This was a multicenter population-based cross-sectional study. Participants were randomly selected from 6 urban areas in Argentina, using cluster sampling, in order to establish the prevalence of COPD and evaluate the clinical and sociodemographic characteristics of patients, treatment, and risk factors.

The study was conducted between August 2014 and May 2016. The following urban clusters were selected: La Plata, Rosario, Autonomous City of Buenos Aires, Northern Region of Gran Buenos Aires, Córdoba and Mendoza. The sample size was calculated on the basis of an 8% prevalence, as determined by the PLATINO study,9 with a 95% confidence interval, a design effect of 1.5, and a total accuracy (6 urban areas) of ±1.15%, for which 3207 patients with completed questionnaires and postbronchodilator spirometries needed to be recruited. Taking into account an estimated response rate of 75%, the number of households to be contacted was 4276.

Men and women ≥40 years of age were included from the 6 selected urban clusters. The sample was selected by multistage probability cluster sampling based on cartographic area units, described in greater detail in the supplementary material. All selected individuals (1 single subject per household) were invited to participate in the study, and if they agreed, they were asked to sign an informed consent form. The following exclusion criteria were defined and applied: mental disorders or inability to make decisions, history of thoracic or abdominal surgery in the last 3 months, residence in institutions, current COPD exacerbation, and others described elsewhere9 (see supplementary material). Contraindications for performing spirometry were also taken into account.13,14

Data collected included an evaluation of bronchial obstruction by pre- and postbronchodilator spirometry as a primary study objective. COPD was defined as a postbronchodilator FEV1/FVC ratio<0.7, and the GOLD 2017 classification was used to define the degree of obstruction and multidimensional ABCD assessment.15 The analysis was also performed using the lower limit of normal (LLN) method.16 All subjects were asked about any previous medical diagnosis of emphysema, chronic bronchitis or COPD.

A study team consisting of an interviewer and a spirometry technician performed the procedures in the home of each subject. The structured interview, CAT questionnaire,17 and mMRC18 were first completed by all subjects (see supplementary material). Blood pressure and anthropometric measurements were obtained using a previously validated portable ultrasonic measuring board and digital weighing scale (SECA 804, Hamburg, Germany) (see supplementary material).

Spirometry was then performed before and after inhalation of 400mcg salbutamol (Ventolin® HFA, GSK), using a measured dose device with spacer. As in similar studies,9 spirometers equipped with an ultrasonic sensor (EasyOne NDD; Zurich, Switzerland) were used according to the standard technique.14 Devices were calibrated every day with a 3 L syringe. Studies were analyzed according to the 3 criteria previously described in other publications.14,19,20 The acceptability of the spirometries was determined according to Enright's A, B, and C criteria,19 with the NHANES III reference equation.21 During the field work period, spirometries were sent electronically to the Central Spirometry Committee once a week, where they were reviewed by 2 blinded evaluators who analyzed each spirometry and determined the number of acceptable maneuvers obtained according to ATS/ERS criteria.14

The performance of the technicians was evaluated every week according to the acceptability criteria of the spirometries they had conducted, and retraining was offered, if necessary (see supplementary material).

The study and written informed consent form were initially reviewed and approved by the Provincial Bioethics Committee (#344/14), Rosario City, Argentina. The ethical, regulatory, teaching and research aspects of the study were then reviewed and approved by regional and local authorities (ethics committees from each urban area).

The estimated prevalence of each variable was presented with the absolute value, percentage and respective 95% confidence interval (CI) for the overall study population and for each subgroup of interest.

A bivariate analysis was performed to establish the possible factors associated with COPD. Odds ratio (OR) with CI was used as a measure of association. A multivariate analysis was then performed using logistic regression, and variables that were considered relevant were included in the adjustment of the model. The logistic regression analysis model was performed using the stepwise method, and the adjusted ORs obtained with the complete model were analyzed. The ORs of the adjusted model were also calculated with their CI, and for these, the level of significance of the differences between odds was analyzed using the Chi-square method, with a level of significance set at 5%.

COPD prevalence was 14.5% (CI: 13.4–15.7), 18.4% (CI: 16.4–20.4) in men and 11.7% in women (CI: 10.3–13.1), and increased in each decade of life analyzed, from 3.2% (CI: 2.0–4.4) in subjects younger than 50 years of age to 30.4% (CI: 23.3–37.5) in patients 80 years of age or older (Table 3 and Fig. 3). Prevalence was higher in subjects with lower educational levels: 16.7% in subjects with primary education only (CI: 14.5%–18.8%); and lower socio-economic status: 17.8% in the lowest category (CI: 15.8%–19.9%). Overall COPD prevalence and differences among the urban clusters did not vary significantly after direct adjustment for sex, age, and education (Appendix B, Fig. A2 of the supplementary material).

Fig. 3.

COPD prevalence. Distribution by age and sex.

(0.06MB).

When the LLN criterion was applied, prevalence was 9.4%, n: 325 (CI: 8.4%–10.3%) (Table 4 and Appendix B, Tables A3 and A4 of the supplementary material), and a good correlation for spirometric diagnosis of obstruction was observed between both methods (k=0.735; CI: 0.700–0.770; P<.05). However, 187 cases (37.1%) classified as obstructive according to the GOLD criteria were considered normal according to the LLN method (Appendix B, Table A4 of the supplementary material), and when both spirometric criteria were compared, the prevalence of COPD in the older subgroups was lower with the LLN method (Table 4).

Severity of COPD according to spirometric obstruction (GOLD 2017) was: (1) mild 38% (CI: 34–43); (2) moderate 52% (CI: 47–56), (3) severe 10% (CI: 7–13); and (4) very severe 1% (CI: 0–2). The GOLD 2017 ABCD multidimensional assessment15,22 distribution was as follows: (A) 52% (CI: 47–56), (B) 43% (CI: 39–48), (C) 1% (CI: 0–2), and (D) 4% (CI: 2–6) (Fig. 4).

Fig. 4.

Distribution of COPD cases according to GOLD 2017 multi-dimensional assessment. Prevalence of each ABCD subgroup with grades 1, 2, 3, and 4 for all COPD patients (n: 471). Patients in subgroup A (grade 1: 24%; grade 2: 25%; grade 3: 2%; grade 4: 0%); B (grade 1: 14%; grade 2: 24%; grade 3: 6%; grade 4: 0%); C (Total 1%*: grade 1: 0%; grade 2: 1%; grade 3: 0%; grade 4: 0%) and D (total 4%*: grade 1: 0%; grade 2: 2%; grade 3: 1%; grade 4: 0%).

(0.04MB).

Only 22.6% (114/504) of the subjects with obstructive spirometry results had a previous diagnosis of COPD, chronic bronchitis or emphysema made by a doctor. This is equivalent to an underdiagnosis rate of 77.4% (CI: 73.7–81.1), with no differences between men (77%) and women (78%). Of the 504 patients with a spirometric diagnosis of COPD, only 190 (38.1%) had performed at least 1 previous spirometry.

In contrast, 176 of the 290 subjects who reported that a doctor had given them a diagnosis of pulmonary emphysema, chronic bronchitis, or COPD, did not show an obstructive spirometric pattern. This is equivalent to a diagnostic error rate of 60.7% (CI: 55.1–66.3), lower in men (49%) than in women (69%), P<.001.

Only 27.6% (505/1828) of the subjects who reported symptoms of COPD (dyspnea, cough, expectoration, and wheezing) had performed at least 1 previous spirometry.

COPD prevalence varied when only the definition based on spirometric obstruction was taken into consideration (14.5%; CI: 13.4–15.7). Prevalence reduced progressively (7.2%; CI: 6.3–8.1), as exposure (smoking, occupational exposure, and indoor pollution) and symptoms (mMRC≥2 or CAT≥10) were added to the model (Table 4).

Smoking was found to be highly prevalent; 35% of all respondents were active smokers (33.5% women; 37% men), while 35.3% were former smokers (30% women; 42.6% men) and 29.7% were never-smokers (36.5% women; 20.4% men).

COPD prevalence in the study population was 16.9% among the 2439 active or former smokers, and 8.9% among the 1030 never smokers.

Among the 504 subjects who met spirometry criteria for COPD, 412 (82%) were active or former smokers and 92 (18%) had no history of an active or former smoking habit (Table 5).

The association between smoking and COPD was significant (OR: 1.95; CI: 1.49–2.54), both in the crude analysis and after adjustment using logistic regression (Table 5). This association was maintained when smokers and former smokers were compared by sex, with a closer association observed for men. In total, 248 men (21.3%) and 164 women (12.9%) with COPD and a history of smoking were identified (Fig. 5). The corresponding OR for the association between smoking and COPD was 3.95 for men, and 1.42 for women.

Fig. 5.

COPD prevalence by smoking history (active or former smoker) or no history of smoking (never-smoker). Distribution by sex.

(0.05MB).

Exposure to tobacco was similar in the 6 urban clusters (Appendix B, Table A5 of the supplementary material).

Table 5 shows the association between numbers of COPD cases with selected characteristics and spirometry-confirmed COPD, along with the OR for the crude association and the association adjusted by binary logistic regression. Sex, age, smoking (current or previous), socioeconomic status, history of tuberculosis, and a family history of asthma showed a statistically significant association with the presence of COPD. The associations with COPD persisted after adjustment by logistic regression, taking into account the concomitant presence of all the other variables. Diabetes and obesity showed an inverse association: fewer COPD patients presented concomitant diabetes and obesity (Appendix B, Tables 5 and A6 of the supplementary material).