Cardiovascular diseases as causes of death: towards coherence and comparability
Agnieszka Fihel, Institut National d'Études Démographiques (INED)
France Meslé, Institut National d'Études Démographiques (INED)
Mortality data by single causes of death allow for the analysis of epidemiological trends and impacts of health policies across countries. Comparing mortality trends in international perspective makes an important methodological challenge for three particular reasons: 1) the cause-specific data lack longitudinal continuity due to changes of classifications of causes of death, 2) each country differs with regard to the way the information on causes of death are collected, coded and registered, 3) causes registered under so-called ‘garbage codes’ are used with different frequency and in different context across countries. In case of cardiovascular diseases, in particular, data exhibit large dissimilarities in historical and geographical dimension with regard to the scale of mortality and its structure by specific causes of death. This study presents how the cause-specific data may be corrected and adjusted to international analyses by redistributing deaths registered under garbage codes across selected cardiovascular diseases. Possible impacts of unusual circumstances affecting registering practices, such as introduction of automatic coding, are also investigated. We study mortality: – in the period covered by the 10th ICD revision, 1994–2013; – across four countries with different coding practices and epidemiologic situations (Czech Republic, Poland, Russia and the United Kingdom), ¬– within and between all adult age groups, – due to four ‘garbage code’ causes: atherosclerotic cardiovascular or heart disease (ICD-10 code: I25.0, .1), cardiac arrest (I46), heart failure (I50), atherosclerosis (I70). Preliminary results of adjustments prove that mortality due to garbage codes can be to some extent redistributed across well-defined cardiovascular diseases that are related to the respective garbage codes in a pathophysiological and statistical sense. Consequently, for well-defined causes of death we can smooth disruptions in time series resulting from sudden changes in coding practices.
Presented in Session 106: Advances in cause of death analysis