Over the past few years, the European Medicines Agency (EMA) has been doing its best to involve industry, consumers, researchers and government bodies when defining and introducing its new clinical trial transparency architecture. This stakeholder involvement was essential to EMA Policy 0070 in terms of record keeping and public disclosure of clinical trials information (including results), while protecting data privacy.
Phase 1 of the policy, which came into force on 1 January 2015, has led to a lot of raised eyebrows. The general belief is that the legislation skews far too heavily towards transparency and raising awareness, and not enough towards protecting individual patient data privacy and recognising that clinical trial information may also be commercially valuable and warrant protection.
Phase II, which remains to be implemented, pertains to the publishing of individual patient data (IPD), and will pose an even tougher challenge for information protection.
According to Christopher Marshallsay, director and head of medical writing and public disclosure at Grünenthal, EMA misses the mark with 0070.
The policy, which anticipates EMA posting redacted clinical trial and submission-related documents on the internet, will have a huge effect on medical writing. As Marshallsay tells me from the company’s base in Aachen, Germany, it is well meant, but – along with other forms of public disclosure – may cause additional risk to IPD, and create considerable additional expense for every new trial.
“Perhaps the industry wasn’t as transparent as it should have been in the past,” Marshallsay says. “There were already trial registrations, trial status updates, and reports of aggregated trial results accessible to people with no restrictions, and now there is also proactive public disclosure of detailed clinical trial documentation via the policy.”
Anon and on and on
Under Phase I of the EMA Policy 0070, clinical trial documentation will be posted on the internet and will include comprehensive information and results, not just whatever a company wants to share, or the evidence it needs to substantiate product-label claims. Access will be subject to terms of use, but it is unsure how, or by whom, compliance will be policed. Posted trial documentation may include IPD, which could be difficult to protect, but which the marketing authorisation holder must try to anonymise.
Marshallsay thinks this won’t be as simple as it sounds. “Protecting IPD requires marketing authorisation holders to identify all IPD in the to be posted documents; to assess the risk of patient reidentification based on the found IPD (based on each data item alone and in combination with other data); to define an acceptable level of anonymisation based on this calculated risk, residual data utility and published data on IPD attack methods and adversaries; to calculate the residual data utility after anonymisation of the high risk IPD; to justify this strategy in a report; and to apply the proposed anonymisation (data masking) before finally submitting the proposed anonymised documentation to the EMA and following the EMA consultation procedure,” he explains. “This is a process requiring resources and awareness of
IPD attack methods. The cost-benefit ratio for posting anonymised clinical documentation on the internet is clearly skewed towards cost.”
Marshallsay believes the biggest challenge will be to protect patient information, especially when the data sets become publically available in Phase II of EMA Policy 0070. “This will be a quantum leap,” he predicts. “The real benefit is that it will enable analyses to be replicated, confirmed and extended, but, hopefully, with better controls to prevent abuse.
Marshallay describes the policy as “well-meant”, but fears the EMA has created a whole new risk in terms of patient-identification that could result in hefty fines being imposed. Indeed, under European law, failing to protect personal data can lead to fines of up to 5% of the global turnover of the product on trial.
New requirements
On top of Policy 0070, by 2018, EU regulations will also require that a trial’s protocol, investigator’s brochure, report, and summaries of clinical safety and efficacy from the clinical trial authorisation application are posted on the internet in a staggered manner. These documents must also be anonymised to protect IPD, and redaction to protect commercially confidential information (CCI). Trial sponsors will also be obliged to submit lay summaries of the clinical trial results for posting.
As Marshallsay sees it, these new requirements will require more work, larger teams and extra oversight: “We’ve been waiting for the technical logistics to be set up; you will have to submit for posting a short expert summary, and any lay summaries, for all trials you perform.”
Medical writers drafting clinical trial documentation will have to consider the needs of the new readers, as well as those of the staff who must anonymise the clinical trial documentation prior to posting on the internet, and the potential risk connected to IPD and CCI. Another challenge will be maintaining consistency across the clinical trial registries and the numerous clinical trial documents.
Given the added value and necessity for transparency, companies must carefully consider the content and structure of their clinical documentation; they will have to reduce the amount of CCI and IPD included, while maintaining full scientific utility.
“We want to share useful information without too much patient detail, and without giving away everything, or giving the negative impression that we’re hiding things.” Marshallsay says. This means that IPD in clinical documentation will have to be anonymised, or redacted, on an industrial scale.
Anonymisation is the next major step. “We need tools to go through thousands of pages of information to make a proposal,” explains Marshallsay. “They have to be approved and reviewed by EMA; uploading this information using tools is far easier, but you need a flux,” he says.
Another issue that could result from the new workload being placed on the pharma companies.
“This policy works to the advantage of big pharma, rather than smaller companies, which, for one or two trials, won’t have the infrastructure to identify and redact information in the 10,000- page reports.” When you have dozens of trials to run and report, you can establish specialised teams and exploit tools for the work, but such teams and tools may not be an option for smaller companies with fewer trials. This could result in inadequate protection of IPD, or delayed compliance with the EMA policy, and either could see companies suffering penalties.
The pitfalls
As of 2018, trial sponsors will have to provide lay (readable) summaries of the clinical trial results for posting. This sounds like a good thing, one that is really targeting the patient.
The summaries, which will be posted online, must be formulated in such a way that ‘lay’ persons are able to understand how a trial was performed, and what results were obtained.
“This will be difficult to do,” Marshallsay concludes. Writing a scientific report for trained experts is one thing – and medical writers are schooled in the technique – but producing an accessible version that retains all the necessary linguistic and numerical aspects, is quite another.
Elements may differ considerably between regions and readers – the lay English summaries will have to be translated into readable versions in every language used in a trial – and the summary must be completed within a six or 12-month reporting deadline.
While these new requirements are meant to increase transparency, a Marshallsay anticipates “banana-skins on the ground” as every new requirement creates a fresh challenge.
Marshallsay also expects difficulties in disseminating summaries in patient-sensitive ways. Patient centricity requires that patients are informed when the results of their trial are posted. However, simply mailing trial patients to inform them that a lay summary has been posted could cause distress to relatives if the trial patient has deceased since the trial. Trial patients may also have questions after reading lay summaries, so patient-centric companies must establish suitable contacts that are able to answer them in lay terms in the relevant language.
As Marshallsay says, the larger the amount of IPD made publically accessible, the easier it becomes to reidentify a trial patient. The more people with access to the data, the larger the number of possible adversaries who might want to reidentify a trial patient.
Although IPD posted under the EMA Policy 0070 is anonymised, as the technology develops, IPD that was robustly anonymised today will be reidentifiable tomorrow. Also, patients themselves may aid such reidentification by posting their news, for example, that they participated in a trial on X for sponsor Y, on social media. It’s as simple as someone logging reading this information and being able to join the dots.
Future of medical writing
Marshallsay advises his fellow medical writers to hold off on penning protocols until they know exactly what they want. “Identify what result you can expect, and the objective of the trial before you start,” he explains. “Be aware of the disclosure regulations and requirements. Even within the EU, there are still national specificities.”
Marshallsay urges writers to “think globally” when writing reports in order to avoid customising disclosed information by regions.
“We need to plan ahead and think up front,” he says. “Remember that these documents will be going public; so think intelligently about how you write protocols, and understand the registry needs and expectations.” He also urges those involved in writing the reports to respect tight timelines; think ahead and factor in those dates.
Marshallsay is keen to point out that the new public disclosure regulations and requirements are not all bad; in fact, he believes there will be real benefits in this level of openness: “The benefits of providing access to the trial documentation and data for research, for patients, for healthcare professionals, and the potential reduced patient burden, due to the avoidance of unwarranted repetition, are clear.”
Patients will be able to feel more included, trials will be more transparent and the industry as a whole could work better together.
Companies will also have the opportunity to profit from this new transparency. They can learn from each other and work together wherever appropriate; researchers can see what is happening in industry and vice-versa; and colleagues with the same interests can create networks and have valuable discussions.
“By putting this information out there, we are creating exciting new research possibilities,” concludes Marshallsay.