Last updated Feb 24, 2019
Researchers have used UK Biobank data to help develop a score that predicts an individual’s risk of dying within 5 years for people aged between 40 and 70 years old in the UK, The Lancet, a prestigious medical journal, has reported.
The score, which uses measures that can be obtained by simple questionnaires without any need for physical examination, such as self-rated health and usual walking speed, could be used by individuals to improve awareness of their health status, and by doctors to identify high-risk individuals for further treatment, say the authors.
The study uses UK Biobank data, but was undertaken by scientists in Sweden. Along with around 1,800 scientists so far, they have registered with UK Biobank and asked for data to help with their research. Their results will have been reviewed by other scientists before being published and will be returned to the resource so that other scientists can benefit from them.
The UbbLE study (UK Longevity Explorer) has generated extensive coverage around the world, and generated lots of interest in UK Biobank. Professor Rory Collins, UK Biobank Principal Investigator, said: “UK Biobank will only be limited by the imagination of the researchers using it.
“As an open access resource, scientists will be able to scrutinise each other’s work, build upon their results or undertake their own novel studies to improve the health of future generations.”
Individuals can calculate their personalised 5-year mortality risk and what the study authors call an ‘Ubble age’ ―the age where the average mortality risk in the population is most similar to the estimated risk of the individual―through an online questionnaire, using the UbbLE dedicated interactive website The site was developed with Sense About Science, a UK charity that that equips people to make sense of scientific and medical claims in public discussion.
“The Ubble website allows anybody in the UK between forty and seventy years old to calculate their ‘Ubble age’ and their risk of dying within the next five years compared to the general population, using a simple set of a dozen or so questions,” says study co-author Professor Erik Ingelsson from Uppsala University, Sweden. “This is the first study of its kind which is based on a very large study sample, and is not limited to specific populations, single types of risk, or requiring laboratory testing.”
To create the score, Professor Ingelsson and Dr Andrea Ganna, from the Karolinska Institutet in Sweden, analysed data collected by UK Biobank between 2006 and 2010. They used a statistical survival model to assess the probability that 655 specific demographic, lifestyle, and health measurements could predict death from any cause and six specific causes, in men and women separately.
The results allow researchers to compare the predictive ability of a large number of health-related risk factors which were formerly studied separately and using different populations, although the predictive ability of risk factors does not necessarily imply they are causally related to the risk of dying.
For example, the research shows that self-reported information, such as usual walking pace and illness and injuries in the past 2 years, is generally a stronger predictor of death and survival than biological measurements, such as pulse rate and blood pressure. Surprisingly, self-reported walking pace is a stronger predictor of death risk in both men and women than smoking habits and other lifestyle measurements.
Self-rated overall health emerged as the single most powerful predictor of death in men, and previous cancer diagnosis the strongest predictor in women. When excluding individuals with serious diseases or disorders, smoking habits were the strongest predictors of mortality from any cause.
Using these findings, the researchers then developed an easy-to-calculate risk score for an individual’s risk of dying in the next 5 years, based on the most predictive self-reported information, including 13 questions for men and 11 for women. The performance of the score was validated in 35810 participants enrolled at two Scottish centres which were not used to develop the score, and was found to have around 80% accuracy in men and women. UK life tables and census information were used to calibrate the score for the overall UK population.
According to study co-author Dr Ganna: “The fact that the score can be measured online in a brief questionnaire, without any need for lab tests or physical examination, is an exciting development. We hope that our score might eventually enable doctors to quickly and easily identify their highest risk patients, although more research will be needed to determine whether it can be used in this way in a clinical setting. Of course, the score has a degree of uncertainty and shouldn’t be seen as a deterministic prediction. For most people, a high risk of dying in the next five years can be reduced by increased physical activity, smoking cessation, and a healthy diet.”
Writing in a linked Comment, Simon Thompson and Peter Willeit from the University of Cambridge say: “One appealing aspect of the website is the representation of an individual’s estimated 5 year mortality risk as what the authors call “Ubble age” (the age where the average risk in the population is most similar to the estimated risk of the individual); this concept is similar to that of heart age developed for communicating cardiovascular risk. Whether this will help individuals improve self-awareness of their health status, however, or only lead to so-called cyber-chondria, is a moot point. Moreover, 5 year mortality is easier to predict than long-term morbidity, or quality of life and life expectancy, all of which are more important to individuals and to society.”