Prediction Model Helps Flag Sound-Alike, Look-Alike Drugs

Marcia Frellick

June 14, 2019

A new prediction model could alert prescribers that a drug looks like, sounds like, or is packaged like another, new research indicates.

The model, introduced in a poster at the American Society of Health-System Pharmacists Summer Meetings and Exhibition in Boston, Massachusetts, this week, was developed by Harvard researchers to help reduce medication errors.

According to the researchers, rates of sound-alike, look-alike (SALA) errors — which they say contribute to 250,000 hospitalizations a year — may be reduced if medications for which there is a high risk for SALA confusion are identified before marketing.

Qoua Her, PharmD, a research analyst at the Harvard Pilgrim Health Care Institute, a Harvard Medical School affiliate that funded the study, led a team that created a prediction model to identify SALA medication pairs.

He told Medscape Medical News that the team used medication pairs from the Institute for Safe Medication Practice's list of confused drug names as SALA medications pairs. For a control group, the researchers randomly selected non-SALA medication pairs from a medication database (First DataBank's MedKnowledge).

The model comprises 13 predictors, such as two drug names having the same first letter, or two drugs from the same manufacturer or that have the same package unit size. The model was tested in 20,000 samples of medication pairs and was found to be highly accurate in identifying potential SALA medications. The mean c-statistic and positive predictive values were 0.987 and 0.939, respectively.

Integration in Decision Support Tools

Within a month, the researchers will submit the study to a journal for publication. Her said the hope is that the model will be made available to download for free on a website and will eventually be integrated into decision support tools such that when a prescription is entered, the prescriber would be asked whether he or she meant to prescribe that drug rather than others for which it could be confused.

Current models take into account similarity of words, Her said.

"In our model, we use a number of different predictors. That's why we have a higher predictive accuracy," he said.

Her's team analyzed factors such as manufacturer logos, colors of the medication bottle, strength of the medication, and routes of administration.

The confusion can come anywhere during the delivery process, he noted, from the prescriber, to the pharmacist, to the nurse who administers it, to the patient.

Other strategies have been used to stop SALA errors.

The researchers note that in 2016, after 55 reports of incidents in which the antidepressant Brintellix (Takeda) was confused with the blood thinner Brilinta (AstraZeneca), including two incidents that resulted in serious adverse events, the US Food and Drug Administration approved a change of the brand name from "Brintellix" to "Trintellix."

The authors have not yet validated the model externally.

Her has disclosed no relevant financial relationships.

Follow Medscape on Facebook, Twitter, Instagram, and YouTube


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.