Socioeconomic Status and Prognosis of COPD in Denmark

Abstract

We investigated the association between length of school education and 5-year prognosis of chronic obstructive lung disease (COPD), including exacerbations, hospital admissions and survival. We used sample of general population from two independent population studies: The Copenhagen City Heart Study and Copenhagen General Population Study. A total of 6,590 individuals from general population of Copenhagen with COPD defined by the Global initiative for obstructive lung disease criteria were subdivided into 4 groups based on the length of school education: 1,590 with education < 8 years; 3,131 with education 8–10 years, 1,244 with more than 10 years, but no college/university education and 625 with college/university education. Compared with long education, short education was associated with current smoking (p < 0.001), higher prevalence of respiratory symptoms (p < 0.001) and lower forced expiratory volume in the first second in percent of predicted value (FEV₁%pred) (p < 0.001).

Adjusting for sex, age, FEV₁%pred, dyspnea, frequency of previous exacerbations and smoking we observed that shortest school education (in comparison with university education), was associated with a higher risk of COPD exacerbations (hazards ratio 1.65, 95% CI 1.15–2.37) and higher risk of all-cause mortality (hazards ratio 1.96, 95% CI 1.28–2.99). We conclude that even in an economically well-developed country with a health care system (which is largely free of charge), low socioeconomic status, assessed as the length of school education, is associated with a poorer clinical prognosis of COPD.

Keywords :

• COPD
• prognosis
• socioeconomic status
• survival
• exacerbations

Introduction

Previous studies have shown that socioeconomic status (SES) is associated with development of COPD (1–9). A recent systematic review by Gershon et al. reported that individuals with lowest school education had on average twice as high risk of COPD than those with highest level of school education (10). Lifestyle and environmental factors like smoking, poor diet, prenatal exposures, high number of respiratory infections in childhood and work related exposures to dust and fumes have been suggested as possible mechanisms (11).

The influence of SES on the clinical outcomes of COPD is, however, not as well described as the association between SES and incidence and prevalence of COPD. Once COPD has developed, SES may influence the prognosis of the disease if SES is related to available treatment, adherence to medication and smoking status. Yet, a possible negative influence of SES on prognosis of COPD could also be mediated through the disease severity as such, since previous reports have suggested that low SES is associated with more severe COPD (10).

The aim of the present study was to investigate the influence of SES, assessed as the length of school education, on the clinical prognosis of COPD. We investigated the risk of exacerbations, hospital admissions and mortality in individuals with COPD as a function of different lengths of school education.

We adjusted our analyses for severity of disease by including clinical characteristics of COPD used to stage the patients according to the most recent GOLD document, including the level of lung function, severity of dyspnea and history of previous exacerbations (12). We utilized the data from two large population-based studies in the area of Copenhagen. Some of the findings have been reported earlier in a form of an abstract (13).

Methods

Study population

We pooled data from two similar but independent studies: The 4th examination of The Copenhagen City Heart Study (CCHS) in 2001–2003 and the examination of The Copenhagen General Population Study (CGPS) in 2003–2008. These studies were approved by institutional review boards and Danish ethics committees (KF100.2039/91, H-KF01-144/01), and were conducted according to the Declaration of Helsinki. Written informed consent was obtained from all participants.

Copenhagen City Heart Study (CCHS)

The CCHS is a prospective epidemiologic study initiated in 1976–1978. A sample of 19,698 subjects aged 20 to 100 years was selected at random from the national Danish Civil Registration System, after age stratification in 5-year age groups, from residents of inner Copenhagen, and 14,223 subjects participated in the initial survey at Copenhagen University Hospital (14). They were all re-invited to participate in later surveys along with additional subjects in the youngest age groups. A total of 6,237 attended the 4^th survey in 2001–2003 (response rate 50%) (15).

Copenhagen General Population Study (CGPS)

CGPS is a prospective epidemiologic study that aims to recruit more than 100,000 subjects representative of the general population and collect genotypic and phenotypic data of relevance to a wide range of health-related problems. It is designed almost identical to the CCHS. Recruitment began in 2003 and is still ongoing (response rate 45%) (16).

Study parameters

In both studies, participants filled in an extensive questionnaire concerning lifestyle and health topics, including medication use at home. A physical examination was performed, where basic variables such as weight and height were measured. A Vitalograph spirometer (Maids Moreton, Buckinghamshire, UK) was used in the CCHS and in the first 14,624 participants in the CGPS. An EasyOne Diagnostic Spirometer (ndd Medizintechnik, Switzerland) was used in the remaining participants in the CGPS.

Spirometry was performed in the standing position without the use of a nose clip. Three sets of forced expiratory volume in the first second of expiration (FEV₁) and forced vital capacity (FVC) values were obtained and as a criterion for correct performance of the procedure at least two measurements differing by less than 5% had to be produced together with the correct visual appearance of the spirometry tracings. The highest obtained values for every single participant of both FEV₁ and of FVC were used. Only pre-bronchodilator measurements were available.

Study design

Individuals with COPD were identified according to international COPD guidelines, based on FEV₁/FVC < 0.7. In accordance with the GOLD document we excluded individuals with self-reported asthma (12). In the present analyses, we only included individuals older than or equal to 40 years. Among 5,919 participants of the CCHS and 55,731 participants of the CGPS we identified 849 individuals from CCHS and 5,741 from CGPS fulfilling the abovementioned COPD criteria who also had sufficient information on the length of school education and household income. They were subdivided according to the length of school education into following groups:

– Short school education: shorter than 8 years: 1,590 individuals;

– Middle school education: 8–10 years: 3,131 ­individuals;

– Long school education: longer than 10 years, but no university or college: 1,244 individuals;

– University or college education: 625 individuals

These categories correspond to lower primary school, higher primary school, secondary school, and university or college respectively. We have chosen to use the length of school education as the variable characterizing SES, since this variable has been used more frequently in previous studies on prevalence and incidence of COPD than income (10), and is considered a more stable indicator of the socioeconomic position than household income, which can change periodically and profoundly after retirement.

In the analyses we included various demographic and clinical variables describing the participants with COPD on the basis of the information obtained in the questionnaire at the time of the examination. For the clinical grouping we used the GOLD 2011 classification with groups A-D on the basis of FEV₁ in percent of predicted value (FEV₁%pred), the self-reported severity of dyspnea (modified Medical Research Council (mMRC)) and the frequency of COPD exacerbations in the last year (12). To define exacerbations, both in the year before the examination and during the follow-up, we utilized ­information regarding hospital admissions due to COPD and the use of oral corticosteroids and antibiotics related to treatment of exacerbations. The methodology for obtaining these data has been described in detail previously (17).

Outcomes

After the examination date, the participants were followed by means of the nationwide registries up to 8.9 years with an average of 4.3 years.

The following clinical outcomes were defined:

• COPD exacerbation (defined on the basis of the medication used) until December 31, 2009.

• Admission to hospital due to COPD until December 31, 2009.

• All-cause mortality until August 17, 2010.

This was done using the unique personal civil registration number assigned to all Danish inhabitants and by linking our surveys to two national registries: The Danish National Patient Registry covering all hospital contacts in Denmark and the Danish Registry of Medicinal Product Statistics, which contains information on all prescriptions dispensed in all Danish pharmacies (18). We identified the relevant medications using the Anatomic Therapeutic Chemical (ATC) code. From the Danish National Patient Registry, we identified hospital admissions with a discharge diagnosis of COPD (International Classification of Diseases 10th edition: J41-44). An exacerbation of COPD was defined as a short course treatment with prednisolone alone or in combination with an antibiotic or an acute admission to hospital because of COPD. Survival data was obtained by connecting our databases to Danish Central Person Register, which contains information on all deaths in Denmark.

Statistics

All analyses were performed with R version 2.15.1 (R Foundation for Statistical Computing, Vienna, Austria). For demographics, ANOVA was used for continuous variables and chi-square tests were used for categorical variables. To account for censoring of data, the Kaplan-Meier estimator was used to estimate the percentage of events during follow-up. Hence, all percentages presented are 100% minus the Kaplan-Meier estimate of being event-free. The log-rank test was used to compare the differences in exacerbations, hospital admissions, and mortality within the four subgroups with short, middle, long and university/college education.

To estimate the relative risk of the clinical events for different school education groups we used Cox regression model with age as underlying time scale in order to adjust for age differences between the four education groups. In so-called univariate models, only sex was included in addition to the length of school education, which was the variable of main interest, whereas the most comprehensive model also included clinical characteristics of COPD: FEV₁%pred., dyspnea as assessed by mMRC, previous exacerbations, smoking, exposure to dust and fumes, dietary intake of vegetables, cohabitation, use of inhaled medication for COPD, use of cardiac medication, body mass index (BMI), and cough and sputum.

Results

A total of 6,590 individuals fulfilled our COPD inclusion criteria based on spirometry, age and absence of asthma and had the necessary information on the length of school education.

Demographics, social and lifestyle characteristics of individuals with COPD according to the length of school education are shown in Table 1. As expected, those belonging to the group with shortest school education also had the lowest household income. They were also older and had been more exposed to occupational dusts and fumes, had the lowest daily consumption of fruits and vegetables and were more often sedentary compared to individuals with longer education. In general, the proportion of ever-smokers is very high, which is typical for an urban Danish population of this age and there is a strong relationship between current smoking and short school education.

Table 1.   Demographic and lifestyle variables of 6 590 individuals with COPD at baseline according to length of school education

	Length of school education
	< 8 yearsN = 1 590	8–10 yearsN = 3 131	> 10 yearsN = 1 244	UniversityN = 625	p value^1
Males,%	50	46	41	64	<0.001
Age, years (SD)	72 (8)	66 (11)	62 (13)	63 (13)	<0.001
Living alone,%	41	33	32	25	<0.001
High household income,%^2	2	14	30	56	<0.001
Smoking					<0.001
Never,%	14	18	24	28
Former,%	41	39	41	46
Current,%	45	43	35	27
Exposed to dusts and fumes at work,%	27	16	7	2	<0.001
Sedentary,%^3	12	10	7	7	<0.001
Fruit consumption ≥ 1 a day,%	45	50	58	59	<0.001
Vegetable consumption ≥ 1 a day,%	28	37	51	62	<0.001

¹p values were calculated using ANOVA for continuous variables and chi-square test for categorical variables.

²Defined as those belonging to the highest income tertile of all participants in The Copenhagen City Heart Study and The Copenhagen General Population Study, which corresponds to more than 80,000 Euros per year before taxation.

³Sedentary individuals were defined as individuals who only engaged in light physical activity as walking for less than 2 hours a week.

The clinical characteristics of individuals with COPD differed across the length of school education (Table 2). As our study comprises individuals from the general population, the average FEV₁%pred was in general high in all education groups and the prevalence of individuals treated with inhaled medications was low (10%). The distribution of GOLD A-D was dependent on the education group, e.g., significantly more individuals from the shortest school education group belonged to the most severe GOLD D category (Table 2).

Table 2.  Disease characteristics of the 6 590 individuals with COPD at baseline according to length of school education

	Length of school education
	< 8 years N = 1,590	8–10 years N = 3 131	> 10 years N = 1 244	University N = 625	p value^1
FEV1,% pred (SD)	76 (21)	79 (19)	82 (18)	85 (18)	<0.001
mMRC > 2,%	28	18	12	9	<0.001
Chronic bronchitis,%	25	20	16	12	<0.001
Wheezing,%	32	30	24	18	<0.001
BMI, kg/m2 (SD)	26 (4)	26 (4)	25 (4)	25 (3)	<0.001
GOLD grouping					<0.001
A, N (%)	1,071 (67)	2,424 (77)	1,053 (85)	548 (88)
B, N (%)	318 (20)	444 (14)	126 (10)	41 (7)
C, N (%)	78 (5)	134 (4)	37 (3)	22 (4)
D, N (%)	123 (8)	129 (4)	28 (2)	14 (2)
On any inhaled medication,%	13	10	8	6	<0.001
On LABA, LAMA or ICS,%	9	7	6	5	0.001
On cardiac medication,%	15	9	8	8	<0.001

¹p values were calculated using ANOVA for continuous variables and chi-square test for categorical variables.

Abbreviations: mMRC: modified Medical Research Council Dyspnea scale, BMI: body mass index, GOLD: global initiative for obstructive lung disease, LABA: long-acting beta-2-agonist, LAMA: long-acting antimuscarinic agent, ICS: inhaled corticosteroid.

Also, the prevalence of respiratory symptoms and use of respiratory and cardiac medication was highest in the group with shortest education. The prevalence of treatment with inhaled medications was related to the GOLD group: only 5.3% of individuals in GOLD group A were treated, whereas the corresponding prevalence was 19.6% in GOLD B, 18.8% in GOLD C and 52.3% in GOLD D. There were however, no differences between the frequencies of treatment in relation the length of school education within similar GOLD groups which suggests that the differences in the use of inhaled medications shown in Table 2 are related to different distribution of the GOLD A-D groups within the four education group, rather than to differential treatment within similar GOLD groups.

Table 3 shows the number of end-points observed during follow-up, whereas Table 4 shows the results of the Cox regression. Both in the univariate and multivariate models, the group with the shortest school education had the highest risk of the clinical outcomes studied (Table 4). Inclusion of GOLD A-D groups reduced the hazard ratios of all outcomes (Model 2), in particular COPD hospital admission, but the association between short school education and poor prognosis remained significant (Table 4).

Table 3.  Five-year prognosis of the participants according to length of school education

	Length of school education
	< 8 yearsN = 1,590	8–10 yearsN = 3 131	> 10 yearsN = 1 244	UniversityN = 625	p value
COPD exacerbations, N (%)	231 (17.7)	328 (13.0)	88 (9.3)	37 (7.2)	<0.001
COPD admissions, N (%)	146 (11.8)	165 (7.2)	32 (3.3)	14 (2.6)	<0.001
All-cause mortality, N (%)	218 (17.6)	279 (12.3)	64 (7.7)	28 (5.7)	<0.001

All percentages are Kaplan-Meier estimates.

p values are based on log-rank trend test for time to event data.

Table 4.  The risk of exacerbations, hospital admissions and death during 5 years follow-up among participants with COPD according to level of school education

	Model 1Univariate^1	Model 2^2 Adjusted for variables defining GOLD A-D	Model 3^3 Adjusted for variables defining GOLD A-D and smoking	Model 4^4Adjusted for disease severity and other relevant confounders
	HR (95% CI)	HR (95% CI)	HR (95% CI)	HR (95% CI)
COPD exacerbations
<8 years	1.82 (1.28–2.59)	1.73 (1.22–2.47)	1.65 (1.15–2.37)	1.62 (1.10–2.37)
8–10 years	1.54 (1.09–2.16)	1.61 (1.14–2.27)	1.56 (1.10–2.22)	1.54 (1.06–2.23)
>10 years	1.24 (0.84–1.82)	1.45 (0.98–2.14)	1.41 (0.95–2.09)	1.40 (0.93–2.12)
University	1 (ref)	1 (ref)	1 (ref)	1 (ref)
COPD admissions
<8 years	2.65 (1.52–4.60)	1.89 (1.09–3.30)	1.74 (0.98–3.11)	1.55 (0.86–2.81)
8–10 years	1.99 (1.15–3.45)	1.66 (0.96–2.88)	1.60 (0.91–2.84)	1.41 (0.79–2.52)
>10 years	1.31 (0.70–2.46)	1.30 (0.69–2.45)	1.28 (0.67–2.45)	1.15 (0.60–2.22)
University	1 (ref)	1 (ref)	1 (ref)	1 (ref)
All-cause mortality
<8 years	2.19 (1.48–3.26)	1.89 (1.27–2.82)	1.96 (1.28–2.99)	1.86 (1.20–2.90)
8–10 years	1.92 (1.30–2.85)	1.79 (1.21–2.65)	1.89 (1.25–2.87)	1.90 (1.24–2.92)
>10 years	1.51 (0.96–2.36)	1.42 (0.91–2.24)	1.51 (0.94–2.42)	1.50 (0.93–2.42)
University	1 (ref)	1 (ref)	1 (ref)	1 (ref)

¹Adjusted for age (underlying time scale) and sex.

²Adjusted for age, sex and FEV₁%pred., mMRC, frequent exacerbations, the latter three included as binary variables defined according to GOLD A-D.

³Adjusted for variables in model 2 and smoking (never smoker, former smoker and current smoker).

⁴Adjusted for age, sex, FEV₁%pred. (continuous), mMRC (groups 0-4), exacerbations (at least 2 vs. less than 2 last year), smoking (never smoker, former smoker and current smoker), former or current substantial exposure to dust and fumes at work (yes vs. no), dietary intake of vegetables (at least 1 vs. less than 1 a day), cohabitation (cohabitating with a partner vs. living alone), use of inhaled medication (yes vs. no), use of cardiac medication (yes vs. no), BMI (4 groups according to World Health Organisation), and daily cough and sputum (yes vs. no).

Additional inclusion of smoking (Model 3) did not change the estimates significantly for exacerbations and death, but short school education was no longer significantly related to COPD admissions. The inclusion of additional variables as confounders in the most advanced model (Model 4) did not change the estimates much and the hazards ratios were almost unchanged compared to the model including age, sex, smoking and GOLD groups (Model 3) (Table 4).

Discussion

The present study shows that the length of school education is significantly associated with the clinical prognosis of COPD in the general population. This is in line with a number of previous studies, showing a higher risk of COPD mortality and COPD health service utilization in individuals with lowest SES (10). In addition to being associated with high prevalence and incidence of COPD, low SES has also been associated with reduced lung function, high prevalence of respiratory symptoms and reduced health related quality of life (10,19–22). Yet, as most previous studies have not adjusted for the severity of COPD, the poor outcomes could have been caused by a more severe disease in individuals with low SES. The present study expands previous findings by showing that, even after meticulous adjustment for the severity of COPD including the GOLD A-D grouping comprising the level of lung function, dyspnea and frequency of exacerbations, the individuals with shortest school education had the highest risk of exacerbations, hospital admissions due to COPD, and death. Thus, although COPD was most severe among those with the shortest education, the severity as such could not solely explain the worst prognosis in this group.

After inclusion in the Cox model of a number of additional variables describing both COPD characteristics (cough, sputum, and BMI), occupational exposures, lifestyle variables (smoking and diet) and treatment of both COPD and cardiac comorbidities, education remained significantly related to exacerbations and survival and the effect size with regard to COPD admission remained virtually unchanged.

Although there are likely to be multiple yet unknown underlying mechanisms responsible for the association between low SES and high risk of COPD, it is important to consider what can be done in order to improve the prognosis of COPD in individuals with low SES and thus to reduce the apparent disparity in health (11).

Firstly, a possible explanation for the association between short school education and poor prognosis could be insufficient medical treatment. The latter could be caused by under-diagnosis of COPD and under-treatment, which both are most pronounced in individuals with low SES (23). Under-diagnosis and under-treatment are without doubt also present in our cohort, as only app. 50% of individuals with COPD and FEV₁%pred below 50 were treated with inhaled ­medication (24).

Yet, we could not demonstrate any social gradient with regard to being treated with COPD medications after adjusting for severity by GOLD A-D grouping. Although treatment with inhaled medications in the present cohort seems to be unrelated to the length of school education, we cannot exclude that adherence to prescribed medication was higher among individuals with the longest school education, as a number of studies has previously reported lowest medicine adherence in subjects with low SES together with a lower likelihood of being enrolled in a rehabilitation program (25–27).

Secondly, different smoking habits across the different lengths of school education could also be responsible for differences in COPD prognosis. A Spanish study by Miravitlles et al. including app 4,500 patients with COPD reported the highest prevalence of never-smokers among those with the lowest SES (22).

This is in contrast to the present study, where we found that 45% of those with the shortest school education were still smoking compared to 27% among those with the longest school education. This difference between Spain and Denmark can be attributed to the fact that Spain as other south European countries entered the state of smoking epidemic decades later than did Denmark. However, inclusion of smoking habits in the Cox regression only resulted in minor modifications of the hazard ratios for length of education on all study outcomes (Table 4).

In addition to differences regarding treatment and smoking, factors like unemployment, lack of social relations and living alone, which are frequent in those with low SES could negatively influence prognosis. Yet, although related to mortality as such and to increased use of health services, these factors are difficult to intervene on in a clinical setting.

Finally, the frequency and severity of comorbidities, like heart disease, could also be responsible for poorer outcome of COPD in individuals with the shortest school education (28). Although we in the present study used treatment for ischemic heart disease as an indicator for presence of an important cardiovascular comorbidity, we cannot completely rule out such residual ­confounding.

The present study of the general population has certain weaknesses and strengths. The strengths include the prospective design, large number of outcomes, comprehensive registration of respiratory risk factors and complete follow-up regarding hospital admissions and mortality. Regarding diagnosis of COPD admissions based on hospital records, a previous study has identified a high specificity and an adequate completeness of COPD diagnosis in Danish Hospitals (29). Diagnostic inaccuracy may bias results if it is related to socioeconomic status but we find this unlikely, particularly as healthcare is free of charge in Denmark. As we probably underestimated the true incidence of hospital admission due to COPD, we also included milder COPD exacerbations defined by medical treatment (17).

The weaknesses include lack of information on possible changes in smoking habits during the follow-up and the fact that we only had prebronchodilatatory spirometric values, but we do not think that this influences our conclusions regarding the role of SES to any major extent. Another weakness is the response rate of just below 50%. We know that in particular individuals with shortest education have a lower attendance than those with longest educations, but we have no reasons to believe that the relationship between school education and COPD prognosis differs between the responders and nonresponders. Finally, our statistical approach can be debated, as inclusion of a great number of variables in the advanced models, could result in over-adjustment and weaken the role of socioeconomic status. Yet, it was our aim to adjust in our analyses for the severity of COPD as thoroughly as possible.

In conclusion, our analysis of a large sample of individuals with COPD selected from the general population shows that short education is associated with poor prognosis in already established COPD, and that neither disease severity nor modifiable risk factors such as smoking are fully mediating this effect. Further studies are needed to explain the apparent disparity in prognosis of COPD related to socioeconomic status and there is a need to develop clinical strategies in order to reduce the inequality concerning COPD across social strata. We suggest, that clinicians attending patients with COPD should pay additional attention to patients with low socioeconomic status with regard to future control and adherence to medications and healthy ­lifestyle.

Declaration of Interest Statement

Funding: This work was supported by Herlev Hospital, Copenhagen University Hospital, Copenhagen County Foundation, and University of Copenhagen, all from Denmark.

Ethical approval: The studies were approved by institutional review boards and Danish ethics committees (KF100.2039/91, H-KF01-144/01), and were conducted according to the Declaration of Helsinki. Written informed consent was obtained from all participants.

Conflicts of Interest: None of the authors has any conflicts of interest to declare.

Author contributions: None of the funders had any role in the design and conduct of the study; collection, management, analysis and interpretation of the data; or preparation, review and approval of the manuscript.

Guarantor: PL had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analyses.