 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
To the Editor.—
We read with concern Flores et al’s findings on errors in medical interpreting in the January 2003 edition of Pediatrics.1 However, upon closer inspection, we find several flaws in the study’s methods and analyses, which may unnecessarily alarm limited English proficiency (LEP) patients and the clinicians who work with them.
Errors are an acknowledged part of medical practice.2 There exists a well-developed literature of the analysis of error in clinical practice in general.2–6 Among the error classification system used by the authors, we hold that omission, addition, substitution, editorialization, and false fluency are highly debatable categories, they lack reproducibility, and neither correlate necessarily with quality of interpretation nor the quality of care provided. There are other striking flaws in the study that lead to misleading conclusions:
As professional medical interpreters, we welcome all efforts to improve communication between patients and clinicians. However, Flores et al’s critique misses its mark, both in its methodology and its analysis. Additional research should study other factors that put an interpreter at risk of making errors of potential clinical consequence. Such factors might include changes in health care legislation and budgets allocated to interpreter services, training and familiarity of providers for working with interpreters, and years of experience and training of the interpreter. Otherwise, studies such as Flores et al, which pinpoint "interpreters" as the sole source of errors, generate a lack of trust in the quality of work that interpreters do, and may lead to an unnecessary public health alarm.8
REFERENCES
In Reply.—
We have the utmost respect for the important and challenging work of medical interpreters. We also appreciate the desire of Abadía-Barrero et al to not "unnecessarily alarm" limited English proficient (LEP) patients and their clinicians. We strongly disagree, however, with Abadía-Barrero et al’s comments and address them in order.
Abadía-Barrero et al assert that the error classification system that we developed is "highly debatable," lacks reproducibility, and neither correlates with "quality of interpretation nor the quality of care." The basic error category framework we used was derived 31 years ago.1 It was successfully used in a previous rigorous study of interpreter errors published in a major peer-reviewed medical journal.2 In a systematic review of the published literature on medical interpretation since 1966, we did not encounter any other framework as rigorous in the 2640 citations reviewed. To this framework, we added the false fluency category, a problem that most interpreters readily recognize and acknowledge. Published reports of single false fluency errors that resulted in inappropriate social services placement of children3 and a $71 million lawsuit4 indicate that there is little debate about this error category’s validity and relevance to quality of interpretation and the quality of care.
As detailed in the "Methods" section of our article, we subjected our error classification system to a rigorous, multistep validation process that included: 1) analysis of a test set of transcripts by 3 independent reviewers (2 of whom were blinded to the study aims); 2) calculations of agreement matrices and kappa indices; 3) further refinement of the system, based on preliminary piloting; and 4) a consensus process when there was observer disagreement. The final mean agreement of 99% and the mean kappa of 0.99 would suggest that this error classification system is both valid and reproducible.
Abadía-Barrero et al believe that our study "lacks proper controls" and maintain that we should have had 2 additional comparison groups, English-proficient (EP) patients, and LEP patients who had no interpreter. As stated in the beginning of our article, one of the principal study goals was to compare the quality of interpretation by professional hospital versus ad hoc interpreters. Thus, the only 2 appropriate groups to study are professional hospital interpreters and ad hoc interpreters. Indeed, it is not apparent to us how one could perform an interpreter error analysis of clinical encounters with EP patients, because EP patients, by definition, do not need or use interpreters.
Abadía-Barrero et al argue that interpreter errors are presented "out of context," in that the mean of 31 errors per encounter represents "only 0.8% of potential word errors," which they view as "a far cry from the ‘alarmingly common’ frequency claimed by the authors." We are troubled by these views for the following reasons: 1) few would tolerate a surgeon who averaged 31 errors per operation or an airline pilot who committed a mean of 19 errors of potential air safety consequence per flight, so it follows that a similar average number of errors of interpretation and interpreter errors of potential clinical consequence per clinical encounter should be considered alarming; 2) high quality and safety in health care demand placing the needs and welfare of patients above all other considerations. Trivializing an average of 31 interpreter errors and 19 interpreter errors of clinical consequence per encounter by "pseudo-contextualizing" the errors with an inappropriate denominator of "potential word errors" is not only a harmful disservice to our patients and their families but also misses the ultimate goal of patient safety, which is to have a health system in which patients are subjected to no errors that compromise the quality of their care or their safety; and 3) the literature documents that misinterpretation of a single word can have devastating consequences for patients and their families, including inappropriate child abuse charges, mistaken social services custody of children, quadriplegia, and multimillion-dollar lawsuits.3,4 Such cases vividly demonstrate that the ultimate goal should be error-free care.
Abadía-Barrero et al express concerns about inclusion of errors committed by clinicians in our analysis. As was stated in the "Methods" section of our article, we took the patient-centered approach of counting errors made by clinicians as errors of interpretation because 1) these errors are errors of interpretation; 2) these errors were found to affect patient care; and 3) these errors often went uncorrected. Furthermore, errors of interpretation by clinicians comprised only 10% of all errors, and exclusion of these errors in statistical analyses did not alter the study findings. To our knowledge, this is the first study to report that clinicians are responsible for most false fluency errors, a valuable finding because it points out the need to train clinicians on how to work with professional medical interpreters, as well as the dangers of clinicians who try to use limited foreign language skills.
Abadía-Barrero et al claim that "stated policies in hospitals across Massachusetts prohibit the use of untrained ‘interpreters’ in medical settings." Although we wish that this were true, we are not aware of any such statewide policy that universally pertains to all "medical settings."
Abadía-Barrero et al make the assertion that interpreter services in the state have "undergone notable change" since Massachusetts legislation requiring interpreter services in emergency departments was enacted in 2001 (the legislation was actually passed in 20005). They claim that this legislation "has raised the quality and availability of interpreters in Massachusetts hospitals." Abadía-Barrero et al, however, provide no data or studies to document any changes in the quality or availability of interpreters in Massachusetts since the passage of this act. We welcomed this important legislation but observed that, as with similar federal initiatives,6,7 institutional compliance appears to be low, probably because of concerns about excessive cost8 and because of lack of enforcement.
We sincerely hope that state legislation raises the quality of care for LEP families in Massachusetts. But a study in 2001–2002 that we recently completed of the 2 largest pediatric emergency departments in Massachusetts suggests that there is still much room for improvement.9 We found that many LEP children and their families either have no interpreter whatsoever or use ad hoc interpreters, most of whom are family members.
The principal conclusions of our article were: 1) errors committed by ad hoc interpreters are significantly more likely to have clinical consequences than those committed by professional hospital interpreters; 2) interpreter errors may be a root cause of medical errors; 3) more research and policy work is needed on what type of interpreter training is most effective in reducing interpreter errors; 4) clinicians should receive training on the proper technique for working with interpreters; and 5) third-party reimbursement for trained interpreters should be considered for all LEP patients. The overwhelmingly positive response to our article from interpreters, health care providers, patients and families, the media, and policymakers across the United States underscores the value of studies that call attention to the importance of providing high-quality care to the 19 million LEP Americans.
REFERENCES
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
||
|
||
Home | My Pediatrics | Journal Information | Current Issue | Past Issues | Subscriptions & Services | Contact Us Click here for faster international access © 2003 American Academy of Pediatrics. All rights reserved. |
||
|
||
