PRACTICE RESEARCH REPORTS 1460 AM J HEALTH-SYST PHARM | VOLUME 75 | NUMBER 19 | OCTOBER 1, 2018 Detection of potential look-alike/sound-alike medication errors using Veterans Affairs administrative databases Jessica M. Zacher, Pharm.D., BCPS, VA Center for Medication Safety/Pharmacy Benefts Management Services, Hines, IL. Francesca E. Cunningham, Pharm.D., VA Center for Medication Safety/Pharmacy Benefts Management Services, Hines, IL. Xinhua Zhao, Ph.D., VA Center for Health Equity Research and Promotion, VA Pittsburgh Healthcare System, Pittsburgh, PA. Muriel L. Burk, Pharm.D., VA Center for Medication Safety/Pharmacy Benefts Management Services, Hines, IL. Von R. Moore, Pharm.D., VA Center for Medication Safety/Pharmacy Benefts Management Services, Hines, IL. Chester B. Good, M.D., M.P.H., VA Center for Medication Safety/Pharmacy Benefts Management Services, Hines, IL, and VA Center for Health Equity Research and Promotion, VA Pittsburgh Healthcare System, Pittsburgh, PA. Peter A. Glassman, M.B.B.S., M.Sc., VA Center for Medication Safety/Pharmacy Benefts Management Services, Hines, IL, and VA Greater Los Angeles Healthcare System, Los Angeles, CA. Sherrie L. Aspinall, Pharm.D., M.Sc., BCPS, VA Center for Medication Safety/Pharmacy Benefts Management Services, Hines, IL, and VA Center for Health Equity Research and Promotion, VA Pittsburgh Healthcare System, Pittsburgh, PA. Address correspondence to Dr. Aspinall (sherrie.aspinall@va.gov). DOI 10.2146/ajhp170703 Purpose. Results of a study to estimate the prevalence of look-alike/ sound-alike (LASA) medication errors through analysis of Veterans Affairs (VA) administrative data are reported. Methods. Veterans with at least 2 flled prescriptions for 1 medication in 20 LASA drug pairs during the period April 2014–March 2015 and no his- tory of use of both medications in the preceding 6 months were identifed. First occurrences of potential LASA errors were identifed by analyzing dispensing patterns and documented diagnoses. For 7 LASA drug pairs, potential errors were evaluated via chart review to determine if an actual error occurred. Results. Among LASA drug pairs with overlapping indications, the pairs associated with the highest potential-error rates, by percentage of treat- ed patients, were tamsulosin and terazosin (3.05%), glipizide and gly- buride (2.91%), extended- and sustained-release formulations of bupro- pion (1.53%), and metoprolol tartrate and metoprolol succinate (1.48%). Among pairs with distinct indications, the pairs associated with the highest potential-error rates were tramadol and trazodone (2.20%) and bupropion and buspirone (1.31%). For LASA drug pairs found to be associated with actual errors, the estimated error rates were as follows: lamivudine and lamotrigine, 0.003% (95% confdence interval [CI], 0–0.01%); carbamaz- epine and oxcarbazepine, 0.03% (95% CI, 0–0.09%); and morphine and hydromorphone, 0.02% (95% CI, 0–0.05%). Conclusion. Through the use of administrative databases, potential LASA errors that could be reviewed for an actual error via chart review were iden- tifed. While a high rate of potential LASA errors was detected, the number of actual errors identifed was low. Keywords: database management systems, medical order-entry sys- tems, medication errors, Veterans Administration Am J Health-Syst Pharm. 2018; 75:1460-6 L ook-alike/sound-alike (LASA) med- ications are medications whose names have similar spellings and/or pronunciations. A LASA error occurs when a LASA medication is incor- rectly prescribed, dispensed, or ad- ministered instead of the intended medication. 1 Currently, little is known about the prevalence or impact of LASA errors, but it has been estimated that 7–15% of all medication errors are due to LASA confusion. 2 However, this is likely an underestimation, because LASA medication errors are often dis- covered only if they are reported. There are very few published stud- ies of potential or actual LASA er- rors. Phatak and colleagues 2 quanti- fied potential errors in dispensing of medications with similar names by reviewing adjacent claims for speci- fied pairs of drugs (not necessarily known LASA pairs) in a state Medi- caid database. They found that 106 of the 300 evaluated drug pairs were as- sociated with a total of 1,138 potential