Reexamining the Dominance of Birth Cohort Effects on Mortality

Do Japanese and Italian Women Live Longer than Women in Scandinavia?

Life expectancies at birth are routinely computed from period life tables. When mortality is falling, such period life expectancies will typically underestimate real life expectancies, that is, life expectancies for birth cohorts. Hence, it becomes problematic to compare period life expectancies between countries when they have different historical mortality developments. For instance, life expectancies for countries in which the longevity improved early (like Norway and Sweden) are difficult to compare with those in countries where it improved later (like Italy and Japan). To get a fair comparison between the countries, one should consider cohort data. Since cohort life expectancies can only be computed for cohorts that were born more than a hundred years ago, in this paper we suggest that for younger cohorts one may consider the expected number of years lost up to a given age. Contrary to the results based on period data, our cohort results then indicate that Italian women may expect to lose more years than women in Norway and Sweden, while there are no indications that Japanese women will lose fewer years than women in Scandinavia. The large differences seen for period data may just be an artefact due to the distortion that period life tables imply in times of changing mortality.

Decomposing Current Mortality Differences Into Initial Differences and Differences in Trends: The Contour Decomposition Method

This study proposes a new decomposition method that permits a difference in an aggregate measure at a final time point to be split into additive components corresponding to the initial differences in the event rates of the measure and differences in trends in these underlying event rates. For instance, when studying divergence in life expectancy, this method allows researchers to more easily contrast age-specific mortality trends between populations by controlling for initial age-specific mortality differences. Two approaches are assessed: (1) an additive change method that uses logic similar to cause-of-death decomposition, and (2) a contour decomposition method that extends the stepwise replacement algorithm along an age-period demographic contour. The two approaches produce similar results, but the contour method is more widely applicable. We provide a full description of the contour replacement method and examples of its application to life expectancy and lifetime disparity differences between the United States and England and Wales in the period 1980–2010.

Why did Danish women's life expectancy stagnate? The influence of interwar generations' smoking behaviour

The general health status of a population changes over time, generally in a positive direction. Some generations experience more unfavourable conditions than others. The health of Danish women in the interwar generations is an example of such a phenomenon. The stagnation in their life expectancy between 1977 and 1995 is thought to be related to their smoking behaviour. So far, no study has measured the absolute effect of smoking on the mortality of the interwar generations of Danish women and thus the stagnation in Danish women's life expectancy. We applied a method to estimate age-specific smoking-attributable number of deaths to examine the effect of smoking on the trends in partial life expectancy of Danish women between age 50 and 85 from 1950 to 2012. We compared these trends to those for women in Sweden, where there was no similar stagnation in life expectancy. When smoking-attributable mortality was excluded, the gap in partial life expectancy at age 50 between Swedish and Danish women diminished substantially. The effect was most pronounced in the interwar generations. The major reason for the stagnation in Danish women's partial life expectancy at age 50 was found to be smoking-related mortality in the interwar generations.

Rise, stagnation, and rise of Danish women's life expectancy

Health conditions change from year to year, with a general tendency in many countries for improvement. These conditions also change from one birth cohort to another: some generations suffer more adverse events in childhood, smoke more heavily, eat poorer diets, etc., than generations born earlier or later. Because it is difficult to disentangle period effects from cohort effects, demographers, epidemiologists, actuaries, and other population scientists often disagree about cohort effects' relative importance. In particular, some advocate forecasts of life expectancy based on period trends; others favor forecasts that hinge on cohort differences. We use a combination of age decomposition and exchange of survival probabilities between countries to study the remarkable recent history of female life expectancy in Denmark, a saga of rising, stagnating, and now again rising lifespans. The gap between female life expectancy in Denmark vs. Sweden grew to 3.5 y in the period 1975-2000. When we assumed that Danish women born 1915-1945 had the same survival probabilities as Swedish women, the gap remained small and roughly constant. Hence, the lower Danish life expectancy is caused by these cohorts and is not attributable to period effects.

Bayesian Population Forecasting: Extending the Lee-Carter Method

In this article, we develop a fully integrated and dynamic Bayesian approach to forecast populations by age and sex. The approach embeds the Lee-Carter type models for forecasting the age patterns, with associated measures of uncertainty, of fertility, mortality, immigration, and emigration within a cohort projection model. The methodology may be adapted to handle different data types and sources of information. To illustrate, we analyze time series data for the United Kingdom and forecast the components of population change to the year 2024. We also compare the results obtained from different forecast models for age-specific fertility, mortality, and migration. In doing so, we demonstrate the flexibility and advantages of adopting the Bayesian approach for population forecasting and highlight areas where this work could be extended.

Lifespan variation by occupational class: compression or stagnation over time?

Cross-sectional analyses of adult lifespan variation have found an inverse association between socioeconomic position and lifespan variation, but the trends by social class are unknown. We investigated trends in lifespan variation over four decades (1971-2010) by occupational social class (manual, lower nonmanual, upper nonmanual, other) using Finnish register data. We performed age and cause-of-death decompositions of lifespan variation for each sex (a) by occupational class over time and (b) between occupational classes at a shared level of life expectancy. Although life expectancy increased in all classes, lifespan variation was stable among manual workers and decreased only among nonmanual classes. These differences were caused by early-adult mortality: older-age lifespan variation declined for all the classes, but variation in early-adult mortality increased for all classes except the highest. The manual class's high and stagnant lifespan variation was driven by declines in circulatory diseases that were equally spread over early mortality-compressing and older mortality-expanding ages, as well as by high early-adult mortality from external causes. Results were similar for men and women. The results of this study, which is the first to document trends in lifespan variation by social class, suggest that mortality compression is compatible with increasing life expectancy but currently achieved only by higher occupational classes.

Mortality cohort effects from mid-19th to mid-20th century Britain: did they exist?

PURPOSE

Identification is a central problem with age-period-cohort analysis. Because age + cohort = period, there is no unique solution to the linear effect using generalized linear modeling, but cohort effects have caused greater controversy than age and period effects. To illustrate the magnitude of cohort effects given the presence of collinearity, we reanalyze data from the seminal study by Kermack et al, with an update.

METHODS

Relative mortality data in England and Wales between year 1845 and 1995 were analyzed using partial least squares regression. There were seven age groups ranging from 5 to 74 years old and 16 periods with 22 cohorts.

RESULTS

Our reanalysis seemed to support the existence of cohort effects in the mortality trends. Period and cohort effects were generally consistent with changes in the social, economic, and environmental factors taking place in the last two centuries. Our analysis also showed a declining trend in period effects up to 1950s.

CONCLUSIONS

Partial least squares and related methods provide intuitive pointers toward the separation of linear age, period, and cohort effects. Because statistical algorithms cannot distinguish between relative and actual mortality rates, cohort effects may be underestimated because of contamination by negative age effects.

Period-Based Mortality Change: Turning Points in Trends since 1950

We investigate a major turning point in mortality trends at adult ages that occurred for many low‐mortality countries in the late 1960s or early 1970s. We analyze patterns of total and cause‐specific mortality over the past 60 years using data from the Human Mortality Database and the World Health Organization. We focus on four broad categories of causes of death: heart diseases, cerebrovascular diseases, smoking‐related cancers, and all other cancers. We use a two‐slope regression model to assess the timing and magnitude of turning points in mortality trends over this era, making separate analyses by sex, age, and cause of death. The age pattern of temporal changes is given particular attention. Our results demonstrate convincingly that period‐based factors were very significant in the onset of the “cardiovascular revolution” in the years around 1970. In general, although cohort processes cannot be ruled out as a driver of mortality change in recent decades (especially for mortality due to smoking‐related cancers), the evidence reviewed here suggests that period factors have been the dominant force behind the mortality trends of high‐income countries during this era.

Year of birth effects in the historical decline of tuberculosis mortality: a reconsideration

Birth cohort patterns in mortality are often used to infer long-lasting impacts of early life conditions. One of the most widely accepted examples of a birth cohort effect is that of tuberculosis mortality before the late 1940s. However the evidential basis for claims of cohort-specific declines in tuberculosis mortality is very slight. Reanalysis of original or enhanced versions of datasets used previously to support claims of cohort effects in tuberculosis mortality indicated that: 1. where the initial decline in tuberculosis mortality occurred within the period of observation, onset of decline occurred simultaneously in many age groups, in a pattern indicative of 'period' not cohort-dependent effects. 2. there was little evidence of 'proportional hazard'-type cohort patterns in tuberculosis mortality for any female population studied. Therefore any mechanisms proposed to underlie this type of cohort pattern in male mortality must be sex-specific. 3. sex ratios of tuberculosis mortality at older ages peaked in cohorts born around 1900, and resembled cohort sex ratios of lung cancer mortality. This analysis indicates that age-specific patterns in the decline in tuberculosis mortality before 1950 are unlikely to reflect improvements in early life conditions. The patterns observed are generally more consistent with the influence of factors that reduced mortality simultaneously in most age groups. Additional influences, possibly smoking habits, impeded the decline of tuberculosis in older adult males, and produced the sex-specific shifts in age distributions of mortality that were previously interpreted as evidence of cohort-dependent mortality decline.

Use of an age-period-cohort model to reveal the impact of cigarette smoking on trends in twentieth-century adult cohort mortality in England and Wales

We use an age-period-cohort (APC) model to estimate the contribution of smoking-related mortality to cohort changes in adult mortality in Britain since 1950. We show that lung cancer and overall mortality can be satisfactorily modelled using cohort relative risk and a fixed age pattern. The results of the model suggest that smoking by itself can account for a substantial fraction of change in cohort mortality for those born around the first half of the twentieth century. In particular, smoking provides an explanation for the higher-than-average improvement in the mortality of both males and females born around 1930. Our confidence in the correctness of the results of the models is strengthened by the fact that they are very similar to those of the Peto-Lopez and Preston-Glei-Wilmoth models that estimate the contribution of smoking-related to overall mortality.

Human actuarial aging increases faster when background death rates are lower: a consequence of differential heterogeneity?

Many analyses of human populations have found that age-specific mortality rates increase faster across most of adulthood when overall mortality levels decline. This contradicts the relationship often expected from Williams' classic hypothesis about the effects of natural selection on the evolution of senescence. More likely, much of the within-species difference in actuarial aging is not due to variation in senescence, but to the strength of filters on the heterogeneity of frailty in older survivors. A challenge to this differential frailty hypothesis was recently posed by an analysis of life tables from historical European populations and traditional societies that reported variation in actuarial aging consistent with Williams' hypothesis after all. To investigate the challenge, we reconsidered those cases and aging measures. Here we show that the discrepancy depends on Ricklefs' aging rate measure, ω, which decreases as mortality levels drop because it is an index of mortality level itself, not the rate of increase in mortality with age. We also show unappreciated correspondence among the parameters of Gompertz-Makeham and Weibull survival models. Finally, we compare the relationships among mortality parameters of the traditional societies and the historical series, providing further suggestive evidence that differential heterogeneity has strong effects on actuarial aging.

The ‘Golden Generations’ in Historical Context

Assumptions about future mortality are more important than those for factors such as fertility, migration, disability trends or real interest rates for cost projections of the U.S. Old Age, Survivors, Disability and Health Insurance scheme. Recently, one factor has been assumed to be the key driver of future mortality in both official British population projections and actuarial ones: a ‘cohort effect’ associated with a group who were born in a period centred on the early 1930s who have been identified as having experienced particularly rapid improvements in mortality rates and are often referred to as the ‘golden generations’ or ‘golden cohorts’. The concept of ‘cohort effects’ is discussed; limitations of national-level cohort data considered; and methods for identifying such effects are reviewed. Particular attention is given to the analysis of populations which have been identified as having clear cut cohort effects; those of Britain and Sweden in the later part of the nineteenth century and early twentieth century, as well as the contemporary British population. The likely magnitude of such effects is discussed using a stylised model to assess the extent to which members of the ‘golden generations’ are especially privileged.

Forecasting Mortality, Different Approaches for Different Cause of Deaths? The Cases of Lung Cancer; Influenza, Pneumonia, and Bronchitis; and Motor Vehicle Accidents

Most of the methods of mortality forecasting have been assessed using performance on overall mortality, and few studies address the issue of identifying the appropriate forecasting models for specific causes of deaths. This study analyses trends and forecasts mortality rates for three major causes of death — lung cancer, influenza-pneumonia-bronchitis, and motor vehicle accidents — using Lee–Carter, Booth–Maindonald–Smith, Age-Period-Cohort, and Bayesian models, to assess how far different causes of death need different forecasting methods. Using data from the Twentieth and Twenty-First Century Mortality databases for England and Wales, results show major differences among the different forecasting techniques. In particular, when linearity is the main driver of past trends, Lee–Carter-based approaches are preferred due to their straightforward assumptions and limited need for subjective judgment. When a clear cohort pattern is detectable, such as with lung cancer, the Age-Period-Cohort model shows the best outcome. When complete and reliable historical trends are available the Bayesian model does not produce better results than the other models.

The relative effects of shocks in early- and later-life conditions on mortality

The relative importance of cohorts' early-life conditions, compared to later period conditions, on adult and old-age mortality is not known. This article studies how cohort-level mortality depends on shocks in cohorts' early- and later-life (period) conditions. I use cohorts' own mortality as a proxy for the early-life conditions, and define shocks as deviations from trend. Using historical data for five European Countries i find that shocks in early-life conditions are only weakly associated with cohorts' later mortality. This may be because individual-level health is robust to early-life conditions, or because at the cohort level scarring, selection, and immunity cancel each other. Shocks in period conditions, measured as deviations from trend in period child mortality, are strongly and positively correlated with mortality at all older ages. The results suggest that at the cohort level changing period conditions drive mortality variation and change.