A better world towards convergence of longevity?

www.thelancet.com Published online February 21, 2017 http://dx.doi.org/10.1016/S0140-6736(17)30314-8 1 Forecasting life expectancy by age and sex is broadly used for research and planning of health sevices, social services, pensions, and economics, and has been developed at the national and multicountry levels. The basic idea for predicting life expectancy is closely related to the concept of epidemiological transition. However, the idea is neither entirely based on evidence nor well-defi ned methodologically. Improvement in life expectancy is achieved through reductions in infant and younger age mortality and the progressive delay of mortality among older people. Given the lack of comparable longitudinal mortality data and the diffi culty in understanding the outputs from compound forecasting modelling, it is clear that better longitudinal data about age-specifi c mortality are needed to reliably predict extending longevity. In The Lancet, Vasilis Kontis and colleagues have tried to avoid the pitfalls of such modelling. Using probabilistic projections from 21 diff erent models with a 90% complete dataset, the authors predict that life expectancy will increase uninterruptedly in 35 industrialised countries by 2030 with a persistent female advantage. The sex diff erence is predicted to diminish by 2030 in most industrialised countries, except in Mexico, Chile, France, and Greece. A high probability of surpassing the maximum human lifespan that was considered as a potential maximum by some researchers, is predicted in men and women. The study notes the extensive shortage of relevant data about unexpected events (eg, climate fl uctuations and disease outbreaks) and changes in social determinants (eg, migration patterns and economic crises), which lead to uncertainties in projections of life expectancy. Most population projections are deterministic models, using simple extrapolations, and take into account the non-linearity observed. Hence, these models are unable to determine the probability of the uncertainty of future trends. This study proposes a Bayesian method for probabilistic projection of multiple models to fully capture the uncertainty of future trends in mortality and life expectancy. The study predicted uninterrupted gains of life expectancy with a higher probability for men (85%) than for women (65%) in 35 countries, with sex gaps in life expectancy expected to shrink in all countries, except in Mexico. Female life expectancy would break the 90-year barrier with a more than 50% probability by 2030, a barrier believed unreachable at the turn of the 21st century. The forecasts of life expectancy at birth are broadly similar to the UN predictions. The fi ndings raise crucial issues about which responses are appropriate to tackle such worsening disparity in terms of health policy and provision of health services. Identifying the best model for forecasting life expectancy will help assess appropriate health prevention strategies in diff erent populations. In terms of relevant policy implications, this study provides substantial evidence of longevity gains and identifi es the groups predicted to have a greater contribution to gains. Accurate forecasting of life expectancy is needed. Country diff erences in life expectancy over 25 years remain. Global Health 2035 indicates that most low-income and middle-income countries achieved a “grand convergence” in health. Countries are moving towards universal health coverage. Forecasting life expectancy at birth and at age 65 years can help governments and health services to make the right investments in health, such as averting deaths due to infectious diseases and reducing maternal and child mortality. Achieving universal health coverage is worthy, plausible, and needs to be continued. A better world towards convergence of longevity?