Elsevier

Brazilian Journal of Physical Therapy

Volume 24, Issue 6, November–December 2020, Pages 524-531
Brazilian Journal of Physical Therapy

Original Research
Tackling the language barrier to implementing research into practice: A survey of usage of the Physiotherapy Evidence Database

https://doi.org/10.1016/j.bjpt.2019.10.003Get rights and content

Highlights

  • Language is a barrier to implementing research evidence into practice.

  • Physiotherapy Evidence Database (PEDro) content is available in 13 languages.

  • English, Portuguese, Spanish, German, and French PEDro content is used the most.

  • Region-specific promotion of the underused PEDro content may facilitate global usage.

  • Awareness and collaboration between stakeholders could reduce the language barrier.

Abstract

Background

Language is a barrier to implementing research evidence into practice. Whilst the majority of the world’s population speak languages other than English, English has become the dominant language of publication for research in healthcare.

Objective

The aim of this study was to quantify the usage of the Physiotherapy Evidence Database (PEDro) web-site (www.pedro.org.au) and training videos by language, including the use of online translation, and to calculate relative usage of the different sections of the web-site.

Methods

Google Analytics was used to track usage of the PEDro web-site for July 2017 to June 2018. The number of views of each of the PEDro training videos was downloaded from YouTube for January 2015 to August 2018. The pageviews and videos were categorized by language and, for pageviews, web-site section.

Results

2,828,422 pageviews were included in the analyses. The English-language sections had the largest number of pageviews (58.61%), followed by Portuguese (15.57%), and Spanish (12.02%). Users applied online translation tools to translate selected content of the PEDro web-site into 41 languages. The PEDro training videos had been viewed 78,150 times. The three most commonly viewed languages were English (58.80%), Portuguese (19.83%), and Spanish (6.13%).

Conclusions

There was substantial use of some of the translated versions of the resources offered by PEDro. Future efforts could focus on region-specific promotion of the language resources that are underutilized in PEDro. The developers of PEDro and PEDro users can work collaboratively to facilitate uptake and translate resources into languages other than English to reduce the language barrier in using research to guide practice.

Introduction

In the late twentieth century, English became the ‘lingua franca’ of science.1 Over 80% of peer-reviewed journals across all scientific disciplines in the SCOPUS database are written in English.1 Similarly, data from across all areas of healthcare indicate that 80–90% of articles are published in English. English also predominates in published physical therapy research. For example, the Physiotherapy Evidence Database (PEDro) identifies and indexes published articles reporting the results of trials, reviews and guidelines evaluating physical therapy interventions, regardless of the language of publication; yet approximately 90% of the content is published in English.2

The predominance of English in scientific publications means that proficiency in English enables researchers, clinicians and other users of research to capitalize on research findings to a greater extent. This may explain why a global survey found that a country’s English proficiency correlates with its level of innovation.3

The most recent data show that approximately 20% of the world’s working age population speaks English and less than 5% have English as their first language.4 An evaluation of global internet utilization reveals that English accounts for 25% of usage, followed by Chinese (19%), Spanish (8%), and Arabic (5%).5

Clinicians, policy makers, educators, and researchers may encounter difficulty in identifying and reading high-quality research if they are not able to access resources in their native language.6 Recent surveys of Brazilian physical therapists have found that whilst they are motivated to incorporate research into their clinical practice,7, 8 the predominance of resources in English appears to be a barrier as resources in Portuguese are accessed preferentially.7 This language barrier could result in unnecessary duplication of research, as well as introducing bias into clinical decision-making.9

The mismatch between the proportion of articles published in English (80–90%) and the global proficiency in the English language (20–25%) points to the need for strategies to enable equitable access to research. Advances in machine learning capabilities have led to the development of online translation tools, but the accuracy of machine translations may be insufficient to support clinical decision-making. Whilst translations from Portuguese and German into English are reasonably accurate, machine translation from Hebrew, Japanese, Spanish, and Chinese into English are relatively poor.10 Translations from English into French, Italian, Swedish, and German have reasonable accuracy, whilst translation from English into Japanese, Persian, and Vietnamese rate particularly poorly.11 Translation issues include difficulty with finding equivalent vocabulary and working with an entirely different syntax, as well as difficulty displaying the language script.12 Translating sections of text out of context may fail to yield a reliable translation, which may be critical in understanding the research. Professional translation services would address these issues, but creates a financial barrier for researchers.13

To help reduce the language barrier to consuming research, providers of bibliographic databases have started using translation and multi-language strategies. For example, the Cumulative Index of Nursing and Allied Health Literature (CINAHL) database allows users to search for articles using any of 24 different languages.14 Ovid Technologies provide natural language search interfaces in eight languages (English, French, German, Spanish, Simplified Chinese, Traditional Chinese, Japanese, Korean).15 PubMed is developing multilingual search engine interfaces,16 as well as allowing users to search for transliterated titles using English characters.17 The Cochrane Collaboration established its multi-language strategy in 2014, where the Cochrane Community provides human (and some machine) translation of Cochrane content into 15 languages, with 23,006 translations of abstracts and plain language summaries published by December 2017.18 Wikipedia, a highly utilized source of medical information online,19 includes content in 255 languages — the largest amount of medical content being in English, German, French, Spanish, Polish, Italian, Portuguese, Russian, Dutch, Japanese, Arabic, and Swedish.20 The PEDro evidence resource provides human translations of its web-site into 12 languages: English, Simplified Chinese, Traditional Chinese, French, German, Italian, Japanese, Korean, Spanish, Portuguese, Tamil, and Turkish.21 The training videos on the PEDro YouTube channel have also been translated into these languages except Turkish, but have also been translated into Dutch.

Evaluation of the impact and usage of multi-lingual evidence resources is required to guide their development and to optimize efforts to reduce the language barrier to using published research. An analysis of PubMed and Medline Plus in English and Spanish found higher use of the English versions in countries with greater English proficiency, and unsurprisingly, higher use of the Spanish platform in countries with greater Spanish proficiency.22 However, we have been unable to identify evaluations of healthcare databases by user language that are available in the public domain, other than the country of origin of the users of resources.23

The aim of this study was to quantify the usage of the PEDro web-site (www.pedro.org.au) and training videos by language. Use of online translation of web-site pages was also quantified. A secondary aim was to calculate relative usage of the different sections of the web-site to identify the most used elements that could be targeted in future translation efforts.

Section snippets

Extraction and processing of pageviews

Google Analytics was used to track usage of the PEDro web-site (www.pedro.org.au). The metric used in the evaluation was “pageviews,” or the number of times each universal resource locator (URL) was viewed by a user.24 The number of pageviews for each URL in the www.pedro.org.au domain for a 12-month period, from 1 July 2017 to 30 June 2018, were downloaded from Google Analytics. Two authors independently categorized the pageviews to confirm accurate classification; disagreements were resolved

Results

The total pageviews for the www.pedro.org.au domain between 1 July 2017 and 30 June 2018 was 2,840,053. Of these, 11,631 (0.41%) pageviews were omitted from the analysis because they contained error messages (3860), mistakes that would not take the user to a page (3097), queries (2829), URL paths produced by the content management system (868), and paths for pages that were under construction (639) or not publicly available (338). The remaining pageviews were included in the analysis: 2,825,557

Discussion

This study achieved its aims of quantifying use of the various sections of the PEDro web-site by language, online translations of the web-site, and use of the training videos by language. The English-language sections of the PEDro web-site had the largest number of pageviews, followed by Portuguese, Spanish, German, and French. The language sections with the lowest usage were Tamil, Simplified Chinese, and Korean. Users applied online translation tools to translate pages in the PEDro web-site

Conclusions

This study has demonstrated substantial use of translated versions of the resources offered by PEDro. Variations in usage of PEDro content by language suggest avenues for further translations and other changes to facilitate usage internationally. Implications for the developers of PEDro (or other resources), professional organizations, policymakers, and physical therapists in clinical or research roles include investing in translation into non-English languages, being aware of available

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Ethics and study design

Ethics approval was not required as the project analyzed anonymous user data obtained from Google Analytics and YouTube. The STROBE checklist was utilised for reporting of this observational study.

Conflicts of interest

The authors are all involved in the production of the Physiotherapy Evidence Database (PEDro). AM is employed as the PEDro Research Officer and MRE, SJK and AMM sit on the Steering Committee for the PEDro Partnership. They have no other conflicts of interest to declare.

Acknowledgements

PEDro is funded by industry groups including the Australian Physiotherapy Association, Motor Accident Insurance Commission, Transport Accident Commission, Cerebral Palsy Alliance, Chartered Society of Physiotherapy, and over 40 other physiotherapy member organizations.

References (27)

  • T. Amano et al.

    Languages are still a major barrier to global science

    PLoS Biol

    (2016)
  • E.M. Balk et al.

    Accuracy of data extraction of non-English language trials with Google Translate

    (2012)
  • E.M. Balk et al.

    Data extraction from machine-translated versus original language randomized trial reports: A comparative study

    Syst Rev

    (2013)
  • Cited by (3)

    • Using research to guide practice: The Physiotherapy Evidence Database (PEDro)

      2020, Brazilian Journal of Physical Therapy
      Citation Excerpt :

      Many features have been implemented within and alongside the original database to assist users to develop these skills. To facilitate the use of PEDro we have made the web-site and YouTube channel available in 13 languages.13 Both the web-site and YouTube channel are available in Portuguese, English, Simplified Chinese, Traditional Chinese, French, German, Italian, Japanese, Korean, Spanish, and Tamil.

    View full text