Back Translation for Clinical Trial Documentation - Informed Consent Forms

It is a recommendation of some IRB/ECs that back translation is included as a necessary step for the translation of informed consent forms, as well as other documents that are to be used in a clinical trial. Back translation should also be considered if the trial is particularly risky, is complex or there is any reason for wanting to be very confident that the consent form or other documents are being accurately translated without loss of meaning. It is not always necessary but there are situations when it would be important. Each trial should be considered in terms of the risk and complexity of the study, what is being asked of the participants and why it might be important to ensure there are no differences in means between different versions of study documents. If you want advice on this please ask through the discuss forums. However, the processes surrounding back translation are not well discussed in the literature, and so it can be difficult for researchers to accurately carry out a back translation review, without resorting to paying commercial translation companies for this process.

A back translation involves taking an existing translation, and then having a separate translator translate it back into the language of origin, but with no knowledge of the original text. Because the second translator (the ‘back translator’) has no knowledge of the original version, the back translation is ‘blinded’ – the translator is simply telling the project manager what the translated version says. The project manager can then use this back translation to check against the original document, to check whether there are any discrepancies in meaning.

It is well-known that nuances in meaning are often ‘lost in translation’, and the reason that ethics committees and regulatory bodies advocate the back translation of informed consent forms in particular is that there is perhaps no scenario more important for conveying meaning than when someone is submitting to participate in a clinical trial. Because translation is so subjective, it can often be the case that even an accurate and experienced bilingual translator will miss a nuance in meaning, either from a simple misjudgement, subjectivity of language, or simply through human error. Therefore, adding a back translation and review adds a quality assurance step.

This article aims to provide a comprehensive step-by-step guide to help researchers understand the aims and reasons for a back translation review, as well as providing the tools to facilitate researchers to accurately review their own documents. The article is split into three parts. Part one focuses on the process, and how to oversee the full translation and back translation. Part two focuses on how to perform a back translation review. Part three outlines how to find a suitable translator for each job.

Part 1: Overseeing the translation process

An outline of the process is shown in figure 1. To create a reviewed translation which is suitable for use in a trial, the following stages must take place:

1)      The source text is translated into the target language by a qualified translator; this step is known as the forward translation. For example, for a trial in India you may wish to translate an English IC form into Hindi – English is the source language and Hindi is the target.

N.b. it is also possible to perform a dual forward-translation approach, in which two separate translators perform a forward translation and then one or both of them review the two independent translations and combine them to create a ‘best of both’ version. This is a requirement/suggestion of some ethics committees and it’s worth checking with each local committee prior to beginning the process.

2)      The forward translation is then back translated into the original source language by a separate translator who has no knowledge of the original source text. The back translator should be native speaker of the source text language – in the example this would be English, but they should be fluent in the target (in this case Hindi) as well. 

The translator should be told that the text is being used as a back translation, and should be asked to be a little more literal than usual. This is because the document is going to be used as a ‘window’ into the forward translation. 

3)      The project manager will then review the back translation one sentence at a time. The person allocated to perform this review should ideally be independent to the process – some translators say that they will perform the back translation review themselves, but it is preferable for someone separate to perform this step since they are more likely to be able to perform an objective review.

The reviewer does not need to speak the target language (Hindi) because the back translation is showing them what is in the Hindi text. However, they should be a native speaker of the source language (in this case English). They should also be someone with a good grasp of linguistics or good language skills.

4)      If necessary, the reviewer will then discuss any discrepancies in meaning with the original forward translator, who should revise their translation if need be. They should explain exactly what they revised, and should give a new back translation of the new and improved sentence. The entire discussion process should be documented, to create an audit trail of the document showing which changes were made and why. This is explained in more detail in section 2.

5)      The final version can then be proofread - however it is worth noting that, having had the translation back-translated and checked, the proofreader should in this case only be looking for minor typographical errors and the like, rather than making stylistic changes to the meaning.

 There are several points to keep in mind when overseeing this process:

·         Always choose translators who are native speakers of the language that they are translating into.

·         Make sure the back translator is not given the source text, and that they have never seen this text before.

·         Make sure translators are suitably experienced. Ideally they should have, as a minimum, a degree in translation and two years’ experience of translation, or five years’ experience of translation (this is based on the translation requirements of ISO – the International Organisation for Standardisation). They should also be specialised in a suitable area (for example, for clinical trial documentation you could choose a translator who is used to translating documents in the following subjects: medical, legal and business, healthcare, or psychology).

·         Try to choose a translator who is from the location where the document will be used. For example, if a Spanish translation is being created for use in Costa Rica, choose a Costa Rican translator if possible, or at least a Latin American translator, rather than a European Spanish translator. This is because language can vary from place to place, and it’s particularly important in a clinical trial setting to use language that is familiar and comfortable to the patient. An example of this is that the terms used for ‘doctor’ and ‘medication’ can vary between European Spanish and some Latin American Spanish countries.

·         It’s important to make sure that the translator’s names and contact details are recorded, as well as a copy of their CV being kept. This will allow any queries to be easily checked with them, and will facilitate the process if the ethics committee request any of these details. 

It is also worth checking the requirements for each individual ethics committee or review board, since some advocate the use of two separate back translations. If this is the case, step two is simply duplicated – the back translators work separately and are both blinded – and then the reviewer will simply compare both translations simultaneously to the source text (explained further in section 2). 

Additional considerations about finding an appropriately qualified translator, and how to budget for a translation and keep in mind timings, are considered in section 3.

Part 2: Conducting the Review

Once the back translator has returned their back translation, there will be three versions of the document – the source text (the original document – in this example, English), the forward translation (for example, Hindi), and the back translation (for example, English).  

For simplicity and to make the whole process clear, it is a good idea to input the texts sentence by sentence into a basic grid. While this may take a short amount of time, it will vastly facilitate the review process and will create a clear document which can be provided to the regulatory authorities to show the changes made, if they so require. For example, a grid could look like this:

The next step is to begin conducting the review. The important thing to remember is that the back translation is to be used as a window into the Hindi. The back translation does not need altering or criticising – it is a tool to check whether the Hindi matches the source English text.

It is also important to keep in mind that the aim is to create a version of the translation which is conceptually equivalent to the source text. It is futile to expect the back translation to match the original word-for-word, so the reviewer needs to make judgements about whether the intended meaning of the source text is captured in the translation, or whether changes need to be made.

To perform the review, the back translation should be compared with the source text, row by row. So, for row 2, it is clear that there may be a problem with the Hindi – ‘name of doctor’ is not necessarily a suitable translation for ‘name of investigator’, because the investigator may not also be a doctor – in fact, this could confuse patients who may think that this row should contain their family doctor’s name.

This could have arisen because of one of three reasons:

1)      There is an error in the Hindi

2)      The Hindi is accurate but there is an error in the back translation

3)      Both the Hindi and the back translation are correct, because there is no specific word for investigator in Hindi and the term ‘doctor’ is always used.

When there is a discrepancy like this, the next step is to ask a query of the translator. This can be placed in the empty column on the right. For example, for this document the grid sent back to the forward translator may appear as follows:

Here the reviewer has picked out the discrepancies and is checking them with the translator. The translator should be asked to respond underneath the reviewer’s comments, to create a clear communication channel. If they need to alter the translation, they should be asked to do so underneath the old translation rather than just overwriting it – this will allow the reviewer to see what’s changed. The translator should also provide a new back translation for the new wording.

An example of the completed discussion grid could look as follows:

The grid above shows several scenarios. In row 1, the translator has had to update something because of a forward translation error. In row 2, the lack of context for the back translator has led to a mismatch in wording, and in row 4, the literal back translation ‘study plan’ does not quite match the English, but does show the word that would be used in this context in Hindi. It may sometimes be necessary to go back and forth several times with a translator to achieve a translation that is the optimum wording for both parties.

It is important to remember that the resulting back translations (as in the revised text in row 1) do not have to be exact, as long as the meaning is conveyed correctly (conceptual equivalence).

If two back translations are required, the system is much the same, except that the reviewer will need to focus on both back translations at once, as shown in the grid below.  

The examples above show that there are several options which can occur with dual back translations:

1)      Error in forward translation, picked up by both back translators

2)      Error in forward translation, picked up by one back translator

3)      Error in one or both back translations

As with a single back translation process, it’s important to remember that the aim is to produce a conceptually equivalent translation, rather than matching the source text word-for-word.

The Global Health Trials team are more than happy to help you with the initial stages of the back translation review or to answer specific queries – please do not hesitate to contact us.

Part 3: How to find a translator:

Although it may at first appear unnecessary, it is advisable to use a registered, qualified individual for the translation of trial documents, rather than someone who is untrained but bilingual – for example, colleagues and family. There are several reasons for this; firstly, it is very important in a clinical trial that the exact meaning is conveyed, and a translator will have been trained and will be well-practised in this art. Secondly, some ethics committees and regulatory authorities request certificates of translation and/or to see the CV of the translator, and both these aspects require that trained individuals are documented.

As mentioned above, it’s important to choose translators who are suitably qualified (a degree in translation plus at least two years’ experience, or five years’ experience as a minimum). They should always speak the language that they are translating into as their first language, and it is wise to choose someone who is very familiar with the location which the document will be used in (for example, use an Indian Tamil translator rather than a Singaporean Tamil translator if your trial is in India).

Translators can be easily found through online resources, and can be contracted for freelance jobs. Ideally it’s sensible to give them several days’ warning, and to allow as long as possible for the job. Typically, a translator is able to translate 2,000-3,000 words a day, and to ensure quality it’s recommended that the work is not rushed, so researchers should make sure to allow enough time for the translation of documentation.

There are several useful resources that will help researchers to find reliable individual translators, of which www.proz.com is the largest; this can be searched by language, location and many other parameters including choosing the translator’s first language and speciality subject. Proz.com also has a useful feedback mechanism, showing what other clients have said about the translator. Most translators have their CVs online already, but if this is not the case it is sensible to request their CV prior to assigning the job. Another large resource for finding translators is www.translatorscafe.com.

Through googling your language and ‘translation’, you will probably also find some agencies dealing in your local language. Agencies usually charge a premium so will be a little more expensive than working with an individual translator, but are useful for assigning multiple translations, since there will be only one point of contact. Similarly, if a back translation of the document is required, it will be possible to send out the forward translation and then request the back translation from the same agency (as long as the agency are told that a separate translator must perform the back translation). It is therefore possible to handle the entire process with just one point of contact and one invoice.

Most translators charge a similar fee to one another, and the fee will be based on the number of words (so when requesting a quote, a word count should be given rather than, for example, the number of pages) but each will be slightly different – this should be checked before giving the go-ahead on a job. There are several things to consider checking, depending on the job:

·         The length of your document – most translators charge a minimum fee for jobs under around 600 words, or may give discounts for large orders.

·         Whether the individual charges by the source text or the target text; if by the source, clients will more accurately be able to estimate how much each job is going to cost.

·         Some translators charge more for more technical texts, so it’s worth clarifying this with them before starting. It is wise to give them a sample of the text so that they can also check they’re comfortable translating the subject matter.

·         Payment terms – most translators request payment within thirty days, but this can vary dramatically and both parties should agree on the terms before moving ahead with the work.

If a back translation is planned and it may be necessary to revert back to the original translator for queries, it is worth mentioning this to the forward translator in the first instance – they may wish to invoice on completion of the entire job (back translation review included), and charge a small hourly rate to cover the time they spend discussing the queries.

The overall rule is to make sure you give clear requests and explain the job thoroughly. 

References:

Grunwald, D and Glodfarb, N: Back Translation for Quality Control of Informed Consent Forms, Journal of clinical Research Best Practices, vol 2:Feb 2006

FDA guidelines: A Guide to Informed Consent - Information Sheet http://www.fda.gov/RegulatoryInformation/Guidances/ucm126431.htm10/18/2010