in six medication errors involves name confusion (e.g., Vioxx/Videx). Drug companies and regulators screen names prior to approval, but the screening process is itself error prone due to an overreliance on subjective assessments of similarity. Our long-term objective is to minimize the incidence of name confusion errors. Our short-term goal is to develop an empirically validated, user-friendly software tool that can be used to screen proposed drug names against databases of existing drug names. Given a name as input, the software will return a list of existing names ranked in descending order of confusability. Confusability ratings will be based on validated, objective criteria derived from studies of clinicians' and lay persons' auditory perceptual errors. Auditory perception experiments will be based on Luce's Neighborhood Activation Model (NAM). The NAM predicts that errors in auditory perception depend on the intelligibility of the target word as well as the similarity and frequency of words in the target word's perceptual """"""""neighborhood."""""""" This prediction is embodied in Luce's Frequency-Weighted neighborhood Probability Rule (FWNPR). Using NAM as the theoretical framework, we will test three hypotheses: 1) The number of errors in auditory perceptual identification will increase as frequency-weighted neighborhood probability decreases. 2) The effects of frequency-weighted neighborhood probability on auditory perceptual identification will be the same among adult lay people, physicians, nurses, and pharmacists. 3) A model can be developed that accurately predicts a drug name's probability of confusion in an auditory perceptual identification task. To test these hypotheses, we propose studies with the three specific aims: 1) to generate confusion data from pharmacists, physicians, nurses, and adult lay people using a noisy auditory perceptual identification task; 2) to use the confusions-in conjunction with computational tools and the theoretical model-to develop and refine a model for predicting confusions; 3) to incorporate the best predictive models into a user-friendly software tool that can be used to support decision-making during the drug name approval process.
| Lambert, Bruce L; Dickey, Laura Walsh; Fisher, William M et al. (2010) Listen carefully: the risk of error in spoken medication orders. Soc Sci Med 70:1599-608 |
