ABSTRACT
Objectives
Although the association between air pollutants and chronic rhinitis has been extensively investigated, causal inference is lacking. We aimed to investigate the causal relationship between air quality index (AQI) and outpatient visits for chronic rhinitis using instrumental variable (IV) analysis, employing thermal inversion.
Methods
We used National Health Insurance Service data collected between January 1, 2014, and December 31, 2017, to determine the short‐term causal relationship between air pollution and outpatient visits due to chronic rhinitis. A two‐stage generalized method‐of‐moments Poisson regression model was employed to estimate the causal effects. Stratified analyses by age group and robustness tests for IV with negative controls were conducted.
Results
A total of 81,210,447 outpatient visits for chronic rhinitis were analyzed. IV analysis demonstrated significant positive associations between AQI and outpatient visits on lag days 0, 3, 5, and 6, with the strongest effect observed at Lag 0 (relative risk [RR] 1.078, 95% confidence interval [CI] 1.045–1.113). Stratified analysis revealed that individuals aged 10–19 years were the most vulnerable, showing statistically significant relative risks across all lag days (RR: 1.039–1.161). Analyses using negative control outcomes and exposures supported the validity of the proposed instrumental variable approach, suggesting a robust causal effect of AQI on outpatient visits for chronic rhinitis.
Conclusions
We identified a causal relationship between increased air pollution and outpatient visits for chronic rhinitis. Unlike a simple association potentially confounded by various factors, our analysis provides robust evidence of the causal impact of air pollution on disease burden.
Level of Evidence
3.
Keywords: air pollution, causality, rhinitis
Using a natural environmental phenomenon, thermal inversion, as an instrumental variable, we investigated the causal effect of air pollution on outpatient visits for chronic rhinitis. Instrumental variable analysis revealed a significant causal increase in outpatient visits associated with higher air quality index, with the strongest effect observed on the same day of exposure. These findings provide robust causal evidence that short‐term air pollution directly increases the healthcare burden of chronic rhinitis.
1. Introduction
Chronic rhinitis is one of the most common inflammatory diseases of the upper airway, affecting ~20%–50% of the population depending on regional factors, and its prevalence has increased over the past several decades [1, 2, 3, 4]. Chronic rhinitis can be classified into allergic and nonallergic rhinitis, both of which significantly impair patients' quality of life [4, 5]. Air pollution is associated with symptom aggravation in both patients with allergic and those with nonallergic rhinitis, and those patients with chronic rhinitis may experience higher exposure levels to air pollutants [4, 6]. Environmental pollutants directly damage nasal epithelial cells and trigger the release of inflammatory mediators, such as chemokines and cytokines, thereby exacerbating rhinitis symptoms [4, 7, 8]. Particularly, long‐term exposure to fine particulate matter, including PM2.5 and NO2, has been consistently correlated with increased rhinitis severity [6, 9, 10].
The associations between air pollution and chronic rhinitis were investigated by adjusting for known confounders, such as ambient temperature, relative humidity, and seasonality of the diseases [11, 12, 13]. However, the appropriate assessment of the causal relationship between air pollution and chronic rhinitis requires controlling for potential confounders through randomization, which is unethical and impractical in environmental epidemiology. Pseudo‐randomization can be achieved by appropriately utilizing domain knowledge and statistical techniques, such as instrumental variable (IV) analysis. IV analysis utilizes domain knowledge to identify an IV that serves as a randomizer or controller for the exposure variable. This methodology is widely used in various settings, including Mendelian randomization, pharmacoepidemiological studies, social sciences, and per‐protocol analyses of randomized trials [14, 15]. Moreover, environmental epidemiological studies have employed IV analysis to assess the causal relationship between air pollution and health outcomes [16, 17].
Thermal inversion (TI) is an atmospheric phenomenon characterized by an increase in temperature with increasing altitude. Typically, the temperature in the troposphere decreases with altitude as the air near the surface is warmed by ground heat. The TI layer, which is typically induced by a cold front during the daytime and/or radiative cooling of the Earth's surface at night, restricts air circulation to the height of the inversion. When TI occurs at low altitudes, air pollutants are trapped within the layer, leading to increased concentrations at the ground level [18, 19]. Beard et al. and Trinh et al. investigated the association between air pollutants, TI, and health outcomes, specifically focusing on asthma emergency department visits and acute respiratory and cardiovascular diseases [20, 21]. Liu et al. utilized IV analysis to establish causality between total costs due to respiratory diseases and air pollution [22]. These studies provided a strong foundation for using TI as an IV for air pollution. However, no previous studies have investigated the association between TI and chronic rhinitis.
This study aimed to establish a causal relationship between air pollution and outpatient visits for chronic rhinitis using TI as an instrumental variable. We evaluated the robustness of our findings through falsification and sensitivity analyses.
2. Materials and Methods
2.1. Data and Study Outcome
This study included South Koreans who visited hospitals located in Seoul, the largest city in South Korea, between January 1, 2014, and December 31, 2017, for chronic rhinitis. Data were de‐identified and were provided by the National Health Insurance Service (NHIS). Chronic rhinitis was defined as a clinical visit for vasomotor or allergic rhinitis (ICD‐10: J30.x). This study was conducted in accordance with the 1964 Declaration of Helsinki and was approved by the Institutional Review Board (IRB) of Seoul National University Hospital in Seoul (IRB No. E‐1902‐045‐1008). This study adhered to strict confidentiality guidelines established by the National Health Insurance Service. Additional details on the data for the participants, negative control analysis, meteorological conditions, air pollution, and TI are provided in the Supporting Information.
2.2. Instrumental Variable Analysis
IV analysis is a feasible method that utilizes a third variable to examine the causal relationship between two other variables. The third variable, IV, has a strong association or causation with the exposure variable and affects the outcome variable through the exposure variable. The identification of an effective IV is challenging. When found, IV can effectively remove confounding factors and yield local average treatment effects, also referred to as complier‐average causal effects (CACE). This indicates that the causation can be established for those who adhere to the relationship between the IV and exposure variable [23]. Three basic assumptions underpin IV analysis: relevance, exclusion, and exchangeability [24]. Details about these assumptions and the applicability of TI as an IV can be found in the Supporting Information. The application of IV analysis in this study is schematically shown in Figure S1.
2.3. Thermal Inversion
TI, also known as temperature inversion, is a phenomenon in which temperature increases with altitude. The Earth's surface cooling overnight can alter the atmospheric stability, resulting in temperature increases with altitude and creating a TI layer. Air pollutants become trapped below the inversion layer as air circulation is restricted within this layer, leading to an increase in the pollutant concentration [18]. Other than affecting the air pollutant concentration, TI does not directly influence health outcomes [25]. Hence, we selected TI as the IV, as it increases the air pollutant concentration and affects health outcomes only indirectly through air pollution. A schematic TI is shown in Figure S2. Further information is provided in the Supporting Information.
2.4. Modeling and Variable Definition
The IV analysis was conducted using a two‐stage estimation approach. In the first stage, the exposure variable, mean air quality index (AQI), was modeled against IV and other explanatory variables. In the second stage, the outcome, the daily number of visits, was modeled against the exposure and explanatory variables, where IV was excluded in the second model. Residual confounding factors were also tested. The air pollutants, including particulate matter with diameters of less than 10 μm (PM10) and less than 2.5 μm (PM2.5), nitrogen dioxide (NO2), sulfur dioxide (SO2), carbon monoxide (CO), and ozone (O3), were transformed into AQI using the “con2aqi” package in R. The resulting AQIs were averaged to obtain the mean AQI, which served as the exposure variable. Further details regarding the modeling can be found in the Supporting Information.
2.5. Sensitivity Analysis and Robustness Tests
The robustness of the causal relationships between the AQI and daily visits due to chronic rhinitis was tested using various methods. Negative control analysis is a type of falsification test that can assess and validate the assumptions of the IV analyses, including the exclusion and exchangeability assumptions [24, 26]. The daily number of out‐of‐hospital cardiac arrest (OHCAs) due to non‐air‐pollutant‐related accidents (non‐disease related) was used as the negative control outcome. To eliminate confounding factors, OHCA cases resulting from traffic and suicidal accidents were excluded as they are known to be associated with air pollution. Two negative control exposure variables were also considered. First, the AQIs from 6 months in the future for each district in Seoul were additionally adjusted. Second, the AQI in Melbourne, Australia, measured on the same day as the AQI in Seoul, was used as negative control exposure [22, 24, 26]. Further detailed explanation can be found in the Supporting Information.
3. Results
3.1. Participant Demographics and Air Pollution and Meteorological Data
Table 1 presents the demographic characteristics of the participants. A total of 81,210,447 chronic rhinitis outpatient visits between January 1, 2014, and December 31, 2017, were analyzed. Among the 81,210,447 individuals with chronic rhinitis, the proportion of those aged 10–19 years was the highest (48%), followed by those aged 0–9 years (28%). Descriptive statistics of the air pollutants and meteorological variables are shown in Table 1. A total of 540 TI days occurred during the 1335‐day period (40.4%). TI was more prevalent from mid‐fall to early spring (428 of 673 days [63.6%], October to March) than during the other half of the year (112 of 662 days [16.9%], April to September) because of the mechanisms underlying the formation of TI.
TABLE 1.
Demographic and environmental characteristics of the study population and exposure variables.
| Values | |
|---|---|
| Male | 36,425,559 (45%) |
| Female | 44,784,888 (55%) |
| Age 0–9 | 22,743,059 (28%) |
| Age 10–19 | 7,916,932 (10%) |
| Age 20–64 | 38,746,065 (48%) |
| Age ≥ 65 | 11,804,391 (14%) |
| SO2 (ppb) | 5.2 ± 1.7 |
| CO (ppb) | 530.0 ± 205.9 |
| O3 (ppb) | 23.4 ± 12.7 |
| NO2 (ppb) | 320.5 ± 13.0 |
| PM10 (μg/m3) (daily mean) | 45.8 ± 28.7 |
| PM2.5 (μg/m3) (daily mean) | 24.4 ± 13.2 |
| Mean AQI | 47.4 ± 13.7 |
| Precipitation (daily sum) (mm) | 2.6 ± 10.3 |
| Temperature (daily mean) (°C) | 13.6 ± 10.3 |
| Relative humidity (daily mean) (%) | 63.4 ± 14.9 |
| Pressure (daily mean) (hPa) | 1006.0 ± 7.9 |
| Pollen by tree (grains/m3) | 21.1 ± 106.1 |
| Pollen by grass (grains/m3) | 2.6 ± 19.7 |
| Pollen by weed (grains/m3) | 2.9 ± 12.9 |
| Fungi (CFU/m3) | 13.8 ± 51.0 |
| Influenza (n) | 15,610 ± 33,800 |
Abbreviations: AQI, air quality index; CO, carbon monoxide; NO2, nitrogen dioxide; O3, ozone; PM, particulate matter; SO2, sulfur dioxide.
3.2. Causal Estimation Through IV Method
Figure 1 illustrates the relative risks (RRs) of the interquartile range (IQR) increase in the AQI on daily chronic rhinitis outpatient visits across different lag days for both IV and non‐IV analyses. The colored 95% confidence interval (CI) plot in Figure 1 represents the RRs from the IV analysis, reflecting CACE. Conversely, the gray CI plot in Figure 1 demonstrates the RRs from the non‐IV analysis, corresponding to a correlation. The highest RRs were observed on Lag 0 in both IV and non‐IV analyses (IV: RR, 1.078; 95% CI: 1.045–1.113; and non‐IV: RR, 1.027; 95% CI: 1.019–1.035), followed by insignificant associations in lags 1 and 2, which may indicate the harvesting effect of the AQI on chronic rhinitis outpatient visits. Significant causal relationships reappeared on Lag days 3, 5, and 6. The non‐IV analysis demonstrated a similar pattern, showing significant associations from Lag days 3 to 7. However, the RRs estimated from the non‐IV analysis reflected mere correlations rather than causal effects, and their magnitudes were consistently smaller than those from the IV analysis.
FIGURE 1.
RR of IQR increase in the AQI with the corresponding CI on daily outpatient visits due to chronic rhinitis based on the instrumental variable (IV) analysis and non‐IV analyses in Seoul from 2014 to 2017. * denotes statistical significance at the 0.05 level for estimates derived from the IV analyses. AQI, air quality index; CI, confidence interval; IQR, inter quartile range; RR, risk ratio.
Figure 2 shows the RR associated with the IQR increase in the moving average AQI during chronic rhinitis outpatient visits. While no significant causal associations were observed between the 2‐ and 3‐day moving averages, significance emerged from the 4‐day moving average onward and persisted through the 7‐day moving average. This finding supports the hypothesis that a harvesting effect may exist between AQI levels and chronic rhinitis outpatient visits. The unadjusted RRs values are presented in Table S1.
FIGURE 2.
RR of IQR increase in the moving AQI average with the corresponding CI on daily outpatient visits due to chronic rhinitis based on the instrumental variable (IV) analysis and non‐IV analyses in Seoul from 2014 to 2017. * denotes statistical significance at the 0.05 level for estimates derived from the IV analyses. AQI, air quality index; CI, confidence interval; IQR, inter quartile range; RR, risk ratio.
3.3. Stratified Analyses
Table 2 shows the RR associated with an IQR increase in the AQI during daily visits due to chronic rhinitis stratified into four age groups: 0–9, 10–19, 20–64, and ≥ 65 years. The 10–19‐year age group exhibited the highest susceptibility to air pollution, with statistically significant relative risks across all lag days (RR: 1.039–1.161). The 0–9‐year and 20–64‐year age groups followed, although their risk estimates were not consistently significant across all lags. Interestingly, the ≥ 65‐year age group demonstrated negative effects except on Lag 0 (the day the TI occurred). A stratified analysis by sex was also performed (Table S2), and similar patterns were observed in both male and female groups.
TABLE 2.
RR of the IQR increase in the AQI with the corresponding CI for daily outpatient visits due to chronic rhinitis by age groups through an IV analysis.
| Age 0–9 | Age 10–19 | Age 20–64 | Age ≥ 65 | |
|---|---|---|---|---|
| Lag 0 | 1.080** | 1.161** | 1.074** | 1.043** |
| (1.041–1.120) | (1.113–1.211) | (1.041–1.107) | (1.013–1.074) | |
| Lag 1 | 0.984 | 1.048* | 0.973* | 0.928** |
| (0.955–1.014) | (1.011–1.086) | (0.949–0.998) | (0.906–0.951) | |
| Lag 2 | 1.014 | 1.039* | 0.992 | 0.953** |
| (0.984–1.045) | (1.002–1.078) | (0.966–1.019) | (0.929–0.978) | |
| Lag 3 | 1.051** | 1.082** | 1.041** | 0.991 |
| (1.022–1.081) | (1.046–1.120) | (1.016–1.067) | (0.968–1.014) | |
| Lag 4 | 1.026* | 1.099** | 1.009 | 0.950** |
| (1.000–1.052) | (1.066–1.133) | (0.987–1.032) | (0.930–0.971) | |
| Lag 5 | 1.041** | 1.110** | 1.034** | 0.993 |
| (1.014–1.069) | (1.065–1.136) | (1.011–1.057) | (0.972–1.014) | |
| Lag 6 | 1.043** | 1.091** | 1.044** | 0.997 |
| (1.020–1.066) | (1.059–1.124) | (1.022–1.067) | (0.977–1.018) | |
| Lag 7 | 1.039** | 1.084** | 1.013 | 0.969** |
| (1.015–1.065) | (1.053–1.115) | (0.992–1.034) | (0.950–0.988) |
Note: RR values are expressed as point estimates with corresponding 95% CI. * denotes statistical significance at the 0.05 level. ** denotes statistical significance at the 0.01 level.
Abbreviations: AQI, air quality index; CI, confidence interval; IQR, interquartile range; IV, instrumental variable; RR, relative risk.
3.4. Negative Control Analysis
Table 3 presents the results of the negative control analysis. First, the analyses of negative control exposure for daily outpatient visits for chronic rhinitis are presented. After incorporating the AQI from 6 months in the future or the AQI of Melbourne into the model (2) in the Supporting Information, the results before and after adjusting for negative control exposure did not change the direction or size of the RR from the IV analysis. Additionally, the negative control outcome analyses showed no significant causal association. The results of the two types of negative control analyses confirmed that the proposed IV analysis provided a robust estimation of the causal effect of the AQI on daily outpatient visits for chronic rhinitis.
TABLE 3.
RR of the IQR increase in the AQI with the corresponding CI for daily outpatients due to chronic rhinitis through a negative control analysis.
| Negative control outcome analysis | Negative control exposure analysis | |||
|---|---|---|---|---|
| Out of hospital cardiac arrest | Main results | AQI from 6 months in the future | AQI of Melbourne | |
| Lag 0 | 1.105 | 1.078** | 1.081** | 1.088** |
| (0.801–1.524) | (1.045–1.113) | (1.047–1.116) | (1.054–1.123) | |
| Lag 1 | 0.998 | 0.976 | 0.976 | 0.979 |
| (0.761–1.308) | (0.951–1.002) | (0.951–1.002) | (0.954–1.004) | |
| Lag 2 | 1.045 | 0.996 | 0.997 | 0.997 |
| (0.807–1.354) | (0.969–1.023) | (0.971–1.024) | (0.971–1.024) | |
| Lag 3 | 0.963 | 1.038** | 1.034** | 1.042** |
| (0.765–1.214) | (1.013–1.065) | (1.008–1.061) | (1.016–1.069) | |
| Lag 4 | 1.072 | 1.013 | 1.017 | 1.007 |
| (0.855–1.343) | (0.990–1.036) | (0.994–1.040) | (0.984–1.029) | |
| Lag 5 | 1.054 | 1.036** | 1.044** | 1.032** |
| (0.841–1.320) | (1.013–1.060) | (1.020–1.068) | (1.007–1.056) | |
| Lag 6 | 1.132 | 1.043** | 1.046** | 1.038** |
| (0.902–1.420) | (1.020–1.066) | (1.023–1.070) | (1.015–1.061) | |
| Lag 7 | 1.098 | 1.020 | 1.015 | 1.013 |
| (0.878–1.374) | (0.999–1.042) | (0.994–1.037) | (0.991–1.035) | |
Note: RR values are expressed as point estimates with corresponding 95% CI. * denotes statistical significance at the 0.05 level. ** denotes statistical significance at the 0.01 level.
Abbreviations: AQI, air quality index; CI, confidence interval; IQR, interquartile range; RR, relative risk.
4. Discussion
This study utilized IV analysis to establish the short‐term causal relationship between air pollution and chronic rhinitis outpatient visits between January 1, 2014, and December 31, 2017. Although several observational studies have reported associations between air pollution and rhinitis symptoms or related healthcare utilization [6, 11, 12, 13], these findings have been limited by potential confounding factors and reverse causality. Recent studies utilizing Mendelian randomization have attempted to draw causal relationships using genetic variants as instrumental variables to investigate the causal relationship between air pollution and chronic rhinitis [27, 28]. However, Mendelian randomization has inherent limitations. Genetic instruments reflect long‐term exposure tendencies and are therefore less suitable for assessing short‐term health effects such as daily fluctuations in rhinitis symptoms. Furthermore, the interpretability of Mendelian randomization estimates is often limited because the instruments represent genetic proxies rather than direct environmental measurements.
To our knowledge, this is the first study to apply a non‐genetic instrumental variable approach, using a natural environmental phenomenon, thermal inversion (TI), to evaluate the causal relationship between air pollution and chronic rhinitis. We demonstrated that TI satisfies the assumptions required for a valid instrument, thereby addressing potential confounders [24]. The use of TI offers several key advantages over genetic instruments. First, TI captures short‐term, real‐world variations in air pollution, making it suitable for detecting acute health responses such as rhinitis‐related outpatient visits. Second, TI is directly linked to the measured concentrations of air pollutants, enabling causal estimates that are more policy‐relevant and clinically interpretable. Finally, the validity of nongenetic IV can be empirically assessed using negative control exposure and outcomes, thereby enhancing the robustness of causal inference.
A significant positive causal effect was observed in the AQIs of chronic rhinitis outpatient visits on Lag days 0, 3, 5, and 6. This finding indicates a causal relationship between air pollution and chronic rhinitis outpatient visits. An abrupt increase in air pollution may directly affect nasal epithelial cells and aggravate rhinitis symptoms, leading to significantly positive estimates in Lag 0. However, the non‐significant estimates on Lag days 1 and 2 may indicate a harvesting effect. This finding implies that patients with chronic rhinitis in poor condition responded immediately to high air pollution levels and visited the hospital. This immediate response may account for the subsequent decrease in the number of visits in Lag days 1 and 2. Additionally, the moving averages of the AQI showed increasing, significant, and positive estimates from Lag days 0–3 to 0–7, further strengthening the causal association.
In the stratified analysis by age, patients aged 0–9 years and 20–64 years exhibited patterns similar to those observed in the overall population. Among adults aged 20–64 years, the significantly decreased RR at Lag 1 may represent a compensatory estimate of the elevated RR at Lag 0. Statistically significant relative risks across all lag days indicated that adolescents aged 10–19 were the most susceptible to air pollution. Notably, significantly elevated RRs were observed at Lags 1 and 2, when a harvesting effect was typically present. Unlike other age groups, which may have more flexibility in avoiding air pollution exposure, school‐aged children are often required to attend school and engage in outdoor activities, making it difficult to avoid exposure. This may explain the higher RRs observed in this group.
Patients aged ≥ 65 years showed markedly different patterns compared to the other age groups. Except for Lag 0, which was the day of high air pollution when the AQI increased, this group exhibited a decrease in outpatient visits for rhinitis. The reduced RRs at Lags 1 and 2 may partly reflect the harvesting effect, but could also be due to behavioral responses, such as refraining from going outdoors during periods of severe air pollution, including visits to outpatient clinics. As chronic rhinitis, although burdensome in terms of quality of life, is not life threatening, elderly individuals may have prioritized avoiding exposure to air pollution over seeking care [29]. Additionally, significantly lower RRs were observed at Lags 4 and 7. When examining the moving averages, all periods from Lag 0–1 through 0–7 showed significantly reduced RR values, suggesting that the elderly population tended to minimize outdoor activities, including outpatient visits, during periods of poor air quality. These lower RR values at Lags 4 and 7 may have influenced the overall population trends, potentially contributing to the lack of significant increases in RR at these lags in the total population.
This study has several strengths. First, it provided a CACE by utilizing TI in IV analysis of chronic rhinitis. The results indicate that the estimated risks of the AQI in the non‐IV analyses were substantially underestimated. Many confounding factors influence acute exacerbation of chronic rhinitis, including medication adherence, disease severity, and comorbidities. However, big data studies often have limited information on these confounders [30], making it challenging to remove their effects and infer CACE. Our proposed IV methodology may effectively account for these confounding factors and estimate the causal relationships. Second, we estimated the overall effect of air pollution on daily outpatient visits due to chronic rhinitis using the mean AQI of the six air pollutants. These pollutants share common sources, are highly correlated, and often act as effect modifiers. By collectively assessing the causal effects of air quality on chronic rhinitis, this study may alert policymakers and health authorities and provide strong guidelines for action. Third, we employed a negative control analysis to enhance the robustness of our causal inference. As a negative control, we used out‐of‐hospital cardiac arrest, a condition that is theoretically unrelated to short‐term air pollution exposure, and successfully demonstrated a null effect. Additionally, we selected AQI in Melbourne on the same day and AQI in Seoul 6 months into the future as negative control exposures, based on the theoretical implausibility of any causal relationship between these exposures and outpatient visits in the target population. Including these negative control exposures resulted in minimal changes in the effect estimates, and the negative control exposures themselves were not statistically significant, suggesting that any potential confounding was negligible.
Despite its strengths, this study had some limitations. We utilized a preexisting environmental cohort constructed from administrative claims data for patients with chronic rhinitis, which did not include information to distinguish between allergic and nonallergic rhinitis. As a result, subgroup analyses based on rhinitis subtype could not be performed. In addition, although we adjusted for several potential confounders associated with chronic rhinitis, such as pollen, fungal activity, influenza, and calendar date, the dataset lacked information on underlying medical conditions, limiting our ability to control for comorbidities. Finally, exposure was summarized as mean AQI to reflect inversion‐related pollutant mixtures and improve interpretability; however, this limits pollutant‐specific inference (e.g., PM2.5 and other components). In addition, because our IV strategy relied on a single instrument (thermal inversion), we could not fit a multi‐pollutant IV model, which would require additional valid instruments; future work will incorporate multiple instruments to better distinguish pollutant‐specific versus mixture effects.
5. Conclusion
This study established a causal relationship between AQI and chronic rhinitis through IV analysis using TI as the IV. This approach moves beyond simple associations and addresses confounding biases, thereby offering a more accurate assessment of the true impact of air pollution on chronic rhinitis. This study presents a methodological advancement in evaluating the impact of air pollution on outpatient healthcare utilization for chronic rhinitis.
Funding
This work was supported by a National Research Foundation of Korea (NRF) grant (2018R1A5A1060031) to H.J.Y, funded by the Korean government and the Basic Science Research Program. This study was supported by the Korea Health Industry Development Institute (KHIDI, RS‐2025‐02263702) to J.Y.K.
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information
Data S1: Supplementary Methods. Detailed description of the data sources, instrumental variable analysis, definition of thermal inversion, modeling strategy, and sensitivity analyses.
Table S1: Unadjusted RR of the IQR increase and the moving average of AQI on daily outpatient visits due to chronic rhinitis through an IV analysis.
Table S2: RR of the IQR increase in AQI along with the CI for daily outpatient visits due to chronic rhinitis by sex group through an IV analysis.
Figure S1: Application of instrumental variable analysis in this study. (A) The direct association between air pollution and chronic rhinitis does not permit causal inference due to potential confounding variables, such as disease severity and comorbid conditions. (B) Instrumental variable analysis enables causal inference by utilizing an instrumental variable (thermal inversion) that influences the exposure (air pollution) but is not directly related to the outcome.
Figure S2: Schematic representation of thermal inversion. Thermal inversion can occur due to cold fronts during the day or radiative cooling of the Earth's surface at night, leading to the formation of an inversion layer that suppresses vertical air movement. This phenomenon results in the accumulation of air pollutants near the ground level.
Acknowledgments
The authors have nothing to report.
Contributor Information
Jin Youp Kim, Email: kjyoup0622@gmail.com.
Hyung‐Jin Yoon, Email: hjyoon@snu.ac.kr.
Data Availability Statement
The data that support the findings of this study are available from the National Health Insurance Service. Restrictions apply to the availability of these data, which were used under license for this study. Data are available from https://opendata.hira.or.kr/home.do with the permission of the National Health Insurance Service.
References
- 1. Katelaris C. H., Lee B. W., Potter P. C., et al., “Prevalence and Diversity of Allergic Rhinitis in Regions of the World Beyond Europe and North America,” Clinical and Experimental Allergy 42 (2012): 186–207. [DOI] [PubMed] [Google Scholar]
- 2. Bousquet J., Khaltaev N., Cruz A. A., et al., “Allergic Rhinitis and Its Impact on Asthma (ARIA) 2008 Update (In Collaboration With the World Health Organization, GA(2)LEN and AllerGen),” Allergy 63, no. Suppl 86 (2008): 8–160. [DOI] [PubMed] [Google Scholar]
- 3. Sly R. M., “Changing Prevalence of Allergic Rhinitis and Asthma,” Annals of Allergy, Asthma & Immunology 82 (1999): 233–248; quiz 248–252. [DOI] [PubMed] [Google Scholar]
- 4. Rosario C. S., Urrutia‐Pereira M., Murrieta‐Aguttes M., et al., “Air Pollution and Rhinitis,” Frontiers in Allergy 5 (2024): 1387525. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Choi B. Y., Kim S. Y., Kim C. H., et al., “Effectiveness of ClariFix (Cryoablation) of the Posterior Nasal Nerve on Nasal Symptoms in Patients With Chronic Rhinitis: A Systematic Review and Meta‐Analysis,” Journal of Rhinology 31 (2024): 57–66. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Burte E., Leynaert B., Marcon A., et al., “Long‐Term Air Pollution Exposure Is Associated With Increased Severity of Rhinitis in 2 European Cohorts,” Journal of Allergy and Clinical Immunology 145 (2020): 834–842.e836. [DOI] [PubMed] [Google Scholar]
- 7. Choi H., Yang M. J., Lee H. J., et al., “Particulate Matter Induces NLRP3 Inflammasome‐Mediated Pyroptosis in Human Nasal Epithelial Cells,” Journal of Rhinology 31 (2024): 106–113. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Choi T., Choi B. Y., Park J. H., et al., “Epithelial‐Mesenchymal Transition in Chronic Rhinosinusitis,” Journal of Rhinology 31 (2024): 67–77. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Zhang Y., Chen J. H., Han D. H., et al., “Association Between Ambient Air Pollution and Allergic Rhinitis: A Systematic Review and Meta‐Analysis,” Laryngoscope 134 (2024): 1012–1022. [Google Scholar]
- 10. Gao W., Wu J., Zhao J., et al., “Ambient Air Pollution and the Risk of Chronic Rhinosinusitis: A Systematic Review and Meta‐Analysis,” Laryngoscope 134 (2024): 31–40. [Google Scholar]
- 11. Chu H., Xin J., Yuan Q., Wang M., Cheng L., and Zhang Z., “The Effects of Particulate Matters on Allergic Rhinitis in Nanjing, China,” Environmental Science and Pollution Research 26 (2019): 11452–11457. [DOI] [PubMed] [Google Scholar]
- 12. Pénard‐Morand C., Raherison C., Charpin D., et al., “Long‐Term Exposure to Close‐Proximity Air Pollution and Asthma and Allergies in Urban Children,” European Respiratory Journal 36 (2010): 33–40. [DOI] [PubMed] [Google Scholar]
- 13. Kim J. Y., Park Y., Kim S. H., Kim S. P., Park S. W., and Yoon H. J., “Effect of Ambient Air Pollutants on the Medical Costs of Allergic Rhinitis in Seoul, Korea,” 133 (2023): 1828–1833. [DOI] [PubMed] [Google Scholar]
- 14. Labrecque J. and Swanson S. A., “Understanding the Assumptions Underlying Instrumental Variable Analyses: A Brief Review of Falsification Strategies and Related Tools,” Current Epidemiology Reports 5 (2018): 214–220. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Swanson S. A., Holme Ø., Løberg M., et al., “Bounding the Per‐Protocol Effect in Randomized Trials: An Application to Colorectal Cancer Screening,” Trials 16 (2015): 541. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Arceo E., Hanna R., and Oliva P., “Does the Effect of Pollution on Infant Mortality Differ Between Developing and Developed Countries? Evidence From Mexico City,” Economic Journal 126 (2016): 257–280. [Google Scholar]
- 17. Schwartz J., Fong K., and Zanobetti A., “A National Multicity Analysis of the Causal Effect of Local Pollution, NO2, and PM2.5 on Mortality,” Environmental Health Perspectives 126 (2018): 87004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Wallace J. and Kanaroglou P., “The Effect of Temperature Inversions on Ground‐Level Nitrogen Dioxide (NO2) and Fine Particulate Matter (PM2.5) Using Temperature Profiles From the Atmospheric Infrared Sounder (AIRS),” Science of the Total Environment 407 (2009): 5085–5095. [DOI] [PubMed] [Google Scholar]
- 19. Feng X., Wei S., and Wang S., “Temperature Inversions in the Atmospheric Boundary Layer and Lower Troposphere Over the Sichuan Basin, China: Climatology and Impacts on Air Pollution,” Science of the Total Environment 726 (2020): 138579. [DOI] [PubMed] [Google Scholar]
- 20. Beard J. D., Beck C., Graham R., et al., “Winter Temperature Inversions and Emergency Department Visits for Asthma in Salt Lake County, Utah, 2003‐2008,” Environmental Health Perspectives 120 (2012): 1385–1390. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Trinh T. T., Trinh T. T., Le T. T., Nguyen T. D. H., and Tu B. M., “Temperature Inversion and Air Pollution Relationship, and Its Effects on Human Health in Hanoi City, Vietnam,” Environmental Geochemistry and Health 41 (2019): 929–937. [DOI] [PubMed] [Google Scholar]
- 22. Liu Y. M. and Ao C. K., “Effect of Air Pollution on Health Care Expenditure: Evidence From Respiratory Diseases,” 30 (2021): 858–875. [DOI] [PubMed] [Google Scholar]
- 23. Angrist J. and Imbens G., Identification and Estimation of Local Average Treatment Effects (National Bureau of Economic Research, 1995). [Google Scholar]
- 24. Swanson S. A. and Hernán M. A., “Commentary: How to Report Instrumental Variable Analyses,” Epidemiology 24 (2013): 370–374. [DOI] [PubMed] [Google Scholar]
- 25. Garratt J. R., “The Atmospheric Boundary Layer,” Earth‐Science Reviews 37 (1994): 89–134. [Google Scholar]
- 26. Shi X., Miao W., and Tchetgen E. T., “A Selective Review of Negative Control Methods in Epidemiology,” Current Epidemiology Reports 7 (2020): 190–202. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Zhong J., Li W., Yang S., Shen Y., and Li X., “Causal Association Between Air Pollution and Allergic Rhinitis, Asthma: A Mendelian Randomization Study,” Frontiers in Public Health 12 (2024): 1386341. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Chen F., Zhang J., Xie J., and Fu X., “Causal Relationships Between Air Pollutants and Upper Respiratory Tract Infections: A Two‐Sample, Mendelian Randomization Study,” Journal of the Chinese Medical Association 88 (2025): 481–485, 10.1097/JCMA.0000000000001242. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Ryu G., Shin H. W., Kim D. K., et al., “Survey on Treatment‐Seeking Patterns in Patients With Allergic Rhinitis,” Journal of Rhinology 31 (2024): 138–144. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Park M. J., Kim J. H., Kim S. H., et al., “Risk of Biliary Tract and Pancreatic Cancer Following Obstructive Sleep Apnea Diagnosis: Analysis of a National Health Insurance Database,” Journal of Rhinology 32 (2025): 48–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data S1: Supplementary Methods. Detailed description of the data sources, instrumental variable analysis, definition of thermal inversion, modeling strategy, and sensitivity analyses.
Table S1: Unadjusted RR of the IQR increase and the moving average of AQI on daily outpatient visits due to chronic rhinitis through an IV analysis.
Table S2: RR of the IQR increase in AQI along with the CI for daily outpatient visits due to chronic rhinitis by sex group through an IV analysis.
Figure S1: Application of instrumental variable analysis in this study. (A) The direct association between air pollution and chronic rhinitis does not permit causal inference due to potential confounding variables, such as disease severity and comorbid conditions. (B) Instrumental variable analysis enables causal inference by utilizing an instrumental variable (thermal inversion) that influences the exposure (air pollution) but is not directly related to the outcome.
Figure S2: Schematic representation of thermal inversion. Thermal inversion can occur due to cold fronts during the day or radiative cooling of the Earth's surface at night, leading to the formation of an inversion layer that suppresses vertical air movement. This phenomenon results in the accumulation of air pollutants near the ground level.
Data Availability Statement
The data that support the findings of this study are available from the National Health Insurance Service. Restrictions apply to the availability of these data, which were used under license for this study. Data are available from https://opendata.hira.or.kr/home.do with the permission of the National Health Insurance Service.