Mortality and socio-economic differences in Denmark: a competing risks proportional hazard model
Introduction
Since the 1970s life expectancy in Denmark has increased less than in most other advanced countries. Among the 15 EU-countries before expansion, only three had a higher life expectancy than Denmark in 1980, while in 2002 only one country had a lower life expectancy.2 At the same time GDP per capita in 2002 in Denmark is third in the EU, only preceded by Luxembourg and Ireland according to the OECD, 2004 statistic. Generally, life expectancy and GDP per capita have a positive correlation (Sickles and Taubman, 1997); hence, it is remarkable that Denmark does not have a higher life expectancy. The explanation is presumably a complex mixture of factors like individual behaviour, life-style (e.g. attitudes toward smoking and alcohol use), access to medical care and quality of treatment. A step in the direction of better understanding the Danish “excess” mortality problem is provided by Juel et al. (2000). Based on aggregate data they show how mortality rates due to different causes of death have developed in Denmark compared to seven other European countries in the recent decades. They find that the excess mortality in Denmark is due to higher mortality rates from cancer (women only), circulatory diseases,3 and ill-defined conditions. Thus, their study identifies which illnesses are at the heart of the problem, but it does not cast light on factors that may have caused the disappointing development for these illnesses.
In the present study, we take a closer look at the relationship between the causes of death identified by Juel et al. (2000) and a variety of factors. We focus on the association between socio-economic status and mortality at the individual level, since in addition to the well-known correlation between life expectancy and GDP per capita at the population level, it is also found that high socio-economic status supports individual longevity (Moore and Hayward, 1990, Martikainen, 1995).
The available panel data set allows us to estimate a duration model, thus capturing the mortality experiences of different cohorts as they age. Furthermore, we employ a flexible competing risks proportional hazard model to allow for different causes of death. In addition, we formulate a duration model that handles left-truncated observations. Thereby, we correct for the fact that observed individuals have survived different time periods when they are observed.
Our findings reveal that for men the negative correlation between socio-economic status and mortality prevails for some diseases, but for women we find that factors such as being married, income, wealth and education are not significantly associated with higher life expectancy. Moreover, marriage increases the likelihood of dying from cancer for women, early retirement prolongs survival for men, and homeownership increases life expectancy in general.
Section snippets
Data
We use a representative sample of the Danish population to examine how mortality is related to factors such as education, skill level, sector of occupation, and income for the years 1992–1997.4 The data set is formed by merging a number of
Methodology
As previous studies on mortality we use a duration model (Moore and Hayward, 1990, Panis and Lillard, 1995). The mortality rate is given by the hazard function, which is the probability of dying in the interval (t; t + dt] given survival until t. We allow for different causes of death by specifying cause-specific hazard functions. The vector of explanatory variables, x, is included to capture the influence of the various exogenous socio-economic factors on the mortality rate. Let the continuous
Results
A positive (negative) effect implies that a given explanatory variable increases (decreases) the mortality rate (Table 3, Table 4).10 The effect of a given explanatory variable is always conditional on all the other variables that are included in the analysis, so the measured effect is, therefore, the net effect of that variable.11
Conclusion
We investigate the relationship between mortality and demographic and socio-economic indicators taking into account the distinction between three causes of death that have been identified in previous research to be at the root of the Danish excess mortality problem. Juel et al. (2000) showed that the excess mortality in Denmark compared to seven other European countries could be attributed largely to higher mortality due to cancer (women only), circulatory diseases and ill-defined conditions.
Acknowledgements
Financial support from the Danish Social Science Research Council is gratefully acknowledged. We thank Niels Henning Bjørn, Mette Ejrnæs, Anne Kristine Høj, Michael Rosholm and four anonymous referees very for useful comments and Jens Chr. Thellesen for excellent research assistance.
References (22)
- et al.
The social and private micro-level consequences of homeownership
J. Urban Econ.
(2003) - et al.
Mortality and morbidity among adults and the elderly
- Ellwood, D.T., Kane, T.J., 1989. The American way of aging: An event history analysis. NBER Working Paper No....
Marriage selection and mortality patterns: Inferences and fallacies
Demography
(1993)- et al.
Mortality among Japanese singles: A reinvestigation
Popul. Stud.
(1995) - et al.
Occupational influences on retirement, disability, and death
Demography
(1989) - et al.
Mortality difference by marital status: an international comparison
Demography
(1990) - Hurd, M. D., McFadden, D., Merril, A., 1999. Predictors of mortality among the elderly. NBER working paper no....
- et al.
Mortality and life expectancy in Denmark and in other European countries
Eur. J. Public Health
(2000) - et al.
The size of mortality differences associated with educational level in nine industrialized countries
Am. J. Public Health
(1994)
Marital status and mortality: the role of health
Demography
Cited by (21)
Marital and cohabitation status as predictors of mortality: A 10-year follow-up of an Italian elderly cohort
2008, Social Science and MedicineCitation Excerpt :Third, data on economic condition were unavailable in the ILSA study, thus this potentially important factor could not be verified. However, according to results of previous studies, in the elderly income and wealth seem to have little association with health or longevity (Goldman et al., 1995; Munch & Svarer, 2005). We believe that these weaknesses are balanced by ILSA strengths, which include: its prospective design, the large number of followed-up subjects (5376 subjects), the follow-up duration (10 years), the richness and accuracy of baseline data, the inclusion of resident in aged care facilities – avoiding the potential bias due to the disproportionate selection of unmarried subjects in better health conditions (Goldman et al., 1995).
Competing risk model for mobile phone service
2008, Technological Forecasting and Social ChangeCause-specific mortality at young ages: Lessons from Finland
2008, Health and PlaceToward generation XL: Anthropometrics of longevity in late 20th-century United States
2005, Economics and Human BiologyDoes early retirement lead to longer life?
2007, Ageing and SocietyMarital Status, Living Arrangements and Mortality at Older Ages in Chile, 2004–2016
2022, International Journal of Environmental Research and Public Health
- 1
Tel.: +45 8942 1598; fax: +45 8613 6334.