Hindi in Clinical Trials: Making Hindi Patient-Friendly

Over the years that I’ve been running around the clinical research circuit, I’ve heard countless references to raters verbally rephrasing COA items, or parts of items, to clarify their meaning for patients.

It is not a myth. Most of us know it happens. What we do not know is how often it happens, because it is not formally measured. We also do not know, in any concrete way, what effect it has on trial data. It may introduce noise. It may also, in some cases, reduce noise by rescuing an otherwise unusable translation, at least in part. Either way, once an already approved wording has to be orally reconstructed at the site level, the consistency in translation is outside of the controlled workflow.

As the translation industry adage goes, a translation is only as good as its translator. In the context of clinical trials, a translation is also only as good as the assumptions built into the scope of work, the register chosen for the target population, the extent of localization permitted by the sponsor’s internal policies, and the feedback loop between the teams producing the text and the teams using it with patients.

This feedback loop is often weak or non-existent. We sometimes see pre-trial edits from independent translation reviews. We sometimes see late-stage comments from sponsor reviewers. But we rarely get systematic feedback from where it matters most: the research sites. Cognitive debriefing cleans up a lot of errors, but in its standard form, it is simply not thorough enough to catch everything. Yet we rarely learn what did not work locally, why it did not work, and whether investigators had to verbally replace the approved wording with something patients could actually understand.

A translation of a clinical outcome measure is a living thing, in the same way that a newly created assessment is. One that has not yet been tested across multiple demographic groups in the country of origin, or proven in what it measures across those groups. The translation can be perfectly worded, but that doesn’t necessarily mean that it will do its job equally well, or at all, in a new language and cultural context.  The expectation should be that translations will need to be revised and updated as they are used over time across multiple demographics in the new target languages. Hence, the feedback loop is critical for supporting maintenance of this kind of content.

This became especially interesting in the context of Hindi.

Over the last several years, we noticed what seemed to be a decline in requests for Hindi translations for India, while Tamil, Kannada, Gujarati, Marathi, Malayalam and Telugu continued with strong traction in clinical trial scopes. Punjabi and Bengali also appeared, though always with lighter frequency. Hindi was not absent, but it no longer seemed to be the automatic headliner it once was.

This is only an observation from Santium’s own project flows. It may not reflect the wider market. Hindi may have been sourced elsewhere. Many commonly used outcome measures are already available in Hindi. But that explanation is only partly satisfying, because the same should also be true for other major Indian languages. Whatever the reason, the shift registered. 

Why is Hindi – a language recorded as the mother tongue of more than 43% of India’s population – not always included in the linguistic validation scope of work for clinical trials planned across India?

Some time later, a clue to this apparent decline came by coincidence.

I was at a conference, and tuned in to a conversation at the next table among an international group of research site investigators, some of whom were from India. They were discussing the quality of translations reaching the sites. The Indian clinicians were especially not enamoured with Hindi translations of PROs. Their view was that they were “not very usable.” Why? Because it was the “wrong language.” The rest of the group nodded, and the conversation steered to other challenges.

The wrong language?

At the time, I did not fully understand what that meant. What I found opened a part of the linguistic world that we probably should have known about much earlier. My team and I had made the same assumptions Westerners tend to make: that a major language used within a country can be used more or less the way English, French, German or Spanish are used across their linguistic territories. Hindi certainly cannot, and here I’ll explain why.