See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/319164571
Global, regional, and national deaths, prevalence, disability-adjusted life
years, and years lived with disability for chronic obstructive pulmonary
disease and asthma, 1990–2015:...
Article  in  The Lancet Respiratory Medicine · August 2017
CITATIONS READS
21 236
123 authors, including:
Amanuel Abajobir Kalkidan Hassen Abate
African Population and Health Research Center Jimma University
148 PUBLICATIONS   8,778 CITATIONS    172 PUBLICATIONS   11,621 CITATIONS   
SEE PROFILE SEE PROFILE
Semaw Ferede Abera Anurag Agrawal
Mekelle University and University of Hohenheim Institute of Genomics and Integrative Biology
164 PUBLICATIONS   23,387 CITATIONS    269 PUBLICATIONS   9,147 CITATIONS   
SEE PROFILE SEE PROFILE
Some of the authors of this publication are also working on these related projects:
ILD-India Registry View project
Mater-University of Queensland Study of Pregnancy (MUSP) View project
All content following this page was uploaded by Tesfay Mehari Atey on 18 August 2017.
The user has requested enhancement of the downloaded file.
Articles
Global, regional, and national deaths, prevalence, 
disability-adjusted life years, and years lived with disability 
for chronic obstructive pulmonary disease and asthma, 
1990–2015: a systematic analysis for the Global Burden of 
Disease Study 2015
GBD 2015 Chronic Respiratory Disease Collaborators*
Summary
Background Chronic obstructive pulmonary disease (COPD) and asthma are common diseases with a heterogeneous Lancet Respir Med 2017
distribution worldwide. Here, we present methods and disease and risk estimates for COPD and asthma from the Global Published Online 
Burden of Diseases, Injuries, and Risk Factors (GBD) 2015 study. The GBD study provides annual updates on estimates August 16, 2017 
of deaths, prevalence, and disability-adjusted life years (DALYs), a summary measure of fatal and non-fatal disease http://dx.doi.org/10.1016/ 
S2213-2600(17)30293-X
outcomes, for over 300 diseases and injuries, for 188 countries from 1990 to the most recent year.
See Online/Comment 
http://dx.doi.org/10.1016/ 
Methods We estimated numbers of deaths due to COPD and asthma using the GBD Cause of Death Ensemble modelling S2213-2600(17)30308-9
(CODEm) tool. First, we analysed data from vital registration and verbal autopsy for the aggregate category of all chronic *Collaborators listed at the end 
respiratory diseases. Subsequently, models were run for asthma and COPD relying on covariates to predict rates in of the Article
countries that have incomplete or no vital registration data. Disease estimates for COPD and asthma were based on Correspondence to: 
systematic reviews of published papers, unpublished reports, surveys, and health service encounter data from the USA. Prof Theo Vos, Institute for 
We used the Global Initiative of Chronic Obstructive Lung Disease spirometry-based definition as the reference for Health Metrics and Evaluation, 
University of Washington, 
COPD and a reported diagnosis of asthma with current wheeze as the definition of asthma. We used a Bayesian meta- Seattle, WA 98121, USA 
regression tool, DisMod-MR 2.1, to derive estimates of prevalence and incidence. We estimated population-attributable tvos@uw.edu
fractions for risk factors for COPD and asthma from exposure data, relative risks, and a theoretical minimum exposure 
level. Results were stratified by Socio-demographic Index (SDI), a composite measure of income per capita, mean years 
of education over the age of 15 years, and total fertility rate.
Findings In 2015, 3·2 million people (95% uncertainty interval [UI] 3·1 million to 3·3 million) died from COPD 
worldwide, an increase of 11·6% (95% UI 5·3 to 19·8) compared with 1990. There was a decrease in age-standardised 
death rate of 41·9% (37·7 to 45·1) but this was counteracted by population growth and ageing of the global population. 
From 1990 to 2015, the prevalence of COPD increased by 44·2% (41·7 to 46·6), whereas age-standardised prevalence 
decreased by 14·7% (13·5 to 15·9). In 2015, 0·40 million people (0·36 million to 0·44 million) died from asthma, 
a decrease of 26·7% (–7·2 to 43·7) from 1990, and the age-standardised death rate decreased by 58·8% (39·0 to 69·0). 
The prevalence of asthma increased by 12·6% (9·0 to 16·4), whereas the age-standardised prevalence decreased by 17·7% 
(15·1 to 19·9). Age-standardised DALY rates due to COPD increased until the middle range of the SDI before reducing 
sharply. Age-standardised DALY rates due to asthma in both sexes decreased monotonically with rising SDI. The relation 
between with SDI and DALY rates due to asthma was attributed to variation in years of life lost (YLLs), whereas DALY 
rates due to COPD varied similarly for YLLs and years lived with disability across the SDI continuum. Smoking and 
ambient particulate matter were the main risk factors for COPD followed by household air pollution, occupational 
particulates, ozone, and secondhand smoke. Together, these risks explained 73·3% (95% UI 65·8 to 80·1) of DALYs due 
to COPD. Smoking and occupational asthmagens were the only risks quantified for asthma in GBD, accounting for 
16·5% (14·6 to 18·7) of DALYs due to asthma.
Interpretation Asthma was the most prevalent chronic respiratory disease worldwide in 2015, with twice the number of 
cases of COPD. Deaths from COPD were eight times more common than deaths from asthma. In 2015, COPD caused 
2·6% of global DALYs and asthma 1·1% of global DALYs. Although there are laudable international collaborative efforts 
to make surveys of asthma and COPD more comparable, no consensus exists on case definitions and how to measure 
disease severity for population health measurements like GBD. Comparisons between countries and over time are 
important, as much of the chronic respiratory burden is either preventable or treatable with affordable interventions.
Funding Bill & Melinda Gates Foundation.
Copyright © The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY 4.0 license.
www.thelancet.com/respiratory   Published online August 16, 2017   http://dx.doi.org/10.1016/S2213-2600(17)30293-X 1
Articles
Research in context
Evidence before this study We also provide an analysis of how sociodemographic 
Chronic obstructive pulmonary disease (COPD) and asthma have development has a different effect on the burden of COPD and 
been identified as important contributors to fatal and non-fatal asthma. We show that mortality but not prevalence of asthma 
disease burden in all iterations of the Global Burden of Disease is strongly related to sociodemographic development. For 
study (GBD). Since the 1990s, two landmark epidemiological COPD, the burden increases from low sociodemographic 
studies in asthma, the International Study of Asthma and development to the mid-range of our Socio-demographic Index 
Allergies in Childhood and the European Community Respiratory and decreases with increasing development, most likely 
Health Survey, provided comparable evidence of the asthma through the pathways of exposure to smoking and 
prevalence in children and adults, respectively, but in a limited environmental risks. We also present risk factor estimates and 
number of countries. Similarly, for COPD, international initiatives discuss potential new risks that can be added in future GBD 
such as PREPOCOL, PLATINO, BOLD, IBERPOC, and EPI-SCAN iterations.
used standardised population spirometry to quantify COPD and 
Implications of all the available evidence
its severity. An absence of consensus on case definitions and 
COPD and asthma are important contributors to the burden of 
other sources of measurement bias between data sources 
non-communicable diseases. Although much of the burden is 
complicates their estimations.
either preventable or treatable with affordable interventions, 
Added value of this study these diseases have received less attention than other 
In this study, we provide details on the methods used in GBD to non-communicable diseases. Up-to-date population 
minimise measurement error introduced by heterogeneous information on these diseases is key to policy making to improve 
cause of death and prevalence data on COPD and asthma. access to and quality of existing intervention strategies.
Introduction five Colombian cities (PREPOCOL)10 and in five Latin 
Chronic respiratory diseases are among the leading American capital cities (PLATINO)11 provided more 
causes of mortality and morbidity worldwide. Of all population estimates. Although all these studies used 
chronic respiratory diseases, chronic obstructive comparable methods, there is still no universal consensus 
pulmonary disease (COPD) and asthma are the most about the thresholds of spirometry findings to define 
common. These diseases ranked among the top COPD.12,13 The two dominant case definitions for airflow 
20 conditions causing disability globally and were ranked limitation compatible with COPD are a value of less 
eighth (COPD) and 23rd (asthma) as causes of disease than 0·70 for the ratio of FEV1 and forced vital capacity 
burden as measured by disability-adjusted life years (FVC), or the lower limit of normal (LLN) method of 
(DALYs) in 2015.1,2 Yet the measurement of mortality, deriving a threshold as the fifth percentile of FEV1:FVC in 
prevalence, and other population indicators of these two a healthy reference population.14 No universal LLN 
diseases is complicated by misclassification and an threshold exists because it is thought to vary between 
absence of consensus about case definitions. Both death populations.15 Because most people identified with COPD 
rates and prevalence of COPD steeply increase with age. based on spirometry findings report not having been 
The age pattern of asthma mortality resembles that of diagnosed prior to survey, population screening and case-
COPD rather than the relatively steady prevalence in adults finding in symptomatic smokers have been suggested to 
seen in asthma surveys and health service encounter data. provide an opportunity for smoking cessation interventions 
This difference in age patterns between cause of death and before the disease has progressed.16,17
prevalence data sources has been attributed to a range of Most surveys of asthma use a case definition based on 
factors including the commonly reported misclassification self-report of a diagnosis of asthma by a physician and 
of asthma in the elderly as COPD, variable and temporal wheeze (with other respiratory symptoms) in the past 
effects of smoking, and an actual overlap of asthma and 12 months.18 Others have suggested that wheezing 
COPD (asthma COPD overlap; ACO).3,4 However, no symptoms in the past year and bronchial hyper-
consensus exists on the definition of ACO to date.5 Also, responsiveness to inhalation of methacholine or histamine 
evidence from a longitudinal study6 did not show a larger that is reversible with a bronchodilator is a better case 
reduction in lung function in those patients with COPD definition for clinically relevant asthma.19 This case 
and asthma than those without asthma, whereas others definition has been used to measure asthma prevalence in 
have challenged the concept of ACO altogether.7,8 a few surveys, but has not been universally adopted, partly 
Spirometry is the fundamental tool used to define for logistical reasons, but also because of concern about 
and stage COPD and, accordingly, establish population poor specificity and poor prediction of future risk of 
prevalence in surveys. Under the umbrella of the burden asthma in individuals without symptoms.20 However, the 
of obstructive lung disease (BOLD) initiative, surveys have use of biological measurements to improve the validity of 
been done in 29 countries, with surveys in a further the asthma definition depends on the aim of the study. For 
nine countries still in progress.9 Two previous initiatives in instance, bronchial hyper-responsiveness has similar or 
2 www.thelancet.com/respiratory   Published online August 16, 2017   http://dx.doi.org/10.1016/S2213-2600(17)30293-X
Articles
better specificity, but much worse sensitivity, than [ICD]-9 to 490–492, 494, and 496). We used 7301 prevalence 
symptom questionnaires, making it a less suitable method datapoints and 22 incidence datapoints covering 15 of 
for the measurement of prevalence.21,22 21 GBD world regions. No data were available for 
Misclassification and varying case definitions are Andean Latin America, the Caribbean, central Asia, central 
commonly encountered in population health measure- and east sub-Saharan Africa, and Oceania. We used the 
ment.2 A key component of the Global Burden of Disease Global Initiative for Chronic Obstructive Pulmonary 
(GBD) analyses is to identify and correct for such sources Disease (GOLD) spirometry-based definition for COPD 
of measurement bias. (a ratio of FEV1:FVC <0·70 after bronchodilation)14 and 
In this Article, we present the results of estimating modelled overall prevalence and the proportions in COPD 
mortality, prevalence, and disease burden in DALYs and spirometry stages mild (FEV1 ≥80% of normal), moderate 
years lived with disability (YLDs) for COPD and asthma (FEV1 50–79% of normal), and severe or very severe 
from the GBD 2015 study. We also report on the attribution combined (FEV1 <50% of normal) in DisMod-MR 2.1, 
of risk factors for these diseases and the relation between a Bayesian meta-regression tool. DisMod-MR 2.1 takes all 
disease burden and the Socio-demographic Index (SDI), available data on prevalence, incidence, remission (defined 
a compound measure of income, years of education, and in GBD as the cure rate), and cause of death rates jointly 
total fertility rate. into account and forces a consistent set of estimates for 
each parameter. Before entering data into DisMod-MR 2.1, 
Methods we adjusted survey data using different spirometry case 
Mortality definitions. We adjusted datapoints from 14 studies 
The methods of the GBD 2015 study have been extensively reporting on the GOLD case definition without a 
reported elsewhere.1,2,23 Briefly, deaths, incidence, preva- bronchodilator after fitting an exponential curve to age-
lence, and DALY rates were estimated for 310 diseases and specific ratios of both measurements from three studies.24–26 
injuries for 195 countries and territories by age group and Using a similar approach, we adjusted datapoints from 
sex from 1990 to 2015. All-cause mortality was derived six studies reporting LLN pre-bronchodilator data based on 
from vital registration systems, censuses, and surveys, and one study,24 three studies with LLN post-bronchodilator 
analysed with demographic methods to correct for data based on five studies,12,24,27–29 and two studies using an 
incompleteness. Causes of death, derived from an older version of LLN by the European Respiratory Society 
extensive database of vital registration and verbal autopsy based on two studies.30,31 We used the meta-regression 
data, were analysed using GBD’s Cause Of Death component of DisMod-MR 2.1 to determine an adjustment 
Ensemble modeling (CODEm) tool to calculate mixed factor for data based on physician diagnosis and the US 
effects or spatiotemporal Gaussian process regression health service encounter data. We included a scalar for the 
models of rates or cause fractions with varying combined exposure to all risks estimated for COPD as a 
combinations of predictive covariates. Predictive validity predictive covariate. We included corresponding datapoints 
testing determined the optimal ensemble of models. for excess mortality rate estimated as the ratio of cause-
Covariates included smoking prevalence, cigarettes per specific mortality rate and prevalence corresponding to the 
capita, the proportion of the population exposed to same year and age range of the datapoint. We used lag-
household air pollution, mean exposure to ambient distributed income per capita as a predictive covariate for 
particulate matter (defined as the population-weighted excess mortality, forcing a negative coefficient on the 
annual average mass concentration of particles with a assumption that case fatality decreases with increasing 
diameter less than 2·5 μm [PM2·5] in a m³ of air) from wealth in a country. Prevalence by GOLD class was 
outdoor air pollution, a scalar of the combined exposure to available from only 24 countries in 14 GBD world regions.
risks for COPD (and asthma), and SDI. Because the The proportions of people in GOLD classes I, II, and III 
sensitivity of verbal autopsy algorithms to detect specific or IV were modelled separately in DisMod-MR 2.1 and 
chronic respiratory diseases is poor, we only modelled data then scaled to a sum of 1 and multiplied by the overall 
on deaths from all chronic respiratory diseases in CODEm prevalence of COPD. In GBD, severity of COPD is 
and constrained the estimates for specific chronic classified into health states (appendix p 4). To map the See Online for appendix
respiratory diseases to the estimates for all chronic prevalence by GOLD class into health states representing 
respiratory deaths. We constrained estimates for all symptoms, we used the Medical Expenditure Panel Survey 
individual causes to the all-cause mortality rates derived (MEPS) data32 for 2001–11 from the USA. MEPS is an 
from demographic estimation. ongoing data collection project with new panels recruited 
every 2 years. Respondents report on all health service 
Non-fatal estimation for COPD contacts and the reasons for those contacts. We identified 
Non-fatal estimates for COPD were based on systematic individuals with an ICD-9 diagnosis of COPD. We 
reviews of published papers, unpublished reports, surveys translated scores from a generic quality-of-life instrument, 
available in GBD’s Global Health Data Exchange the 12-Item Short Form Health Survey (SF-12),33 into GBD 
repository, and health service encounter data from the disability weight values based on convenience samples of 
USA (coded in International Classification of Diseases research fellows at the Institute for Health Metrics and 
www.thelancet.com/respiratory   Published online August 16, 2017   http://dx.doi.org/10.1016/S2213-2600(17)30293-X 3
Articles
Evaluation and annual GBD workshop participants filling availability of exposure data, potential for modification, 
in SF-12 for a selection of 60 of the 235 health states used and policy interest are criteria for choosing risks and 
in GBD 2015. Health states were presented as lay associated outcomes in GBD. Population-attributable 
descriptions that had been the basis of the pairwise fractions of disease outcomes were estimated from 
comparisons presented to respondents to the GBD exposure data, relative risks of outcomes, and a theoretical 
disability weight surveys. After controlling for comorbidity, minimum level of exposure. Population surveys were the 
we assigned a specific disability weight to each individual main source of exposure data on smoking, second-hand 
with a diagnosis of COPD. We categorised cases into smoke, and household air pollution. Exposure to PM2·5 was 
asymptomatic (disability weight value of 0), mild COPD measured from satellite data on aerosols in the atmosphere 
(disability weight value between 0 and the midpoint of and calibrated to observations from ground monitors. We 
GBD disability weights for mild and moderate COPD), based exposure to ozone on a chemical transport model of 
moderate COPD (disability weight value greater than the satellite data.34 Occupational exposures were based on the 
midpoint between mild and moderate and midpoint proportion of the working population exposed to 
between moderate and severe COPD disability weights), asthmagens and particulates based on distribution of the 
and severe COPD (the remainder). We took the prevalence population in nine occupational groups as reported by the 
estimates for the USA in 2005 (at the mid point of MEPS International Labor Organization.35 We derived relative 
data range) and mapped the distribution of cases by GOLD risks from meta-analyses of cohort studies. The theoretical 
classes into the distribution of severity from MEPS minimum exposure level was set as zero for smoking, 
(appendix p 5). This gave us a mapping from GOLD class second-hand smoke, and the occupational exposures. For 
into GBD health states, which could then be applied to the household air pollution, the minimum was defined as no 
prevalence data by GOLD class from all other countries household reporting use of solid fuel for cooking. For 
and time periods. ambient particulate matter, the minimum was set as a 
uniform distribution between the lowest and fifth 
Non-fatal estimation for asthma percentile exposure level from all data sources. For ozone, 
The main data sources for asthma were population surveys the minimum was set as a uniform distribution between 
and US health service encounter data on the diagnoses for the lowest and fifth percentile exposure measured in the 
any health service contact for 42 million people. We used American Cancer Society’s Cancer Prevention Study II.36 
9219 prevalence, 29 incidence, and 32 remission datapoints Unlike disease estimates that are mutually exclusive and 
and population death rates from asthma estimated in collectively exhaustive in GBD, risk estimates are based on 
CODEm and scaled to total death rates with all other cause- a counterfactual analysis (what if past exposure to a risk 
specific estimates. Data on prevalence were available for had been at the theoretical minimum level?) and are, 
121 countries covering all 21 GBD world regions. Our therefore, not additive. Estimates of combinations of risks 
case definition for asthma was a reported diagnosis by take mediation into account based on the difference in 
a physician, with wheezing in the past 12 months. In relative risks from cohort and trial data that did and did not 
DisMod-MR 2.1, we adjusted data based on reported control for another risk as a confounder. After adjustment 
wheezing only and US health service encounter data, and for mediation, risks were combined using a multiplicative 
used a scalar of the combined exposure to risk factors for function to avoid the sum of risks exceeding the total 
asthma. Similar to the COPD model, we added excess amount of disease.23 Additional details on the estimation 
mortality rates corresponding to all prevalence datapoints process for COPD and asthma risks can be found in the 
with lag-distributed income per capita as a predictive appendix (pp 29–57).
covariate. The health states and disability weights for 
three asthma health states are listed in the appendix (p 4). DALY estimation
The distribution between the three asthma health states We calculated years of life lost (YLLs) by multiplying 
and an asymptomatic health state was analysed in MEPS. the number of deaths for a cause by the remaining 
In the absence of comparable epidemiological severity life expectancy in GBD’s standard life table based 
distribution data, a simplifying assumption had to be on the lowest observed mortality rates at each age in 
made that the US distribution of severity for asthma can be any population over 5 million.1 We calculated YLDs by 
generalised to all countries. Additional details on the multiplying the prevalence of each sequela by the 
estimation process for COPD and asthma can be found in disability weight that quantifies the relative severity of the 
the appendix (pp 15–28). sequela on a scale between 0 and 1. We derived disability 
weights from nine population surveys and an open-access 
Risk estimation internet survey using pairwise comparison methods.37 
Estimates were made of six risk factors for COPD DALYs are the sum of YLLs and YLDs. We estimated 
(smoking, second-hand smoke, household air pollution, uncertainty by recalculating every outcome of interest 
ambient particulate matter, ozone, and occupational 1000 times, drawing from distributions of the sampling 
particulates) and two risk factors for asthma (smoking and error around input data, corrections for measurement 
occupational asthmagens). Sufficient evidence of causality, error, and estimates of residual non-sampling error and, 
4 www.thelancet.com/respiratory   Published online August 16, 2017   http://dx.doi.org/10.1016/S2213-2600(17)30293-X
Articles
in the case of cause of death estimates, model selection. study design, data collection and analysis, interpretation of 
Uncertainty intervals (UIs) were defined as the 25th and data, decision to publish, or preparation of the manuscript.
975th values of the posterior distributions. We computed 
differences between estimates at the 1000-draw level and Results
reported them as significant if more than 95% of values In 2015, 3·2 million people (95% UI 3·1 million to 
for the difference were either positive or negative. We 3·3 million) died from COPD worldwide, an increase of 
computed age-standardised rates using the GBD standard 11·6% (5·3–19·8) compared with 1990, despite a 
population.1 decrease in the age-standardised death rate of 41·9% 
SDI is an index of sociodemographic development (37·7–45·1). Population growth and ageing of the global 
consisting of lagged distributed income per capita, mean population outweighed the downward trend in age-
years of education over the age of 15 years, and total fertility standardised death rates. The greatest reduction in age-
rate.1 Each component was given equal weight and rescaled standardised death rates occurred in countries in the 
from 0 (for the lowest value observed during 1980–2015) to high-middle-SDI quintile and middle-SDI quintile. 
1 (for the highest value observed) for income per capita and From 1990 to 2015, the prevalence of COPD increased 
average years of schooling, and the reverse for the total by 44·2% (95% UI 41·7–46·6) to 174·5 million 
fertility rate. The final SDI score was computed as the individuals (160·2 million to 189·0 million). The 
geometric mean of each of the components. We classified decrease in age-standardised prevalence of 14·7% 
countries into five quintiles based on the entire distribution (13·5–15·9) was much smaller than the decrease in age-
of location-year combinations between 1980 and 2015. We standardised death rates. The greatest decrease in age-
present results on each country’s position based on its standardised prevalence was seen in countries in the 
2015 SDI value. A LOESS regression on all data from 1980 high-middle-SDI quintile and the middle-SDI quintile 
to 2015 was done to define the expected relationship (table 1).
between SDI and each health outcome. We contrast In 2015, 0·40 million people (95% UI 0·36 million to 
observed disease rates against this expected level to identify 0·44 million) died from asthma, a decrease of 26·7% 
world regions performing better or worse than expected (–7·2 to 43·7) compared with 1990. The decrease in 
based on their development status. age-standardised death rates was 58·8% (39·0–69·0) 
This study is compliant with the Guidelines for Accurate between 1990 and 2015. The greatest reduction in 
and Transparent Health Estimates Reporting (GATHER) age-standardised death rates occurred in countries in the 
with details provided in the appendix (pp 58–60).38 high-SDI and low-middle-SDI quintiles. From 1990 
to 2015, the prevalence of asthma increased by 12·6% 
Role of the funding source (9·0–16·4) to 358·2 million individuals (323·1 million to 
This research was supported by funding from the Bill & 393·5 million). The decrease in age-standardised 
Melinda Gates Foundation. The funders had no role in the prevalence by 17·7% (15·1–19·9) was smaller than the 
Number of deaths Percentage change in Percentage change in Number of prevalent cases Percentage change in Percentage change in 
(thousands) all-age deaths, age-standardised death (thousands) all-age prevalence, age-standardised 
1990–2015 rates, 1990–2015 1990–2015 prevalence, 1990–2015
COPD
Global 3188 (3084 to 3293) 11·6 (5·3 to 19·8) –41·9 (–45·1 to –37·7) 174 483 (160 205 to 188 952) 44·2 (41·7 to 46·6) –14·7 (–15·9 to –13·5)
High SDI quintile 482 (468 to 505) 31·6 (27·8 to 38·2) –26·2 (–28·2 to –22·5) 43 105 (39 912 to 46 414) 35·3 (31·8 to 39·1) –7·3 (–9·3 to –4·9)
High-middle SDI quintile 626 (602 to 651) –11·1 (–17·4 to –4·3) –57·8 (–60·8 to –54·7) 44 923 (41 215 to 48 803) 42·3 (39·3 to 45·1) –20·2 (–21·5 to –19·0)
Middle SDI quintile 1110 (1055 to 1169) –3·4 (–11·0 to 5·8) –53·5 (–57·2 to –49·2) 52 209 (47 430 to 57 154) 104·8 (101·6 to 108·0) –22·6 (–23·9 to –21·4)
Low-middle SDI quintile 907 (850 to 965) 51·5 (24·1 to 89·5) –25·7 (–38·1 to –7·9) 30 058 (27 495 to 32 719) 74·3 (71·6 to 77·2) –3·7 (–4·8 to –2·7)
Low SDI quintile 61 (52 to 71) 68·8 (40·3 to 111·3) –16·3 (–30·1 to 3·5) 4223 (3795 to 4656) 36·1 (33·2 to 38·9) –1·6 (–3·1 to –0·1)
Asthma
Global 397 (363 to 439) –26·7 (–43·7 to 7·2) –58·8 (–69·0 to –39·0) 358 198 (323 134 to 393 466) 12·6 (9·0 to 16·4) –17·7 (–19·9 to –15·1)
High SDI quintile 22 (20 to 24) –53·2 (–56·8 to –48·9) –71·8 (–73·7 to –69·5) 63 883 (59 724 to 68 309) –13·8 (–17·0 to –10·2) –26·0 (–28·4 to –23·0)
High-middle SDI quintile 54 (50 to 61) –3·2 (–12·9 to 9·0) –49·4 (–54·9 to –43·0) 76 935 (69 650 to 84 654) 8·4 (4·1 to 13·0) –15·2 (–18·0 to –12·1)
Middle SDI quintile 120 (110 to 132) –12·2 (–28·7 to 16·0) –38·4 (–49·5 to –19·4) 91 375 (82 505 to 100 370) 8·3 (4·2 to 12·6) –14·5 (–16·3 to –12·3)
Low-middle SDI quintile 159 (136 to 186) –40·5 (–61·4 to 19·0) –69·6 (–81·3 to –32·9) 90 605 (79 887 to 101 371) 28·9 (24·6 to 33·4) –18·4 (–20·7 to –15·9)
Low SDI quintile 41 (34 to 51) 22·1 (2·1 to 55·4) –54·3 (–63·9 to –38·3) 35 011 (30 065 to 40 255) 94·8 (86·2 to 106·1) –7·3 (–11·2 to –3·0)
Data in parentheses are 95% uncertainty intervals. SDI is calculated for each location (all 188 countries, seven territories, and 519 subnational locations estimated in GBD 2015) as a function of lag-distributed 
income per capita, average educational attainment in the population aged over 15 years, and the total fertility rate. SDI of 0 represents the lowest level of income per capita, educational attainment, and highest 
total fertility rate observed from 1980 to 2015, and SDI of 1 represents the highest income per capita, educational attainment, and lowest total fertility rate with an effect on health over the same period. Cutoffs 
on the SDI scale for the quintiles have been selected based on their 2015 values by location. COPD=chronic obstructive pulmonary disease. SDI=Socio-demographic Index.
Table 1: Deaths due to asthma and COPD and number of prevalent cases of disease in 2015 and percentage change in all-age and age-standardised rates in locations grouped by SDI quintile
www.thelancet.com/respiratory   Published online August 16, 2017   http://dx.doi.org/10.1016/S2213-2600(17)30293-X 5
Articles
Number of YLLs, all ages Number of YLDs, all ages Number of DALYs, all Percentage change in Percentage change in 
(thousands) (thousands) ages (thousands) DALYs, 1990–2015, age-standardised 
all ages DALY rates, 
1990–2015
COPD
Global 51 803 (49 898 to 53 611) 12 047 (10 207 to 13 725) 63 850 (61 215 to 66 289) –1·0 (–7·1 to 6·2) –43·7 (–47·0 to –39·8)
High SDI quintile 5914 (5762 to 6180) 2214 (1890 to 2545) 8128 (7755 to 8530) 12·7 (9·9 to 16·6) –28·2 (–30·1 to –25·9)
High-middle SDI quintile 9058 (8693 to 9446) 2500 (2103 to 2882) 11 661 (11 093 to 12 226) –19·8 (–25·0 to –14·2) –58·5 (–61·2 to –55·6)
Middle SDI quintile 17 918 (16 979 to 18 887) 4050 (3422 to 4623) 21 812 (20 738 to 22 908) –16·4 (–22·3 to –9·2) –55·8 (–58·9 to –52·1)
Low-middle SDI quintile 17 444 (16 260 to 18 652) 2954 (2493 to 3345) 20 399 (19 079 to 21 673) 32·0 (8·0 to 61·0) –27·0 (–40·0 to –10·6)
Low SDI quintile 1433 (1203 to 1685) 374 (316 to 429) 1806 (1559 to 2064) 55·7 (31·3 to 86·9) –18·0 (–30·5 to –0·5)
Asthma
Global 10 270 (9369 to 11 448) 15 899 (10 371 to 22 344) 26 169 (20 501 to 32 583) –14·6 (–26·0 to 2·1) –42·8 (–52·0 to –29·5)
High SDI quintile 384 (366 to 408) 2818 (1838 to 3905) 3203 (2221 to 4299) –25·4 (–29·9 to –21·9) –35·9 (–40·2 to –32·6)
High-middle SDI quintile 1227 (1130 to 1400) 3419 (2228 to 4795) 4766 (3508 to 6154) –3·8 (–9·1 to 0·9) –30·3 (–35·6 to –25·9)
Middle SDI quintile 2912 (2655 to 3223) 4061 (2657 to 5704) 6855 (5464 to 8453) –12·5 (–22·9 to –0·7) –40·7 (–49·3 to –30·2)
Low-middle SDI quintile 4327 (3728 to 5073) 4020 (2621 to 5682) 8350 (6705 to 10 088) –27·4 (–44·6 to 9·1) –60·8 (–71·8 to –31·9)
Low SDI quintile 1402 (1162 to 1692) 1563 (1009 to 2238) 2961 (2343 to 3687) 45·9 (27·2 to 67·9) –31·4 (–41·2 to –18·4)
Data in parentheses are 95% uncertainty intervals. YLLs=years of life lost. YLDs=years lived with disability. DALYs=disability-adjusted life years. COPD=chronic obstructive 
pulmonary disease. SDI=Socio-demographic Index.
Table 2: YLLs, YLDs, and DALYs due to asthma and COPD in 2015 and percentage change in all-age counts and age-standardised DALY rates from 1990 to 
2015 in locations grouped by SDI quintiles
Age-standardised DALY 
rate per 100 000 people
100–300
301–600
601–1000
1001–2000
2001–4500
ATG VCT Barbados Comoros Marshall Isl Kiribati
West Africa Eastern 
Mediterranean
Solomon Isl FSM
Dominica Grenada Maldives Mauritius Malta
Vanuatu Samoa
Caribbean LCA TTO TLS Seychelles Persian Gulf Singapore Balkan Peninsula Fiji Tonga
Figure 1: Age-standardised DALY rate per 100 000 people due to COPD by country, both sexes, 2015
DALYs=disability-adjusted life years. COPD=chronic obstructive pulmonary disease. ATG=Antigua and Barbuda. FSM=Federated States of Micronesia. Isl=islands. LCA=Saint Lucia. TLS=Timor-Leste. 
TTO=Trinidad and Tobago. VCT=Saint Vincent and the Grenadines.
6 www.thelancet.com/respiratory   Published online August 16, 2017   http://dx.doi.org/10.1016/S2213-2600(17)30293-X
Articles
Age-standardised DALY 
rate per 100 000 people
100–200
201–300
301–500
501–800
801–1200
1201–3000
ATG VCT Barbados Comoros Marshall Isl Kiribati
West Africa Eastern 
Mediterranean
Solomon Isl FSM
Dominica Grenada Maldives Mauritius Malta
Vanuatu Samoa
Caribbean LCA TTO TLS Seychelles Persian Gulf Singapore Balkan Peninsula Fiji Tonga
Figure 2: Age-standardised DALY rate per 100 000 people due to asthma, by country, both sexes, 2015
DALYs=disability-adjusted life years. ATG=Antigua and Barbuda. FSM=Federated States of Micronesia. Isl=islands. LCA=Saint Lucia. TLS=Timor-Leste. TTO=Trinidad and Tobago. VCT=Saint Vincent and 
the Grenadines.
overall decrease in age-standardised death rates. The The 63·9 million DALYs (95% UI 61·2 million to 
age-standardised death rate for asthma in 2015 was higher 66·3 million) due to COPD represented 2·6% (95% UI 
in males (6·7 [5·9–7·5] per 100 000 people) than in females 2·4–2·8) of the entire global burden of disease in 2015. 
(5·6 [4·8–6·4] per 100 000 people). A greater reduction in 26·2 million DALYs (20·5 million to 32·6 million) due to 
age-standardised prevalence was seen in countries in the asthma contributed 1·1% (0·9–1·3) of the total burden 
high-SDI and low-middle-SDI quintiles (table 1). in 2015 (table 2). The greatest decrease in age-standardised 
Globally, COPD affected 104·7 million males (95% UI DALY rates due to COPD occurred in countries in the 
96·0 million to 113·8 million) and 69·7 million females high-middle-SDI and middle-SDI quintiles. The biggest 
(64·2 million to 75·4 million) in 2015. Age-standardised reduction in age-standardised asthma DALY rates occurred 
prevalence was 3·2% (2·9–3·5) in males and 2·0% in the low-middle-SDI quintile (table 2).
(1·8–2·1) in females. Age-standardised DALY rates in Age-standardised DALY rates due to COPD in 2015 were 
males (1273·0 [95% UI 1215·5–1328·3] per 100 000 people) estimated to exceed 2000 per 100 000 people in Papua New 
were almost twice as high as those in females (717·4 Guinea, India, Lesotho, and Nepal. Rates below 300 per 
[677·7–759·3] per 100 000 people) reflecting a higher male- 100 000 people were seen in some countries in high-income 
to-female ratio for deaths than for prevalence. Conversely, Asia Pacific, central Europe, north Africa and Middle East, 
age-standardised DALY rates due to asthma were similar the Caribbean, western Europe, and Andean Latin America 
between male individuals (365 [290–451] per 100 000 people) (figure 1). Age-standardised asthma DALY rates in excess of 
and female individuals (368 [286–461] per 100 000 people). 1200 per 100 000 people were estimated for Afghanistan, 
In 2015, more females (190·2 million [172·2 million to Central African Republic, Fiji, Kiribati, Lesotho, Papua New 
208·9 million]) than males (168·0 million [150·8 million to Guinea, and Swaziland. Countries in eastern and central 
185·1 million]) had asthma; a reversal of the higher male- Europe, China, Italy, and Japan had asthma DALY rates 
to-female ratio during adolescence. between 100 and 200 per 100 000 people (figure 2). DALY 
YLLs contributed more than 80% of DALYs due to estimates for COPD and asthma by country and the 
COPD. Conversely, asthma is highly prevalent at all ages percentage change in DALYs and age-standardised DALY 
and leads to fewer deaths than COPD and thus YLDs rates between 1990 and 2015 are presented in the appendix 
formed the larger component of DALYs, at just over 60%. (p 61–67).
www.thelancet.com/respiratory   Published online August 16, 2017   http://dx.doi.org/10.1016/S2213-2600(17)30293-X 7
Articles
Male Female Male Female
1·00 COPD 1·00 COPD
Asthma Asthma
0·75 0·75
0·50 0·50
0·25 0·25
0 0
1600 1200 800 400 0 400 800 1200 1600 3000 2250 1500 750 0 750 1500 2250 3000
All-age DALY rate per 100 000 people Age-standardised DALY rate per 100 000 people
Figure 3: Expected relationship between all-age DALY rates due to COPD and Figure 4: Expected relationship between age-standardised DALY rates due to 
asthma and SDI by sex, 2015 COPD and asthma and SDI by sex, 2015
DALYs=disability-adjusted life years. COPD=chronic obstructive pulmonary DALYs=disability-adjusted life years. COPD=chronic obstructive pulmonary 
disease. SDI=Socio-demographic Index. disease. SDI=Socio-demographic Index.
a combined effect of population growth, ageing, and 
YLLs YLDs
1·00 COPD variation in prevalence. The change in age-standardised 
Asthma DALY rates with SDI shows an increase in DALY rates due 
to COPD until the middle range of SDI values and then a 
sharp decline. DALY rates due to asthma in both sexes 
0·75 decreased monotonically with rising SDI (figure 4). The 
relationship between DALY rates due to asthma and SDI 
largely reflected variation in YLLs, whereas DALY rates 
due to COPD varied similarly for YLLs and YLDs across 
0·50 the SDI continuum (figure 5).
The GBD regions of Oceania, east Asia, south Asia, and 
high-income North America had higher age-standardised 
DALY rates due to COPD in both sexes than expected 
based on their SDI. Male individuals in eastern Europe 
0·25
also had higher than expected DALY rates. Regions with 
better-than-expected COPD DALY rates included eastern, 
central, and western sub-Saharan Africa; central and 
Andean Latin America and the Caribbean; and north 
0
3000 2250 1500 750 0 250 500 Africa and the Middle East (figure 6). Age-standardised 
DALY rates due to asthma in Oceania were much higher 
Age-standardised DALY rate per 100 000 people than expected based on SDI. Australasia, southeast Asia, 
Figure 5: Expected relationship between age-standardised DALY rates due to the Caribbean, and southern sub-Saharan Africa also had 
COPD and asthma and SDI by YLLs and YLDs, 2015 higher DALY rates than expected. The asthma DALY rates 
DALYs=disability-adjusted life years. COPD=chronic obstructive pulmonary in south Asia were higher than expected in 1990 (when 
disease. SDI=Socio-demographic Index. YLLs=years of life lost. YLDs=years lived 
with disability. SDI was lowest), but converged with expected values 
in 2015. Central Europe, east Asia, and western and eastern 
Examining the expected relationship between SDI and sub-Saharan Africa had lower than expected asthma DALY 
all-age DALY rates showed a reduction in asthma rates rates (figure 7).
with increasing SDI in both sexes, whereas DALY rates Smoking and ambient particulate matter were the main 
due to COPD increased up until around 0·5 SDI, decreased risks for COPD followed by household air pollution, 
to the lowest values at an SDI value of 0·75, after which occupational particulates, ozone, and second-hand smoke 
they slightly increased (figure 3). These patterns reflect (figure 8). Together, these risks explained 73·3% (95% UI 
8 www.thelancet.com/respiratory   Published online August 16, 2017   http://dx.doi.org/10.1016/S2213-2600(17)30293-X
SDI SDI
SDI
Articles
Male
5000 Global
High-income Asia Pacific
High-income North America
4000 Western Europe
Australasia
Andean Latin America
3000 Tropical Latin America
Central Latin America
Southern Latin America
2000 Caribbean
Central Europe
Eastern Europe
1000 Central Asia
North Africa and Middle East
South Asia
0 Southeast Asia
East Asia
Female Oceania
5000
Western sub-Saharan Africa
Eastern sub-Saharan Africa
Central sub-Saharan Africa
4000
Southern sub-Saharan Africa
3000
2000
1000
0
0 0·4 0·6 0·8
SDI
Figure 6: Age-standardised DALY rates due to COPD by 21 GBD world regions and the expected value based on the SDI by sex, 1990–2015
The black line represents the expected value of a disease rate based on a LOESS regression of all years of estimates by GBD locations and their SDI value. 
DALYs=disability-adjusted life years. COPD=chronic obstructive pulmonary disease. GBD=Global Burden of Disease. SDI=Socio-demographic Index. LOESS=locally 
weighted regression and smoothing scatterplots. 
65·8–80·1) of DALYs due to COPD. Smoking and affected in 2015. Deaths from COPD were eight times 
occupational asthmagens were the only risks quantified more common than deaths from asthma. YLLs contributed 
for asthma in GBD, explaining just 16·5% (14·6–18·7) of 81·2% of the 63·8 million global DALYs due to COPD, 
the asthma DALYs. The contribution of risks to the burden whereas YLDs represented the largest proportion of the 
of COPD varied by SDI quintiles. In high-SDI countries, 26·2 million global DALYs due to asthma. COPD ranked 
the behavioural risks (smoking and second-hand smoke) eighth (2·6% of global DALYs) and asthma 23rd (1·1% of 
were the most important, whereas environmental risks global DALYs) among the 315 GBD causes in 2015. Age-
and, to a lesser extent, occupational risks explained most of standardised DALY rates from COPD and asthma declined 
the burden in lower-SDI quintiles. The proportions of significantly by 43·7% (39·8–47·0) for COPD and by 
COPD burden not explained by any of the GBD risks 42·8% (29·5–52·0) for asthma annually between 1990 and 
showed little variation between SDI quintiles (figure 9, 2015. Most of the reductions have come from a reduction 
table 3). in mortality, by 41·9% for COPD and 58·8% for asthma. 
More detailed GBD 2015 results are available for The reductions in YLDs have been much smaller. This For the GBD 2015 results see 
download online and in online visualisation tools, and the finding reflects greater improvements in reducing case http://ghdx.healthdata.org/gbd-
GBD 2015 code can be accessed online. fatality rather than a change in incidence and prevalence. results-tool
The absence of a relationship between asthma prevalence For the online visualisation 
Discussion tools see https://vizhub.and asthma death rates, and our knowledge about asthma healthdata.org/gbd-compare
Asthma was the most prevalent chronic respiratory pathophysiology and clinical trial findings, add evidence For the GBD 2015 code see 
disease, affecting an estimated 358 million people in 2015. that most asthma deaths at all ages are preventable by http://ghdx.healthdata.org/gbd-
COPD was half as common, with 174 million people treatment with low-dose inhaled corticosteroids and other 2015-code
www.thelancet.com/respiratory   Published online August 16, 2017   http://dx.doi.org/10.1016/S2213-2600(17)30293-X 9
Age-standardised DALY rate per 100 000 people Age-standardised DALY rate per 100 000 people
Articles
Male
3000 Global
High-income Asia Pacific
High-income North America
Western Europe
Australasia
2000 Andean Latin America
Tropical Latin America
Central Latin America
Southern Latin America
Caribbean
1000 Central Europe
Eastern Europe
Central Asia
North Africa and Middle East
South Asia
0 Southeast Asia
East Asia
Female Oceania
3000
Western sub-Saharan Africa
Eastern sub-Saharan Africa
Central sub-Saharan Africa
Southern sub-Saharan Africa
2000
1000
0
0 0·4 0·6 0·8
SDI
Figure 7: Age-standardised DALY rates due to asthma by 21 GBD world regions and the expected value based on the SDI by sex, 1990–2015
The black line represents the expected value of a disease rate based on a LOESS regression of all years of estimates by GBD locations and their SDI value. 
DALYs=disability-adjusted life years. GBD=Global Burden of Disease. SDI=Socio-demographic Index. LOESS=locally weighted regression and smoothing scatterplots. 
prevalence of asthma. The relationship between SDI and 
Smoking COPD is less monotonic. Higher COPD death rates and 
Ambient particulate matter prevalence at the middle range of SDI values reflect the 
Household air pollution increase in smoking and outdoor air pollution observed 
Occupational particulates in countries going through the demographic and 
epidemiological transition.
Ozone
Between the GBD 2013 and GBD 2015 iterations, 
Occupational asthmagens COPD a methodological change led to a significant difference in 
Secondhand smoke Asthma prevalence and YLDs due to COPD. In the GBD 2013 
0 50 100 150 200 250 300 350 400 450 study, we estimated 328·5 million prevalent cases and 
DALY rate per 100 000 people 26·1 million YLDs for the year 2013, whereas the GBD 
2015 estimates for 2015 were 174·5 cases and 12·0 million 
Figure 8: Age-standardised DALY rates due to COPD and asthma attributable to seven risk factors, both sexes, 
2015 YLDs. This difference is due to a shift from taking LLN 
COPD=chronic obstructive pulmonary disease. DALYs=disability-adjusted life years. estimates as the reference case definition to using the 
fixed-ratio definition of GOLD. This change in the methods 
management strategies or, to a lesser extent, avoidance of led to much lower prevalence estimates in 2015 than in 
risk factors. Indeed, the observed low asthma mortality in 2013 because the GOLD criteria identified lower prevalence 
high-income countries reflects better access to health in younger adults than older adults. Younger adults 
services and better treatment options following represent a much larger proportion of the world’s 
international asthma guidance.39 This fact is reflected by a population, and therefore a higher estimate of prevalence 
strong relationship between SDI and mortality, but not at these ages affects the total prevalence more than an 
10 www.thelancet.com/respiratory   Published online August 16, 2017   http://dx.doi.org/10.1016/S2213-2600(17)30293-X
Risk factors
Age-standardised DALY rate per 100 000 people Age-standardised DALY rate per 100 000 people
Articles
equivalent change in the prevalence at older ages. The 
2500 Environmental, and occupational
main motivation to revert to the fixed-ratio GOLD Behavioural, environmental, and occupational
definition for GBD 2015 was that we aimed to estimate Behavioural
2000
symptomatic disease. Our severity distributions are Unattributed
derived from epidemiological data on GOLD classes, 1500
which fit better with the estimation of prevalence based on 
GOLD’s fixed ratio criteria than LLN. Most of the 1000
arguments for using an LLN case definition are based on 
future risk of disease to identify people with early signs of 500
disease, who could be prevented from developing 
symptomatic disease by measures such as smoking 0
cessation. Both in primary and secondary care, clinicians High SDI High-middle Middle SDI Low-middle Low SDI
rely on respiratory symptoms, exposure to major known SDI SDI
SDI quintile
risks, and airflow limitation to diagnose COPD clinically. 
Accurate LLN estimation requires prediction reference Figure 9: Contribution of behavioural and environmental and occupational 
equations, and to date there are no universal prediction risks to DALYs due to COPD per 100 000 people in locations grouped by SDI 
equations for LLN because spirometry is not only variable quintiles, 2015
Environmental and occupational: ambient particulate matter, household air 
by age and sex, but is also race-related and affected by pollution, occupational particulates, and ozone. Behavioural: smoking and 
different local environments.15,40 To date, the fixed ratio has second-hand smoke. Behavioural environmental: ambient particulate matter, 
gained more popular use in clinical practice because it is household air pollution, occupational particulates, ozone, smoking, and 
easy to calculate, thus helping to remove barriers to the second-hand smoke. DALYs=disability-adjusted life years. COPD=chronic obstructive pulmonary disease. SDI=Socio-demographic Index.
widespread use of spirometry and diagnosis of fixed 
airflow obstruction. Furthermore, nearly all evidence on 
efficacy and safety of respiratory drugs and other Percentage Percentage Percentage Percentage contribution from contribution from contribution from unattributed
treatments comes from randomised trials with patients environmental behavioural and behavioural risks
identified using a fixed ratio definition.41 and occupational environmental and 
No comparable estimates of the global prevalence of risks occupational risks
COPD exist other than those made for previous iterations High SDI quintile 7·0 15·1 54·8 23·2
of GBD. A decade ago, findings from a meta-analysis42 of High-middle SDI quintile 23·5 20·3 27·1 29·2
COPD prevalence studies showed large heterogeneity Middle SDI quintile 32·5 21·3 17·5 28·7
depending on case definitions based on spirometry, Low-middle SDI quintile 37·0 26·3 12·7 24·0
physician diagnosis, symptoms, and radiology. No Low SDI quintile 40·4 17·8 7·9 34·0
attempt was made to pool estimates between different 
COPD=chronic obstructive pulmonary disease. DALYs=disability-adjusted life years. SDI=Socio-demographic Index.
study methods and diagnostic thresholds. The various 
initiatives (BOLD, PLATINO, EPISCAN, and PREPOCOL) Table 3: Proportional contribution of behavioural and environmental and occupational risks for COPD in 
for population-representative COPD spirometry surveys DALYs per 100 000 people in locations grouped by SDI quintiles, 2015
have been combined for pooled analyses to compare 
prevalence estimates between sites, but estimation of been treated for asthma or currently using asthma 
global prevalence has not been attempted.16 medication) were pooled. This showed a difference of 
An estimate of 300 million prevalence cases of asthma around two times between higher estimates for wheeze 
was made in 2004 as part of the Global Initiative for compared with a reported physician diagnosis, whereas 
Asthma (GINA) based on the International Study of prevalence of clinical asthma was only marginally higher 
Asthma and Allergies in Childhood (ISAAC) and European than for physician diagnosis. However, the pooling method 
Community Respiratory Health Survey (ECRHS) estimates in this study was not explained.44 ISAAC Phase Three was 
of wheezing prevalence and an arbitrary 50% reduction of completed in 233 centres in 97 countries (80% low-income 
these estimates for clinical asthma.43 This estimate is a and middle-income countries), and ISAAC repeated 
lower than our estimate of 329 million prevalent cases surveys of prevalence of asthma have been done in 
in 2000, and 327 million cases in 2005, and our current 106 centres in 56 countries to date. We have made use of all 
2015 estimate of 358 million cases. A comparison between publicly available survey results from ISAAC and ECRHS.
the ISAAC study estimates in children and the ECRHS We estimated the highest age-standardised DALY rates 
study in adults showed a high correlation between due to COPD in 2015 in Papua New Guinea, India, 
childhood and adult prevalence estimates within countries, Lesotho, and Nepal. Findings from three verbal autopsy 
but large variation between countries.18 In an analysis of studies in the 1980s in Papua New Guinea showed high 
the World Health Surveys done in the early 2000s in chronic respiratory disease mortality. We decided to retain 
70 countries,44 estimates of the prevalence of wheeze, these three studies as they provide the only data on causes 
reported physician diagnosis, and clinical asthma (based of death for this country; there were no reasons to exclude 
on questions of a physician diagnosis and ever having them based on an assessment of their quality. The high 
www.thelancet.com/respiratory   Published online August 16, 2017   http://dx.doi.org/10.1016/S2213-2600(17)30293-X 11
DALY rate per 100 000 people
Articles
rates of mortality and morbidity in Lesotho and Nepal air pollution will be re-evaluated as risk factors for 
were based on predictive covariates as we do not have asthma.55,56
primary data for COPD from these two countries. The Other newly established individual risk factors of COPD, 
high rates in India were driven by mortality data sources such as low level of physical activity, could also have 
and two small spirometry studies in Pune and Mumbai.45 contributed to the unexplained COPD burden, as it has 
Our knowledge of the natural history of COPD and been related both to an increased risk of COPD among 
asthma is extensive yet incomplete. For asthma, over smokers57 and to a higher risk of COPD mortality.58
100 cohorts focusing on asthma and allergy have been We estimate only a small proportion of asthma burden 
initiated worldwide over the past 30 years.46 These long- due to risks quantified in GBD: 10·1% from occupational 
term birth cohort studies are essential to understanding asthmagens and 7·8% due to smoking.59,60 Evidence from 
the life course and childhood predictors of asthma and long-term observational studies and birth cohorts have 
allergy and the complex interplay between genes and rendered three hypotheses on other causes and triggers of 
the environment (including lifestyle and socioeconomic asthma, namely the hygiene, westernisation, and obesity 
determinants). However, information to quantify or sedentarism hypotheses. Comparative studies61,62 of 
population-level exposure to allergens in a comparable rural and urban populations gave rise to the hygiene theory 
manner is incomplete, and it has therefore not been that exposure to infections in early childhood explains 
possible to add it as a risk in GBD. Such natural history the lower prevalence of asthma in rural areas. The 
evidence is mostly missing for COPD.47 second theory is that socioeconomic development or 
The contribution of modifiable risk factors to COPD is westernisation predisposes to the development of asthma, 
large, yet much less for asthma. There are preventive but it is not clear which pathways other than those 
interventions to reduce exposure to smoking, second-hand described in the hygiene theory have a role.63 Obesity has 
smoke, air pollution, biomass for cooking or heating, been linked to a higher prevalence of asthma in children64 
occupational exposures, or any combination of these and an increased risk of developing new asthma in adults.65
factors. Additionally, other risk factors have been identified As concluded by Fuchs and colleagues in their 
such as parental or sibling history of asthma and atopy, low 2017 Review,66 we need to better understand underlying 
birthweight, lower respiratory infections in childhood, mechanisms of associations of asthma onset or remission 
education, day care, pet ownership, and other exposures, with risk and protective factors, and future asthma 
among others suggested. However, we are still far from research should integrate both paediatric and adult 
eliminating these as major contributors to the burden populations and longitudinal studies.
of COPD. Smoking is the largest contributor to the The general limitations of GBD studies have been 
COPD burden in countries at the higher end of the reported elsewhere and apply to estimates of obstructive 
SDI (69·4% of COPD burden in high-SDI quintile airways disease as well,1,2,23 and there are a number of 
countries), whereas the proportion of COPD explained by limitations specific to COPD and asthma. The first 
environmental exposures is highest in countries with low concerns the poor consensus on a case definition of 
SDI (58·1% of COPD burden in low-SDI quintile COPD. There was a difference of more than two times in 
countries). Given the importance of smoking as a risk YLDs between the GBD 2013 study, which used LLN as its 
factor of disease, monitoring national and international case definition, and the current study’s YLD results based 
trends and projections in smoking remains paramount for on the fixed ratio of the GOLD definition. As the survey 
worldwide health surveillance.48,49 Smoking prevalence has data on GOLD class distributions is largely based on a 
decreased in men and women since 1990 worldwide, but fixed ratio estimate of overall prevalence, we advocate that 
progress in tobacco control has not been universal.50 for GBD estimation purposes, use of the fixed ratio is 
Globally, exposure to household air pollution from solid preferable. Additionally, defining the cutoff value for LLN 
fuels has decreased since 1990, but exposure to ambient air as the fifth percentile in a healthy reference population 
pollution has increased since 1990.23 makes the arbitrary assumption that prevalence cannot be 
However, a considerable proportion of COPD remains lower than 5%.
unexplained and cannot be attributed to the risks Second, to make use of all spirometry surveys that 
quantified in GBD. In the next iteration of GBD, we plan reported COPD prevalence using different thresholds, 
to quantify a past history of pulmonary tuberculosis as an and with or without bronchodilation, we had to adjust data 
additional risk factor for COPD as there is growing sources to the expected values of our reference case 
evidence for a causal relationship.51,52 We did not estimate definition. We used a limited set of surveys that presented 
air pollution as a risk for asthma, because we had data with the reference and alternative case definitions. 
insufficient evidence for an increased risk of disease. Each of those adjustments showed a strong age pattern, 
Repeated lower respiratory infections in childhood and which we tried to capture with regression methods. Such 
the long-term effects of asthma have been reported as adjustments add uncertainty, which would be avoided if 
other explanatory factors, but it is not quite apparent how estimates were all reported in a standard manner.
estimation of these effects can be operationalised in Third, because no physiological measurement is 
GBD.53,54 In future GBD iterations, ambient and household considered a gold standard, diagnosis of asthma relies on 
12 www.thelancet.com/respiratory   Published online August 16, 2017   http://dx.doi.org/10.1016/S2213-2600(17)30293-X
Articles
clinical assessment and self-report, a physician diagnosis, excess mortality rates based on a country’s income. This 
or both. Thus, measurement of asthma prevalence can be addition had little effect on the estimates in high-income 
affected by the limitations of recall bias, access to health countries, but increased estimates in low-income and 
services, and different interpretations of survey questions middle-income countries considerably. We believe that 
inherent in self-reported health measurements.67 Access this approach is an improvement in the estimation strategy 
to clinical care is a challenge in low-income and middle- and that future estimates of global prevalence of asthma 
income countries, as well as in rural settings, therefore, will be more consistent with the GBD 2015 finding.
defining asthma by reported symptoms and a doctor COPD and asthma are important contributors to the 
diagnosis could lead to an underestimate of asthma burden of non-communicable disease. Although much of 
prevalence. We refined our assessment of asthma studies the burden is either preventable or treatable with affordable 
in GBD 2015 to better deal with nuances in self-report interventions, these diseases have received less attention 
measures and adjusted for three instead of just one non- than other prominent non-communicable diseases like 
reference case definition. However, we cannot exclude cardiovascular disease, cancer, or diabetes. Up-to-date 
that some residual measurement bias has affected the population information on these common diseases is key 
comparability of estimates between countries. to policy decision making to improve access to and quality 
Fourth, mapping severity in MEPS to COPD GOLD class of existing intervention strategies. We call for greater 
prevalence assumes this relationship can be generalised standardisation in data collection with regard to case 
from the USA to the rest of the world. However, we only definition and severity distributions of all non-
use the relationship between epidemiological data on communicable diseases in general, and of asthma and 
GOLD class distributions in the USA to the severity pattern COPD in particular. More, and updated, population 
of cases with COPD in MEPS as reflected in respondents’ measurements of COPD and asthma are needed to better 
answers to the SF-12. Our epidemiological models of the quantify the size of the problem, to benchmark with other 
GOLD class distribution allow us to differentiate severity chronic conditions associated with smoking and ageing, 
by age, sex, year, and location in as far as the sparse and with any other environmental and air pollution 
information on GOLD class prevalence allows. Our exposures.
measurements of COPD severity would benefit from GBD 2015 Chronic Respiratory Disease Collaborators
increased use standardised measures in surveys that Joan B Soriano, Amanuel Alemu Abajobir, Kalkidan Hassen Abate, 
reflect the lay descriptions on which the GBD disability Semaw Ferede Abera, Anurag Agrawal, Muktar Beshir Ahmed, 
weights are based or that use a generic quality-of-life Amani Nidhal Aichour, Ibtihel Aichour, Miloud Taki Eddine Aichour, Khurshid Alam, Noore Alam, Juma M Alkaabi, Fatma Al-Maskari, 
measure like SF-12. Nelson Alvis-Guzman, Alemayehu Amberbir, Yaw Ampem Amoako, 
Fifth, our measurement of asthma severity completely Mustafa Geleto Ansha, Josep M Antó, Hamid Asayesh, 
relies on MEPS data and therefore, unlike COPD, assumes Tesfay Mehari Atey, Euripide Frinel G Arthur Avokpaho, 
the same distribution for every location, year, age, and sex. Aleksandra Barac, Sanjay Basu, Neeraj Bedi, Isabela M Bensenor, Adugnaw Berhane, Addisu Shunu Beyene, Zulfiqar A Bhutta, 
This assumption is highly unlikely as treatments have a Stan Biryukov, Dube Jara Boneya, Michael Brauer, David O Carpenter, 
large effect on severity of asthma. For this reason, we Daniel Casey, Devasahayam Jesudas Christopher, Lalit Dandona, 
found no relationship between SDI and YLDs from Rakhi Dandona, Samath D Dharmaratne, Huyen Phuc Do, 
asthma, counter to the expectation that increased access to Florian Fischer, Ayele Geleto, Aloke Gopal Ghoshal, Richard F Gillum, Ibrahim Abdelmageem Mohamed Ginawi, Vipin Gupta, Simon I Hay, 
treatment, particularly steroid inhalers, would impact Mohammad T Hedayati, Nobuyuki Horita, H Dean Hosgood, 
asthma severity and hence disability. Researchers are Mihajlo (Michael) B Jakovljevic, Spencer Lewis James, Jost B Jonas, 
encouraged, in future surveys, to collect information on Amir Kasaeian, Yousef Saleh Khader, Ibrahim A Khalil, 
Ejaz Ahmad Khan, Young-Ho Khang, Jagdish Khubchandani, 
the proportion of cases that would fall into the lay Luke D Knibbs, Soewarta Kosen, Parvaiz A Koul, G Anil Kumar, 
description categories for controlled, partially controlled, Cheru Tesema Leshargie, Xiaofeng Liang, Hassan Magdy Abd El Razek, 
and uncontrolled asthma. Azeem Majeed, Deborah Carvalho Malta, Treh Manhertz, Neal Marquez, 
Sixth, for many countries in the world that do not have Alem Mehari, George A Mensah, Ted R Miller, 
Karzan Abdulmuhsin Mohammad, Kedir Endris Mohammed, 
functional vital registration systems, we had to rely on Shafiu Mohammed, Ali H Mokdad, Mohsen Naghavi, 
death estimates of all chronic respiratory diseases from Cuong Tat Nguyen, Grant Nguyen, Quyen Le Nguyen, 
verbal autopsy studies because these studies cannot Trang Huyen Nguyen, Dina Nur Anggraini Ningrum, 
distinguish between asthma, COPD, or other chronic Vuong Minh Nong, Jennifer Ifeoma Obi, Yewande E Odeyemi, 
Felix Akpojene Ogbo, Eyal Oren, Mahesh PA, Eun-Kee Park, 
respiratory diseases. Initiatives to strengthen vital George C Patton, Katherine Paulson, Mostafa Qorbani, 
registration systems are key to improving population Reginald Quansah, Anwar Rafay, Mohammad Hifz Ur Rahman, 
health measurement because verbal autopsy can only Rajesh Kumar Rai, Salman Rawaf, Nik Reinig, Saeid Safiri, 
identify a restricted set of diseases.68 Rodrigo Sarmiento-Suarez, Benn Sartorius, Miloje Savic, 
Monika Sawhney, Mika Shigematsu, Mari Smith, Fentaw Tadese, 
Seventh, the estimate of the global prevalence of asthma George D Thurston, Roman Topor-Madry, Bach Xuan Tran, 
changed from 242 million in 2013 based on GBD 2013 to Kingsley Nnanna Ukwaja, Job F M van Boven, 
358 million in 2015 for GBD 2015. In GBD 2015, cause- Vasiliy Victorovich Vlassov, Stein Emil Vollset, Xia Wan, 
specific mortality rates were added to the DisMod-MR 2.1 Andrea Werdecker, Sarah Wulf Hanson, Yuichiro Yano, 
Hassen Hamid Yimam, Naohiro Yonemoto, Chuanhua Yu, 
model with income per capita as a covariate to differentiate 
www.thelancet.com/respiratory   Published online August 16, 2017   http://dx.doi.org/10.1016/S2213-2600(17)30293-X 13
Articles
Zoubida Zaidi, Maysaa El Sayed Zaki, Christopher J L Murray, and University, Washington, DC, USA (R F Gillum MD); College of 
Theo Vos. Medicine, University of Hail, Hail, Saudi Arabia (I A Ginawi MD); 
Affiliations Department of Anthropology, University of Delhi, Delhi, India 
(V Gupta PhD); Oxford Big Data Institute, Li Ka Shing Centre for Health 
Instituto de Investigación Hospital Universitario de la Princesa (IISP), 
Information and Discovery, University of Oxford, Oxford, UK 
Madrid, Spain (Prof J B Soriano MD); Universidad Autónoma de 
(Prof S I Hay DSc); Department of Medical Mycology and Parasitology, 
Madrid, Madrid, Spain (Prof J B Soriano MD); School of Public Health, 
School of Medicine, Mazandaran University of Medical Sciences, Sari, 
University of Queensland, Brisbane, QLD, Australia (A A Abajobir MPH, 
Iran (Prof M T Hedayati PhD); Department of Pulmonology, Yokohama 
L D Knibbs PhD); Department of Epidemiology, College of Health 
City University Graduate School of Medicine, Yokohama, Japan 
Sciences (M B Ahmed MPH), Jimma University, Jimma, Ethiopia 
(N Horita MD); Albert Einstein College of Medicine, Bronx, NY, USA 
(K H Abate MS); School of Public Health, College of Health Sciences 
(Prof H D Hosgood PhD); Faculty of Medical Sciences, University of 
(S F Abera MSc), College of Health Sciences (K E Mohammed MPH), 
Kragujevac, Kragujevac, Serbia (Prof M B Jakovljevic PhD); Denver 
Mekelle University, Mekelle, Ethiopia (T M Atey MS); Food Security and 
Health/University of Colorado, Denver, CO, USA (S L James MD); 
Institute for Biological Chemistry and Nutrition, University of 
Department of Ophthalmology, Medical Faculty Mannheim, Ruprecht-
Hohenheim, Stuttgart, Germany (S F Abera MSc); CSIR—Institute of 
Karls-University Heidelberg, Mannheim, Germany (Prof J B Jonas MD); 
Genomics and Integrative Biology, Delhi, India (A Agrawal PhD); 
Endocrinology and Metabolism Population Sciences Institute 
Department of Internal Medicine, Baylor College of Medicine, Houston, 
(A Kasaeian PhD), Hematology-Oncology and Stem Cell Transplantation 
TX, USA (A Agrawal PhD); University Ferhat Abbas of Setif, Setif, 
Research Center, Tehran University of Medical Sciences, Tehran, Iran 
Algeria (A N Aichour BS); National Institute of Nursing Education, Setif, 
(A Kasaeian PhD); Department of Community Medicine, Public Health 
Algeria (I Aichour MS); High National School of Veterinary Medicine, 
and Family Medicine, Jordan University of Science and Technology, 
Algiers, Algeria (M T Aichour MD); Murdoch Childrens Research 
Irbid, Jordan (Prof Y S Khader ScD); Health Services Academy, 
Institute, The University of Melbourne, Parkville, VIC, Australia 
Islamabad, Pakistan (E A Khan MD); Department of Health Policy and 
(K Alam PhD); The University of Sydney, Sydney, NSW, Australia 
Management, Seoul National University College of Medicine, Seoul, 
(K Alam PhD); Department of Health, Queensland, Brisbane, QLD, 
South Korea (Prof Y Khang MD); Institute of Health Policy and 
Australia (N Alam MAppEpid); College of Medicine and Health Sciences, 
Management, Seoul National University Medical Center, Seoul, South 
United Arab Emirates University, Al Ain, United Arab Emirates 
Korea (Prof Y Khang MD); Department of Nutrition and Health Science, 
(J M Alkaabi MD, Prof F Al-Maskari PhD); Universidad de Cartagena, 
Ball State University, Muncie, IN, USA (J Khubchandani PhD); 
Cartagena de Indias, Colombia (Prof N Alvis-Guzman PhD); Dignitas 
Center for Community Empowerment, Health Policy and Humanities, 
International, Zomba, Malawi (A Amberbir PhD); Department of 
National Institute of Health Research & Development, Jakarta, Indonesia 
Medicine, Komfo Anokye Teaching Hospital, Kumasi, Ghana 
(S Kosen MD); Sher-i-Kashmir Institute of Medical Sciences, Srinagar, 
(Y A Amoako MD); West Hararghe Zonal Health Department, Chiro, 
India (Prof P A Koul MD); Chinese Center for Disease Control and 
Ethiopia (M G Ansha MPH); Barcelona Institute for Global Health 
Prevention, Beijing, China (Prof X Liang MD); Mansoura Faculty of 
(IS Global), Barcelona, Spain (Prof J M Antó MD); Department of 
Medicine, Mansoura, Egypt (H Magdy Abd El Razek MBBCH); 
Medical Emergency, School of Paramedic, Qom University of Medical 
Department of Primary Care & Public Health (Prof A Majeed MD), 
Sciences, Qom, Iran (H Asayesh MS); Institut de Recherche Clinique du 
Imperial College London, London, UK (Prof S Rawaf MD); Universidade 
Bénin (IRCB), Cotonou, Benin (E F G A Avokpaho MPH); Laboratoire 
Federal de Minas Gerais, Belo Horizonte, Brazil (Prof D C Malta PhD); 
d’Etudes et de Recherche-Action en Santé (LERAS Afrique), Parakou, 
Center for Translation Research and Implementation Science, National 
Benin (E F G A Avokpaho MPH); Faculty of Medicine, University of 
Heart, Lung, and Blood Institute, National Institutes of Health, 
Belgrade, Belgrade, Serbia (A Barac PhD); Stanford University, Stanford, 
Bethesda, MD, USA (G A Mensah MD); Pacific Institute for Research & 
CA, USA (S Basu PhD); College of Public Health and Tropical Medicine, 
Evaluation, Calverton, MD, USA (T R Miller PhD); Centre for Population 
Jazan, Saudi Arabia (N Bedi MD); University of São Paulo, São Paulo, 
Health, Curtin University, Perth, WA, Australia (T R Miller PhD); 
Brazil (I M Bensenor PhD); College of Health Sciences, Debre Berhan 
University of Salahaddin, Erbil, Iraq (K A Mohammad PhD); ISHIK 
University, Debre Berhan, Ethiopia (A Berhane PhD); College of Health 
University, Erbil, Iraq (K A Mohammad PhD); Health Systems and 
and Medical Sciences (A S Beyene MPH), Haramaya University, Harar, 
Policy Research Unit, Ahmadu Bello University, Zaria, Nigeria 
Ethiopia (A Geleto MPH); Centre of Excellence in Women and Child 
(S Mohammed PhD); Institute of Public Health, Heidelberg University, 
Health, Aga Khan University, Karachi, Pakistan (Z A Bhutta PhD); 
Heidelberg, Germany (S Mohammed PhD); Department of Public 
Centre for Global Child Health, The Hospital for Sick Children, Toronto, 
Health, Semarang State University, Semarang City, Indonesia 
ON, Canada (Z A Bhutta PhD); Institute for Health Metrics and 
(D N A Ningrum MPH); Graduate Institute of Biomedical Informatics, 
Evaluation (S Biryukov BS, Prof M Brauer ScD, D Casey MPH, 
College of Medical Science and Technology, Taipei Medical University, 
Prof L Dandona MD, Prof R Dandona PhD, Prof S I Hay DSc, 
Taipei City, Taiwan (D N A Ningrum MPH); Pulmonary Medicine, 
I A Khalil MD, T Manhertz BA, N Marquez BA, Prof A H Mokdad PhD, 
Howard University Hospital, Washington, DC, USA (J I Obi MBBS, 
Prof M Naghavi PhD, G Nguyen MPH, K Paulson BS, N Reinig BS, 
Y E Odeyemi MBBS); Centre for Health Research, Western Sydney 
M Smith MPA, Prof S E Vollset DrPH, X Wan PhD, 
University, Sydney, NSW, Australia (F A Ogbo MPH); University of 
S Wulf Hanson MPH, Prof C J L Murray DPhil, Prof T Vos PhD), Center 
Arizona, Tucson, AZ, USA (Prof E Oren PhD); JSS Medical College, JSS 
for Health Trends and Forecasts, Institute for Health Metrics and 
University, Mysore, India (Prof M PA DNB); Department of Medical 
Evaluation (Prof M B Jakovljevic PhD), University of Washington, 
Humanities and Social Medicine, College of Medicine, Kosin University, 
Seattle, WA, USA; Department of Public Health (D J Boneya MPH), 
Busan, South Korea (E Park PhD); Murdoch Childrens Research 
Debre Markos University, Debre Markos, Ethiopia (C T Leshargie MPH); 
Institute, Department of Paediatrics, University of Melbourne, 
University of British Columbia, Vancouver, BC, Canada 
Melbourne, VIC, Australia (Prof G C Patton MD); Non-Communicable 
(Prof M Brauer ScD); University at Albany, Rensselaer, NY, USA 
Diseases Research Center, Alborz University of Medical Sciences, Karaj, 
(Prof D O Carpenter MD); Christian Medical College, Vellore, India 
Iran (M Qorbani PhD); University of Ghana, Accra, Ghana 
(Prof D J Christopher MD); Public Health Foundation of India, 
(R Quansah PhD); Noguchi Memorial Institute of Medical Research, 
Gurugram, India (Prof L Dandona MD, Prof R Dandona PhD, 
Accra, Ghana (R Quansah PhD); Contech International Health 
G A Kumar PhD); Department of Community Medicine, Faculty of 
Consultants, Lahore, Pakistan (A Rafay MS); Contech School of Public 
Medicine, University of Peradeniya, Peradeniya, Sri Lanka 
Health, Lahore, Pakistan (A Rafay MS); International Institute for 
(S D Dharmaratne MD); Institute for Global Health Innovations, 
Population Sciences, Mumbai, India (M H U Rahman MPhil); Society 
Duy Tan University, Da Nang, Vietnam (H P Do MSc, C T Nguyen MSc, 
for Health and Demographic Surveillance, Suri, India (R K Rai MPH); 
Q L Nguyen MD, T H Nguyen MSc, V M Nong MSc); School of Public 
Managerial Epidemiology Research Center, Department of Public 
Health, Bielefeld University, Bielefeld, Germany (F Fischer PhD); 
Health, School of Nursing and Midwifery, Maragheh University of 
University of Newcastle, Newcastle, NSW, Australia (A Geleto MPH); 
Medical Sciences, Maragheh, Iran (S Safiri PhD); Universidad Ciencias 
National Allergy Asthma Bronchitis Institute, Kolkata, India 
Aplicadas y Ambientales, Bogotá DC, Colombia 
(A G Ghoshal DNB); College of Medicine (A Mehari MD), Howard 
14 www.thelancet.com/respiratory   Published online August 16, 2017   http://dx.doi.org/10.1016/S2213-2600(17)30293-X
Articles
(R Sarmiento-Suarez MPH); Public Health Medicine, School of Nursing 9 BOLD-COPD International Research platform. 
and Public Health, University of KwaZulu-Natal, Durban, South Africa http://www.boldstudy.org/index.html (accessed Feb 7, 2017).
(Prof B Sartorius PhD); UKZN Gastrointestinal Cancer Research Centre, 10 Caballero A, Torres-Duque CA, Jaramillo C, et al. Prevalence of 
South African Medical Research Council, Durban, South Africa COPD in five Colombian cities situated at low, medium, and high 
(Prof B Sartorius PhD); Center for Disease Burden altitude (PREPOCOL study). Chest 2008; 133: 343–49.
(Prof S E Vollset DrPH), Norwegian Institute of Public Health, Oslo, 11 Menezes AMB, Victora CG, Perez-Padilla R, PLATINO Team. 
Norway (M Savic PhD); Department of Public Health, Marshall The Platino project: methodology of a multicenter prevalence 
University, Huntington, WV, USA (M Sawhney PhD); National Institute survey of chronic obstructive pulmonary disease in major 
of Infectious Diseases, Tokyo, Japan (M Shigematsu PhD); Latin American cities. BMC Med Res Methodol 2004; 4: 15.
Sandia National Laboratories, Albuquerque, NM, USA 12 Vollmer WM, Gíslason T, Burney P, et al. Comparison of spirometry 
(M Shigematsu PhD); Department of Public Health, Wollo University, criteria for the diagnosis of COPD: results from the BOLD study. 
Dessie, Ethiopia (F Tadese MPH); Nelson Institute of Environmental Eur Respir J 2009; 34: 588–97.
Medicine, School of Medicine, New York University, Tuxedo, NY, USA 13 Quinn CE, Soriano JB, Chen R, Zhou Y. Epidemiology of COPD. 
(Prof G D Thurston ScD); Institute of Public Health, Faculty of Health In: Quinn CE, ed. 100 questions & answers about chronic obstructive pulmonary disease (COPD). Burlington, MA: Jones & 
Sciences, Jagiellonian University Medical College, Kraków, Poland Bartlett Learning, 2005.
(R Topor-Madry PhD); Faculty of Health Sciences, Wroclaw Medical 14 Vogelmeier CF, Criner GJ, Martinez FJ, et al. Global strategy for the 
University, Wroclaw, Poland (R Topor-Madry PhD); Johns Hopkins diagnosis, management, and prevention of chronic obstructive lung 
University, Baltimore, MD, USA (B X Tran PhD); Hanoi Medical disease 2017 report: GOLD executive summary. Eur Respir J 2017; 
University, Hanoi, Vietnam (B X Tran PhD); Department of Internal 49: 1750214.
Medicine, Federal Teaching Hospital, Abakaliki, Nigeria 15 Quanjer PH, Stanojevic S, Cole TJ, et al. Multi-ethnic reference 
(K N Ukwaja MD); University of Groningen, Groningen, Netherlands values for spirometry for the 3–95-yr age range: the global lung 
(J F M van Boven PhD); National Research University Higher School of function 2012 equations. Eur Respir J 2012; 40: 1324–43.
Economics, Moscow, Russia (Prof V V Vlassov MD); Department of 16 Lamprecht B, Soriano JB, Studnicka M, et al. Determinants of 
Global Public Health and Primary Care, University of Bergen, Bergen, underdiagnosis of COPD in national and international surveys. 
Norway (Prof S E Vollset DrPH); Institute of Basic Medical Sciences, Chest 2015; 148: 971–85.
Chinese Academy of Medical Sciences, Beijing, China (X Wan PhD); 17 Soriano JB, Zielinski J, Price D. Screening for and early detection 
Competence Center Mortality-Follow-Up of the German National of chronic obstructive pulmonary disease. Lancet 2009; 
Cohort, Federal Institute for Population Research, Wiesbaden, Germany 374: 721–32.
(A Werdecker PhD); Department of Preventive Medicine, Northwestern 18 Pearce N, Sunyer J, Cheng S, et al. Comparison of asthma 
University, Chicago, IL, USA (Y Yano MD); Mizan Tepi University, prevalence in the ISAAC and the ECRHS. ISAAC Steering 
Mizan Teferi, Ethiopia (H H Yimam MPH); Department of Biostatistics, Committee and the European Community Respiratory Health 
School of Public Health, Kyoto University, Kyoto, Japan Survey. International Study of Asthma and Allergies in Childhood. 
(N Yonemoto MPH); Department of Epidemiology and Biostatistics, Eur Respir J 2000; 16: 420–26.
School of Public Health (Prof C Yu PhD), Global Health Institute 19 Toelle BG, Peat JK, Salome CM, Mellis CM, Woolcock AJ. Toward a 
(Prof C Yu PhD), Wuhan University, Wuhan, China; University Hospital definition of asthma for epidemiology. Am Rev Respir Dis 1992; 
of Setif, Setif, Algeria (Prof Z Zaidi DSc); and Faculty of Medicine, 146: 633–37.
Mansoura University, Mansoura, Egypt (Prof M E Zaki PhD). 20 Riiser A, Hovland V, Carlsen K-H, Mowinckel P, 
Lødrup Carlsen KC. Does bronchial hyperresponsiveness in 
Contributors childhood predict active asthma in adolescence? 
TV and JBS prepared the first draft. CJLM conceived the study and Am J Respir Crit Care Med 2012; 186: 493–500.
provided overall guidance. All other authors provided data, developed 21 Pekkanen J, Pearce N. Defining asthma in epidemiological studies. 
models, reviewed results, initiated modelling infrastructure, or reviewed Eur Respir J 1999; 14: 951–57.
and contributed to the report. 22 Pattemore PK, Asher MI, Harrison AC, Mitchell EA, Rea HH, 
Stewart AW. The interrelationship among bronchial 
Declaration of interests hyperresponsiveness, the diagnosis of asthma, and asthma 
We declare no competing interests. symptoms. Am Rev Respir Dis 1990; 142: 549–54.
References 23 GBD 2015 Risk Factors Collaborators. Global, regional, and national 
1 GBD 2015 Mortality and Causes of Death Collaborators. comparative risk assessment of 79 behavioural, environmental and 
Global, regional, and national life expectancy, all-cause mortality, occupational, and metabolic risks or clusters of risks, 1990–2015: 
and cause-specific mortality for 249 causes of death, 1980–2015: a systematic analysis for the Global Burden of Disease Study 2015. 
a systematic analysis for the Global Burden of Disease Study 2015. Lancet 2016; 388: 1659–724.
Lancet 2016; 388: 1459–544. 24 Tilert T, Dillon C, Paulose-Ram R, Hnizdo E, Doney B. 
2 GBD 2015 Disease and Injury Incidence and Prevalence Estimating the US prevalence of chronic obstructive pulmonary 
Collaborators. Global, regional, and national incidence, prevalence, disease using pre- and post-bronchodilator spirometry: the National 
and years lived with disability for 310 diseases and injuries, Health and Nutrition Examination Survey (NHANES) 2007–2010. 
1990–2015: a systematic analysis for the Global Burden of Disease Respir Res 2013; 14: 103.
Study 2015. Lancet 2016; 388: 1545–602. 25 Johannessen A, Omenaas ER, Bakke PS, Gulsvik A. Implications of 
3 Burney P, Jarvis D, Perez-Padilla R. The global burden of chronic reversibility testing on prevalence and risk factors for chronic 
respiratory disease in adults. Int J Tuberc Lung Dis 2015; obstructive pulmonary disease: a community study. Thorax 2005; 
19: 10–20. 60: 842–47.
4 Yáñez A, Cho S-H, Soriano JB, et al. Asthma in the elderly: 26 Pérez-Padilla R, Hallal PC, Vázquez-García JC, et al. Impact of 
what we know and what we have yet to know. World Allergy Organ J bronchodilator use on the prevalence of COPD in population-based 
2014; 7: 8. samples. COPD 2007; 4: 113–20.
5 Wurst KE, Rheault TR, Edwards L, Tal-Singer R, Agusti A, Vestbo J. 27 Shirtcliffe P, Weatherall M, Marsh S, et al. COPD prevalence in a 
A comparison of COPD patients with and without ACOS in the random population survey: a matter of definition. Eur Respir J 2007; 
ECLIPSE study. Eur Respir J 2016; 47: 1559–62. 30: 232–39.
6 de Marco R, Marcon A, Rossi A, et al. Asthma, COPD and overlap 28 Mohamed Hoesein FAA, Zanen P, Sachs APE, Verheij TJM, 
syndrome: a longitudinal study in young European adults. Lammers J-WJ, Broekhuizen BDL. Spirometric thresholds for 
Eur Respir J 2015; 46: 671–9. diagnosing COPD: 0·70 or LLN, pre- or post-dilator values? 
7 Barnes PJ. Asthma-COPD overlap. Chest 2016; 149: 7–8. COPD 2012; 9: 338–43.
8 Sin DD, Miravitlles M, Mannino DM, et al. What is asthma-COPD 29 Ko FWS, Woo J, Tam W, et al. Prevalence and risk factors of airflow 
overlap syndrome? Towards a consensus definition from a round obstruction in an elderly Chinese population. Eur Respir J 2008; 
table discussion. Eur Respir J 2016; 48: 664–73. 32: 1472–78.
www.thelancet.com/respiratory   Published online August 16, 2017   http://dx.doi.org/10.1016/S2213-2600(17)30293-X 15
Articles
30 Swanney MP, Ruppel G, Enright PL, et al. Using the lower limit of 49 Ng M, Freeman MK, Fleming TD, et al. Smoking prevalence and 
normal for the FEV1/FVC ratio reduces the misclassification of cigarette consumption in 187 countries, 1980–2012. JAMA 2014; 
airway obstruction. Thorax 2008; 63: 1046–51. 311: 183–92.
31 Lindberg A, Bjerg A, Bjerg-Bäcklund A, Rönmark E, Larsson L-G, 50 GBD 2015 Tobacco Collaborators. Smoking prevalence and 
Lundbäck B. Prevalence and underdiagnosis of COPD by disease attributable disease burden in 195 countries and territories, 
severity and the attributable fraction of smoking report from the 1990–2015: a systematic analysis from the Global Burden of Disease 
Obstructive Lung Disease in Northern Sweden Studies. Respir Med Study 2015. Lancet 2017; 389: 1885–906.
2006; 100: 264–72. 51 Buist AS, McBurnie MA, Vollmer WM, et al. International variation 
32 US Department of Health and Human Services. Medical expenditure in the prevalence of COPD (the BOLD Study): a population-based 
panel survey home. https://meps.ahrq.gov/mepsweb/ (accessed prevalence study. Lancet 2007; 370: 741–50.
Feb 9, 2017). 52 Lee C-H, Lee M-C, Lin H-H, et al. Pulmonary tuberculosis and 
33 Ware J, Kosinski M, Keller SD. A 12-Item Short-Form Health delay in anti-tuberculous treatment are important risk factors for 
Survey: construction of scales and preliminary tests of reliability chronic obstructive pulmonary disease. PLoS One 2012; 7: e37978.
and validity. Med Care 1996; 34: 220–33. 53 Shaheen SO, Barker DJ, Holgate ST. Do lower respiratory tract 
34 Brauer M, Freedman G, Frostad J, et al. Ambient air pollution infections in early childhood cause chronic obstructive pulmonary 
exposure estimation for the Global Burden of Disease 2013. disease? Am J Respir Crit Care Med 1995; 151: 1649–51.
Environ Sci Technol 2016; 50: 79–88. 54 Mannino DM, Buist AS. Global burden of COPD: risk factors, 
35 International Labour Organization. Key indicators of the labour prevalence, and future trends. Lancet 2007; 370: 765–73.
market. http://www.ilo.org/ilostat/faces/ilostat-home/home?_adf. 55 Guarnieri M, Balmes JR. Outdoor air pollution and asthma. 
ctrl-state=190tjch1f7_4&_afrLoop=1612006259110819#! (accessed Lancet 2014; 383: 1581–92.
Aug 1, 2017). 56 Khreis H, Kelly C, Tate J, Parslow R, Lucas K, Nieuwenhuijsen M. 
36 Jerrett M, Burnett RT, Pope CA, et al. Long-term ozone exposure Exposure to traffic-related air pollution and risk of development of 
and mortality. N Engl J Med 2009; 360: 1085–95. childhood asthma: a systematic review and meta-analysis. 
37 Salomon JA, Haagsma JA, Davis A, et al. Disability weights for the Environ Int 2017; 100: 1–31.
Global Burden of Disease 2013 study. Lancet Glob Health 2015; 57 Garcia-Aymerich J, Lange P, Benet M, Schnohr P, Antó JM. 
3: e712–23. Regular physical activity modifies smoking-related lung function 
38 Stevens GA, Alkema L, Black RE, et al. Guidelines for Accurate and decline and reduces risk of chronic obstructive pulmonary disease: 
Transparent Health Estimates Reporting: the GATHER statement. a population-based cohort study. Am J Respir Crit Care Med 2007; 
Lancet 2016; 388: e19–23. 175: 458–63.
39 Global Initiative for Asthma. 2017 GINA report: global strategy for 58 Garcia-Aymerich J, Lange P, Benet M, Schnohr P, Antó JM. 
asthma management and prevention. http://ginasthma.org/2017- Regular physical activity reduces hospital admission and mortality 
gina-report-global-strategy-for-asthma-management-and- in chronic obstructive pulmonary disease: a population based 
prevention/ (accessed March 23, 2017). cohort study. Thorax 2006; 61: 772–78.
40 Burney PGJ, Hooper RL. The use of ethnically specific norms for 59 Verlato G, Nguyen G, Marchetti P, et al. Smoking and new-onset 
ventilatory function in African-American and white populations. asthma in a prospective study on Italian adults. 
Int J Epidemiol 2012; 41: 782–90. Int Arch Allergy Immunol 2016; 170: 149–57.
41 Soriano JB, Parkes G. Remember elephants and icebergs... 60 Polosa R, Morjaria JB, Caponnetto P, et al. Persisting long term 
‘Your lung function should be here, but it is there!’ Eur Respir J benefits of smoking abstinence and reduction in asthmatic smokers 
2009; 33: 715–16. who have switched to electronic cigarettes. Discov Med 2016; 
42 Halbert RJ, Natoli JL, Gano A, Badamgarav E, Buist AS, 21: 99–108.
Mannino DM. Global burden of COPD: systematic review and 61 Wong GWK, Chow CM. Childhood asthma epidemiology: insights 
meta-analysis. Eur Respir J 2006; 28: 523–32. from comparative studies of rural and urban populations. 
43 Masoli M, Fabian D, Holt S, Beasley R, Global Initiative for Asthma Pediatr Pulmonol 2008; 43: 107–16.
(GINA) Program. The global burden of asthma: executive summary 62 Strachan DP. Family size, infection and atopy: the first decade of 
of the GINA Dissemination Committee report. Allergy 2004; the ‘hygiene hypothesis’. Thorax 2000; 55 (suppl 1): S2–10.
59: 469–78. 63 Douwes J, Pearce N. Asthma and the westernization ‘package’. 
44 To T, Stanojevic S, Moores G, et al. Global asthma prevalence in Int J Epidemiol 2002; 31: 1098–102.
adults: findings from the cross-sectional world health survey. 64 Guibas GV, Manios Y, Xepapadaki P, et al. The obesity-asthma link 
BMC Public Health 2012; 12: 204. in different ages and the role of body mass index in its 
45 Burney P, Jithoo A, Kato B, et al. Chronic obstructive pulmonary investigation: findings from the Genesis and Healthy Growth 
disease mortality and prevalence: the associations with smoking Studies. Allergy 2013; 68: 1298–305.
and poverty—a BOLD analysis. Thorax 2014; 69: 465–73. 65 Antó JM, Sunyer J, Basagaña X, et al. Risk factors of new-onset 
46 Bousquet J, Anto J, Sunyer J, et al. Pooling birth cohorts in allergy asthma in adults: a population-based international cohort study. 
and asthma: European Union-funded initiatives—a MeDALL, Allergy 2010; 65: 1021–30.
CHICOS, ENRIECO, and GA²LEN joint paper. Int Arch Allergy 66 Fuchs O, Bahmer T, Rabe KF, von Mutius E. Asthma transition 
Immunol 2013; 161: 1–10. from childhood into adulthood. Lancet Respir Med 2017; 5: 224–34.
47 Kohansal R, Soriano JB, Agusti A. Investigating the natural history 67 van Schayck CP, van der Heijden FMMA, van den Boom G, 
of lung function: facts, pitfalls, and opportunities. Chest 2009; Tirimanna P, van Herwaarden CLA. Underdiagnosis of asthma: 
135: 1330–41. is the doctor or the patient to blame? The DIMCA project. Thorax 
48 Bilano V, Gilmour S, Moffiet T, et al. Global trends and projections 2000; 55: 562–65.
for tobacco use, 1990–2025: an analysis of smoking indicators from 68 Bloomberg Philanthropies. Data for health. 
the WHO Comprehensive Information Systems for Tobacco https://www.bloomberg.org/program/public-health/data-health/ 
Control. Lancet 2015; 385: 966–76. (accessed March 23, 2017).
16 www.thelancet.com/respiratory   Published online August 16, 2017   http://dx.doi.org/10.1016/S2213-2600(17)30293-X
View publication stats