COPD-Related Mortality and Co-morbidities in Northeastern Italy, 2008-2012: A Multiple Causes of Death Analysis

Abstract

Introduction: Analysis of COPD mortality based only on the underlying cause of death (UCOD) derived from death certificates underestimates disease burden. We analyzed the burden of COPD, as well as the pattern of reporting COPD and its co-morbidities in death certificates, using multiple-cause of death (MCOD) records.

Methods: All 220,281 death certificates of decedents aged ≥40 years in the Veneto region (northeastern Italy) were analyzed through 2008–2012. The UCOD was selected by the Automated Classification of Medical Entities software. COPD was defined by ICD-10 codes J40-J44 and J47 based either on the UCOD or on any mention of COPD in death certificates (MCOD). Annual age-standardized COPD death rates were computed for 40–85 year-old subjects.

Results: COPD was mentioned in 7.9% (and selected as the UCOD in 2.7%) of death certificates. In about half of these, COPD was mentioned in Part II only. After circulatory and neoplastic diseases, the most frequent chronic diseases reported in certificates with any mention of COPD were diabetes (15.2%) and dementia/Alzheimer (8.9%). Between 2008 and 2012, age-standardized death rates (/100,000/year) decreased from 39.8 to 34.0 in males and from 12.7 to 11.3 in females in the UCOD analyses. These trends were confirmed, although figures were three times greater, in the MCOD analyses.

Conclusions: MCOD analysis should be adopted to fully evaluate the burden of COPD-related mortality. Our findings support a decreasing trend in COPD-related mortality in northeastern Italy between 2008 and 2012, in line with other recent studies in Europe and beyond.

Keywords:

• cause of death
• chronic obstructive pulmonary disease
• co-morbidity
• death certificate
• medical records
• observational study

Introduction

Chronic obstructive pulmonary disease (COPD) is a global leading cause of mortality (1). Assessments of COPD mortality based on death certificates are thought to underestimate the mortality burden associated with the ­disease. This may be due to different reasons: COPD is widely undiagnosed even during its severe stage (2, 3); COPD diagnosis is underreported on death certificates (2); lastly, even when mentioned in death certificates, COPD is not selected as the underlying cause of death (UCOD) (4).

In fact, several causes of death can be reported on death certificates by certifiers. Part I of a certificate reports the causal sequence from the immediate cause of death, to intermediate cause(s), to a single underlying cause of death, defined as the disease or injury which initiated the train of morbid events leading directly to death. Part II of a certificate includes other significant conditions contributing to death. Standard mortality statistics are based on internationally adopted algorithms that identify the UCOD from all the conditions reported on a certificate (5). The UCOD generally ­corresponds to the underlying cause stated by a certifier, but it could also be another disease reported in Part I or II, or a derived condition (6). Analyses based not only on the underlying cause, but on all diseases reported on a certificate (multiple causes of death, MCOD), can more fully represent the burden of COPD-related mortality. Moreover, MCOD analysis makes it possible to describe the association between COPD and co-morbidities at the time of death (4).

MCOD analyses for COPD have been carried out in the United States (7), in Western and Northern Europe (4, 8, 9), in China (10), and Australia (11). However, from Italy as well as from the whole southern Europe, MCOD data on COPD are lacking, and even analyses based on the UCOD are scarce (12–14). Moreover, in contrast to other chronic diseases such as diabetes (15,16), to date analyses aimed at describing the pattern of COPD reporting in death certificates and at estimating the associated probability of selection of COPD as the UCOD, have not been carried out.

Although the number of deaths due to COPD has been projected to raise worldwide (1), some recent studies showed a decrease in COPD mortality over the last years, both in Europe (13, 17), Asia (10, 18), and the United States (19). Because of this conflicting information, up-to-date estimates of COPD mortality are needed.

The aims of this study are to analyze recent mortality data from the Veneto region (northeastern Italy) in order to investigate the frequency and patterns of COPD mention and COPD co-morbidities in different sections of death certificates, and the burden of COPD mortality according to analyses based on the UCOD or on MCOD.

Methods

A copy of death certificates of all decedents in the Veneto Region (about 4,900,000 inhabitants) is transmitted to the Regional Epidemiology Department for coding of the causes of death according to the International Classification of Diseases, 10^th Edition (ICD-10). Since 2008 the regional mortality database includes not only the UCOD, but all the diseases mentioned, and the UCOD is selected by the Automated Classification of Medical Entities (ACME) software, which has become an international standard adopted in many countries (16).

Mention of COPD was searched among MCOD records of the period 2008–2012 to retrieve all COPD-related deaths among subjects aged ≥40 years according to the following ICD-10 codes: bronchitis, not specified as acute or chronic (J40); simple and mucopurulent chronic bronchitis (J41); unspecified chronic bronchitis (J42); emphysema (J43); other chronic obstructive pulmonary disease (J44); bronchiectasis (J47). Based on the position where COPD was reported, the following classification into mutually exclusive categories was adopted: COPD in the line reserved for the underlying cause 1) alone, 2) mentioned before other causes of death, 3) mentioned after other causes of death); 4) COPD not reported as the underlying cause, but mentioned in another line of Part I; and 5) COPD only reported in Part II of a certificate.

Proportional mortality (percentage of all registered deaths) and age- (5-year classes between 40 and 89 years, and ≥90 years) and gender-specific mortality rates were computed for COPD selected as the UCOD, and for COPD mentioned anywhere in a certificate. Population data were derived from the National Institute for Statistics (http://demo.istat.it/). Weekly and monthly figures were extracted to assess seasonal variations in COPD-related deaths, measured both as absolute numbers and as proportional mortality rates.

Age-standardized mortality rates, based both on the UCOD and on the MCOD analyses, were estimated for men and women and for every year studied, to explore temporal trends in COPD mortality. Mortality rates were truncated at <85 years and the European standard was used as the reference population with the purpose of comparing death rates in the Veneto region with the data reported by López-Campos et al. (17).

Lastly, among all COPD-related deaths, the distribution of the UCOD according to major disease categories, and the prevalence of common chronic co-morbidities reported in any position of death certificates, were investigated.

Results

Out of 220,281 decedents aged ≥40 years in the Veneto region through 2008-2012, COPD was mentioned in 17,474 (7.9%) death certificates. COPD was reported in Part I in 48% of these death certificates, and only in Part II in the remaining 52%. Overall, COPD was selected as the UCOD in 34% of deaths with any mention of the disease, corresponding to 2.7% of all regional deaths. When COPD was mentioned in Part I, the probability of selection of COPD as the UCOD was 62% overall. This probability increased to 77% when COPD was reported in the line corresponding to the underlying cause, and to more than 85% when COPD was the first or the only cause reported in this line, whereas it dropped to 33% when COPD was mentioned in other lines of Part I (Figure 1). When only mentioned in Part II, selection of COPD as the UCOD was limited to 8% of deaths, mainly when pneumonia, respiratory failure, and unspecified respiratory disorders were reported in Part I.

Figure 1.  Number of deaths with any mention of COPD, by position where the disease was mentioned in death certificates, and selection as the underlying cause of death (UCOD): subjects aged ≥40 years, Veneto region (Italy), 2008–2012.

Proportional mortality from COPD was higher for men than for women, both in the UCOD- and in the MCOD-based analyses (Table 1). COPD was mentioned among the diseases leading or contributing to death in more than 13% and 7% of certificates in males and females aged ≥80 years, respectively. Mortality rates increased steeply with age, with a male to female ratio peaking at 3.5 in the 75–79-age class. The ratio between rates based on the MCOD analysis and rates based on the UCOD analysis tended to decline with age in males (from >4 in those aged 60–69 to <3 in those aged ≥85 years), and to be stable in females.

Table 1.  Number of deaths, proportional mortality (% of all deaths), and mortality rates (/100,000/year) for COPD selected as the underlying cause of death (UCOD) or mentioned in any position of death certificates (MCOD): Veneto region (Italy), 2008–2012

	Number of deaths		Proportional mortality		Rate (/100,000 /year)
	UCOD	MCOD	UCOD	MCOD	UCOD	MCOD
Males
40–44 yrs	3	7	0.3%	0.6%	0.3	0.7
45–49 yrs	5	13	0.3%	0.7%	0.5	1.3
50–54 yrs	7	28	0.3%	1.1%	0.8	3.4
55–59 yrs	31	105	0.8%	2.7%	4.2	14.1
60–64 yrs	48	204	0.8%	3.3%	6.7	28.5
65–69 yrs	103	423	1.2%	4.8%	16.4	67.3
70–74 yrs	246	907	1.9%	7.2%	45.1	166.3
75–79 yrs	549	1702	3.3%	10.1%	137.1	425.2
80–84 yrs	941	2804	4.5%	13.5%	350.3	1043.9
85–89 yrs	884	2589	4.8%	14.1%	680.3	1992.4
≥90 yrs	628	1524	5.4%	13.1%	1477.8	3586.1
Overall	3445	10306	3.3%	9.9%	54.3	162.5
Females
40–44 yrs	0	1	0.0%	0.2%	0.0	0.1
45–49 yrs	3	6	0.3%	0.5%	0.3	0.6
50–54 yrs	8	17	0.5%	1.1%	1.0	2.1
55–59 yrs	8	27	0.4%	1.3%	1.1	3.6
60–64 yrs	30	76	0.9%	2.3%	4.0	10.1
65–69 yrs	47	157	1.0%	3.4%	6.8	22.8
70–74 yrs	115	334	1.6%	4.5%	17.8	51.7
75–79 yrs	223	673	1.9%	6.2%	40.0	120.8
80–84 yrs	512	1479	2.4%	7.1%	107.6	310.8
85–89 yrs	700	2030	2.4%	7.0%	222.5	645.2
≥90 yrs	812	2368	2.4%	7.1%	538.2	1569.4
Overall	2458	7168	2.1%	6.2%	34.4	100.3

Deaths from COPD showed a pronounced seasonal variability (Figure 2). Considering men and women together, the ratio between the number of deaths occurring in January-March (cold season) and those registered in July-September (warm season) through the study period was 1.20 for all causes of mortality, 1.50 when COPD was mentioned in a death certificate, and 1.81 when COPD was selected as the UCOD. Accordingly, the proportional mortality rates ranged from 2.2% in the warm season to 3.3% in the cold season in the UCOD-based analysis, and from 7.1% to 8.9% in the MCOD-based analysis.

Figure 2.  Weekly number of deaths of subjects aged ≥40 years with COPD selected as the underlying cause of death (UCOD), or mentioned in any position of death certificates (MCOD): Veneto region (Italy), 2008–2012.

In men, death rates for COPD in the Veneto region decreased between 2008 and 2011 (Figure 3, panel A). In the same period, a small decrease was evident in women (Figure 3, panel B). In both genders, death rates slightly increased back or reached a plateau in 2012. This was likely due to a cold spell occurred in Italy in February 2012 (20), when a peak in COPD-related deaths was observed beyond baseline seasonality (Figure 2). This peculiar time trend is evident both in the UCOD and in the MCOD analyses. The generally decreasing tendency observed for both sexes in our study is in line with the trend reported by López-Campos et al., who analyzed COPD mortality in the EU using the UCOD (Figure 3) (17); however, UCOD-based death rates were about half in the Veneto region, with respect to the EU.

Figure 3.  Age-standardized COPD mortality rates (/100,000/year) in the Veneto region between 2008 and 2012, in men and women (panel A and panel B, respectively) aged 40–84 years, based on UCOD and MCOD analyses (dashed and solid black lines, respectively), and in EU between 2000 and 2010 based on UCOD analysis^a (solid gray line).

Table 2 shows the distribution of the most common UCOD in subjects with any mention of COPD in their death certificates. Deaths from respiratory diseases, including COPD, were outweighed by those from circulatory disorders. Among the latter, chronic or acute ischemic heart diseases were the most common category. The third nosological sector was represented by neoplasms, mainly lung cancer. The prevalence of the most common chronic disorders retrieved in any position of death certificates among decedents with any mention of COPD is reported in Table 3. After the main categories of circulatory and neoplastic diseases, the most frequent COPD co-morbidities mentioned were diabetes (15.2%) and dementia/Alzheimer (8.9%).

Table 2.  Underlying cause of death (UCOD) according to the International Classification of Diseases, 10^th Edition (ICD-10) in 17,474 certificates mentioning COPD: subjects aged ≥40 years, Veneto region (Italy), 2008–2012

Disease (ICD–10)	n	%
Diseases of the respiratory system (J00–J99)	6221	35.6%
COPD (J40–J44, J47)	5903	33.8%
Asthma (J45–J46)	22	0.1%
Other respiratory diseases (J00–J39, J50–J99)	296	1.7%
Circulatory diseases (I00–I99)	6300	36.1%
Hypertensive diseases (I10–I15)	1083	6.2%
Ischemic heart diseases (I20–I25)	2719	15.6%
Other heart diseases (I00–I09, I26–I51)	1429	8.2%
Cerebrovascular diseases (I60-I69)	856	4.9%
Other circulatory diseases (I70–I99)	213	1.2%
Neoplasms (C00–D48)	2505	14.3%
Lung cancer (C33–34)	1022	5.8%
Other neoplasms (C00–C32, C35–D48)	1483	8.5%
Diabetes (E10–E14)	381	2.2%
Dementia, Alzheimer (F01–F03, G30)	455	2.6%
Diseases of the digestive system (K00–K99)	468	2.7%
Other causes of death	1144	6.5%

Table 3.  Specific chronic diseases reported according to the International Classification of Diseases, 10th Edition (ICD–10) in 17,474 certificates mentioning COPD: subjects aged ≥40 years, Veneto region (Italy), 2008–2012

Disease (ICD–10)	n	%
Asthma (J45–J46)	121	0.7%
Hypertensive diseases (I10–I15)	4806	27.5%
Ischemic heart diseases (I20–I25)	5211	29.8%
Cerebrovascular diseases (I60–I69)	2551	14.6%
Neoplasms (C00–D48)	3949	22.6%
Lung cancer (C33–34)	1248	7.1%
Diabetes (E10–E14)	2666	15.2%
Dementia, Alzheimer (F01–F03, G30)	1560	8.9%

Discussion

The burden of COPD-related mortality was estimated to be three times greater when assessed by means of a MCOD-based analysis with respect to when it is assessed using standard UCOD statistics. The large proportion of deaths where COPD was mentioned but not selected as the UCOD partly depends on patterns of death certification. Due to ageing of the regional population and the associated increase in several coexisting chronic diseases, death often results from a complex interaction between multiple factors with no simple etiologic chain leading to the identification of a single underlying cause. It may therefore be difficult for certifiers to decide whether to report COPD in Part I (which implies that the disease directly caused death), or in Part II (which indicates that COPD contributed to death but was not part of the causal chain directly causing death). For other chronic diseases, patterns of death certification have been investigated, and were found to influence differences in UCOD-based statistics between countries (16). For COPD, these data are lacking. The present study shows that, in northeastern Italy, more than half of death certificates report COPD in Part II, greatly reducing its probability of selection as the UCOD.

Even in MCOD analyses, mention of COPD greatly varies between countries. In France, in the period 2000–2002, COPD was mentioned on death certificates in only 3% of all deaths of adults aged 45 years or older (4). By contrast, obstructive lung diseases (including COPD and asthma) were mentioned on about 8% of death certificates in England and Wales in 1993–1999 (8), and in the United States in 1979–1993 (7). Proportional mortality in our study (7.9%) was more similar to the latter estimates. We did not include asthma codes in our analysis to achieve comparability with recent studies on COPD mortality based on ICD-10 (17). In our data, asthma was rarely mentioned in COPD-related deaths (Table 3), and adding asthma codes did not substantially change UCOD or MCOD-based mortality rates (data not shown). However, some deaths recorded as caused by

COPD may have been due to undiagnosed asthma or to asthma with fixed airflow obstruction (4). In fact, doctors propensity to diagnosing more frequently COPD than asthma tends to increase with patient's age (21). Thus, particularly later in life when it is more difficult to distinguish between the two conditions (8), certifying physicians may be more likely to attribute deaths to COPD rather than to asthma.

Within a clinical trial on patients with moderate to very severe COPD, rates of exacerbations showed a two-fold increase in winter with respect to summer months; all-causes mortality followed a similar seasonality (22). Physiologic effects of weather, seasonal variations in respiratory tract infections and air pollution may be responsible for this pattern (22). The present mortality data confirm a strong seasonal variability of the risk of death especially when COPD is the UCOD, but also when COPD is mentioned in any position of a death certificate. Moreover, the excess risk in the cold season, with respect to the warm season, seems to be greater for deaths from (and with) COPD than for all-cause mortality.

Death rates from COPD tended to decrease between 2008 and 2012, especially in men (Figure 3). This updates and corroborates recent evidence by López-Campos et al. (17), who reported that COPD mortality declined in men between 1994 and 2010 in the EU and in Italy, while it remained stable in women. It must be remarked that our analysis also showed that death rates based on the UCOD were about half in the Veneto region with respect to the EU, for both men and women, and that death rates increased by about three times in MCOD-based analyses (Figure 3). It is therefore important that the MCOD analysis confirmed the downward trends observed in the UCOD analysis, indicating that not only deaths from COPD but also deaths of subjects with COPD are declining in northern Italy. One major reason for these temporal trends is the change in tobacco consumption during the last decades (23, 24). However, further studies should be performed to evaluate alternative hypotheses, such as temporal changes in coding habits of certifiers, or improving treatment of patients with COPD (14).

Previous reports have already shown that the most common causes of death where COPD is mentioned but not selected as the UCOD are ischemic heart diseases and lung cancer, due both to common risk factors (tobacco), and to a higher mortality in patients with lung cancer or cardiovascular disorders who are also affected by COPD (4, 8). COPD is related to several other co-morbidities, including cerebrovascular diseases, diabetes, osteoporosis, anxiety and depression, which may interact with COPD and increase mortality (17). In the present study, many of the above co-morbidities displayed a high prevalence in patients with COPD, unveiling the complex net of disorders contributing to death especially in the ageing population.

The main limit of the study is the lack of sensitivity and specificity measures for COPD mention in death certificates. In a sample of COPD patients participating in the TORCH trial, there was good agreement between causes of death reported in certificates and cause of death assigned by an independent committee, but COPD was markedly underreported in death certificates (3). In hospital discharge records, it has been documented that algorithms underdetect acute exacerbations, which are an important terminal cause of death in COPD (25). In a cohort study on mortality of subjects evaluated by spirometry at baseline, the probability of COPD mention in death certificates increased with COPD severity, but it was still below 50% in subjects with very severe COPD (GOLD stage 4) (2). Furthermore, deaths attributed to COPD occurred also in subjects with normal spirometry at baseline, possibly due to COPD development during follow-up, or to disease misclassification especially in subjects with cardiac disorders causing breathlessness. Thus, care must be taken when interpreting epidemiological figures based on mortality statistics alone.

Conclusions

This study underlines that MCOD analyses should be adopted to more fully evaluate the burden of COPD-related deaths. Moreover, further research is needed on how different patterns of COPD mention within death certificates may affect the probability of selection of COPD as the UCOD. Lastly, MCOD analyses can help to understand the role that the co-existence of multiple chronic diseases play in increasing the risk of mortality. These are all fundamental steps to understand the limits of COPD-related mortality data, to assess co-morbidities associated to COPD, and to properly compare figures from different countries or time periods. Finally, the decreasing trend in COPD-related mortality in the Veneto region confirms the findings from other studies, possibly supporting the effectiveness of anti-smoking campaigns in Italy and Europe.

Abbreviations

GOLD, Global initiative for obstructive lung diseases; ICD-10, International classification of diseases, X edition; MCOD, multiple causes of death; UCOD, underlying cause of death

Declaration of Interest Statement

The authors report no conflicts of interest. UF and AM conceived the idea for this paper, planned and conducted the statistical analysis and drafted the manuscript. AM, MS, UF discussed the analyses, contributed to manuscript drafting and critically reviewed the manuscript.

The authors alone are responsible for the content and writing of the paper.