Finding the ideal process for pharmaceutical translations

Over the past few years, the amount of translations in the pharmaceutical industry has grown considerably. This has been driven by global regulatory requirements as well as new opportunities that have opened up in new markets, whether for marketing medicinal products or conducting clinical trials. This article reviews some of the current issues in pharmaceutical translations related to marketing authorization and finding the optimum translation process. As always, it is useful to consider the big picture of the pharmaceutical industry and what is moving and shaking it, since it ultimately and necessarily trickles down to how translations are managed and affected.

An unprecedented number of branded medicines within the industry are set to lose patent protection in the near future. This so-called “patent cliff” means some $250 billion in sales may be at risk between 2011 and 2015. This includes such blockbusters as the world’s best-selling drug brand Lipitor, a cholesterol-lowering medicine from Pfizer, with sales of well over US$11 billion a year and whose patent expired in June 2011. Plavix, marketed by Bristol-Myers Squibb and Sanofi-Aventis and the second best-selling drug globally with sales of US$9 billion annually, will see its patent expiring in 2012.

Pharmaceutical companies are working hard to tackle this looming revenue drop and to fight off the biotechnology and generic industry by ever-increasing investment in research. However, the overall resource and development (R&D) productivity of the industry has been in decline for some time, and the new drugs being discovered don’t seem to be replacing the sales income of those medicines that are set to go off-patent soon.

It is not surprising in light of this that the pharmaceutical industry is one of the most active in regards to mergers and acquisitions, as acquiring another company’s new drug pipeline can be more cost-effective and possibly less risky than investing in its own R&D. The ways in which these trends manifest themselves include major efforts to speed up the process for new drugs to the marketplace, including the process of obtaining regulatory approval. A growing number of clinical trials are increasingly conducted in emerging markets such as Central and Eastern Europe, Latin America, Russia, India or China. Cost is one reason, but these regions can equally benefit from large treatment populations or trial-naïve populations, unexposed to medications that could interfere with the given trial drug. Overall, there is a drive towards cost efficiency and cost management, including increased outsourcing of research and clinical trials to contract research organizations (CROs), rather than managing these in-house.

It is probably safe to say that when it comes to the maturity of managing its translations, the pharmaceutical industry would not traditionally score high and would lag behind what may be the average in high-tech or related industries. This is perhaps not surprising, as the industry’s main focus has traditionally been on ensuring the utmost quality and accuracy of translated materials rather than overall efficiency and transparency of processes. Where even a small translation error may potentially have dire and literally fatal consequences, it is understandable that the cost would not come first and foremost. Instead, quality, accuracy and regulatory compliance have been the rules of the day so far.

As a result, many pharmaceutical companies may work in a highly decentralized model, with several functional groups overseeing their specific translation efforts. The adoption of translation management systems or centralized translation memories is relatively low, as much of the work related to translations is manual and based on Microsoft Word documents, spreadsheets and e-mails — though with a clear audit trail available.

But there is change afoot, and the previous factors are some of the main driving forces. There is a recent but accelerating trend toward centralizing translations and their purchasing. Many pharmaceutical companies have taken steadfast efforts to better organize and optimize their translation setup.


The life of PIM

A major impetus to improving the overall efficiency of one particular area — marketing authorization — was recently provided by the European Medicines Agency (EMA) and its Product Information Management (PIM) project — or so it seemed. Via its Centralised Procedure (CP), the EMA administers the process of marketing authorization in Europe, and this procedure allows submitting one application that is valid in all EU and EEA-EFTA states (Iceland, Liechtenstein and Norway).

The medicinal product information needs to be submitted in all EU languages plus Icelandic and Norwegian, which includes materials such as the summary of product characteristics (SPC), labeling and package leaflets. SPCs are used by physicians and pharmacists to advise patients on drug side effects, how the drug should or should not be taken, dosage, storage and whether a prescription is  required or not.

Translation volumes related to submissions are not huge, but there is a fair degree of complexity arising from the fact that the same information is often duplicated in multiple documents to account for various dosage forms, product strengths and presentations. This results in a large set of documents that need to be managed, kept up to date and synchronized.

Within the 24 languages that are currently involved in the CP, there may be as many as 1,000 documents for a single trade name, which need to exist on paper or as Microsoft Word documents. This increases the likelihood of creating an error in the process and is a major burden for the EMA, member states’ competent authorities and pharmaceutical companies, as well as providers of market authorization translations.

Aware of the complexity and cost of ensuring product information translations for its marketing authorization, the EMA has a program in place for providing such translations for free to SMEs (micro, small and medium-sized enterprises, as defined by the EU). The agency will do so for all the required languages except for Norwegian and Icelandic, according to the normal timelines, through the Centre for Translation (CdT) in Luxembourg. These free-of-charge translations provided by the EMA to SMEs in the pharmaceutical industry are complemented by other concessions offered, such as substantial (90%) reduction on certain fees or payment deferrals.

The idea behind PIM has been to simplify the submission and review process and to establish an XML-based electronic exchange of information, designed to reuse common content globally and eliminate the need to use paper documents and materials in Microsoft Word formats, as is the case now. Currently, a large amount of effort needs to be dedicated to ensure accurate formatting and presentation in the required Word-based Quality Review of Documents templates, as to the actual content.

As a by-product, the translation process could be made more efficient by adopting PIM. The use of structured formatting in XML and the electronic exchange of information throughout the authoring, review and submission processes would lend itself readily to the adoption of more advanced translation management systems and TMs. Last but not least, this would help companies reach the short window to provide translations and their post-review finalization.

PIM was introduced by the EMA in December 2005, and the project had been through a pilot phase that attracted a few major pharmaceutical companies and led to positive results. The agency had been making all the right noises about the standard’s upcoming adoption until the PIM project came to an abrupt end in March 2011, when the EMA pulled the plug on the project. The agency cited a review of its business strategy in light of the new legislation and a budgetary review as reasons for halting the project, while saying it remains committed to the concept of the exchange of structured product information and suggesting it will return to this once the review process is completed.

At this moment, the jury is still out on what this is going to mean. This decision has certainly removed one of the most immediate drivers for pharmaceutical companies to adopt systems and technologies to better manage their translation processes. A number of companies have put their initiatives in this area on hold. Submission to the EMA for market authorization in Europe will continue to be done electronically with Microsoft Word documents or in paper for the foreseeable future.

But the underlying idea has been sound, and there are clear benefits in adopting a centralized translation system. While we may have to wait a bit longer for a standardized XML format to be prescribed for electronic submissions in Europe, the number of languages that will require translation as well as the number of sundry documents needed is not going down any time soon. Clearly, the current e-mail-centered model of communication and delivery is not sustainable indefinitely, considering the way other industries are rapidly developing.

Inspiration and hope may be drawn from the structured product labeling (SPL) system that exists for the United States and has been mandatory there since November 2005. This XML-based standard was adopted by the US Food and Drug Administration, which requires electronic labeling submissions as well as annual report submissions to be done using the SPL format.


No silver bullet

The quality of translations is clearly paramount in the pharmaceutical industry. What is not always so clear is how to achieve this required quality effectively.

This does not concern only translations for regulatory submission, but applies equally to translations required during clinical trials, including pre-clinical studies and the individual phases of trials. The best practice in the industry has seen several types of the translation process developed and adopted. Each has a different method or level of quality assurance (QA). There seems to be no one-size-fits-all process. The use of a particular process may depend on the type of the material in question and whether this is a patient-facing document or not, as well as any specific company standard operating procedures in place.

For nonpatient-facing materials as well as some patient-facing documents, the typical translation process is rather traditional — a qualified/certified clinical translator doing the initial forward translation into the target language (TL), followed by a complete review of the translated document for linguistic quality as well as for factual accuracy, cultural and technical language suitability for the specific target country market. These reviews are conducted by a combination of qualified linguists and medical professionals and practitioners who have the practical experience in the subject area in the given country.

Other types of patient-facing materials, such as informed consent forms, would see a back translation of the previously translated content into the original language added, followed by a comparative review of the source and back translation, and the usual QA reviews.

In a related scenario, used typically for materials such as patient reported outcomes (PROs) — for questionnaires collecting responses directly from patients during trials, multiple (typically two) forward translations are required. These are followed by reconciliation — merging more than one forward translation into a single forward translation, and then one, or in some cases two, back translations, and subsequent harmonization.

In this case, the International Society for Pharmacoeconomics and Outcomes Research has taken great strides in harmonizing the process, after it had identified a general lack of consistency in current methods and published guidelines. The ground rules have been summarized in The Principles of Good Practice for the Translation and Cultural Adaptation Process for Patient-Reported Outcomes Measures, published by its Task Force for Translation and Cultural Adaptation.


Backing off back translations

The introduction of back translation is a safety-check for companies to ensure nothing has been “lost in translation,” but it is also a major cost driver, effectively doubling the cost of any particular translation.

Finding qualified back translators may also be a challenge. The requirement for a native speaker of the source language (SL), most frequently English, with relevant experience in the given therapeutic area, and high proficiency in the TL, may not be a problem for a major language. But it will be for some of the emerging or minority languages.

While for some companies back translation is an integral part of their procedure or an institutional review board requirement, some are trying to find the right situational approach, aiming for an intelligent application of back translations suiting their particular purpose. If forward translation may introduce errors, so can back translation. Back translations should not be seen as a replacement of the established QA process or a compensation for inferior original forward translations. As always, it pays to put maximum effort into ensuring quality upstream — with the forward translation.

When they are assigned to the work, translators should be made aware they are conducting a back translation so that they can adjust their approach accordingly and focus on an accurate, more literal reflection of the content in the SL and the general readability of the text. They should also be able to provide commentary to the back translation, which will greatly facilitate subsequent reconciliation.


Cognitive dissonance

about debriefing

Similar judgments may need to be made for PRO measures and the linguistic validation step of cognitive debriefing (pilot testing). The idea behind this step is to ensure that data collected from PROs can be comparable across various language groups used during trials, by verifying how the translated questionnaire is comprehensible to patients from the specific target population. This is achieved via structured linguistic validation interviews, during which a group of patients completes the translated questionnaires and then answers questions related to how well they understood the questionnaire.

In a way, this step represents an ideal combination of checking the quality of translated text with the ultimate end users from the linguistic as well as factual perspective. But it is also an area where a degree of latitude remains in real life. Recruitment of patients is one aspect of this. There has been and continues to be some healthy debate about whether patients used for linguistic validation need to be in a given disease state or not. Will they understand the questions and complete the questionnaire differently if they don’t have the “tactile” experience with the specific disease the given trial addresses?

Like with back translations, experience shows that the key aspect of the process is the review of cognitive debriefing results and the finalization of translations. The objective is to remove any terms, phrases or wording that could possibly be misunderstood by the patient and would ultimately distort the collected outcomes.

It is equally important at this stage to communicate with the developers of the original source questionnaire and ensure that the final translated measure continues to be a true reflection of the original. The final report, an integral component of PRO translations, needs to include an explanation of all key translation decisions made during the process as well as the methodology used.

The way translations are managed by pharmaceutical companies, large or small, is currently experiencing major development, and the industry as a whole is moving up the maturity ladder rapidly. At the same time, other industries can certainly learn from the hard-won experience the life sciences sector has in ensuring quality translations. Prevention is always better than a cure, and the same applies to optimizing the overall translation and quality process, regardless of the industry in question.