Why clinical trials need data utility comparisons of healthcare systems data

02 May 2024

In a new paper published in Contemporary Clinical Trials, researchers from the MRC Clinical Trials Unit at UCL and colleagues around the UK explore how data utility comparison studies (DUCkS) could help clinical trials run more efficiently by making better use of healthcare systems data.

Clinical trials require extensive data collection about trial participants. Much of this data is already collected and stored, elsewhere, during routine interactions with healthcare systems, such as the NHS in the UK.

Re-using data which is already collected by healthcare systems could streamline the conduct of clinical trials by avoiding duplication of data collection. This would also lessen the immense pressures placed on public health staff by reducing the burden of research activities.

Furthermore, healthcare systems data could improve the quality of trial data: for example, by ensuring that medical events such as hospitalisations are captured even when they occur in hospitals that are not taking part in the trial.

However, several obstacles stand in the way of widespread of use healthcare systems data in trials. One challenge is determining how useful and appropriate healthcare systems datasets are for particular aspects of a trial, i.e. the data utility.

This is where DUCkS come in. These studies compare the relevant data collected through clinical trials to data acquired from healthcare systems, to see how well they match up. This helps researchers understand where trial-specific data collection could be replaced with healthcare systems data, while still accurately capturing what is needed for trial outcomes.

But despite the importance of DUCkS, there are few published examples using UK healthcare systems data, and few known to be ongoing. Those that do exist use varying methods to quantify data utility.

More DUCkS are still needed to build a wider body of evidence for the use of healthcare systems data in clinical trials. Using knowledge and experience gained from running these studies, a team led by researchers at the MRC Clinical Trials Unit at UCL has set out considerations for the planning, development, data collection, analysis and reporting of DUCkS.

Firstly, researchers must ensure trial participants are correctly linked to the necessary healthcare systems, in line with their consent and with the appropriate ethical approvals. It is also important that researchers take snapshots of both datasets as close together timewise as possible, ensuring a like-for-like comparison. Thirdly, trial teams should carefully consider whether to conduct DUCkS at the level of individual participant data, at the level of the whole trial, or both.

The MRC CTU at UCL, along with partners from the Centre for Trials Research in Cardiff and Oxford Big Data Institute, also recently ran a workshop to help clinical trial teams run their own DUCkS. The workshop took place online on Thursday 25th April 2024.

Further information: