We investigated mortality risks in relation to the GOLD classification of obstructive pulmonary disease in a large population-based study of middle-aged Swedish men and women. The main findings were firstly that among smokers, symptoms of chronic bronchitis were associated with increased total mortality risk among individuals with normal pulmonary function (corresponding to GOLD stage 0) and among those with mild to moderate COPD (GOLD stage 1 and 2). Secondly, among male smokers of stage 0, cause-specific mortality showed increased HR for all cancer (except lung cancer), lung cancer and other diseases. Thirdly, the total mortality hazard ratio in stages 1–4 increased stepwise with a pronounced increase in risk at a level of FEV1 below 50% predicted, suggesting a threshold effect in mortality risk at this level. Individuals with comorbidity of cardiovascular disease, diabetes and cancer at baseline were excluded and we have adjusted for age and smoking habits as completely as possible including tobacco consumption and inhalation habits.
Comparison with other studies
Prevalence of stage 0
The prevalence of individuals with stage 0 in our study was 2.2% in men and 2.4% in women. In the Copenhagen City Heart study the prevalence was slightly higher; criteria for GOLD Stage 0 were met in 5.8% of the total adult population and in 7.2% of smokers . In the National Health and Nutrition Examination Study (NHANES) study  the prevalence of respiratory symptoms among individuals with normal pulmonary function was substantially higher (approximately 13%). This is probably mostly dependent on different selection criteria due to variation in the definition of stage 0. Vestbo and coworkers limited the selection of individuals to those with symptoms of chronic bronchitis but included all individuals without airflow limitation irrespective of the level of FEV1. On the other hand the NHANES study had higher prevalence rates probably due to a broader definition of the respiratory symptoms . In our study a specific definition of symptoms of chronic bronchitis was used and we only selected individuals with normal pulmonary function, both with respect to FEV1 and the FEV1/FVC ratio. Consequently the prevalence of stage 0 was lower in our study than in the NHANES  and the Copenhagen City Heart study .
Mortality risks associated with symptoms of chronic bronchitis
Among male smokers of our study, the HR of total mortality in GOLD stage 0 was strongly significant irrespective of whether we used "ever" or "recent" symptoms of chronic bronchitis" to define stage 0. The HR:s were of a similar magnitude as for stage 2. Among men with GOLD stage 1 in men and among women with stage 2 in women, increased HR:s were noticed in smokers who reported symptoms of chronic bronchitis, implying that respiratory symptoms might be important for an adverse prognosis. However, in GOLD stage 3 we were not able to show an additive risk associated with symptoms of chronic bronchitis possibly due to low statistical power.
Similar results have previously been demonstrated in some studies. In the Copenhagen City Heart study, chronic mucus hypersecretion was associated with an increased risk of all cause mortality statistically significant in men only (relative risk: 1.3 in men and 1.1 in women) . There was no interaction between the effect of chronic mucus hypersecretion and FEV1 percent predicted with regard to total mortality which is in agreement with our results. However, also in the same study, regarding obstructive lung disease mortality, the effect of chronic mucus hyper-secretion varied with the level of ventilatory function, being weak in individuals with normal ventilatory function but more pronounced in individuals with reduced ventilatory function. This interaction was statistically significant . Furthermore, another study using the Copenhagen City Heart study population showed that chronic mucus hyper-secretion was associated with later hospitalization due to COPD, relative risks were 2.4 (1.3 to 4.5) for men and 2.6 (1.2 to 5.3) for women .
In contrast, in the National Health and Nutrition Examination Study (NHANES) the presence of respiratory symptoms was not associated with an increased mortality risk in subjects with normal lung function . The relative risk (1.2) was elevated but not statistically significant probably due to the fact that a much wider definition of respiratory symptoms was used which included as well individuals with wheeze and a diagnosis of asthma. In our study based on a similar number of deaths, but also on a more specific definition of respiratory symptoms (symptoms of chronic bronchitis) a HR of 1.6 was statistically significant among smoking men.
Occupational studies have shown results in agreement with our study. In a population of 1061 men working in the Paris area and followed for 22 years, chronic phlegm production was significantly associated with mortality: Relative risk 1.35; (p less than 0.01) . In a group of gold miners, chronic mucus hyper-secretion was not related to COPD mortality, but related to mortality in ischemic heart disease and other causes after adjustment for tobacco and dust exposure .
On the contrary, in another occupational cohort of 2,718 British men examined between 1954 and 1961, and followed for 20 to 25 years, the risk of death from COPD was correlated with the initial degree of air-flow obstruction. Among men with similar initial air-flow obstruction, age-specific COPD death rates were not significantly related to initial mucus hypersecretion .
The explanation for the relation between chronic productive cough and future mortality risk is still unclear. Previously, in the classic study by Fletcher and coworkers, conducted in a cohort of working men in London, chronic phlegm production was unrelated to the development of airway obstruction and regarded as a less important condition . Furthermore, the presence of stage 0 at baseline was not a meaningful predictor for the subsequent development of COPD in later stages after 5 and 15 years follow-up in a Danish prospective population based study .
Thus it seems that the symptoms of chronic bronchitis do not affect the FEV1 decline and development of COPD to the same extent that it affects all-cause and obstructive lung disease mortality and risk of COPD hospitalization. An explanation could be that symptoms of chronic bronchitis may be an independent pulmonary disorder associated with inflammatory and/or infectious exacerbations  and possibly also with the development of other diseases such as lung cancer  and thus not specifically associated with COPD development and progression. We were able to analyse cause specific mortality in GOLD stage 0 in male smokers. Our findings of significantly increased HR for lung cancer, all cancers, and other diseases in individuals of stage 0 support this view.
Mortality risks in stages 1–4
GOLD stage 1 in our study, as well as in the NHANES study , conveyed only a slight increase in total mortality risk of borderline significance, even lower than found for GOLD stage 0. Among former smokers there was a tendency towards a decreased risk. Thus among individuals who quit smoking when still in GOLD stage 1, the future mortality risk will be normal. In GOLD stages 2 to 4 our findings of increased mortality HR are in agreement with other previously conducted studies, as they actually reflect a successive reduction in FEV1 [21–23]. Mortality risks were also increased among never smokers, and these results are entirely in line with other studies . The NHANES study showed increased mortality risks associated with moderate to severe COPD, among smokers and former smokers of approximately the same magnitude as we report .
A methodological limitation of these findings is the measurement of pulmonary function. The protocol of the lung function tests did not meet the standards of current recommendations . The values of the internal regression equation based on height and age and the standard deviation of the equation is similar to that of the Working party of the European Community for Steel and Coal . Hence, we believe that the methodological error is similar to the one achieved after optimal spirometric technique. However there remains the possibility of misclassification due to decreased specificity in the spirometries which could have resulted in an underestimation of the results rather than the reverse.
Another possible limitation of this study is dependent on the fact that smoking prevalence in Sweden has decreased since the 1970's . Some smokers in this study are likely to have stopped smoking during the follow-up time. Smoking habits were assessed at one point in time and not repeated thereafter. Hence, we have not been able to calculate accumulated smoking exposure in terms of individual pack-years. However, even the use of pack-years has its limitations, especially with respect to recall biases in subjects with a long history of smoking .
Furthermore, we have no follow-up data on symptoms of chronic bronchitis and thus we were not able to determine whether the symptoms of chronic bronchitis were stable or not. According to previous studies symptoms of chronic bronchitis may disappear in some individuals . In those individuals, symptoms of chronic bronchitis may have a limited influence on mortality risks and could cause an underestimation of the effect of persistent symptoms of chronic bronchitis in this study. In addition, we have no spirometric follow-up data and hence we are not able to determine whether individuals with GOLD stage 0 at baseline have developed COPD in the later GOLD stages during the follow up time.
Another limitation of the study was the lower statistical power in the analyses of women. Our findings in women, especially in former smokers and never smokers should therefore be interpreted with caution.
Implications of our findings
We believe that symptoms of chronic bronchitis should be regarded as a marker for an increased all-cause mortality risk. Our data show that among middle-aged smokers, symptoms of chronic bronchitis might represent an independent risk factor for an early mortality equal to that of having moderate COPD.
It is however uncertain from previous studies whether symptoms of chronic bronchitis really are related to the development and progression of COPD . An increased risk of COPD mortality associated with chronic mucus hyper secretion among individuals with established COPD has been shown previously . In our study, symptoms of chronic bronchitis increased the risk of all cause mortality and mortality in cancer, lung cancer and other causes in smoking individuals with normal lung function (stage 0) and the risk of total mortality in individuals with GOLD stage 1 and 2. Our findings indicate that symptoms of chronic bronchitis might be an independent pulmonary disorder that may or may not be associated with COPD. It is therefore questionable whether GOLD stage 0 should be used as an "at risk" stage for COPD development in future COPD classifications.
Finally, GOLD stage 0 indicates a group of individuals with a very unfavourable prognosis similar to that of individuals with GOLD stage 2. Among smokers, symptoms of chronic bronchitis contributed to an adverse prognosis not only in individuals "at risk", but also in those with early stages of COPD with GOLD 1 and 2.