Is Mortality Spatial or Social?

Mortality modelling for the purposes of demographic forecasting and actuarial pricing is generally done at an aggregate level using national data. Modelling at this level fails to capture the variation in mortality within country and potentially leads to a mis-specification of mortality forecasts for a subset of the population. This can have detrimental effects for pricing and reserving in the actuarial context. In this paper we consider mortality rates at a regional level and analyse the variation in those rates. We consider whether variation in mortality rates within a country can be explained using local economic and social variables. Using Northern Ireland data on mortality and measures of deprivation we identify the variables explaining mortality variation. We create a population polarisation variable and find that this variable is significant in explaining some of the variation in mortality rates. Further, we consider whether spatial and non-spatial models have a part to play in explaining mortality differentials.

Structural Breaks in Mortality Models and their Consequences

In recent years, the issue of life expectancy has become of upmost importance to pension providers, insurance companies and the government bodies in the developed world. Significant and consistent improvements in mortality rates and, hence, life expectancy have led to unprecedented increases in the cost of providing for older ages. This has resulted in an explosion of stochastic mortality models forecasting trends in mortality data in order to anticipate future life expectancy and, hence, quantify the costs of providing for future aging populations. Many stochastic models of mortality rates identify linear trends in mortality rates by time, age and cohort, and forecast these trends into the future using standard statistical methods. The modeling approaches used failed to capture the effects of any structural change in the trend and, thus, potentially produced incorrect forecasts of future mortality rates. In this paper, we look at a range of leading stochastic models of mortality and test for structural breaks in the trend time series.

Identifying Structural Breaks in Stochastic Mortality Models

In recent years, the issue of life expectancy has become of utmost importance to pension providers, insurance companies, and government bodies in the developed world. Significant and consistent improvements in mortality rates and hence life expectancy have led to unprecedented increases in the cost of providing for older ages. This has resulted in an explosion of stochastic mortality models forecasting trends in mortality data to anticipate future life expectancy and hence quantify the costs of providing for future aging populations. Many stochastic models of mortality rates identify linear trends in mortality rates by time, age, and cohort and forecast these trends into the future by using standard statistical methods. These approaches rely on the assumption that structural breaks in the trend do not exist or do not have a significant impact on the mortality forecasts. Recent literature has started to question this assumption. In this paper, we carry out a comprehensive investigation of the presence or of structural breaks in a selection of leading mortality models. We find that structural breaks are present in the majority of cases. In particular, we find that allowing for structural break, where present, improves the forecast result significantly.

Repeated measures of body mass index and C-reactive protein in relation to all-cause mortality and cardiovascular disease: Results from the consortium on health and ageing network of cohorts in Europe and the United States (CHANCES)

Obesity has been linked with elevated levels of C-reactive protein (CRP), and both have been associated with increased risk of mortality and cardiovascular disease (CVD). Previous studies have used a single ‘baseline’ measurement and such analyses cannot account for possible changes in these which may lead to a biased estimation of risk. Using four cohorts from CHANCES which had repeated measures in participants 50 years and older, multivariate time-dependent Cox proportional hazards was used to estimate hazard ratios (HR) and 95 % confidence intervals (CI) to examine the relationship between body mass index (BMI) and CRP with all-cause mortality and CVD. Being overweight (≥25–<30 kg/m2) or moderately obese (≥30–<35) tended to be associated with a lower risk of mortality compared to normal (≥18.5–<25): ESTHER, HR (95 % CI) 0.69 (0.58–0.82) and 0.78 (0.63–0.97); Rotterdam, 0.86 (0.79–0.94) and 0.80 (0.72–0.89). A similar relationship was found, but only for overweight in Glostrup, HR (95 % CI) 0.88 (0.76–1.02); and moderately obese in Tromsø, HR (95 % CI) 0.79 (0.62–1.01). Associations were not evident between repeated measures of BMI and CVD. Conversely, increasing CRP concentrations, measured on more than one occasion, were associated with an increasing risk of mortality and CVD. Being overweight or moderately obese is associated with a lower risk of mortality, while CRP, independent of BMI, is positively associated with mortality and CVD risk. If inflammation links CRP and BMI, they may participate in distinct/independent pathways. Accounting for independent changes in risk factors over time may be crucial for unveiling their effects on mortality and disease morbidity.

Models of Mortality - Analysing the Residuals

The area of mortality modelling has received significant attention over the last 20 years owing to the need to quantify and forecast improving mortality rates. This need is driven primarily by the concern of governments, professionals, insurance and actuarial professionals and individuals to be able to fund their old age. In particular, to quantify the costs of increasing longevity we need suitable model of mortality rates that capture the dynamics of the data and forecast them with sufficient accuracy to make them useful. In this paper we test several of those models by considering the fitting quality and in particular, testing the residuals of those models for normality properties. In a wide ranging study considering 30 countries we find that almost exclusively the residuals do not demonstrate normality. Further, in Hurst tests of the residuals we find evidence that structure remains that is not captured by the models.

Excess mortality in Type 1 diabetes diagnosed in childhood and adolescence: a systematic review of population-based cohorts

Aims: Systematic review of mortality in childhood-/adolescent-diagnosed Type 1 diabetes and examination of factors explaining the mortality variation between studies. 
Methods: Relevant studies were identified from systematic searches of MEDLINE and EMBASE. Observed and expected numbers of deaths were extracted, and standardised mortality ratios (SMRs) and 95 % confidence intervals (CIs) were calculated. Negative binomial regression was used to investigate association between mortality and study/country characteristics.
Results: Thirteen relevant publications with mortality data were identified describing 23 independent studies. SMRs varied markedly ranging from 0 to 854 (chi-squared = 70.68,df = 21, p<0.0001). Significant associations were observed between SMR and mid-year of follow-up [incidence rate ratio (IRR) 0.95, 95 % CI 0.91–0.99 equivalent to a 5 % decrease per year], between SMR and infant mortality rate (IRR 1.07, 95 % CI 1.02–1.12, a 7 % increase for each death per 1,000 live births) and, after omitting an outlier, between SMR and health expenditure as a percentage of gross domestic product (GDP) (IRR 0.79, 95 % CI 0.68–0.93, a 21 % decrease for each one percent increase in GDP). No relationship was detected between SMR and a country’s childhood diabetes incidence rate or GDP.
Conclusions: Excess mortality in childhood-/adolescent diagnosed Type 1 diabetes is apparent across countries worldwide. Excesses were less marked in more recent studies and in countries with lower infant mortality and higher health expenditure.

The association between glucose lowering drug use and mortality among breast cancer patients with type 2 diabetes.

This study assessed the association between glucose-lowering drug (GLD) use, including metformin, sulphonylurea derivatives and insulin, after breast cancer diagnosis and breast cancer-specific and all-cause mortality. 1763 breast cancer patients, diagnosed between 1998 and 2010, with type 2 diabetes were included. Cancer information was retrieved from English cancer registries, prescription data from the UK Clinical Practice Research Datalink and mortality data from the Office of National Statistics (up to January 2012). Time-varying Cox regression models were used to calculate HRs and 95 % CIs for the association between GLD use and breast cancer-specific and all-cause mortality. In 1057 patients with diabetes before breast cancer, there was some evidence that breast cancer-specific mortality decreased with each year of metformin use (adjusted HR 0.88; 95 % CI 0.75–1.04), with a strong association seen with over 2 years of use (adjusted HR 0.47; 95 % CI 0.26–0.82). Sulphonylurea derivative use for less than 2 years was associated with increased breast cancer-specific mortality (adjusted HR 1.70; 95 % CI 1.18–2.46), but longer use was not (adjusted HR 0.94; 95 % CI 0.54–1.66). In 706 patients who developed diabetes after breast cancer, similar patterns were seen for metformin, but sulphonylurea derivative use was strongly associated with cancer-specific mortality (adjusted HR 3.64; 95 % CI 2.16–6.16), with similar estimates for short- and long-term users. This study provides some support for an inverse association between, mainly long-term, metformin use and (breast cancer-specific) mortality. In addition, sulphonylurea derivative use was associated with increased breast cancer-specific mortality, but this should be interpreted cautiously, as it could reflect selective prescribing in advanced cancer patients.

Urban/rural variation in the influence of widowhood on mortality risk: a cohort study of almost 300,000 couples

Death of a spouse is associated with increased mortality risk for the surviving partner (the widowhood effect). We investigated whether the effect magnitude varied between urban, rural and intermediate areas, assembling death records (2001-2009) for a prospective cohort of 296,125 married couples in Northern Ireland. The effect was greatest during the first six months of widowhood in all areas and for both sexes. Subsequently, the effect was attenuated among men in rural and intermediate areas but persisted in urban areas (HRs and 95% CIs: rural 1.09 [0.99, 1.21]; urban 1.35 [1.26, 1.44]). Among women the effect was attenuated in all areas (rural 1.06 [0.96, 1.17]; urban 1.09 [1.01, 1.17]). The impacts of spousal bereavement varied between urban and more rural areas, possibly due to variation in social support provided by the wider community. We identify men in urban areas as being in greatest need of such support and a possible target for health interventions.

Dietary fat and breast cancer mortality: a systematic review and meta-analysis

Background: The influence of dietary fat upon breast cancer mortality remains largely understudied despite extensive investigation into its influence upon breast cancer risk 

Objective: To conduct meta-analyses of studies to clarify the association between dietary fat and breast cancer mortality Design: MEDLINE and EMBASE were searched for relevant articles published up to March 2012. Risk of all-cause or breast cancer specific death was evaluated by combining multivariable adjusted estimates comparing highest versus lowest categories of intake; and per 20 gram increase in intake of total and/or saturated fat (g/day) using random-effects meta-analyses. 

Results: Fifteen prospective cohort studies investigating total fat and/or saturated fat intake (g/day) and breast cancer mortality were included. There was no difference in risk of breast cancer specific death (n = 6; HR = 1.14; 95% CI: 0.86, 1.52; P = 0.34) or all cause death (n = 4; HR = 1.73; 95% CI: 0.82, 3.6; P = 0.15) for women in the highest versus lowest category of total fat intake. Breast cancer specific death (n = 5; HR = 1.63; 95% CI: 1.19, 2.24; p <0.01) was higher for women in the highest versus lowest category of saturated fat intake. 

Conclusions: These meta-analyses have shown that saturated fat intake negatively impacts upon breast cancer survival.