As people interact with the healthcare system throughout their lives, they leave a long trail of electronic data. These data end up in disparate databases housed in different places, whether it’s the hospital where they were born, the family doctor’s they visit regularly, or in a clinical trial site where they participated in a study. People would get better and more personalised care if these datasets were combined because it would give their doctors a bigger picture of their state of health. It would also help those that make decisions about the ‘worth’ of different treatments, because it would reveal more about the outcomes patients care about the most (as opposed to the clinical endpoints that researchers focus on in their scientific research). It would help us understand the economic value of different kinds of treatments and how much public money should be allocated to each. All of this is why, in 2016, IMI launched a programme called BD4BO to figure out new, better and ethical ways for researchers to unleash the collective power of healthcare datasets, with the ultimate aim of producing real-life positive results for patients. The programme was built around four different IMI projects, each centred on a particular disease. DO->IT was not a typical IMI research project but rather a coordinating ‘entity’ that was intended to help with two things: firstly, to coordinate the different activities of the four data-driven IMI projects that make up the BD4BO programme, raise awareness of their existence, and communicate to the scientific community their findings, results and educational materials. Secondly, DO->IT’s mission was to sort out thorny issues relating to informed consent forms, the documents that patients fill in to signal their agreement about who is allowed to use their data. All this was done at a time when new EU laws were impacting the way data privacy is treated. DO->IT spent two years building recommendations, tools and knowledge repositories and engaging with groups concerned about questions of real-world evidence in research, and, together with patients, came up with a new GDPR-friendly informed consent form template to make progress towards Europe-wide harmonisation.
DO->IT was responsible for coordinating the activities of the BD4BO ‘big data’ projects within IMI. The objective of the BD4BO projects is to answer some tricky questions about the practical ways of going about using data from diverse sources about real-life patient-reported outcomes, and using them to make healthcare better. Importantly, they also set out to figure out how to make their proposed new methods and standards stick.
Big data describes the practice of combining large datasets so that the numbers can be crunched to reveal patterns and associations. In healthcare, datasets from electronic health records and clinical studies can answer questions about what’s happening in our healthcare systems, whether or not these systems offer value for money, as well as what kind of outcomes really constitute a good result for patients who have to live with a given disease or condition. Getting access to these datasets requires collaboration across different stakeholder groups, including patients, health care providers, payers, authorities, academia, and industry.
But first, what are outcomes?
The BD4BO projects that DO->IT coordinated are working with specific diseases, each of which is studying the potential of data to help come to some conclusions about how to make life better for patients. The researchers are interested in real-world outcomes, which are best described as the results of treatment that patients care about – not blood test results about hormone or cholesterol levels, for example, but rather things like their ability to move around, pain levels, and general quality of life.
Figuring out the outcomes that patients care about the most is crucial; public money, in the form of research funding or healthcare insurance coverage, should be spent on treatments that offer the most benefits to people. This kind of value-based approach ensures that the healthcare system makes the right choices with regards to public funds when deciding between different treatment alternatives.
Biggest issues and core outcome sets
One of DO->IT’s missions was to find out what the data-driven projects considered to be their most pressing big-data challenges. Five main problematic issues emerged, the most pressing of which was the lack of agreement about what big data is really worth to those who are responsible for making decisions. It turns out that their views were very varied. Lack of interoperability between the different systems that house and format the data was a big issue, as was the lack of ‘sharing culture’ (the IMI EHDEN project is currently working on this question of interoperability between different systems that house and format data, including addressing the lack of ‘haring culture and legal issues around the data sharing. See the EHDEN project factsheet for more information.)
Significant legal issues kept cropping up, and the problem of what can be done to make the projects' outcomes permanent was also tackled. The DO->IT project produced a document that outlines the challenges but also provides guidance and direction for those in the BD4BO programme, proposing potential activities and tactics that might mitigate risks and take advantage of opportunities. For example, with regards to lack of interoperability, they concluded that consistent use of ‘core outcome sets’ - meaning, priority patient-reported outcomes - is key for comparing interventions, comparing across studies, and for linking data sources.
DO->IT conducted a case study review to see if these core outcome sets can be replicated in several different disease areas. They also drew up a toolkit, or a practical guide for BD4BO projects that would help them to identify, select and measure the core outcome sets for their own disease area. This was done together with patient focus groups and a panel of representatives from data protection authorities and ethics committees from various countries.
The toolkit describes the six steps required for developing a core outcome set, from scoping to dissemination, with a focus on ensuring that all stages stakeholders are included. It is intended to stimulate standardisation of outcomes in Europe more widely. They published a webinar recording that introduces these core outcome sets and their role in harmonising outcomes generated in real-world settings; the webinar video includes explanations of the experiences from the different BD4BO projects including, for example, on involving patient perspectives on the standardisation of outcomes.
DO->IT also left behind the ‘Knowledge Hub’ - an online space for BD4BO projects to share their work with each other and the wider big data community, as well as a communications toolkit that includes patient stories, presentations, a video and a series of webinars on outcomes actually are. DO->IT spread awareness of the BD4BO project through a website, social media channels, newsletters and infographics, helping to draw more followers in the research community so that they can benefit from the knowledge being generated in the BD4BO projects.
Data privacy and consent forms
The other aspect of the DO->IT mission was to address data privacy issues that arise in the use of patient informed consent forms. An informed consent form is the agreement which every patient has to give before he or she enters a clinical trial that stipulates what can be done with his or her data. The forms cover the use of both data and human samples. DO->IT sought to create transparency in the way data privacy laws are applied, and develop uniform standards and guidance about data privacy issues. There has hitherto been a lot of disagreement as to what form this form should take, but the work of the project demonstrated that there is a clear interest in developing new models.
New models of consent could really bring a lot of advantages to research, considering the emerging data and analysis environment. DO->IT carried out an empirical investigation into current informed consent practice in order to take stock of existing and emerging ways of doing things. They reviewed scientific journal articles and findings from a survey that probed different stakeholders for their experiences and opinions of informed consent procedures; they produced a report that highlighted some of the main new kinds of consent models tailored to an ever-evolving research environment.
They then created a proposal for what EU-wide and even globally harmonised data protection clauses could look like under the new General Data Protection Regulation (GDPR), which came into effect around the same time as the work of DO->IT was getting started. Different interpretations of the GDPR rules across countries makes it difficult to develop universally accepted standards. DO->IT discussed their proposed clauses with patient focus groups, representatives of data protection authorities and ethics committees from various countries.
They came up with an informed consent form template, meant to cover all of the information required by the GDPR rules, that addresses the use of study data collected in clinical trials for future research projects including biobanks. They also created an accompanying document that explains the rationale and background context on the complex choices behind the form, as well as training materials with key questions and answers on data privacy requirements and concerns.
Peer-reviewed journal articles and research reports continue to disseminate and raise awareness of the activities of the DO-IT project. These publications are helping further spread the project findings, raise awareness of the BD4BO agenda, and contribute to academic thinking on big data. Coordination activities have partly been taken over by the IMI BD4BO project EHDEN, particularly in regard to data definition, standards and collection and thanks to additional funding from Bayer, the Knowledge Hub has been kept accessible and updated beyond the project timeframe.
Recommendations by the DO->IT project on metrics and data categories for health outcome data that includes real-world evidence have been incorporated into standard procedures in a number of pharmaceutical companies. The informed consent templates have set an international standard for future clinical trials in line with GDPR, including the option for future (secondary) use of data, thus supporting data sharing in the big data era. The IMI projects IDEA-FAST and PIONEER have incorporated the DO->IT best practice guidelines on ICFs, and the template forms the basis of the updated ICF model recommended by German Association of Ethics Committees. It was also recommended by the Spanish pharma association and has fed insights into the pan-European Code of Conduct harmonisation initiative driven by BBMRI-ERIC, the European research infrastructure for biobanking.