Purpose: Accurate medical interpretation services are essential for providing health care to the approximately 12 million U.S. children whose families primarily speak a non-English language at home. However, the additional time medical interpretation adds to appointments, as well as the variation in interpretation accuracy has yet to be thoroughly explored. This study has two primary aims: (1) to evaluate the time interpretation adds to patient care, and (2) to quantify translation errors made during medical interpretation within a controlled environment. Methods: Researchers scripted and simulated an appointment between a pediatrician and a parent of an infant with hyperbilirubinemia. A native Spanish-speaking member of the research team (researcher 1) played the role of the parent and another researcher (researcher 2) played the role of the pediatrician. A total of 57 timed simulations were conducted. Figure 1 illustrates the distribution of control and experimental trials. Video remote interpretation (VRI) was provided by professional interpreters who had undergone 120 hours of medical interpretation training. Each experimental trial was audio-recorded and conducted with a different interpreter. Deviations from the script in the experimental trials with VRI were measured by quantifying: (1) linguistic additions, (2) omissions, (3) substitutions, (4) requests by interpreters to look up medical terminology, (5) clarification requests from the interpreter, and (6) requests from the research team to repeat secondary to poor audio quality. Deviations were categorized as English-to-Spanish or Spanish-to-English based on the language of the speech being interpreted. Results: The English-English control trials ranged from 211 to 226 seconds and were on average 257.2% longer than the experimental trials (220 s vs. 788 s, p < .0001), which ranged from 704 to 1237 seconds. The English-Spanish control trials ranged from 230 to 246 seconds and were on average 231.1%% longer than the experimental trials (220 s vs. 788 s, p < .0001). Out of the 19 experimental trials with VRI, 16 had adequate audio quality for deviation analysis. Interpreters made linguistic errors (additions, omissions, substitutions) for English-to-Spanish interpretation at a significantly higher rate than for Spanish-to-English interpretation (2.71 errors per trial vs. 1.28 errors per trial, p < .001). Table 1 summarizes average frequency of interpretation errors and deviations from the script. Conclusion: The present study highlights challenges to providing adequate health care to limited-English proficient (LEP) patients. Our simulated patient interviews with professional video remote medical interpretation required significantly extended appointment times and showed variation in interpretation accuracy based on deviations from a script. Pediatricians must be aware of the barriers to health care that LEP patients face, consider the variation in interpretation accuracy, and account for the additional time required to provide adequate care to LEP patients.
Distribution of simulations into control trials and audio-recorded experimental trials with video remote interpretation (VRI).
Fifty-seven simulations using the same script were grouped into three sets of 19 trials: two control sets without interpretation in which the parent’s lines were read in English (English-English) and Spanish (English-Spanish) respectively and one audio-recorded experimental set in which the parent’s lines were read in Spanish with a different trained, professional video remote interpreter for each trial. The physician’s lines were read in English in all 57 trials.
Distribution of simulations into control trials and audio-recorded experimental trials with video remote interpretation (VRI).
Fifty-seven simulations using the same script were grouped into three sets of 19 trials: two control sets without interpretation in which the parent’s lines were read in English (English-English) and Spanish (English-Spanish) respectively and one audio-recorded experimental set in which the parent’s lines were read in Spanish with a different trained, professional video remote interpreter for each trial. The physician’s lines were read in English in all 57 trials.
Comments