The last decade has seen an explosion in the volume and types of health data. Electronic healthcare record systems, with structured codes and labels for different diseases and events, are now commonplace. Yet concerns remain about the quality, data privacy and transparency of these systems. Furthermore, there is still no consensus on how to apply the ‘FAIR’ (findable, accessible, interoperable and reusable) principles to structured healthcare data. And while checklists to address some of these issues do exist, they do not set minimum standards and are often used as a box ticking exercise to facilitate publication in journals.
The COVID-19 pandemic illustrated how powerful healthcare data can be; information on the relationship between COVID-19 and cardiovascular disease came from research that combined primary care data, hospital data, death records, and COVID-19 testing in over 54 million people. However, the pandemic also highlighted the challenges to sharing data in close to real time that could have directed care and helped design clinical trials.
To address these issues, Innovative Medicines Initiative (IMI) project BigData@Heart teamed up with the European Society of Cardiology (ESC) to develop ‘pragmatic advice’ on the use of structured healthcare data in clinical trials and observational research. The goal was to deliver something that would be applicable across different disease areas, and meet the needs and expectations of all stakeholders, including the general public.
To do this, they first held meetings involving all key stakeholders, including regulators, government agencies, medical journals, patient organisations, the pharmaceutical industry, payers, academic institutions, and professional societies. Consensus positions were further developed and agreed on via virtual work, and another meeting was held to finalise the outputs.
The result, dubbed the CODE-EHR framework, aims to improve the quality of studies using structured healthcare data, and provide confidence in the results of these studies for use in clinical decision-making.
The framework focuses on five key areas:
- Dataset construction and linkage (to provide an understanding of how the data were identified and used);
- Data fit for purpose (to ensure transparency on the coding systems used);
- Disease outcomes and definitions (detailing how conditions and outcome events were defined);
- Analysis (detailing how outcome events were analysed);
- Ethics and governance (covering processes for consent, privacy, and patient and public involvement).
For each area, the framework sets out both minimum and preferred standards, and details the information that researchers should provide when writing up their study.
‘With the support of patients and the public, routinely collected healthcare information provides an exciting opportunity to answer important clinical questions in populations representative of our communities,’ said author Professor Dipak Kotecha of the University of Birmingham and University Hospitals Birmingham NHS Foundation Trust, UK. ‘Our ability to apply findings from studies that use these data sources is critically dependent on transparency at every stage. This international framework will enable robust and effective use of healthcare data for clinical research and provide those working in this field with guidance on how to design better studies for maximal benefit to patient care.’
Looking to the future, the group plans to evaluate the framework in two years’ time, and adapt it as needed to any new developments in this rapidly-evolving field.
BigData@Heart is supported by the Innovative Medicines Initiative, a partnership between the European Union and the European pharmaceutical industry.