For the general population analysis, 3 databases KNHANES, KCHS and NHIS-NSC, were used. The data did not include any identifiable information about the individual patients. The KNHANES is a nationwide health and nutrition survey that has been maintained since 1998. This survey has collected various data including physical examination findings by physicians, clinical and laboratory tests, information from personal interviews, and related measurements.9 The KNHANES is comprised of 3 types of questionnaires that address health, nutrition and screening. This survey includes some items related to asthma such as questions about asthma diagnoses. To improve the accuracy of asthma diagnoses in this study, asthma was considered only when the study participant provided a positive response to the question: “Have you been diagnosed with asthma by your doctors?” To obtain appropriate estimates, we followed the guidelines for reporting sample weight and stratification as designated by the KCDC. On the basis of the information collected, the prevalence of asthma was analyzed from 1998 to 2013 and by age and sex. We obtained the data regarding age and sex distribution from the Population and Housing Census conducted by Statistics Korea in 2005 and adjusted the data.

The KCHS is another nationwide health survey conducted in 253 cities, provinces and local health centers. This survey also includes asthma-related questions that examine whether the subject experienced wheezing symptoms during the last year, whether they had been diagnosed with asthma by their physicians and received treatment, or whether they had experienced a worsening of their condition over the past year. From the KCHS data, the prevalence of asthma was defined as the experience rate of diagnosed asthma by a physician during the lifetime period of 2008–2013 and was standardized by sex and age. We also compared asthma prevalence among different regions in Korea.

The NHIS-NSC database is a cohort database created in 2002, excluding noncitizens and special purpose employees with an unidentifiable income level. It links an individual's social and economic variables (i.e., qualification data) with their medical history and health screening data. A sample of 1 million people (1,025,340; 2.2% of the total population in 2002) in the NHIS database followed participants for 11 years unless their eligibility was disqualified because of death or emigration. Systematic stratified random sampling with proportional allocation within each stratum was conducted using the individual's total annual medical expenses. Within each stratum, systematic sampling was conducted after sorting population data by the value of annual medical expenses and maintaining a sampling rate of 2.2%. During the follow-up period, the cohort was updated annually by adding a representative sample of newborns using the 2.2% sampling rate.10 The NHIS-NSC database can be longitudinally analyzed for 11 years (2002–2012). We defined eligible asthma patients for this study as those whose medical claims included International Classification of Disease, Tenth Revision, Clinical Modification (ICD-10-CM) diagnosis codes (J45–J46) for asthma during the identification period of January 1, 2002 through December 31, 2012. In addition, the prescription of 1 or more asthma medications and the performance of designated asthma-related tests were considered an operational definition of asthma (Supplementary Table 1). In other words, asthma was defined as a condition that meets an asthma diagnosis code (J45–J46) at the same time asthma-related tests were performed or medications were prescribed. The prevalence was determined by the number of patients with asthma divided by the total number of patients registered by the NHIS-NSC and was analyzed annually by age and sex. Incidence was defined as the number of patients with a new case of asthma diagnosed in a specific year, with the denominator as the number of patient-years of observation (calculated from the number of patients registered with practices and their length of registration). The incidence was analyzed for all ages, except children under 1 year of age.

Among these defined patients with asthma, an episode of acute exacerbation was defined as an event requiring an asthma-related hospital admission, emergency room visit or the use of prescribed systemic steroids in an outpatient-based treatment setting. We divided these patients by the total number of patients defined as having asthma to calculate the annual asthma exacerbation rate. In addition, all the patients with asthma were divided into 3 groups according to severity levels that were defined according to their medications: mild asthma, low-dose inhaled corticosteroid (ICS); moderate asthma, medium-dose ICS; and severe asthma, high-dose ICS. Each ICS dose was converted to the budesonide-equivalent dose and was then classified as low-, medium- or high-dose ICS according to the annual Global Initiative for Asthma (GINA) guidelines.11 This study was approved by the Ethics Committee of the National Institute of Health (NIH).

For COREA, a total of 1,832 patients with asthma were recruited by allergists or pulmonologists at 23 referral centers in diverse areas of the Republic of Korea.812 These patients had experienced dyspnea, cough, sputum production or wheezing for >3 months. All the patients showed positive airway hyperresponsiveness as indicated by a 20% fall in the forced expiration volume in 1 second (FEV1) with a methacholine dose of 16 mg/mL (PC₂₀) during a bronchial provocation test, or positive bronchodilator responsiveness (12% and 200 mL improvement in FEV1 after inhalation of 180 µg of albuterol). At enrollment, the patients' conditions were stable with regular medications. Patients with severe lung damage, bronchiectasis or a history of lung resection were excluded.11 Another independent cohort of patients was registered at the SCH (Bucheon, Korea), and their asthma diagnosis was based on GINA guidelines. All the patients were aged >18 years, and they were regularly followed up and treated with appropriate medications based on the GINA guidelines. Unlike the COREA cohort analysis, longitudinal data for 1 and 3 years were not obtained and analyzed. The design and protocol for this cohort study were approved by the Institutional Review Board of each center (2012-0234). Asthma exacerbation was defined as the presence of 1 or more of the following during the previous year: 1) emergency department visit, 2) unexpected outpatient visit, 3) inpatient or intensive care unit treatment, or 4) use of systemic corticosteroids (>3 consecutive days). According to this definition, we compared the differences between patients who experienced asthma exacerbations and stable patients with asthma without exacerbations, and we considered variables with significant differences to be risk factors.

To obtain unbiased estimates of asthma in the general population, we applied the KNHANES sampling weights. We followed the guidelines for reporting sample weight (sampling weights) and stratification designed by the KCDC. This information is also available on the KNHANES website (http://knhanes.cdc.go.kr). In addition, the weights were modified by adjusting for sex and age. Given that samples in the same areas exhibited high correlations, an analysis of prevalence was performed using the generalized linear mixed model method while considering the spatial correlation. We conducted logistic regression and univariate analyses to determine risk factors for asthma exacerbation. Next, we selected statistically significant variables satisfying P<0.05 in the univariate analysis. We then performed a multiple regression with these selected variables to determine the association of asthma exacerbation with each variable. Next, we extracted significant variables that correlated strongly with the deterioration of the patient's asthma and calculated the odds ratios. We further analyzed and compared the risk factors for asthma exacerbation between male and female patients. All the statistical analyses were performed using R software (R Project for Statistical Computing, Vienna, Austria; www.r-project.org). A P value of <0.05 was considered statistically significant.

According to the KNHANES database from 1998 to 2013, the annual asthma prevalence among individuals older than 19 years ranged from the lowest rate of 1.2% in 1998 to the highest rate of 3.1% in 2010. The prevalence of asthma was consistently higher among women than among men (Fig. 1 and Supplementary Fig. 1). In particular, the largest sex-based difference in prevalence occurred in 2013, with rates of 2.3% and 3.6% among male and female subjects, respectively (Table 1). An analysis by age revealed a significantly higher prevalence in the elderly group (≥65 years old) than that in the younger group (Table 1 and Supplementary Fig. 1A and B). In particular, the prevalence of asthma among female elderly patients (≥65 years old) in 2013 was 9.7%, or 3-fold higher than that in men (Table 1 and Supplementary Fig. 1C). The KCHS data showed an increasing trend in overall prevalence, and a regional analysis revealed that Chungnam Province had the highest prevalence of 3.0% in 2013. The annual regional trends of asthma prevalence varied, and the prevalence in Jeju Province increased continuously (Table 2). According to the NHIS-NSC database from 2002 to 2012, the prevalence of asthma also increased gradually from 4.5% to 6.2%. Similar to the results of the KNHANES analysis, the prevalence of asthma was significantly higher in women than in men in each year (Table 3 and Supplementary Fig 2A). In addition, the prevalence of asthma was remarkably higher in the elderly group (≥65 years old) than in any other age group of adult patients with asthma (Table 3, Supplementary Fig. 2A and B).

According to the NHIS-NSC database analysis, the overall incidence of asthma per year has exceeded 20 per 1,000 person-years since 2009. Interestingly, the asthma incidence was always higher among women than among men (Table 4 and Fig. 2).

Approximately 80% of the patients had moderate asthma; accordingly, this condition was most prevalent. Approximately 11% of the patients had severe asthma (Supplementary Table 2 and Fig. 3A). The annual asthma exacerbation rates ranged between 25% and 35%, with a decreasing trend over time until 2009, followed by a plateau since 2010 (Supplementary Table 2 and Fig. 3B).

Among COREA's 1,832 patients with asthma, 213 (11.6%) experienced 1 or more exacerbations during the year after enrollment. A univariate analysis identified sex, baseline pulmonary function, treatment duration, and lung function variability or changes in Asthma Control Test (ACT) scores over 1 and 3 years as significant associated factors (Table 5). A multiple regression analysis identified 6 risk factors for acute asthma exacerbation. These factors included sex, baseline pulmonary function (FEV1, FVC, FEV1/FVC), treatment duration, lung function variability over a 1 year period, changes in ACT score over a 1-year period and changes in ACT score over a 3-year period (Table 5). In other words, female sex, a lower pulmonary function at baseline, longer treatment duration, or wide fluctuations in the FEV1 or ACT during the treatment course were identified as risk factors for an acute asthma exacerbation within 1 year. A lower baseline lung function and longer treatment duration were identified as important risk factors for asthma exacerbations, especially in women (Fig. 4A and B). The results of analysis of the SCH cohort were similar to that of COREA. Of the total 2,195 individuals with asthma, 283 (12.9%) experienced exacerbation during the year after enrollment. Lower baseline lung functions (FEV1 and FVC) were also identified as a risk factor. Female sex showed a higher risk of acute exacerbation than male sex, but it was not statistically significant (Supplementary Table 3).