Articles
Mortality trends in chronic obstructive pulmonary disease in Europe, 1994–2010: a joinpoint regression analysis Jose Luis López-Campos, Miguel Ruiz-Ramos, Joan B Soriano
Summary Lancet Respir Med 2014; 2: 54–62 Published Online December 6, 2013 http://dx.doi.org/10.1016/ S2213-2600(13)70232-7 See Comment page 4 Unidad Médico-Quirúrgica de Enfermedades Respiratorias. Instituto de Biomedicina de Sevilla (IBiS), Hospital Universitario Virgen del Rocio / Universidad de Sevilla, Spain (J L López-Campos MD); CIBER de Enfermedades Respiratorias (CIBERES), Instituto de Salud Carlos III, Madrid, Spain (J L López-Campos); Consejería de Salud y Bienestar Social de Andalucía. Seville, Spain (M Ruiz-Ramos MD); and Fundación Caubet-Cimera, Bunyola, Islas Baleares, Spain (J B Soriano MD) Correspondence to: Dr Jose Luis López-Campos, Hospital Universitario Virgen del Rocio, Avda Manuel Siurot, s/n, 41013 Seville, Spain [email protected]
Background Findings from studies done over the past 20 years suggest that mortality from chronic obstructive pulmonary disease (COPD) is decreasing worldwide, but little information is available for trends in Europe. We aimed to describe COPD mortality trends by sex and calendar year for the period of 1994 to 2010. Methods We extracted data for COPD deaths between 1994 and 2010 in the 27 countries in the European Union (EU) from the statistical office of the EU (Eurostat), using the International Classification of Diseases 10 (ICD-10) codes J40–J44 and J47. We estimated age-standardised mortality rates (ASR), and analysed data using joinpoint regression, for women and men in the EU overall and by individual country for each year. We used the standard European population as the reference and present our findings as deaths per 100 000 person-years. We compared findings for each country with the EU average by calculating standardised rate ratios (SRR) and 95% CIs. Findings Between 1994 and 2010, there were 2 348 184 recorded COPD deaths in the EU. COPD mortality was higher in men than in women throughout the study period in all EU countries. In the EU overall, deaths per 100 000 population decreased in men almost linearly from 90·07 in 1994 to 61·33 in 2010, and in women from 26·99 in 1994 to 25·15 in 2010, representing a narrowing in gender gap over the study period. Several countries had a higher SRR mortality than the EU average—eg, Ireland, Hungary, and Belgium for men and Denmark, the UK, and the Netherlands for women. Our joinpoint regression analysis identified no statistically significant changes in the trend for the whole EU, but several countries had changing trends over the study period. In men, we recorded a 2·56% constant and statistically significant decrease in ASRs in the EU. Five countries had an increase in ASR. Overall, in women, we recorded a 0·76% statistically significant decrease in ASRs. 14 countries had an increase in ASR. Interpretation Our findings indicate a downward trend in COPD mortality in Europe between 1994 and 2010. The data also suggest a narrowing of the gap between COPD mortality in men and in women. The wide heterogeneity in mortality rates within European countries could serve as a reference to allow informed policy making. Funding None.
Introduction Chronic obstructive pulmonary disease (COPD) is a chronic respiratory disorder that causes much morbidity, mortality, and socioeconomic burden.1 COPD mortality has been constantly increasing, with projections that it will become the fourth leading cause of death by 2030.2 However, despite concordant publications indicating an increase in the number of deaths due to COPD worldwide, in the past few years several publications have reported decreases and downward projections of COPD mortality rates in various countries, such as Japan, USA, Australia, and Spain.3–6 An update of the Global Burden of Diseases, Injuries, and Risk Factors Study in 2010 estimated the annual deaths from 235 causes for 21 world regions between 1980 and 2010, by age and sex. Although COPD was the third leading cause of death in 2010, age-standardised death rates decreased by 43·3% in the past two decades, falling from 77·4 per 100 000 in 1990 to 43·8 per 100 000 in 2010.7 Worldwide, an estimated 328 million people had COPD in 2010, and COPD was the fifth major cause of years lived with disability, ranking behind back pain, major depressive disorders, iron-deficiency anaemia, and neck pain.8 54
Although very valuable, these global figures do not inform of trends within countries, many of which have different health-care systems and are at different phases of the smoking epidemic; great variability is, therefore, expected between countries. Furthermore, these studies do not provide estimates of when a change in disease mortality rate, if any, occurred. We aimed to describe trends in the risk of dying from COPD in people older than 40 years in the European Union (EU) by sex and calendar year between 1994 and 2010. This estimation might help to identify high-risk areas and provide health-care policy makers with evidence for more focused health intervention in COPD, from prevention to palliative care.
Methods Deaths in the 27 EU countries are those detailed by the statistical office of the EU (Eurostat; Luxemburg) where the International Classification of Diseases 10 (ICD-10) codes J40–J44 and J47 were selected from the years 1994 to 2010. The database was updated on April 8, 2013. We used the underlying cause of death for this analysis, which is defined by Eurostat as “the disease or injury which www.thelancet.com/respiratory Vol 2 January 2014
Articles
initiated the train of morbid events leading directly to death, or the circumstances of the accident or violence which produced the fatal injury”. Depending on the country, coding is done manually or with automated coding systems, but, in most countries, coding is done centrally in the Eurostat statistics office. Unfortunately, a dataset was not complete for all countries in the EU because some countries declared no deaths for several years in our study period and not all countries were included in the EU for the complete study period (appendix). Therefore, to infer the deaths in these years, we calculated the expected deaths by taking an average of the preceding and following year, if data were available.9 We calculated expected deaths as the product of the specific death rates in each age group and the population in each age group for those countries and years for which no data were available. Thus, once the populations were estimated, all deaths by age and sex were obtained for the entire series. We obtained populations sizes from country censuses when available for a specific year, and from the inter-censual estimates by Eurostat for other years. Age was aggregated in 5-year groups from 40 years to 85 years by sex. We estimated age-standardised mortality rates (ASR) for women and men in the EU and within each country for every year studied, using the standard European population as the reference, giving our findings as number of deaths per 100 000 person-years. Thus, ASR allowed direct comparisons between countries in this and in previous studies by adjusting for a reference population. We calculated standardised rate ratios (SRR) of countries versus the whole EU and 95% CIs using the approximation of Miettinen:10 (T1 / T2)1±Zα/2/χ, where Zα/2 = 1·96 for 95% CI and χ = (T1 – T2) / (S1 + S2)1/2. Where T1 is the country standardised rate, T2 is the EU standardised rate for each year and sex, and S1 and S2 are SEs of T1 and T2. The SRR indicates the number of times
that the mortality of a country is increased (if above 1) or decreased (if below 1) as compared with the EU average. Thus, the product of the SRR of a country in a given year and for a given sex by the standardised mortality rate for the EU is equal to the standardised rate of that country for that year and sex. We established the statistical significance of these standardised rate ratios using 95% CIs. The standardised rate trends by sex and age were estimated by joinpoint regression analysis.11 We used joinpoint regression analysis to identify years when changes occurred in the linear slope of the temporal trend. Best-fitting points (the so-called joinpoints) were chosen when the rate changed significantly. This approach has two major advantages, it identifies the time when there are changes in the trend, and it estimates the magnitude of the increase or the decrease seen in each interval by estimating the percentage of annual change (PAC). We used The Surveillance Epidemiology and End Results statistical software (Joinpoint Regression Program, version 3.5.1) to calculate PAC, trends, and graphs, and the statistical significance of ASRs. We built segmented models of Poisson regression to estimate the trend changes over time. In this model, we used standardised rates as the dependent variable and the year of death as the independent variable. To identify a future timepoint when trends in men and women converge, we used the model equation to estimate yearly rates of mortality for both sexes until the projected death rate for men equalled that for women. In all analyses, a p value lower than 0·05 was regarded as statistically significant.
Role of the funding source There was no funding source for this study. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.
100 Age-standardised death rate per 1 00 000 person-years
See Online for appendix
Men (joinpoint regression lines) Men (age-standardised mortality) Women (joinpoint regression lines) Women (age-standardised mortality)
90 80 70 60 50 40 30 20 10 0 1994
1995
1996
1997
1998
1999
2000
2001
2002 Year
2003
2004
2005
2006
2007
2008
2009
2010
Figure 1: Age-standardised COPD mortality trends in the European Union (1994–2010) Solid lines are joinpoint regression lines. Dotted lines are age-standardised mortality.
www.thelancet.com/respiratory Vol 2 January 2014
55
Articles
Results Between 1994 and 2010, there were 1 475 061 recorded COPD deaths in men and 873 123 recorded COPD deaths in women. Age-standardised COPD mortality rates in men decreased linearly (figure 1), although there was much variability between countries (table 1). The downward trend over the same period in women was less pronounced (figure 1; table 2). The male-to-female ratio in COPD mortality decreased from 3·3 in 1994 to 2·4 in 2010. If these trends in both sexes were to continue, we estimate that COPD mortality would not differ between sexes by 2031 (appendix). Countries such as Belgium, Denmark, Hungary, Lithuania, Netherlands, Spain, and the UK had significantly higher COPD mortality rates in men than the EU average (table 1, figure 2). This excess mortality was present in the initial years of the time period studied and was sustained over time. Countries such as Bulgaria, Cyprus, the Czech Republic, Greece, and France had SRRs significantly below the EU average, whereas some countries—eg, Germany, Austria, or Luxembourg—have changes in SRRs in line with EU average (table 1). In
European Union
women (table 2), Denmark had the highest risk of dying from COPD each year, reaching a three-times increase by 2010 compared with the EU average. Ireland, UK, and the Netherlands had the next highest mortality rates. Small countries such as Malta and Cyprus had the lowest mortality SRRs. Figure 3 (men) and figure 4 (women) show maps of SSRs for each country in the EU. Findings from our joinpoint regression analysis, assessing trend changes over time, are shown in table 3 for men and table 4 for women; figure 2 shows graphs for trends in both sexes. In men, we recorded a 2·56% constant and statistically significant decrease in ASRs in the EU, with no joinpoints identifiable (table 3). By country, during the entire period, Slovenia, France, and Ireland had the steepest decreases in ASR (table 3). Only five countries had an increase in ASR, with Cyprus, Slovakia, and Greece having the steepest increases. Over the entire study period, PACs in some countries such as Austria, Denmark, Finland, France, Latvia, the Netherlands, Poland, Slovakia, and Sweden peaked and then decreased, with an overall downward trend.
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
73·84
64·86
64·78
61·33
90·07
92·36
89·09
87·09
86·79
87·68
78·32
72·64
78·19
69·48
71·91
66·65
67·48
Austria
0·66
0·70
0·69
0·71
0·71
0·75
0·83
0·89
0·97*
1·12
1·09
1·06
1·09
1·10
0·99*
0·96*
Belgium
1·53
1·57
1·61
1·56
1·63
1·53
1·71
1·84
1·63
1·54
1·58
1·52
1·45
1·47
1·50
1·46
1·54
Bulgaria
0·58
0·57
0·59
0·71
0·64
0·56
0·64
0·58
0·55
0·55
0·62
0·78
0·84
0·92
0·85
0·70
0·65
Cyprus
0·27
0·26
0·28
0·28
0·28
0·27
0·31
0·33
0·32
0·30
0·52
0·46
0·49
0·54
0·48
0·65
0·53
Czech Republic
0·67
0·52
0·41
0·46
0·56
0·56
0·70
0·71
0·68
0·68
0·69
0·89
0·75
0·84
0·84
0·93
0·80
1·01*
Denmark
1·23
1·34
1·22
1·47
1·48
1·48
1·61
1·75
1·58
1·48
1·44
1·33
1·44
1·45
1·51
1·58
1·67
Estonia
0·77
0·65
0·59
0·51
0·56
0·59
0·67
0·61
0·64
0·67
0·92*
0·67
0·86
0·84
0·93*
0·68
0·78
Finland
0·72
0·79
0·87
0·90
0·95*
0·86
0·93
0·97*
0·91
0·83
0·79
0·85
0·94*
0·88
0·83
0·84
0·82
France
0·67
0·69
0·74
0·72
0·75
0·73
0·45
0·45
0·45
0·46
0·45
0·47
0·45
0·44
0·45
0·42
0·44
Germany
0·98
0·99*
0·98
0·90
0·85
0·83
0·92
0·89
0·91
0·91
0·91
0·88
0·89
0·87
0·89
0·96
0·98
Greece
0·27
0·28
0·27
0·29
0·28
0·35
0·43
0·48
0·40
0·30
0·36
0·33
0·51
0·52
0·56
0·58
0·57
Hungary
1·67
1·57
1·48
1·35
1·26
1·44
1·35
1·22
1·26
1·38
1·41
1·76
1·94
1·95
1·88
1·82
1·89
Ireland
1·88
1·91
1·80
1·90
1·78
1·77
1·63
1·52
1·55
1·41
1·38
1·21
1·33
1·32
1·27
1·36
1·18
Italy
0·99*
0·89
0·85
0·93
0·90
0·88
0·93
0·90
0·89
1·03
0·94
0·91
0·99*
0·97
0·96
0·96
0·95
Latvia
0·54
0·52
0·60
0·62
0·53
0·49
0·49
0·51
0·59
0·54
0·61
0·64
0·63
0·77
0·64
0·55
0·64
Lithuania
1·65
1·51
1·45
1·42
1·41
1·30
1·42
1·55
1·50
1·45
1·54
1·43
1·60
1·69
1·47
1·35
1·34
Luxemburg
1·23*
1·14*
1·00*
0·99*
0·94*
1·03*
1·15*
1·16*
1·17*
1·00*
1·00*
1·03*
0·90*
0·89*
0·84*
1·05*
1·01*
Malta
1·11*
1·23*
1·17*
1·33
1·42
1·47
1·19*
0·79*
1·35
1·36
1·47
1·46
1·47
1·21*
1·15*
1·11*
1·08*
Netherlands
1·41
1·48
1·55
1·48
1·59
1·45
1·60
1·58
1·50
1·39
1·34
1·38
1·42
1·39
1·32
1·28
1·23
Poland
0·68
0·69
0·72
0·79
0·79
0·79
0·87
0·89
0·89
0·94
1·01*
1·03
1·07
1·12
1·09
1·09
0·94
Portugal
0·81
0·80
0·87
0·83
0·78
0·87
0·83
0·83
0·88
0·81
0·82
0·90
0·75
0·83
0·83
0·89
0·87
Romania
1·03*
1·01*
1·04
1·07
1·07
1·13
1·13
1·20
1·32
1·18
1·31
1·19
1·22
1·17
1·19
1·21
1·23
Slovakia
0·27
0·26
0·34
0·25
0·64
0·77
0·76
0·74
0·74
0·76
0·83
0·85
0·74
0·85
0·75
0·80
0·88
Slovenia
1·42
1·49
1·54
1·40
1·36
1·49
1·48
1·54
1·40
1·42
1·20
1·16
0·95*
0·83
0·94*
0·83
0·89
Spain
1·30
1·35
1·41
1·39
1·46
1·50
1·45
1·40
1·48
1·38
1·37
1·45
1·25
1·30
1·24
1·29
1·28
Sweden
0·49
0·48
0·51
0·53
0·55
0·55
0·59
0·62
0·63
0·59
0·63
0·62
0·64
0·61
0·64
0·60
0·63
UK
1·32
1·37
1·30
1·32
1·28
1·30
1·30
1·35
1·31
1·28
1·29
1·25
1·27
1·24
1·31
1·23
1·26
Data are deaths per 100 000 population for the European Union (EU) and rate ratios for individual countries (95% CIs are given in the appendix). *Not statistically significant versus EU average.
Table 1: Standardised COPD mortality rates in the European Union (men)
56
www.thelancet.com/respiratory Vol 2 January 2014
Articles
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
26·99
28·34
27·88
27·93
28·54
29·84
25·92
24·46
25·38
27·93
24·73
26·43
24·86
25·73
25·45
25·72
25·15
Austria
0·70
0·74
0·71
0·69
0·67
0·75
0·92
1·02*
1·07
1·24
1·23
1·11
1·22
1·12
1·01
1·13
1·12
Belgium
1·05*
1·18
1·21
1·20
1·24
1·03*
1·19
1·26
1·19
1·41
1·36
1·39
1·30
1·39
1·37
1·31
1·34
Bulgaria
0·79
0·75
0·75
0·78
0·77
0·63
0·74
0·62
0·47
0·50
0·59
0·74
0·69
0·72
0·69
0·47
0·47
Cyprus
0·33
0·32
0·34
0·33
0·32
0·32
0·37
0·38
0·37
0·33
0·64
0·46
0·49
0·43
0·41
0·40
0·40
Czech Republic
0·68
0·45
0·42
0·46
0·49
0·55
0·71
0·71
0·66
0·70
0·67
0·88
0·73
0·78
0·85
0·88
0·82
Denmark
2·54
2·68
2·72
3·01
3·28
3·21
3·82
3·87
3·69
3·13
3·15
3·00
3·22
3·27
3·03
3·22
3·29
Estonia
0·44
0·37
0·30
0·35
0·38
0·28
0·29
0·29
0·32
0·32
0·36
0·28
0·36
0·40
0·34
0·22
0·32
Finland
0·42
0·48
0·46
0·52
0·55
0·55
0·54
0·61
0·54
0·45
0·49
0·55
0·55
0·50
0·56
0·60
0·57
France
0·72
0·73
0·78
0·79
0·80
0·77
0·38
0·35
0·35
0·35
0·36
0·37
0·35
0·35
0·35
0·34
0·34
Germany
0·88
0·88
0·89
0·81
0·82
0·80
0·89
0·90
0·95
0·93
0·96
0·95
0·98*
0·98*
1·00*
1·11
1·15
Greece
0·06
0·06
0·05
0·04
0·04
0·01
0·01
0·00
0·01
0·01
0·01
0·00
0·00
0·00
0·00
0·00
0·00 1·90
European Union
Hungary
1·94
1·74
1·64
1·51
1·40
1·57
1·49
1·31
1·31
1·47
1·60
1·99
2·01
2·12
1·86
1·92
Ireland
2·59
2·75
2·61
2·70
2·68
2·59
2·53
2·45
2·05
1·91
2·13
2·03
1·91
2·23
1·90
2·09
1·91
Italy
0·88
0·77
0·73
0·78
0·79
0·76
0·80
0·73
0·77
0·92
0·86
0·80
0·85
0·83
0·84
0·84
0·80
Latvia
0·42
0·40
0·40
0·45
0·33
0·33
0·30
0·27
0·28
0·30
0·31
0·24
0·28
0·25
0·27
0·26
0·19
Lithuania
1·30
1·34
1·23
1·07*
0·95*
0·78
0·92*
0·91*
0·95*
0·77
0·75
0·76
0·62
0·77
0·55
0·50
0·54
Luxemburg
1·20*
1·22*
0·86*
0·94*
0·72
0·93*
0·77*
1·02*
1·18*
0·98*
0·86*
0·96*
1·22*
1·24*
1·28*
1·43
0·99*
Malta
0·59
0·46
0·14
0·63
0·59
0·62
0·68
0·43
0·25
0·30
0·50
0·46
0·34
0·26
0·51
0·28
0·38
Netherlands
1·31
1·41
1·45
1·48
1·56
1·53
1·74
1·74
1·64
1·62
1·59
1·70
1·74
1·70
1·72
1·66
1·65
Poland
0·49
0·47
0·49
0·57
0·56
0·54
0·61
0·59
0·57
0·61
0·63
0·63
0·65
0·71
0·71
0·72
0·62
Portugal
0·77
0·73
0·80
0·75
0·70
0·73
0·77
0·73
0·77
0·71
0·68
0·72
0·63
0·66
0·66
0·70
0·65
Rumania
1·38
1·31
1·34
1·33
1·30
1·50
1·40
1·44
1·44
1·20
1·27
1·18
1·15
1·05*
0·97*
0·99*
0·94
Slovakia
0·23
0·22
0·29
0·18
0·61
0·65
0·66
0·58
0·56
0·53
0·55
0·58
0·54
0·54
0·51
0·47
0·59
Slovenia
0·90*
1·10*
0·98*
1·06*
1·09*
1·07*
1·02*
1·21
1·03*
1·06*
0·86
0·79
0·70
0·69
0·58
0·62
0·56
Spain
0·90
0·87
0·86
0·86
0·83
0·86
0·80
0·74
0·75
0·67
0·64
0·67
0·55
0·56
0·54
0·53
0·53
Sweden
0·75
0·86
0·87
0·85
0·90
0·97*
1·09
1·18
1·26
1·13
1·25
1·27
1·27
1·24
1·27
1·25
1·27
UK
1·84
1·95
1·96
1·99
1·97
2·02
2·10
2·25
2·21
2·18
2·18
2·12
2·18
2·15
2·24
2·06
2·15
Data are deaths per 100 000 population for the European Union (EU) and rate ratios for individuals countries (95% CIs are given in the appendix). *Not statistically significant versus EU average.
Table 2: Standardised COPD mortality rates in the European Union (women)
Overall, in women, we recorded a 0·76% statistically significant decrease in ASRs, with no joinpoints identifiable (table 4). 14 countries had an increase in ASR, with Slovakia, Luxembourg, and Hungary having the steepest increases. The remaining EU countries had decreasing trends between 1994 and 2010, with Greece having the steepest decrease.
Discussion Our findings substantiate other reports of decreasing COPD mortality rates in the EU,7 but with some additional key findings. First, although men have a much higher COPD mortality rate than women in all countries, the trend, in the EU overall and by country, is a decrease in the mortality rates for both sexes, with steeper decreases for men than women, indicating a narrowing of the disparity between sexes. Second, these mortality trends differ between countries and sexes, with some countries having higher mortality rates than the EU average. Third, the joinpoint analysis allowed identification of timepoints at which trends changed in different countries. This information, along with knowledge of local circumstances www.thelancet.com/respiratory Vol 2 January 2014
at each timepoint, might generate hypotheses about why such changes occurred, and could help local health-policy makers to identify local factors that could affect this trend, and act accordingly. Our findings are derived from data in the Eurostat database, which are obtained from locally pooled death certificates. The quality of death certificates in different EU countries is a key limiting factor of attempts to map COPD, and has not been thoroughly assessed within chronic lung diseases to date. We know of no European study that has analysed the quality of death certificates for COPD in Europe. In 1987, Mackenbach and colleagues identified limitations with European death certificate data that could lead to systematic error.12 Findings from a 2006 study suggested that COPD is underreported on most European death certificates, which could lead to underestimation of the mortality burden attributable to COPD.13 Similar findings have been reported in the UK14 and China (panel).15 In both studies, the investigators concluded that improvements are needed in the quality of cause-of-death recording for deaths in health-care facilities and deaths at home. Periodic assessments of the 57
80
80
60
60
40
40
Belgium
160
20
20
0
0
Cyprus
120
50 40
10 0
Czech Republic
70
40 30
20
20 10
Estonia
80
120
60
80
40
40
20
0
0
0
Finland
France
80
Germany
100
60
60 40 40 20
20
0
Hungary
0
Ireland
200
120
Italy
100
Greece
45 40 35 30 25 20 15 10 5 0
80
Latvia
60 50
160
80
120
60
80
40
40
20
10
0
0
0
40
80
30
Luxembourg
120
120
100
80
80
60
80
60
40
40
20
20
0
0
Portugal
100
40 0
Romania
120
60 80
60 40
40
60 40
Sweden
60
120
120
100
40
20
Year Men (joinpoint regression lines)
10
08
20
06
20
20
02
04 20
00
20
98
20
19
19
94
0
96
20
0
19
10
08
20
06
Year Men (age-standardised mortality)
20
20
02
04 20
00
20
98
20
19
94
96
19
19
08 20 10
06
20
20
02
04 20
00
20
98
20
19
19
19
94
0
96
0
40
10
Year Women (joinpoint regression lines)
10
40
60
08
40
80
30
20
80
06
80
20
120
United Kingdom
140
50
20
Spain
160
0
02
0
04
0
20
20
00
Slovenia
160
20
20
20
0
40
98
20
Slovakia
80
100
80
60
Netherlands
160
120
100
Poland
Malta
140
20
Lithuania
19
0
94
40
20
96
Age-standardised death rates per 100 000 person-years Age-standardised death rates per 100 000 person-years Age-standardised death rates per 100 000 person-years Age-standardised death rates per 100 000 person-years
Denmark
10
0
80
0
160
50 30
180 160 140 120 100 80 60 40 20 0
30 20
40
60
40
160
Bulgaria
70 60
80
90 80 70 60 50 40 30 20 10 0
Age-standardised death rates per 100 000 person-years
Austria
100
19
50
European Union
19
Age-standardised death rates per 100 000 person-years
100
Age-standardised death rates per 100 000 person-years
Articles
Year Women (age-standardised mortality)
Figure 2: Age-standardised COPD mortality trends, by country (1994–2010) Solid lines are joinpoint regression lines. Dotted lines are age-standardised mortality.
58
www.thelancet.com/respiratory Vol 2 January 2014
Articles
quality of cause-of-death statistics might enhance their usability for health policy and epidemiological research. In the TORCH trial,16 the quality of death reports was assessed by an external expert panel. The cause of death was assigned by an independent committee, and then compared with the information given on the death certificate. The combined primary and secondary listing of death certificates presented a good concordance (610 [80%] of 911 certificates) with the final cause of death. However, COPD was under-reported in the death certificates because only 474 (52%) of 911 death certificates listed COPD as a primary cause of death. Additionally, this under-reporting was more common when the primary cause of death was not a pulmonary one.17 Therefore, our data might have underestimated the mortality due to COPD. This underestimation has been reported in Poland,18 but we know of no other EU studies that have assessed the quality of death certificates related to COPD. Additionally, we do not have information about quality control from different EU countries. Our findings are based on the underlying cause of death. We did not include multiple cause data. At the individual level, it might be difficult to establish the specific cause that led to a person’s death, especially in elderly people with multiple coexisting contributing chronic diseases. Having multi-cause data would inform our analysis of the number of people who died from COPD and also the number of people who died with COPD, which would have given a more comprehensive picture of the COPD burden. In a study done in Italy in 1996–2000, COPD death rates doubled by including not only the underlying cause but also concurrent mortality from death certificates.19 However, such information is not yet available in Eurostat. Between 1994 and 2010, there was a change of version of the International Classification of Diseases (ICD). During the late 1990s and early 2000s, nearly all EU countries switched from using the ninth to using the tenth revision of ICD. Since then, many bridge studies have been done to establish the effect of the new version of the ICD on reporting of causes of death. The results of these studies show that there has been no change in mortality from COPD.20,21 However, looking at the different countries in figure 2, it is possible to speculate on the effect of this change in ICD, for example in France, where the change in mortality trends occurred between 1999 and 2000, with a substantial drop in registered COPD deaths. In this regard, a pan-European study assessing the potential effect of the new classification of disease on the quality of death statistics is lacking. Despite findings from previous well-publicised studies indicating an upward global trend,2,22 COPD mortality in Europe is decreasing. This downward trend in mortality rates is in accordance with findings from the Global Burden of Diseases, Injuries, and Risk Factors Study (GBD) 2010,2 which reported a fall in the ASR of 43·3% in the past two decades, but contradicts GBD projections www.thelancet.com/respiratory Vol 2 January 2014
Standardised rate ratios ≥2·00 1·50–1·99 1·00–1·49 0·80–0·99 0·50–0·79
Mortality trends in chronic obstructive pulmonary disease in Europe, 1994–2010: a joinpoint regression analysis Jose Luis López-Campos, Miguel Ruiz-Ramos, Joan B Soriano
Summary Lancet Respir Med 2014; 2: 54–62 Published Online December 6, 2013 http://dx.doi.org/10.1016/ S2213-2600(13)70232-7 See Comment page 4 Unidad Médico-Quirúrgica de Enfermedades Respiratorias. Instituto de Biomedicina de Sevilla (IBiS), Hospital Universitario Virgen del Rocio / Universidad de Sevilla, Spain (J L López-Campos MD); CIBER de Enfermedades Respiratorias (CIBERES), Instituto de Salud Carlos III, Madrid, Spain (J L López-Campos); Consejería de Salud y Bienestar Social de Andalucía. Seville, Spain (M Ruiz-Ramos MD); and Fundación Caubet-Cimera, Bunyola, Islas Baleares, Spain (J B Soriano MD) Correspondence to: Dr Jose Luis López-Campos, Hospital Universitario Virgen del Rocio, Avda Manuel Siurot, s/n, 41013 Seville, Spain [email protected]
Background Findings from studies done over the past 20 years suggest that mortality from chronic obstructive pulmonary disease (COPD) is decreasing worldwide, but little information is available for trends in Europe. We aimed to describe COPD mortality trends by sex and calendar year for the period of 1994 to 2010. Methods We extracted data for COPD deaths between 1994 and 2010 in the 27 countries in the European Union (EU) from the statistical office of the EU (Eurostat), using the International Classification of Diseases 10 (ICD-10) codes J40–J44 and J47. We estimated age-standardised mortality rates (ASR), and analysed data using joinpoint regression, for women and men in the EU overall and by individual country for each year. We used the standard European population as the reference and present our findings as deaths per 100 000 person-years. We compared findings for each country with the EU average by calculating standardised rate ratios (SRR) and 95% CIs. Findings Between 1994 and 2010, there were 2 348 184 recorded COPD deaths in the EU. COPD mortality was higher in men than in women throughout the study period in all EU countries. In the EU overall, deaths per 100 000 population decreased in men almost linearly from 90·07 in 1994 to 61·33 in 2010, and in women from 26·99 in 1994 to 25·15 in 2010, representing a narrowing in gender gap over the study period. Several countries had a higher SRR mortality than the EU average—eg, Ireland, Hungary, and Belgium for men and Denmark, the UK, and the Netherlands for women. Our joinpoint regression analysis identified no statistically significant changes in the trend for the whole EU, but several countries had changing trends over the study period. In men, we recorded a 2·56% constant and statistically significant decrease in ASRs in the EU. Five countries had an increase in ASR. Overall, in women, we recorded a 0·76% statistically significant decrease in ASRs. 14 countries had an increase in ASR. Interpretation Our findings indicate a downward trend in COPD mortality in Europe between 1994 and 2010. The data also suggest a narrowing of the gap between COPD mortality in men and in women. The wide heterogeneity in mortality rates within European countries could serve as a reference to allow informed policy making. Funding None.
Introduction Chronic obstructive pulmonary disease (COPD) is a chronic respiratory disorder that causes much morbidity, mortality, and socioeconomic burden.1 COPD mortality has been constantly increasing, with projections that it will become the fourth leading cause of death by 2030.2 However, despite concordant publications indicating an increase in the number of deaths due to COPD worldwide, in the past few years several publications have reported decreases and downward projections of COPD mortality rates in various countries, such as Japan, USA, Australia, and Spain.3–6 An update of the Global Burden of Diseases, Injuries, and Risk Factors Study in 2010 estimated the annual deaths from 235 causes for 21 world regions between 1980 and 2010, by age and sex. Although COPD was the third leading cause of death in 2010, age-standardised death rates decreased by 43·3% in the past two decades, falling from 77·4 per 100 000 in 1990 to 43·8 per 100 000 in 2010.7 Worldwide, an estimated 328 million people had COPD in 2010, and COPD was the fifth major cause of years lived with disability, ranking behind back pain, major depressive disorders, iron-deficiency anaemia, and neck pain.8 54
Although very valuable, these global figures do not inform of trends within countries, many of which have different health-care systems and are at different phases of the smoking epidemic; great variability is, therefore, expected between countries. Furthermore, these studies do not provide estimates of when a change in disease mortality rate, if any, occurred. We aimed to describe trends in the risk of dying from COPD in people older than 40 years in the European Union (EU) by sex and calendar year between 1994 and 2010. This estimation might help to identify high-risk areas and provide health-care policy makers with evidence for more focused health intervention in COPD, from prevention to palliative care.
Methods Deaths in the 27 EU countries are those detailed by the statistical office of the EU (Eurostat; Luxemburg) where the International Classification of Diseases 10 (ICD-10) codes J40–J44 and J47 were selected from the years 1994 to 2010. The database was updated on April 8, 2013. We used the underlying cause of death for this analysis, which is defined by Eurostat as “the disease or injury which www.thelancet.com/respiratory Vol 2 January 2014
Articles
initiated the train of morbid events leading directly to death, or the circumstances of the accident or violence which produced the fatal injury”. Depending on the country, coding is done manually or with automated coding systems, but, in most countries, coding is done centrally in the Eurostat statistics office. Unfortunately, a dataset was not complete for all countries in the EU because some countries declared no deaths for several years in our study period and not all countries were included in the EU for the complete study period (appendix). Therefore, to infer the deaths in these years, we calculated the expected deaths by taking an average of the preceding and following year, if data were available.9 We calculated expected deaths as the product of the specific death rates in each age group and the population in each age group for those countries and years for which no data were available. Thus, once the populations were estimated, all deaths by age and sex were obtained for the entire series. We obtained populations sizes from country censuses when available for a specific year, and from the inter-censual estimates by Eurostat for other years. Age was aggregated in 5-year groups from 40 years to 85 years by sex. We estimated age-standardised mortality rates (ASR) for women and men in the EU and within each country for every year studied, using the standard European population as the reference, giving our findings as number of deaths per 100 000 person-years. Thus, ASR allowed direct comparisons between countries in this and in previous studies by adjusting for a reference population. We calculated standardised rate ratios (SRR) of countries versus the whole EU and 95% CIs using the approximation of Miettinen:10 (T1 / T2)1±Zα/2/χ, where Zα/2 = 1·96 for 95% CI and χ = (T1 – T2) / (S1 + S2)1/2. Where T1 is the country standardised rate, T2 is the EU standardised rate for each year and sex, and S1 and S2 are SEs of T1 and T2. The SRR indicates the number of times
that the mortality of a country is increased (if above 1) or decreased (if below 1) as compared with the EU average. Thus, the product of the SRR of a country in a given year and for a given sex by the standardised mortality rate for the EU is equal to the standardised rate of that country for that year and sex. We established the statistical significance of these standardised rate ratios using 95% CIs. The standardised rate trends by sex and age were estimated by joinpoint regression analysis.11 We used joinpoint regression analysis to identify years when changes occurred in the linear slope of the temporal trend. Best-fitting points (the so-called joinpoints) were chosen when the rate changed significantly. This approach has two major advantages, it identifies the time when there are changes in the trend, and it estimates the magnitude of the increase or the decrease seen in each interval by estimating the percentage of annual change (PAC). We used The Surveillance Epidemiology and End Results statistical software (Joinpoint Regression Program, version 3.5.1) to calculate PAC, trends, and graphs, and the statistical significance of ASRs. We built segmented models of Poisson regression to estimate the trend changes over time. In this model, we used standardised rates as the dependent variable and the year of death as the independent variable. To identify a future timepoint when trends in men and women converge, we used the model equation to estimate yearly rates of mortality for both sexes until the projected death rate for men equalled that for women. In all analyses, a p value lower than 0·05 was regarded as statistically significant.
Role of the funding source There was no funding source for this study. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.
100 Age-standardised death rate per 1 00 000 person-years
See Online for appendix
Men (joinpoint regression lines) Men (age-standardised mortality) Women (joinpoint regression lines) Women (age-standardised mortality)
90 80 70 60 50 40 30 20 10 0 1994
1995
1996
1997
1998
1999
2000
2001
2002 Year
2003
2004
2005
2006
2007
2008
2009
2010
Figure 1: Age-standardised COPD mortality trends in the European Union (1994–2010) Solid lines are joinpoint regression lines. Dotted lines are age-standardised mortality.
www.thelancet.com/respiratory Vol 2 January 2014
55
Articles
Results Between 1994 and 2010, there were 1 475 061 recorded COPD deaths in men and 873 123 recorded COPD deaths in women. Age-standardised COPD mortality rates in men decreased linearly (figure 1), although there was much variability between countries (table 1). The downward trend over the same period in women was less pronounced (figure 1; table 2). The male-to-female ratio in COPD mortality decreased from 3·3 in 1994 to 2·4 in 2010. If these trends in both sexes were to continue, we estimate that COPD mortality would not differ between sexes by 2031 (appendix). Countries such as Belgium, Denmark, Hungary, Lithuania, Netherlands, Spain, and the UK had significantly higher COPD mortality rates in men than the EU average (table 1, figure 2). This excess mortality was present in the initial years of the time period studied and was sustained over time. Countries such as Bulgaria, Cyprus, the Czech Republic, Greece, and France had SRRs significantly below the EU average, whereas some countries—eg, Germany, Austria, or Luxembourg—have changes in SRRs in line with EU average (table 1). In
European Union
women (table 2), Denmark had the highest risk of dying from COPD each year, reaching a three-times increase by 2010 compared with the EU average. Ireland, UK, and the Netherlands had the next highest mortality rates. Small countries such as Malta and Cyprus had the lowest mortality SRRs. Figure 3 (men) and figure 4 (women) show maps of SSRs for each country in the EU. Findings from our joinpoint regression analysis, assessing trend changes over time, are shown in table 3 for men and table 4 for women; figure 2 shows graphs for trends in both sexes. In men, we recorded a 2·56% constant and statistically significant decrease in ASRs in the EU, with no joinpoints identifiable (table 3). By country, during the entire period, Slovenia, France, and Ireland had the steepest decreases in ASR (table 3). Only five countries had an increase in ASR, with Cyprus, Slovakia, and Greece having the steepest increases. Over the entire study period, PACs in some countries such as Austria, Denmark, Finland, France, Latvia, the Netherlands, Poland, Slovakia, and Sweden peaked and then decreased, with an overall downward trend.
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
73·84
64·86
64·78
61·33
90·07
92·36
89·09
87·09
86·79
87·68
78·32
72·64
78·19
69·48
71·91
66·65
67·48
Austria
0·66
0·70
0·69
0·71
0·71
0·75
0·83
0·89
0·97*
1·12
1·09
1·06
1·09
1·10
0·99*
0·96*
Belgium
1·53
1·57
1·61
1·56
1·63
1·53
1·71
1·84
1·63
1·54
1·58
1·52
1·45
1·47
1·50
1·46
1·54
Bulgaria
0·58
0·57
0·59
0·71
0·64
0·56
0·64
0·58
0·55
0·55
0·62
0·78
0·84
0·92
0·85
0·70
0·65
Cyprus
0·27
0·26
0·28
0·28
0·28
0·27
0·31
0·33
0·32
0·30
0·52
0·46
0·49
0·54
0·48
0·65
0·53
Czech Republic
0·67
0·52
0·41
0·46
0·56
0·56
0·70
0·71
0·68
0·68
0·69
0·89
0·75
0·84
0·84
0·93
0·80
1·01*
Denmark
1·23
1·34
1·22
1·47
1·48
1·48
1·61
1·75
1·58
1·48
1·44
1·33
1·44
1·45
1·51
1·58
1·67
Estonia
0·77
0·65
0·59
0·51
0·56
0·59
0·67
0·61
0·64
0·67
0·92*
0·67
0·86
0·84
0·93*
0·68
0·78
Finland
0·72
0·79
0·87
0·90
0·95*
0·86
0·93
0·97*
0·91
0·83
0·79
0·85
0·94*
0·88
0·83
0·84
0·82
France
0·67
0·69
0·74
0·72
0·75
0·73
0·45
0·45
0·45
0·46
0·45
0·47
0·45
0·44
0·45
0·42
0·44
Germany
0·98
0·99*
0·98
0·90
0·85
0·83
0·92
0·89
0·91
0·91
0·91
0·88
0·89
0·87
0·89
0·96
0·98
Greece
0·27
0·28
0·27
0·29
0·28
0·35
0·43
0·48
0·40
0·30
0·36
0·33
0·51
0·52
0·56
0·58
0·57
Hungary
1·67
1·57
1·48
1·35
1·26
1·44
1·35
1·22
1·26
1·38
1·41
1·76
1·94
1·95
1·88
1·82
1·89
Ireland
1·88
1·91
1·80
1·90
1·78
1·77
1·63
1·52
1·55
1·41
1·38
1·21
1·33
1·32
1·27
1·36
1·18
Italy
0·99*
0·89
0·85
0·93
0·90
0·88
0·93
0·90
0·89
1·03
0·94
0·91
0·99*
0·97
0·96
0·96
0·95
Latvia
0·54
0·52
0·60
0·62
0·53
0·49
0·49
0·51
0·59
0·54
0·61
0·64
0·63
0·77
0·64
0·55
0·64
Lithuania
1·65
1·51
1·45
1·42
1·41
1·30
1·42
1·55
1·50
1·45
1·54
1·43
1·60
1·69
1·47
1·35
1·34
Luxemburg
1·23*
1·14*
1·00*
0·99*
0·94*
1·03*
1·15*
1·16*
1·17*
1·00*
1·00*
1·03*
0·90*
0·89*
0·84*
1·05*
1·01*
Malta
1·11*
1·23*
1·17*
1·33
1·42
1·47
1·19*
0·79*
1·35
1·36
1·47
1·46
1·47
1·21*
1·15*
1·11*
1·08*
Netherlands
1·41
1·48
1·55
1·48
1·59
1·45
1·60
1·58
1·50
1·39
1·34
1·38
1·42
1·39
1·32
1·28
1·23
Poland
0·68
0·69
0·72
0·79
0·79
0·79
0·87
0·89
0·89
0·94
1·01*
1·03
1·07
1·12
1·09
1·09
0·94
Portugal
0·81
0·80
0·87
0·83
0·78
0·87
0·83
0·83
0·88
0·81
0·82
0·90
0·75
0·83
0·83
0·89
0·87
Romania
1·03*
1·01*
1·04
1·07
1·07
1·13
1·13
1·20
1·32
1·18
1·31
1·19
1·22
1·17
1·19
1·21
1·23
Slovakia
0·27
0·26
0·34
0·25
0·64
0·77
0·76
0·74
0·74
0·76
0·83
0·85
0·74
0·85
0·75
0·80
0·88
Slovenia
1·42
1·49
1·54
1·40
1·36
1·49
1·48
1·54
1·40
1·42
1·20
1·16
0·95*
0·83
0·94*
0·83
0·89
Spain
1·30
1·35
1·41
1·39
1·46
1·50
1·45
1·40
1·48
1·38
1·37
1·45
1·25
1·30
1·24
1·29
1·28
Sweden
0·49
0·48
0·51
0·53
0·55
0·55
0·59
0·62
0·63
0·59
0·63
0·62
0·64
0·61
0·64
0·60
0·63
UK
1·32
1·37
1·30
1·32
1·28
1·30
1·30
1·35
1·31
1·28
1·29
1·25
1·27
1·24
1·31
1·23
1·26
Data are deaths per 100 000 population for the European Union (EU) and rate ratios for individual countries (95% CIs are given in the appendix). *Not statistically significant versus EU average.
Table 1: Standardised COPD mortality rates in the European Union (men)
56
www.thelancet.com/respiratory Vol 2 January 2014
Articles
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
26·99
28·34
27·88
27·93
28·54
29·84
25·92
24·46
25·38
27·93
24·73
26·43
24·86
25·73
25·45
25·72
25·15
Austria
0·70
0·74
0·71
0·69
0·67
0·75
0·92
1·02*
1·07
1·24
1·23
1·11
1·22
1·12
1·01
1·13
1·12
Belgium
1·05*
1·18
1·21
1·20
1·24
1·03*
1·19
1·26
1·19
1·41
1·36
1·39
1·30
1·39
1·37
1·31
1·34
Bulgaria
0·79
0·75
0·75
0·78
0·77
0·63
0·74
0·62
0·47
0·50
0·59
0·74
0·69
0·72
0·69
0·47
0·47
Cyprus
0·33
0·32
0·34
0·33
0·32
0·32
0·37
0·38
0·37
0·33
0·64
0·46
0·49
0·43
0·41
0·40
0·40
Czech Republic
0·68
0·45
0·42
0·46
0·49
0·55
0·71
0·71
0·66
0·70
0·67
0·88
0·73
0·78
0·85
0·88
0·82
Denmark
2·54
2·68
2·72
3·01
3·28
3·21
3·82
3·87
3·69
3·13
3·15
3·00
3·22
3·27
3·03
3·22
3·29
Estonia
0·44
0·37
0·30
0·35
0·38
0·28
0·29
0·29
0·32
0·32
0·36
0·28
0·36
0·40
0·34
0·22
0·32
Finland
0·42
0·48
0·46
0·52
0·55
0·55
0·54
0·61
0·54
0·45
0·49
0·55
0·55
0·50
0·56
0·60
0·57
France
0·72
0·73
0·78
0·79
0·80
0·77
0·38
0·35
0·35
0·35
0·36
0·37
0·35
0·35
0·35
0·34
0·34
Germany
0·88
0·88
0·89
0·81
0·82
0·80
0·89
0·90
0·95
0·93
0·96
0·95
0·98*
0·98*
1·00*
1·11
1·15
Greece
0·06
0·06
0·05
0·04
0·04
0·01
0·01
0·00
0·01
0·01
0·01
0·00
0·00
0·00
0·00
0·00
0·00 1·90
European Union
Hungary
1·94
1·74
1·64
1·51
1·40
1·57
1·49
1·31
1·31
1·47
1·60
1·99
2·01
2·12
1·86
1·92
Ireland
2·59
2·75
2·61
2·70
2·68
2·59
2·53
2·45
2·05
1·91
2·13
2·03
1·91
2·23
1·90
2·09
1·91
Italy
0·88
0·77
0·73
0·78
0·79
0·76
0·80
0·73
0·77
0·92
0·86
0·80
0·85
0·83
0·84
0·84
0·80
Latvia
0·42
0·40
0·40
0·45
0·33
0·33
0·30
0·27
0·28
0·30
0·31
0·24
0·28
0·25
0·27
0·26
0·19
Lithuania
1·30
1·34
1·23
1·07*
0·95*
0·78
0·92*
0·91*
0·95*
0·77
0·75
0·76
0·62
0·77
0·55
0·50
0·54
Luxemburg
1·20*
1·22*
0·86*
0·94*
0·72
0·93*
0·77*
1·02*
1·18*
0·98*
0·86*
0·96*
1·22*
1·24*
1·28*
1·43
0·99*
Malta
0·59
0·46
0·14
0·63
0·59
0·62
0·68
0·43
0·25
0·30
0·50
0·46
0·34
0·26
0·51
0·28
0·38
Netherlands
1·31
1·41
1·45
1·48
1·56
1·53
1·74
1·74
1·64
1·62
1·59
1·70
1·74
1·70
1·72
1·66
1·65
Poland
0·49
0·47
0·49
0·57
0·56
0·54
0·61
0·59
0·57
0·61
0·63
0·63
0·65
0·71
0·71
0·72
0·62
Portugal
0·77
0·73
0·80
0·75
0·70
0·73
0·77
0·73
0·77
0·71
0·68
0·72
0·63
0·66
0·66
0·70
0·65
Rumania
1·38
1·31
1·34
1·33
1·30
1·50
1·40
1·44
1·44
1·20
1·27
1·18
1·15
1·05*
0·97*
0·99*
0·94
Slovakia
0·23
0·22
0·29
0·18
0·61
0·65
0·66
0·58
0·56
0·53
0·55
0·58
0·54
0·54
0·51
0·47
0·59
Slovenia
0·90*
1·10*
0·98*
1·06*
1·09*
1·07*
1·02*
1·21
1·03*
1·06*
0·86
0·79
0·70
0·69
0·58
0·62
0·56
Spain
0·90
0·87
0·86
0·86
0·83
0·86
0·80
0·74
0·75
0·67
0·64
0·67
0·55
0·56
0·54
0·53
0·53
Sweden
0·75
0·86
0·87
0·85
0·90
0·97*
1·09
1·18
1·26
1·13
1·25
1·27
1·27
1·24
1·27
1·25
1·27
UK
1·84
1·95
1·96
1·99
1·97
2·02
2·10
2·25
2·21
2·18
2·18
2·12
2·18
2·15
2·24
2·06
2·15
Data are deaths per 100 000 population for the European Union (EU) and rate ratios for individuals countries (95% CIs are given in the appendix). *Not statistically significant versus EU average.
Table 2: Standardised COPD mortality rates in the European Union (women)
Overall, in women, we recorded a 0·76% statistically significant decrease in ASRs, with no joinpoints identifiable (table 4). 14 countries had an increase in ASR, with Slovakia, Luxembourg, and Hungary having the steepest increases. The remaining EU countries had decreasing trends between 1994 and 2010, with Greece having the steepest decrease.
Discussion Our findings substantiate other reports of decreasing COPD mortality rates in the EU,7 but with some additional key findings. First, although men have a much higher COPD mortality rate than women in all countries, the trend, in the EU overall and by country, is a decrease in the mortality rates for both sexes, with steeper decreases for men than women, indicating a narrowing of the disparity between sexes. Second, these mortality trends differ between countries and sexes, with some countries having higher mortality rates than the EU average. Third, the joinpoint analysis allowed identification of timepoints at which trends changed in different countries. This information, along with knowledge of local circumstances www.thelancet.com/respiratory Vol 2 January 2014
at each timepoint, might generate hypotheses about why such changes occurred, and could help local health-policy makers to identify local factors that could affect this trend, and act accordingly. Our findings are derived from data in the Eurostat database, which are obtained from locally pooled death certificates. The quality of death certificates in different EU countries is a key limiting factor of attempts to map COPD, and has not been thoroughly assessed within chronic lung diseases to date. We know of no European study that has analysed the quality of death certificates for COPD in Europe. In 1987, Mackenbach and colleagues identified limitations with European death certificate data that could lead to systematic error.12 Findings from a 2006 study suggested that COPD is underreported on most European death certificates, which could lead to underestimation of the mortality burden attributable to COPD.13 Similar findings have been reported in the UK14 and China (panel).15 In both studies, the investigators concluded that improvements are needed in the quality of cause-of-death recording for deaths in health-care facilities and deaths at home. Periodic assessments of the 57
80
80
60
60
40
40
Belgium
160
20
20
0
0
Cyprus
120
50 40
10 0
Czech Republic
70
40 30
20
20 10
Estonia
80
120
60
80
40
40
20
0
0
0
Finland
France
80
Germany
100
60
60 40 40 20
20
0
Hungary
0
Ireland
200
120
Italy
100
Greece
45 40 35 30 25 20 15 10 5 0
80
Latvia
60 50
160
80
120
60
80
40
40
20
10
0
0
0
40
80
30
Luxembourg
120
120
100
80
80
60
80
60
40
40
20
20
0
0
Portugal
100
40 0
Romania
120
60 80
60 40
40
60 40
Sweden
60
120
120
100
40
20
Year Men (joinpoint regression lines)
10
08
20
06
20
20
02
04 20
00
20
98
20
19
19
94
0
96
20
0
19
10
08
20
06
Year Men (age-standardised mortality)
20
20
02
04 20
00
20
98
20
19
94
96
19
19
08 20 10
06
20
20
02
04 20
00
20
98
20
19
19
19
94
0
96
0
40
10
Year Women (joinpoint regression lines)
10
40
60
08
40
80
30
20
80
06
80
20
120
United Kingdom
140
50
20
Spain
160
0
02
0
04
0
20
20
00
Slovenia
160
20
20
20
0
40
98
20
Slovakia
80
100
80
60
Netherlands
160
120
100
Poland
Malta
140
20
Lithuania
19
0
94
40
20
96
Age-standardised death rates per 100 000 person-years Age-standardised death rates per 100 000 person-years Age-standardised death rates per 100 000 person-years Age-standardised death rates per 100 000 person-years
Denmark
10
0
80
0
160
50 30
180 160 140 120 100 80 60 40 20 0
30 20
40
60
40
160
Bulgaria
70 60
80
90 80 70 60 50 40 30 20 10 0
Age-standardised death rates per 100 000 person-years
Austria
100
19
50
European Union
19
Age-standardised death rates per 100 000 person-years
100
Age-standardised death rates per 100 000 person-years
Articles
Year Women (age-standardised mortality)
Figure 2: Age-standardised COPD mortality trends, by country (1994–2010) Solid lines are joinpoint regression lines. Dotted lines are age-standardised mortality.
58
www.thelancet.com/respiratory Vol 2 January 2014
Articles
quality of cause-of-death statistics might enhance their usability for health policy and epidemiological research. In the TORCH trial,16 the quality of death reports was assessed by an external expert panel. The cause of death was assigned by an independent committee, and then compared with the information given on the death certificate. The combined primary and secondary listing of death certificates presented a good concordance (610 [80%] of 911 certificates) with the final cause of death. However, COPD was under-reported in the death certificates because only 474 (52%) of 911 death certificates listed COPD as a primary cause of death. Additionally, this under-reporting was more common when the primary cause of death was not a pulmonary one.17 Therefore, our data might have underestimated the mortality due to COPD. This underestimation has been reported in Poland,18 but we know of no other EU studies that have assessed the quality of death certificates related to COPD. Additionally, we do not have information about quality control from different EU countries. Our findings are based on the underlying cause of death. We did not include multiple cause data. At the individual level, it might be difficult to establish the specific cause that led to a person’s death, especially in elderly people with multiple coexisting contributing chronic diseases. Having multi-cause data would inform our analysis of the number of people who died from COPD and also the number of people who died with COPD, which would have given a more comprehensive picture of the COPD burden. In a study done in Italy in 1996–2000, COPD death rates doubled by including not only the underlying cause but also concurrent mortality from death certificates.19 However, such information is not yet available in Eurostat. Between 1994 and 2010, there was a change of version of the International Classification of Diseases (ICD). During the late 1990s and early 2000s, nearly all EU countries switched from using the ninth to using the tenth revision of ICD. Since then, many bridge studies have been done to establish the effect of the new version of the ICD on reporting of causes of death. The results of these studies show that there has been no change in mortality from COPD.20,21 However, looking at the different countries in figure 2, it is possible to speculate on the effect of this change in ICD, for example in France, where the change in mortality trends occurred between 1999 and 2000, with a substantial drop in registered COPD deaths. In this regard, a pan-European study assessing the potential effect of the new classification of disease on the quality of death statistics is lacking. Despite findings from previous well-publicised studies indicating an upward global trend,2,22 COPD mortality in Europe is decreasing. This downward trend in mortality rates is in accordance with findings from the Global Burden of Diseases, Injuries, and Risk Factors Study (GBD) 2010,2 which reported a fall in the ASR of 43·3% in the past two decades, but contradicts GBD projections www.thelancet.com/respiratory Vol 2 January 2014
Standardised rate ratios ≥2·00 1·50–1·99 1·00–1·49 0·80–0·99 0·50–0·79
