Professor of Nutritional Epidemiology at the University of Leeds, Janet Cade, helps to shed some light on what an epidemiologist does and how data lies at the centre of it all.
‘An epi….demi….ologist, is that something to do with skin?’ Understandably, until recently, most people had never heard of epidemiology. The root of the word is with epidemic – a widespread disease in a community, sadly we now know the word pandemic too, a cross-country epidemic. An epidemiologist studies the distribution (frequency, pattern) and determinants (causes or risk factors such as diet, smoking) of health related conditions and diseases in specified populations. The focus of their work is on overall public health aiming to identify population-based risks to support health policy. Clinical medicine, in contrast, is focussing on the individual and their physical signs and symptoms to diagnose, treat and prevent disease.
An epidemiologist has to rely on data, the bigger the better, to calculate rates and risks. We are always aware of measurement error, and factor this into our data interpretation. So the presentation of COVID rates which we see daily in the news, needs to be interpreted with an eye for the accuracy of that data. The fact that we are now testing far more people for COVID than we were in April 2020 is likely to mean that the very high numbers identified then were in fact much higher than we thought. On top of this, the tests are not 100% reliable, another form of error. We can try to minimise this error by testing people in a standardised fashion and having appropriate sample sizes. However, as yet, I do not think we have found clear ways to present the nuances of this flawed data to the public.
In my own field of nutritional epidemiology, some people have said that measurement errors are so great that we cannot draw reliable conclusions. Results from studies are often conflicting, with one study showing a certain vitamin ‘causes cancer’ and another study shows it ‘cures cancer’. Media headlines often sensationalise study findings. Whilst we would like to have gold standard randomised controlled trial data, that is often not possible in relation to food and nutrient intake that may take many years to influence risk of disease. In long-term cohort studies, which are possible, how we measure diet is critical. RIRO – rubbish in, rubbish out.
Many studies use questionnaires to measure food intake, these are often not adequately assessed in relation to their ability to measure what they are intending to measure (a particular food or nutrient). Newer technologies are now available, such as myfood24, this online dietary assessment tool has been developed by academics with a focus on making sure the underlying food database is comprehensive; putting the tool through a rigorous comparison against independent biomarkers. No tool is perfect, and we still have to be aware of potential self-reported measurement errors. However, it provides us with an opportunity to collect reliable data on large numbers whilst minimising costs.
Epidemiologists study populations. They rely on data, which is often flawed. Better measurement tools are needed, combined with careful analysis, interpretation and clear communication of our findings. Educating people about epidemiology, as has been forced on us by the COVID pandemic, will allow for more nuanced interpretation and debate of our findings. This ‘Cinderella’ science is fundamental to our promotion of public health.