A new prediction model developed to identify drugs with similar names and appearances may reduce medication errors<\/p>\n\n\n\n
Potato, patato. Slather on rich\nsour cream speckled with fresh chopped chives and who cares what you call the\nstarchy plant tuber! Just call it delish<\/em>.\n<\/p>\n\n\n\n It\u2019s a different story with drugs.\nA medication\u2019s name \u2013 as well as its look and packaging \u2013 matters a great deal.\nAccording to the Harvard Pilgrim Health Care Institute (HPHCI) team, a Harvard Medical School affiliate, sound-alike look-alike\n(SALA) medication confusion effects as many as 250,000 Americans annually. To mitigate\nthat stat, the HPHCI team developed a new prediction model that supports\nprescribers by flagging any prescribed medication that shares a similar name\nand\/or appearance with another drug. The model, still in its infancy, debuted\nas a poster during the June 2019 American Society of Health-System Pharmacists Summer Meetings\nand Exhibition in Boston. <\/p>\n\n\n\n How serious is SALA? Maybe you recall the 2016 case\nwhen 55 reports noted confusion between Brintellix<\/em>, an\nantidepressant, and Brilinta<\/em>, a blood thinner. Two reports cited serious\nadverse events. Soon thereafter, the U.S. Food and Drug Administration (FDA)\napproved a name change for Brintellix<\/em>\nto Trintellix<\/em>. <\/p>\n\n\n\n \u201cIt may be surprising that something as simple as a name can lead to serious medical prescription errors,\u201d says Qoua Her, PharmD, MSPharm, research analyst at HPHCI and lead author of the prediction model poster. \u201cBut even well-trained healthcare professionals can mistake one medication for another with an almost identical name. Our new prediction model has the potential to reduce critical errors by identifying high-risk sound-alike-look-alike medication pairs.\u201d <\/p>\n\n\n\n Aiming for a model that fosters preventative healthcare \u2013 rather than reactive action \u2013 the HPHCI team evaluated 82 medication name similarities and seven product attribute measures with 40,000 samples of medication pairs. Next the team created a prediction model comprised of 13 predictors for potential SALA medication errors, with predictors including similar medication names and product attributes, such as route, class, and strength. The team tested the model with 20,000 samples of medication pairs and found the model to have a high accuracy in identifying possible mistakes . The final prediction model was able to identify almost all SALA medication pairs, with mean c-statistic and positive predictive value at 0.987 and 0.939, respectively <\/p>\n\n\n\n The\nresearch team\u2019s next step is to publish their study\u2019s results. Then,\nlooking off into the future, Dr. Her adds, \u201cThis tool may someday be used to\nevaluate new medication names adjunct to what is already in place.\u201d <\/p>\n\n\n\n That, as they say, is no small potatoes.\n<\/p>\n\n\n\n