Towards better prediction of individual longevity
For a single individual the remaining length of life seems highly unpredictable. However, factors may be identified that characterise groups of individuals with shorter or longer longevity. The characterisation of such groups may contribute to the discussion on how to insure longevity risk. This study aims to maximise predictive value and to quantify the remaining uncertainty. We identify the predictive value of a broad selection of potential predictors, based on a 24-year mortality follow-up in the Longitudinal Aging Study Amsterdam, which is representative for the 55-85 years age group in the Netherlands (n=3,088, period 1993-2017). Potential predictors involved six domains: socio-demographics, disease history and medication use, physical functioning, lifestyle, psychosocial factors, and blood markers. We found significant predictors across all domains, including both self-reported and objectively tested measures. The significant predictors in the first five domains explained 21.3% of the variance in longevity. Additional predictive value of 3.7% was contributed by blood markers of disease processes and a genetic marker. We conclude that the prediction of individual longevity requires a broad set of variables, including both subjective and objective information. Yet, 75% of the variance in longevity remains unexplained, so that a large error margin remains in the prediction of an individual’s longevity.