Dr Richard Emsley
MRC fellow with a Career Development Award in Biostatistics at the University of Manchester.
Dr Richard Emsley develops new ways to design and analyse clinical trials to determine how treatments work and has had the opportunity to visit Harvard University as part of his fellowship.
It wasn’t until the final semester of his undergraduate degree in mathematics and statistics at the University of Manchester that a module in medical statistics piqued Dr Richard Emsley’s interest in his future career. “I found it to be by far the most interesting thing in my degree,” he says. Richard later applied to stay at the university’s Health Methodology Research Group for an MRC PhD studentship in biostatistics and has remained there ever since.
Unpicking clinical trials
Richard’s work centres on finding better ways to determine how treatments work - or why they don’t. Unravelling the exact mechanisms by which treatments work can allow them to be tailored to particular patients, he says. “Equally, if a trial shows that a new treatment isn't effective, we can explore why there is a negative finding - is it because the treatment didn't work via the mechanism suggested?” asks Richard.
He says that lots of clinical trials don’t take into account ‘hidden’ variables that could be influencing the results.“One of the key things is mediation: does the treatment act through a particular mechanism and how can you assess that?” he asks. For example, a form of cognitive behavioural therapy (CBT) to help people with psychosis should, in theory, reduce symptoms such as delusions by reducing the extent to which people jump to conclusions. So testing this idea by providing people with CBT and then assessing their level of delusions seems reasonable.
But standard methods don’t, for example, take into account that there could be other hidden factors that affect both jumping to conclusions and delusions independent of the CBT. This means researchers can’t tell whether CBT reduces delusions directly, via reducing someone’s tendency to jump to conclusions, or through another mechanism.
Richard and his colleagues try to get involved in trials at the earliest possible stage as co-applicants or trial statisticians. “That way we can embed these ideas in the concept of the trial. It’s much better if you know in advance that you want to do this kind of analysis.”
Exchanging ideas
One of the benefits of his MRC fellowship is that it gives Richard the freedom to get involved in other research in his department - much of it MRC-funded. “One of the great things that the MRC did was to invest in methodological research and we’ve been beneficiaries of that,” he says.
The training element of the fellowship is also key. “It gives you the protected time to say ‘if I want to make a contribution in this area, I need to go and learn about this’,” says Richard.
As part of his fellowship, in 2011 he spent four months at the Harvard School of Public Health. “Without the fellowship that opportunity wouldn’t have been available to me. It was a tremendous learning experience and hopefully will allow a collaborative opportunity going forward.” “I learnt about their latest methods and research. You can learn from the literature, but it’s much more effective to talk to people.”
Communicating results
Richard and his colleagues disseminate their methods by traditional routes, but they are also keen to ensure that they deliver their statistical techniques into the hands of those who need them. “It’s no use just publishing methods in statistics journals where they won’t be read by people who might be applying them,” says Richard.
Part of this means updating existing software packages so that everyone who uses them has access to the latest possible methods, as well as alerting users to the hidden assumptions in many trials. “We also arrange free workshops where we disseminate the ideas that we’ve been developing, not necessarily in a technical way, but to reach researchers who might want to apply the methods,” says Richard.
Psychology researchers are probably the main audience for this, he adds, but the techniques could be applied in surprising areas. For example, scientists interested in the use of biomarkers - substances used to measure biological processes, including disease - might be interested in Richard’s methods. “A lot of the of the statistical criteria for judging whether something is a surrogate outcome [biomarker] is almost the same as judging whether there was mediation,” he says.
And for the future? Richard hopes that a permanent academic post is on the horizon and intends to continue his career in Manchester. After all, he says, “Once you’re in a good place, why move?”
Find out more about the MRC’s support for research training and careers.
Read more about the MRC and methodology research.
Published January 2012