Observational studies: the good, the bad and the sensational
We're all used to picking up the newspaper and seeing a headline screaming that food X is linked with cancer or that food Y can prolong our lives by a few years. How on earth do scientists work this out?
The answer is observational studies – the science of crunching the diet, health and lifestyle data of thousands of individuals to look for statistically significant patterns called 'associations'.
Observational studies are valuable
It wouldn’t be feasible or affordable to recruit healthy people for decades into an active intervention study to see how a particular diet, supplement or food affected their risk of developing an illness, like stroke or type 2 diabetes. For one, there would be a rebellion if, for example, they had to eat spinach daily for 10 years. Also, it's unlikely ethics committees would approve a study which restricts the lives of participants for the many years it takes for chronic conditions to develop.
The advantage of prospective observational studies is that researchers collect lifestyle and health data using questionnaires at regular intervals leaving participants to get on with their normal lives. Then, they use statistical analysis to examine associations between certain dietary patterns – such as high vegetable intakes or use of vitamin D supplements – and the risk of developing illnesses.
Risk is presented as 'odds ratios' or 'relative risk' which is a way of expressing the risk of the high/low consumers of a food against a baseline group of people. So, if the media say that high fruit consumers have a 20% lower risk of heart disease, this is in comparison with low fruit consumers, not in absolute terms.
But there are downsides
The main issue with observational studies is that associations are not the same as cause and effect. Remember participants in observational studies are only being observed, not controlled in any way. So, the person who is eating a lot of fruit could also be exercising regularly and avoiding alcohol. If we see an association between high fruit intake and lower risk of heart disease, we can't tell if it's the fruit, the exercise or something else. This is called confounding.
Researchers try to control it using statistics but it's not 100% effective and only works for the confounders they have measured. To be certain, they would need a randomised controlled trial where people are asked to eat more fruit versus a placebo. All that observational studies can tell us is that there is a link between certain factors and risk of a disease which deserves further examination. Remember that next time you pick up a newspaper and read a sensational headline about diet.
If you'd like to know more about this topic, check out my conference report in Proceedings of the Nutrition Society https://pubmed.ncbi.nlm.nih.gov/35042570/