Decreased COPD prevalence in Sweden after decades of decrease in smoking

Chronic obstructive pulmonary disease (COPD) is a common disease worldwide and a major public health problem. Globally, the prevalence of COPD has increased during the past century with tobacco smoking being the most important risk factor especially in high income countries [1]. Epidemiological data have indicated that up to 50% of smokers may develop COPD sooner or later if they continue to smoke [2, 3], and the incidence of COPD is high among smokers already at younger ages [4, 5].

Since the 1970s, a substantial and ongoing decrease in smoking prevalence has been observed particularly in North America, Australia, and in several northern and western European countries [6], at least partly due to different public health interventions. Positively, some recent studies based on spirometry have indicated the prevalence trend of COPD to be levelling off or even tending to decrease [7‐10]. The substantial reduction in smoking will probably contribute to a reduced future burden of COPD.

Along with spirometric classification of chronic airway obstruction and risk of exacerbations, the recent GOLD strategy documents also emphasize the importance of assessing respiratory symptoms in the diagnosis and management of COPD [11]. However, the knowledge on prevalence trends and risk factors for COPD based on both airway obstruction confirmed by spirometry and respiratory symptoms is limited.

The aim of the current study was to estimate the prevalence, characteristics and risk factors for COPD in the Swedish general population using clinical and physiological methods according to the GOLD 2020 document [11], by using also the LLN-criterion. A further aim was to estimate the prevalence trend of COPD in Northern Sweden from 1994 to 2009. We hypothesized that the prevalence of COPD had decreased and that the risk factor pattern was altered following the major decrease in smoking in Sweden.

The study was performed in Norrbotten in northern Sweden and in Västra Götaland in south-western Sweden within two large scale population-based research projects, the Obstructive Lung Disease in Northern Sweden (OLIN) Studies, and the West Sweden Asthma Study (WSAS). The study comprises clinical examinations performed in 2009–2012 [12] including structured interviews and spirometry with reversibility testing in the age-range 21–78 years. The Masterscope (Jaeger) spirometer was used in both study areas. Approval was received from the Regional Ethical Review Boards at the Universities of Umeå and Gothenburg, Sweden. All participants signed informed consent.

The study population (Additional file 1: Fig. S1) included 1839 randomly selected responders from two large scale questionnaire surveys, which well reflected the age and gender distribution of the population in the two areas. Furthermore, the results from northern Sweden were also compared with a previous methodologically similar study in 1994 (n = 660) in the same area and using same age-span [9]. Methods are described in more detail in the e-Appendix.

The prevalence of COPD was 7.0% in ages 21–78 years in 2009–2012, and the prevalence of moderate to severe (GOLD stage ≥ 2) COPD was 3.5% according to the GOLD [11] fixed ratio definition in combination with respiratory symptoms. When the lower limit of normal definition of COPD was applied, the prevalence estimates were about 30% lower. The prevalence of COPD, particularly of moderate to severe COPD, decreased from 1994 to 2009 and the prevalence in the latter survey is low in comparison with most other studies on COPD prevalence. These results follow a substantial three to fourfold decrease in smoking prevalence in Sweden over 30 years (Fig. 3). Still, COPD was more common among men, and although current smokers were more common among women, the number of packyears and prevalence of occupational exposure was greater among men.

The scientific debate on how to best define COPD is still ongoing [15, 16], which is not surprising as relative to most diseases, COPD is an indeed “young” disease. Most guidelines on COPD rely on the fixed ratio definition of FEV₁/FVC < 0.70 recommended by GOLD [17]. Clinical physiologists recommend the use of the 5th or 2.5th percentile of the reference value for the FEV₁/FVC ratio as lower limit of normal (LLN) [18, 19] as also recommended by two ERS task forces [20, 21]. The GOLD strategy document has been revised over the years with nowadays a more pronounced role for patient-reported outcomes, e.g. symptoms and exacerbation history, and not only obstructive spirometry [11].

Differences in criteria for defining COPD together with differences in study designs, age distributions of the studied samples, and regional differences in smoking prevalence and other exposures contribute to different prevalence estimates [20, 22]. Taking all this into account, there are nevertheless possibilities for comparisons, and globally the prevalence estimates of CAO vary from about 10% to 20% among subjects > 40 years of age [23, 24]. The Burden of Obstructive Lung Disease (BOLD) study used identical methods worldwide, and studies following the BOLD protocol have resulted in a prevalence of 15–20% in most European countries in ages > 40 years, and of moderate to severe COPD on average 10%, following the GOLD fixed ratio criterion [25, 26], which is twice as high compared to our results on CAO. However, our results of about 30% lower prevalence based on LLN compared to the fixed ratio definition of COPD are quite in line with the 40% lower prevalence in BOLD [26]. A meta-analysis of European studies estimated the CAO prevalence to be 13.7% in 2010 [27], and recent studies in Canada and within the US NHANES have ended up in similar prevalence [28, 29]. Similar but slightly variable data were reported from Scandinavian countries after the millennial shift, as summarized in Table 5 [3, 8‐10, 30‐34]. Regarding severity of COPD, several of these studies have found considerably higher prevalence of moderate to severe COPD defined as GOLD grade ≥ 2 than our results, although it should be noted that some of these referred studies are based on pre-bronchodilator spirometry [8, 10, 32].

In contrast to the large number of cross-sectional surveys studying prevalence of CAO, only few have allowed for studies of trends in prevalence. The Spanish IBERPOC [35] was followed after 10 years by the EPI-SPAN and showed a substantial decrease in the prevalence of CAO GOLD grade ≥ 2 [7], and the Norwegian HUNT study found a decreased prevalence of CAO GOLD grade 2 over eleven years in parallel with decreased smoking prevalence from 29 to 17% [10]. The repeated studies from US NHANES and Finland with follow-up periods of about 20 years found no major prevalence change but a trend towards a decrease in both CAO and smoking [8, 28]. Also based solely on spirometry results, a previous study in the Northern Sweden area revealed a decrease in the prevalence of CAO GOLD grade ≥ 2 over 15 years following a decrease in smoking prevalence from 27 to 16% [9]. Thus, CAO prevalence seems to be decreasing in some high income countries with decades of decreasing smoking prevalence [6, 36].

In line with results from studies in high income countries worldwide [11], smoking remained by far the dominating risk factor for COPD in our study. The magnitude of smoking as risk for developing COPD remained high, but in contrast to our first study on COPD prevalence [3], less than half of elderly smokers had developed COPD. This may be a consequence of both lower numbers of cigarettes smoked per day and a decrease of exposure to environmental tobacco smoke in work places, homes, restaurants, and public places as a result both of large scale society actions at schools and workplaces, in media, and of legislation as well. The proportion of never smokers among subjects with COPD remained low and on similar level as in our previously performed studies [3, 37]. Still COPD was more common among men than women, although the smoking prevalence in Sweden has not been higher in men for decades. Our results indicate that women might have benefited more from the substantial decrease in smoking prevalence, results supported by studies showing that women are more susceptible to the harmful effects of smoking than men [38]. The higher COPD prevalence among men is probably a consequence of the higher number of packyears among smoking men than smoking women along with differences in occupational exposures. Men had a history of more occupational exposure to GDF, which was independently associated with COPD, in line with previous results [12, 39]. Several large industries are located in the study areas, and previous Swedish studies have shown an increased COPD mortality among construction workers [40]. Further, although we did not see independent associations between low socioeconomic status and COPD, it has been shown in previous studies [3, 41, 42]. Thus, it is important that, besides continuing to target smoking rates, public health strategies also aim to reduce the levels of occupational exposures and socioeconomic differences.

As the major part of subjects with CAO but no symptoms were > 60 years of age, adding symptoms to the spirometric definition of chronic airway obstruction, in accordance with recent GOLD recommendations [11], reduces the age-related bias associated with the fixed ratio definition of COPD. As found by others [43, 44], under-diagnosis of COPD was still huge in 2009, although our results show a decreasing trend regarding under-diagnosis. Early recognition of COPD is important, also in younger ages [45, 46], and as it improves, the burden of COPD on the health care system will remain huge despite decreasing prevalence. In line with results from studies by us and others [47, 48], multi-morbidity was common among subjects with COPD, inferring further challenges for care providers. However, a positive result was that the mean age among those with COPD was higher in 2009 than in 1994, implying disease onset at older age and thus improved public health, likely due to successful interventions and legislation targeting smoking.

Overall, the results strongly emphasize the importance of continuous smoking cessation measures, not only among subjects with COPD but among all smokers, along with continuous measures to prevent young people from smoking initiation and political measures to ban smoking from society. Although prevalence decreases, the under-diagnosis of COPD is still substantial. Spirometry is accessible and should be performed in all individuals with respiratory symptoms and/or those extensively exposed to risk factors, in order to enable early intervention and treatment of COPD.

Our study has several strengths, both the OLIN and the WSAS are large studies well reflecting the general population with high participation rates. Studies of non-response have indicated good representativeness of the results for the populations in both the WSAS [49] and OLIN [50] areas, although potential healthy volunteer effects never can be completely ruled out. The study staffs were trained together in order to avoid inter-observer bias, and the same spirometer brand, the Jaeger´s Masterscope, was used in both areas. Furthermore, post-bronchodilator spirometry results were used and quality assurance of the spirometry curves was performed. A further strength is the simultaneous analysis of the fixed ratio and LLN to define CAO. Reassuringly, when including symptoms in the COPD definition, only few subjects were differently classified by LLN. However, there are also weaknesses with our study, e.g. that repeated surveys to enable analyses of time trends only was available from northern Sweden and that there always is a possibility for inter-observer bias when performing structured interviews.

The prevalence of COPD has decreased in Sweden, and it was considerably lower in 2009–2012, i.e. 7.0%, compared to most previous studies in high income countries including the Scandinavian countries. The prevalence of moderate to severe COPD was particularly low, only 3.5%. The low prevalence follows a three to fourfold decrease in smoking prevalence over the past 30 years. However, smoking still remained by far the most dominating risk factor for COPD.