Professor Duncan Young, clinical advisor for NIHR Efficacy and Mechanism Evaluation (EME) programme

Welcome to the future - the push to get more out of health research


Professor Duncan Young, clinical advisor for NIHR Efficacy and Mechanism Evaluation (EME) programme

Date: 21 July 2017


With austerity in public funding, an ageing population requiring more evaluated treatments, and the productivity crisis afflicting the pharmaceutical industry the need for efficient studies has never been greater. These challenges are making the timely translation of promising, newly-discovered treatments into clinical practice a priority, driving policy in both academic and commercial biomedical research.

Clinical evaluations of the efficacy and effectiveness of new treatments are an increasingly expensive and time-consuming part of this “translational pathway”. As a major public funder of this phase of research the NIHR is very interested in efficient study designs that allow clinical research to be conducted more rapidly, or at lower cost, while maximising the delivery of robust data to guide NHS or patient decisions.

The Health Technology Assessment (HTA) programme has previously put out commissioning calls for efficient study designs, in 2014 and 2016. Both attracted large numbers of applications and had high funding success rates, suggesting the clinical trials community in the UK is both ready and able to address the challenges of designing novel, streamlined, and cost-efficient studies.

Using routine patient data

All efficient study designs make best use of one or more of time, data or patient populations to achieve cost-effectiveness and the maximum increment in knowledge. Patient recruitment rates and follow-up time largely determine study duration, so this is an obvious area to seek efficiency. Trials that recruit rapidly tend to cost less per patient studied and produce timely results.

In general the techniques to maximise recruitment after patients have been identified are well understood, so the challenge is to find eligible patients quickly and at low cost. Routinely collected electronic healthcare data, disease registries established for surveillance, audit or research, or patients identified for other research studies can all provide lists of potentially eligible patients to approach. Often the records also contain enough detail to allow checks on inclusion and exclusion criteria or define sub-groups of interest.

Routinely-held electronic data can also be used to determine outcomes in some studies. In the past this was largely limited to mortality but now national hospital episode data and national audit databases such as the Renal Registry, cardiac surgical databases, or the National Joint Registry and many other data sources are available.

Novel study design

Whilst the two arm parallel randomised controlled study is likely to remain the optimal study design for evaluating most treatments, there are other more efficient designs that can be used in some circumstances. Where the study primary outcome is determined soon after the intervention, a sequential design can be used which essentially evaluates the intervention repeatedly as data accumulate. This may result in an answer earlier than a parallel design, though at the cost of an uncertainty around sample size.

Adaptive designs allow knowledge gained in an early phase of a study to determine the process used in later phases, such as a play-the-winner design where the randomisation ratio to treatment and control is varied depending on emerging knowledge about a treatment. Factorial designs can answer two research questions within one overall trial. Umbrella, bucket and multi-arm multi-stage (MAMS) designs effectively share the fixed costs of trial management across multiple smaller studies and allow resources to be diverted towards promising treatments and away from those showing less potential benefit.

Studies of treatments for rare diseases may be difficult to design because the available patient population is too small for standard clinical trial designs. Obtaining a large enough population may require a major international effort. In these cases cross-over, n-of-1 or other designs conducted on a more limited scale may be appropriate. Non-randomised evaluations where bias can be reduced to acceptable levels may be needed.

A standard two arm randomised controlled study may simply not be feasible for some treatments. If the treatment is widely used in practice already, there may be insufficient equipoise to run a study. If the treatment involves a large capital expenditure a randomised study may be too expensive for the knowledge gained. In these cases, if adequate records are available, a non-randomised retrospective comparison may be the only option. Techniques such as case-matching exist to minimise the effects of known confounders, though the problem of “unknown unknowns” remains.

The knowledge gained in a study is communicated as study outputs such as papers. Efficient studies can maximise the outputs, by carefully analysing the results, by combining data with other studies to decrease uncertainties, and by reducing waste by using both the results and data as starting points for other studies.

Professor Duncan Young is clinical advisor for NIHR's Efficacy and Mechanism Evaluation (EME) programme.

The views and opinions expressed in this blog are those of the authors and do not necessarily reflect those of the NIHR, NHS or the Department of Health.
  • Summary:
    Duncan Young explores the need to develop new health research and efficient studies against the backdrop of public funding austerity, an ageing population and a productivity crisis in the pharmaceutical industry.
  • Year:
  • Author:

    Professor Duncan Young, clinical advisor for NIHR Efficacy and Mechanism Evaluation (EME) programme