The Randomised Evaluation of COVid-19 Therapy (RECOVERY) trial was launched on 23 March 2020. Led by Sir Martin Landray, Professor of Medicine and Epidemiology, Oxford Population Health, and Research Director for HDR UK Oxford, and Sir Peter Horby, Professor of Emerging Infectious Diseases, Nuffield Department of Medicine, it’s identified four life-saving COVID-19 treatments to date.

These discoveries have changed clinical practice worldwide and been credited with saving hundreds of thousands, if not millions, of lives internationally. And, by harnessing the power of routine data, it’s helping to pave the way to a future for better, faster and more efficient clinical trials.

Marion Mafham, Associate Professor at Oxford Population Health and who leads on RECOVERY’S data linkage, explained, “When the pandemic began, there were no specific treatments available for those severely affected by COVID-19. The trial was set up very rapidly – in just nine days – to see if any existing therapies would be effective for patients hospitalised with the disease.

“For any clinical trial, it’s important to have access to robust, complete, high-quality data. For RECOVERY, the main challenge was having enough follow up information on the participants so we could compare their outcomes and figure out whether any of the treatments worked.

“Usually, this is done by asking hospital staff to collect data 28 days after the treatment has been given. But because the trial treatments were given for just 10 days, it meant patients wouldn’t necessarily still be in hospital, or the same hospital, at the 28-day follow-up point, making accurate data collection difficult.

“Staff were also incredibly busy, and we worried they wouldn’t have capacity to collect the complete data we needed to make rapid decisions on treatments.

“That prompted the decision to create a ‘hybrid’ data collection system. Onsite staff were still asked to collect data at 28 days about whether the patient was in hospital, discharged, or had died, but this data was linked with additional information from patient’s electronic health care records.

“These records contain live data about hospitalisations, underlying medical conditions, prescribed treatments, and more. Using them as an additional source of data relieved the pressure from frontline health care workers, meaning they could complete a much briefer form onsite and save precious time in the midst of the pandemic.

“The system also allowed us to find out extra information about the patients, like their ethnicity, and how this may influence any outcomes of the treatments. And, with the patient’s consent, we can continue to analyse their data in relation to the RECOVERY trial to find out what happens to them in the long-term.”

Source: Understanding Patient Data

In the words of Martin Landray, Co-Chief Investigator of the trial, “Making use of routinely collected NHS data has been key to the success of the RECOVERY trial.” But it raises the question as to why this powerful resource isn’t harnessed for more trials.

Marion added, “This kind of hybrid data collection isn’t suitable for all trials, but it could definitely be used more.

“The problem is, data linkages like this aren’t straight forward to set up. We were lucky at Oxford to have lots of in-house expertise, but most trial teams don’t have access to the knowledge of how to collect, use and interpret this kind of data.

“Another challenge is convincing people this approach is still robust. Clinical trials have been done in the same way for decades – using carefully collected research-specific data with a lot of source data verification. This is a process where trial data is collected, sent off, and cross-checked by someone external from the trial site to make sure the data has been entered correctly. It’s very laborious, but it’s tried and tested and has formed the basis of thousands of treatment approvals to date.

“Moving away from this system would be a major shift. But it’s a shift Health Data Research UK’s clinical trials programme is working to support.

“The programme provides training and resources to help researchers learn how to utilise routine clinical data, but it’s also looking to support studies that compare newer data methods with traditional ones. Work like this could help provide evidence to regulators, industry and funders that these methods are reliable.”

Matt Sydes, Professor of Clinical Trials and Methodology at the MRC Clinical Trials Unit at UCL and BHF Data Science Centre added, “One thing we need are more case studies to show this kind of innovative data collection can be done safely and robustly.

“There are a number of real challenges in using routinely collected clinical data at the moment, but there are plenty of perceived challenges too. What RECOVERY has done is demonstrate how this kind of data can be successfully linked and used in a trial to lead to clinically important patient impact. It’s a trial that has really helped to open people’s eyes to what’s possible in the UK.

“RECOVERY was efficient in other ways too. It used a multi-arm approach to ask several clinically important questions rather than a traditional approach that would have just tested one. And it added in new questions so they could be addressed at pace, rather than in later trials, or not at all. These multi-arm platform approaches are increasingly being used and should be seen as an additional weapon in the arsenal of modern trials.

“I hope we can start to see a mindset change so that instead of defaulting to traditional approaches, there is at least a conversation at the start of every trial about multi-arm approaches or using routine data, to bring new possibilities for long-term follow up, reduce the data collection burden on hospital staff, and more.”

Join our mailing list to stay up to date on our work to use health data for better, faster and more efficient clinical trials.