COPD PREVALENCE IN A RANDOM POPULATION SURVEY: A MATTER OF DEFINITION

Abstract

A recent ATS and ERS joint taskforce report recommends using a lower limit of normal (LLN) of FEV₁/FVC as opposed to a fixed ratio of <0.7 to diagnose airflow obstruction, to reduce false positive diagnoses of COPD as defined by the Global Initiative for Obstructive Lung Disease (GOLD). To date there is no reliable spirometry-based prevalence data for COPD in New Zealand and the effect of different definitions of airflow obstruction based on post-bronchodilator spirometry is not known.

Detailed written questionnaires, full pulmonary function tests (including pre- and post-bronchodilator flow volume loops) and atopy testing were completed in 749 people recruited from a random population sample.

The GOLD-defined age-adjusted prevalence for adults ≥ 40 years old was 14.2% (95% CI, 11.0-17.0). This compared to a LLN-defined, age-adjusted, post-bronchodilator prevalence, in the same group of 9.0% (95% CI, 6.7 to 11.3).

The prevalence of COPD varied markedly depending on the definition used. Further research using longitudinal rather than cross-sectional data will help decide the preferred approach in COPD prevalence surveys.

BACKGROUND

The significance of chronic obstructive pulmonary disease (COPD) as a cause of global morbidity and mortality is undisputed. Within the next twenty years COPD is projected to move from the sixth to the third most common cause of death worldwide, whilst rising from fourth to third in terms of morbidity within the same time frame [1].

For such a significant condition, there is a relative dearth of accurate prevalence information, a major difficulty being a lack of consensus about the definition of COPD [2]. A number of different approaches have been used including self reporting/doctor diagnosis, diagnosis based on the presence of respiratory symptoms and a diagnosis based on the presence of various definitions of airflow limitation (pre- or post-bronchodilator). The Global Initiative for Chronic Obstructive Lung Disease (GOLD), published in 2001 and now updated yearly, has resulted in agreement on spirometry thresholds for diagnosis and severity and has become the “gold standard”, at least for epidemiological purposes. GOLD defines COPD as a post-bronchodilator ratio of forced expiratory volume in one second to forced vital capacity (FEV₁/FVC) of < 0.7[3]. A very similar definition (post-bronchodilator FEV₁/FVC ≤0.7) has been agreed on by the American Thoracic Society (ATS) and European Respiratory Society (ERS) [4].

Whilst it is generally accepted that FEV₁/FVC is the most important guide when identifying airflow obstruction, the practice of classifying values of FVC and FEV₁ less than 80% predicted and a fixed FEV₁/FVC ratio of <0.7 as abnormal has no statistical basis [5]. Since FEV₁/FVC ratios fall with age a fixed ratio results in an apparent increase in the prevalence of impairment associated with aging or with age-confounded factors such as cigarette smoking [5]. However one statistically acceptable approach for establishing lower limits for any spirometric measure is to define the lowest 5% of the reference population as below the lower limit of normal (LLN) [5]. In contrast to the ATS,ERS and GOLD definitions of COPD a recent joint ATS/ERS Taskforce has proposed using a cut-off for the FEV₁/VC ratio set at the 5^th percentile of the normal distribution rather than at a fixed value of 0.7 in an attempt to reduce the number of false-positive diagnoses [6]. They also noted that a slow VC manoeuvre may be more accurate than using FVC to diagnose airflow obstruction. However, the two studies on which the LLN recommendation was made did not include post-bronchodilator spirometry [7,8], which is a pre-requisite for the definition of airflow limitation that is not fully reversible [3,4,9].

In this study we aimed to establish the GOLD-defined prevalence of COPD in an adult urban NZ population for the first time. Secondly we aimed to compare the GOLD-defined prevalence rate to that obtained with the LLN definition of FEV₁/SVC and FEV₁/FVC using both pre- and post-bronchodilator values.

METHODS

Study subjects

A total population of 3,500 people, randomly selected and equally divided between each of 5 age groups (25-34, 35-44, 45-54, 55-64, 65-74 years) at age of selection, were sent a postal screening questionnaire (SQ). Subjects were identified from the New Zealand electoral register for five electoral wards in the greater Wellington region plus the separate Maori electoral register (the main ethnic minority) constrained for the same geographical area. Subjects who completed the SQ were invited to attend the research centre to complete an interviewer administered questionnaire (MQ) followed by visits to undergo detailed pulmonary function testing, exhaled nitric oxide measurements, peak flow recordings over a one week period, skin prick testing, blood tests (full blood count, carboxyhaemoglobin level, total serum IgE level, alpha-1 antitrypsin level, DNA extraction), urinary cotinine and a CT scan of the chest. The Wellington Ethics Committee approved the study and written informed consent was obtained from each subject.

Questionnaire

All participants completed a detailed written questionnaire compiled from a series of validated questionnaires [10] administered by a trained interviewer in a standardized manner. Data obtained by the questionnaire included demographic information, respiratory history and symptoms, smoking history (including exposure to marijuana and environmental tobacco smoke), allergy, family history, occupation, medication and use of health services. Smoking status was a calculated field based on smoking of tobacco cigarettes. The pack year field was based only on the tobacco cigarette history with one pack year defined as equivalent to 20 cigarettes a day for one year.

Pulmonary function tests (PFTs)

These have been described in detail elsewhere [11,12]. In brief pulmonary function tests were carried out on one site using two Jaegar Master Screen body volume constant plethysmography units with pneumotachograph and diffusion unit (Masterlab 4.5 and 4.6 Erich-Jaegar, Wurzburg, Germany) by trained operators. All measurements were carried out in accordance with ATS and ERS guidelines [13-15]. Measurements of lung volumes and spirometry were repeated 45 minutes after the administration of 400mcg of salbutamol (Ventolin™ GlaxoSmithKline) via a spacer (Space Chamber™). Results were corrected to BTPS and expressed as a percentage of predicted based on local formulae derived using linear regression techniques. The reference sample was taken from within the subject group of this survey and from a concurrent study investigating the pulmonary effects of marijuana smoking. This represented a convenience sample of adults aged between 18 and 70 years recruited through newspaper and radio advertisements and informal contacts. Normal subjects from both studies were defined using ATS guidelines and were required to self identify as “New Zealand European” and be never smokers with no diagnosis of respiratory disease, no recent respiratory symptoms and no use of inhaled medication [12].

Skin prick testing

All subjects received testing to the following nine locally relevant allergens- house dust mite (Dermatophagoides pteronyssinus 30,000 Au/ml), Pine (Lodge Pole Pine, Western Yellow Pine 1:20), Birch 1:20, Grass Mix (10,000 BAU/ml, Aspergillus Fumigatus (1:10), Dog hair (1:10), Feather mix (chicken, duck, goose 1:10), cat pelt (10,000 BAU/ml), cockroach mix), plus a positive control (histamine dihydrochoride) and negative control (saline). The subject’s arm was cleaned with soap and water and then ten numbered points were marked at 2cm intervals on the anterior aspect of the forearm with pen. A drop of each allergen extract was placed alongside the point and a sterile lancet was used to make a prick through the centre of the drop. A new lancet was used for each allergen. The forearm was blotted with tissue paper and the tests were read 15 minutes later. Reactions were assessed by the degree of erythema and the size of the wheal produced. A positive result was defined as a wheal 2mm or more than that of the negative control.

Diagnostic criteria for COPD

In the present analyses we used the definition of COPD proposed by GOLD for our primary outcome: a post-bronchodilator ratio of FEV₁/FVC < 0.7. Severity of disease was also defined according to the GOLD guidelines [3].

In the LLN definition the cut-off value of the post-bronchodilator ratio was set at the 5^th percentile of the normal distribution [5]. The reference equation for the LLN defined by the FEV₁/SVC ratio was: FEV₁/SVC predicted - (1.65*6.48) where FEV₁/SVC predicted = 90.16-(0.271*Age) for males and FEV₁/SVC predicted = 92.55-(0.271*Age) for females and where age is in years at time of pulmonary function tests [12]. The reference equation for LLN defined by the FEV₁/FVC ratio was: FEV₁/FVC predicted - (1.65*5.94) where FEV₁/FVC predicted = 108.1-(0.24*Age)-(10.6*Height) with age in years at time of pulmonary function tests and height is in metres [12]. Post bronchodilator values for SVC were used in the LLN equations.

A doctor’s diagnosis of COPD was based on positive answers to the questions “Did your doctor ever tell you that you had chronic bronchitis?”, “Did a doctor ever tell you that you had emphysema?”, and “have you ever been told by a doctor that you had chronic obstructive respiratory disease?”

Those with bronchiectasis or sarcoidosis and airflow obstruction were not counted as having COPD.

Statistical analysis

Confidence limits for proportions were by an exact method. The Kappa coefficient was used to describe agreement between different methods to define COPD. The prevalence of COPD was adjusted to the age distribution of the Wellington population using the method cited in [16]. Where appropriate t-tests are used to compare continuous variables between groups. Differences between post and pre-bronchodilator FEV₁ are expressed as a percentage change from pre-bronchodilator FEV₁. The FEV₁/FVC ratio is expressed as a percentage. Odds ratios in the univariate and multivariate analysis were calculated by logistic regression. SAS version 9.1 was used for all analyses.

RESULTS

A total of 3,500 people were invited to complete a screening questionnaire (SQ) between 14 April 2003 and 3 June 2004. A total of 2,319 people returned the SQ out of 2,978 with valid contact details, giving a response rate of 77.9%.

Of those completing the SQ, 1,017 (43.9% of 2,319 returning the SQ, 34.2% of the original 2,978 with valid contact details) went on to complete the main questionnaire (MQ). All 1,017 were invited to undergo pulmonary function testing (PFT) of whom 795 attended. A full set of completed pre- and post-bronchodilator flow volume loops were obtained for the 749 people whose results are analysed in this paper (see Figure 1).

Figure 1.

Showing flow of subjects

Table 1 compares respiratory symptoms and selected risk factors for those completing the SQ, the MQ and those who underwent PFT. Generally differences were small but responders to the MQ were more likely to complain of ever having breathing trouble (35% vs 25.3%), were more likely to be ex-smokers (42% vs 36.1%), had a higher prevalence of doctor diagnosed asthma (23.9% vs 19.1%) and had less wheeze in the last 12 months (26.0% vs 22.3%) than those in the SQ group. There were only small differences between the MQ and those with full PFTs.

TABLE 1.

Baseline characteristics of subjects completing the screening questionnaire (SQ), the main questionnaire (MQ) and pulmonary function tests (PFTs) to illustrate possible differences in subject characteristics by various phases of the study programme

	SQ (N, %)	MQ (N, %)	PFTs (N, %)
Male	1097, 47.3	514, 50.5	412, 54.4
Wheezing in the last 12 months	603, 26.0	227, 22.3	165, 21.8
Cough without cold usually	533, 23.0	247, 24.3	181, 23.9
Cough 3 months each year	544, 23.5	188, 18.5	141, 18.6
Phlegm 3 months each year	223, 9.6	125, 12.3	95, 12.5
Breathing trouble ever	587, 25.3	356, 35.0	261, 34.4
Doctor diagnosed chronic bronchitis	198, 8.5	102, 10.0	75, 13.9
Doctor diagnosed emphysema	34, 1.5	9, 0.9	6, 0.8
Doctor diagnosed asthma	443, 19.1	243, 23.9	182, 24.0
Current smoker	279, 12.0	123, 12.1	96, 12.7
Ex-smoker	837, 36.1	427, 42.0	332, 43.8
Response rate ≥ 45 years	1613, 69.6	741, 72.9	557, 73.5
Total N	2319	1017	758

Characteristics of the study population by age, gender, ethnicity, smoking status and pack years are detailed in Table 2. Slightly more men than women participated in the study (54.2% vs 45.8%). The mean age (SD) of the study population was 54.9 years (12.8 years). Just less than half the participants (46.2%) had never smoked tobacco cigarettes. Of the current or ex- smokers, 28.4% had a pack year history for tobacco cigarette smoking of twenty or more years. Nearly half the participants were atopic (47.1%) based on a positive result to one of nine locally relevant allergens and 9.2% had a diagnosis of childhood asthma.

TABLE 2.

Characteristics of the study population by age, gender, ethnicity, smoking status and pack years

	N (%)	Mean (SD)	Median (IQR)
Age (years)   Overall   25-29   30-39   40-49   50-59   65-69   70+	749 15 (2.0) 110 (14.7) 133 (17.8) 199 (26.6) 188 (25.1) 104 (13.9)	54.9 (12.8)	56.5 (44.7-65.7)
Gender   Male   Female	406 (54.2) 343 (45.8)
Ethnicity   NZ European   Maori 1   Other 2   Not stated	654 (87.3) 26 (3.5) 68 (9.1) 1
Smoking status   Never3   Current4   Ex-smoker4   Overall smokers4	346 76 297 373	23.7 (18.6) pack yrs 32.2 (17.2) pack yrs 15.3 (17.9) pack yrs	17.8 (9.6-33.9) 8.0 (1.8-18.5) 10.0 (3.0-21.8)
Pack years4   0-9   10-19   20+	185 (49.6) 82 (22.0) 106 (28.4)
Atopy5	350/743 (47.1)
Prematurity6	22/749 (2.9)
Early hospitalization	11/749 (1.6)
Child asthma7	69/749 (9.2)
TOTAL	749

¹Maori are the main indigenous ethnic minority in New Zealand.

²Mainly Asian and Pacific Islanders

³Refers to non- smokers of tobacco cigarettes. There were 30 smokers of tobacco who did not smoke cigarettes.

⁴Refers to tobacco cigarette smokers.

⁵Based on a positive skin prick test to one of test substances not including the positive control

⁶Based on answer to question “Were you born prematurely (at least one month before the date expected for your birth)”

⁷Based on response to two questions “Did a doctor ever tell you that you had asthma?” and “How old were you when you had your first attack of asthma” (<18 years)

The overall raw and age-adjusted prevalence rates of COPD are presented in Table 3 and Figure 2. The GOLD-defined prevalence for adults ≥ 40 years, adjusted to the age distribution of the Wellington population, was 14.2% (95% CI, 11.0 to 17.0). The LLN-defined age adjusted prevalence for adults aged over 40 years, using slow vital capacity (SVC) was 9.0% (95% CI, 6.7 to 11.3). This was very similar to a LLN definition using forced vital capacity (FVC) of 9.5% (95% CI, 7.1 to 11.8). A LLN definition using FVC but pre-bronchodilator values gave a prevalence of 15.2% (95% CI, 12.0 to 18.5). The estimate for COPD prevalence based on a doctor diagnosis in the same over 40 age group was 10.5 (95% CI, 7.8 to 13.2).

TABLE 3.

Overall raw and age adjusted prevalence and 95% confidence intervals (CI) of chronic obstructive respiratory disease (COPD) by different criteria

Method of diagnosis	N with COPD/N total	Raw prevalence (95% CI)	Age adjusted prevalence1 (95% CI)
GOLD2
All ages	116/749	15.5 (13.0 to18.3)	9.3 (7.2 to 11.4)
≥40 years	109/624	17.5 (14.6 to 20.7)	14.2 (11.0 to 17.0)
LLN FEV1/SVC3
All ages	73/746	9.8 (7.8 to 12.2)	7.0 (5.0 to 9.0)
≥40 years	66/621	10.6 (8.3 to 13.3)	9.0 (6.7 to 11.3)
LLN FEV1/FVC (post-bronchodilator)
All ages	78/749	10.4 (8.3 to 12.8)	7.7 (5.5 to 9.9)
≥40 years	69/624	11.1 (8.7 to 13.8)	9.5 (7.1 to 11.8)
LLN FEV1/FVC (pre-bronchodilator)
All ages	120/749	16.0 (13.5 to 18.9)	14.5 (11.3 to 17.7)
≥40 years	100/624	16.0 (13.2 to 19.1)	15.2 (12.0 to 18.5)
Doctor’s diagnosis4
All ages	79/749	10.6 (8.4 to 13.0)	9.8 (7.1 to 12.5)
≥40 years	67/624	10.7 (8.4 to 13.4)	10.5 (7.8 to 13.2)

¹Adjusted to the age distribution of the Wellington population at the 2001 census [12]

²Diagnostic criteria as per the Global Initiative for Chronic Obstructive Lung Disease (GOLD) guidelines

³Definition of an obstructive pulmonary defect based on FEV₁/VC ratio and cut-off value of this ratio is set at the 5^th percentile of the normal distribution rather than at a fixed value of 0.7 (lower limit of normal, LLN). This ratio can be used with either pre- or post bronchodilator values. Three subjects missing data for slow vital capacity (SVC)

⁴Based on the questions “Did your doctor ever tell you that you had chronic bronchitis?”, “Did a doctor ever tell you that you had emphysema?”, and “have you ever been told by a doctor that you had chronic obstructive respiratory disease?”

Figure 2.

Prevalence of COPD by diagnostic definition and age group. COPD: chronic obstructive pulmonary disease. △: GOLD; □: Lower limit of normal post bronchodilator; ○: Lower limit of normal pre bronchodilator; ●: Doctor’s diagnosis. See text for more complete definitions. Ages 25-29 excluded due to small numbers (N=15).

Table 4 shows the moderate degree of agreement between GOLD and the other definitions. The GOLD criteria classified more subjects as having COPD than the post-bronchodilator LLN for either FEV₁/FVC or FEV₁/SVC. The odds ratio per decade older for a diagnosis of COPD by GOLD compared to the LLN (defined by post-bronchodilator FEV₁/SVC) was 2.5 (CI, 1.8 to 3.6), P< 0.0001. For the pre-bronchodilator LLN for FEV₁/FVC there were a substantial proportion of subjects ‘misclassified’ in both directions. For the post-bronchodilator LLN for FEV₁/FVC fewer subjects were classified as having COPD than by the pre-bronchodilator LLN for FEV₁/FVC.

TABLE 4.

Agreement between various definitions of COPD

		LLN FEV1/FVC Post-Bronchodilator		LLN FEV1/FVC Pre-Bronchodilator		LLN FEV1/SVC Post-Bronchodilator1
		Yes	No	Yes	No	Yes	No
GOLD	Yes	75	41	88	28	70	44
	No	3	630	32	601	3	629
Kappa (95% CI)		0.74 (0.67 to 0.81)		0.70 (0.63 to 0.80)		0.71 (0.64 to 0.79)
LLN FEV1/FVC post-bronchodilator)			Yes	72	6	68	10
			No	48	623	5	663
Kappa (95% CI)				0.69 (0.61 to 0.76)		0.89 (0.83 to 0.94)
LLN FEV1/FVC (pre-bronchodilator)					Yes	67	51
					No	6	622
Kappa (95% CI)						0.66 (0.58 to 0.74)

¹Three subjects missing data for SVC

The prevalence of GOLD defined COPD is presented by age, gender, ethnicity, smoking status and pack years in Table 5. The prevalence was higher in men than women and increased with increasing age. COPD was more frequent in current and ex-smokers and increased with increasing pack years.

TABLE 5.

Prevalence of chronic obstructive respiratory disease (COPD)^† by age, gender, ethnicity, smoking status and pack years with 95% confidence intervals (CI)

	N with COPD/N (%)	95% CI
Age (years)   Overall   25-39   40-49   50-59   60-69   70+	116/749 7/125 (5.6) 6/133 (4.5) 23/199 (11.6) 47/188 (25.0) 33/104 (31.7)	2.2 to 11.2 1.7 to 9.6 7.5 to 16.8 19.0 to 31.8 23.0 to 41.6
Gender   Male   Female	80/406 (19.7) 36/343 (10.5)	16.0 to 23.9 7.5 to 14.2
Ethnicity   NZ European   Maori   Other1	98/654 (15.0) 6/26 (23.1) 12/56 (17.7)	12.3 to 18.0 9.0 to 43.7 9.5 to 28.9
Smoking status2   Non smoker   Current   Ex-smoker	45/376 (12.0) 20/76 (26.3) 51/297 (17.2)	8.9 to 43.7 16.9 to 37.7 13.1 to 22.0
Pack years   0-9   10-19   20+	17/185 (9.2) 16/82 (19.5) 38/106 (35.9)	5.4 to 14.3 11.6 to 29.7 26.8 to 45.7

^†COPD: Defined by GOLD criteria, post-bronchodilator FEV₁/FVC <0.7

¹“Other” includes Asian and Pacific Island people

²A further 30 subjects were tobacco smokers but not of cigarettes

The prevalence of COPD according to GOLD severity stages was 53/116 (45.7%) mild (FEV₁ ≥80% predicted), 52/116 (44.8%) moderate (50% ≤ FEV₁ <80% predicted), 8/116 (6.9%)severe (30% ≤ FEV₁ <50% predicted) and 3/116 (2.6%) very severe (FEV₁ <30% predicted). Thus 105/116 (90.5%) of subjects with GOLD-defined COPD were in the mild or moderate categories.

Amongst the subjects with COPD defined by GOLD, the mean percentage change in FEV₁ (SD) from baseline after bronchodilator was 10.9 (11.3). There were 35/116 (30.2%) subjects with COPD who met the criteria for a positive bronchodilator response [6]. This compares to a mean change (SD) of 3.9% (4.4) in the group without COPD of whom only 23/633 (3.6%) met the criteria for a positive bronchodilator response.

DISCUSSION

This study found that the prevalence of GOLD-defined COPD in adults ≥40 years of age was 14.2%. To date there is no reliable data for COPD from New Zealand (NZ) population surveys based on either pre- or post- bronchodilator lung function criteria with which to compare this figure [17].

Previous prevalence studies in other countries have been summarized in two recent reviews of the literature [18,19]. In the first, 32 studies were identified between 1962 and 2001, only 11 of which were based on spirometry, with just three clearly stating that post-bronchodilator values were measured [18]. In the most recent review and meta-analysis of 37 prevalence studies between 1990 and 2004, a pooled prevalence estimate in adults ≥40 years was 9.0%. Only 9 studies included post-bronchodilator lung function and only 6 were from the Western Pacific region [19]. Since 2004 there have been two general population-based studies worldwide that have applied post-bronchodilator values. The PLATINO study identified crude rates in adults ≥40 years of 7.8% (Mexico City) to 19.7% (Montevideo) [20] and a Scandinavian study reported a prevalence of 7% in those aged 26-82 years [21]. The ongoing Burden of Lung Disease (BOLD) initiative is designed primarily as a COPD prevalence survey among non-institutionalized adults 40 years and older and should facilitate direct comparison of GOLD-defined COPD prevalence rates between countries [22].

The prevalence of COPD was also estimated by reference to the LLN for the FEV₁/VC ratio as proposed by the ATS/ERS taskforce to reduce the number of false positive diagnoses that occurs as FEV₁/FVC ratios fall with age. In the same group of adults ≥ 40 years (using post-bronchodilator values) the estimate was only 9.0% giving a discordant rate of 5.2%. Allocation by GOLD to COPD was more likely for men and older adults than by the LLN definition. Looking at this another way, the LLN for the FEV₁/FVC in our study for a 70 year old was 0.65, and for a 75 year old 0.63.

We are not aware of any studies which specifically consider the difference between a fixed ratio of 0.7 and the LLN using post-bronchodilator values. Margolis et al performed a retrospective review of pulmonary function tests (no post-bronchodilator values) at a Veteran’s Administration hospital comparing a fixed ratio of 0.7 and 95^th percentile based numeric criteria and found discordant readings in 7.2% of the 664 individual tests [23]. Roberts et al found similar discordant rates of 6.9% to 7.5% depending on the reference range used but again did not include post-bronchodilator values [24].

In contrast to our findings, in their study developing reference ranges for post-bronchodilator lung function, Johannessen et al found that the LLN FEV₁/FVC for both men and women after reversibility testing exceeded 0.7 across all ages. This might suggest that the LLN would diagnose more people with COPD but this finding is more likely just a consequence of their small number of observations in elderly men [25]. The findings could also suggest that the GOLD cut off point is useful as long as post bronchodilator values are used. In terms of the clinical impact of the difference between the two definitions, Mannino et al noted that elderly subjects classified as normal using the LLN but abnormal using the fixed ratio died at a similar rate to the cohort classified as abnormal using the LLN criteria. They concluded that the fixed ratio may identify at-risk patients [26]. This analysis was however limited by the unavailability of post-bronchodilator values.

It is not surprising that the use of pre-bronchodilator values for the LLN definition gave a higher COPD prevalence of 15.2%. Our 31% reduction in the prevalence estimate using post-bronchodilator values. is similar to the 27% difference noted by Johannessen et al [21]. The issue of whether one uses FVC or SVC appears to make little difference as the values for these variables were very similar for most subjects although this difference may be greater in a population group with a higher proportion of subjects with more severe COPD.

The range of prevalence estimates depending on definition has been considered by other groups. All reported a wide range depending on definition but none of these included post-bronchodilator testing [27-29]. The importance of using post-bronchodilator lung function is debated and comes down to an attempt to distinguish between asthma and COPD. Both disease complexes share similar symptoms and demonstrate airflow obstruction. However the variability of symptoms and changes in airflow limitation that occur spontaneously or in response to treatment has traditionally been ascribed to asthma. The degree of reversibility in FEV1 which indicates a diagnosis of asthma is generally accepted as ≥ 12% (or ≥ 200ml) from the pre-bronchodilator value [30]. However the 1995 ATS official statement on COPD mentions that a significant increase in FEV1 after an inhaled beta-adrenergic agonist has been observed in up to one third of COPD patients during single testing sessions [31]. The recent ERS/ATS taskforce did not achieve a consensus on the interpretation of bronchodilator responsiveness in subjects with airflow obstruction although comment that values ≥ 12% and 200ml are “significant”. It also states that even though asthmatics tend to show larger responses to bronchodilators, this response has never been shown to be capable of clearly separating the two classes of patients [6]. .

The prevalence estimates in our study were based only on the results of spirometry. We acknowledge that a clinical diagnosis of COPD requires a history of chronic progressive symptoms, possible abnormalities on physical examination and a consideration of risk factors, however objective evidence of airflow obstruction determined by forced expiratory spirometry is the standard for demonstrating and quantifying airflow obstruction [3, 26, 32]. It is well recognized that COPD is under-diagnosed in the community, at least in part because it is silent clinically until the disease process is well advanced. In the third National Health and Nutrition Examination Survey (NHANES III), 44% of those with an FEV₁ of less than 50% of the predicted value did not have a current diagnosis of COPD [33]. In our study only 17/116 (15%) of those who met the GOLD criteria had a doctor’s diagnosis of COPD. 99/116 (85%) met the GOLD criteria but did not report a doctor diagnosis of COPD. Of those with a GOLD-defined COPD and who were in the severe categories (11 subjects), 4 had a doctor’s diagnosis of COPD (36%).

The main strength of this study was the wide range of objective measures made in individuals from a random population survey. The major problem with this study was the large drop-out between those selected from the electoral roll to those with a full complement of tests. However subjects completing the investigative modules were broadly similar to those completing the screening questionnaire. More detailed analysis by age-band (results not shown) indicated a difference between the electoral roll sampling frame and the SQ where response rate increased by age and this was accounted for by adjusting the prevalence rates to the age distribution of the Wellington population. However there was little difference in the proportions by age-band participating in the SQ vs MQ vs those completing PFT. The relatively small sample size and small number of subjects with COPD by any definition lead to the wide confidence intervals for prevalence as illustrated in Table 5. We acknowledge that the ongoing discussion around the topic of a fixed ratio versus a LLN definition is focused on the elderly patient with mild disease. Our sample has few individuals in this group. A limitation of performing such a study in New Zealand is that we are a relatively young population limiting the number of older adults accessible through the electoral roll.

In conclusion this study has provided a first estimate of the GOLD-defined prevalence of COPD in an urban New Zealand population. This prevalence is broadly comparable with the few other studies worldwide that similarly report post-bronchodilator values. We have shown a range of prevalence rates for the same population group from 9.0-15.2% depending on definition. Longitudinal studies using post-bronchodilator spirometry are required to determine whether a LLN approach would better estimate the true COPD prevalence albeit at the cost of simplicity.