In the last article, I commented on poorly localised figurative language in Clinical Outcome Assessments. This article examines other common localisation pitfalls that we have encountered when testing COAs around the world. We offer advice about how to resolve these problems.

1.       Descriptions of Symptoms

Some medical or subjective terms around the experience of a symptom do not have direct equivalents in all languages. We have commented in other articles that some languages that do not distinguish between “pain” and “discomfort” are examples of this. Similarly, frequently used response choices also have a large amount of semantic overlap, such as “mild”, “moderate” and “somewhat”. We have seen occasions where ‘mild’, ‘moderate’ and ‘somewhat’ have been translated as the same adjective in the target language. So, the response choices back-translate as follows:

“No pain – mild pain – mild pain – mild pain – much pain – severe pain”.

Yes, you did read ‘mild pain’ three times.

One way to improve this is to allow descriptive clarifications in source and target versions to remove any ambiguity, such as:

·       Mild pain (noticeable but does not interfere with daily activities)

·       Moderate pain (some interferences with daily activities but manageable)

·       Severe pain (significantly disrupts daily activities)

The addition of these functional descriptions retains accuracy and ensures that the choices are interpreted consistently across languages.

2.       Figurative language that does not translate well

Idioms, metaphors and culturally specific phrases that make perfect sense in the source context often do not have a direct equivalent in other languages. This can lead to a literal translation that makes no sense or misleads participants in target locales. A common example of this is ‘feeling blue’ and I have seen this translated literally, which causes endless problems in cognitive debriefing. More often than not, it is interpreted literally, such as ‘appearing blue’, like a Smurf.

There are two strategies to resolve this issue.

i) with existing COAs, this item needs to be flagged in an exercise like a concept elaboration that it must not be translated literally. A translatability assessment dedicated to each language could provide advice about local equivalents of the figurative language, such as ‘to sit in the well’ in Dutch, or ‘to be in a ditch’ in Portuguese.

ii) For COAs in development, do not use figurative language, if you want to make sure that all participants interpret the question in the same way. Use straightforward language instead with clear, neutral phrasing, such as ‘I am feeling unhappy today’, rather than ‘I am feeling blue today’. Care needs to be taken not to unintentionally introduce figurative imagery when intending to be neutral. For example, feeling ‘low’ or ‘down’ is using figurative imagery that is culturally bound.

3.       Inappropriate Sports, Activities and Functional Tasks.

COAs often include questions about mobility, exercise or daily activities that are culturally irrelevant. I have lost count how many times we have had to test items such as ‘Do you have difficulty skiing for 30 minutes?’ in locales in sub-tropical to equatorial regions where the average person has little experience of skiing. The result is that the item is poorly adapted, since the skiing reference is unchanged in translation. This makes the question meaningless in the target context.

For existing COAs, there needs to be some discussion prior to translation with developers about what scope there is to localise to a local sport or activity that involves equivalent demands on the patient.

For new COAs, culturally specific activities could be replaced with commonly performed movements like ‘How much have you been able to walk on uneven ground for 30 minutes?’

Other sports-related activities commonly found in COAs relate to sports common in the source country, such as golf, catch or swimming. All of these taken-for-granted activities in the source context need cultural modification.

A similar issue arises when the source contains items relating to chores and activities that are misaligned to the target cultures. COAs frequently include questions to assess daily activities, but some tasks are not universal. Many of them focus on activities, such as ‘yard work’ or appliances, e.g. ‘dishwasher’ or ‘microwave’ that are more common to the source context. A similar issue arises with questions that mention certain kinds of public transport that may not exist in the target setting.

4.       Food-related Issues

COAs often contain examples of food and dietary practices for nutritional assessment questions that may also misalign with the target cultures. Not all cultures eat the same food, such as beef and pork, or use the same utensils. Not all cultures consume dairy products as they may in the source context. Therefore, failure to localise, such as by using a direct translation potentially excludes entire groups of people. A solution to this could be to replace specific foods with neutral, adaptable alternatives, such as ‘How often do you eat red meat or your preferred protein source?’

5.       Weights, Measures and Inappropriate Units of Currency

A common yet often overlooked issue in COA localisation is the use of inappropriate or non-standard weights, measures, and currencies. Many COAs include questions that refer to body weight (pounds), medication dosages, distances (feet and inches), or financial expenditures, but these units may not align with local conventions. For instance, a questionnaire developed in the United States may refer to weight in pounds, while most countries use kilograms. Similarly, monetary amounts may be given in US dollars without adaptation, such as dollars, cents, even nickels, dimes. It is shocking how often these terms are poorly localised or not localised at all. Direct translation without conversion or contextualisation can lead to inaccurate responses or participant disengagement.

To avoid these pitfalls, COAs should be adapted to use the metric system where applicable and either local currencies or generic financial references (e.g., “your weekly grocery expenses” instead of a fixed dollar amount). Where direct adaptation is not feasible, including a conversion chart or explanation within the questionnaire can ensure clarity.

Summary

How is it possible that these important issues slip through the linguistic validation process? Yet, they do. Poor localisation compromises data quality, reduces validity and leads to participants misinterpreting the responses. Poor localisation issues such as awkward phrasing or cultural irrelevance can make the entire COA lose reliability.

Here are some obvious statements:

·       Localisation in COAs must involve using native-speaker translators with clinical expertise to minimise terminology errors

·       The cognitive debriefing process must investigate instances of culturally misaligned items more thoroughly; in our experience, it often fails to do this

·       Culturally inappropriate references must be adapted to fit local environments and, before that, discussed in detail with translators, developers and other stakeholders by means of a non-superficial concept elaboration and translatability exercises.

·       In COA development, symptom descriptions could be replaced by functional impact, rather than vague terms

·       Do not translate culturally-specific items literally, such as idioms, weights and measures, sports, household activities, but use clear, neutral wording instead.

Thank you for reading,

Mark Gibson

Leeds, United Kingdom, February 2025