Objectives: Although mortality differences between groups of different socioeconomic status (SES) are known to decrease at higher age, the reasons for this are unknown. This study reviews the existing arguments for convergence versus divergence of SES mortality differences in old age, differentiates between increasing age and worsening health, and discusses explanations for the observed convergence.
Methods: Register data of all Danish men aged ≥ 59 years between 1980 and 2002 (n = 938,427), and survival analysis, are used to examine mortality differences between income groups, controlling for seven variables. Interaction models describe the change of SES mortality differences with age or health, days in hospital being a proxy for the latter.
Results: Large mortality differences were found with relative risks of 0.28-0.36 for income groups above the 25th income percentile compared with those below the 10th percentile. Mortality differentials are stable across age groups but converge when health deteriorates.
Discussion: The idea of "age as leveler" has to be revised because it is illness and not age that levels SES mortality differences. That the transition from poor health to death is independent from SES suggests that social mortality differences originate in early stages of the health deterioration process. Intensive treatment for very sick people will not decrease SES mortality differences.