It appears to be a given that the controlled clinical trial is no longer the exclusive provider of data to support the authorisation, marketing and reimbursement of a medicinal product. Even the European Medicine Agency (EMA) has taken a position accepting that there is value and merit in real-world evidence, and in a recent expert group report, the EMA and the EC point out the uses of such data in supporting pharmacovigilance, comparative efficacy and pricing negotiations for new treatments.
Perhaps it might have been a good idea for them to include the notion that taking a look back at existing medicines could be an opportunity to evaluate how they fit in the current healthcare delivery systems. Many well-established healthcare routines are not based on the best evidence, but on the belief that it is considered best practice.
Accepting that controlled trials are still the gold standard for proof of efficacy and tolerability, whatever else can be discovered/demonstrated is all in the data. Real-world evidence is healthcare data that does not originate in controlled studies and comes from a variety of sources including electronic healthcare records, disease/treatment registries, health insurance data and web/social media. There are many attractions including the potential cost saving compared to clinical trials, and the possibility that the numbers of patients involved may be comparatively huge for a marketed medicine. However, it is important to note that some of these methods will create problems over the governance and protection of data, whether there is consent to use pre-existing data for potentially commercial reasons, and if there is a commercial reason, what is the perceived worth of the data to the buyer and what did it cost the owner to create the data?
If a real-world evidence question is to be based on existing data such as existing patient records, then the study had to be designed around what already exists: it is not possible to synthesise the contents of information gaps. The gaps can be qualitative or timeline longitudinal – if data was not collected at a particular timepoint under evaluation, then a form of potential bias is introduced, which means that poor or erratic record-keeping confounds the data set.
Patient registries – multi-purpose opportunities
A significant amount of attention has been given to patient registries. These are well-loved because they offer clinicians the opportunity to have an overview of their management of patients with a particular condition, and they permit a bigger picture to be created which can set benchmarks for the standards of management, and they can provide a real-world evidence platform. These have become the proposed platforms for studies that may be agreed with competent authorities, but are a potential headache for sponsors because existing disease and treatment registries may not collect the information that is being sought. When disease registries are created, they are typically conceived with the idea of setting national or international benchmarks for outcomes of treatment. Rarely are they created with the flexibility to accept bolt-on activities for unplanned studies. This has sparked the creation of new product/disease specific registries sponsored by commercial interests, but there are many frustrations potentially in store when this occurs. If similar registries already exist, it is unlikely that clinicians will be happy to enter data twice into two registries in different formats (unless incentivised). A key factor in the delivery of healthcare is inequality and the fact that what is standard in one geographic region may not be universally true. The frequency of collection of data in registries may additionally cause frustration: pharma is used for the regular repeated collection of data over a defined time period, and this may not be the way it has operated historically for a particular registry.
National Electronic Health Records – an alternative source
Several publications of RWE studies have used the strengths of national electronic health records (EHR). These have operated best when there is a consistent, reliable collection of outcome data relevant to a therapeutic outcome, and a single socialised healthcare provision model, such as in the United Kingdom. To enhance this data, additional information may be sought prospectively, creating the so-called hybrid study, such as the Salford Lung Study. Once again, the completeness and accuracy of the EHR is critical, and the consistency of recording and collection is paramount, making the data potentially as complex to collect as in a controlled clinical study. For example it is typical for a patient in primary care not to see the same family doctor on two occasions, and because the primary intention of the EHR is not to be the basis of collection of outcomes study data, there may be inconsistencies, errors, omissions and even confusion about whose data was collected on any one occasion. I know for a fact that my own primary care record is riddled with transcribing errors, and because of the way the records are kept, they cannot be corrected, but rather can only be the subject of an explanatory note.
Social media – a novel approach for the future
Still in its infancy, and a long way from current utility in regulatory and fiscal decision-making is a growing interest in how patient/user-generated information that interfaces with social media on the Internet may be used. Almost as a form of group therapy, people in contact with illness are keen to share their good and bad experiences, and the relevance of this data in relation to other, more conventional health records cannot be underestimated. Information that may never come the way of a doctor, eg when self-treatment is used, fits comfortably into the spectrum of what may become real-world evidence. At its most simple, a statement ‘It did not work for me’ offers a benefit/risk comment that might not otherwise be documented.
A number of methods are currently under investigation for the collection of real-world evidence using online presence and social media. These may be as diverse as wearable/self-operated technology, which uploads information, to the completion of self-rating scales online using typical connected devices such as a smartphone or tablet. This latter method has been successfully used for diabetes, and by Orbital Media in the collection of information which helps update legacy information from a very old and no-longer relevant clinical trial of treatment of nappy rash and also provided an up-to-date demographic of product users. Unlike clinical trials, these RWE studies cover the real user population rather than a prospectively selected group, and the data is mostly retrospective, because the interventions have already been utilised. A significant advantage of using social media platforms is the speed of data collection, enabling study completion in weeks instead of years, and access to big numbers of participants, making study results more credible and more likely to represent the whole population. The proportion of the population engaging with social media via some device drives the big numbers, and generates real-world data that can be used to support efficacy claims, safety oversight and pricing negotiations. It is easy to see that it may also be incorporated into regulatory documents in the near future.
Whatever its source, real-world evidence is more representative of the outcome of healthcare interventions in the real world, where there is a wider variety of users, compared to regulatory clinical studies, no patient exclusions and no critical attention to the pattern of usage of products. In addition such studies, in support of reimbursement, may answer the simple question of how well the medicine worked for the individual. The question: ‘Did it make them better?’ is rarely asked in clinical trials.