Demographic Research

BackgroundLifespan inequality arises both from heterogeneity (e.g., in sex or race) and from unavoidable individual stochasticity. By treating a heterogeneous population as a mixture we can (and many have) partition variance in lifespan into a between-group component due to heterogeneity and a within-group component due to chance. Until now, such studies have treated factors singly. It is now possible to analyze multiple factors and their contributions to variance. ObjectiveThis paper is the first to exploit the new analysis for multi-factor studies. Multi-factor data are painfully rare, but a remarkable study by Bergeron-Boucher et al. presented U.S. life tables under all 54 combinations of four factors (sex, marital status, education, race). Our objective is to quantify the contributions of these factors and their interactions to lifespan inequality. MethodsThe population is treated as a mixture of 54 groups, with a mixture distribution either flat or proportional to population size of the different factor combinations. Components of the variance in remaining longevity, for starting ages from 30 to 85 years, are calculated using marginal mixture distributions. ResultsEven accounting for four factors and their interactions, between-group heterogeneity accounts for only 7% (population-weighted mixing) to 10% (flat mixing) of lifespan variance. Education and its interactions make the largest contribution. Contributions of two-way, three-way, and four-way interactions are orders of magnitude smaller. This suggests new ways of displaying, summarizing, and interpreting inequality as measured in multi-factor studies. ContributionMulti-factor studies can now be used to identify sources of variance in longevity and other demographic outcomes.

We investigate the consequences of the COVID-19 pandemic on other underlying causes of death in Brazil in 2020 and 2021. We estimate monthly age-standardized mortality rates for 2010-2021 and decompose those time series into three additive components: trend, seasonality, and remainder. Given the long-term trend and historical seasonal fluctuations, we assume that any impact from the pandemic will be left on the remainder. We also decompose the contributions of COVID-19 deaths (direct effect) and those from other causes (indirect effects) to the annual change in life expectancy at birth (0) from 2017 to 2021. Broadly, the remainder mirrors the trajectory of pandemic waves. The impact of the COVID-19 pandemic on other causes of death was not limited to increases but also decreases. The direct effects of the pandemic reduced 0 by 1.89 years between 2019 and 2020 and 1.77 between 2020 and 2021. Indirect effects increased 0 by 0.44 between 2019 and 2020 and had virtually no impact on 0 between 2020 and 2021. Whether trajectories in mortality rates and annual gains in 0 will quickly return to pre-pandemic levels depends on governmental actions to mitigate the consequences of the COVID-19 pandemic.

The analysis in this paper uses the new Understanding Society COVID-19 survey. The key advantage of these data is that they allow us to examine infection rates for people with particular characteristics. We study how reported symptoms vary in the population and relate reported symptoms to a positive Covid-19 test in the small sample in the survey who were tested. Combining these probabilities we find that the chances of infection increase with a persons education level, are lower and declining with age among those aged over 55, and were higher in the West Midlands and London and lower in the North East than in the rest of the country, and tended to increase with regional population density. There is also evidence that the infection rate was lower among those of a Caribbean origin. A suitably cautious estimate of the mean infection rate is that, during the period up to the end of April 2020, it was between 2% and 8%, with a central rate of about 5%.

Epidemiological studies suggest that age distribution of a population has a non-trivial effect on how morbidity rates, mortality rates and case fatality rates (CFR) vary when there is an epidemic or pandemic. We look at the empirical evidence from a large cohort of countries to see the sensitivity of Covid-19 data to their respective median ages. The insights that emerge could be used to control for age structure effects while investigating other factors like cross-protection, comorbidities, etc.

Data for ten European countries which provide detailed distribution of COVID-19 cases by sex and age show that among people of working age, women diagnosed with COVID-19 substantially outnumber infected men. This pattern reverses around retirement: infection rates among women fall at age 60-69, resulting in a cross-over with infection rates among men. The relative disadvantage of women peaks at ages 20-29, whereas the male disadvantage in infection rates peaks at ages 70-79. The elevated infection rates among women of working age are likely tied to their higher share in health- and care-related occupations. Our examination also suggests a link between womens employment profiles and infection rates in prime working ages. The same factors that determine womens higher life expectancy account for their lower fatality and higher male disadvantage at older ages.

The growing number of series on COVID-19 deaths classified by age and sex, released by national health authorities, has allowed us to compute age and sex patterns of its mortality, based on 183,619 deaths from Western Europe and the USA. We highlight the specific age schedule of COVID-19 mortality and its pronounced excess male mortality and we then apply these COVID-19 death rates to world populations, in 2020. Our results underscore that considerable variations exist between world regions, as concerns the potential impact of COVID-19 mortality, because of their demographic structures. When compared to younger countries in Sub-Saharan Africa, the vulnerability to COVID-19 mortality is shown to be 17 times higher in several industrialized countries of East Asia and Europe. There is a high correlation (r2= .44) between demographic vulnerability to COVID-19 mortality and current COVID-19 death rates. AbstractCOVID-19, mortality, age structures, death rates, Europe, USA. BackgroundThe data available on infection and death rates from COVID-19 have pointed to the elderlys vulnerability to pandemics, especially elderly men. However, current models have not yet incorporated the growing volume of information on deaths by age and sex that is being released by statistical offices and health authorities. These newly available data allow us to examine the specific age and sex patterns of COVID-19 mortality and to estimate the impact of specific demographic structures on potential COVID-19 mortality worldwide. MethodsWe use the data available on May 15, 2020, from the nine countries with the largest series of deaths, disaggregated by age and sex: Belgium, France, Germany, Italy, Netherlands, Spain, Sweden, the UK, and the USA. Using 183,619 deaths (60.2% of all currently estimated COVID-19 deaths), we estimate the sex-specific death rates, by 5-year ranges, for two large death samples: USA and Western Europe. We compare these mortality rates with Gompertz models and with life tables of the worlds population, estimated by the United Nations. We apply these COVID-19 mortality rates by sex and 5-year group to the 2020 age and sex structures of world countries and regions, and obtain an index summarizing the relative magnitude of their potential vulnerability to COVID-19 mortality because of their demographic structures. FindingsCOVID-19 death rates cannot be computed below age 15. COVID-19 death rates from age 15-19 years to 90+ increase by a factor of 3 of every ten years, at a rate that is faster than general mortality. Male mortality from COVID-19 is systematically higher than female mortality, with a peak of excess male mortality occurring among 55-59-year-olds. Age and sex structures show considerable variations across countries, in terms of vulnerability to COVID-19 mortality. It transpires that the youngest countries in Central Africa are 17 times less vulnerable than aging countries, such as Japan. InterpretationWhereas the true intensity of the ongoing COVID-19 pandemic remains underestimated by existing statistics, this unique mortality dataset shows that the regularity of the distribution of COVID deaths by age and sex is in line with the standard Gompertz mortality equation; thus confirming the quality of the first death samples and the unique age and sex patterns of COVID-19 mortality. COVID-19 death rates tend to be negligible below age 15 and cannot be used for analysis. The rate of progression of death rates by age is faster than that of general mortality. This feature places the elderly population in a particularly vulnerable situation compared to younger adults. Male excess mortality from COVID-19 also appears far more pronounced than in general mortality patterns, with men aged 40-59 years being almost 2.5 times more likely to die than women of the same age. Our analysis also points to considerable variations between world regions, as concerns the potential impact of COVID-19 mortality, because of their demographic structures. The COVID-19 structural vulnerability index ranges from .28 in Western or Middle Africa to 2.6 in Southern Europe. There is a high correlation between demographic vulnerability to COVID-19 mortality and current COVID-19 death rates (r2= .44 for 188 countries).

Conventional measurements of aging do not take into account the dynamic nature of aging-related characteristics over time. Therefore, in order to refine the estimates of aging, demographers have proposed prospective measures based on remaining life expectancy, Sanderson and Scherbov (2007). We compared these new measures with conventional aging measures using the data from the Census of India 2011 and Sample Registration System life tables 2009-2013. In conventional aging measures, we used life expectancy at age 60 and the old age dependency ratio (OADR), whereas for new measures of aging, we applied the threshold of old age based on the remaining life expectancy and prospective old age dependency ratio (POADR). Both measures of aging provided different estimates of the aging population at the national and subnational levels. At national level, application of prospective measures increased the number of older dependents from 66.4 million to 71.8 million (OADR: 8.6% vs. POADR: 10.6%). We observed profound variation at sub-national level in India. We also observed that the prospective ageing measures not only provided higher estimates of ageing burden in India, but also altered the gender and rural urban differential in ageing. Considering the heterogeneity of life expectancies across Indian states, prospective measures provide more accurate refined estimates of aging burden in India as they are based on length of life expectancy. Application of these measures has great policy relevance in India.

Using data from two generations of British women followed from birth through childhood and into adulthood, we investigate risk factors for the onset of natural menopause before the age of 45 (known as early menopause). We focus on key stages during the life course to understand when risk factors are particularly harmful. We find that earlier cessation of menstruation is influenced by circumstances at birth. Women born in lower social class families, whose mother smoked during the pregnancy or who were short-term breastfed (one month or less) were more likely to undergo menopause before 45. Early menopause is also associated with poorer cognitive ability and smoking in childhood. Adult health behaviour also matters. Smoking is positively correlated with early menopause, while regular exercise (one to several times a week) and moderate frequency of alcohol drinking (one to three times a month) in womens early thirties are associated with a reduced risk of early menopause. The occurrence of gynaecological problems by womens early thirties is also linked to early menopause. We note that some of these factors (e.g. health behaviours) are modifiable and thus the risks may be preventable.

Age is a key variable for sciences and public planning. The demographic consequences of not measuring age correctly are manifold, including errors in mortality rates and population estimates, particularly at older ages. It also affects public programs because target populations depend on reliable population age distributions. In Brazil, the start of the vaccination campaign against COVID-19 marked the collection of new administrative data. Every citizen must be registered and need to show an identity document to get vaccinated. The requirement of proof-of-age documentation provides a unique opportunity for measuring the elderly population using a different database. This article examines the reliability of age distributions of men and women 80 years and older. We calculate various demographic indicators using data from the vaccination registration system and compare them to those from the target population estimates of the National Vaccination Plan, censuses, and population projections for Brazil and countries with high-quality population data. We show that requiring proof-of-age, such as in the vaccination records, increases data quality, mainly through the reduction of age heaping and age exaggeration. However, I.D. cards cannot fully solve wrong birth dates that result from weak historical registration systems. Thus, one should be careful when using estimates of the old age population living in some of the Brazilian regions, particularly the North, Northeast, and Center-West. Also, our analysis reveals a mismatch between the projected population by age, sex, and region, which guided the vaccination plan, and the number of vaccinated at ages 80 and older. The methodology developed to adjust the mortality rates used in the demographic projections is probably the main factor behind the disparities found.

On September 22nd the US officially recorded 200,000 COVID-19 deaths. It is unclear how many deaths might have been expected in the case of an early and effective response to the pandemic. We aim to provide a best-case estimate of COVID-19 deaths in the US by September 22nd using the experience of Germany as a benchmark. Our methods accommodate the differences in demographics between Germany and the US. We match cumulative incidence of COVID-19 deaths by age group in Germany to non-Hispanic whites in the US and project the implied number of deaths in this population and among the black and Hispanic populations under observed racial/ethnic disparities in cumulative COVID-19 mortality in the US. We estimate that if the US had been as successful as Germany in managing the pandemic we would have expected 22% of the deaths actually recorded. The number of deaths would have been lower by a further one-third if we could have eliminated racial/ethnic disparites in COVID-19 outcomes. We conclude that almost 80 percent of the COVID-19 deaths in the US by September 22nd could have been avoided with an early and effective response producing similar age-specific death rates among non-Hispanic whites as in Germany.

BackgroundPrevious kinship models analyze female kin through female lines of descent, neglecting male kin and male lines of descent. Because males and females differ in mortality and fertility, including both sexes in kinship models is an important unsolved problem. ObjectivesThe objectives are to develop a kinship model including female and male kin through all lines of descent, to explore approximations when full sex-specific rates are unavailable, and to apply the model to several populations as an example. MethodsThe kin of a focal individual form an agexsex-classified population and are projected as Focal ages using matrix methods, providing expected age-sex structures for every type of kin at every age of Focal. Initial conditions are based on the distribution of ages at maternity and paternity. ResultsThe equations for two-sex kinship dynamics are presented. As an example, the model is applied to populations with large (Senegal), medium (Haiti), and small (France) differences between female and male fertility. Results include numbers and sex ratios of kin as Focal ages. An approximation treating female and male rates as identical provides some insight into kin numbers, even when male and female rates are very different. ContributionMany demographic and sociological parameters (e.g., aspects of health, bereavement, labor force participation) differ markedly between the sexes. This model permits analysis of such parameters in the context of kinship networks. The matrix formulation makes it possible to extend the two-sex analysis to include kin loss, multistate kin demography, and time varying rates.

The COVID-19 pandemic has had overwhelming global impacts with deleterious social, economic, and health consequences. To assess the COVID-19 death toll researchers have estimated declines in 2020 life expectancy at birth. Because data are often available only for COVID-19 deaths, the risks of dying from COVID-19 are assumed to be independent of those from other causes. We explore the soundness of this assumption based on data from the US and Brazil, the countries with the largest number of reported COVID-19 deaths. We use three methods. One estimates the difference between 2019 and 2020 life tables and therefore does not require the assumption of independence. The other two assume independence to simulate scenarios in which COVID-19 mortality is added to 2019 death rates or is eliminated from 2020 rates. Our results reveal that COVID-19 is not independent of other causes of death. The assumption of independence can lead to either an overestimate (Brazil) or an underestimate (US) of the decline in e0, depending on how the number of other reported causes of death changed in 2020.

Keyfitz entropy is a widely used metric to quantify the shape of survivorship of populations, from plants, to animals, and microbes. Keyfitz entropy values < 1 correspond to life histories with an increasing mortality rate with age (i.e., actuarial senescence), whereas values > 1 correspond to species with a decreasing mortality rate with age (negative senescence), and a Keyfitz entropy of exactly 1 corresponds to a constant mortality rate with age. Keyfitz entropy was originally defined using a continuous-time model, and has since been discretised to facilitate its calculation from discrete-time demographic data. In this short note, we show that the previously used discretisation of the continuous-time metric does not preserve the relationship with increasing, decreasing, or constant mortality rates. To resolve this discrepancy, we propose a new discrete-time formula for Keyfitz entropy for age-classified life histories. We show that this new method of discretisation preserves the relationship with increasing, decreasing, or constant mortality rates. We analyse the relationship between the original and the new discretisation, and we find that the existing metric tends to underestimate Keyfitz entropy for both short-lived species and long-lived species, thereby introducing a consistent bias. To conclude, to avoid biases when classifying life histories as (non-)senescent, we suggest researchers use either the new metric proposed here, or one of the many previously suggested survivorship shape metrics applicable to discrete-time demographic data such as Gini coefficient or Hayleys median.

Analysis of some extensive individual-record data using a demographically informed model suggests constructing a general population model in which the lifetime of a person, beyond a certain threshold age, follows an extreme value distribution with a finite upper bound, and with that upper bound randomized over the population. The resulting population model incorporates heterogeneity in life-lengths, with lifetimes being finite individually, but with extremely long lifespans having negligible probability. Our findings are compared in detail with those of related studies in the literature, and used to reconcile contradictions between previous studies of extreme longevity. While being consistent with currently reported analyses of human lifetimes, we nevertheless differ with those who conclude in favour of unbounded human lifetimes.

The behavioral economics literature suggests that exposure to traumatic events shifts preference features like risk aversion and time preference. Drawing on this literature, this study explored the relationship between early life exposure to traumatic events and self-control at older ages. The data were obtained from the Health and Retirement Study, which offers retrospective data on trauma exposure and a measure of self-control. The results showed that the experience of serious physical attacks or assaults was associated with a 3.1% reduction in self-control, above and beyond the influence of demographic characteristics and childhood socioeconomic disadvantages. The mean number of years elapsed since the physical attack was about 30, conditional on exposure, indicating that traumatic experiences early in life exert a lasting influence on self-control. Our findings were consistent with evidence that experiences of natural disasters and armed conflicts increased impatience among survivors. JEL classification codeD12, D14, D91

Population ageing increases the importance of cognitive capacity for making decisions about retirement and living independently beyond it. We tested whether post-war educational expansion and working-life social mobility eliminate the association between social class of origin and cognition in early old age using the 1958 National Child Development Study. Two outcomes were analysed at age 62: standard episodic memory (immediate + delayed word recall) and long-term episodic memory, capturing accurate half-century recall of childhood household facts (rooms and people at age 11 validated against mothers' responses). Social mobility trajectories derived in prior work were classified into predominantly manual versus non-manual class trajectories. Models were estimated separately for women and men across three specifications: (i) social origin and controls, (ii) adding social mobility, and (iii) adding weighting to address healthy survivor bias. Education was consistently associated with both outcomes. For long-term episodic memory, social origin gradients were clearer than for short-term episodic memory, with men from service/professional origins showing a 13 percentage-point higher probability of accurate half-century recall than men from manual origins. These findings indicate that education expansion and working-life social mobility failed to release the grip of social origin on long-term episodic memory.

BackgroundSubjective Survival Probabilities (SSP) are known to be associated with mortality but little is known about the relationship they might have with employment categories and job satisfaction. We assess such a relationship looking at the fifty-plus population in Japan that is characterized by a stratified labour market for the older workers and high working time intensity. MethodWe use the four waves (2007-2013) of the Japanese Study of Aging and Retirement (JSTAR), a panel dataset tracking 7,082 50-plus respondents in 10 Japanese prefectures. We use a mixed-effects quantile regression model to investigate the relationship between SSP and employment status (model 1) and job satisfaction (model 2). Both models additively control for demographic and socio-economic cofounders as well as other health measurements. Multiple imputations are used to correct sample attrition. ResultsIn model 1, retirement (-0.27, 95%CI =-0.51;-0.03) and contract work (-0.51, 95%CI=-0.79;-0.23) are negatively associated with SSP in comparison with full-time employment. In model 2, low job satisfaction appears to be strongly associated with SSP (-1.37, 95%CI=-1.84;-0.91) in comparison with high job satisfaction. The same trend is observed regardless of the way job satisfaction is calculated. Both working time and employment category are not significantly associated with SSP after controlling for job satisfaction which indicates that job satisfaction is a main driver of SSP discrepancies. DiscussionSSP variations can be explained by employment category with contract work more at risk. Job dissatisfaction is a main explanation of low SSP. Both work and employment explain SSP variations.

It is natural to question the impact of COVID-19 on life expectancy. However, a newborn during the 2020-2021 period need not experience the same level of adult mortality found in 2020-2021 because there may be zero COVID-19 related deaths when the newborn reaches adulthood. Thus, life expectancy lost due to COVID-19 cannot be found simply by incorporating excess deaths due to COVID-19 and re-doing the life table computations because: (1) we know that the COVID-19 deaths need not occur every year for the next 20-25 years; and (2) once an adult, a newborn in 2021/2022 need not experience the same mortality rate that current middle and older aged COVID-19 patients experience. Using U.S. data as an example, we estimate an average of 29.68 years of life was lost to those aged 18-64 who died from COVID-19 in the U.S., noting that 74 % of the reported deaths of 18-64 occurred among 50-64 years and 10 % below 40 years. Instead of computing life expectancy years lost due to COVID-19, we recommend computing life years lost due to COVID-19.

Mathematical kinship demography is an expanding area of research. Most models explore the expected number of kin without accounting for demographic stochasticity. Recently, a paper provided a method to calculate the complete number-distribution of kin in a one-sex time-invariant demography. We extend this method to the case of two-sexes and to time-variant demographic rates. Drawing from the mathematical tools of Fourier and convolution theory as well as basic probability and matrix algebra, we derive closed form expressions which capture the recursive nature of kin replen-ishment, generation-by-generation. Formulae presented here extend arbitrary genealogical distances to recover relatives considered in the leading frameworks of kinship. All we require as inputs are age, sex, and time-specific mortality and fertility schedules. This research presents the first kinship model able to predict the probable numbers of relatives, structured by age and sex within a time-varying demography. As well as producing the probable numbers of living kin, the model flexibly extends to give the probable numbers of deaths an individual experiences. Such a detailed analysis of the kin-network will be useful in many fields.

The COVID-19 pandemic, caused by tens of millions of SARS-CoV-2 infections world-wide, has resulted in considerable levels of mortality and morbidity. The United States has been hit particularly hard having 20 percent of the worlds infections but only 4 percent of the world population. Unfortunately, significant levels of misunderstanding exist about the severity of the disease and its lethality. As COVID-19 disproportionally impacts elderly populations, the false impression that the impact on society of these deaths is minimal may be conveyed by some because elderly individuals are closer to a natural death. To assess the impact of COVID-19 in the US, I have performed calculations of person-years of life lost as a result of 194,000 premature deaths due to SARS-CoV-2 infection as of early October, 2020. By combining actuarial data on life expectancy and the distribution of COVID-19 associated deaths we estimate that over 2,500,000 person-years of life have been lost so far in the pandemic in the US alone, averaging over 13.25 years per person with differences noted between males and females. Importantly, nearly half of the potential years of life lost occur in non-elderly populations. Issues impacting refinement of these models and the additional morbidity caused by COVID-19 beyond lethality are discussed.