Language Translation Apps in Health Care Settings: Expert Opinion (Preprint)

Abstract

Background: Currently, over 300 languages are spoken in Australian homes. People from non-English speaking (NES) countries without proficient English may not receive equitable care if their health care workers do not speak the same language. Use of professional interpreters is considered the gold standard, but, for a variety of reasons, is often limited to specific aspects of care (e.g. diagnosis and consent). People from NES backgrounds may be disadvantaged when health care workers try to "get by" with minimal means of communication. With the emergence of mobile technologies, health care workers are increasingly using digital translation tools to fill this gap. However, many of these technologies were not developed for or evaluated in health care settings, and may pose a risk if used for certain situations and conversations. Objective: The aim of this study was to evaluate language translation applications that have the potential to be used in health care settings. The applications will be analysed for their capabilities and potential risks. Methods: Translation applications were identified by searching the Apple iTunes Store, and published and grey literature. Applications that met inclusion criteria were reviewed in two stages. Stage 1: was conducted by two independent researchers who evaluated technology features (e.g. input and output methods, languages available). Stage 2: was conducted by two independent experts in translation and cross-cultural communication, who evaluated the potential risk when used in a health care setting. Highrisk health care conversations involved clinical assessment, provision of diagnoses, conversations about treatment and care planning, discharge planning, and medico-legal information such as seeking consent for medical treatments. Results: Fifteen applications were evaluated. Eight of the 15 applications contained voice to voice and voice to text translation options. Six applications were restricted to using pre-set health phrases only, while one application used a combination of free input and pre-set phrases. Five applications were excluded prior to Stage 2. Six of the 10 remaining applications reviewed in Stage 2 were specifically designed for health care translation purposes. Of these, two applications were rated as low risk for use in the health care setting -CALD Assist and Talk To Me. Both of these applications contained simple and appropriate pre-set health phrases and did not enable free input. Conclusions: No translation application was without risk and none should replace professional interpreters. However, some applications may be suitable for low risk everyday conversations, such as those that enable pre-set health phrases to be translated on subject matters that pose little to no risk in the continuum of care. Further research into the use of translation technology for these types of conversations is needed and clinicians should use translation technology cautiously and consider the risks before use.