Doctors who treat non-English-speaking patients have long been frustrated with the poor accuracy of computer programs pharmacies use to translate instructions for patients, and a new study in Pediatrics supports a shocking reality. For the study, researchers reviewed 76 medicine labels generated by 13 popular English-to-Spanish programs, and found an overall error rate of 50 percent.
Unlike translations by a careful medical interpreter, the computer-generated instructions often featured an unhelpful form of "Spanglish." For example, the English "once" wouldn't be distinguished from the Spanish "once"--the difference between telling a patient to take a drug one vs. 11 times a day. Or "boca"--"by mouth"--would be misspelled to read "poca"--"a little." And terms such as "dropperfuls," "take with food," "apply topically," "for 7 days" and "apply to affected areas" were not translated into Spanish at all, the researchers found.
"It's scary how high the error rate is," said lead author Dr. Iman Sharif, chief of the division of general pediatrics at the Nemours/Alfred I. duPont Hospital for Children in Wilmington, Del. "If we can't do this right in Spanish--the most commonly spoken non-English language in the U.S.--I'm afraid to think what happens with the other languages."
One of the biggest problems, Sharif says, is that English prescription instructions aren't standardized, making it difficult for databases to correctly interpret the multiple ways the same directions may be written. For that reason, she explains, physicians and pharmacists must make a point to go over prescription instructions with all patients, regardless of what language they speak.
"It doesn't only apply to someone who has a language barrier--it applies to everybody," agrees Akil Ghoghawala, the pharmacist manager of Bienestar Pharmacy in Chicago. Ghoghawala speaks five languages, including Spanish, and says he makes it a point to verbally explain to patients how to take their medications. "I override [the program], each and every time," he says.