A major weakness in most nutrition and diet studies is that we have no true way of measuring what individuals eat. We rely solely on people keeping logs of their daily diets – but studies suggest around 60 percent of individuals misreport what they eat to some extent.

So although we've big data collections about diet and health, using patient log books, we do not have very accurate information because of widespread misreporting. This can be for various reasons, which include being too embarrassed in truth or simply since it can be hard to record all you consume over days and weeks.

With this in mind, my team at Imperial alongside researchers from Newcastle University and Aberystwyth University chose to develop a completely independent test that may reliably reveal the caliber of a person's diet. We began with a simple premise – asking 19 volunteers to come right into a clinic for three days after which eat and drink very defined diets, which range from very healthy to very unhealthy. They were formulated using World Health Organisation dietary guidelines, which advise on the very best diets to avoid conditions for example obesity, diabetes and heart disease.

Throughout this three day period, we collected urine samples each morning, afternoon and evening and then assessed the urine for hundreds of compounds, called metabolites, produced when certain foods are broken down in your body.

By analysing the thousands of metabolites in the volunteers' urine, we were able to develop an accurate test that measures biological markers in urine developed by the introduction to foods for example steak, chicken, fish and fruit and veggies. Case study also gives an indication of methods much fat, sugar, fibre and protein a person has eaten. We tested the precision of the test on data from a previous study. This included 225 UK volunteers as well as 66 people from Denmark.
All the volunteers had provided urine samples, and kept information on their daily diets.

Analysis of those urine samples enabled us to accurately predict the diet plan of the 291 volunteers.

As a control, we developed a urine metabolite profile of the healthy, balanced diet with a decent intake of fruit and vegetables. The concept is that this ‘healthy diet’ profile could be when compared to diet profile from an individual’s urine, to supply an immediate indicator of whether they are eating healthily.

We still have work to do in order to be sure that the test is going to be accurate on a greater number of individuals but we are certain that it will be available for use within clinical practice within two years. We estimate that it will cost around lb20 per test and that results will be available within a couple of days. The results could pinpoint if a person was not eating sufficient fruit and vegetables for example, or eating a lot of refined carbohydrates and fats.

In particular, it will be useful to help people have an accurate understanding of their own diets and just how they might improve their nutrition. This is useful to people who have pre-existing conditions for example heart disease and diabetes, but also people who are overweight and obese and can not slim down. Included in our research, we've developed a urine metabolite profile of a healthy, balanced diet with a decent consumption of fruit and vegetables. This ‘healthy diet’ profile could be when compared to diet profile from an individual’s urine, to supply an instant indicator of whether they are eating healthily.

The study, funded by the Scientific research Council and the National Institute for Health Research, was published in the journal Lancet Diabetes and Endocrinology and conducted in the MRC-NIHR National Phenome Centre.