Giulia Vilone looks at how sodium and potassium intake are affecting stroke risk in the UK
Nutrition is one of those areas that can benefit from the new techniques in big data analytics. Despite the evidence that diet and health are closely related, the exact mechanisms through which diet affects people's health have remained obscure to date, owing to the lack of long-term data on the almost infinite combinations of foods and diet-related diseases.
Cracking the mystery of nutrition to provide healthy, personalised diets will significantly improve quality of life and long-term health outcomes for millions of people, and decrease the associated cost for medical care currently sustained by governments, private individuals and other entities. Health and life insurers are among those that can benefit from the long-term improvement of people's diet and health and from more refined models that can be used in underwriting, pricing and claim processes.
Several countries have attempted to overcome the lack of data by carrying out regular dietary surveys of their population, but, to date, the data collected has been limited. Obstacles to collecting long-term data include the increased running costs over long periods of time and the human aversion to carrying out repetitive tasks, such as filling out daily food diaries. Historical evidence has shown that after two or three days, the quality of data collected significantly worsens.
Other ways to collect long-term data should be considered. National surveys could be completed with data coming from other indirect sources from consumers, such as apps, sensors, social media and businesses (retail and manufacturers). These sources can represent a cheaper alternative to collecting data and a more engaging way for people to share long-term information on their diet, but they usually provide lower-quality or biased data. Moreover, once this information has been collected, it needs to be harmonised and linked in order to make dietary analyses feasible. Therefore, the only way available today to assess long-term diets is still to develop predictive models based on the short-term food consumption diaries.
An example of these models can be represented by that developed by Creme Global, together with the European Food Safety Authority, to create the Compiled European Food Consumption Database. This database contains data retrieved from 34 national food consumption surveys, representing the short-term diet of 66,492 individuals from 22 EU member states. The data was harmonised and fed into an adapted version of methods developed by the National Cancer Institute to assess the usual intake distributions of 589 food groups covering most of the diet.
This article presents the results obtained from two dietary assessments on the consumption of two nutrients by the adult population of the UK.
The two nutrients chosen, sodium and potassium, have a known association with blood pressure and hence cardiovascular disease and stroke. Whereas sodium raises the risk of stroke, potassium reduces it. These assessments analysed each nutrient separately.
The food consumption data was retrieved from the UK's National Diet and Nutrition Survey (Public Health England, 2014) carried out from 2008 to 2012, covering around 7,000 individuals aged 1.5 years and older, randomly selected from 799 postcodes to represent the entire UK population, infants excluded. Some 3,450 adults aged 19+ years completed at least three days of the food and drink diary.
This analysis is based on 2,083 adults, selected as core cases for the main UK sample; the other subjects were selected to increase the sample sizes in Wales, Scotland and Northern Ireland.
Cutting out sodium
According to the Stroke Association, the incidence of strokes in the UK ranges from 115 to 150 per 100,000 people depending on the study. The overconsumption of sodium, mainly consumed as table salt, is linked with higher blood pressure, which is a risk factor for cardiovascular diseases. A daily average consumption of salt higher than 5g is associated with a 23% greater stroke risk. Salt is made of 40% sodium; hence the daily average consumption of sodium should be lower than 2g per day (2,000mg/day). Some studies have shown that a reduction of salt intake by 2-2.3g (corresponding to 0.8-0.92g of sodium) is associated with a reduction in stroke by 20%.
Table 1 displays the results obtained from the sodium intake assessment in the adult UK population. Both the average and median consumption of sodium already exceeds the maximum recommended intake.
The assessment results below show the incremental risk of stroke for each statistic measure.
This has been calculated assuming that the increment of the risk remains constant over the range of intakes, so an increase of 0.9g of sodium leads to a 20% increase in the risk of stroke.
The estimated increase in the risk for each statistic has been extrapolated from the 23% increase that occurs when the sodium intake exceeds 2g per day.
The recommended daily intake of potassium is 3.5g per day for adults (British Nutrition Foundation, 2015) and, according to scientific studies, a potassium intake higher than 1.64g per day is associated with a 21% reduced stroke risk. Moreover, every 1g per day increase in potassium intake decreases the stroke risk by 11%.
In this case, both the average and median potassium intakes are below the minimum recommended intake. However, the great majority of the UK population eats enough potassium to enjoy its positive effects on the stroke risk.
As per sodium, the results above show the relative reduction in stroke risk for each statistic, based on extrapolating the 21% decrease that occurs when potassium intake exceeds 1.64g per day.
More advanced models, analysing the effects of the entire diet on people's health, might become real in the future, when more long-term, more robust and harmonised data will be available.
These analyses can be integrated into the existing actuarial predictive models in order to obtain more accurate estimates on the incidence of impairing diseases carrying high costs for society and the insurance companies.
If health insurers and pension funds can easily capture the diet of their clients (and many people are doing this themselves anyway) they can use these techniques to assess the likelihood of chronic diseases. Several aspects of the insurance business can benefit from these insights; new health-coverage, diet-dependent products can be designed to match the needs of a wider public; risks can be quantified with more precision, leading to a more strategic allocation of the company's capital; medical costs to be refunded by the insurers can be reduced by associating the proper diet with the disease treatments. This is just to name a few of these benefits, but the list might be much longer.
Governments can also benefit from these models by using them to allocate healthcare resources in a more effective manner.
Giulia Vilone is an IFoA student, employed as senior data analyst at Creme Global