KNHANES V (2011-2012) was a nationwide non-institutionalized civilian population-based survey in South Korea. The protocol was approved by the Institutional Review Board of the Korea Centers for Disease Control and Prevention (2011-02CON-06-C, 2012-01EXP-01-2C), and all individuals signed informed consent forms. Stratified multistage cluster random sampling designed by the rolling-survey sampling method was used in the survey.17 Therefore, the rolling samples collected are the probabilities that represent the Korean general population, which are independent of and homogeneous to each other. In 2011-2012, weighted 15,308 individuals participated in the otorhinolaryngologic surveys. Among the 15,308 participants, we excluded the following participants: those younger than 19 years (n=3,717) and those who did not completely answer questions about rhinitis symptoms, mental health, sociodemographic factors, and comorbid diseases (n=1,705). Thus, the final total number of weighted participants (n=11,154; unweighted n=23,893,740) was eligible for our study (Fig. 1).

The questionnaires were administered face to face by the physicians or interviewers. Age was clustered into 7 subgroups by 10-year periods. Body mass index (BMI) was calculated by dividing body weight in kilograms by the square of height in meters. BMI and sleep time were compared by using mean values. Current cigarette smokers were categorized as "yes" for participants who had smoked more than 100 cigarettes during their lifetime and current smoking state. The amount of pure alcohol consumed (in grams per day) was estimated by the average quantity of alcoholic beverages consumed and the frequency of alcohol consumption. Participants who drank more than 30 g/day were categorized as heavy drinkers.18 Participants who commit to moderate physical activity at least 5 days per week for more than 30 minutes each day and/or strenuous physical activity for more than 20 minutes per session at least 3 days per week were classified as the regular exercise group.19 Residency was categorized into urban and rural areas according to the official address of the subjects. The category of 'low income' corresponded to the lowest quartile of annual household income. The education level of the subject was classified as high if the participant finished education beyond middle school (higher than the ninth grade). Marital status was divided into single or married.

Since prevalence based on symptoms alone may represent an overestimation,20 the self-report rhinitis measure used in our study can be considered to be physician-diagnosed AR. The following question was used to assess physician-diagnosed AR for each participant: "Have you been diagnosed with AR by a doctor?"19 Physician-diagnosed depression was also determined by using similar questions: "Have you been diagnosed with depression by a doctor?" Stress levels assessed by the questionnaire were categorized into low (none or small amount) or high (some or extreme). A separate set of questionnaires assessed participants for the presence of depressive symptoms and suicidal thoughts. We asked about depressive symptoms: "During the past year, have you felt feelings of sadness or hopelessness that persisted for at least 2 weeks and that disrupted your social life?" (yes/no). For suicidal thoughts, participants were asked, "During the past year, have you ever felt that you were willing to die?" (yes/no). For psychological consultation, "During the past year, have you ever received any consultation through interview, telephone, or online because of psychological problems" (yes/no). We also used the EuroQol 5-Dimension (EQ-5D) that has been widely used for the measurement of general health-related quality of life. The EQ-5D records the level of self-reported problems evaluated in 5 dimensions (i.e., mobility, self-care, usual activities, pain/discomfort, and anxiety/depression). For each dimension, participants chose 1 of the 3 levels of severity (1, no problem; 2, some/moderate problems; 3, severe/extreme problems). We used the anxiety/depression dimension of the EQ-5D to evaluate mental health.

Statistical analyses were performed by using Statistical Analysis Software (SAS) version 9.3 (SAS Institute, Cary, NC, USA). To calculate unbiased national estimates representing the general Korean population, we used KNHANES sample weights accounting for the complex sampling design to each participant.21 We measured the mean±SE or the proportion (SE). For the differences between general characteristics of AR and healthy participants, continuous variables were tested with the t test using the SURVEYREG procedure, while categorical variables were tested with the Rao-Scott chi-square test using the SURVEYFREQ procedure. The weighted prevalence of AR was calculated in the Korean population after adjustment for strata and after clustering at the primary sampling unit level. Using multiple logistic regression analysis with the SURVEYLOGISTIC procedure, we tested the correlation between AR and mental health after adjustment for confounding factors. Odds ratios (ORs) and 95% confidence intervals (CIs) were estimated after adjustment for potential confounders. Adjusted confounding factors were age, sex, BMI, smoking status, alcohol use status, and exercise status (model 1); age, sex, BMI, smoking status, alcohol use status, exercise status, region of residence, income, education, and marital status (model 2); and age, sex, BMI, smoking status, alcohol use status, exercise status, region of residence, income, education, marital status, asthma, and atopic dermatitis (model 3). A P value of <0.05 was considered statistically significant.

A total of weighted 11,154 participants (unweighted n= 23,893,740) were included in the study, and the general characteristics of this study population are presented in Table 1. Of these 11,154 participants, 1,467 (unweighted n=3,518,737) were diagnosed with AR. In univariate analysis, the significantly higher AR prevalence was found in demographic characteristics, socioeconomic status, and comorbid diseases, including age (P<0.0001), sex (P<0.001), BMI (P<0.0001), sleep (P<0.0001), current cigarette smoker (P=0.00001), urban residency (P=0.0144), high education level (P<0.0001), low income (P<0.0001), married state (P<0.0001), asthma (P<0.0001), and atopic dermatitis (P<0.0001). Thus, demographic characteristics, socioeconomic status, and comorbid diseases had the possibility of acting as potential confounders.

Among components explaining mental health, AR prevalence was high only in stress component (P<0.001). This result could be explained by the age factor. Fig. 2 shows the prevalence of AR and mental health characteristics stratified by age among the KNHANES (2011-2012) population aged ≥19 years. As age increased, AR prevalence decreased (P for trend<0.0001), while mental health components except stress component showed an increasing trend. Therefore, univariate analysis resulted in a weak relationship between AR and mental health without adjusting for potential confounders, such as age.

To assess the relationship between AR and 6 components for evaluating mental health, the prevalence of AR was analyzed depending on numbers of components (Fig. 3). In categories with components, the prevalences of AR were 13.8% (0.6), 15.8% (1.0), 16.6% (1.6), and 19.8% (4.5) (P for trend=0.0048).

Among 1,467 AR participants, 1,209 were further tested for subanalysis according to the ARIA guidelines. Subanalysis was performed on AR participants reporting at least 1 subjective allergic symptom, such as sneezing, watery rhinorrhea, nasal obstruction, or itching, without fever or sore throat during the last year. The participants without AR were defined as healthy. According to the ARIA guidelines, the prevalences of intermittent mild, intermittent moderate/severe, persistent mild, and persistent moderate/severe AR were 44.1% (1.9), 26.1% (1.9), 11.3% (1.2), and 18.4% (1.6), respectively. The values of mental health components were higher in the moderate/severe AR group than in the mild AR group and also in the persistent AR group than in the intermittent AR group. After adjustment for age, sex, BMI, smoking status, alcohol use status, and exercise status, all components had a significant P for trend (Table 2).

In multiple logistic regression analysis after adjusting for demographic confounding factors, the odds ratios for stress and suicidal thoughts remained significantly higher in participants with AR. After adjustment for socioeconomic status, borderline results about depressive mood showed more changes in model 2 than in model 1. In model 3, regardless of adjustment for comorbid diseases, there remained significant associations between AR and 3 components of mental health: stress, depressive mood, and psychological consultation components (Table 3).