Background International rheumatology research often requires the use of patient reported outcome measures (PROM) such as questionnaires. These need translation into many languages and it is essential that the translated versions have a meaning consistent with the original.
Objectives We sought to quantify the extent to which difficulties may occur in translation.
Methods The 20-question Bristol Rheumatoid Arthritis Fatigue Multi-Dimensional Questionnaire (BRAF-MDQ), and the 3-question BRAF Numerical Rating Scale (BRAF-NRS) [1,2] were devised in English from qualitative interviews, focus groups and cognitive interviewing, and have since been translated into many languages. A professional PROM translation company, collaborating with the developers, used recommended iterative cycles of forward translation, back translation, review by developers, harmonization meetings and pilot testing with patients. The procedural reports from the entire process for the first 25 translations of the BRAF scales were scrutinised to identify which parts of the questionnaires caused problems.
Results Of 42 instructions, questions or response items translated, only 13 caused no difficulties with any language. Thirteen raised issues in one translation, 11 in 2 translations and 5 in 3 or more translations. The two most problematic questions regarded being embarrassed or upset by fatigue, where a culture had no directly comparable concept to the UK meaning. Difficulties were identified mainly during initial forward translations, or when these were back-translated into English by another translator. The back translations of embarrassed initially included a range of concepts, including ashamed (Canadian French, Japanese, US Spanish), uncomfortable (Dutch, Canadian French, US Spanish), bewildered (Korean), and awkward (Russian). Upset was conceptually difficult to capture in other languages and initial back translations included angry (Czech, Korean), mad (Czech), disturbed (Canadian French), antagonised (Canadian French), annoyed or irritated (Belgian Dutch, Polish, Swedish), bothered (Belgian French, Swedish), frustrated or shaken (Belgian French), bad tempered or disgruntled (German), agitated or worried (Italian), frightened or shocked (Dutch) and sad (Polish).
Even some simple instructions and words required discussion. In Belgian French turn the page was correctly translated as turn over a new leaf but the connotation of “change your life” meant it required re-wording. Hungarian translations for thinking can include insanity or mental state prior to death. Difficulties were resolved using pre-defined statements exemplifying the concepts intended in the original English version, in discussion between the BRAF developers, translation professionals, native language speakers, and in pilot testing.
Conclusions There are hidden pitfalls inherent in the process of PROM translations, which require great care, with iterative cycles of translations by professional translators and ideally with input from the developers. These contextual difficulties differ for different countries and can substantially alter the original meanings. Detailed attention is required to convey the original meaning of questionnaires.
Nicklin et al. Arth Care Res 2010:62:1552-8;
Nicklin et al. Arth Care Res 2010:62:1559-68
Disclosure of Interest None declared