Background International studies necessitate validated translations and adaptations of PROMs. However, the exact technique to translate and validate a PROM is not well defined. The full procedure can be complex and costly: the Bristol Rheumatoid Arthritis Fatigue Multi-dimensional Questionnaire and Numerical Rating Scales (BRAF-MDQ; BRAF-NRS)1 were translated by professional PROM translators using an extensive protocol. A simplified procedure can also be used as was the case for the Rheumatoid Arthritis Impact of Disease Score (RAID) - translated by the clinician developers, using a brief protocol.2
Objectives To compare the full versus simplified procedures used for these 2 questionnaires, in 5 languages, using formal cognitive interviewing (CI).3
Methods In each language (Dutch, English, French, German, Spanish, Swedish), 10 patients per country completed all 3 PROMS, using CI (verbalising their thoughts and audio-taped). Each item was analysed for Understanding, Retrieval (of information), Judgement and Response (options). The steering committee interpreted all the data; a cut-off for “difficult item” was set at 30% ie if >3/10 patients in a country had difficulties with an item. The number of difficult items was compared across the questionnaires. The difficult items were rephrased where appropriate.
Results As planned, 60 patients participated, of whom 50 assessed all 3 questionnaires (BRAFs not assessed in English, RAID not assessed in Swedish): 72% women; 17% aged <39 yrs, 50% aged 40-59, 33% aged >60. (a) After the extensive translation process (BRAFs), the number of difficult items identified by CI were: BRAF-MDQ, 0 difficult items identified out of 20 questions and 5 languages (0%); and BRAF-NRS had 1 difficult item out of 3 questions and 5 languages (6.7%). Specifically, Dutch patients were confused by the BRAF-NRS Coping response option, which is reverse scored (high = good) compared to the other NRS. Overall percentage of difficult items for the BRAFs was 0.87% (1 item of 23 x 5 languages, or 1 in 115). (b) After the simplified translation process (RAID questionnaire) 2 difficult items out of 7 questions and 6 languages were identified (4.7% or 2 in 42). The difficult questions were Question 2 (functional disability) in Spanish and Q5 (physical wellbeing) in Dutch. For one item, the initial translation was widely different from the original version indicating a real translation error.
Conclusions More difficult items were found for the RAID questionnaire, which had been translated using a simplified procedure, than for the BRAFs (professionally translated). The data thus suggest that while both translation methods are possible, professional PROM translation might be preferable. These problems were only identified by formal evaluation, demonstrating that cognitive interviewing is crucial in PROM translations.
Nicklin et al. Arth Care Res 2010.
Gossec et al. ARD 2009.
Drennen J. J Adv Nurs 2003.
Disclosure of Interest None declared