Spatial-temporal comparability of mortality structure by cause of death in Russia:: the role of regional cause-of-death coding practices

Abstract

The study of mortality by cause of death is an important area of demography and public health that has practical significance for the formulation of health policy. However, as previous studies on Russia and other countries have shown, objective dynamics and regional differences in the structure of mortality by cause may conceal subjective factors, including “established” features of the selection of the initial cause of death. This paper assesses the spatial and temporal stability of mortality structures and cause-of-death coding practices in Russian regions. Rosstat data on mortality from 67 main causes for 71 regions of the Russian Federation were used. An indirect regression method (separately for men and women) was applied, modified to compare consecutive periods: 2000–2004, 2005–2009, 2010–2014, and 2015–2019. Regional mortality structures were compared with the Russian average to identify: a) regions with distinct mortality structures, b) causes of death with high interregional variability in their contribution to overall mortality, and c) changes in these patterns over time. The method made it possible to identify causes of death whose differences in contribution between regions may be due to coding practices, as well as to track the dynamics of these differences. For 39 causes of death, significant differences in their contribution to overall mortality between regions were identified throughout the period 2000–2019. Only for ten causes did the degree of variability change significantly in at least one of the periods. Among the regions, 14 (for men) and 22 (for women) demonstrated specificity relative to the average Russian mortality structure throughout 2000–2019, and in seven regions, changes in this specificity were recorded in at least one of the periods. The degrees of variability in the contributions of causes of death and the specificity of regional mortality structures are significantly correlated between periods. The results are stable when the sample of regions is expanded and external causes of death are excluded, and are consistent with previous studies of interregional differences in the coding of causes of death in Russia. The differences obtained are at least partly due to cause of death coding practices. Therefore, it can be argued that the ten causes of death for which significant changes in the interregional variability of their contributions were found are characterized by persistent interregional inconsistencies in coding practices over time. This indicates the long-standing nature of differences in coding practices associated with the decentralization of the system for selecting the initial cause of death in Russia. To improve data comparability, it is necessary to unify practices through methodological recommendations, working with regions, and introducing automated coding systems.