View all news

The whole picture of your health

26 October 2017

With conflicting public messages about aspects of our health today, it is a real challenge to know which advice to follow. At Bristol, population health scientists are developing a ground-breaking health study method that will disentangle cause from effect, and help us to answer some of the world’s most pressing health questions.

Twenty-one years ago, over 14,000 pregnant mothers agreed to share one of the most personal and life-changing experiences of their lives – for science. Through years of questionnaires, blood samples, DNA tests and more, they gave the Children of the 90s epidemiology study an unprecedented bank of information. The scale and depth of the study makes it unique across the world. It has been the source of countless discoveries. We now know it is safer for babies to sleep on their back; that 15 minutes of vigorous exercise a day can reduce the risk of obesity by half; and that teenage girls who are too thin may be putting their bones at risk later on in life.

Bristol’s population health research has revolutionised the way we live, but the Children of the 90s study is just the beginning. This study took place on a vast scale, over several decades, yet very few research projects can replicate the time or resources. As a result, we often hear about a new study that claims to have the answer to the latest health advice, even if it is less reliable than the work which has been done before.

The challenge is to conduct a study with the depth and scale of Children of the 90s, but quicker and on a lower budget – and Bristol researchers have a solution. They are building an unprecedented genetic database called MR-Base. Using a principle called Mendelian randomisation, this bank of genetic information could transform the way in which we conduct health studies, leading to more robust and reliable health advice.

Is our health advice reliable?

‘New study suggests light drinking is healthy during pregnancy’ – does this sound familiar? These sorts of headlines are commonplace now. They claim to hold new information on a range of health issues, and capture the attention of anyone affected. Expectant mothers in the UK recently faced headlines both for and against light drinking during pregnancy.

Some studies suggested light drinking could help the unborn do better in school, while one newspaper was even so bold as to print the headline ‘Pregnant women told to ignore drinking advice’. But which advice is it referring to? With similar conflicts in reporting on cancer, obesity, high blood pressure and other health issues, how do we know which advice to follow and which to ignore?

Misleading studies

The problem with these studies is that they are observational: the researchers recorded a set of data for children of pregnant drinkers and another for children of pregnant non-drinkers, and looked for patterns in the results. However, this tells us nothing about cause and effect, and so rarely tells the whole story.

Professor George Davey Smith is a leading clinical epidemiologist working on the MRBase project. He explains: ‘It’s likely that mothers who drink more during pregnancy will differ in many ways from mothers who don’t. In the studies that have been reported, the mothers who drank more frequently during pregnancy were likely to have higher educational attainment, and healthier lifestyles.’ These social factors have the potential to affect the health, wellbeing and educational achievements of a child.

Finding the truth

The challenge is to conduct a studyunaffected by these external factors – knownas confounding risk factors. The best way of doing this is via a randomised control trial. This is a study where participants are randomly allocated into two groups, who will differ in no other way than the factor being studied. For example, to find out the effect of a new drug, one group may be administered the drug while the other is administered a placebo. This method makes it far more likely that any observed differences have been caused by the drug.

So why doesn’t everyone conduct this sort of study? ‘This is a great method,’ says Davey Smith, ‘but it’s very expensive, and in many situations you simply couldn’t do this for logistical or ethical reasons. You couldn’t randomise pregnant women to drink more or drink less for example.’ However, there is an elegant solution – and it comes from genetics.

Harnessing genetics

Without knowing it, we have all been recruited into an experiment from the moment we were conceived. As the ‘father of modern genetics’ Gregor Mendel famously discovered, the genes we inherit from our parents – influencing everything from our appetites, to our sleep patterns, to our drinking habits – are essentially randomised. They are generally transmitted independently of environmental factors, such as social class or lifestyle.

Mendelian randomisation groups people together by their genetic code and compares the results. If we compare a set of people with a version of a gene associated with heavier drinking and another set of people with a version associated with lighter drinking, any differences are likely to be a result of the gene itself and not any other environmental factors. Researchers use this method to interrogate a vast ‘library’ of people’s health and genetic data in MR-Base. They can analyse thousands of health characteristics, simply by splitting the library of people into groups based on which version of the related gene they have. They then observe the differences between these groups.

This can be used to produce clear, reliable answers on health causes and effects, such as the impacts of drinking more or less alcohol. In fact, it was shown in a 2012 genetic study using data from the Children of the 90s project that even moderate alcohol consumption during pregnancy was linked with children having a lower IQ.

Mendelian randomisation has also been used to debunk a number of other myths surrounding alcohol intake, such as the idea that drinking a moderate amount might lower the risk of heart disease – in fact, alcohol has an adverse effect across the range of intake.

Bristol’s role to play

Bristol’s Population Health Science Institute is second-to-none. It’s one of seven pioneering Specialist Research Institutes that builds on the University’s world-leading reputation for population health research that addresses some of society’s greatest challenges. It seems fitting that as the Children of the 90s generation comes of age – many now with their own children to provide a new generation to the ongoing study – the field of population health at Bristol is celebrating something of a ‘coming of age’ as well. The founding of this Institute will provide projects such as MR-Base the opportunity to truly flourish.

This will build a vast database of genetic associations which allows millions of potential causal relationships to be studied. By combining data from over one thousand genetic studies on diseases and other traits, the researchers have created a platform from which they can use Mendelian randomisation to predict health outcomes due to different factors.

Not only has this advance made it quick, cheap and easy to perform this analysis, but the output of each analysis should also be more reliable than using observational data alone. This will ultimately have a profound impact on both the treatments we can develop and on the reliability of the health advice we are given, perhaps even helping to restore trust in what we are told.

Further information

You can listen and download the audio version here (mp3).