Back translations - useful or waste of time?
Written by Andres Heuberger on Wednesday, April 08, 2009
Anybody doing translation work related to clinical trials knows that so-called "back translations" are a fact of life - most drug companies rely on back translations as a QA tool by comparing the back translation and the source to one another.
However, any translation that is done well will depart from the semantics and structure of the original source language. Grammar is inherently diverse across different languages, and the translator will retain the meaning of the translation, rather than the original grammatical properties of the source text.
Not surprisingly then, there exists a healthy debate on the use and usefulness of back translations.
In our experience, the usefulness of back translations is highly dependent on the content that is being translated. In the context of informed consent forms, the consensus is that back translation should be performed to ensure that the language is a correct translation of the original, i.e., the meaning was not changed but also that the readability of the text has not been affected through the translation process. Additionally some Institutional Review Boards have made back translations a necessary step in the approval process.
For patient reported outcomes, back translations are a necessary part of the validation and cultural adaptation of the instrument for specific markets. The process includes conducting two back translations and harmonization with the original translation to accommodate the findings of the back translation.
Other content, for instance technical material or advertising and marketing material, may be better suited for an in-country review or an additional copy edit or adaptation step.
The bottom line? Use back translations with caution. It is not a silver bullet, and it can be a time-consuming and expensive task. But when employed correctly, back translations are a highly effective validation tool.
For more on the case for - and against - back translations, take a look at the following:
- Darren Clayson explains the intricate (and mandatory) process for back translations in Translation & Linguistic Validation of PRO Measures
- Back translations - tackling discrepancy in meaning highlights some of the linguistic challenges involved
- Our own Hilary Davies presented the terrific Best Practices for Back Translation as a QA Tool
Like this post? Subscribe to Medical Translation Insight via email or RSS.
Another consideration is time. It is very time-consuming to go over the possible mismatches, or issues and see if the original translation needs to be changed or not.
Careful review of a translation aligned with the source document by an expert linguist may be a less expensive alternative, but a standard operating procedure would be useful as a guide to such a review.
But, QA is not (or should not be) the main point. The point of a back translation is that multiple-language versions of (in many cases) a medical questionnaire are generated. I would agree with Dan Lufkin who writes that people who hate doing back translations should stay away from them. Back translations should be taken extremely seriously. A proper back translation should include lots of comments and remarks about the content of the forward translation. It should focus on 'equivalence' between the original and the translation. A back translation should only start after the forward translation is complete. There are organisations managing series of medical questionnaires that require a method of double forward translations and editing, then reconciliation of those 2 translations into 1 ideal version, and only then (and not earlier) start with the back translation, again by 2 back translators and followed by again reconciliation of those 2 back translations. Back translators also must be aware that they are making a back translation. Otherwise they would silently iron away weak parts and errors in the forward translation.
To do a cheap forward translation and then have it back translated just for the sake of having it done, creates a very false sense of safety. CROs and translation vendors who simple do a back translation to check the quality of a poor forward translation miss the point completely.
If the end client had gotten this "raw" BT, it would not have reflected the original message that is being delivered to the target audience and thus quite misleading.
So now I know that a BT only has value as a QA step if the translator doing the BT is competent! I agree with Simon that the person doing the BT should go beyond only translating the text and aim to ensure that the correct meaning is being conveyed... then add comments where ever there are discrepancies.