Estimating trends in working life expectancy based on health insurance data from Germany – Challenges and advantages

Against the backdrop of population aging and growing strain on pension systems, monitoring the development of Working Life Expectancy (WLE) is vital to assess whether the policies taken are effective. This is the first study investigating time trends and educational inequalities in WLE based on German health insurance data. The analyses are based on the data of the AOK Lower Saxony (N = 3,347,912) covering three time periods (2006-08, 2011-13, and 2016-18). WLE is defined as years spent in the labor force (i.e. in employment and unemployment) and was calculated for each age between 18 and 69 years for the three periods to depict changes over time using multistate life table analysis. Educational inequalities in 2011-13 are reported for two educational levels (8–11 years and 12–13 years of schooling). WLE increased in both sexes with increases being stronger among women. This holds irrespective of whether WLE at age 18 (35.8–38.3 years in men, 27.5–34.0 years in women) or the older working-age (e.g. at age 50 10.2–11.7 years in men, 7.8–10.5 years in men) is considered. Among women at all ages and men from their mid-20s onwards, WLE was higher among higher-educated individuals. Inequalities were most pronounced among women (e.g. Δ3.1 years in women, Δ1.3 years in men at age 50). The study supports previous research indicating that measures to extend working life are effective, but that noticeable inequalities in WLE exist. Health insurance data represent a valuable source for such research that has so far remained untapped. The data provide a suitable basis to investigate trends and inequalities in WLE. Future research should build on the strengths of the data by broadening the research towards a more comprehensive analysis of the development of WLE from a health perspective.

•

Working Life Expectancy has increased substantially since the mid-2000s.

•

The gender gap narrowed over time, but men still spend considerably more years in the labor force than women.

•

Educational inequalities in WLE exist in both sexes, but are larger in women.

•

Measures to increase labor force participation should focus especially on lower-educated women, as WLE is low in this group.

•

Prevention should be strengthened for vulnerable groups to reduce health inequalities and early labor market exits.

Development of Comorbidities in Type 2 Diabetes: Morbidity Expansion or Dynamic Equilibrium?

Background

Different hypotheses have been proposed about the future development of morbidity associated with the temporal improvement in life expectancy. However, many studies have focused on general morbidity while testing these hypotheses. In type 2 diabetes (T2D), one of the most prevalent chronic diseases, research points towards an increasing prevalence over time with a simultaneous increase in life expectancy among this population. While morbidity compression in T2D can thus be ruled out, it remains unclear whether morbidity expansion or dynamic equilibrium applies, which could be the case with a decrease in personal disease burden despite increasing prevalence. Against this background, this study aims to examine how the prevalence of T2D-related comorbidities is developing over time.

Methods

Using claims data of a large statutory insurance provider in the state of Lower-Saxony, Germany, the period prevalence of nine T2D-related comorbidities was examined for the time periods 2005-2007, 2010-2012, and 2015-2017 in 240241, 295868, and 308134 individuals with T2D, respectively. The temporal development of comorbidities was examined by logistic regression analysis. The change in the number of comorbidities over time was examined by ordered logistic regression. Analyses were stratified by gender and three age groups.

Results

Over the three time periods, age-adjusted predicted probabilities for more severe cardiovascular diseases (CVDs) decreased in men and women with T2D while those for less severe CVDs and other vascular diseases such as retinopathy, polyneuropathy any nephropathy increased significantly in all subgroups. Among all subgroups, the predicted probability of having more comorbidities over time also increased significantly.

Conclusions

Despite differences in the way CVDs are developing, the results are in favour of the morbidity expansion hypothesis for this population. Further research is needed to examine the reasons behind the observed trends.

Key messages

The developement of comorbidities in individuals with type 2 diabetes points towards morbidity expansion among this population. Future studies should examine whether the change is socioeconomic status, medication use and lifestyle risk factors are contributing to these trends.

Backfitting tests in generalized structured models

We introduce bootstrap tests for semiparametric generalized structured models. These can be used for testing different kinds of model specifications like separability, functional forms and homogeneity of effects, or for performing variable selection in a large class of semiparametric models. The test statistics are based on the comparison of non- and semiparametric alternatives in which both the null hypothesis and the alternative are non- or semiparametric. All estimators are obtained by smooth backfitting. Simulation studies show excellent performance of the test procedures.

Socioeconomic inequalities in type 2 diabetes among different population subgroups

Background

Type 2 diabetes (T2D) is a rising global epidemic with lower socioeconomic status (SES) groups being more affected. Considering specific population subgroups to examine prevalence and SES inequalities in T2D is rare. In addition, using only one indicator to depict SES inequalities in health has been a common practice despite evidence on differences in what different indicators measure. This study examines the prevalence of and SES (school education, occupation and income) inequalities in T2D in the three population subgroups: employed individuals, nonworking spouses and pensioners. This study also determines the SES indictor with the highest explanatory power.

Methods

This study is based on claims data from a statutory health insurance provider in Lower Saxony, Germany. T2D prevalence in the period between 2013 and 2017 was examined in 1,345,841 employed individuals, 180,949 nonworking spouses and 773,427 pensioners. Multivariate logistic regression analysis was applied to examine SES inequalities in T2D in the three subgroups. Explanatory power of the three SES indicators was compared by deviance analysis.

Results

T2D prevalence was four times higher in male nonworking spouses (24.2%) and 2.6 times higher in female nonworking spouses (12.7%) compared to employed men (6.4%) and women (4.7%) respectively, while it accounted for 40% of men and 36% of women in pensioners. Clear gradients in T2D inequalities emerged for all three SES indicators and were observed in the three population subgroups. School education had the highest explanatory power in employed men and women and male nonworking spouses.

Conclusions

Nonworking spouses are an important target group in T2D prevention interventions. The three SES indicators differ in their explanatory power where low school education appears to be a major risk factor. It can be discussed that health literacy and the associated health behavior play a role in mediating the association between education and T2D.

Key messages

The population subgroup “nonworking spouses” is an important target group for type 2 diabetes prevention interventions. The level of school education is a substantial determinant of socioeconomic inequalities in type 2 diabetes.

[Why do mothers smoke? Analyzing the influence of living circumstances and psychological factors on tobacco consumption among mothers with minor children]

Using a cross-sectional population sample of 3,129 women with minor children, this paper analyzes the impact of social and psychological factors on maternal smoking. Pearson's χ(2) was used for testing significance of differences between smoking prevalence and social as well as psychological factors. Adjusted effects of these factors were calculated by means of multiple logistic regression analysis. About 30% of mothers smoked daily and about 11% were heavy smokers ( ≥ 20 cigarettes/day). Lower educational degree was the most important predictor for daily tobacco consumption. But also fulltime employment, low income, distress due to conflicts with (former) partner, job-related burdens, and single motherhood enhanced the risk of maternal smoking, while household-related stress and having a young child was associated with less tobacco use. With respect to personal factors, dysfunctional coping patterns and depression showed a significant impact on tobacco use. Against expectations, active coping styles were not associated with lower but with increased smoking rates. In conclusion, it can be stated that maternal smoking has multiple causes, and that intervention programs should target both current living circumstances and psychological factors.

[Social living conditions and health among mothers in Germany : findings from a population sample]

This paper reports findings from a population sample of mothers with underage children living in Germany (n=3,129). The objective of the study was to analyze whether social and familiar living conditions are associated with enhanced health risks for mothers. The sample is representative with respect to German federal states, school education, marital status, age of mothers, and number of children. Health problems were assessed in terms of physical disabilities and discomforts, anxiety and depression, and self-rated health. About 27% of mothers perceived physical disabilities and discomfort, 21% reported high levels of anxiety, and 22% high levels of depression. About 6% assessed their health as poor or very poor. Particularly single motherhood, unemployment, sole responsibility for household and family as well as low income and low school education increased the risk of poor health. The findings suggest that for specific subgroups of mothers parenting may be a vulnerable phase of life reflecting the need for a life stage orientation in health inequality research.

[Continuative approaches of social structure for health inequality research--which perspective opens up the concept of living conditions?]

For the last 30 years the appropriateness of social strata concepts has been discussed controversially in Germany. It was hypothesised that changes in the social structure resulting in a greater heterogeneity of social living conditions has decreased the relevance of social strata concepts. However, socio-epidemiological research still revealed a strong social gradient, indicating that health risks are still depending on individual social background. Nevertheless, enhanced models of social stratification could be fruitful for socio-epidemiological research, particularly with respect to the objective of reducing health inequality. The 'concept of living conditions' is one of the continuative social strata approaches, which is based on a multidimensional concept of social inequality. First use of this concept provided promising results in obtaining a more precise description of health-related living conditions. In this paper, the concept is presented in more detail, spotlighting on questions about conceptual realisation as well as empirical implementation. The paper aims to encourage a wider discussion about the use of these concepts for socio-epidemiological research and medical sociological theory.

[Correlation between social status, empowerment processes and the development of psychological health]

The aim of this paper is to deepen our knowledge about empirical evidence of empowerment on the basis of follow-up data from mother-child rehabilitation centres. The study focusses on three questions: i) to what extent can psychological health in different socioeconomic groups be improved by the intervention?, ii) Is there an impact of social position on the success of empowerment and iii) Is there any relevance of empowerment for sustainable improvement of psychological health? The study is based on data from 6 095 patients of 39 mother-child rehabilitation centres in Germany, collected up to one year after intervention. Empowerment was assessed by two self-defined scales, measuring positive changes i) in living circumstances, such as partnership or household ('Verhältnis-Empowerment') and ii) in health-related behaviours and competencies, such as better coping with problems and higher health awareness ('Verhaltens-Empowerment'). Health status was assessed by psychological symptoms and measured by SCL-K-9, a short form of the SCL-90-R. The socioeconomic position was determined by cluster analysis, the influence of social position and empowerment for psychological health was computed by analysis of regression. Before intervention mothers from the lower class showed higher degrees of psychological symptoms. After intervention they reached higher short-term effects, but six months later psychological symptoms increased again. In particular improvement of health-related behaviours and competencies was associated with better psychological health. The highest impact on health was found among mothers from the middle and lower class. These mothers, however, showed less success in optimising health-related behaviours and competencies. Twelve months after treatment the health-related impact of empowerment decreased. The theoretical evidence of empowerment could be empirically confirmed. However, decreasing health effects of empowerment after twelve months show that supporting help after clinical intervention is necessary to maintain the positive impact of empowerment. Only small differences in empowerment between lower and middle class lead to the conclusion that terms of social class are less appropriate for the clinical sample of mothers investigated in this study.

Krankheitsprofile und Therapieeffekte von Patientinnen in Mutter-Kind-Einrichtungen

Within the setting of external quality management of mother-child rehabilitation centres the short-term and medium-term therapy effects as well as the disease profiles of patients were analysed. This paper is focused on the changing of the main symptoms among the mothers. The patients' state of health is essentially characterized by physical and psychological exhaustion as well as by multimorbidity. It is strongly associated with the mothers' current situation of living. Before intervention the patients' health status was significantly impaired. Directly after the treatment high or very high therapy effects were found. All indicators are still higher after 6 months than they were at the beginning of the treatment. The results of a quasi-control population give evidence that these effects are caused by the treatment. First trends about the children treated show that behaviour problems can be reduced and the quality of life enhanced.

[Reactivity of patients under lithium-therapy/assessment of multivariate relationships (author's transl)]

A representative sample of patients under long-term lithium therapy was examined with respect to performance in reaction task, mood, and attitude toward lithium therapy. Reaction time to simple acoustic or visual stimuli as well as reactivity in complex settings under power- and speed-conditions requiring psychomotoric response represent the bulk of the data. Significant effects of additional medication, sex and duration of lithium therapy are assessed by univariate and multivariate data analysis. A hierarchic cluster analysis results in 5 well interpretable clusters. The clusters, furthermore, demonstrate the importance of attitude and other psychological variable, e.g. morning-/evening-type.

[A comparative study of patients following myocardial infarction assessed by the freiburger personality inventory (author's transl)]

The relationship between coronary disease and psychological factors is generally accepted. Numerous psychiatric studies have shown that there are typical basic personality features which predispose to heart disease. A personality inventory - the Freiburger Persönlichkeitsinventar (FPI) - was applied to a group of patients attending the Cardiology Clinic of the University of Vienna for a follow-up control examination after a heart attack. The results were compared with those of a control group (taken from the control population of the FPI). With respect to one personality dimension of the FPI, the so-called,,Nervositätsskala" (nervousness scale), significant differences were observed between the two groups. This scale includes so-called ,,psychosomatic" distrubances and disorders. According to these results,patients with coronary disease were more likely to experience such ,,psychosomatic distrubances" (as far as could be measured by this dimension of the FPI) than members of the control group. A comparison of patients graded according to the physical stamina level showed that those with a greatly-reduced work capacity as a consequence of myocardial infarction had a significantly higher score than patients whose work capacity was impaired to a lesser extent. This study introduces a new aspect in the relationship between coronary disease and personality dimensions.