Trends in Age-Specific Coronary Heart Disease Mortality
Trends in Age-Specific Coronary Heart Disease Mortality
Data on cause-specific numbers of deaths and population numbers, by sex and in 5-year age groups (up to 85 and over) for each country of the EU, were extracted from the WHO global mortality database. Age- and sex-specific mortality rates were calculated and aggregated rates were standardized to the European Standard Population using the direct method. Coronary heart disease as a cause of death was defined according to the following International Classification of Disease codes: ICD-10 (International Classification of Diseases, tenth revision) codes I20-I25; ICD-9 and ICD-8 codes 4100–4149.
Data were extracted and analysed for the years 1980 (or the establishment of the present-day country if later) to 2009 where available. In some countries, death registration or population data were missing for some years. Cyprus was excluded from analyses due to a high level of unreliable and missing data. Data for Germany for years prior to 1990 were obtained by combining death registration and population data for the former Democratic Republic of Germany and the former Federal Republic of Germany and calculating overall rates. England, Wales, Scotland, and Northern Ireland were analysed together as the UK. The age groups analysed were (i) <45 years, (ii) 45–54 years, (iii) 55–64 years, and (iv) 65 years and over. All age groups up to 45 years were combined due to the low numbers of deaths before this age, and because subgroups that contain any year with 0 deaths must be excluded from joinpoint analyses.
Age-standardized mortality rates by country, sex, and age group were calculated. In the descriptive tables, 5-year average mortality rates are presented, to show more representative and stable rates and minimize the effect of year-to-year fluctuations in death rates, (particularly among younger age groups where rates are relatively low and therefore percentage changes over time are more affected by small absolute changes in rates). The average rate in the five most recent years of data (2005–09) was compared with earlier 5-year periods to describe crude proportional changes in death rates over time. Data for mortality rates in all individual years were, however, used in the joinpoint analysis.
Joinpoint regression was performed to identify periods with statistically distinct log-linear trends in death rates from CHD over time within each age group, by sex and country. This analysis identifies inflexion points ('joinpoints') at which there is a significant change in the trends, using a series of permutation tests, with Bonferroni adjustment for multiple comparisons. The two-sided significance level was set at p < 0.05 for all tests. The number and location of significant joinpoints for each country by sex and age group (maximum of 3) was determined using a log-linear model, and the annual percentage change within each segment calculated. Use of a log-linear model enables the analysis of constant percentage (rather than absolute) change in prevalence over time. The log-linear model is also more robust in rare conditions, as CHD generally is in younger subpopulations. Evidence to suggest a recent plateau in a population group was defined as the situation where the final joinpoint segment (in the best fitted model) was less steeply negative compared with the preceding segment (or indeed, was flat or positive), that is, that the final segment showed less of a decrease than the segment before it. Average annual percentage changes (AAPCs) were also calculated overall (1980–2009) and separately for each decade with respect to the underlying joinpoint model.
All analyses were conducted using Stata version 11.2 (StataCorp, TX, USA) and Joinpoint Regression Program version 3.5.4 (Statistical Research and Applications Branch, National Cancer Institute, USA).
Methods
Data on cause-specific numbers of deaths and population numbers, by sex and in 5-year age groups (up to 85 and over) for each country of the EU, were extracted from the WHO global mortality database. Age- and sex-specific mortality rates were calculated and aggregated rates were standardized to the European Standard Population using the direct method. Coronary heart disease as a cause of death was defined according to the following International Classification of Disease codes: ICD-10 (International Classification of Diseases, tenth revision) codes I20-I25; ICD-9 and ICD-8 codes 4100–4149.
Data were extracted and analysed for the years 1980 (or the establishment of the present-day country if later) to 2009 where available. In some countries, death registration or population data were missing for some years. Cyprus was excluded from analyses due to a high level of unreliable and missing data. Data for Germany for years prior to 1990 were obtained by combining death registration and population data for the former Democratic Republic of Germany and the former Federal Republic of Germany and calculating overall rates. England, Wales, Scotland, and Northern Ireland were analysed together as the UK. The age groups analysed were (i) <45 years, (ii) 45–54 years, (iii) 55–64 years, and (iv) 65 years and over. All age groups up to 45 years were combined due to the low numbers of deaths before this age, and because subgroups that contain any year with 0 deaths must be excluded from joinpoint analyses.
Age-standardized mortality rates by country, sex, and age group were calculated. In the descriptive tables, 5-year average mortality rates are presented, to show more representative and stable rates and minimize the effect of year-to-year fluctuations in death rates, (particularly among younger age groups where rates are relatively low and therefore percentage changes over time are more affected by small absolute changes in rates). The average rate in the five most recent years of data (2005–09) was compared with earlier 5-year periods to describe crude proportional changes in death rates over time. Data for mortality rates in all individual years were, however, used in the joinpoint analysis.
Joinpoint regression was performed to identify periods with statistically distinct log-linear trends in death rates from CHD over time within each age group, by sex and country. This analysis identifies inflexion points ('joinpoints') at which there is a significant change in the trends, using a series of permutation tests, with Bonferroni adjustment for multiple comparisons. The two-sided significance level was set at p < 0.05 for all tests. The number and location of significant joinpoints for each country by sex and age group (maximum of 3) was determined using a log-linear model, and the annual percentage change within each segment calculated. Use of a log-linear model enables the analysis of constant percentage (rather than absolute) change in prevalence over time. The log-linear model is also more robust in rare conditions, as CHD generally is in younger subpopulations. Evidence to suggest a recent plateau in a population group was defined as the situation where the final joinpoint segment (in the best fitted model) was less steeply negative compared with the preceding segment (or indeed, was flat or positive), that is, that the final segment showed less of a decrease than the segment before it. Average annual percentage changes (AAPCs) were also calculated overall (1980–2009) and separately for each decade with respect to the underlying joinpoint model.
All analyses were conducted using Stata version 11.2 (StataCorp, TX, USA) and Joinpoint Regression Program version 3.5.4 (Statistical Research and Applications Branch, National Cancer Institute, USA).
Source...