Chronic Airflow Obstruction in a Black African Population: Results of BOLD Study, Ile-Ife, Nigeria

Abstract

Global estimates suggest that Chronic Obstructive Pulmonary Disease (COPD) is emerging as a leading cause of death in developing countries but there are few spirometry-based general population data on its prevalence and risk factors in sub-Saharan Africa. We used the Burden of Obstructive Lung Disease (BOLD) protocol to select a representative sample of adults aged 40 years and above in Ile-Ife, Nigeria. All the participants underwent spirometry and provided information on smoking history, biomass and occupational exposures as well as diagnosed respiratory diseases and symptoms. Chronic Airflow Obstruction (CAO) was defined as the ratio of post-bronchodilator (BD) one second Forced Expiratory Volume (FEV₁) to Forced Vital Capacity (FVC) below the lower limit of normal (LLN) of the population distribution for FEV₁/FVC.

The overall prevalence of obstruction (post-BD FEV₁/FVC < LLN) was 7.7% (2.7% above LLN) using Global Lung Function Initiative (GLI) equations. It was associated with few respiratory symptoms; 0.3% reported a previous doctor-diagnosed chronic bronchitis, emphysema or COPD. Independent predictors included a lack of education (OR 2·5, 95% CI: 1.0, 6.4) and a diagnosis of either TB (OR 23.4, 95% CI: 2.0, 278.6) or asthma (OR 35.4, 95%CI: 4.9, 255.8). There was no association with the use of firewood or coal for cooking or heating. The vast majority of this population (89%) are never smokers. We conclude that the prevalence of CAO is low in Ile-Ife, Nigeria and unrelated to biomass exposure. The key independent predictors are poor education, and previous diagnosis of tuberculosis or asthma.

Keywords:

• CAO
• COPD
• Nigeria
• prevalence
• spirometry

Introduction

Chronic obstructive pulmonary disease (COPD) is emerging as one of the leading causes of death in developing countries. However, due to lack of survey data, the estimates of prevalence of airflow obstruction for Africa are based mainly on expert opinion (1,2).

Some studies have examined the prevalence of COPD in various African cities and countries; most of these data are substantially limited in external validity because they were either restricted to occupational groups or samples of patients from outpatient clinics (3–6). Other prevalence studies have attempted to estimate the burden of COPD from self-reported chronic symptoms of cough and sputum as indicators for chronic bronchitis (7–9). However COPD prevalence varies with the criteria used to define it, and self-reported respiratory symptoms substantially underestimate the presence of chronic airflow obstruction (CAO) (9–11).

The Burden of Obstructive Lung Disease (BOLD) initiative is an international collaboration aimed at investigating the prevalence of COPD and associated risk factors in countries around the world, using standardized methods for collection of COPD related data and post-bronchodilator (post-BD) spirometry in representative general population samples (12). In Africa, only one population-based survey in a suburb of Cape Town using the BOLD protocol has been reported (13,14). Since Africa is a large continent with wide disparities in the distribution of potential risk factors, filling the gaps in knowledge relating to COPD prevalence is a priority for informing public health policy.

Our study sought to estimate the prevalence and risk factors of CAO in a suburban community in Nigeria.

Methods

The study was conducted using the BOLD protocol, previously described in detail (12) and approved by the ethics committee of Obafemi Awolowo University ­Hospital.

The study population

Ile-Ife, the site of the survey, is an ancient city in southwest Nigeria. It comprises two local government districts (Ife East and Ife Central) with an estimated population of 250,000 (15). In order to achieve our goal of recruiting at least 760 participants with acceptable spirometry, we used a three stage random sampling method to obtain an initial sample of 1120 adults aged at least 40 years, accounting for a probable 20% non-response and 15% non-usable spirometry rates. Institutionalized individuals were excluded, as were those precluded from lung function testing for safety reasons.

The sampling units were enumeration areas (EA), households and individuals in the first, second and third sampling stage respectively. They were selected randomly and proportionately with replacement from each of the two strata; Ife Central and Ife East. In total, 76 EAs were randomly selected from a combined total of 1658 EAs within the two strata. The ratio of adults aged 40 and above per household in the EAs was low (0.6), so in order to achieve our sample, all households and eligible individuals in the sampled EAs were included (100% sampling in 2^nd and 3^rd stage).

Spirometry

Spirometry was performed according to the ERS/ATS recommendations by trained and certified technicians using the EasyOne spirometer (ndd Medizintechnik; Zurich, Switzerland), an ATS-certified, portable, battery operated and ultrasound-based spirometer (12). Participants were tested in a seated position and separate measurements were made before and at least 15 minutes after two puffs of Salbutamol (200 microgram) had been administered via metered-dose inhaler with a valve spacer (Volumatic; GlaxoSmithKline; Research Triangle Park, NC). The spirometers were checked for calibration daily with a 3.00-L syringe. Spirometry data were sent periodically via a secure internet transfer to the central quality control centre in London, UK, where each spirogram tracing was carefully reviewed and graded according to the ATS guideline for acceptability (at least three trials and two acceptable- that is, free of zero flow errors, artefacts, early termination without a plateau, extra breaths, cough in first second and back-extrapolated volumes of >5.0% of FVC) and repeatability tests for both FEV₁ and FVC (16). We accepted a difference between largest and second largest values providing they were <200 ml., a slight relaxation of the ATS/ERS criteria. Only spirometry tests meeting these criteria were included in the analysis. Study technicians were monitored regularly to ensure the quality of the spirograms met quality control standards. When a technician's quality score dropped below a preset level, he/she was temporarily withdrawn from the field, retrained and recertified.

Questionnaire data

We used the BOLD study questionnaires to obtain information about respiratory symptoms, health status, exposure to potential risk factors including use of biomass fuels like wood, coal or crop residue, cigarette smoking, occupational exposures, activity limitation, co-morbidities, medication use, health care utilization and other respiratory diagnosis. The questionnaires were translated into Yoruba, the local language, and administered face-to-face by trained and certified field staff. Standard methods for translation using forward and backward translation, reconciliation and piloting are part of the BOLD standardized methods. For participants who did not complete the full set of questionnaires, minimal data on smoking history and co-morbidities were obtained.

Data analysis

We performed all analyses using Stata 13 (Stata Corp., College Station, TX, USA). Data describing population demographics included all responders. For estimation of prevalence of CAO in Ife, only responders who had usable spirometry were included. We defined the response rate as the proportion of the eligible sample that completed post-bronchodilator spirometry (regardless of quality control scores) and the core questionnaire, while the cooperation rate was defined as the proportion of those who provided full data and spirometry when successfully contacted. We defined pack-years as the number of cigarettes smoked per day divided by 20 and multiplied by the number of years that the participant smoked. CAO was defined using post-BD FEV₁/FVC less than the lower limit of normal (LLN), which is the fifth percentile in a standard distribution of a non-smoking population without overt respiratory disease.

We derived the LLN using the Global Lung Function Initiative (GLI) equations for African Americans. We excluded respondents aged above the reference ­equation's upper age limit for men (85 years) and women (87 years). The GLI equation was used in the primary analysis (17). To be consistent with BOLD standard, we conducted a secondary analysis using equations derived from participants in the Third US National Health and Nutrition Examination Survey (NHANES III) (18) and from local participants. Local equations were developed from non-smoking asymptomatic respondents in the present survey. To identify the risk factors of chronic airflow obstruction, we developed univariate and multivariate models adjusting for possible confounders. We took into account the sampling design using the ‘svy’ set of commands in the Stata statistical program to correct for the effect of clustering and we derived the sample weights by estimating the probability of selection. These weights were used to generate the population prevalence estimates.

Results

Out of 1704 potential participants selected for recruitment, 1545 were confirmed resident in the area, 1169 participated and 1148 provided complete data (Figure 1). Of these, 883 individuals performed acceptable post-bronchodilator spirometry that met the quality control criteria. Thereafter we excluded 8 respondents aged above the age range for GLI African Americans reference equation. In the end, 875 respondents were included in the primary analysis. Overall, the response rate was 76% (1169/1545) and the cooperation rate among those reached was 97% (1169/1200).

Figure 1.  Schematic representation of the sampling process.

Table 1 is a summary of the comparison of responders and non-responders. Compared with non-responders, those who responded were more likely to be women (p < 0.001), never smokers (p = 0.006) and reported ­having less doctor-diagnosed chronic respiratory disease (p = 0.01). Of the responders, 89% had never smoked and only 2–3% were current smokers.

Table 1.  Comparison of responders and non-responders

		Responders^1	Non-responders^2
Age*	40–49	406 (35%)	150 (40%)
	50–59	282 (25%)	108 (29%)
	60–69	257 (22%)	72 (19%)
	70+	203 (18%)	46 (12%)
Gender*	Male	424 (37%)	185 (49%)
	Female	724 (63%)	191 (51%)
Smoking status	Current	26 (2.3%)	2 (15.4%)
	Ex	97 (8.4%)	0
	Never	1025 (89.3%)	11 (84.6%)
Doctor-diagnosed asthma, emphysema, CB or COPD	Yes	10 (1%)	1 (8%)
	No	1138 (99%)	12 (92%)
Other co-morbid conditions	Yes	42 (4%)	1 (8%)
	No	1106 (96%)	12 (92%)

1Responders are those who completed post-bronchodilator spirometry (regardless of Quality Control scores) and the core questionnaire.

2Non-responders are eligible individuals who are missing the core questionnaire and/or post-bronchodilator spirometry, but for whom the tabulated variable is known.

*p < 0.05. CB: Chronic Bronchitis, COPD: Chronic Obstructive Pulmonary Disease.

Table 2 describes the clinical profile of the respondents who reported the presence of respiratory symptoms or diagnosis. Few respondents had clinical symptoms: 9.2%, 6.7% and 2.2% for cough, phlegm or wheeze but among these, the estimated proportion with chronic airflow obstruction were: 10.7%, 11.5% and 22.7%, respectively. Fewer respondents reported chronic cough or chronic phlegm for at least 3 months in 2 consecutive years: 0.5% and 0.3%, respectively, and none had airflow obstruction. Table 3 shows the prevalence of obstruction (FEV₁/FVC <LLN) by age group, years of schooling, body mass index, smoking status and other exposures. Overall, 7.7% of the respondents had CAO, ranging from 5.2% in those aged 40–49 years to 10.4% in those aged 70 years and above. The rate is slightly lower (6.9%) if Caucasian norms are used but much lower (3.5%) with local equations.

Table 2.  Clinical profile of respondents with chronic airflow obstruction (FEV₁/FVC < LLN)

Variable	All Respondents		Post-Bronchodilator FEV1/FVC < LLN (GLI)
	N	%	%
Cough	81	9.7	10.7
Phlegm	59	7.5	11.5
Wheeze	19	1.9	22.7
Chronic cough^**	4	0.5	–
Chronic phlegm^**	3	0.3	–
Self-reported chronic bronchitis	1	0.1	–
Doctor-diagnosed chronic bronchitis	3	0.3	19.7
Doctor-diagnosed respiratory disease	8	0.9	61.1

Respondents answered to the questions: ‘Do you usually cough when you don't have a cold’, ‘Do you usually bring up phlegm from your chest’ ‘Have you had wheezing or whistling in your chest at any time in the last 12 months.

Estimates include adjustment for survey design and clustering effect.

Global Lung Function Initiative, LLN - Lower Limit of Normal, FEV₁ –Forced Expiratory Volume in 1 second, FVC- Forced Vital Capacity. Post-bronchodilator FEV₁/FVC < LLN was defined as chronic airflow obstruction.

**Chronic cough and chronic phlegm was defined by the presence of symptom for most days in a 3-month period in 2 consecutive years.

Table 3.  Estimated population prevalence of chronic airflow obstruction (FEV₁/FVC <LLN)

				Post-Bronchodilator FEV1/FVC <LLN
		All Respondents		GLI Equation (African Americans)		NHANES Equation (White)		Local Equation
Variable	Group	N	%	%	95% CI	%	(95% CI)	%	95% CI
Age group	40–49	325	37.1	5.2	3.2–8.1	5.2	3.2–8.1	3.4	1.9–6.2
	50–59	231	26.4	9.5	5.8–15.1	7.8	4.6–12.8	2.9	1.3–6.1
	60–69	213	24.4	10.0	6.0–16.1	8.6	5.0–14.2	4.3	2.2 –8.3
	70 and above	106	12.1	10.4	5.3–19.3	10.1	5.9–16.8	4.5	1.7–11.1
Sex	Male	341	39.0	8.6	6.1–12.0	7.1	4.8–10.3	3.5	2.1–5.9
	Female	534	61.0	6.7	4.9–9.1	6.7	4.9–9.1	3.5	2.2–5.7
Years of schooling	13 years+	191	21.8	5.5	2.7–10.6	5.0	2.5–9.6	3.4	1.4–7.8
	0–12 years	527	60.1	7.5	5.3–10.5	6.2	4.1–9.2	3.1	1.9–5.0
	None	164	18.1	13.1	8.4–19.9	14.2	9.1–21.6	5.6	3.1–10.0
BMI (Kg/m^2)	18.5–23.9	342	39.5	8.5	5.8–12.3	7.6	5.1–11.2	4.2	2.4–7.2
	Under 18.5	44	5.1	13.4	5.6–29.0	8.7	3.4–20.8	5.7	1.7–18.0
	24–29.9	315	36.4	6.5	4.0–10.5	6.2	3.7–10.2	2.8	1.4–5.3
	30 and above	164	19.0	7.1	3.6–13.7	6.6	3.2–13.1	2.9	0.9–8.6
Smoking status	Never	784	89.6	7.4	5.6–9.7	6.5	4.9–8.6	3.3	2.2 –4.9
	Ex	68	7.8	8.4	3.6–18.3	7.3	3.0–16.7	2.7	0.8–8.1
	Current	23	2.6	13.2	3.4–39.1	13.2	3.4–39.1	10.0	2.7–30.5
Ever diagnosed with TB		4	0.5	62	16.9–92.9	62	16.9–92.9	62.0	16.9–92.9
Ever diagnosed with Asthma		6	0.7	68.3	26.2–92.9	68.3	26.2–92.9	26.4	5.6–68.2
Firewood or coal for cooking/heating ^‡		594	67.9	8.3	6.3–10.8	7.4	5.5–9.8	3.9	2.6–5.8
Ever engaged in farming*		386	44.2	7.7	5.3–11.2	6.2	4.1–9.3	3.6	2.2–5.8
Ever worked in dusty job^‡‡		309	35.3	9.6	6.3–14.3	8.4	5.2–13.2	4.2	2.5–7.0
Overall				7.7	6.0–9.8	6.9	5.4–8.8	3.5	2.5–5.0

NB: These estimates are based on responders with acceptable spirometry.

GLI – Global Lung function Initiative, NHANES – National Health And Nutrition Examination Survey, BMI -Body mass index, OR - Odds Ratio, CI- Confidence Interval, TB - Tuberculosis, LLN - Lower Limit of Normal, FEV₁ –Forced Expiratory Volume in 1 second, FVC- Forced Vital Capacity. We defined chronic airflow obstruction by post-bronchodilator FEV₁/FVC <LLN.

*Ever worked as a farmer for at least 3 months.

‡Used indoor open fire with wood, coal, charcoal, crop residues or dung as primary means of cooking or heating home for more than 6 months.

‡‡Ever worked in a dusty job for a year or more.

**Chronic cough and chronic phlegm was defined by the presence of symptom for most days in a 3-month period in 2 consecutive years.

The prevalence of chronic airflow obstruction is higher in older people and in current smokers. Though 67.9% of all the respondents had used either firewood or coal for cooking or heating for at least 6 months, only 8.3% (95% CI: 6.3, 10.8) of them had evidence of chronic airflow obstruction. We found high prevalence of obstruction in those with no schooling; 13.1% (95% CI: 8.4, 19.9), underweight; 13.4% (95% CI: 5.6, 29.0), current smokers; 13.2% (95% CI: 3.4, 39.1), and previously diagnosed TB; 62.0% (95% CI: 16.9, 92.9) or asthma; 68.3% (95% CI: 26.2, 92.9). After adjusting for co-variates such as age, sex, body mass index, occupation and exposures to indoor pollutants, the key independent predictors of CAO were a lack of education; OR: 2.5 (95% CI: 1.0, 6.4) and previous diagnosis of TB; OR: 23.4 (95% CI: 2.0, 278.6) or asthma; OR: 35.4 (95% CI: 4.9, 255.8), albeit with wide confidence intervals because of the low numbers. There was no significant association; OR: 1.2 (95% CI: 0.5, 2.7) between CAO and use of firewood or coal for domestic cooking or heating (Table 4).

Table 4.  Multivariate logistic regression of chronic airflow obstruction (FEV₁/FVC <LLN)

		Post-Bronchodilator FEV1/FVC <LLN (GLI)
		Unadjusted			Adjusted
Variable	Group	OR	95% CI		OR	95% CI		p value
Age group	40–49 (ref)	–	–	–	–	–	–	–
	50–59	1.9	0.9	4.0	1.8	0.9	4.0	0.137
	60–69	2.0	0.9	4.4	1.9	0.9	4.0	0.078
	70 and above	2.3	0.8	5.3	1.8	0.9	5.7	0.305
Sex	Female	0.8	0.5	1.2	0.7	0.4	1.4	0.367
Years of schooling	13 years+ (ref)	–	–	–	–	–	–	–
	0–12 years	1.4	0.6	3.3	1.2	0.5	2.8	0.633
	None	2.7	1.1	6.7	2.5	1.0	6.4	0.059
BMI (Kg/m^2)	18.5–23.9 (ref)	–	–	–	–	–	–	–
	Under 18.5	1.7	0.6	4.8	1.6	0.5	5.3	0.407
	24–29.9	0.7	0.4	1.5	0.8	0.4	1.7	0.519
	30 and above	0.8	0.3	2.0	0.8	0.4	1.8	0.588
Smoking status	Never (ref)	–	–	–	–	–	–	–
	Ex	1.2	0.4	3.0	0.8	0.2	2.9	0.776
	Current	1.9	0.4	8.7	2.4	0.4	12.5	0.308
Ever diagnosed with TB		20.3	2.5	168.1	23.4	2.0	278.6	0.013
Ever diagnosed with Asthma		27.6	4.5	170	35.4	4.9	255.8	0.001
Firewood or coal for cooking or heating‡		1.3	0.7	2.5	1.2	0.5	2.7	0.658
Ever engaged in farming*		1.0	0.6	1.8	0.8	0.4	1.6	0.596
Ever worked in dusty job^‡‡		1.5	0.8	2.9	1.5	0.7	3.0	0.269

GLI – Global lung function Initiative, NHANES – National Health And Nutrition Examination Survey, BMI -Body mass index, OR - Odds Ratio, CI- Confidence Interval, TB - Tuberculosis, LLN - Lower Limit of Normal, FEV1 –Forced Expiratory Volume in 1 second, FVC- Forced Vital Capacity. We defined chronic airflow obstruction by post-bronchodilator FEV₁/FVC <LLN.

A pack-year of cigarette smoking was defined as smoking of 20 sticks of cigarettes every day for a year.

*Ever worked as a farmer for at least 3 months.

‡Used indoor open fire with coal, charcoal, wood, crop residues or dung as primary means of cooking or heating home or water for more than 6 months.

‡‡Ever worked in a dusty job for a year or more.

Discussion

We found a low prevalence of chronic airflow obstruction (FEV₁/FVC <LLN) in the population aged 40 years and above, low prevalence of respiratory symptoms or diagnosis and low cigarette smoking rates. The key independent predictors of obstruction were a lack of education and previous diagnosis of tuberculosis or asthma.

Little is known about the burden of COPD in Africa, especially sub-Saharan Africa because of lack of high quality population data. We used the standardized spirometry-based protocol developed by the BOLD ­initiative (13) to execute this survey. We applied ­rigorous methods to select a representative sample of the general population and to achieve high quality questionnaire and spirometry data. These quality control measures included daily calibration checks for the spirometers, completion of central training in study methods and quality control measures, ongoing certification and monitoring of technicians, standardized translation of questionnaires and strict adherence to protocol in the collection and cleaning of data for analysis (12). We accepted slightly greater variation in highest and next highest spirometries than recommended by the ATS/ERS (200 ml rather than 150 ml) a relaxation that has been shown to have very little impact on quality and reduces missing data (19).

The Global initiative on Obstructive Lung Disease (GOLD) defines COPD spirometrically, by a fixed ratio of FEV₁ to FVC less than 70%. To take account of the expected age related decline in this ratio, we defined COPD in this study as CAO, the ratio of FEV₁ to FVC below the lower limit of normal (9). The LLN is defined as that value below which 5% of an asymptomatic non-smoking population lies, so a prevalence of 5% is expected in an entirely normal population and 7.7%, the prevalence we noted in Ile-Ife is only 2.7% above this value. Normal values of the FEV₁/FVC ratio are relatively constant across all ethnic and social groups, and we have used the GLI equations for African Americans to define these. For comparison with other BOLD publications and in consonance with BOLD standards, we have also used the NHANES equations for white Americans and the locally derived equations. The prevalence of CAO remains low when local or NHANES equations are applied, further supporting a low prevalence of chronic airflow obstruction in this population irrespective of the choice of reference equation.

The prevalence of obstruction varies widely between populations but seem consistently related to the prevalence of cigarette smoking. In a study of adults aged at least 30 years in a rural district in Uganda, 16.2% had FEV₁/FVC ratio below the LLN using the GLI equations (20). Interestingly, by the fixed GOLD criterion (FEV₁/FVC < 0.7), they found a lower estimate (12.4%) of obstruction. Cigarette smoking and early exposure to biomass smoke from childhood were implicated. In another survey of adults aged 45years and above in Rwanda, Musafiri and colleagues (21) reported a ­prevalence of 9.6% (FEV₁/FVC <LLN) using locally derived equations and found it closely associated with cigarette smoking, male sex and older age. But in a BOLD site in Cape Town, South Africa that used a similar protocol as the present study, 16.3% had evidence of CAO; a rate which is much higher than the prevalence presently reported and attributable to higher prevalence of both cigarette smoking and tuberculosis in that suburb in Cape Town (13,14).

Two Nigerian surveys have reported a low prevalence of obstruction in adults (22,23). Only 2.3% of our study population are current smokers, while 89% are never smokers. This low prevalence of smoking is consistent with the low prevalence of chronic airflow obstruction, an observation previously reported in other BOLD sites with similar smoking rates (13,24). However, because Nigeria is a large and diverse country of over 170 million people, more studies are needed to provide a robust estimate of the national burden of CAO and cigarette smoking (25).

As in previous studies, we also observed that there is a low prevalence of respiratory symptoms and diagnosis among those with spirometric evidence of obstruction (26,27). CAO is often asymptomatic and unrecognized even when present.

Interestingly, we did not find evidence to support an independent association between exposure to biomass fuels and development of CAO. Biomass fuels especially wood smoke is a common source of fuel both for domestic heating and cooking in many developing countries. It is estimated that at least 50% of the world's households depends on biomass as source of domestic energy for cooking and/or heating, and probably even more in sub-Saharan Africa (28–30). Chronic exposure to biomass has been associated with respiratory infections in children (31,32) and reduced lung function in adults (28,33) but there are serious debates and conflicting reports about its role in the aetiology of chronic airflow obstruction. In a meta-analysis of studies on the risk of COPD associated with use of biomass, Hu et al. reported odds ratio of 2.44 (95% CI, 1.90, 3.33) for exposure to biomass compared with non-exposed group in those with COPD (34). However, that analysis did not include data from sub-Saharan Africa where biomass is a leading source of domestic energy, and COPD was defined in many of the studies included in that analysis by self-reported chronic symptoms or a fixed ratio definition of CAO; failing to adjust for the expected decline in lung function with age. Generally, studies with a positive association tend to be published more readily.

In another report on risk of CAO, Hooper et al. (35) using the BOLD global data, analyzed results from 14 sites defining CAO by the more rigorous criteria of FEV₁/FVC <LLN. They did not find any significant association between use of biomass fuels for cooking or heating, and development of CAO. However, this report included only two sites with a high exposure to biomass, potentially making it underpowered to detect an effect. In our study, 67.9% of all the respondents had used either firewood or coal for cooking or heating for at least 6 months, out of which 66% were women and 34% were men.

We did not observe any association between occupational exposures and CAO. Farmers who have a variable occupational exposure to dust, fumes and other chemicals like pesticides, made up 43.5% of the respondents (data not shown) and they were not at increased risk of CAO (OR: 0.8, 95%CI: 0.4, 1.6) (36). Working in a dusty job was also not a significant determinant of airflow obstruction in our population, probably reflecting the fact that Ile-Ife is a non-industrial city.

However, those reporting a previous diagnosis of Tuberculosis (TB) had a high risk for CAO. Previous authors have alluded to this finding (37–40). Only a small number of respondents volunteered this information and as such, the confidence interval is wide. TB is a highly stigmatized clinical condition and the proportion presently reported may be an underestimate. In addition, a previous diagnosis is not a sensitive method of identifying TB, as access to health care is variable and often low in low-income countries. However, like with other studies that have looked at this risk factor (37,40,41), previous diagnosis of TB is notably an important risk factor for developing CAO in many developing countries where the burden of TB is substantial.

We also noted that self-reported asthma diagnosis is a risk factor for chronic airflow obstruction. There is substantial evidence that shows that chronic asthma is both a risk factor for irreversible airway obstruction (42) and for accelerated rate of decline in FEV₁ compared with non-asthma patients, irrespective of cigarette smoking (43) or treatment status (44,45). As smoking was uncommon, we have no reason to think that these respondents may represent misclassified cases of COPD. In the absence of further details about treatment and control status, we are unable to fully interpret this finding.

Those with CAO were more than twice as likely to have no education. Years of education serves as a proxy measure for socioeconomic status and may represent many risk factors associated with deprivation in childhood including nutritional status, respiratory infections, birth weight and possibly, access to health care (39).

This study, being a questionnaire based cross-sectional survey is subject to some element of recall bias. Participants are more likely to overlook or forget previous exposures to possible risk factors that date back to many years such as childhood hospitalization for respiratory conditions. Potentially, this may result in under estimation of effects. Also in cross-sectional designs it is difficult to estimate the temporal relationship of exposure to potential risk factors and development of CAO.

Improvement in socio-economic conditions and the adequate treatment of asthma and tuberculosis are essential for the prevention of chronic airflow ­obstruction in this African population. In addition, prevention of smoking in the younger cohorts is of the greatest importance for maintaining a low prevalence of obstruction.

Conclusion

In conclusion, the prevalence of CAO in Ile-Ife, Nigeria is 7.7% (2.7% above the prevalence defined as normal by the LLN), consistent with a low smoking prevalence and the key independent predictors are poor education and a diagnosis of tuberculosis or asthma.

Declaration of Interest Statement

DOO reports grant from the Wellcome Trust, PGB reports grants from Wellcome Trust, during the conduct of the study; grants from MRC, MRC-PHE, Wellcome Trust, Glaxo Smithkline, and BLF outside the submitted work. GEE, ASB, LG and OOA report no relevant conflict of interest. The sponsor of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.

The authors alone are responsible for the content and writing of the paper.

Acknowledgments

We thank the members of BOLD Coordinating Centre at Imperial College London, including Anamika Jithoo, Sonia Coton, Hadia Azhar, Bernet Kato and James Potts for their assistance with spirometry training, quality control and data management for the study. We are also highly indebted to the American Thoracic Society for initiating this work through the global research training program (MECOR). We appreciate the invaluable comments from William Vollmer of Kaiser Permanente Center for Health Research, William Beckett of Harvard Medical School, Boston and Jane Carter of International Union Against Tuberculosis and Lung Disease (IUATLD) in the writing of this manuscript.

Funding

We wish to thank the Wellcome Trust for the Master's training fellowship awarded to the corresponding author in support for this research (REF: 089405/Z/09/Z).