Occupational Chronic Obstructive Pulmonary Disease in a Danish Population-Based Study

Abstract

The aim was to explore the impact of occupation on chronic obstructive pulmonary disease (COPD) in a cross-sectional population-based study among subjects aged 45 to 84 years. In a stratified sampling 89 general practitioners practices (GPP) in Denmark recruited 3106 males and 1636 females through the Danish Civil Registration System. COPD was defined by spirometry by the 2.5^th-centile Lower Limit of Normal of FEV₁ and FEV₁/FVC. Information about smoking, occupational exposure and the respective occupations were obtained from questionnaires. Occupations followed the Danish adaptation of The International Standard Classification of Occupations, revision 1988 (DISCO-88). Exposure to vapour, gas, dust (organic and inorganic), and fume (VGDF) in each occupation (yes/no) was evaluated by two independent specialist in occupational medicine. Exposures were divided in no, low, medium, and high exposure as 0, <5, 5–14, and ≥ 15 years in the job, respectively. Data was analysed by a mixed random effect logistic regression model. The age-standardised COPD study prevalence was 5.0%. Of 372 DISCO-88 codes 72 were identified with relevant exposure to VGDF. 46% of the participants reported at least one occupation with VGDF exposure. Adjusted for smoking, age, sex, and GPP a dose-dependent association of COPD was found among workers in jobs with high organic dust exposure, with OR 1.56 (95% CI 1.09–2.24). Restricted to agriculture the OR was 1.59 (95% CI: 1.08–2.33). No association was observed for workers in jobs with inorganic dust, fume/gas, or vapour exposures. In summary, occupational organic dust exposure was associated to the prevalence of COPD.

Keywords :

• epidemiology
• occupational exposure
• organic dust
• prevalence study

Introduction

Chronic Obstructive Pulmonary Disease (COPD) is a common and widespread disease (1,2), with a wide range of co-morbidities (3). In the Global Burden of Diseases, Injuries, and Risk Factors Study 2010 (1) COPD is ranked as 7^th in disability-adjusted life years (DALY) in Western Europe. Smoking is the major risk factor for COPD, but worldwide, up to half of COPD cases are in fact due to nonsmoking causes, e.g., exposure to biomass smoke; occupational exposure to vapour, gases, dust and fumes (VGDF); history of pulmonary tuberculosis or chronic asthma; outdoor air pollution, and poor socioeconomic status (4). Growing evidence suggests that about 15% of COPD could be attributable to occupational exposure (5–7). Currently available treatments have minimal impact on progression of the disease (8) so attempts to prevent COPD are attractive and favourable, and occupational exposures can be prevented or reduced in many trades. The aim of the present study was to estimate the occupational contribution to the risk of COPD in a cross-sectional population-based Danish study of 45–84-year-old participants.

Methods

Population

The study is based on baseline data from the North Jutland COPD Prevention Study (NCPS) from 2004 (9). The population was selected from two former counties in Denmark (North Jutland and Viborg) with together 299,000 inhabitants aged 45–84 years (January 2005). In Denmark all citizens have free access to medical care provided by a general practitioner. All 480 general practitioners (GP) in the two counties were invited to participate, Figure 1 (Flow chart). The study recruited patients from 155 GP situated in 89 practices (GPP). The Danish individual personal 10-digit Civil Registration System was used to select a randomly, but age and sex stratified sample of persons aged 45–84 with an overweight group of elderly and men, based on the expected Danish prevalence of COPD in 10-year groups (10).

Figure 1.  Flow chart of the study enrolment population.

The selected subjects received an invitation by mail in which they were requested to contact their general practitioner in order to participate in the study. The participants consisted of a mixed urban/rural population. In the period October 2004–September 2006, in all 4742 (36%) of the invited 13,087 persons entered the study, and consulted their general practitioner. Applied baseline data consisted of a self-administered questionnaire including a question of prior asthma; “Have you ever had asthma?” (Yes/No), spirometry tests, height and age. Participants with a prior lung cancer were excluded (n = 25), leaving 4717 participants for analysis in the current study.

Spirometry

Pulmonary function tests by spirometry [forced expiratory volume per second (FEV₁) and forced vital capacity (FVC)], were performed by the general practitioner or a trained member of the practice staff with the general practitioners own spirometer. Volume and time calibration of the spirometers was performed before study start and every six months by trained staff using a one litre syringe. Adequate spirometry test instructions followed the statement from The European Respiratory Society (ERS) (11) and the standard from The American Thoracic Society (ATS) (12). For participants with a FEV₁/FVC ratio below 0.70 a reversibility test was performed with eight inhalations of Combivent (1 dose = 100 μg salbutamol and 20 μg ipratropium) and spirometry test was assessed after 30 min. The reference population used was The Global Lung Function 2012 equations (13) in the “GLI-2012 Desktop Software for Large Data Sets version 1.3.4 build 3” (14), which takes age, sex and race-ethnicity into account to estimate the subsequent applied z-scores.

COPD

We defined COPD by lung function measurements according to the method of Lower Limit of Normal (LLN) for FEV₁/FVC, as recommended by the ATS and ERS (15), and used the GLI-2012 prediction equations (13). In a screening setting of normal subjects the LLN is the 2.5^th-centile (z-score = −1.96). To distinguish a group of more severe airways obstruction we also added the screening criterion of the LLN for FEV₁ of a z-score = −2.0 as recommended by Quanjer et al. (16). COPD defined by LLN was estimated using the prebronchodilator values. Prevalence data was additionally presented based on the Global Initiative for Chronic Obstructive Lung Diseases (GOLD) criteria with a fixed FEV₁/FVC ratio < 0.70 and FEV₁ < 80% of the predicted value (moderate airways obstruction; GOLD 2+) (3). COPD defined by GOLD was estimated using the postbronchodilator values if prebronchodilator FEV₁/FVC ratio was < 0.70 otherwise prebronchodilator values were used.

Occupational exposure assessment

The occupational exposure assessment was based on a self-administered questionnaire, where the participants were asked for duration of exposure (5-year span) to organic dust, inorganic dust, fume/gas, and vapour, respectively. Furthermore, for each exposure type the participants could state up to three job titles and appertaining durations (5-year span). Where exposure to dust, gas, vapours or fumes were known to be present occupational codes were selected from the Danish adaptation of The International Standard Classification of Occupations, revision 1988 (DISCO-88) (17) by two specialists in occupational medicine. Final decision was agreed upon by the two specialists.

For each of the exposure agents, the total period of employment was calculated by adding the mean years from each 5-year span of each stated employment. The total duration of employment per exposure was then calculated to define four categories of cumulated duration of exposure: no exposure (0 years), low exposure (<5 years), medium exposure (5–14), and high exposure (≥15 years). When summarising the four exposures in one overall VGDF occupational exposure, the “low exposure” was defined to have specific low exposures only, “medium exposure” to have specific medium exposure but no high exposures, and finally “high exposure” to have any specific high exposure.

Smoking habits

From the questionnaire three different smoking variables were estimated, i.e, status (never, ever or current smoking), duration (years) and cumulated smoking as pack-years: ((number of cigarettes smoked per day × number of years smoked)/20). Different types of smoking were transformed to cigarettes by the following equations: one cheroot = three cigarettes; one cigar = four cigarettes, one pipe bowl = three cigarettes. Pack-years were reduced to three categories: below 10, 10–20, and above 20 pack-years, respectively.

Statistics

The chi-square test for categorical variables was used to assess differences between sub-groups of the study population. The COPD prevalence was age-standardised to the Danish population, January 2006. Univariate and mixed random effect logistic regression model (18) with GPP as random variable was used to estimate the association between COPD and occupational exposures with adjustment for pack-year, sex and age as fixed effects, and likelihood ratio tests for interaction. The McNemar test for matched data was used to compare the two methods of assessing COPD; LLN and GOLD as defined earlier. Sensitivity analyses were performed by recoding all the missing occupational exposures (i) into no exposure and (ii) into high exposure, or by (iii) excluding all participants with prior self-reported asthma. The 95% confidence intervals (CI) were calculated using a normal approximation. The significance level was set at 5%. Statistical analyses were conducted in Stata12.1 (StataCorp LP, 2011).

Ethics

The NCPS-study has been performed in accordance to the Helsinki Declaration and approved by the Danish Scientific Ethics Committee (VN2003/62) and the Danish Data Protection Agency (updated in 2007 before follow-up: 2007-41-1576). Written informed consent was obtained from all participants.

Results

Occupational exposure

Of the existing 372 DISCO-88 codes 72 were considered by the two specialists to include relevant exposure to VGDF. In the study population 27 of the 34 selected DISCO-88 codes with relevant exposure to organic dust were identified, likewise12 of 20 to inorganic dust, 2 of 3 to fume/gas, 5 of 5 to vapour, 7 of 7 to inorganic dust and fume/gas, 0 of 2 to organic and inorganic dust, and 1 of 1 to inorganic dust, fume/gas and vapour (Documented in the supplementary Table S1). About half of the population, 49%, reported no DISCO-88 code with relevant exposure, 31% reported one DISCO-88 code with relevant exposure, and 15% (n = 693) reported between two and six DISCO-88 codes with relevant exposures, while 5% did not answer the occupational question. The occupational exposures are separately illustrated by sex and level of exposure in Figure 2. Figure 2 underlines the fact that many participants (15%), especially the men, have several exposures. The exposure pillars in the figure section with VGDF are less than the sum of the specific exposure pillars.

Figure 2.  Occupational exposure in males (n) and females (n) (N = 4717). Vapour, Gas, Dust and/or Fume (VGDF) exposure: Low, “Only having specific low exposure”; Medium, “Having specific medium exposure but no specific high exposure”; and High, “Having any specific high exposure” Organic dust, inorganic dust, fume/gas and vapours occupational exposure means of duration: Low, <5 years in exposed job (excluding no exposure); Medium, 5–14 years in exposed job; and High, ≥15 years in exposed job.

An overview of the age distribution, occupational exposures and smoking habits is given by sex in Table 1. The majority of women (81%) had no relevant exposure to VGDF, while 40% of the men had a relevant high exposure to VGDF. Organic dust was the most frequent occupational exposure in both men (56%) and women (19%). A considerable number of men were also relevant exposed to other agents: Inorganic dust 24%, fume/gas 16% and vapour 6%. Overall, 23% of all subjects were current smokers and 43% were ex-smokers. Men were somewhat more likely than women to be ex-smokers (p < 0.001), whereas never smokers were more prevalent among women (p < 0.001). The mean smoking duration and number of pack-years were significantly higher in men compared to women (p < 0.001).

Table 1.  Description of the study population, N = 4717

Characteristic	Female		Male		Total
	n	(%)	n	(%)	n	(%)
Age
Age groups
45–54 years	509	(31)	400	(13)	909	(19)
55–64 years	420	(26)	649	(21)	1069	(23)
65–74 years	426	(26)	1359	(44)	1785	(38)
75–84 years	271	(17)	683	(22)	954	(20)
Total	1626	3091	4717
Mean age (SD)	62.4	(10.8)	67.0	(9.3)	65.5	(10.1)
Occupational exposure
VGDF duration of occupational exposure (n = 4474)
No exposure	1226	(81)	1104	(37)	2330	(52)
Only low^a exposures	61	(4)	309	(10)	370	(8)
Medium^b exposure (and no high)	63	(4)	360	(12)	423	(9)
Any high^c exposure	155	(10)	1196	(40)	1351	(30)
Any organic dust (n = 3883)	244	(19)	1469	(56)	1713	(44)
Any inorganic dust (n = 3776)	9	(1)	563	(24)	572	(15)
Any fume/gas (n = 3826)	23	(2)	384	(16)	407	(11)
Any vapour (n = 3701)	17	(1)	152	(6)	169	(5)
Smoking habits
Status (n = 4692)
Never smokers	722	(45)	853	(28)	1575	(34)
Ex-smokers	537	(33)	1490	(48)	2027	(43)
Current smokers	354	(22)	736	(24)	1090	(23)
Duration (n = 4614)
Mean smoked years (SD)	15.4	(17.9)	25.5	(20.8)	22.0	(20.4)
Cumulated smoking (n = 4371)
Mean smoked pack-years (SD)	8.7	(12.7)	26.0	(29.8)	20.1	(26.6)

^aLow: Exposure but < 5 years.

^bMedium: 5 to 14 years of exposure.

^cHigh: ≥15 years of exposure.

VGDF: Vapour, gas, dust and fume.

SD: Standard deviation.

All variables between sexes, p < 0.001.

Prevalence of COPD

Twenty participants either missed (n = 2), were incapable of accomplishing (n = 12) or had an insufficiently performed spirometry test (n = 6), leaving 4697 with a prebronchodilator spirometry, and 726 also had a postbronchodilator spirometry. Table 2 stratifies the cases by sex and age, and according to the LLN 2.5^th centile method the overall study prevalence of COPD was 5.9% (n = 279), and when age-standardised to the Danish population 5.0%. The study prevalence among never smokers was 1.7% (n = 26).

Table 2.   The study and age-standardised prevalence of COPD by sex and age group (n = 4697)

	COPD defined by LLN^a
	Female			Male			Total
	n	%	(95% CI)	n	%	(95% CI)	n	%	(95% CI)
Age groups
45–54	17	3.4	(1.8–4.9)	13	3.3	(1.5–5.0)	30	3.3	(2.1–4.5)
55–64	18	4.3	(2.4–6.3)	26	4.0	(2.5–5.5)	44	4.1	(2.9–5.3)
65–74	21	5.0	(2.9–7.0)	105	7.8	(6.3–9.2)	126	7.1	(5.9–8.3)
75–84	20	7.5	(4.3–10.6)	59	8.7	(6.6–10.8)	79	8.3	(6.6–10.1)
All ages	76	4.7	(3.7–5.7)	203	6.6	(5.7–7.5)	279	5.9	(5.3–6.6)
Age-standardised^b	4.6	(4.6–4.6)		5.0	(5.0–5.0)			5.0	(5.0–5.0)

COPD: Chronic Obstructive Pulmonary Disease.

CI: Confidence Interval.

^aLLN: Lower Limit of Normal; FEV₁/FVC z-score <−1.96 and FEV₁ z-score <–2.0.

^bPrevalence age-standardised to the Danish population, January 2006.

The GOLD 2+ COPD prevalence was also estimated: A postbronchodilator spirometry was conducted in 726 (77%) of the 949 participants with a prebronchodilator FEV₁/FVC ratio < 0.70. For the GOLD definition of COPD the remaining 223 with a prebronchodilator FEV₁/FVC < 0.70 and no reversibility test were excluded. The excluded subjects were by prebronchodilator lung function test distributed as: COPD by GOLD n = 119, and COPD by LLN n = 32 (discordance n = 87). The overall study prevalence of GOLD 2+ COPD was 8.6% (n = 386), and 6.8% when age-standardised to the Danish population.

In a paired comparison between the two criteria for defining COPD there was a significant difference both overall and when stratified by sex, p < 0.001. Table 3 display the overall comparison, with 173 discordant diagnoses between the LLN and GOLD definitions.

Table 3.  Comparison of LLN and GOLD methods to define COPD by spirometry in the study population

		GOLD		Total
		Normal or GOLD I	GOLD II or more severe^a
LLN	Normal or mild COPD	4071	156	4227
	Moderate or more severe^b	17	230	247
Total	4088	386	4474

LLN: Lower Limit of Normal.

GOLD: Global Initiative for Chronic Obstructive Lung Diseases.

COPD: Chronic Obstructive Pulmonary Disease.

^aFEV₁/FVC ratio<0.70 and FEV₁< 80% of the predicted value.

^bFEV₁/FVC z-score <−1.96 and FEV₁ z-score <−2.0.

Occupational exposures and COPD (LLN)

The results from the univariate and multiple regression analyses on the association between occupational VGDF exposure, organic dust exposure, and COPD are shown in Table 4. In the VGDF analysis an association between COPD and occupational exposure was indicated. The odds ratio (OR) decreased from crude to adjusted (ORadj) (pack-years, age group, sex and GPP), but was still significant in the medium exposed group, ORadj 1.61 (95% CI: 1.03–2.51) and also the trend analysis maintained significance, p < 0.05. Looking at the ­different categories of occupational exposure the ORadj for high organic dust exposure was associated to COPD and the trend in ORadjs was significant; 1.56 (95% CI: 1.09–2.24), p < 0.02, ­respectively.

Table 4.  Occupational and covariates associations to COPD in the population-based study.

	Odds Ratio (OR)
	Crude		Adjusted^a
	OR	95% CI	OR	95% CI
VGDF occupational exposure	n = 4457		n = 4132
No	1.00	Reference	1.00	Reference
Low	1.08	(0.66–1.78)	1.05	(0.61–1.80)
Medium	1.87	(1.27–2.77)	1.61	(1.03–2.51)
High	1.55	(1.17–2.05)	1.38	(0.99–1.93)
Score test for trend	p < 0.001		p = 0.031^b
Covariates
<10 pack-years			1.00	Reference
10–20 pack-years			2.60	(1.54–4.38)
>20 pack-years			7.26	(4.95–10.66)
Age 45–54			1.00	Reference
Age 55–64			1.22	(0.73–2.02)
Age 65–74			2.09	(1.33–3.29)
Age 75–84			2.28	(1.38–3.76)
Males			1.00	Reference
Females			1.78	(1.24–2.56)
Organic dust exposure	n = 3867		n = 3607
No	1.00	Reference	1.00	Reference
Low	1.22	(0.73–2.05)	1.15	(0.65–2.03)
Medium	1.49	(0.95–2.33)	1.24	(0.74–2.07)
High	1.50	(1.10–2.04)	1.56	(1.09–2.24)
Score test for trend	p = 0.006		p = 0.017^b
Covariates
<10 pack-years			1.00	Reference
10–20 pack-years			2.36	(1.34–4.16)
>20 pack-years			6.82	(4.55–10.22)
Age 45–54			1.00	Reference
Age 55–64			1.23	(0.71–2.14)
Age 65–74			2.10	(1.28–3.44)
Age 75–84			2.13	(1.23–3.69)
Males			1.00	Reference
Females			1.89	(1.30–2.77)

COPD: Chronic Obstructive Pulmonary Disease.

CI: Confidence Interval.

VGDF: Vapour, gas, dust and fume.

Bold estimates: Beyond the significance level of 5%.

^aRandom effect logistic regression adjusted for; Pack-year group: < 10, 10–20, and > 20 smoked pack-years; Age group: 45–54, 55–64, 65–74, 75–84 years of age; Sex; General practitioner practice (random variable).

^bWithout the random variable.

Excluding participants with prior asthma (n = 497) reduced the high exposure association and trend of occupational organic dust exposure to COPD into insignificance, ORadj 1.47 (0.92–2.34), p = 0.15, respectively. Restricting the analysis to patients with mild COPD and probably vague or no symptoms slightly increased the association between high organic dust exposure and COPD, ORadj 1.88 (1.13–3.13). Two main exposure groups were identified in the organic occupational dust exposure; Agriculture as the main group followed by “wood” (timber industry and carpenters). Of a total of 1926 reports agriculture was reported 1495 times (78%) and “wood” 289 times (15%). When the analysis was restricted to agriculture alone ORadj and trend remained significant; 1.59 (95% CI: 1.08–2.33), p < 0.02, respectively.

Although no evidence of associations was seen in the smaller ‘wood’ group of timber industry and carpenters. Specific inorganic dust, fume/gas and vapour occupational exposures showed no trend or evidence for an association with COPD, data not shown. Sensitivity analyses related to the multiple regression analysis of the VGDF exposure and COPD added 198 participants with missing occupational exposure to no exposure. The medium exposure and trend changed to borderline significance, ORadj 1.55 (95% CI: 1.00–2.41) p = 0.05 and p = 0.06, respectively. When adding the 723 participants with missing occupational organic dust exposure to the unexposed the high exposure estimate became insignificant in the sensitivity analyses of organic dust; both as ORadj (1.30 (95% CI: 0.93–1.80)) and in the trend analysis (p = 0.15).

Discussion

In this study we found a LLN defined age-standardised prevalence of COPD of 5.0%. Occupational organic dust exposure was in a dose-dependent manner associated to the prevalence of COPD.

COPD

In epidemiologic studies the definition of disease is often simplified and not identical with the corresponding clinical diagnose. We defined COPD by lung function measurements. The LLN approach is recommended concurrently by the ERS and ATS (15). As opposed to the GOLD fixed method, which is somewhat biased by participants age, sex and height (19), the LLN definition ensures that the participants do have a degree of airflow obstruction outside accepted population norms. However, the 2.5^th centile as a diagnostic criterion is a conservative approach, aiming to minimize the percentage of false positives at the expense of an increased number of false negatives.

Furthermore, COPD severity described in the ATS/ERS recommendation utilise the fixed FEV₁ percent predicted (15). In the present study we have analysed the data based on Quanjer et al. (16). They have recently recommended a new grading of the obstructive lung disease that is clinically relevant and free of biases related to age, height, sex and ethnic group (16). Adding the restriction of FEV₁ z-score <−2 we obtain a group of subjects having moderate airways obstruction, ­corresponding to the ATS/ERS moderate airways obstruction (16).

Measuring COPD or asthma

The 1993 official statement from ERS (11) and the 2005 ATS/ERS standardisation of lung function testing (15) do not judge a significant positive bronchodilator response to differentiate between asthma and COPD in similar ways. Furthermore, there is no evidence to clearly differentiate asthma and COPD patients by bronchodilator response (15). The response to bronchodilators varies within and between individuals (20) and when predicted reference values were established as prebronchodilator values, this lead to some overestimation of the reversibility of a low FEV₁/FVC ratio. In this study asthma was managed in sensitivity analyses by excluding participants with self-reported prior asthma.

Despite that asthma predisposes to the development of COPD (21,22) excluding these subjects from the analysis might underestimate the true association to occupation as an inclusion of the subjects is likely to overestimate the association. The best estimate might therefore be in-between these ORs.

Occupational exposure assessment

The occupational exposures were assessed by a self-administered questionnaire on exposures and occupation validated with an expert judgement. This approach is considered more sound than questionnaires alone (23), and our results support that the often used self-reported exposures might be prone to bias. This or similar combined approaches is well established and have been utilized in several population-based retrospective studies (24–26). Recall bias of occupational exposure might be introduced tending to overestimate the association between exposure and disease, although the risk was considered low, as the questionnaires were filled out before the GP examination, and thus only participants with well known COPD, might have been more aware of job exposures.

As patients with mild COPD have no symptoms, this assumption was tested by estimating the association of mild COPD and high organic dust exposure, which slightly increased the OR. This result support that our findings is not skewed by recall bias. Further, few Danes aged 45 to 84 have knowledge of an association between occupational exposure to VGDF and COPD, and the a priori selected DISCO-88 codes and expert management of each job title into DISCO-88, without awareness of COPD status, have further blinded the exposure assessment.

Smoking assessment

No common conversion of different types of smoking to amount of cigarettes were found (e.g. from the ERS). The utilised equations are clinically used at Aalborg University Hospital, Denmark and nearly the same as Bernaards et al. used in a comparison study of calculating pack-years prospectively and retrospectively (27).

Prevalence

The COPD prevalence's estimated in this study, i.e. an age-standardised prevalence of 5.0% are in accordance with other studies from Europe and US where the prevalence is reported to be between 4.5 and 10% using different diagnostic criteria (24, 28–31). We have taken a conservative approach with the FEV₁/FVC LLN of 2.5^th centile instead of the clinical used 5^th centile to reflect a screening setting. This approach gives rise to more false negative and less false positive subjects. Using the clinical FEV₁/FVC<LLN 5^th centile (z-score < −1.64) and FEV₁ < LLN −2 the age-standardised prevalence was 5.9% (n = 329). As expected the GOLD 2+ prevalence was lower in this study compared to the prior NCPS article estimating the GOLD 1+ prevalence (9) on the same data.

Occupational exposure association to COPD

A novel review by Omland et al. (32) identified 29 former population-based studies of occupational exposure to VGDF and COPD, restricting inclusion in their analysis to studies that have used ATS/ERS approved spirometric criteria for defining abnormalities in airflow obstruction and level of FEV₁ (15). There were differences in associations and exposures in the included studies, but only two with a nonsignificant association both carried out in younger populations (33,34). The present study found positive associations within the organic occupational exposure, maybe reflecting the dominating position of agriculture in Northern Denmark as the major occupation for exposure to VGDF in the study.

Associations were similar when restricting the analysis from organic dust exposure to agriculture exposure alone. Several cross-sectional studies have found an association between occupational organic dust exposure in agriculture and COPD (35–37). We found a weaker association between high organic dust exposure and COPD when excluding subjects with prior self-reported history of asthma OR 1.47 (0.92–2.34), trend p = 0.15. In the SAPALDIA study, only including nonasthmatics, the incidence rate ratio was 2.76 (95% CI: 1.32–5.75) with any organic dust exposure in ever-smokers and COPD (38). The differences in observation might be due to difference in the concentration of the exposure, low statistical power in our study or may reflect healthy-worker selection where subjects with possible prior asthma leave farming to more manageable jobs with lower occupational exposures. These associations should be addressed in prevention of new cases and deterioration of working cases.

Strengths and limitations

The limited reduction of OR in the sensitivity analyses of the high organic exposure emphasises the occupational association, although the associations become insignificant. The external study validity is considered to be high in a Western world setting with similar occupational distribution, due to expected similar occupational exposures. The enrolled study population included more young women and fewer participants in the oldest group than among nonresponders. This might introduce an age-dependent healthier study population tending to underestimate the associations. Spirometry error measurements have been managed by regular calibration, but the variations of brands among GPP were neglected for the benefit of a local experienced operator. However, the internal biologically variability in lung function was addressed by requiring three sufficient measurements as recommended by the ERS and the ATS (15).

Possible misclassifications would be of nondifferential nature and tend to underestimate the associations. Moreover, when using a spirometrically defined COPD, some COPD patients with compliance difficulties might be excluded from the analysis resulting in false low associations. Information bias of exposure was reduced by using the specialist-assessed exposure on the basis of job titles, instead of the commonly used self-reported exposure assessment. A validation of the self-reported exposures showed that job titles were often stated within the wrong exposure and, furthermore, the specialists assessed many of the job titles as having no occupational exposure.

Conclusion

In this population-based study involving 4697 subjects we found an age-standardised prevalence of COPD of 5.0%. Organic dust exposure was in a dose-dependent manner associated to the prevalence of COPD, independent of smoking habits, although the study found no associations to other less-common exposures.

Declaration of Interest Statement

The authors declare no conflicts of interest. The authors alone are responsible for the content and writing of the paper.

Supplementary material for this article is available online and can be accessed at http://dx.doi.org/10.3109/15412555.2014.974739