Guangquan Li

Future life expectancy in 35 industrialised countries: projections with a Bayesian model ensemble

Summary Background Projections of future mortality and life expectancy are needed to plan for health and social services and pensions. Our aim was to forecast national age-specific mortality and life expectancy using an approach that takes into account the uncertainty related to the choice of forecasting model. Methods We developed an ensemble of 21 forecasting models, all of which probabilistically contributed towards the final projections. We applied this approach to project age-specific mortality to 2030 in 35 industrialised countries with high-quality vital statistics data. We used age-specific death rates to calculate life expectancy at birth and at age 65 years, and probability of dying before age 70 years, with life table methods. Findings Life expectancy is projected to increase in all 35 countries with a probability of at least 65% for women and 85% for men. There is a 90% probability that life expectancy at birth among South Korean women in 2030 will be higher than 86·7 years, the same as the highest worldwide life expectancy in 2012, and a 57% probability that it will be higher than 90 years. Projected female life expectancy in South Korea is followed by those in France, Spain, and Japan. There is a greater than 95% probability that life expectancy at birth among men in South Korea, Australia, and Switzerland will surpass 80 years in 2030, and a greater than 27% probability that it will surpass 85 years. Of the countries studied, the USA, Japan, Sweden, Greece, Macedonia, and Serbia have some of the lowest projected life expectancy gains for both men and women. The female life expectancy advantage over men is likely to shrink by 2030 in every country except Mexico, where female life expectancy is predicted to increase more than male life expectancy, and in Chile, France, and Greece where the two sexes will see similar gains. More than half of the projected gains in life expectancy at birth in women will be due to enhanced longevity above age 65 years. Interpretation There is more than a 50% probability that by 2030, national female life expectancy will break the 90 year barrier, a level that was deemed unattainable by some at the turn of the 21st century. Our projections show continued increases in longevity, and the need for careful planning for health and social services and pensions. Funding UK Medical Research Council and US Environmental Protection Agency.

Trends and mortality effects of vitamin A deficiency in children in 138 low-income and middle-income countries between 1991 and 2013: a pooled analysis of population-based surveys

BACKGROUND Vitamin A deficiency is a risk factor for blindness and for mortality from measles and diarrhoea in children aged 6-59 months. We aimed to estimate trends in the prevalence of vitamin A deficiency between 1991 and 2013 and its mortality burden in low-income and middle-income countries. METHODS We collated 134 population-representative data sources from 83 countries with measured serum retinol concentration data. We used a Bayesian hierarchical model to estimate the prevalence of vitamin A deficiency, defined as a serum retinol concentration lower than 0·70 μmol/L. We estimated the relative risks (RRs) for the effects of vitamin A deficiency on mortality from measles and diarrhoea by pooling effect sizes from randomised trials of vitamin A supplementation. We used information about prevalences of deficiency, RRs, and number of cause-specific child deaths to estimate deaths attributable to vitamin A deficiency. All analyses included a systematic quantification of uncertainty. FINDINGS In 1991, 39% (95% credible interval 27-52) of children aged 6-59 months in low-income and middle-income countries were vitamin A deficient. In 2013, the prevalence of deficiency was 29% (17-42; posterior probability [PP] of being a true decline=0·81). Vitamin A deficiency significantly declined in east and southeast Asia and Oceania from 42% (19-70) to 6% (1-16; PP>0·99); a decline in Latin America and the Caribbean from 21% (11-33) to 11% (4-23; PP=0·89) also occurred. In 2013, the prevalence of deficiency was highest in sub-Saharan Africa (48%; 25-75) and south Asia (44%; 13-79). 94 500 (54 200-146 800) deaths from diarrhoea and 11 200 (4300-20 500) deaths from measles were attributable to vitamin A deficiency in 2013, which accounted for 1·7% (1·0-2·6) of all deaths in children younger than 5 years in low-income and middle-income countries. More than 95% of these deaths occurred in sub-Saharan Africa and south Asia. INTERPRETATION Vitamin A deficiency remains prevalent in south Asia and sub-Saharan Africa. Deaths attributable to this deficiency have decreased over time worldwide, and have been almost eliminated in regions other than south Asia and sub-Saharan Africa. This new evidence for both prevalence and absolute burden of vitamin A deficiency should be used to reconsider, and possibly revise, the list of priority countries for high-dose vitamin A supplementation such that a countrys priority status takes into account both the prevalence of deficiency and the expected mortality benefits of supplementation. FUNDIN Bill & Melinda Gates Foundation, Grand Challenges Canada, UK Medical Research Council.

Inference from ecological models: Estimating the relative risk of stroke from air pollution exposure using small area data

Maheswaran et al. (2006) analysed the effect of outdoor modelled NO(x) levels, classified into quintiles, on stroke mortality using a Poisson Bayesian hierarchical model with spatial random effects. An association was observed between higher levels of NO(x) and stroke mortality at the small area (enumeration district) level. As this model is framed in an ecological perspective, the relative risk estimates suffer from ecological bias. In this paper we use a different model specification based on Jackson et al. (2008), modelling the number of cases of mortality due to stroke as a binomial random variable where p(i) is the probability of dying from stroke in area i. The within-area variation in outdoor modelled NO(x) levels is used to determine the proportion of the population in area i falling into each of the five exposure categories in order to estimate the probability of an individual dying from stroke given the kth level of NO(x) exposure assuming a homogeneous effect across the study region. The inclusion of within-area variability in an ecological regression model has been demonstrated to help reduce the ecological bias (Jackson et al., 2006, 2008). Revised estimates of relative risk are obtained and compared with previous estimates.

Evaluating the No Cold Calling zones in Peterborough, England: application of a novel statistical method for evaluating neighbourhood policing policies

Some police forces in the UK institute ‘No Cold Calling’ (NCC) zones to reduce cold callings (unsolicited visits to sell products or services), which are often associated with rogue trading and distraction burglary. This paper evaluates the NCC-targeted areas chosen in 2005–06 in Peterborough and reports whether they experienced a measurable impact on their burglary rates in the period up to 2008. Time-series data for burglary at the census output area level are analysed using a Bayesian hierarchical modelling approach to address issues of data sparsity and lack of randomised allocation of areas to treatment groups that are often encountered in small-area quantitative policy evaluation. To ensure internal validity, we employ the interrupted time-series quasi-experimental design embedded within a matched case-control framework. Results reveal a positive impact of NCC zones on reducing burglary rates in the targeted areas compared with the control areas.

The ecology of outdoor rape: The case of Stockholm, Sweden

The objective of this article is to report the results of an ecological study into the geography of rape in Stockholm, Sweden, using small area data. In order to test the importance of factors indicating opportunity, accessibility and anonymity to the understanding of the geography of rape, a two-stage modelling approach is implemented. First, the overall risk factors associated with the occurrence of rape are identified using a standard Poisson regression, then a local analysis using profile regression is performed. Findings from the whole-map analysis show that accessibility, opportunity and anonymity are all, to different degrees, important in explaining the overall geography of rape - examples of these risk factors are the presence of subway stations or whether a basområde is close to the city centre. The local analysis reveals two groupings of high risk of rape areas associated with a variety of risk factors: city centre areas with a concentration of alcohol outlets, high residential population turnover and high counts of robbery; and poor suburban areas with schools and large female residential populations where subway stations are located and where people express a high fear of crime. The article concludes by reflecting upon the importance of these results for future research as well as indicating the implications of these results for policy.

Future inequalities in life expectancy in England and Wales – Authors' reply

We noted with interest the much greater increase in life expectancy for men than for women in England and Wales forecast to 2030 by James Bennett and colleagues (July 11, p 163). The expected rate of increase in life expectancy for men not only far exceeds the rate of increase in world-record life expectancy (4 months per year according to Bennett and colleagues vs 3 months per year according to the increase in world-record life expectancy), but also induces a sharp reduction in the life expectancy gap between the sexes towards values that could be considered biologically implausible. Although the authors primarily aim to predict mortality consistently at the subnational level, these sex-specific inconsistencies at the national level reveal substantial shortcomings of their model. The inconsistencies are probably due to the choice of an age–period– cohort model that was fitted to a short reference period, separately for men and women. This complex model was preferred over simpler models because of its smaller error in an out-of-sample forecast in the years 2002–12. In the presence of strong temporal serial correlation, such validation approaches tend to be biased towards the selection of models that are too flexible, potentially resulting in overfitting. Thus, the model is particularly sensitive towards the choice of the reference period, inducing a misleading extrapolation of short-term fluctuations into the future. More coherent and robust forecasts could be generated with a hierarchical model that jointly estimates parameters at different levels of aggregation, such as country, gender, and region, and, if applicable, also accounts for the diff erential eff ect of the smoking epidemic.