As the number and complexity of clinical trials increase, participants are becoming an increasingly precious resource, necessitating a change in the research toolkit.
There has never been a greater need for clinical trial participants. Developments like precision medicine, and the increasing number of agents being developed for rare and orphan diseases are saving lives – but all need clinical testing in specific and potentially hard to recruit patient populations.
By design, the key to any useful clinical trial is to (a) recruit the right number of (b) the right kind of participants and (c) for them to stay the course and not drop out.
The solutions that follow leverage technology to overcome the inherent difficulty of recruiting niche patient populations and keeping them actively engaged in the clinical trial.
The starting point for any trial is ‘N’. Statical analysis determines ‘N’ – the number required to achieve sufficient ‘power’ to test a hypothesis.
For instance, estimating the magnitude of the effect it is expected a drug will have on, say, planned outcome measures, and then applying standard approaches to work out how many people you need to show the anticipated effect with confidence.
There is a lot to consider. Drop-out rates need to be factored in. The UK National Institute for Health Research (NIHR) in 2020 put the average rate of participants dropping out of clinical trials in the UK at around 15%, based on data from over 1,100 clinical trials.
And a further factor is the number of patients who are likely to have changes in their condition meaning they must be excluded from the trial.
The number of participants needed is also determined by the likelihood of each responding. A chunky potential cohort can be whittled down by all these factors, and more.
What, then, are some potential solutions?
Electronic health records
Integration of electronic health record (EHR) data into participant recruitment systems helps to recruit the most relevant participants, on the basis of real-world patient data, thereby minimizing N and the inherent difficulty to recruit.
Much like a US-style pre-trial jury selection where the prosecution and defence will select jurors most likely to deliver a fair verdict, the ability to be more selective in recruitment means N can be reduced.
The overall effect of an agent will be more pronounced in a more relevant participant base, with the right disease characteristics and biomarkers.
This is leading to the rise of software platforms like the popular EDC (Electronic Data Capture) which integrates EHR data into clinical trial recruitment, allowing searchers to be more specific in their recruitment criteria.
Clinical trial recruitment platforms which use machine learning algorithms and natural language processing to identify potential patients based on EHR data are now available to automate patient screening and recruitment, and reducing the time and effort required to identify eligible patients.
Exchange of patient data
Health Information Exchange (HIE) platforms are also increasingly important.
HIEs provide a way to securely exchange patient data between different providers, integrating data from multiple sources and allowing researchers to access a larger pool of potential study participants.
Given the lack of common APIs (application programming interfaces – the way software communicates with other software) between providers, information sharing has always been a challenge both in research and clinical practice. This is now changing.
Deploying AI
Once you have all this additional data, of course, it needs to be sifted through, a task for which AI is remarkably well suited.
AI can analyse vast amounts of data, at speed, and identify participants who meet the eligibility criteria for a trial, otherwise a time consuming semi-/manual task.
When used to effectively match potential participants to trials for which they are a good match, the quality of the trial increases – as does the likelihood of patients remaining with the trial, notwithstanding that EHR data is seldom perfect at the point of data entry.
In the future, AI will be able to help establish eligibility criteria that will optimise the magnitude of the effect of a certain agent, on the basis of similar agents.
External control arms
Using external control arms, i.e. using existing patient data from other clinical trials or the real world as a comparison, is another good solution.
This has the bonus of allowing all participants to receive active treatment – a boon where patient recruitment is difficult or the consequences of not pursuing active treatment would be very deleterious, as in chemotherapy.
Oncology specialist Celsion is exploring this approach with New-York-headquartered Medidata.
Remote monitoring
Huge strides have been made in recent years in remote monitoring technology, and (partially) decentralised trial designs are becoming more common.
This more closely resembles how an agent will perform in the real-world. Intensive monitoring by experts and use of state-of-the-art equipment does not mirror everyday clinical practice.
An increasing number of drug approval applications incorporate elements of ongoing community monitoring.
While this reduces the burden on patients associated with participation, however, it carries the risk that the data will be of lower quality when compared against in-person outcome monitoring.
The challenge is to replicate the capabilities of a research centre spread throughout the cohort’s homes – and not every trial is suitable.
World at our fingertips
What of the future? Virtual trials make the reach of clinical trials potentially global and data only ever a few keystrokes away.
Software like eClinical’s elluminate Operational Insights is designed to improve the oversight and management of digital dataflows by combining data from disparate data sources taking into consideration agile or hybrid clinical trials where data collection may differ between sites or countries.
The utopia of a global network of data all being pooled and shared for the betterment of humanity may seem a world away, but the first steps have been taken and that world is looking smaller – and more connected – every day.
Dr Leonid Shapiro is Managing Partner at Candesic, Floris Wentholt is Principal at Candesic and Dr Joe Taylor is Senior Advisor at Candesic
