The Complexity of Linguistic Validation

Linguistic validation through the eyes of a language service provider

Authors: Éva Tóth-Tölgyesi and Andrea Varga

Introduction

Translation is always needed once a product is intended for the global market. Clinical studies involve translations at many points in the whole process.

In general, a medical translation’s quality is crucial, not only because it can save lives, but also, in case of clinical studies, for example, it can determine the success of patient recruitment. Patient enrollment is an essential part of a study cycle, which makes high quality patient-facing translations and thus linguistic validation an inescapable step. Without linguistic validation, clinical research trials are in danger of being rejected due to patient misunderstandings of poorly translated clinical instruments (Language Scientific, 2016).

When translating patient-facing materials, the following points should be kept in mind:

• the target audience
• the client’s preferences
• the target culture
• the cultural differences between the source and the target locales.

The Linguistic Validation (LV) process of Hunnect Outcomes

Firstly, we receive different kinds of files from the client. They can be source files, reference materials, glossaries, instructions, style guides etc. They must be prepared before they are shared with the linguists. Aspect 1 is the IT-side, when our experts make the source editable and ready to be translated. Aspect 2 is the content-wise validation, in which we read through the document and query the client about any ambiguous or inconsistent phrases. An important consideration at this point is to make it clear to the client that patient-facing materials need to be localized not just translated (e.g., Beaton et al., 2000).

Let’s see a quick example for localization vs. translation:

Source: Can you hold an object of a baseball bat’s weight?

If the literal Hungarian translation would be used in this case, Hungarian patients would not be able to give an adequate answer to this question. Baseball is not popular in Hungary, an average Hungarian patient simply cannot estimate the approximate weight of a baseball bat, so here localization is necessary and the Hungarian target translation would be something like: Meg tud tartani egy futball-labdához hasonló tömegű tárgyat?

Backtranslation: Can you hold an object of a football’s weight?

Now, let’s see the linguistic steps one by one in the order Hunnect performs them.

Forward translation (FT)

In this step, there are 2 translators who work on the same text simultaneously. This is a standard translation procedure. We use CAT (computer-assisted translation) tool in this step to optimize time, ensure consistency, minimize the risks of omission, numeric errors etc., enable the use of custom-made QA tools, and of course, to reduce the budget. This means that we do not work with the total word count, rather we calculate a discounted net word count. Ideally, both translators finish at the same time and two similar, but not completely identical texts go to the next linguist who will be the so-called reconciliator.

Reconciliation (Rec)

During this step, a linguist will merge the two translations into one. This step is also done in a CAT tool. The reconciliator can see both versions and opts for the better one, or they may combine the two translations and thus create a third, improved version. This step also includes the review of the translations against the source, therefore, the reconciliator may implement some further edits to the forward translation.

Proofreading (PR)

In this step, the proofreader performs monolingual proofreading. The linguist uses the source document as reference only and focuses on the target. Proofreaders make sure the translation is fluent, it sounds natural in the target language, it is free of typos and spelling errors, and it is acceptable in the target culture.

Backtranslation (BT)

This is the most powerful quality assurance step. In this step, you translate the target translation back to the source language. For example, if, during the FT step you translate from English into Hungarian, the BT step will mean translation from Hungarian into English. Once you have the BT, you have the chance to spot any mistranslations or any other kind of major or critical errors. It takes no more time than a regular translation step, and the costs can be kept at a relatively acceptable level as BT is performed by life science translators whose native language is the same as that of the target population (see more details about this in the Clinician Review step). BT is rather a literal translation, not a well-written translation that sounds natural; it will never be published, its sole function is to check that the target translation conveys the intended meaning of the source text.

Review and Editing (ReE)

In this step, the BT is checked against the original source and any deviation is recorded. If there is deviation in meaning, it should be determined whether the FT or the BT contains the error, and the necessary corrections should be made. In general, this is not a time-consuming step, but, of course, a lot depends on the quality of the previous steps.

Clinician Review (CR)

This step is not performed by a linguist, but by a practicing clinician, which makes this step both time- and budget-consuming. Let’s see the main reasons. Practice shows that medical doctors focus their time on patients in the first place. Therefore, their linguistic capacities are much lower than that of a full-time translator’s. To compare the capacity of a linguist and a clinician, it is 2500 words per day and 250 words per week, respectively. In Figure 1 you can see that this step alone accounts for 10% of the full process. As for the budget, a clinician’s hourly rate is always higher than that of a linguist. On the other hand, it is quite a challenge to make sure the practicing physician understands all the instructions, the linguistic requirements, the purpose of their task and the reference materials correctly. This is where the Project Manager’s skills are put to the test.

Harmonization (H)

This step can be the last one if a client needs no Cognitive Debriefing. Harmonization means that a linguist goes through all the suggestions made by the clinician and implements the ones they agree with. In addition, the linguist also performs computer-assisted QA checks. This is a quick step in which the linguist either finalizes the text or prepares it for the interviews.

Cognitive Debriefing (CD)

This is the most time-consuming step in the whole process. Patients are recruited to interview them in-person or online. As Figure 1 shows, this step accounts for more than 70% of the validation process. Hunnect only performs Hungarian CD, because the legal background of CD is quite complicated and should be validated locally. In some countries, it is significantly circumstantial or even illegal to involve patients in interviews. In Hungary, we contacted the authorities and learnt the necessary measures we should take before we contact patients. Personal information, type of disease etc., are sensitive data which should be managed confidentially and according to the regulations. This step can be tailored to the client’s needs, for example with regard to the number of patients to be interviewed, the distribution of men and women, the method of interviewing etc. We described our experiences concerning CD in another article.

Finalizing (F)

This is the final step which requires a linguist. No matter what the previous step was, the linguist who finalizes the text will run a final automated QA and perform all the necessary changes that were suggested in the previous step(s). If it was CD, the finalizer will work with the report made by the interviewer about their impressions based on patients’ answers. If CD was left out, we can call the last step either finalization or harmonization, they both involve similar actions.

Reporting (R)

Reporting can vary from simple to complex methods. It is 100% up to the client whether they need this step and in what complexity. The most common and the most simple form is to report the date of each step and briefly describe the process, for example, if a critical error was revealed during BT etc. A complex report can include statistics (for example, interlingual statistics), challenges during the process, any linguistic observations and even a more detailed change report.

Last, but not least, delivery to the client is performed and client feedback is requested.

The added value of the LV process by Hunnect

The benefits of LV for Clinical Outcome Assessments (COAs) are well-established (e.g., culturally appropriate, statistically valid measures across languages where the risks of mistranslations or omission are minimized); and the today, LV is an inescapable process in the localization of clinical trial documents which relies on a concise methodology and clear guidelines (see for example, Wild et al., 2005).

So, if we know that the benefits are numerous, let’s see the value Hunnect can add to the process. As described above, Hunnect has an established LV process which relies on industry standards and guidelines and combines these with the perspective of linguists. This holistic approach further strengthens the success of the linguistic validation process and thus the success of clinical trials.

The added benefits of our process are as follows:

• We use CAT tools which not only reduce the budget and speed up the translation process, but, perhaps more importantly, ensure consistency, minimize the risk of omission, and boost accuracy and precision by allowing for the use of custom-made quality assurance tools.
• Building on the flexibility of industry guidelines and good practices, we have designed our own linguistic validation process. For example, instead of leaving Proofreading to the very end, we introduced this step after the reconciliation of forward translations. By performing monolingual proofreading before backtranslation, we can aid the success of backtranslation by eliminating any possible mistranslation and improving the quality of the translation.
• Including a practicing physician in the validation process is one of our core steps. However, we believe, this step should be followed by a harmonization step when a linguist makes sure the clinician’s suggestions are incorporated into the translation in line with client requirements, source considerations, and the orthography rules of the target language.
• Each and every step performed by Hunnect from preparation to QA check is performed by life science professionals. Our Life Science Department works with reliable and tested in-house and freelance life science linguists and CRAs.

For us, translation which is tailored to the needs of the client and that of the target population is everything. We consider ourselves the bridge that connects the client with their audience. In the world of clinical studies, linguistic validation is a crucial and, if done properly, a powerful step in licensing new medical products.

References

Beaton, D. E., Bombardier, C., Guillemin, F., & Ferraz, M. B. (2000). Guidelines for the process of cross-cultural adaptation of self-report measures. Spine, 25(24), 3186–3191.

Language Scientific. (2016, November 7.). What is linguistic validation? https://www.languagescientific.com/what-is-linguistic-validation/

Wild, D., Grove, A., Martin, M., Eremenco, S., McElroy, S., Varjee-Lorenz, A., & Erikson, P. (2005). Principles of good practice for the translation and cultural adaptation process for patient-reported outcomes (PRO) measures: Report of the ISPOR Task Force for Translation and Cultural Adaptation. Value in Health, 8(2), 94–104.