Since the beginning of the COVID-19 pandemic, only few studies on excess mortality have considered both cause-specific and sub-national differences. Located at the intersection of the European north-south and east-west gradients of (excess) mortality, Germany represents a fascinating context for such detailed analysis, as the German example might provide implications for the overall European pattern.

Our analyses rely on official cause-of-death statistics consisting of 7.74 million individual death records reported for the German population during 2015–2022. We conduct differential mortality analyses by age, sex, cause, month and district (N=400), using decomposition and standardisation methods, comparing each strata of the mortality level observed during 2020–2022 with its expected value.

Our results show remarkable spatial differences to the disadvantage of the south of eastern Germany in both 2020 and 2021. Excess mortality in the most affected districts is driven widely by older ages and deaths reported during the second wave, particularly from COVID-19 but also from cardiovascular and mental diseases. In 2022, however, the spatial pattern completely changed with the northwest showing the highest levels of excess mortality, while the east widely experienced a rise again in life expectancy.

The results for 2020 and 2021 suggest that increased psychosocial stress influenced the outcome of excess mortality in the most affected areas during the second wave of the pandemic. Cause-specific and seasonal data for 2022 will become available by March 2024, hopefully in time for DGD 2024, and will help us understand the fundamentally changed pattern of excess mortality.

Unlike other destination countries, the overall impact of migrants on life expectancy in Sweden has not historically been positive. However, this varies depending on the country of origin of the migrants. The trend started to change in 2019. Immigrants began to positively contribute to the increase in life expectancy in Sweden, and researchers predicted that this trend would continue, except in 2020 the COVID-19 pandemic appeared.

Sweden adopted a distinctive approach to the pandemic, leading to an overall increase in mortality and a decrease in life expectancy for males and females. Prior research indicates that immigrants are more likely to die from COVID-19 in several countries, especially those from non-Western nations.

Our first aim is to understand whether the recent emergence of a positive impact of immigrants on national life expectancy in Sweden can be attributed to refugees. Swede has a diverse immigrant population in terms of country of origin and reasons for migration. As previously mentioned, life expectancy among migrants varies significantly based on country of origin, but it could change due to the reason for migration. Refugee health is more at risk than that of other migrants since the entire migration process—including leaving the country, travelling, and requesting asylum in the destination country— is associated with stressful and risky circumstances.

Our second aim is to understand how much the COVID-19 pandemic and its disproportionate impact upon international immigrants, interrupted the positive contribution that migrants started to make to life expectancy levels.

We fulfil these two aims we will use Swedish register population data that allows very detailed information.

There are claims of significant excess mortality among children (ages 0 to 14) in Europe during 2021 and 2022. Evaluation of raw data and excess mortality analysis show that children deaths in 2021-2022 are completely in-line with deaths from previous years in most European countries, and in many cases even lower. Across 35 European countries, among children ages 0 to 14 up to July 31st 2022 there are 330 less deaths than expected, with a confidence interval of [-770, 104].

Understanding the relationship between life expectancy at birth (e₀) and the gross domestic product per capita (GDPpc) is relevant for cohesion policies in the European Union (EU) because it might imply that economic convergence (or divergence) is accompanied by narrowing (or widening) health gaps. Previous studies have studied the association between GDPpc and e₀ almost exclusively based on national data. It is certainly more appropriate, however, to study the relationship at the subnational level because levels and trends in both variables, e₀ and GDPpc, vary substantially across Europe’s regions. Accordingly, the aim of our study is examining, whether economic performance of regions predict their e₀ level. We build Preston curves from regression models using regional data for 19 European countries, divided into 535 regions. Mortality data comes from statistical offices and GDPpc can be obtained from the Eurostat database. The period 2005 to 2018 is particularly interesting as it immediately follows the EU enlargement to central-eastern European countries in 2004. In our analysis, spatial units refer to NUTS-2 or NUTS-3 regions, depending on the size of the country. Our preliminary results suggest that there is indeed a positive association between GDPpc and e₀. Similarly, to Preston’s original analysis, we observe an upward shift in the curve, indicating that factors exogenous to a region’s GDPpc level play an important role in explaining e₀ gains as well. In the next step, we plan to employ panel data regression techniques to assess the impact of GDPpc on mortality for the period 2005 to 2018. We are confident that our comprehensive dataset and suggested robust regression techniques will enable us to further examine whether levels and trends in e₀ can be linked to the economic development of European regions.

The social gradient in mortality was much more modest in historical Sweden than it is contemporarily, and for men was reversed with higher mortality for white-collar workers than for blue-collar workers before the second world war. In this paper, we present evidence that even in the absence of a modern social gradient in mortality, family members share a survival advantage at adult ages across generations. Social, behavioral and biological factors that promote good health accumulate in families, even if the mechanisms that promote long lives change across time.

We use a unique dataset consisting of digitized and linked historical records for a region in Southern Sweden, reconstructing lives and families of individuals living in five rural parishes and a town 1900-1967 with nationwide follow-up in the Swedish national registers of these individuals and their descendants 1968-2015. We show evidence that individuals (age 30-90) from well-performing families have a mortality advantage across time, even in the absence of a modern social gradient in mortality. Analyses of cause-specific mortality shows that both preventable and non-preventable disease mortality is reduced in descendants of long-lived families, providing evidence that both behavioral and non-behavioral factors are involved.

There has been a long-standing debate concerning the relative importance of space and place vs. individual characteristics for mortality. In this paper, we analyze the association of lifetime SES-status/income and mortality at older ages among retired German men and its variation at the sub-national level. Does the income gradient in mortality differ across sub-national macro-regions and types of settlement? To answer this question, we employ a large administrative dataset of the German Pension Fund consisting of 17 Mio person-years of exposure and 585.9 thousand deaths that have occurred over the period 2012–2017 among men aged 65–84. We estimate relative mortality risk using a Cox proportional hazard model. To quantify the steepness of the gradients and be able to compare them between different places of Germany we estimate the Slope Index of Inequality and the Relative Index of Inequality. Our preliminary results suggest a linear income-mortality relationship across all macro-regions and types of settlement: the risk of dying decreases as income increases. However, the degree of inequalities between different settlements and the steepness of mortality gradients vary substantially within the country. In particular, regardless of the sub-national macro-region, big cities reveal the highest degree of disparities across income groups.

A precise estimate of the target population is inherently important in population statistics. However, factors such as increased immigration, and few incentives for deregistration after emigration mean that population registers may not always accurately reflect the target population. This study aims to identify unregistered emigration using “signs of life”. That is, detecting historical inactivity of individuals who have emigrated, but are still listed as residents in the population register. Unregistered emigration contributes to over-coverage, as the number of actual emigrations exceeds the number of registered emigrants. This estimation error affects not only size and composition of the population, but also impacts demographic indicators, such as death and fertility rates. Statistics on households and families may also become skewed due to these discrepancies. There is still no consensus on how to identify or deal with unregistered emigration. Addressing this, we first provide a comparison of methods adapted from the literature for estimating the number of unregistered emigrations. The Zero-Income Approach provides a method with minimal computational and data quality requirements, which serves as a foundation for the estimation. The Household-Income Approach builds upon this by correcting for household income factors. Finally, the Register Trace Approach provides the most comprehensive and detailed picture of unregistered emigration. Our estimates suggest that unregistered emigrants account for approximately 0.44 percent of the adult population in Norway. Second, we analyse the demographic characteristics of the non-deregistration group. We find that the problem of unregistered emigration is not equally distributed across the population, indicating that some subgroups are more prone to discrepancies than the rest of the population. Among immigrants, the over-coverage due to unregistered emigration is substantially higher, accounting for 2.29 percent of the population.

The German cause-of-death statistics are often used to draw conclusions about the health status of the population and the significance of certain diseases. Unfortunately, cause-of-death statistics - not only in Germany - often show a relatively high proportion of ill-defined deaths. Ill-defined deaths have an invalid or unspecific ICD code as underlying cause of death. This may be the case when the indicated ICD code is intermediate (e.g. heart failure) or non-specific (e.g. unspecified cancer). These ICD codes are not informative for public health planning and for example in the context of burden of disease calculations. They do not adequately reflect the underlying cause of death.

The Global Burden of Disease Study (GBD) of the Institute for Health Metrics and Evaluation (IHME) has a specific list of ICD codes that shall be considered invalid resulting in an ill-defined death. In 2015 the proportion of invalid codes in Germany was 26,6%, in 2017 it was 26.0%.; with quite substantial regional variation (Wengler et al. 2019). Following the GBD classification we want to up-date this analysis, looking at the time from 2012 to 2021 and the share of ill-defined deaths in the German federal states.

In general, a further decrease in the share of ill-defined deaths is expected. Especially since more federal states use automatic (re-)coding offered in the Iris/MUSE-system, which incorporates the WHO rules for coding of causes of death. Having less ill-defined deaths and hence better quality of causes of death data is of high importance for public health planning and efficient measurements.

Background: In Germany, the majority of long-lived individuals (LLI) aged 85+ suffer from dementia at the time of their death. Additional medical costs of people with dementia (PwD) are mainly caused by differences in hospital care. The aim of this study is how the risk of hospitalization of LLI changes with age and age at death .

Methods: We drew a random sample of quarterly data from all AOK insured persons aged 50+ (N=250,000) in 2004 with follow-up to 2019 and followed the 1918 to 1923 birth cohort (n=4,067 males and 13,303 females), who reached age 85 years between 2004 and 2009, to the end of the study period or to death. We estimated a multivariate logistic regression model, clustering variances by person ID, to examine the simultaneous effects of age, age at death and last year of life on the risk of hospitalization, stratified by PwD and non-PwD.

Results: Overall, 41.23% of men and 48.37% of women had received a dementia diagnosis, and more than half had been hospitalized (men: 56.41%, women 57.19%). PwD were more likely to be in hospital (men: 68%, women: 68%) than non-PwD (men: 46%, women: 53%). In the multivariate analysis (Table 1), the risk of hospitalization increased significantly with each year of age, by 7.6% for PwD and by 9.3% for non-PwD. With increasing age at death, the risk decreased significantly by 11% for both. These trends levelled off at the highest ages. In the last year of life, the risk of hospitalization increased more for non-PwD (OR=7.36) than for PwD (OR=4.79); women had a significantly lower risk.

Conclusion: LLI have lower hospitalization risks the later they die, both in people with and without dementia.

Governments are grappling with demographic shifts, such as an aging population and rising old-age dependency, prompting discussions on delaying retirement age. To safeguard vulnerable groups from prolonged unemployment, policies promoting longer careers need careful planning. Existing research has centred on predicting work-life expectancy, serving as a foundation for policy development. This study extends this work, adopting a life course perspective, by recognizing that retirement is a gradual process characterized by complex and multiple transitions. Using comprehensive data from the German Pension Insurance, we employ advanced machine learning techniques like sequence-to-sequence Transformers and LSTM models, along with sequence analysis, to predict and analyse work-life trajectories from ages 50 to 65. These analyses provide valuable insights for understanding complex retirement transitions and show accurate predictions at the aggregate level.

This paper extends the Lee-Carter model to include vanishing jumps on mortality rates, where
the impact is highest in the beginning and then gradually diminishes. Existing models either account
for transitory jumps over one period or permanent shifts. However, there is no literature on estimat-
ing mortality time series with jumps that have a vanishing effect over a small number of periods,
as is typically observed in pandemics. Using COVID-19 data, our proposed Bayesian model out-
performs transitory shock models in terms of in-sample fit, thus providing a more comprehensive
representation of mortality rates during a pandemic.

The Nordic countries have experienced rapid and unexpected fertility declines in the last decade. Future paths of fertility are a key input when charting the sustainability of social security systems. Hence realistic views of possible future paths of fertility, including the uncertainty regarding these paths, is critically important for economic and social planning in the Nordic countries.
An almost universal finding of probabilistic forecasting is that model-based prediction intervals are too narrow. The result is a misleadingly confident view of the future. We approach the problem of prediction uncertainty by generating empirical prediction intervals for our forecasts of Nordic period fertility up to 2050.
We derive the empirical prediction intervals for our forecasts from an out-of-sample cross-validation, utilizing historic forecasting errors. These quantify how uncertain we should be about the forecasts given the demonstrated forecast performance of the past. We then validate the calibration of the empirical prediction intervals and compare them to the performance of model-based prediction intervals. We use the following forecast models applied to data from the Human Fertility Database: First, a na"{i}ve approach fixing the last observed Total Fertility Rate (TFR). Second, a GARCH model of the TFR. Third, the Lee-Carter method for age-specific fertility rates. And fourth, a probabilistic, random-walk based forecast around the assumption that the ongoing fertility postponement would gradually slow down and come to an end by 2050.
In addition to presenting forecasts of Nordic fertility up to 2050, we expect to find that the empirical prediction intervals provide a more realistic view of the forecast uncertainty, exhibiting wider prediction intervals in comparison to the model-based prediction intervals.

After preliminary attempts at the end of the 20th century to compute probabilistic population forecasts (by Cohen, Keyfitz, Lee & Tuljapurkar, Alho, and others), the field became fully developed in the past two decades. I give a brief overview of the various applications of probabilistic demographic forecasts, spanning from national, subnational, and multi-country populations to the labour market, households, immigrants, and long-term care. I sketch the development from a frequentist to a Bayesian approach. Finally, I evaluate ex-post facto the predictive quality of selected population and household forecasts for Norway, the Netherlands, and France.

Population projections for small geographical areas are challenging even when data availability is good. Despite the presence of register data in Norway the current municipality level population projections by Statistics Norway are not satisfactory and are in the process of being replaced from a cohort-component framework to microsimulation. We propose a simpler and generalizable approach for downscaling national level population projections into municipality level projections, leveraging Norwegian register data and other data sources using an innovative neural network-based machine learning model. An additional advantage of this downscaling approach is that additional dimensions can easily be added to sub-national projections. We show this by disaggregating the Wittgenstein Centre population projections. The machine learning model is also trained by categorizing municipalities by special economic activities that might affect the population structure in that area. Such activities are the presence of fish farming, oil production, universities, or a high concentration of agricultural production.