Lost in translation? Helping LEP patients

Interpreter errors are common

A Spanish-speaking mother brings her infant daughter to a hospital emergency department because of a skin rash on her baby’s face. Because a professional interpreter is not available, the woman’s older child first translates the mother’s information about the patient. Then, the child translates doctor’s instructions on treatment to the mother. The recommended treatment, however, will likely never be delivered because the young girl, asked to quickly interpret complicated medical information, instructs her mother to spread the prescribed hydrocortisone cream across the baby’s entire body.

Such misinterpretation may seem comical in this case, but the actual consequences of mistaken translations in health care can be deadly.

Although federal regulations require health care providers to provide assistance to patients with limited English proficiency (LEP) — including translation and interpretation services when necessary — there are no objective standards or guidelines for who may work as an interpreter for health encounters.1

As a result, most hospitals and primary care providers have sketchy programs for communicating with non-English-speaking patients.

"What often happens is they ask a bilingual employee, who may not be fully fluent in both languages, to step back and serve as interpreter," says Beverly Treumann, a state-certified Spanish interpreter at the University of California at Los Angeles Medical Center and president of the California Healthcare Interpreting Association (CHIA). "For a lot of people, Spanish-speakers for example, Spanish may be their first language and the language they hear at home. But that is not the language they learned in school, if they grew up in this country, and they don’t have formal knowledge of the language, both written and verbal, [as well as] its grammar and structure."

Translator can compromise privacy

Asking bilingual employees to serve as interpreters during health care encounters raises important questions about patient privacy and confidentiality as well. But Treumann and others say they are primarily concerned because inaccuracies in translation can lead to dangerous medical mistakes.

Interpreters not only have to be conversational in both languages, but they also must be able to understand complex medical terms and a great deal of technical vocabulary in both tongues. In addition, nontrained interpreters are unlikely to use the basic professional practices of trained interpreters to ensure accuracy, Treumann adds. Medical interpreters, for example, know to restate what each party said, verbatim, if possible, and minimize any shadings of the meaning. "This may sound like common sense, but you’d be surprised how many people, when asked to interpret for someone else, will not understand that they are to repeat, in exact detail as possible, what each party says," she notes.

Professional interpreters often are trained to pause after translating to allow each person to respond, and also are trained to ask for more time to "catch up" in the conversation without disturbing the interaction between patient and physician or inserting themselves into the visit.

"You are always going to get behind . . ., no matter how good you are," Treumann says. "You have to be able to acknowledge that and ask people to slow down, or repeat what they said to ensure you are accurate."

Even when professional interpreters are used, the results are not always perfect, says Glenn Flores, MD, associate professor of pediatrics, epidemiology, and health policy at the Medical College of Wisconsin in Milwaukee and director of the school’s Center for the Advancement of Urban Children.

Study shows many translator mistakes

In 2003, Flores and colleagues conducted a study aimed at evaluating the accuracy of medical interpreters working during clinical visits.2 They audiotaped 13 clinical encounters in a pediatric clinic where Spanish interpreters were used. Investigators then used trained interpreters and documentation of the visits to evaluate the accuracy of the interpreters’ work. Of the 13 visits, there were a total of 396 interpreter errors, an average of 31 errors per encounter, ranging from as few as 10 errors in one visit, to as many as 60.

Most were errors of omission, Flores says, where the interpreters left out information stated by either party. In 16% of the visits, "false fluency" was a problem — interpreters simply invented words that did not actually exist in the language. In 8% of the visits, interpreters added incorrect information that the physician or patient did not state. Other problems emerged when the designated interpreters frequently inserted their own opinions.

Of the errors, 63% were deemed to have potential clinical consequences, Flores says. "These were errors in interpreting information about the history, past medical history, the history of present illness, or their understanding of treatment instructions and follow-up," he explains. "The average was 19 errors of potential clinical consequence per encounter, ranging from five to 49."

Flores and colleagues also found that "ad hoc interpreters," friends, fellow patients, social workers, clinic personnel, or relatives who spoke Spanish but were not trained interpreters were significantly more likely to make errors of potential clinical consequence than the trained hospital interpreters, he adds.

"Their rate was 77% vs. 53%, for the trained interpreters, which is highly statistically significant," he notes. "As an example, there was an 11-year-old girl used as an interpreter for her mother and an infant child, and 84% of the 15 errors she committed had potential clinical consequences, including some pretty impressive ones. "The most pressing ones we saw, were omitting questions about drug allergies, omitting key information about the past medical history, omitting crucial information about the chief complaint or symptoms, and errors about the antibiotic dose, frequency, and duration.

The interpreters working for the clinic often had other difficulties, Flores states. One interpreter, a social worker, actually turned to the patient and indicated that the doctor was going to ask detailed questions about sexual history and drug use, and advised the patient not to answer the questions. And another interpreter told the patient an antibiotic was prescribed for the flu, which is never done. "Then, there was a mom who clearly said to the interpreter that her child had already had a rectal swab for culture and the interpreter did not tell the doctor, so the child got another swab," he adds. Since that time, the hospital has improved training for its professional interpreters, and their rate of errors has dropped significantly, he says.

In many cases, it is unfair to ask bilingual employees to serve as interpreters, as often is the case in many health care settings, adds Treumann.

Hoping to avoid the cost of professional interpretation and translation services, some providers hope to kill two birds with one stone and hire personnel who are bilingual in the hopes they will serve as interpreters. Just because someone is bilingual, it does not mean they will be an accurate and reliable interpreter. And, the pressure from one’s employer to do something they have not been trained to do can be acute. "They may not feel comfortable serving as an interpreter for a person who speaks the same language because of concerns about how well they know the language, and [also may be uncomfortable] taking such a personal role," she says.

If bilingual health care personnel are adequately trained and want to serve in that capacity — that’s another story, she notes. "Many physicians from other countries, while awaiting licensure here, want to serve as interpreters. They can be ideal because they have usually had a high level of formal education in their native language, and they have the medical background as well," she says.

Enormous potential for disaster

Today, more than 47 million people in the United States speak another language and 21.4 million people have LEP. "There has been a tremendous growth over the last decade in those populations. It is not an inconsequential number of patients seeking care in U.S. hospitals," Flores explains. "And, the available data suggest that lack of available trained interpreters is not uncommon for millions of these patients. One study found, for example, that no interpreter was used for 46% of LEP patients and when an interpreter was used, 39% of the interpreters had no training."

Currently, there are no objective certification processes, licensing bodies, or training programs for health care interpreters, says Treumann. Most professional certification comes from learning to interpret and translate for the legal system. However, health care interpreting has its own set of unique challenges that warrants special training.

Court legal proceedings are adversarial in nature, and the interpreter is not required to be an advocate for the patient. However, in healthcare settings, the interpreter may have certain ethical and moral obligations to the person presenting for care that he or she would not have working in the legal system, Treumann says.

Standards of practice for interpreters

Health care interpreters not only need training in how to serve as accurate and reliable interpreters, but also must receive education about the ethical complexities of the patient encounter as well — information about privacy and the principles of informed consent are particularly compelling.

CHIA has developed standards of practice for health care interpreters, and they can be found at www.chia.ws/index.php on the organization’s web site.

The results of Flores’ study and others like it also provide ample justification for third-party reimbursement of translation and interpretation services, he says. The potential reduction in increased health care costs due to preventable medical errors and unnecessary care should more than offset the costs to the health care system, he notes.

Currently, only 10 states require third-party payer coverage of interpreter services.


1. 68 Fed Reg 47,311 (Dec. 16, 2003).

2. Flores G, Laws MB, Mayo SJ, et al. Errors in medical interpretation and their potential for clinical consequences in pediatric encounters. Pediatrics 2003; 111:1,495-1,497.


For more information on preventing interpreter errors, contact:

Glenn Flores, MD, Center for the Advancement of Urban Children, Medical College of Wisconsin, 8701 Watertown Plank Road, Milwaukee, WI 53226. Web: www.mcw.edu/peds/compeds/index.html

Beverly Treumann, The California Healthcare Interpreting Association, One Capitol Mall, Suite 320, Sacramento, CA 95814. Telephone: (916) 669-5305. E-mail: chia@chia.ws