Design, Implementation, and Evaluation of a Computerized System to Communicate with Patients with Limited Native Language Proficiency in the Perioperative Period : Anesthesia & Analgesia

Secondary Logo

Journal Logo

Patient Safety: Research Reports

Design, Implementation, and Evaluation of a Computerized System to Communicate with Patients with Limited Native Language Proficiency in the Perioperative Period

Taicher, Brad M. DO, MBA*; Alam, Rammy I. DO*; Berman, Joshua; Epstein, Richard H. MD

Author Information
doi: 10.1213/ANE.0b013e3182009cd1

Achieving effective communication with patients who have limited proficiency in the native language of the anesthesia care providers during the perioperative period is often challenging. This failure to communicate may increase patient anxiety when actions with unpleasant consequences occur without warning (e.g., discomfort from cuff inflation during the initial blood pressure determination, burning associated with injection of propofol), as well as frustrate the anesthesia care provider who is trying to provide instructions. More importantly, patient safety may be compromised as a result of the language barrier. For example, assessment of patient responsiveness to simple commands and return of adequate muscle strength before removal of the endotracheal tube may be adversely affected, detection of IV injection of local anesthetic via feedback about the onset of tinnitus during performance of a regional or neuraxial nerve block may be missed, and optimal positioning for placement of a neuraxial anesthetic may not be achieved.

Although most hospitals in the United States (US) offer language interpretation services (e.g., bilingual medical staff, trained in-house translators, telephone translation services) for the purpose of obtaining informed consent and providing preoperative and postoperative instructions, such assistance is generally not readily available or is impractical for use in the operating room (OR) or during the performance of nerve blocks. This lack of coverage is in part attributable to resource constraints, inability to provide on-demand coverage, lack of personnel trained in all languages, and access restrictions in the OR.

Because our hospital is located in the US, we formulated our approach based on English, although our process could be modified easily to work in countries with different primary languages. By way of perspective, Thomas Jefferson University Hospital is a large academic center located in downtown Philadelphia, a city in which 20% of the population aged 5 years and older speak a language other than English at home.1 Furthermore, 29% of patients in the immediate vicinity of the hospital are also in this category.2 The hospital is immediately adjacent to the city's Chinatown, accounting for the high percentage of our patients speaking a Chinese dialect (Table 1).

Table 1:
Prevalence of Patients with Limited English Proficiency in Philadelphia, PA

In this article, we describe how we developed, implemented, and validated a computerized system to convey many frequently used prerecorded phrases related to perioperative anesthesia care in the languages we most often encounter in our patients with limited English proficiency. Our system was inspired by Lee Marin, MD, a Navy physician who used pre-recorded Arabic phrases on a laptop to communicate with his patients during Operation Desert Storm. It was subsequently developed into a military grade handheld device (“Phraselator”) by VoxTec (Annapolis, MD) under a Defense Advanced Research Projects Agency grant in 2001.3 Similarly, our system allows anesthesia care providers to convey information to patients about what is being done or what the patients can expect, instructions about what we want them to do, and also allows the ability to engage in simple responsive dialogs with “yes” or “no” answers. Other domains in which communication between English and non–English-speaking subjects has been described include law enforcement (1-way communication using voice recognition of brief phrases as the trigger),4,5 and medical interviewing (using 2-way voice recognition and translation).6 In our system, the ability to translate ad hoc phrases is not supported and there is no facility for its use in obtaining informed consent.


The Patient Services department at Thomas Jefferson University Hospital provided us with a breakdown of the languages that most frequently required interpretation within our patient population (Fig. 1). Based on this information, we targeted Spanish, Chinese Mandarin, Chinese Cantonese, Russian, Vietnamese, and Korean, representing 85% of our limited English proficiency patient population, for language development.

Figure 1:
Frequency distribution of translation services. The percentages listed are the languages for which interpretative services are requested at Thomas Jefferson University Hospital. Mandarin was requested 18%, Cantonese 16%, and an unspecified Chinese dialect 6% of the time.

Phrases were selected for inclusion in the system through a consensus process among anesthesia department members. The initial list was generated by the study authors, and additional phrases were added based on an informal feedback process involving residents, certified registered nurse anesthetists, and attending physicians. These phrases included routine sayings used to inform patients about what they should anticipate, what interventions we are performing, and how the patients can participate. In addition, common questions in which a “yes” or “no” answer was desired were identified. The phrases were organized into categories related to neuraxial anesthesia, cesarean delivery, induction of anesthesia, and emergence, as well as in an instructional section to verify understanding. Within each section, the phrases were arranged in an order related to the usual sequence of events for that activity (Table 2). To facilitate rapid selection, the display window was sized such that all phrases in a given group were visible, and phrases anticipated to be activated most frequently were placed as “hot buttons” at the left side of the frame (Fig. 2).

Table 2:
List of Prerecorded Phrases Within the Communication Program
Figure 2:
CommunicatOR™ user interface. This is a screenshot of the language program developed for patients with limited English proficiency. The desired language is selected from a droplist, and clicking one of the radio buttons in the “Area” section displays a list of phrases specific for that environment. The volume of the output is controlled by the slider, and the “Speak” button causes the highlighted phrase to be output in the selected language. The selection automatically advances to the next phrase, and the prior phrase can be repeated by clicking the “Repeat Last” button. Frequently used phrases can immediately be spoken using the buttons at the left of the window. A phrase can be automatically respoken at the selected interval in seconds by checking the box marked “Repeat every.”.

During the development stage, volunteer residents validated the usefulness and completeness of the phrases by testing during simulated procedures. This resulted in the addition of several instructions that had been neglected and modification of others to improve comprehension.

Clinical faculty and residents from the department of anesthesiology and nurses from the hospital's Chinese Health Information Center fluent in our targeted languages were solicited to translate the list of phrases. We thought it was important to use native speakers who were both knowledgeable medically and familiar with the culture of the patients to provide accurate translations.

Voice recording and sound editing were done using the software program Audacity®, a free audio editor and recorder.a The audio files were recorded into “.wav” files and then edited to improve clarity, volume, pitch, and pace.

We developed a software application (CommunicatOR™) in VB.NET (Microsoft, Redmond, WA) that categorically grouped the phrases and allowed care providers to select a phrase, adjust the volume, and play the associated sound file to the patient. A midrange volume was selected for the default volume and the highest allowable sound level was set so as to avoid the risk of acoustic injury, because testing revealed the maximum output to be excessively loud. A volume test button is provided to allow the provider to sample the headphone output before placing it on a patient. The user interface was designed to require minimal interaction from the provider, with large buttons for the most common phrases, the ability to auto-repeat phrases (e.g., “open your eyes” every 15 seconds), and easy switching among the various environments (Fig. 2). Our anesthesia information management system workstations are touchscreen enabled, which also simplifies the process of phrase selection and playback. The program was deployed on our anesthesia information management system workstations, which are attached to our anesthesia machines, and can be activated from a link on a web page displayed within our system. The software is also installed on several notebook computers to allow portable use in areas lacking an anesthesia machine. Lightweight, inexpensive (<$5), hard plastic “in-ear” headphones (to allow easy decontamination between cases) with 6-ft cable extensions were deployed in each OR.

After receiving IRB approval, a convenience sample of obstetrical patients speaking either Mandarin or Cantonese who used the language program was asked to complete an anonymous questionnaire, translated into Chinese, about their experience (Fig. 3). We focused on this group because of the relatively large number of patients available for evaluation compared with other geographical areas, and the ability to ensure follow-up in 100% of cases. Among Chinese dialects, Cantonese is represented in our Chinatown community at a higher incidence (approximately 40%) than is present worldwide (10%).

Figure 3:
Patient satisfaction quality assurance questionnaire. The English and Chinese versions of the survey instrument are presented. During the study, the Chinese version was completed anonymously by patients to assess acceptance of the language program.

Patients were approached upon their initial interaction with their anesthesia care providers and given the option of using the software and headphones to facilitate communication throughout the perioperative period. After completion of the preoperative interview and after obtaining consent by a hospital interpreter, bilingual telephone interpreter, or through a friend or family member, patients were instructed on the software's application in their native language (as prerecorded within the application) and their understanding of its use was determined through appropriate responses to some simple, responsive questions.

Upon termination of the procedure and recovery from the anesthetic, the questionnaire was given to all participating patients for optional completion. The survey was conducted by a member of the department of anesthesiology or a labor and delivery nurse involved in the care of the patient. Surveys were variously conducted on the labor floor after delivery, or the next day, in the patient's hospital room. The 95% lower confidence limits (LCLs) for proportions were calculated using the method of Blyth-Still-Casella (StatExact 9; Cytel, Inc., Cambridge, MA).


During the study interval, we approached 25 parturients with varying levels of English comprehension, and all agreed to use the language program. Each used it throughout her interaction with the anesthesia care providers during labor and delivery, and all patients completed the survey. Twenty-one patients received a spinal or combined spinal-epidural in the delivery room before cesarean delivery (1 of which proceeded to general anesthesia), 2 received an epidural in their laboring suite before vaginal delivery, and 2 received an epidural in their laboring suite before requiring a cesarean delivery in the delivery room.

Acceptance of the process was high, with all patients indicating that they would like to use it again were they to return for another procedure requiring anesthesia. Eighty-eight percent (LCL = 72%) of patients indicated that having instructions in their native language made them feel more relaxed, whereas the experience was neutral in the remainder. Comprehension of the phrases presented was high, with 96% (LCL = 82%) of patients indicating that they understood all instructions. Ninety-six percent (LCL = 82%) of patients indicated that they would be likely to refer friends and family to our institution based on the availability of this device. Survey results are summarized in Table 3.

Table 3:
Results of Survey of Chinese-Speaking Parturients Using the Communication Program


Nearly all patients reported favorably on the use of the device, which is now in routine use throughout our ORs. Anecdotal reports from staff using the device on patients speaking languages other than a Chinese dialect or outside the labor and delivery suite reveal similar satisfaction with the process. We are in the process of translating our questionnaire into all languages supported by the software.

Department of anesthesiology staff, and labor and delivery nurses who participated in cases in which the program was used provided feedback on the system throughout the implementation phase. They perceived no negative impact from use of the device but requested adding languages not currently supported, and identified the need to add a few additional phrases. In addition, components of our institution's “time-out” procedure (e.g., questions asking the patient to state her name, date of birth, obstetrician's name, and procedure being performed) were also suggested as improvements.

One limitation we noted was that in response to questions asking for a simple “yes” or “no” answer, some patients would reply to us in their native language. When this happens, we repeat the instruction asking that they respond by nodding or shaking their head, or by squeezing our hand once or twice, respectively. We plan to add a specific instruction alerting the patients of our inability to understand what they are saying.

Our system requires physical interaction with the computer screen (either by touch or using a mouse), which may require a second operator if the provider is wearing sterile gloves. Other systems use voice recognition as the trigger, but performance is degraded in the presence of ambient noise6 or during stress4 (when voice pitch and enunciation may be affected), and voice training of the device may be required.4 Although the accuracy of voice recognition has improved greatly, word error rates are still substantial and the quality of ad hoc translation is often poor.7

Other limitations of the study include lack of evaluation in environments outside the labor and delivery suite, and restriction to patients speaking the Chinese dialects of Cantonese and Mandarin. We have no reason to suspect that other groups will report different experiences, but this was not tested. Although the headphones isolated the patient from the frequently loud ambient noise within the OR,8 this connection also shut off the computer's external speakers, thereby removing verbal cues to the OR staff who were communicating with the patient. An audio splitter and a small external speaker could be added to restore this capability. Although some of our providers anecdotally reported successful use through the computer's internal speakers when the headphone was missing, we did not formally evaluate comprehension between the 2 methods. Unless single use is planned, we do not recommend headphones with foam earpieces because of potential cross-contamination.

We hypothesize that patient safety would be improved by use of a communication device such as the one we developed, but our study was not sufficiently powered to measure this. Because patient acceptance was high, we focused on the issue of patient satisfaction as justification for expansion of the process to the remainder of the ORs.

Use of our software is not necessary to implement a similar process of foreign language phrase presentation. However, we think that having a highly interactive user interface that requires minimal provider interaction is beneficial to avoid potential distraction from other clinical tasks. Modifying the program for use in other countries would be straightforward. We have provided in Table 2 the list of our current phrases to assist others wishing to create their own versions. Recorded sound files could be stored and played back on any number of devices, including cell phones, MP3 players, or computers with either local or remote storage of the sound files. User interface considerations with the specific device chosen, however, will need to be addressed, especially those related to the efficient selection of phrases to be played. We have developed a version of the system for smartphones running Android 2.2 (Google, Mountain View, CA) and are in the process of doing the same for iPhones (Apple, Cupertino, CA).

The process we describe was designed to facilitate communication wherever anesthesia care providers attend to patients not fluent in the local language. In addition, it may also be useful for teams participating in surgical missions in foreign countries.

In summary, we developed, implemented, and evaluated a computerized system to facilitate communication during the perioperative period with patients having limited English proficiency. Acceptance of the process, as assessed by a survey of Chinese-speaking parturients, was high.


We recognize the contributions of the following individuals who provided language translation, voice recording, and other material assistance for this project: Alice Chen, Anna Greshner, John Jun, Javier Kaplan, Marybeth Pollock, Jian-Zhong Sun, and Huan Qin Ye.

a Available at: Last accessed 10/27/2010.
Cited Here


1. US Census Bureau. American FactFinder: 2006–2008 American Community Survey, Philadelphia City, Pennsylvania. Available at: Accessed December 18, 2009
2. US Census Bureau. Census 2000. DP-2. Summary File 3. Zip Code Tabulation Area 19107. Available at: Accessed September 19, 2009
3. PTS Banner, DARPA One-Way Phrase Translation System. Available at: Accessed 10/27/2010
4. Cohen MP. The Voice Response Translator: A Valuable Police Tool. Available at: Accessed July 27, 2010
5. Training Systems Division–Orlando Naval AirSystem Command Voice Response Translator (VRT) Support for Prototype Development and Results of Initial Field Testing. 2003. Available at: Accessed July 27, 2010
6. Waibel A, Badran A, Black AW, Frederking R, Gates D, Lavie A, Levin L, Lenzo K, Tomokiyo LM, Reicher J, Schultz T, Wallace D, Woszczyna M, Zhang J. Speechalator: two-way speech-to-speech translation on a consumer PDA. Presented at the 8th European Conference on Speech Communication and Technology, Geneva, Switzerland, September 1–4, 2003. Available at:∼awb/papers/eurospeech2003/speechalator.pdf. Accessed July 27, 2010
7. Precoda K, Zheng J, Vergyri D, Franco H, Richey C, Kathol A, Kajarekar S. IraqComm: A Next Generation Translation System. In INTERSPEECH-2007, 2841–2844. Available at: Accessed August 5, 2010
8. Stringer B, Haines TA, Oudyk JD. Noisiness in operating theatres: nurses' perceptions and potential difficulty communicating. J Perioper Pract 2008;18:384–91
9. US Census Bureau. Census 2000. DP-2. Summary File 3. Geographic Area: Philadelphia City, Pennsylvania. Available at: Accessed September 21, 2009
    10. US Census Bureau. American FactFinder: 2005–2007 American Community Survey, Philadelphia City, Pennsylvania. Available at: Accessed September 21, 2009


      Name: Brad M. Taicher, DO, MBA

      Contribution: Co-first author. This author helped design the study, conduct the study, analyze the data, and write the manuscript.

      Attestation: This author has seen the original study data, reviewed the analysis of the data, approved the final manuscript, and is the author responsible for archiving the study files.

      Name: Rammy I. Alam, DO

      Contribution: Co-first author. This author helped design the study, conduct the study, analyze the data, and write the manuscript.

      Attestation: This author has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.

      Name: Joshua Berman

      Contribution: This author helped design the study.

      Attestation: This author approved the final manuscript.

      Name: Richard H. Epstein, MD

      Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript.

      Attestation: This author has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.

      © 2011 International Anesthesia Research Society