How the FAIR principles can bring value to clinical data

We have already described the two major sources of clinical data from clinical trials and routine clinical practice (i.e. real-world). The value of such clinical data would be greatly increased through FAIR Implementation which will likely include the application of interoperability standards such as CDISC, FHIR and vocabularies which are mandatory for submission to the regulators. FAIR data and metadata from clinical trials and real world sources are more likely to be ready for consumption by machine learning and semantic knowledge graphs which will support secondary reuse over much greater periods, as illustrated in figure 1. This would require the use of mappings to strong semantic standards at data capture time, such as proper use of SNOMED-CT or openEHR archetypes.

Open

Figure 1: Increasing the value of clinical data through FAIR Implementation, breaking up data silos and enhancing flexibility and reusability. The traditional paths (blue arrows) are with no implementation of FAIR. When FAIR is implemented the traditional paths become obsolete and (blue arrows crossed out). FAIR implementation is central to generation of FAIR linked data and metadata which supports the added value, indicated in green.

Realising value for clinical trial data

Pharmaceutical companies conduct clinical trials at great expense, primarily to demonstrate to the regulatory authorities that new treatments are efficacious and safe. Although data standards such as CDISC are mandatory for clinical trial submission to FDA, this does not make data FAIR (as will be shown in section 2) and in fact, could actually limit future reuse. However, if clinical data were FAIRified first, this would unlock much more value (see figure 1). Besides making clinical study data and metadata more Findable, Accessible and Reusable it would also benefit from much greater Interoperability. This semantic enrichment of clinical study data will enhance its value both within a company and outside, when it is shared with external parties, including public registries such as clinicaltrials.gov.

The dynamics of realising value can take account of expected timings. Within an organisation, the value is likely to be in the short to intermediate time scales. For the short term (1-2 years), the effectiveness and speed of electronic data capturing is improved by including the most relevant clinical vocabularies or ontologies for a particular clinical trial study. Interoperability through FAIR clinical vocabularies or ontologies could accelerate data integration and interpretation. Time is critical for this primary usage. In the medium term (2-3 years), it is possible to gain much more value from the clinical data beyond immediate primary usage. Findability, Accessibility and Reuse of clinical data will enable future secondary analysis.

This compares with the value expected for the medium to long term (3-5 years) for sharing data with external parties, such as business partners and public registries, like http://clinicaltrials.gov . FAIR implementation for clinical data will enhance its acquisition, sharing and integration as shown in Table 1.

Table 1: Summarising the expected value of data FAIRification.

Realising value from Real World Data

Here are some qualitative examples of how to leverage Real World Data (RWD), by integrating them with other data types (e.g. by unlocking the insights contained in genomic and phenotypic data), can generate value and help all key stakeholders in the healthcare ecosystem.

1. RWD can help to identify and satisfy unmet needs in Biopharmaceutical Research and Development, such as to inform research decisions; innovate/improve clinical trial design by using synthetic control arms; help defining inclusion-exclusion criteria and end-points; optimise site selection, accelerate recruitment; overall reducing time to market; monitor real world outcomes by quantifying unmet needs and understanding safety and efficacy profiles.

2. RWD accelerates the demonstration of economical value of treatment to payers; enables outcome-based pricing; makes it possible to achieve label expansion, by eliminating the need for randomised controlled trials. 

3. RWD analysis provides the clinical evidence, i.e. Real-World Evidence (RWE), regarding the usage and potential benefits or risks of a medical product. This will improve pharmacovigilance by monitoring real world usage and adverse effects; helping to create quickly risk / benefit overviews. This may strengthen the need for differentiation and inclusion to help highlight the benefits or risks for under-studied populations, or could be useful for investigations on rare diseases. 

Overall RWD can help to improve adherence supporting personalised approaches with engagement techniques such as Electronic Patient Reported Outcomes (e-PROs) and other digital interventions.