By Christelle Giroudet and Ana Alvarez

At first glance, translating clinical outcome assessments (COAs) from their original language (most commonly English) to other languages appears to be no more complicated than other trial planning elements.

However, as scientific instruments, the process is much more complex. COA translation needs to consider conceptual equivalency and cultural relevance. The process may also involve a large range of languages and cultures. It is not uncommon for a study to include 30-40 different language groups. To avoid potential delays in studies due to the COA translation process, sponsors need a strategy that is underpinned by best practices.

Translating COAs is an iterative process involving multiple stakeholders. Therefore, it is crucial that ample time is devoted to validating the translations. Translations must be conceptually faithful to the source while also appropriate for the target populations and cultures. Translations must reflect the intention of the original measurement, which might not be obvious from the wording. A questionnaire asking about patients’ abilities to carry out specific activities may need to be adapted to align with a different cultural context. At the same time, the question’s underlying rationale should not be undermined. Translating a patient question about eating using a knife and fork in Europe to ask about using chopsticks in Asia may be inadequate if the source instrument assesses the patient’s ability to coordinate activities using both hands. Similarly, a question about a common physical activity in one part of the world might be inappropriate elsewhere: a Canadian questionnaire asked about the patient’s difficulty shovelling snow, for Mediterranean countries this was replaced with an alternative physical activity involving a similar level of effort. Some languages have formal and informal modes of address, depending on the age or relationships of the parties involved. Translations must account for this and consider which form is most appropriate.

The goal of linguistic validation of COAs is to preserve the integrity of the source instrument’s psychometric properties. The process should be consistent across all countries and coordinated by a project manager within a language services provider in collaboration with the COA developer. Local in-country native speaker linguists who specialise in linguistic validation should supervise the completion of the work’s five phases.

1. Conceptual analysis: the instructions and instrument’s items and response options are reviewed to clarify the meaning of each and provide translation tips. The language services project manager oversees this, liaising with the instrument’s developer for feedback. This list of concept definitions is developed to harmonise translations across target countries.
2. Forward translation: the instrument is translated by two independent translators without contact with one another. A third linguist, the in-country lead linguist, analyses both versions and will use the best parts of each to create the reconciled translation.
3. Backward translation: a native English speaker translates the reconciled translation back into English. This new English version is compared with the instrument’s English source version by the in-country lead, who prepares a report on the discrepancies identified. The language services project manager reviews the documentation and any issues are resolved. The translation is upgraded and may be reviewed by the COA developer.
4. Cognitive interviews/clinician review: for patient questionnaires, several respondents in each country are given the questionnaire. The lead linguist interviews respondents to assess their understanding of translated items. Where instruments are to be completed by clinicians, a review process is carried out by a clinician in each country, who then provides feedback.
5. Proofreading and finalisation: the lead linguist, language services project manager and an independent proofreader review the final version for formatting, typographical and grammatical errors.

To minimise potential delays in studies, it is essential that sponsors start this process as early as possible. COAs should be selected carefully, some of them may not be practical in the context of an international study with constrained timelines. Sponsors should aim for efficiency, but never compromise on the quality of their COA translations. Doing so could undermine the integrity of the translated COAs and negatively impact the study’s data validity. A conservative approach is advisable, especially when adopting process innovations. Seeking support from experts is crucial. COA translations are not word-for-word translations; they require a sophisticated linguistic validation process.  Language service providers trained and experienced in linguistic validation can help sponsors to avoid mistakes that could negatively impact the outcomes of their studies.

Christelle Giroudet and Ana Alvarez are both COA Linguistic Validation Leaders, ICON Language Services