Background and Objectives There are marked inequalities in mortality rates between areas of Britain. These inequalities have been persistent over long time periods and evaluation of recent area-based social policies in deprived areas has found that mortality rates have proven more resistance to change than other social indicators. Migration has been considered as one process that may underlie the persistence of health inequalities between areas. The geography of contemporary mortality rates is the product of movements across the life course however analysis of mobility over long time periods has been hampered by the limited availability of the necessary data in censuses, surveys and other secondary sources. The aim of this study was to assess if new detailed data sets describing residential histories between birth and death could be created through linkage of historical and contemporary data sources and used to illuminate current mortality geography.

Methods An age- and sex-structured random sample of 250 people dying in York was selected from death registrations in 2000/2001. The addresses of the deceased were traced in birth, marriage and electoral registers, BT phone books, street and trade directories and other sources. The distance between place of birth and death, time at last residence and how migration patterns varied with individual and neighbourhood characteristics were assessed within the context of data describing the health and socio-demographic history of the case study area.

Results Key residential data were successfully traced including place of birth and number of years resident at last address, collected for over 80% of cases. Only a third of those dying within York had been born in York but the majority were born in Yorkshire. Residential histories were shaped by a sub-regional network of movements linked to the development of the local economy, slum clearance and public housing policies. While the majority of the deceased in York were in-migrants half had been at their place of death for over 10 years and some for several decades. An exception to this pattern was those dying in nursing homes who often had moved shortly before death.

Conclusion The data set of life course residential histories created demonstrates that residential movements determining place of death often took place long before death. Recent mobility is concentrated among the young but most deaths are among the least mobile, elderly population. Analyses that consider only recent mobility cannot fully capture the selection processes that underlie contemporary mortality geography.