Background Mortality inequalities demonstrate a double burden: the most deprived socioeconomic groups experience the lowest average age of death and the highest variation in age at death. Two processes generate variation in age at death: individual stochasticity (within-group variance) and heterogeneity (between-group inequality). No known research has quantified how much area level deprivation (between-group inequality) has contributed to total variation in age at death or if this has changed over time. We address this research gap using population and mortality data for the entire population of Scotland stratified by a validated measure of area-level deprivation covering 1981–2011.

Methods Census population estimates, and mortality records were matched with the Carstairs score. Life tables by year, sex and population weighted deprivation quintiles (20%) were constructed. Lifespan variation was calculated and transformed into the standard deviation to allow interpretation in number of years. Markov Chain methods were used to decompose the total lifespan variation in to the within deprivation-group and between deprivation-group component at each Census year, for males and females separately.

Results Total variation in age at death for males in Scotland increased between 1981 and 2011. For females there was limited change. Variation stagnated for the most deprived but decreased for the least deprived. Although the between-group component of variation in age at death is small the relative proportion doubled between 1981 and 2011 for males and females, from 1% to 3% and from 1% to 2% respectively. To give context, the proportion of total variance due to between-sex differences in age at death decreased from 4% in 1981 to 2% in 2011, meaning that inequalities in deprivation are comparable with sex differences when seeking to explain variation in age at death. Analysis repeated for population weighted deciles produced greater increases: quintiles are the preferred analytical grouping in Scotland. Analysis using the most and least deprived quintiles only showed that the contribution from deprivation was 4 times higher than when using all quintiles. Results reported used standard Markov stationary mixture assumptions and were unchanged when allowing empirical age-specific mixing distributions.

Conclusion Relative contributions from area level deprivation to total variance in age at death doubled in Scotland between 1981 and 2011 revealing the changing nature of mortality inequalities. These findings are relevant for governments deciding whether to invest in social policies that intervene at the areal level or at the individual level.