Gabriele Doblhammer Uta Ziegler

1. Trends in Disability Trajectories and Subsequent MortalityA study based on the German Socioeconomic Panel for the Periods 1984-1987 and 1995-1998 with a Three Year Mortality Follow-up. Gabriele Doblhammer Uta Ziegler

2. Studies on Trends in Health • Prevalences of disability in the context of the Sullivan Method • Incidence of health transitions in combination with multi-state life tables • Many population-based longitudinal studies on health and disability do not fully take advantage of the information available. • They use two points in time • Study single transitions • Use a dichotomous measurement of disability

3. Health Trajectories • Individual trajectories of disability (Maddox and Clark 1992; Verbrugge et al. 1994; Li et al. 2000; Liang et al. 2003; Deeg 2005; Nusselder et al. 2007; Doblhammer and Hoffmann 2008, Aldwin, Spiro, Levenson & Cupertino 2001, Clipp, Pavalko, & Elder, 1992 ; Liang et al. 2005; McDonough and Berglund 2003) • more than two points in time and • more than two levels of disability • allow for deterioration and recovery • Use the concept of individual trajectories to study time trends in disability

4. Aims of this Study • Identify the most frequent individual-level disability trajectories in the 1980s and 1990s • Explore whether the frequency of individual-level disability trajectories has changed over time taking into account changes in the age-structure and • the socioeconomic structure of the population. • Analyze the relationship between specific disability trajectories and sub-sequent mortality and whether this relationship has changed over time.

5. Method: Step 1 • Separate linear regression of the year on the disability outcome to • asses the four aspects for each individual. • The level of disability is defined as the intercept of a linear regression • model • 2. The slope of the model is used to indicate the direction of the change: • a positive slope indicates deterioration; • a negative one indicates an improvement in disability. • 3. The concavity/convexity of the time trend is measured by adding • a quadratic term to the equation and by measuring the distance between • the quadratic regression curve and the straight linear regression line: • a positive difference indicates a convex shape, • a negative one indicates a concave shape. • 4. The variability of the trajectory is measured by the root mean square error • of the quadratic function.

6. Method: Step 2 The four aspects are the input variables for two independent cluster analyses for the two time periods In order to assure that each of the four aspects influences the cluster analysis equally, we standardize them, using their mean and standard deviation. We perform a hierarchical agglomerative complete linkage cluster analysis based on Euclidian distances. The number of clusters is decided on the basis of the Calinski-Harabasz pseudo-F statistic. We treat the stable disability trajectories (stable healthy, stable moderate disability, stable severe disability) separately.

7. GSOEP In the years 1984 to 1987, 1992, 1995 to 2001 “Not regarding occasional illnesses, is the fulfilment of everyday activities, e.g. in the household, your job or education hindered by your condition of health, and, if so, to what extent?” Not at all, slightly, to a great extent. Time Trends in Disability Trajectories 1984 to 1987 vs. 1995 to 1998 Three year mortality follow-up of the survivors: 1988 to 1990 1999 to 2001 In both periods only data from West Germany are used.

8. Data GSOEP Ages 50+ Excluded individuals due to missing information 398 232 about their disability levels: Deaths in follow-up period: 94 123 Attrition in follow-up period: 203 222

9. Trajectories of the Survivors

10. Trajectories of the Survivors • Relative frequency of the disability trajectories in 1984-1987 and 1995-1998 by sex

11. Males: Trends in German Disability Trajectories 1984/1987 and 1995/1998 A significant shift towards health trajectories that involve stable moderate disability. 1984/1987 1995/1998 Probability • Adjusted for age, • education & • marital status • Weighted by 1984 & • 1995 survey weigths

12. Females: Trends in German Disability Trajectories 1984/1987 and 1995/1998 A significant shift towards health trajectories that involve stable moderate disability. 1984/1987 1995/1998 Probability • Adjusted for age, • education & • marital status • Weighted by 1984 & • 1995 survey weigths

13. Mortality during follow-up periods • Highest mortality: healthy, continous decline & severe disability, stable • Relationship between type of trajectory and subsequent • mortality remains unchanged over time • Adjusted for age, • education & • marital status • Weighted by 1984 & • 1995 survey weigths

14. Conclusions 1. From the 1980s to the 1990s there is a significant shift towards health trajectories that involve stable moderate disability. 2. In both time periods those lost to follow-up have a similar distribution according to their last-measured health status. This implies that the shift towards the stable moderate disability trajectory is not the result of more individuals with moderate disability remaining in the survey sample. 3. In the GSOEP those with moderate disability do not experience any excess mortality as compared to the healthy. This remains unchanged over time. 4.The shift from severe towards stable moderate disability trajectories that are non-fatal is in accordance with a large number of studies in the US and Europe.

15. This study is part of the project MAGGIE – Major Ageing and Gender Issues financed by the 6th framework of the European Commission.