According to a new study by researchers from the University of California Los Angeles Medical Center and the Memorial Sloan Kettering Cancer Center in New York, Google Translate still isn’t reliable enough to use for medical instructions for people who don’t speak English.
The new study evaluated 400 emergency department discharge instructions translated by Google Translate into seven languages: Spanish, Chinese, Vietnamese, Tagalog, Korean, Armenian, and Farsi. Native speakers read the translations and evaluated their accuracy. Overall, the translated instructions were over 80% accurate. That’s an improvement from 2014, when an analysis found that Google Translate was less than 60% accurate for medical information.
But the study also found that accuracy varied between languages. For example, Google Translate was over 90% accurate for Spanish. Tagalog, Korean, and Chinese had accuracy rates ranging from 80-90%. But there was a big drop-off for Farsi (67% accuracy) and Armenian (55% accuracy). Even languages like Spanish and Chinese that were usually accurate could have Google Translate errors with the potential to confuse patients.
“All you need is one error that creates confusion for a patient, and they don’t take their blood thinner, or they take too much of their blood thinner,” said study author Lisa Diamond, a health disparities researcher at the Memorial Sloan Kettering Cancer Center. “And you end up with a medical emergency.”
Federal guidelines say that hospitals and health care organizations have to provide interpreters and translators for patients who don’t speak English. However, in practice, many hospitals don’t offer interpreters to every patient who needs one, and they’re less likely to have a way to translate written instructions.
“There’s a clear gap in the ability to provide written information for patients,” said study author Breena Taira, an associate professor of clinical emergency medicine at UCLA Health.
“It’s become common for doctors to resort to Google Translate in medical settings,” Taira said. “You can imagine a well-meaning emergency department provider thinking, ‘I really want to provide my patient with instructions in their own language, and my hospital doesn’t have a mechanism to do this—why don’t I use this automated translation software?’”
“One of the main problems with relying on machine translation is that it can’t account for context,” Diamond said. “The program might not recognize that a word is the name of a medication, for example, so it loses the meaning of what you’re trying to say.”
“Instead, doctors should write out instructions in English and have an interpreter go over those instructions verbally with a patient,” Taira said. “But that’s just a stopgap, since health systems should give doctors a way to get professional translations of materials. Each doctor is going to do the best they can with the resources they have available.”
News summaries © copyright 2020 SmithBucklin
