Cause of death mortality in Greece: regional variations and spatial patterns

Stamatis Kalogirou, Harokopio University of Athens
Cleon Tsimbos, University of Piraeus
Georgia Verropoulou, University of Piraeus
George Kotsifakis, Hellenic Statistical Authority

Since the Second World War mortality in Greece has been declining as a result of important socio-economic changes. However, the importance of certain causes of death, such as chronic diseases, has been constantly increasing. Although time trends of mortality at national level have been studied quite extensively, analyses of regional mortality differentials by cause of death are rather scarce. The aim of this paper is threefold. First, to calculate appropriate mortality measures for main categories of causes of death at prefecture level using recent (2006-08) statistical information. Second, to estimate the contribution of selected causes of death to the regional differentials of the overall mortality. Third, to identify possible spatial patterns of specific categories of causes of deaths. The study makes use of vital statistics recorded in 2006-2008 and population estimates for 2007 prepared by the Hellenic Statistical Authority. Deaths are classified by sex, cause (according to the WHO 1975 ICD) and place of residence of the deceased (NUTS 2); the causes of death considered are neoplasms, circulatory diseases, respiratory diseases and external causes. Population is classified by sex, 5-year age groups and place of residence. Standardised Mortality Ratios by selected causes of death are calculated for the 51 prefectures of Greece using the national age-sex-cause specific death rates as “standard” mortality schedule. Exploratory data analysis methods are employed. OLS and Geographically Weighted Regression (GWR) models are estimated. Moran’s I indices are computed and mapping techniques are used. Regional differentials in mortality are more pronounced among females. North-Eastern Greece exhibits high mortality from neoplasms and circulatory diseases. OLS models show that 95.5% of the male and 90% of the female variation in the overall mortality differentials are explained by the four causes of death under investigation. GWR models for males, but not for females, explain better the regional mortality variations.