Electronic health records (EHRs) have gone mainstream over the past decade. The shift from paper to electronic records was primarily intended to ensure healthcare professionals have access to patient information but the digitisation of health data has also opened opportunities in clinical research. These opportunities could improve data quality, streamline processes and lead to better protocols. Yet, while early tests suggest EHRs can deliver on that promise, barriers to their use remain.
The idea of using EHRs to populate electronic case report forms and inform the design and execution of clinical trials dates back to when paper was the predominant technology. Back then, the vision of using EHR data in clinical research was held back by the fact that only a fraction of sites captured and held data electronically.
That barrier is no longer in place. By 2017, 86% of office-based physicians in the United States used an EHR system. Uptake at US hospitals is higher still, with 96% of sites using a certified EHR by 2015. The picture in Europe is similar. A 2016 survey of 15 European Union member states found 80% of primary care practices were using an EHR. That figure masks country-to-country variation but the overall trend is for healthcare systems to capture more health data than ever and store it in EHRs.
“With modern health systems, it’s possible to capture a lot more information about how medicines are taken by patients and the effects they have on patients. This really transforms the whole way that we do medicines research and development,” Brendan Barnes, director of IP and data protection at the European pharma trade group EFPIA, told health data institute i~HD in 2018.
The transformation promises to improve clinical research. One study found entering clinical research results directly into a system, rather than transcribing health records, cut the time taken to capture demographic information by 37% and shrunk the rate of data quality issues from 9% to 0%.
In other scenarios, data captured in routine healthcare and stored in EHRs could enable researchers to show how drugs work in the real world, identify populations with unmet needs and inform study designs.
The challenges to using EHR data
The transformation pointed to by the rise of EHRs is yet to happen. A 2017 survey of 193 biopharma and device companies and 56 CROs found 20% of organisations used EHRs, although 50% planned to have a strategy for the technology in place by 2020.
That low rate of uptake potentially reflects the presence of multiple barriers to the effective use of EHR data in clinical research. At the Leveraging EHR for Clinical Research Now! event, leaders from academia, companies including Amgen and AstraZeneca and the US Food and Drug Administration sought to identify the factors that are holding the field back.
Data quality emerged as a key issue, with the convened experts noting that co-morbidity data in EHRs may be good enough to identify and enroll patients in clinical trials but too inaccurate or incomplete to rely on to control for confounding bias in a comparative effectiveness study.
That example reflects a broader truth: data quality directly affects how EHRs can be used in research. The challenge for sponsors is to accurately characterise the quality of the data available and either design studies around its limitations or take steps to mitigate them. That challenge has been around for as long as EHRs and the industry is still grappling with it.
“Although data is there, there is a question of quality. What is the gap in quality? What can we do about that?” Nadir Ammour, global lead, clinical innovation and external partnership at Sanofi, told i~HD.
Some sponsors that identify EHR systems capable of supporting trials run into operational issues. The UK Medicines and Healthcare products Regulatory Agency (MHRA) has identified some of these issues in its assessments of whether the use of EHRs complies with good clinical practices.
In theory, EHRs can be used in a compliant manner, and European authorities have said the same of eSource direct data entry, but in practice their application to research has led to major failings.
MHRA has encountered EHR systems that prevent internal and external monitors from accessing the data, leading researchers to print out records.
This practice is more than just inefficient; it also results in the use of uncertified copies of source data. MHRA has found fault with the use of such uncertified copes for source data verification, particularly as the files can lack information held in the EHR.
Initiatives to overcome the challenges
The clinical research field has invested time and money to tackle the barriers to using EHRs in trials. From 2011 to 2016, the European Union EHR4CR project, which involved ten pharmaceutical companies, used a €16m (£14m) budget to develop a platform for using de-identified hospital EHR data. The system was designed to pull in data from multiple hospitals and other data sources, thereby helping sponsors to predict where to find patients eligible to participate in their studies.
EHR4CR gave rise to i~HD, which has sought to build on the project’s legacy through initiatives such as a European Network of Excellence for Hospitals designed to improve data quality. Advocates of such initiatives argue that improving data quality both enables hospitals to optimise their own operations and increases the value of their EHRs in research contexts.
The overarching challenge of improving EHR data quality is imposing but it may be possible to break it up into more manageable pieces. Ammour highlighted the need to establish methods of assessing quality shortcomings and create interventions designed to address the highest value data.
Other groups have overlapping objectives. One of the research initiatives at Health Data Research (HDR) UK is called ‘Unlocking the potential of EHRs to deliver better, cheaper, quicker trials’. Under the umbrella of that broad goal, HDR UK is running a project to understand how to address missing data in a standardised, scalable manner.
Other HDR UK projects are focused on topics including the use of EHR data to complement study results in regulatory applications.
There are now examples of companies using such approaches outside of pilot projects. Pfizer, for example, won an expanded FDA label for breast cancer drug Ibrance on the strength of data from sources including EHRs.
MHRA is also working to resolve the problems it encounters when inspecting studies that use EHRs. In October 2019, the UK regulator held a roundtable with EHR vendors to understand the functionality of existing systems and the status of efforts to add key features such as segregated access. Some EHR systems already have these features but users lack the knowledge to properly use them. One goal of the meeting was to understand whether system changes or better education are needed.
Making change happen
The level of public and private interest in resolving the barriers to the use of EHRs in clinical research suggests that ultimately solutions to the current challenges will be found. However, while in theory problems such as data quality are surmountable, in practice bringing all systems up to the required standard may be a hard, long process.
Even in the UK, which has a nationalised healthcare service, there are multiple vendors that provide EHR systems to healthcare facilities, making it harder to ensure every site has the features needed to support clinical research.
The rewards for healthcare systems that work through these challenges, and the downsides for those that do not, are potentially significant. That conclusion is implied in the results of early tests on EHR use and how seriously the subject is taken at groups including MHRA, which is focusing on the field in the belief it will help the UK be seen as ‘an attractive location of excellence’ for studies.
