Prevalence and Risk Factors for Chronic Obstructive Pulmonary Disease According to Symptoms and Spirometry

Abstract

The burden of COPD is quite high and its prevalence is increasing. Few data are available from Latin America. There is no consensus on what criteria should be the gold standard for the definition of this disease. A population‐based study was carried out in a southern Brazilian city, including adults aged 40–69 years. The aim was to measure the prevalence of COPD according to several criteria. From the 1,046 subjects chosen from a multiple‐stage sampling protocol, a sub‐sample of 234 subjects was systematically selected to undergo spirometry. Percentages of COPD according to different criteria were: chronic bronchitis by questionnaire (7.8%); GOLD stage 0 (7.3%); fixed ratio (15.2%); GOLD stage II (9.9%); ERS (27.7%). The relationship among the three lung functional measurements showed that around 70% of all subjects had negative results with all three criteria, and around 10% were positive according to all three. Utilization of symptom‐based or spirometry definitions of COPD provide different prevalence estimates. Use of different spirometric criteria also resulted in different percentages of COPD. According to spirometry, COPD was higher among men, elderly, those with low education and ex‐smokers. On the other hand, current smokers were more likely to present symptoms of chronic bronchitis. The fixed ratio criterion is recommended when population‐specific reference curves are not available, while the GOLD definition is recommended to evaluate severity of COPD.

• COPD
• Epidemiology
• Prevalence
• Chronic bronchitis
• Lung function
• Respiratory symptoms

Introduction

Until recently, Chronic Obstructive Pulmonary Disease (COPD) was a loosely term used to describe several different diseases, being based primarily on clinical assessment of chronic bronchitis, emphysema or both. In the recent past, COPD became essentially defined in functional terms, based on lung function assessments [[1]], although it has been shown that spirometry use was considerably higher in patients with asthma than in patients with COPD [[2]]. A MEDLINE search using the keywords “COPD,” “prevalence,” and “adults” was carried out for papers published since 1990. Thirteen papers relating COPD prevalence were identified as the most relevant [[3]][[4]][[5]][[6]][[7]][[8]][[9]][[10]][[11]][[12]][[13]][[14]][[15]]. Changes in definition of COPD, however, are yet to be taken up in epidemiological studies. Of papers identified, two used only clinical criteria [[5]][[11]], one reported medical diagnosis and symptoms [[7]], nine used spirometry [[3]][[4]][[6]][[8]][[9]][[10]][[11]][[12]][[14]], one used spirometry and medical diagnosis [[15]], and none used both clinical and spirometric diagnoses. Two problems became evident. First, different studies used different criteria and few studies used more than one criterion. Second, even when the spirometry was used, often the cutoffs were different from one study to another. It is therefore difficult, if not impossible, to compare the reported prevalence levels.

The burden of COPD is quite high and its prevalence in many developed countries is increasing [[16]][[17]][[18]]. As it is written in the summary of the workshop on COPD from the NHLBI [[19]], despite the limitations of the epidemiological data, there are no doubts that COPD is an enormous public health problem. The considerable burden of this disease is imposed not only on patients, but also on healthcare services in general and on the society [[20]]. For developed countries, COPD is expected to be the 4th cause for males and the 3rd for females for 2020 for “Disability‐Adjusted Life Year” in 2020 [[21]]. There are very few prevalence studies from developing countries. For example, a review of the literature shows only one population‐based study in Latin America [[22]], which was based on clinical symptoms, and therefore, the outcome was prevalence of chronic bronchitis and not COPD as it is defined nowadays. The present study was aimed at using several spirometric and clinical criteria for measuring COPD prevalence in a population‐based sample of adults living in a southern Brazilian city.

Materials and Methods

A cross‐sectional population‐based study was carried out in Pelotas, a medium size (320,000 inhabitants) southern Brazilian city. A comparison between Pelotas (city) and the whole country with regard to demographic, socioeconomic and health‐related variables is presented in Table 1.

Table 1.  Comparison Between the City (Pelotas) and the Whole Country (Brazil) According to Demographic, Socioeconomic and Health‐Related Variables. Pelotas, 2001

Variable	Pelotas	Brazil
Sex (including all ages)
Men	48.8%	49.2%
Women	51.2%	50.8%
Mean monthly income (head of the family)	US$ 44.9	US$ 43.0
Gini index (income inequality)	0.60	0.64
Schooling (years)
0–3	27.7%	41.1%
4–10	55.8%	45.1%
≥ 11	16.5%	13.8%
Infant mortality rate	20/1,000	29/1,000

*Sources: Brazilian Institute of Geography and Statistics (sex, monthly income, Gini index, and schooling), and local/national surveys (infant mortality rate).

A multiple‐stage sampling protocol was carried out to select a representative sample of households in Pelotas. All 258 urban census tracts in the city were listed, and 44 were systematically sampled. One block was randomly selected in each tract; and within this block, one street corner was also randomly selected. Thereafter, every fifth household was visited following a clockwise direction. All individuals aged 40 to 69 years living in the sampled households were invited to participate in the study. The design effect associated with the cluster sampling was equal to 1.1 for the three spirometry outcomes. Therefore, there was no need to carry out statistical analyses that take the clustering of the sample into account.

All subjects answered a questionnaire at home, based on the ATS‐DLD version 78 [[23]], with minor local adaptations. Information on demographic, socioeconomic and behavioral variables was also collected. Every fourth subject interviewed in each census tract was referred to a hospital to carry out spirometry. Free transportation was provided in order to increase the response rate.

The home interviews were conducted from January to March of 2001 and the lung function testing lasted up to June; subjects who failed to attend the hospital clinic had their lung function assessed at home (21.6% of the sample). A Jaeger spirometer (Masterscope version 4.3, Germany) was used both in hospital and home measurements; the same physician performed all tests. All patients were examined seated, and wearing a nose clip. One curve was selected per subject—after performing three to five maneuvers with a variation of less than 5% and an expiratory time of at least 6 seconds; the largest sum of FEV1 and FVC was recorded [[24]]. Calibration was done daily with a one‐liter syringe supplied by the manufacturer. Age, sex and height‐adjusted reference values were derived from the American Thoracic Society [[24]]. Brazilian reference values are available [[25]], but the equipment used did not allow these values to be used. Both in the hospital and at home, height was measured to the nearest mm using locally made stadiometers. Five criteria were used to estimate COPD prevalence (Table 2).

Table 2.  Definitions of Chronic Obstructive Pulmonary Disease (COPD) According to Symptoms and/or Spirometry. Pelotas, 2001

Criteria (Abbreviation)	Definition
Symptoms of chronic bronchitis	Presence of chronic symptoms (cough with phlegm on most of the days during three months of the year, for at least two consecutive years) regardless of spirometry results
COPD GOLD Stage 0 (GOLD stage 0)	Presence of chronic symptoms (cough with phlegm on most of the days during three months of the year, for at least two consecutive years) and FEV1/FVC ≥ 70%
COPD GOLD Stage II (GOLD stage II)	FEV1/FVC < 70% and FEV1 50–80% predicted
COPD Fixed Ratio** (FR)	FEV1/FVC < 70%
COPD ERS*** (ERS)	FEV1/FVC < 88% predicted (males) and FEV1/FVC < 89% predicted (females)

*In accordance with the Global Obstructive Lung Disease (GOLD) guidelines [[26]].

**From Ref. [[15]].

***In accordance with the European Respiratory Society (ERS) guidelines.

Initially, descriptive analyses, including calculations of proportions, 95% confidence intervals, means and medians, were carried out. Next, a crude analysis was conducted in order to compare the prevalence of COPD (all criteria) stratified by sex, age, schooling and smoking status. Chi‐squares tests for heterogeneity and trend were calculated. The relationships between smoking status and all criteria were also adjusted for sex, age and schooling using a multivariable regression. However, due to the fact that the results were unchanged, data from this adjusted analysis are not shown.

The study protocol was approved by the Ethical Committee of the Faculty of Medicine of the Federal University of Pelotas. Informed consents were obtained from each subject, and individual data were kept anonymous.

Results

A total of 1,146 subjects were found in the sampled households; of whom 1,046 were interviewed (non‐response: 8.7%). A sub‐sample of 234 subjects was referred for lung function tests; 191 tests (81.6%) were successfully performed. Table 3 shows the distribution of the total sample and of the sub‐sample with lung function tests regarding demographic, socioeconomic, smoking variables and symptoms of chronic bronchitis. There were no significant differences between the total sample and the sub‐sample regarding these background characteristics.

Table 3.  Description of the Total Sample (n = 1,047) and the Sub‐sample with Spirometry (n = 191) in Terms of the Presence of Clinical Symptoms and Demographic, Socioeconomic and Behavioral Variables. Brazil, 2001

Variables	Percentage of sample (95% CI)
	Total sample (n = 1,046)	Spirometry sample (n = 191)
Sex
Females	59.1% (56.1–62.1)	61.8% (54.9–68.7)
Males	40.9% (37.9–43.9)	38.2% (31.3–45.1)
Age (years)
40–49	42.4% (39.4–45.4)	47.6% (40.5–54.7)
50–59	35.4% (33.0–38.3)	34.0% (27.3–40.7)
60–69	21.9 (19.4–24.4)	18.3% (12.8–23.8)
Skin colour
White	84.2% (82.0–86.4)	83.8% (78.6–89.0)
Non‐white	15.8% (13.6–18.0)	16.2% (11.0–21.4)
Family income (minimum wages**)
< 1.01	28.1% (25.4–30.8)	29.3% (22.8–35.6)
1.01–3.00	42.7% (39.7–45.7)	43.5% (36.5–50.5)
3.01–6.00	15.7% (13.5–17.9)	15.2% (10.1–20.3)
6.01–10.00	7.2% (5.6–8.8)	8.4% (4.5–12.3)
>10.00	5.4% (4.0–6.8)	3.7% (1.0–6.4)
Schooling (years of formal education)
0	7.4% (5.8–9.0)	7.9% (4.1–11.7)
1–4	28.5% (25.8–31.2)	24.1% (18.0–30.2)
5–8	31.7% (28.9–34.5)	35.1% (28.3–41.9)
≥ 9	32.4% (29.6–35.2)	33.0% (26.3–39.7)
Smoking status (cig/day)
Non‐smokers	43.4% (40.4–46.4)	42.9% (35.9–49.9)
Ex‐smokers	27.2% (24.5–29.9)	25.1% (18.9–31.2)
Smokers
< 20	14.0% (11.9–16.1)	15.2% (10.1–20.3)
≥ 20	15.4% (13.2–17.6)	16.8% (11.5–22.1)
Symptoms of Chronic Bronchitis
No	92.2% (90.6–93.8)	89.5% (85.23–93.8)
Yes	7.8% (6.2–9.4)	10.5% (6.2–14.8)

*9 missing values.

**The value of one minimum wage in 2001 was equivalent to US$ 80 nowadays.

Most subjects in both samples were women and white. As expected, there were fewer subjects aged 60 to 69 years than at younger ages. While the median family income was equivalent to US$ 253 per month in the total sample, it was US$ 245 in the sub‐sample. The average lengths of schooling in years were 6.8 and 6.9 in the total and sub‐sample, respectively. Smokers made up a third of the total sample; half of them smoked 20 or more cigarettes a day. Symptoms of chronic bronchitis were reported by 7.8% of the individuals of the total sample and by 10.5% of the individuals of the sub‐sample.

The values of FEV1 ranged from 0.4 to 4.9 and FVC from 1.1 to 5.6 l/min. The mean values were 2.5 and 3.2, respectively. There were no differences in lung function tests between those who were tested at the hospital (mean FEV1 = 2.5 l/min; mean FVC = 3.2 l/min; and FEV1/FVC ratio = 77.8%) and those tested at home (2.4, 3.2 and 76.0%, respectively).

Table 4 shows the prevalence of all COPD criteria stratified by sex, age, schooling and smoking status. The highest overall prevalence was obtained with the ERS definition, while the lowest was obtained using the GOLD stage 0 definition. The ERS criterion resulted in a prevalence almost twice as high as that measured by the FR definition, and about three times higher than the GOLD stage II definition. The higher prevalence provided by the FR value relative to the GOLD stage II definition was expected, because the latter is a subset of the former, also requiring FEV1 to be lower than 80% of the predicted value. There were four subjects classified as COPD according to the GOLD stage III criteria and no subjects fulfilled the criteria of GOLD stage IV. Due to these small numbers, they were included as COPD/GOLD stage II.

Table 4.  Prevalence of Chronic Obstructive Pulmonary Disease According to Different Criteria Stratified by Sex, Age, Schooling Level and Smoking Status. Brazil, 2001

Variables	Symptoms of chronic bronchitis	COPD GOLD Stage 0	COPD Fixed ratio	COPD GOLD Stage II	COPD ERS
Sex	p = 0.08	p = 0.35	p = 0.04	p = 0.06	p = 0.77
Men	9.6%	8.2%	21.9%	15.1%	26.0%
Women	6.6%	6.6%	11.0%	6.8%	28.0%
Age (years)	p = 0.01**	p = 0.02**	p = 0.01**	p = 0.04**	p = 0.07**
40–49	5.9%	5.9%	7.7%	4.4%	27.5%
50–59	7.4%	6.4%	21.5%	15.4%	24.6%
60–69	12.3%	11.4%	22.9%	14.3%	31.4%
Schooling (years of formal education)	p < 0.001**	p < 0.001**	p = 0.05**	p = 0.01**	p = 0.07**
0	16.9%	15.6%	26.7%	20.0%	33.3%
1–4	11.4%	10.7%	21.7%	15.2%	37.0%
5–8	5.7%	5.1%	11.9%	10.4%	25.4%
≥ 9	4.7%	4.4%	11.1%	3.2%	20.6%
Smoking status	p < 0.001	p < 0.001	p = 0.009	p < 0.001	p < 0.001
Non‐smokers	3.1%	3.1%	6.1%	1.2%	11.0%
Ex‐smokers	7.4%	6.0%	31.3%	25.0%	47.9%
Current smokers	15.3%	14.7%	14.8%	9.8%	32.8%
Overall	7.8%	7.3%	15.2%	9.9%	27.2%

*Chi‐square test for heterogeneity.

**Chi‐square test for trend.

Men were more likely to present COPD according to the FR (p = 0.04) and GOLD stage II (p = 0.06) definitions, as well as being more likely to have symptoms of chronic bronchitis (p = 0.08). No gender differences were found for COPD according to GOLD stage 0 and ERS definitions. While age was positively related to COPD according to all definitions, schooling was negatively associated with all outcomes. These associations were of borderline significance (p = 0.07) for the ERS definition. Current smokers were more likely to present symptoms of chronic bronchitis and COPD GOLD stage 0. On the other hand, ex‐smokers presented the highest frequencies of COPD according to all spirometric criteria. Smoking results were unchanged when adjusted for sex, age and schooling.

Figure 1 shows the relationship between the three spirometric criteria. For 138 subjects (72.3%), lung function was considered normal by all three criteria and for 19 individuals (9.9%), all three tests were positive. For 34 (17.8%) subjects, there was disagreement; 24 subjects (12.6%) were diagnosed by the ERS, but not by either of the two other criteria, and 10 (5.2%) were positive according to the FR and ERS cutoffs, but not according to GOLD stage II. Neither FR nor the GOLD stage II criteria, by themselves or in combination, detected any subjects that ERS had failed to pick up.

Figure 1. Relationship between the spirometric results (Venn's diagram). Pelotas, 2001. FR: Fixed Ratio Criteria, ERS: European Respiratory Society Criteria, GOLD II: Global Obstructive Lung Disease Criteria (Stage II). [From Ref. [[26]].]

The prevalence of clinical symptoms was similar between subjects for whom all tests were negative and for those with a positive ERS criteria only. Three of the 10 subjects with positive FR and GOLD stage II results had clinical symptoms, as did 36.8% of those with three positive tests. Due to small numbers, these results must be interpreted with caution.

Discussion

There was wide variability in COPD prevalence estimates according to different diagnostic criteria. Other authors have also reported similar findings [[4]][[15]][[18]] pointing out how much this can affect international prevalence comparisons.

COPD prevalence levels according to FR, GOLD II and ERS found in the present paper (15.2%, 9.9%, 27.2%) were similar to those found by Celli et al. [[15]] using the NHANES population (18.4%, 8.6%, 17.5%, respectively). In an Italian study including subjects aged 46 to 73 years [[9]], the prevalences of COPD according to the ERS and FR definitions were 28.8% and 12.2%, respectively.

Previous reports did not provide an analysis of the agreement between the three spirometric criteria. About 70% of all subjects had negative results with the three criteria, and 9.9% were positive according to all three. All subjects with obstruction according to GOLD stage II were also positive according to the other two criteria, and all subjects with a positive FR diagnosis were positive according to GOLD II. Thus, GOLD II was the most restrictive criterion, FR came in second place and ERS resulted in the highest number of subjects being diagnosed.

Some methodological aspects of the present study should be taken into account. About 25% of the sampled subjects were invited to undergo spirometric testing, due to budgetary limitations. Nevertheless, this sub‐sample was not different from the total sample (Table 3) in terms of socio‐demographic, smoking variables, nor symptoms of chronic bronchitis. The latter finding is highly relevant, because one might expect symptomatic subjects to be more likely to accept the invitation for lung function tests.

It could be argued that spirometry carried out at home—for 22% of the sample—could have differed from tests performed at the hospital. Special efforts were made to avoid such bias by using the same equipment and the same physician in all tests. The fact that the mean values of spirometry for both groups were statistically similar is very reassuring.

We are aware that subjects diagnosed with obstructive disease may include some asthmatic individuals. The analyses were restricted to subjects above 40 years in order to reduce this likelihood, but even so it is not possible to rule it out.

There were five non‐smokers classified as COPD according to the fixed ratio criteria; two of them reported a medical diagnosis of asthma; four of them were women; their ages were: 40, 44, 50, 54 and 57 years old. Therefore, it is possible that some subjects classified as COPD are, in fact, asthmatics.

Correlates of COPD were highly consistent with a recent study [[27]] and were similar between symptom‐based and spirometry definitions of COPD. A major difference was found in terms of smoking; while ex‐smokers presented the highest risk for spirometry‐based COPD, current smokers showed the highest risk for symptom‐based COPD.

Extrapolation of our results to other populations has to be carefully considered. Although Brazil is a large country and some regions may have different lung function or risk factors patterns, in our view, Pelotas is arguably representative of relatively developed areas of Latin America, undergoing rapid epidemiological transition. Furthermore, it represents a scenario likely to be faced by most parts of our region within the next decades. Probably, our prevalence estimates of COPD are not valid for the whole country, but there is no reason to believe that: 1) the relationship between clinical and spirometric results would be different in other populations; 2) the risk factors for COPD would be different in other Brazilian areas.

It is important to reinforce the need for standardization for diagnosing COPD in order to allow comparison among the studies. The purpose of this study was not to identify the best functional criteria for the diagnosis of this disease—as this would not be possible without a true gold standard—but to compare the performance of spirometric and clinical diagnostic criteria in a population based sample. Agreement should be reached on a standard global criterion, and for that purpose it should not depend on national reference values. At the moment, the only criterion that does not require reference values is the FR and for epidemiological studies it might be the most appropriate. On the other hand, utilization of the GOLD definition provides a unique opportunity to evaluate severity of COPD.