skip to primary navigation skip to content

Long-run cause of death series for national populations

Long-run cause of death series for national populations

The proximate causes of the extraordinary increases in life expectancy since the late nineteenth century are changes in the causes of death, the most significant being the shift from infectious diseases of childhood and early adulthood to chronic diseases of late adulthood. To understand these changes we need long-run and comprehensive series of death rates by cause of death and age at death that allow analysis of changes in the levels and age structure of cause-specific mortality. Cause of death data are inevitably problematic because lay and medical diagnoses and nomenclature are historically specific. However the use of complete cause of death series provides the potential to detect shifts in the distribution of deaths between categories, and to estimate the effects of such changes in nosology on observed rates.

This project has created the longest annual cause of death series available in machine-readable form for any national population using the annual counts of deaths by cause, sex and age from the Registrar-General's Reports for England & Wales 1848-1900 (Economic and Social Data Service SN5705) ( documentation). This dataset can be used in conjunction with the Office of National Statistics' series to create a continuous cause of death series for England from 1848-2010. This enables the study of long-run changes in mortality patterns across the period of secular decline in mortality in England & Wales in great detail. In particular, it permits long-run studies of mortality by cohort and cause, and allows detailed assessment of the extent of redistribution of deaths between categories of cause of death, which is not possible when only isolated causes of death are examined. A similar database for Scotland has also been created covering the years 1855-1949, and constitutes with the WHO data for Scotland from 1950-2010 the second earliest age-specific cause of death series for a national population (see documentation, and download dataset). These datasets have been used in the first instance together with data for a number of other populations to refute longstanding claims that declines in respiratory tuberculosis mortality before antibiotic therapy were caused mainly by improvements in the health of successive birth cohorts that were fixed in their effect during infancy and early childhood. Further work with these data is investigating the roles of smoking patterns and migration in international patterns of tuberculosis mortality in the pre-antibiotic era.


  • Davenport RJ (2013) 'Year of birth effects in the historical decline of tuberculosis mortality: a reconsideration.' PLOS ONE doi:10.1371/journal.pone.0081797
  • Davenport, RJ (2012) Annual deaths by cause, age and sex in Scotland, 1855-1949 (submitted to ESDS, see link above for dataset)
  • Davenport, RJ (2007) Annual deaths by cause, age and sex in England and Wales, 1848-1900. Economic and Social Data Services (SN5705)