Menu Close

School of Health and Related Research (ScHARR)(Unversity of Sheffield)

By Tom Foley.

Dr Clare Relton, Senior Research Fellow, University of Sheffield


How do you define a Learning Health System?

One that learns from the healthcare it provides, helping the healthcare it offers to continuously improve. 

What work your group is doing on Learning Health Systems?

Professor Jon Nicholl, myself and other colleagues have been rethinking how randomised controlled trials are designed and conducted – focussing on improving the efficiency, speed, ethics and cost of pragmatic trials testing the effectiveness of interventions to improve health.

We piloted an innovative design (cohort multiple RCT design), which was published in the BMJ in 2010. Since then interest in using this approach has grown in the UK, and abroad – particularly Canada and the Netherlands. (In parallel, triallists in Sweden and Canada are starting to use registries to conduct multiple RCTs). This approach we loosely describe as Trials within Cohorts (and registeries) (TwiCs).

The essence of this approach is that you take groups of people who have already provided some health related information, observe their outcomes over time (either indirectly or directly), and then within this cohort/ registry/ group multiple trials are embedded/ conducted.

At the point of randomisation, only those randomly selected to something other than routine care are given information about the intervention being tested. Those in the usual care arm are not told about something which they wont then get.

The advantage is this approach is that recruitment is FAST (from your cohort/ registry/ group), you can conduct multiple trials (with usual care comparators), alongside each other or at different times, you can collect long as well as short term outcomes, trial populations are much more representative than they normally are, each trial is cheaper, the results of each trial can easily be compared with each other, and there are very strong arguments that the way information is provided is more ethical than the current ‘full information up front to everyone’ approach.

We have an international network of triallists (and ethicists) using the design and planning to use the design and funding boards (including, NIHR, CIHR) like the approach.

Scalability? We believe that the potential of this approach is huge. There are many naturally occurring cohorts/ registers – i.e. groups of people registered with hospitals, GP practices, local authorities, businesses.

What do you see as the big opportunities?

Using these groups (and data linkage opportunities)

What can we learn from other industries?

A lot – health research has become a deep silo of entrenched behaviours, rules and myths – that need busting! For example, it is a myth that everyone must be told everything before randomisation (ICH GCP guidelines state that information must be ‘relevant’ – not full). We now have ethics board approval for this design in 10 different countries and recently held an international
symposium with bioethicists from the US (NIH) and UK
, health regulators (HRA) and triallists.

What gaps need to be filled?

The next steps are to help make the UK trials community more aware of and more confident in the use of this approach and to support a number of large naturally occurring groups to pilot its use.