Background
Chronic Obstructive Pulmonary Disease (COPD) is recognized as a major public health problem with an increasing morbidity and mortality. It has been forecasted that COPD will be ranked the fourth burden of disease worldwide by year 2030 [
1]. The prevalence of COPD is most often reported in the range of 6-10% of the total adult population [
2], it is strongly correlated to smoking and age, and about 50% of elderly smokers fulfil the spirometric criteria of COPD [
3]. Population studies have shown that a large majority of COPD patients have mild-to-moderate disease [
4].
The underdiagnosis of COPD is well-known. Only about a third of all cases with COPD have been recognized by the health care [
3‐
6], and the proportion of undiagnosed cases decreases with increasing disease severity [
4]. Most reports on mortality in COPD are based on death certificates and hence, due to the underdiagnosis, the true impact of COPD on mortality is probably greatly underestimated. There are only few reports on mortality in COPD based on population studies. In a follow-up over up to 22 years of a large general population cohort in the USA, the NHANES I recruited 1971-75 follow-up in 1992 included totally 923 cases of COPD. The overall mortality in COPD was 44% and in severe COPD 71%. Subjects with mild, moderate and severe COPD, and subjects with restrictive lung function impairment, had an increased risk for death [
7]. There is a 30-year follow up of a large population sample from southern Sweden, both smokers and non-smokers with COPD had a significantly increased risk for death [
8]. The first cohort of the Obstructive Lung Disease in Northern Sweden (OLIN) studies, was recruited in 1985-86, and in a recently published 20-year follow-up an overall mortality of 54% in subjects with COPD was reported, while the mortality in severe and very severe COPD at entry was 81% [
9].
Irrespective of COPD, decline in lung function, expressed as FEV
1, is known to predict death [
10,
11]. Reduced FEV
1 has also been reported to be a marker of cardiovascular death [
12]. However, due to the underdiagnosis and lack of longitudinal epidemiological data, the present impact of COPD identified through spirometric screening in the general population on mortality considering the influence of other factors such as cardiovascular co-morbidity has been ill-described.
Within the OLIN studies, cross-sectional and longitudinal data on respiratory diseases, including lung function, have been collected in several cohorts recruited from the general population at different occasions ever since 1985 [
3,
4,
9,
13,
14]. Previously recruited adult OLIN-cohorts were invited to a clinical follow-up in 2002-2004, and from the participants all subjects with COPD were identified together with age and gender matched subjects without airway obstruction [
15]. The aim of the present paper is to report overall mortality in this cohort based on mortality data collected up to the end of 2007, and to evaluate the impact of COPD, level of FEV
1, gender, smoking habits and heart disease on mortality.
Discussion
The strength and newsworthy of the current study is the up-to date report on the impact of COPD on mortality. The identification of the study population and the observation time took place during the twenty first century when treatment for COPD as well as cardiovascular disease was recommended according to modern and current guidelines. Further, the distribution of disease severity among subjects with COPD in the cohort was representative for what has been reported for the general population, comprising of more than 90% of patients in GOLD stages I and II [
4]. COPD was a significant risk factor of increased mortality in this topical epidemiological context.
There are only a few previously published population based studies on mortality in subjects with COPD defined according to currently accepted spirometric criteria [
7‐
9,
19]. As referred to in the introduction, these study populations were recruited during the nineteen-seventies and eighties. The mortality data in these studies were based on long term follow-ups from about ten years to more than twenty years, thus a healthy survivor effect must be taken into account when evaluating these data. Further, updated guidelines during the last decade for treatment of not only COPD, but also cardiovascular disease, have improved the prognosis compared to thirty years ago. Consequently, the starting point, the time and the length of the observation time are of importance when evaluating and comparing mortality data not only in COPD but also other diseases. There are mortality data collected in two large clinical trials started during the nineties with an observation time over three and four years respectively, the TORCH and the UPLIFT studies. The all-cause mortality reported from the TORCH-study was 14.3% [
20] in a population with moderate- to severe COPD (FEV
1 < 60 percent of predicted). In the UPLIFT-study [
21] subjects with a post-bronchodilator FEV
1 < 70 percent of predicted were included, and the crude mortality in the total population was 15.4% at the end of the treatment period. After the approximately four year's observation time, similar to the UPLIFT, the mortality was 10.9% among subjects with in COPD in our study. However, the UPLIFT-cohort had a lower baseline mean FEV
1 compared with our study, further, as most clinical studies the UPLIFT-study included a selected population with regard not only to lung function but also to other factors such as age and co-morbidity. Further, in a 13-year follow-up of the ISOLDE study [
22], the mortality was 56% in the study population including subjects with moderate- to severe COPD.
The reported mortality in this study can be considered up-to-date as both the identification of the study population and the observation time took place after the turn of the century, where modern and current guidelines for treatment have been well established in Sweden. Subjects with COPD in the cohort are considered representative for the general population with regard to distribution by disease severity [
4]. There are, to the best of our knowledge, no other published similar studies. Further strengths of the study are the large size of the COPD-population, comparable to that of the NHANES I [
7], the accuracy of mortality data and that there was no loss of follow-up. However, possible limitations are the comparatively short time of follow up and that information on heart disease was self-reported and not collected from medical records. Another limitation is that the classification of COPD was strictly made by spirometric criteria without regard to respiratory symptoms. This has to be considered when interpreting the data, as respiratory symptoms are known to affect the prognosis in mild/moderate COPD [
23]. Further, the non-COPD population did also include subjects with restrictive lung function impairment, even though the ability for a dynamic spirometry to identify restrictive lung function can be questioned. The reasons for a restrictive pattern on dynamic spirometry are highly heterogeneous and reflect different underlying disorders as idiopathic pulmonary fibrosis, thoracic deformities, obesity, pleural effusion, cardiac insufficiency and neuromuscular diseases.
As expected, increasing age was the most prominent risk factor for death. Among all subjects born < 1940, the proportion of deceased was significantly larger among subjects with COPD. The use of a fixed ratio to define airway obstruction, FEV
1/FVC < 0.70, has been discussed with regard to identifying clinically relevant COPD among elderly [
24], but the results from our study indicate that the fixed ratio identifies a population with significantly increased mortality also among subjects older than 80 years. There is a recent report on 5-year mortality among subjects aged > 65 years, where COPD according to GOLD was not associated with increased mortality among those older then 75 years [
25]. However, the study population was recruited was from an out patient clinic, and can thus not be considered to reflect COPD in the general population.
When FEV
1 was included in the multivariate analyse model, COPD was no longer a significant risk factor for death, but the level of FEV
1, reflecting disease severity at recruitment, was related to mortality. Our data exemplify that a decrease in FEV
1 in the range of 100 to 50 percent of predicted will continuously increase the risk for death, and further, illustrated by Figure
1, a dramatic increased risk for death occurs when FEV
1 continues to decrease below 50 percent of predicted. It is well-known that tobacco exposure contributes to the development of both COPD and cardiovascular diseases, and cardiovascular death is the most common cause of death in the world. In a multivariate model heart disease was a significant risk factor for death just as COPD, age and male gender, however, when smoking habits were added to the model there was a slight change. Smoking roughly doubled the risk for death while male gender and reported heart disease each increased the risk on a similar level, approximately 40%, even though heart disease did not reach statistical significance as a risk factor. Impaired lung function is a known risk factor for death [
10,
11] and according to our results the risk for death in subjects with COPD, when adjusted for confounders including presence of heart disease, was increased by about 75% compared to in subjects without COPD. Maybe the impact of current treatment guidelines of cardiovascular disease has reduced mortality contributing to the borderline significance of heart disease as a risk factor while we found COPD and impaired lung function still being strong and significant risk factors for death.
Besides smoking, BMI is a known prognostic factor in COPD, and increased loss of weight is associated with a higher mortality in COPD [
26,
27]. In this study BMI could not predict mortality; however, longitudinal data on weight loss were not included. An important message is also the benefit of smoking cessation, i.e. being an ex-smoker did not differ significantly from non-smokers with regard to risk for death, while current smoking roughly doubled the risk for death. According to a 9-year follow-up from the ECRHS non-smoking non-symptomatic young adults with mild/moderate COPD do not have worse outcome than subjects without COPD [
23]. Further follow-up of our cohort will give us corresponding data from middle aged and elderly subjects with mild/moderate COPD. There was surprisingly no significant difference in prevalence of heart disease in subjects with and without COPD, however, the dominance of GOLD stage I and II in the COPD-cohort might contribute to this finding.
Conclusions
In this recently identified cohort, where COPD was mostly represented by GOLD stages I & II, COPD, age and current smoking were the strongest risk factors for death. Male gender and reported heart disease were also, and on a similar level, associated to an increased risk for death, even though heart disease did not reach statistical significance. The results further indicate that not only COPD but also impaired lung function, expressed as level of FEV1, is a significant risk factor for death independent of confounders as age, gender, smoking habits and heart disease.
Acknowledgements
Main funding has been granted from the Swedish Heart-Lung Foundation, the Northern Sweden Regional Health Authorities and Umeå University. The Swedish Research Council has supported the management of the OLIN studies large data bases. Financial support was obtained from GlaxoSmithKline R&D, World Wide Epidemiology department. Additional support has been given by the county council in the county of Norrbotten.
The authors thank Professor Anders Oden for statistical advice and analyses. The authors also thank the research assistants RSN Ann-Christin Jonsson, RSN Sigrid Sundberg and MSc Linnea Hedman for collecting the data, BA Ola Bernhoff for work with creating the data base of the study and Viktor Johansson for computerising the data. Further, the late associate professor MD PhD Staffan Andersson and MD PhD Håkan Forsberg are acknowledged for their contributions.
Competing interests
The authors declare that they have no competing interests. Hana Muellerova is an employee of GlaxoSmithKline, R&D, a producer of pharmaceuticals and owns shares and stock options of GlaxoSmithKline plc. The authors alone are responsible for the content and writing of the paper.
Authors' contributions
AL designed the study, performed the statistical analyses, drafted and revised the manuscript. LGL and HM contributed to the paper by interpretation of data and critically revision of the manuscript. ER participated in the design of the study, and contributed to the paper by interpretation of data and critically revision of the manuscript. BL designed the study, drafted the manuscript, contributed to the paper by interpretation of data and critically revision of the manuscript.
All authors have read and approved the final manuscript.