Lasse Tarkiainen1, Pekka Martikainen2, Mikko Laaksonen3, Mikko Aaltonen4. 1. 1 Population Research Unit, Department of Social Research, University of Helsinki, FIN-00014 Helsinki, Finland lasse.tarkiainen@helsinki.fi. 2. 1 Population Research Unit, Department of Social Research, University of Helsinki, FIN-00014 Helsinki, Finland. 3. 2 Finnish Centre for Pensions, Research Department, Helsinki, Finland. 4. 3 National Research Institute of Legal Policy, Criminological Research Unit, Helsinki, Finland.
Abstract
BACKGROUND: Events and conditions during childhood have been found to affect health and mortality at later stages in life. We studied whether childhood conditions explain the observed all-cause and cause-specific mortality disparity between income groups in adulthood. METHODS: We used a 10% register linked sample of Finnish households in the 1950 census identifying 51 647 children aged 0-14 with at least one sibling of the same sex and followed them for mortality from the age 35 until ages 57-72. Using Cox regression with sibling design, we estimated hazard ratios (HRs) for quintiles of personal income at the age 35. We controlled for observed childhood family sociodemographic characteristics and allowed different baseline hazard functions for each group of siblings in order to control for all shared unobserved characteristics within families. RESULTS: Accounting for the observed childhood characteristics did not attenuate the income disparity in mortality, whereas adjusting for the sociodemographic characteristics in adulthood reduced the difference of the lowest quintiles by ∼70% among men and 30-40% among women. Controlling for the unobserved childhood characteristics in the sibling design did not provide any further explanation to the income differentials in mortality. This applied also for cause-specific mortality among men. HR to the cardiovascular diseases was 38% higher and 73% higher in alcohol, accidental and violent causes in the lowest quintile even after adjusting for all observed and unobserved characteristics. CONCLUSIONS: The excess mortality in the lowest income quintiles persists even after shared childhood family conditions among siblings are accounted for.
BACKGROUND: Events and conditions during childhood have been found to affect health and mortality at later stages in life. We studied whether childhood conditions explain the observed all-cause and cause-specific mortality disparity between income groups in adulthood. METHODS: We used a 10% register linked sample of Finnish households in the 1950 census identifying 51 647 children aged 0-14 with at least one sibling of the same sex and followed them for mortality from the age 35 until ages 57-72. Using Cox regression with sibling design, we estimated hazard ratios (HRs) for quintiles of personal income at the age 35. We controlled for observed childhood family sociodemographic characteristics and allowed different baseline hazard functions for each group of siblings in order to control for all shared unobserved characteristics within families. RESULTS: Accounting for the observed childhood characteristics did not attenuate the income disparity in mortality, whereas adjusting for the sociodemographic characteristics in adulthood reduced the difference of the lowest quintiles by ∼70% among men and 30-40% among women. Controlling for the unobserved childhood characteristics in the sibling design did not provide any further explanation to the income differentials in mortality. This applied also for cause-specific mortality among men. HR to the cardiovascular diseases was 38% higher and 73% higher in alcohol, accidental and violent causes in the lowest quintile even after adjusting for all observed and unobserved characteristics. CONCLUSIONS: The excess mortality in the lowest income quintiles persists even after shared childhood family conditions among siblings are accounted for.