Reid, A., Garrett, E., Dibben, C. & Williamson, L. (2013) British Society for Population Studies (BSPS) Swansea 9-11 Sep 2013.

Other information: The comparison of causes of death over time is extremely difficult. Changes in medical knowledge and fashion mean diagnoses of the same condition may change over time: in most cases they become more precise, or two conditions, initially confused, are later recognised to be different. Cause-of-death classifications have changed over time to reflect medical knowledge and the cause of death structure and, even if causes are regrouped, it often proves impossible to follow causes reliably over time. The use of individual level cause of death data would allow the problem of changing nosologies to be addressed, as individual cause of death strings can be reclassified, but a classification system sensitive and flexible enough to reflect changes in the way the causes of death are recorded over time has not been developed. This paper reports on an attempt to develop such a classification, starting with the basic building blocks of the 4 character ICD10 codes but then adding additional historic codes for historical descriptors (HICOD). Of course mis-diagnoses, ‘unknown’ causes and overly vague descriptions of symptoms mean that it will never be possible to know the exact distribution of historical causes of death, and this may hamper analyses of change over time. However some groups of causes can be followed with more consistency than others, and this paper will give examples of successful and more problematic groupings using a variety of nineteenth century data-sets containing individual level causes of death.