Proportionate translation of study materials and measures in a multinational global health trial: methodology development and implementation

Objectives Current translation guidelines do not include sufficiently flexible translation approaches for different study materials. We aimed to develop a proportionate methodology to inform translation of all types of study materials in global health trials. Design The design included three stages: (1) categorisation of study materials, (2) integration of existing translation frameworks and (3) methodology implementation (Germany, India, Israel, Tanzania and Uganda) and refinement. Participants The study population comprised 27 mental health service users and 27 mental health workers who were fluent in the local language in stage 7 (pretesting), and 54 bilingual mental health service users, aged 18 years or over, and able to give consent as judged by a clinician for step 9 (psychometric evaluation). Setting The study took place in preparation for the Using Peer Support in Developing Empowering Mental Health Services (UPSIDES) randomised controlled trial (ISRCTN26008944). Primary outcome measure The primary outcome measure was the Social Inclusion Scale (SIS). Results The typology identifies four categories of study materials: local text, study-generated text, secondary measures and primary measure. The UPSIDES Proportionate Translation Methodology comprises ten steps: preparation, forward translation, reconciliation, back translation, review, harmonisation, pretesting, finalisation, psychometric evaluation and dissemination. The translated primary outcome measure for the UPSIDES Trial (SIS) demonstrated adequate content validity (49.3 vs 48.5, p=0.08), convergent validity and internal consistency (0.73), with minimal floor/ceiling effects. Conclusion This methodology can be recommended for translating, cross-culturally adapting and validating all study materials, including standardised measures, in future multisite global trials. The methodology is particularly applicable to multi-national studies involving sites with differing resource levels. The robustness of the psychometric findings is limited by the sample sizes for each site. However, making this limitation explicit is preferable to the typical practice of not reporting adequate details about measure translation and validation.