Where medicine meets mathematics
As a child, Rory Collins was fascinated by maths. Today, Professor Sir Rory Collins, co-director of the Nuffield Department of Population Health (NDPH), and his colleagues are using statistics and data to answer some of the biggest challenges in global health. But this journey wasn’t entirely obvious.
“I enjoyed maths a lot at school and found it straightforward at A level,” he says. “When I looked into the kind of jobs I could do with a degree in mathematics, the careers officer told me that my best option was to become an actuary. But I thought that sounded quite boring.”
Instead, Collins turned to medicine and spent two years as an undergraduate at St Thomas’ Hospital Medical School in London. The opportunity to return to his mathematical first love came in the third year, when students were given the option of studying a different subject.
“Some people chose things like anatomy, physiology, biochemistry, or psychology, but I thought I could spend the year doing more maths,” he explains. “The medical school was horrified! They said, ‘What does mathematics have to do with medicine?’”
In search of an alternative option, Collins applied to universities in the US, which he thought might have more academic flexibility.
“I got a book on US universities, turned to the index and picked 36 with a pin – it was my first introduction to random sampling,” he laughs.
One of them turned out to be George Washington University in Washington DC, where Dr Jerome Cornfield had established a biostatistics centre to bring maths and medicine together. Collins was able to take up a place there thanks to a bursary from the Medical Research Council, topping up his funding by working one day a week with Cornfield on the centre’s clinical trials.
From maths to medicine and back again
All too soon it was time for Collins to return home to complete his medical training. But the lure of applying mathematics to medical problems was too strong to resist.
He’d become aware of the work of acclaimed epidemiologist Richard Doll, Regius Professor of Medicine at Oxford University, along with his collaborator – a promising young researcher called Richard Peto (now co-director of the NDPH). Together they were investigating the health risks of tobacco smoking by analysing a major long-term study of British doctors, which first began back in 1951.
“I would describe Richard [Peto] as the brains of modern epidemiology,” Collins says. “I came to work with him for a year in 1981, and that year has just never ended. I keep on learning from him – he’s always asking ‘so, what’s the evidence for this?’ It’s all about quantifying risk – figuring out what is important and what is relatively unimportant. And to do that you need good study design and big enough studies to work it out.”
This attitude of ‘go big or go home’ underpins the work of the NDPH today: in order to see whether a treatment or risk factor is really having an effect in a clinical trial or observational study, it’s important to have enough participants to separate true effects from random fluctuations. But it wasn’t always like this.
“When I first came into the field, the focus of epidemiologists was about ensuring they avoided bias. You had a feeling that your card to practice epidemiology would be torn up if you didn't take bias seriously, but you could do small-scale studies and that was fine,” Collins explains. “But there is no point having a perfect study in terms of the design if it's too small, there's no point in having an enormous study if the methodology is wrong. But if you get them both right for the questions you’re asking, then suddenly a signal emerges out of the mist and you get clear information about whether or not things matter, and how much.”
Making a difference through data
In 1985, Collins became co-director with Peto as of the Clinical Trial Services Unit (CTSU) at the University of Oxford, designing, running and analysing the results of clinical trials in a range of diseases. Many of these studies have focused on cardiovascular disease, most notably a series of major trials and meta-analyses of the cholesterol-lowering effects of statin drugs in reducing the risk of suffering a stroke or heart attack.
Then, in 2013, Collins established the NDPH, bringing together research groups from different parts of the university that were gathering, analysing and interpreting data to answer important questions about health on a grand scale. It’s almost the opposite of the current trend towards ever-more-personalised health interventions.
“If you put all your resources into trying to treat an individual perfectly, that first assumes you know how to do it,” Collins says. “It may well be that you get most of it right, but you may get some of it wrong. Instead, at a population level, you could be better off with some general approaches that apply to far more people. Some individuals may not do quite as well, but individuals on average may do much better – a more equitable and potentially much more effective distribution of health.”
The department’s primary focus is on chronic diseases that occur most commonly during middle and old age, such as cancer, heart disease and diabetes. And while there has been considerable success in some areas – such as a significant reduction in early deaths from cardiovascular disease through statins and other interventions – there are new challenges ahead.
“People just don't realise how good things have got,” says Collins. “If you look at the trends in terms of mortality and related morbidity, not just in the rich countries but across the world, we've been fantastically successful in getting kids through to adolescence and into early middle age.
“Now we’re getting people into middle age and we’re seeing big reductions in the age-specific death rates of cardiovascular disease and many cancers due to improvements in prevention and treatment. But we need to do the same with other debilitating conditions of ageing, such as dementia, joint problems and other degenerative diseases.”
Big data, big challenge
As a statistician, Collins has always been eager to gather and analyse data. As we move into the era of ‘Big Data’, new computing and scientific tools are allowing him and his colleagues to gather and crunch more information than ever before. But this isn’t always helpful. The next big challenge facing Collins is big data.
“As an epidemiologist in the past the complaint was that we had too little data,” he says. “Now the complaint is too much data!”
To get a grip on the data problem, the NDPH and Oxford’s Nuffield Department of Medicine recently opened the new Big Data Institute (BDI), bringing together medical experts, statisticians, geneticists, epidemiologists, bio-informaticians and engineers.
In many ways, Collins sees this as a validation of his student quest to bring maths and medicine together. And he has arguably saved many more lives through his work on clinical trials that are applicable to many millions of people around the world than anything he could have done as a practicing clinician working with individual patients.
“I often think it's only since I left medical school that I really learned medicine,” he reflects. “We have to keep going back and asking for the evidence – to find out not what we think might work, but what really works.”