Ambient Temperature and Kawasaki Disease in Seoul, Korea: Evidence from Nationwide HIRA Data Using Case-Crossover and Time-Series Analyses

Ambient Temperature and Kawasaki Disease in Seoul, Korea: Evidence from Nationwide HIRA Data Using Case-Crossover and Time-Series Analyses | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Ambient Temperature and Kawasaki Disease in Seoul, Korea: Evidence from Nationwide HIRA Data Using Case-Crossover and Time-Series Analyses Jung Hee Hong, Kiook Baek This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7559001/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background Kawasaki disease (KD) is an acute vasculitis of childhood and the leading cause of acquired heart disease in developed countries. The impact of ambient temperature on KD incidence remains unclear. We analyzed nationwide claims data for children aged 0–9 years in Seoul between 2012 and 2019. Associations between ambient temperature and KD were examined using a time-stratified case-crossover design and quasi-Poisson time-series regression within a distributed lag non-linear model framework (lags up to 21 days). Reduced-basis models summarized cumulative effects, and analyses were stratified by age. A total of 28,866 KD cases were identified. In the case-crossover analysis, the cumulative effect over lag 0–21 days was not significant. However, at lag 14–21 days, risk increased relatively linearly above 25°C, with an OR of 1.13 (95% CI: 1.03–1.23) at 29.7°C (97.5th percentile) compared with 25°C, while no significant decrease was observed at − 6.9°C (2.5th percentile). In the time-series analysis, risk decreased below 25°C and increased above 25°C over lag 0–21 days, with a more pronounced cumulative effect at lag 14–21 days (RR 1.14, 95% CI: 1.05–1.25 at 29.7°C). Age-stratified analyses showed consistent results in children aged 1–4 years, while associations were less clear in < 1 and 5–9 years. High ambient temperatures were associated with a delayed increase in KD risk, particularly among children aged 1–4 years, highlighting the potential influence of climate change on pediatric health. Kawasaki disease Ambient temperature Climate change Case-crossover study Time-series analysis Figures Figure 1 Figure 2 Figure 3 Figure 4 Introduction Kawasaki disease (KD) is an acute systemic vasculitis of childhood that predominantly affects children under the age of five years. It is the leading cause of acquired heart disease in developed countries and can result in serious cardiovascular sequelae such as coronary artery aneurysms if not treated promptly 1 . Although intravenous immunoglobulin (IVIG) therapy has substantially reduced the risk of complications, KD continues to impose a considerable health burden on children and families, as well as on health systems due to repeated hospitalizations, long-term cardiology follow-up, and associated medical costs 2 . In parallel, global climate change has intensified in recent decades, and the health impacts of rising temperatures have become an urgent public concern 3 . Temperature has been strongly linked with infectious disease outbreaks and exacerbations 4 , 5 , and emerging evidence indicates that climate-sensitive diseases are increasing in both frequency and severity. Beyond infectious conditions, several pediatric diseases with suspected infectious or inflammatory triggers, including intussusception 6 and appendicitis 7 , have also been reported to exhibit temperature-related variation. These findings suggest that climate factors may play a role in a broader range of childhood diseases than previously recognized. For KD specifically, some epidemiological studies have reported associations with ambient temperature 8 . Some investigations have indicated seasonal and geographic variations in incidence 9 , and a few have reported positive correlations with higher temperatures 10 , 11 . However, the available evidence remains limited, inconsistent, and often constrained by small sample sizes or localized data, leaving uncertainty about the true relationship between temperature and KD occurrence. KD is most prevalent in East Asian countries such as Japan, Korea, and Taiwan, and accordingly, research activity has been particularly strong in this region. In Korea, however, studies specifically examining the association between temperature and KD remain scarce. Given that hygiene, environmental conditions, healthcare access, circulating infectious agents, and genetic background may all influence this relationship, country-specific evidence is essential to understand the impact of temperature on KD incidence in the Korean population. Against this background, we sought to clarify the association between ambient temperature and KD incidence using large-scale health claims data from a metropolitan population. Materials and Methods Study population We used claims data from the National Health Insurance Service (NHIS), provided by the Health Insurance Review and Assessment Service (HIRA). The study population comprised children aged 0–9 years who were registered residents of Seoul and received medical care in hospitals located in Seoul, Gyeonggi, or Incheon, as these areas share the same metropolitan living sphere, between January 2012 and December 2019, KD cases were defined using the Korean Standard Classification of Diseases (KCD) code M30.3 together with administration of IVIG. For Kawasaki disease, the KCD code is identical to the International Classification of Diseases, 10th Revision (ICD-10) code. Seoul covers an area of approximately 605 km² and had a population of around 9–10 million during the study period. The city is one of the most densely populated and highly urbanized metropolitan areas in the world, with a built environment dominated by residential, commercial, and service sectors. There are no large-scale agricultural or industrial complexes within the city boundary, and the population is predominantly ethnically homogeneous. Exposure assessment Daily mean temperature data for Seoul were obtained from the Korea Meteorological Administration (KMA). Temperature was measured at the Seoul weather station located in Jongno-gu. Daily averages were calculated from hourly observations according to standard national meteorological protocols. The citywide daily mean temperature was used as the exposure variable, and the procedure was identical to that applied in a previous study 7 . Study design and statistical analysis We described the demographic characteristics of the study population, including age and sex distributions. In addition, we presented the annual mean daily cases of KD for each study year from 2012 to 2019, as well as the monthly mean daily cases calculated across all years for each calendar month. We applied a time-stratified case-crossover design and a time-series regression analysis. In the case-crossover design, the diagnosis date for each patient was defined as the case day, and all days falling on the same day of the week within the same calendar month and year served as control days. This time-stratified referent selection included both days preceding and following the event day. We applied an extended distributed lag non-linear model (DLNM) framework within the case-crossover design, allowing simultaneous assessment of nonlinear exposure–response relationships and distributed lag effects. Conditional logistic regression models with DLNM specifications were fitted to estimate odds ratios (ORs) and 95% confidence intervals (CIs). Cross-basis functions for daily mean temperature were modeled using natural cubic splines, with knots at the 2.5th percentile (–6.9°C) and the 97.5th percentile (29.7°C). We chose 25.0°C as the reference temperature since it corresponds to a neutral operative condition within the thermoneutral zone and falls squarely in the comfort region for typical activity and attire 12 – 14 . Lags from 0 to 21 days were considered, and both the exposure–response and lag–response dimensions were modeled flexibly using spline functions, thereby accounting for nonlinear as well as constrained lag effects. Effect sizes were expressed in terms of odds ratios (ORs). $$\:logit\left({p}_{\left\{it\right\}}\right)=\:{\alpha\:}_{i}+\:cb\left({T}_{\left\{it\right\}},\:l\right)$$ p {it} is the probability that day t is the case day for subject i . α i is the stratum specific intercept for each subject. T it is the daily mean temperature on day t . cb(T {it} , l) is the cross basis function of temperature T it with lag l from 0 to 21 days. We modeled the variation in risk across lags by comparing temperatures at the 2.5 percentile (–6.9°C) and the 97.5 percentile (29.7°C) with a reference of 25.0°C. For lag intervals where significant associations were observed, we applied a reduced-basis approach to evaluate the overall cumulative effects. In the time-series analysis, daily counts of KD cases were modeled using quasi-Poisson regression with a DLNM framework, producing rate ratios (RRs) and 95% CIs. Calendar time (day) was modeled with a natural cubic spline using 7 degrees of freedom (df) per year, consistent with prior work that typically applies 7 df for annual temporal control and is widely used in environmental health research on daily morbidity 15 – 19 . Indicator variables for day-of-week, holidays (including Saturdays and Sundays), the day before holidays, and the day after holidays were included. The maximum lag was set to 21 days, consistent with the specification used in the case-crossover analysis. To facilitate comparability with the case-crossover analysis, we also applied reduced-basis models in the time-series framework, using the same lag interval as in the case-crossover analysis. Effect sizes were expressed in terms of risk ratios (RRs). $$\:{Y}_{t}\sim\:quasiPoisson\left(m{u}_{t}\right)$$ $$\:log\left({\mu\:}_{t}\right)=\:\alpha\:+\:cb\left({T}_{t},\:l\right)+\:ns\left(time,df=\:7\:\times\:years\right)+\:DO{W}_{t}+\:{H}_{\{0,t\}}+\:{H}_{\left\{-1,t\right\}}+\:{H}_{\left\{+1,t\right\}}$$ Y t is the observed number of KD cases on day t . µ t is the expected number of cases. cb(T t , l) is the cross basis for daily mean temperature with lag 0 to 21 days. ns(time, 7×years) is a natural cubic spline of calendar time with 7 degrees of freedom per year. DOW t is the indicator variables for day of week. H {0,t} is the indicator for holidays. H {−1,t} is the indicator for the day before a holiday. H {+1,t} is the indicator for the day after a holiday. When H {0,t} takes the value of 1, the other two variables are set to 0. Subgroup analysis We additionally conducted subgroup analyses by age group to examine potential effect modification. The study population was stratified into three categories: children aged < 1 year, 1–4 years, and 5–9 years. Both the case-crossover and time-series models were fitted separately within each subgroup, using the same modeling specifications as in the main analysis. Ethical statements This study was approved by the Institutional Review Board of Dongsan Medical Center (DSMC IRB No. 2025-02-010). As only non-identifiable data were used, the requirement for informed consent was waived. Results A total of 22,727 KD cases were identified between 2012 and 2019. The mean age of patients was 2.35 ± 1.70 years, and 13,077 (57.5%) were male. Subgroup distribution by age showed that 4,968 (21.9%) were <1 year, 14,917 (65.6%) were 1–4 years, and 2,842 (12.5%) were 5–9 years (Table 1). The mean daily cases among all periods was 7.78 ± 3.66 cases per day. The monthly average number of Kawasaki disease diagnoses is presented in Figure 1 alongside the mean temperature, showing peaks in December–January and June. These seasonal variations appear to occur independently of temperature changes. The 2.5th percentile of temperature was –6.9 °C, and the 97.5th percentile was 29.7 °C, among 2012–2019. In Figure 2, the regression line estimated using a natural spline is presented, and, in line with the monthly fluctuations observed earlier, seasonality is suggested. In the case-crossover analysis, the DLNM results are shown in Figure 3. The contour plot of the exposure–lag–response surface illustrates the overall pattern of the association. The overall cumulative effect across lag 0–21 days did not show statistically significant associations at temperatures lower or higher than 25.0 °. When examining lag-specific effects of 29.7 °C compared with 25.0 °C, a significant increase in risk was observed beginning at lag 14 days (OR 1.01, 95%CI 1.00–1.02 at 14-day lag point), with the risk continuing to rise with increasing lag. In contrast, for –6.9 °C compared with 25.0 °C, no significant associations were found across any lag period. The reduced-basis analysis restricted to lag 14–21 days indicated that, relative to 25 °C, the risk increased in an approximately linear and statistically significant manner at temperatures exceeding 26.6 °C (OR 1.02, 95% CI 1.00 –1.05 at 26.6 °C). When calculating the cumulative effect over lag 14–21 days, the OR at 29.7 °C compared with 25 °C was estimated to be 1.13 (95% CI, 1.03–1.23) (Table 2). In the time-series regression analysis, corresponding DLNM results are presented in Figure 4. The overall cumulative effect across lag 0–21 days demonstrated a significant decrease in KD cases at temperatures below 25 °C and a significant increase at higher temperatures. For 29.7 °C compared with 25 °C, risk elevation was evident from around lag 5 days and continued to increase with longer lags (RR: 1.18, 95% CI: 1.05–1.33). For –6.9 °C compared with 25 °C, a significant decrease in risk was observed (RR: 0.52, 95% CI: 0.30–0.89). The reduced-basis analysis for lag 14–21 days yielded results consistent with the case-crossover analysis, showing a significant increase in risk above 25 °C. In the cumulative effect over lag 14–21 days, no significant association was observed at –6.9 °C (RR 0.52, 95% CI: 0.30–0.89). The overall cumulative RR at 29.7 °C compared with 25 °C was 1.18 (95% CI: 1.05–1.33). The cumulative RR over lag 14–21 days was 1.15 (95% CI: 1.07–1.24) (Table 2). In the subgroup analysis by age, results for the 1–4 year group were consistent with those in the overall population, showing significant risk elevation at higher temperatures during lag 14–21 days. In contrast, no significant associations were identified in children aged <1 year. For those aged 5–9 years, the time-series design indicated a significant increase in risk, whereas the case-crossover design did not show a significant association. The analyses stratified by age group, along with the corresponding figures, are presented in the supplementary material. Discussion In this study, we evaluated the association between ambient temperature and KD case counts using two complementary statistical approaches: a time-stratified case-crossover design and a time-series regression analysis. Each method has distinct strengths, with the case-crossover design offering efficient control for individual-level fixed confounders and time-invariant characteristics, while the time-series framework provides a population-level perspective with flexible adjustment for long-term and seasonal trends. By combining these approaches, we sought to maximize their respective advantages and compensate for their limitations. Although simple inspection of monthly seasonality did not reveal a clear relationship with temperature, both analyses consistently demonstrated that high ambient temperature was associated with an increased risk of KD after a delayed period. Specifically, at lag days 14–21, exposure to the 97.5th percentile of daily mean temperature (29.7 °C) compared with the reference of 25 °C was significantly associated with elevated ORs and RRs. This pattern was most prominent in children aged 1–4 years, highlighting a particularly susceptible subgroup within the pediatric population. Among children aged five years and older, the time-series analysis revealed a significant increase in KD risk at temperatures exceeding 25 °C, whereas this association was not observed in the case-crossover design. This inconsistency highlights the need for further investigations with larger sample sizes and refined study designs that incorporate multiple potential confounders. In contrast, no significant temperature effect was detected among infants younger than one year, which may be explained by their limited outdoor activity and the tendency for this age group to remain in air-conditioned indoor environments 20 that reduce actual exposure during periods of elevated ambient temperature. Moreover, infants under one year typically remain at home under parental care, whereas children beyond the first year of life often enter daycare or other group settings 21, 22 , where increased social contact and institutional childcare may modify patterns of infectious disease exposure. The use of two complementary statistical approaches in this study is of particular methodological importance. When exposure is common to all subjects on a given day, the case-crossover and time-series frameworks are grounded in closely related statistical principles and, under appropriate control for temporal confounding, can yield comparable effect estimates—particularly when a time-stratified referent selection (same day of week within the same month and year) is used to avoid overlap bias. However, estimates may diverge depending on referent selection in the case-crossover design and on the degrees of freedom chosen for spline functions to adjust long-term and seasonal trends in the time-series approach 23 . The case-crossover design evaluates short-term environmental effects by comparing each subject to him-/herself, thereby removing confounding by time-invariant individual characteristics 24 . In this study, referent days were selected as the same day of week within the same calendar month and year as the event day—a strategy that controls for seasonality, long-term trends, and day-of-week effects and generally yields a balanced distribution of referents around the case day 25 . A practical limitation is that, unless explicitly modeled, short-term fluctuations in healthcare utilization related to holidays and adjacent days may influence case ascertainment in the case-crossover setting. In parallel, the time-series framework provides population-level inference with high statistical efficiency by using all daily cases; calendar time was modeled with natural cubic splines to address long-term and seasonal trends 26 , and indicators for holidays and adjacent days were included to capture short-term utilization shifts. Nonetheless, results in the time-series approach can be sensitive to the selection of degrees of freedom and other modeling choices 23 . By combining both approaches, we aimed to capitalize on their strengths while mitigating their respective limitations. The incorporation of distributed lag nonlinear models (DLNM) into both case-crossover and time-series frameworks represents a key methodological strength of this study. DLNM allow simultaneous modeling of nonlinear exposure–response relationships and complex lag structures, thereby enabling a realistic characterization of delayed environmental effects 27 . Previous epidemiological investigations have reported significant associations between ambient temperature and KD incidence, suggesting that extreme temperature conditions may act as risk 9, 10, 28 . Moreover, Kawasaki disease exhibits marked seasonality across countries modifiers 28, 29 a pattern plausibly linked to temperature fluctuations and the epidemic dynamics of infectious diseases. Although some studies, including those from Japan, have identified significant associations within shorter lags of 0–5 days 8, 10 , the etiological hypothesis that KD may be preceded by infections 30 and the diagnostic requirement of at least five days of fever 31 , indicate that longer lag periods should be considered. In this study, we explicitly extended the lag structure up to 21 days and employed reduced-basis models to examine cumulative effects within specific intervals. This approach revealed significant associations during lag days 14–21, highlighting a critical window of vulnerability. Unlike previous studies that focused on shorter lags, our findings emphasize that elevated temperatures may influence KD occurrence even over comparatively long lag periods. An additional strength of this study lies in the use of a comprehensive claims database derived from the Korean National Health Insurance system, which covers virtually the entire population 32, 33 . Case definition required both an ICD-10 diagnosis code for KD and administration of IVIG, thereby enhancing specificity. By restricting the analysis to residents of Seoul, we were able to achieve near-complete case ascertainment within a well-defined urban population. The compact geographic area of Seoul reduced variation in individual exposure relative to the single meteorological monitoring site, while the city’s high population density ensured a sufficiently large sample size for robust analysis. Although data derived from a single country or region may have limitations in terms of generalizability, they provide valuable information for estimating the health and economic burden of temperature within that setting. In particular, the near-complete data from Seoul enabled direct estimation of the health impact of rising temperatures in an urban Korean population, offering a strong basis for projecting risks in other metropolitan areas of Korea with similar demographic and environmental characteristics. Several limitations should be considered. First, changes in the diagnostic criteria for KD in 2017 31, 34 . may have affected temporal patterns of case identification. Such changes influence only the outcome and not the exposure, and their potential impact on short-term associations is likely to have been minimized by our methodological approaches, namely the time-stratified case-crossover design and the adjustment for seasonality and long-term trends using natural splines in the time-series analysis. We did not adjust for other daily environmental exposures such as air pollution. However, these factors are more plausibly mediators, as they are influenced by temperature and subsequently affect KD risk, rather than independent confounders, since it is unlikely that they directly alter variations in ambient temperature. Under modern epidemiologic theory, a true confounder is defined as a factor associated with both the exposure and the outcome 35 . In contrast, variables that affect only the outcome (e.g., changes in diagnostic criteria) or operate as mediators (e.g., air pollutants) influenced by the exposure do not bias the estimated effect size or direction of the exposure–outcome association. Therefore, while not adjusting for these factors may have led to some overestimation of the direct effect of temperature, it is unlikely to have materially affected our estimation of the overall impact attributable to temperature. Additionally, because we assigned the same daily ambient temperature exposure to all cases occurring on a given day, we were unable to take advantage of higher-resolution exposure data to assess individual-level variation, which would have maximized the strengths of the case-crossover design. This limitation also reflects the inability to account for personal-level exposures, such as time spent indoors or the use of cooling and heating systems, thereby restricting prediction of temperature-related risks at the individual level. Nevertheless, the approach provides valuable insights into the broader epidemiological and public health implications of rising ambient temperatures at the regional scale. Conclusions This study provides novel evidence that elevated ambient temperature is associated with an increased risk of Kawasaki disease, with risks rising above the reference temperature of 25 °C, particularly during lag days 14–21 and most prominently among children aged 1–4 years. These findings highlight the potential impact of rising temperatures on pediatric inflammatory diseases and underscore the importance of incorporating climate factors into public health preparedness and policy planning, especially in metropolitan settings such as Seoul. Declarations Acknowledgment This work was supported by the Dongguk University, College of Medicine Research Fund of 2025. Disclosure The authors declare no conflicts of interest. Ethical statements This study was approved by the Institutional Review Board of Dongsan Medical Center (DSMC IRB No. 2025-02-010). As only non-identifiable data were used, the requirement for informed consent was waived. Author contribution Jung Hee Hong: Conceptualization, Investigation, Validation, Writing- Reviewing and Editing Kiook Baek: Visualization, Methodology, Data curation, Writing- Original draft preparation. Declaration of generative AI in scientific writing During the preparation of this work the authors used ChatGPT in order to English translation and polishing. After using this tool/service, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication References Newburger JW, Takahashi M, Burns JC. Kawasaki disease. Journal of the American College of Cardiology. 2016;67(14):1738-1749. Uehara R, Belay ED. Epidemiology of Kawasaki Disease in Asia, Europe, and the United States. Journal of Epidemiology. 2012;22(2):79-85. Legg S. IPCC, 2021: Climate change 2021-the physical science basis. Interaction. 2021;49(4):44-45. Lafferty KD. The ecology of climate change and infectious diseases. Ecology. 2009;90(4):888-900. Rohr JR, Cohen JM. Understanding how temperature shifts could impact infectious disease. PLoS biology. 2020;18(11):e3000938. Nawa N, Nishimura H, Fushimi K, Fujiwara T. Association between heat exposure and intussusception in children in Japan from 2011 to 2022. Pediatr Res. 2025:1-5. Baek K, Park C. Analyzing the dynamics of complicated and uncomplicated appendicitis during the COVID-19 pandemic in Seoul, Korea: a multifaceted time series approach. Epidemiology and Health. 2024;46:e2024081. Nawa N, Nishimura H, Fushimi K, Fujiwara T. Association between heat exposure and Kawasaki disease: A time-stratified case-crossover study. Environ Res. 2024;263:120231. Celis-Seposo AK, Madaniyazi L, Seposo X, Hashizume M, Yoshida LM, Toizumi M. Incidence and seasonality of Kawasaki disease in children in the Philippines, and its association with ambient air temperature. Frontiers in Pediatrics. 2024;12:1358638. Fujii F, Egami N, Inoue M, Koga H. Weather condition, air pollutants, and epidemics as factors that potentially influence the development of Kawasaki disease. Sci Total Environ. 2020;741:140469. Lin Z, Meng X, Chen R , et al. Ambient air pollution, temperature and Kawasaki disease in Shanghai, China. Chemosphere. 2017;186:817-822. Yin Q, Zhang Y, Liu Y. Investigation on Thermal Comfort and Thermal Adaptive Behaviors of Rural Residents in Suibin Town, China, in Summer. Sustainability. 2023;15(8):6630. Kingma BR, Frijns AJ, Schellen L, van Marken Lichtenbelt WD. Beyond the classic thermoneutral zone: Including thermal comfort. Temperature (Austin). Jul-Sep 2014;1(2):142-149. Mora R, Bean R. Thermal comfort: Designing for people. ASHRAE J. 2018;60(2):40-46. Yang Y, Li X, Wang S , et al. Assessing the impact of temperature on acute exacerbation of chronic obstructive pulmonary disease hospitalizations in residents of Panzhihua City: a multi-districts study using a distributed lag non-linear model. BMC Public Health. 2024;24(1):2151. Gao S, Yang T, Zhang X , et al. A longitudinal study on the effect of extreme temperature on non-accidental deaths in Hulunbuir City based on DLNM model. Int Arch Occup Environ Health. 2023;96(7):1009-1014. Peng RD, Dominici F, Louis TA. Model Choice in Time Series Studies of Air Pollution and Mortality. Journal of the Royal Statistical Society Series A: Statistics in Society. 2006;169(2):179-203. Bhaskaran K, Gasparrini A, Hajat S, Smeeth L, Armstrong B. Time series regression studies in environmental epidemiology. Int J Epidemiol. Aug 2013;42(4):1187-1195. Dominici F, Samet JM, Zeger SL. Combining evidence on air pollution and daily mortality from the 20 largest US cities: a hierarchical modelling strategy. Journal of the Royal Statistical Society Series A: Statistics in Society. 2000;163(3):263-302. Yoon H, Seo J, Kim T , et al. Development of Korean exposure factors for children in Korea. 2017. Ahn J, Shin N. The use of child care center for infants of dual-working families in Korea. Children and Youth Services Review. 2013;35(9):1510-1519. Yun-Jin B. ECEC Statistics of KOREA: Recent trends of Services, Enrollment, and Workforce. 2023. Lu Y, Zeger SL. On the equivalence of case-crossover and time series methods in environmental epidemiology. Biostatistics. 2007;8(2):337-344. Maclure M. The case-crossover design: a method for studying transient effects on the risk of acute events. American journal of epidemiology. 1991;133(2):144-153. Szyszkowicz M. Case-Crossover Method with a Short Time-Window. Int J Environ Res Public Health. Dec 27 2019;17(1). Kim H, Lee J-T, Fong KC, Bell ML. Alternative adjustment for seasonality and long-term time-trend in time-series analysis for long-term environmental exposures and disease counts. BMC Medical Research Methodology. 2021;21(1):2. Gasparrini A. Distributed lag linear and non-linear models in R: the package dlnm. Journal of statistical software. 2011;43:1-20. Lin M-T, Wu M-H. The global epidemiology of Kawasaki disease: review and future perspectives. Global cardiology science & practice. 2017;2017(3):e201720. Burns JC, Cayan DR, Tong G , et al. Seasonality and temporal clustering of Kawasaki syndrome. Epidemiology. 2005;16(2):220-225. Burgner D, Harnden A. Kawasaki disease: what is the epidemiology telling us about the etiology? International journal of infectious diseases. 2005;9(4):185-194. McCrindle BW, Rowley AH, Newburger JW , et al. Diagnosis, Treatment, and Long-Term Management of Kawasaki Disease: A Scientific Statement for Health Professionals From the American Heart Association. Circulation. Apr 25 2017;135(17):e927-e999. Kim L, Kim J-A, Kim S. A guide for the utilization of health insurance review and assessment service national patient samples. Epidemiology and health. 2014;36:e2014008. Cheol Seong S, Kim Y-Y, Khang Y-H , et al. Data resource profile: the national health information database of the National Health Insurance Service in South Korea. International journal of epidemiology. 2017;46(3):799-800. Kobayashi T, Ayusawa M, Suzuki H , et al. Revision of diagnostic guidelines for Kawasaki disease. Pediatrics international. 2020;62(10). Tennant PW, Murray EJ, Arnold KF , et al. Use of directed acyclic graphs (DAGs) to identify confounders in applied health research: review and recommendations. Int J Epidemiol. 2021;50(2):620-632. Tables Table 1. General characteristics of participants Number (%) Total 22,727 (100) Sex Male 13,077 (64.8) Female 9,650 (35.2) Age 2.35 ± 1.70 <1 4,968 (23.3) 1–4 14,917 (64.8) 5–9 2,842 (11.9) Diagnosed year 2012 2,747 (12.1) 2013 3,063 (13.5) 2014 3,129 (13.8) 2015 2,956 (13.0) 2016 2,874 (12.6) 2017 2,591 (11.4) 2018 2,792 (12.3) 2019 2,575 (11.3) Table 2. Effect estimates at the 2.5 and 97.5 percentiles of temperature compared with 25 °C in the case-crossover and time-series models Cumulative lags (days) 0–21 14–21 Temperature (reference: 25°C) −6.9 °C (2.5 percentile) 29.7 °C (97.5 percentile) −6.9 °C (2.5 percentile) 29.7 °C (97.5 percentile) Case-crossover (OR, 95% CI) 1.10 (0.86, 1.42) 1.08 (0.92, 1.27) 1.03 (0.84, 1.25) 1.13 (1.03, 1.23) * Time-series (RR, 95% CI) 0.52 (0.30, 0.89) * 1.18 (1.05, 1.32) * 0.84 (0.64, 1.11) 1.14 (1.05, 1.25) * OR, Odds ratio; RR, Rate ratio * : statistically significant (< 0.05) To obtain effect estimates at specific points, non-linear modeling of lag and exposure variables was performed using a DLNM framework. In the case-crossover design, control days were selected as other days of the same month, whereas in the time-series analysis, calendar time was adjusted using a natural spline with 7 degrees of freedom per year based on daily counts. Supplementary Files supple.pptx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7559001","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":512657238,"identity":"06c1005c-910e-443d-95f5-a350929b88a4","order_by":0,"name":"Jung Hee Hong","email":"","orcid":"","institution":"Keimyung University College of Medicine: Keimyung University School of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Jung","middleName":"Hee","lastName":"Hong","suffix":""},{"id":512657239,"identity":"014c7417-6fa6-44a9-8f0e-6c277b5c3096","order_by":1,"name":"Kiook Baek","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAyUlEQVRIiWNgGAWjYBACAx4GxgcfKmzkQJwDD4jUwmw440yaMVhLApFa2KR52w4nNoB4RGkx5zl82HBm2+H0+WGHHwJtsZPTbSCgxbK3LfHBh3PpuRtvpxkAtSQbmx0g5LDzPMaGM8qsczfOTgBpOZC4jQgtZtI8bMzphrPTPxCp5WwPUEubc4K8dA6Rtlj2HEsGBbLhBumcggMJBkT4xZwn+SAoKuXlZ6dv/vChwk6OoBaEC8EqDYhVDgLyDaSoHgWjYBSMghEFACSSSd23ZMetAAAAAElFTkSuQmCC","orcid":"https://orcid.org/0000-0003-2822-8480","institution":"Dongguk University College of Medicine","correspondingAuthor":true,"prefix":"","firstName":"Kiook","middleName":"","lastName":"Baek","suffix":""}],"badges":[],"createdAt":"2025-09-08 01:34:41","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7559001/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7559001/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":91561236,"identity":"91955ea6-0556-4252-b9e6-d88fb5d24b9f","added_by":"auto","created_at":"2025-09-17 18:47:02","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":284987,"visible":true,"origin":"","legend":"\u003cp\u003eMonthly average number of Kawasaki disease cases and corresponding mean temperatures in Seoul during the study period.\u003c/p\u003e","description":"","filename":"figure1.png","url":"https://assets-eu.researchsquare.com/files/rs-7559001/v1/99f0384670216e638741dd9f.png"},{"id":91561240,"identity":"6ebc8eaa-8e7a-4b31-94e4-7069f34ef60d","added_by":"auto","created_at":"2025-09-17 18:47:02","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":138178,"visible":true,"origin":"","legend":"\u003cp\u003eRegression line estimated using a natural spline, illustrating the overall temporal pattern of Kawasaki disease cases across the entire study period.\u003c/p\u003e","description":"","filename":"figure221.png","url":"https://assets-eu.researchsquare.com/files/rs-7559001/v1/89695a2dca05f99db773b239.png"},{"id":91561244,"identity":"f013631e-a3af-4676-8591-0a36f09a1167","added_by":"auto","created_at":"2025-09-17 18:47:02","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":114500,"visible":true,"origin":"","legend":"\u003cp\u003eDistributed lag non-linear model (DLNM) results from the case-crossover design for the association between ambient temperature and Kawasaki disease incidence, with 25.0 °C set as the reference temperature (OR = 1). (A) Contour plot of the exposure–lag–response surface. (B) Overall cumulative effect across lags 0–21 days. (C) Lag-specific ORs for 29.7 °C (97.5th percentile) compared with 25.0 °C. (D) Lag-specific ORs for –6.9 °C (2.5th percentile) compared with 25.0 °C. (E) Overall cumulative effect restricted to lag days 14–21.\u003c/p\u003e","description":"","filename":"Figure321.png","url":"https://assets-eu.researchsquare.com/files/rs-7559001/v1/181de95314c21c3c821b0649.png"},{"id":91562462,"identity":"728d98bf-c39d-40d8-a387-1f54f495854c","added_by":"auto","created_at":"2025-09-17 18:55:02","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":116877,"visible":true,"origin":"","legend":"\u003cp\u003eDistributed lag non-linear model (DLNM) results from the time-series design for the association between ambient temperature and Kawasaki disease incidence, with 25.0 °C set as the reference temperature (RR = 1). (A) Contour plot of the exposure–lag–response surface. (B) Overall cumulative effect across lags 0–21 days. (C) Lag-specific RRs for 29.7 °C (97.5th percentile) compared with 25.0 °C. (D) Lag-specific ORs for –6.9 °C (2.5th percentile) compared with 25.0 °C. (E) Overall cumulative effect restricted to lag days 14–21.\u003c/p\u003e","description":"","filename":"Figure41.png","url":"https://assets-eu.researchsquare.com/files/rs-7559001/v1/26e47b9cd91d61926e91e975.png"},{"id":92611297,"identity":"c02733ad-17e8-40c8-b456-2bce04149df8","added_by":"auto","created_at":"2025-10-01 16:26:21","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1067720,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7559001/v1/05122235-1b21-49f8-8b09-99e05cdbbe6d.pdf"},{"id":91561251,"identity":"26285bc2-91d2-4f17-858f-0dc14cbed00a","added_by":"auto","created_at":"2025-09-17 18:47:02","extension":"pptx","order_by":8,"title":"","display":"","copyAsset":false,"role":"supplement","size":3082816,"visible":true,"origin":"","legend":"","description":"","filename":"supple.pptx","url":"https://assets-eu.researchsquare.com/files/rs-7559001/v1/c56a60a82b68238da555566e.pptx"}],"financialInterests":"","formattedTitle":"Ambient Temperature and Kawasaki Disease in Seoul, Korea: Evidence from Nationwide HIRA Data Using Case-Crossover and Time-Series Analyses","fulltext":[{"header":"Introduction","content":"\u003cp\u003eKawasaki disease (KD) is an acute systemic vasculitis of childhood that predominantly affects children under the age of five years. It is the leading cause of acquired heart disease in developed countries and can result in serious cardiovascular sequelae such as coronary artery aneurysms if not treated promptly\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e. Although intravenous immunoglobulin (IVIG) therapy has substantially reduced the risk of complications, KD continues to impose a considerable health burden on children and families, as well as on health systems due to repeated hospitalizations, long-term cardiology follow-up, and associated medical costs\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eIn parallel, global climate change has intensified in recent decades, and the health impacts of rising temperatures have become an urgent public concern\u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e. Temperature has been strongly linked with infectious disease outbreaks and exacerbations\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e, and emerging evidence indicates that climate-sensitive diseases are increasing in both frequency and severity. Beyond infectious conditions, several pediatric diseases with suspected infectious or inflammatory triggers, including intussusception\u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u003c/sup\u003e and appendicitis\u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e, have also been reported to exhibit temperature-related variation. These findings suggest that climate factors may play a role in a broader range of childhood diseases than previously recognized.\u003c/p\u003e\u003cp\u003eFor KD specifically, some epidemiological studies have reported associations with ambient temperature\u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e. Some investigations have indicated seasonal and geographic variations in incidence\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/sup\u003e, and a few have reported positive correlations with higher temperatures\u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e. However, the available evidence remains limited, inconsistent, and often constrained by small sample sizes or localized data, leaving uncertainty about the true relationship between temperature and KD occurrence. KD is most prevalent in East Asian countries such as Japan, Korea, and Taiwan, and accordingly, research activity has been particularly strong in this region. In Korea, however, studies specifically examining the association between temperature and KD remain scarce. Given that hygiene, environmental conditions, healthcare access, circulating infectious agents, and genetic background may all influence this relationship, country-specific evidence is essential to understand the impact of temperature on KD incidence in the Korean population. Against this background, we sought to clarify the association between ambient temperature and KD incidence using large-scale health claims data from a metropolitan population.\u003c/p\u003e"},{"header":"Materials and Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003eStudy population\u003c/h2\u003e\u003cp\u003eWe used claims data from the National Health Insurance Service (NHIS), provided by the Health Insurance Review and Assessment Service (HIRA). The study population comprised children aged 0\u0026ndash;9 years who were registered residents of Seoul and received medical care in hospitals located in Seoul, Gyeonggi, or Incheon, as these areas share the same metropolitan living sphere, between January 2012 and December 2019, KD cases were defined using the Korean Standard Classification of Diseases (KCD) code M30.3 together with administration of IVIG. For Kawasaki disease, the KCD code is identical to the International Classification of Diseases, 10th Revision (ICD-10) code.\u003c/p\u003e\u003cp\u003eSeoul covers an area of approximately 605 km\u0026sup2; and had a population of around 9\u0026ndash;10\u0026nbsp;million during the study period. The city is one of the most densely populated and highly urbanized metropolitan areas in the world, with a built environment dominated by residential, commercial, and service sectors. There are no large-scale agricultural or industrial complexes within the city boundary, and the population is predominantly ethnically homogeneous.\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eExposure assessment\u003c/h3\u003e\n\u003cp\u003eDaily mean temperature data for Seoul were obtained from the Korea Meteorological Administration (KMA). Temperature was measured at the Seoul weather station located in Jongno-gu. Daily averages were calculated from hourly observations according to standard national meteorological protocols. The citywide daily mean temperature was used as the exposure variable, and the procedure was identical to that applied in a previous study\u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\n\u003ch3\u003eStudy design and statistical analysis\u003c/h3\u003e\n\u003cp\u003eWe described the demographic characteristics of the study population, including age and sex distributions. In addition, we presented the annual mean daily cases of KD for each study year from 2012 to 2019, as well as the monthly mean daily cases calculated across all years for each calendar month.\u003c/p\u003e\u003cp\u003eWe applied a time-stratified case-crossover design and a time-series regression analysis. In the case-crossover design, the diagnosis date for each patient was defined as the case day, and all days falling on the same day of the week within the same calendar month and year served as control days. This time-stratified referent selection included both days preceding and following the event day. We applied an extended distributed lag non-linear model (DLNM) framework within the case-crossover design, allowing simultaneous assessment of nonlinear exposure\u0026ndash;response relationships and distributed lag effects. Conditional logistic regression models with DLNM specifications were fitted to estimate odds ratios (ORs) and 95% confidence intervals (CIs). Cross-basis functions for daily mean temperature were modeled using natural cubic splines, with knots at the 2.5th percentile (\u0026ndash;6.9\u0026deg;C) and the 97.5th percentile (29.7\u0026deg;C). We chose 25.0\u0026deg;C as the reference temperature since it corresponds to a neutral operative condition within the thermoneutral zone and falls squarely in the comfort region for typical activity and attire\u003csup\u003e\u003cspan additionalcitationids=\"CR13\" citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e. Lags from 0 to 21 days were considered, and both the exposure\u0026ndash;response and lag\u0026ndash;response dimensions were modeled flexibly using spline functions, thereby accounting for nonlinear as well as constrained lag effects. Effect sizes were expressed in terms of odds ratios (ORs).\u003cdiv id=\"Equa\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equa\" name=\"EquationSource\"\u003e\n$$\\:logit\\left({p}_{\\left\\{it\\right\\}}\\right)=\\:{\\alpha\\:}_{i}+\\:cb\\left({T}_{\\left\\{it\\right\\}},\\:l\\right)$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003cem\u003ep\u003c/em\u003e\u003csub\u003e\u003cem\u003e{it}\u003c/em\u003e\u003c/sub\u003e is the probability that day t is the case day for subject \u003cem\u003ei\u003c/em\u003e. \u003cem\u003eα\u003c/em\u003e\u003csub\u003e\u003cem\u003ei\u003c/em\u003e\u003c/sub\u003e is the stratum specific intercept for each subject. \u003cem\u003eT\u003c/em\u003e\u003csub\u003e\u003cem\u003eit\u003c/em\u003e\u003c/sub\u003e is the daily mean temperature on day \u003cem\u003et\u003c/em\u003e. \u003cem\u003ecb(T\u003c/em\u003e\u003csub\u003e\u003cem\u003e{it}\u003c/em\u003e\u003c/sub\u003e, \u003cem\u003el)\u003c/em\u003e is the cross basis function of temperature \u003cem\u003eT\u003c/em\u003e\u003csub\u003e\u003cem\u003eit\u003c/em\u003e\u003c/sub\u003e with lag l from 0 to 21 days.\u003c/p\u003e\u003cp\u003eWe modeled the variation in risk across lags by comparing temperatures at the 2.5 percentile (\u0026ndash;6.9\u0026deg;C) and the 97.5 percentile (29.7\u0026deg;C) with a reference of 25.0\u0026deg;C. For lag intervals where significant associations were observed, we applied a reduced-basis approach to evaluate the overall cumulative effects.\u003c/p\u003e\u003cp\u003eIn the time-series analysis, daily counts of KD cases were modeled using quasi-Poisson regression with a DLNM framework, producing rate ratios (RRs) and 95% CIs. Calendar time (day) was modeled with a natural cubic spline using 7 degrees of freedom (df) per year, consistent with prior work that typically applies 7 df for annual temporal control and is widely used in environmental health research on daily morbidity\u003csup\u003e\u003cspan additionalcitationids=\"CR16 CR17 CR18\" citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e. Indicator variables for day-of-week, holidays (including Saturdays and Sundays), the day before holidays, and the day after holidays were included. The maximum lag was set to 21 days, consistent with the specification used in the case-crossover analysis. To facilitate comparability with the case-crossover analysis, we also applied reduced-basis models in the time-series framework, using the same lag interval as in the case-crossover analysis. Effect sizes were expressed in terms of risk ratios (RRs).\u003cdiv id=\"Equb\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equb\" name=\"EquationSource\"\u003e\n$$\\:{Y}_{t}\\sim\\:quasiPoisson\\left(m{u}_{t}\\right)$$\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Equc\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equc\" name=\"EquationSource\"\u003e\n$$\\:log\\left({\\mu\\:}_{t}\\right)=\\:\\alpha\\:+\\:cb\\left({T}_{t},\\:l\\right)+\\:ns\\left(time,df=\\:7\\:\\times\\:years\\right)+\\:DO{W}_{t}+\\:{H}_{\\{0,t\\}}+\\:{H}_{\\left\\{-1,t\\right\\}}+\\:{H}_{\\left\\{+1,t\\right\\}}$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003cem\u003eY\u003c/em\u003e\u003csub\u003e\u003cem\u003et\u003c/em\u003e\u003c/sub\u003e is the observed number of KD cases on day \u003cem\u003et\u003c/em\u003e. \u003cem\u003e\u0026micro;\u003c/em\u003e\u003csub\u003e\u003cem\u003et\u003c/em\u003e\u003c/sub\u003e is the expected number of cases. \u003cem\u003ecb(T\u003c/em\u003e\u003csub\u003e\u003cem\u003et\u003c/em\u003e\u003c/sub\u003e, \u003cem\u003el)\u003c/em\u003e is the cross basis for daily mean temperature with lag 0 to 21 days. \u003cem\u003ens(time, 7\u0026times;years)\u003c/em\u003e is a natural cubic spline of calendar time with 7 degrees of freedom per year. \u003cem\u003eDOW\u003c/em\u003e\u003csub\u003e\u003cem\u003et\u003c/em\u003e\u003c/sub\u003e is the indicator variables for day of week. \u003cem\u003eH\u003c/em\u003e\u003csub\u003e\u003cem\u003e{0,t}\u003c/em\u003e\u003c/sub\u003e is the indicator for holidays. \u003cem\u003eH\u003c/em\u003e\u003csub\u003e\u003cem\u003e{\u0026minus;1,t}\u003c/em\u003e\u003c/sub\u003e is the indicator for the day before a holiday. \u003cem\u003eH\u003c/em\u003e\u003csub\u003e\u003cem\u003e{+1,t}\u003c/em\u003e\u003c/sub\u003e is the indicator for the day after a holiday. When \u003cem\u003eH\u003c/em\u003e\u003csub\u003e\u003cem\u003e{0,t}\u003c/em\u003e\u003c/sub\u003e takes the value of 1, the other two variables are set to 0.\u003c/p\u003e\n\u003ch3\u003eSubgroup analysis\u003c/h3\u003e\n\u003cp\u003eWe additionally conducted subgroup analyses by age group to examine potential effect modification. The study population was stratified into three categories: children aged\u0026thinsp;\u0026lt;\u0026thinsp;1 year, 1\u0026ndash;4 years, and 5\u0026ndash;9 years. Both the case-crossover and time-series models were fitted separately within each subgroup, using the same modeling specifications as in the main analysis.\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eEthical statements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study was approved by the Institutional Review Board of Dongsan Medical Center (DSMC IRB No. 2025-02-010). As only non-identifiable data were used, the requirement for informed consent was waived.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003eA total of 22,727 KD cases were identified between 2012 and 2019. The mean age of patients was 2.35 \u0026plusmn; 1.70 years, and 13,077 (57.5%) were male. Subgroup distribution by age showed that 4,968 (21.9%) were \u0026lt;1 year, 14,917 (65.6%) were 1\u0026ndash;4 years, and 2,842 (12.5%) were 5\u0026ndash;9 years (Table 1). The mean daily cases among all periods was 7.78 \u0026plusmn; 3.66 cases per day. The monthly average number of Kawasaki disease diagnoses is presented in Figure 1 alongside the mean temperature, showing peaks in December\u0026ndash;January and June. These seasonal variations appear to occur independently of temperature changes. The 2.5th percentile of temperature was \u0026ndash;6.9 \u0026deg;C, and the 97.5th percentile was 29.7 \u0026deg;C, among 2012\u0026ndash;2019. In Figure 2, the regression line estimated using a natural spline is presented, and, in line with the monthly fluctuations observed earlier, seasonality is suggested.\u003c/p\u003e\n\u003cp\u003eIn the case-crossover analysis, the DLNM results are shown in Figure 3. The contour plot of the exposure\u0026ndash;lag\u0026ndash;response surface illustrates the overall pattern of the association. The overall cumulative effect across lag 0\u0026ndash;21 days did not show statistically significant associations at temperatures lower or higher than 25.0 \u0026deg;. When examining lag-specific effects of 29.7 \u0026deg;C compared with 25.0 \u0026deg;C, a significant increase in risk was observed beginning at lag 14 days (OR 1.01, 95%CI 1.00\u0026ndash;1.02 at 14-day lag point), with the risk continuing to rise with increasing lag. In contrast, for \u0026ndash;6.9 \u0026deg;C compared with 25.0 \u0026deg;C, no significant associations were found across any lag period. The reduced-basis analysis restricted to lag 14\u0026ndash;21 days indicated that, relative to 25 \u0026deg;C, the risk increased in an approximately linear and statistically significant manner at temperatures exceeding 26.6 \u0026deg;C (OR 1.02, 95% CI 1.00\u0026nbsp;\u0026ndash;1.05 at 26.6 \u0026deg;C). When calculating the cumulative effect over lag 14\u0026ndash;21 days, the OR at 29.7 \u0026deg;C compared with 25 \u0026deg;C was estimated to be 1.13 (95% CI, 1.03\u0026ndash;1.23) (Table 2).\u003c/p\u003e\n\u003cp\u003eIn the time-series regression analysis, corresponding DLNM results are presented in Figure 4. The overall cumulative effect across lag 0\u0026ndash;21 days demonstrated a significant decrease in KD cases at temperatures below 25 \u0026deg;C and a significant increase at higher temperatures. For 29.7 \u0026deg;C compared with 25 \u0026deg;C, risk elevation was evident from around lag 5 days and continued to increase with longer lags (RR: 1.18, 95% CI: 1.05\u0026ndash;1.33). For \u0026ndash;6.9 \u0026deg;C compared with 25 \u0026deg;C, a significant decrease in risk was observed (RR: 0.52, 95% CI: 0.30\u0026ndash;0.89). The reduced-basis analysis for lag 14\u0026ndash;21 days yielded results consistent with the case-crossover analysis, showing a significant increase in risk above 25 \u0026deg;C. In the cumulative effect over lag 14\u0026ndash;21 days, no significant association was observed at \u0026ndash;6.9 \u0026deg;C (RR 0.52, 95% CI: 0.30\u0026ndash;0.89). The overall cumulative RR at 29.7 \u0026deg;C compared with 25 \u0026deg;C was 1.18 (95% CI: 1.05\u0026ndash;1.33). The cumulative RR over lag 14\u0026ndash;21 days was 1.15 (95% CI: 1.07\u0026ndash;1.24) (Table 2).\u003c/p\u003e\n\u003cp\u003eIn the subgroup analysis by age, results for the 1\u0026ndash;4 year group were consistent with those in the overall population, showing significant risk elevation at higher temperatures during lag 14\u0026ndash;21 days. In contrast, no significant associations were identified in children aged \u0026lt;1 year. For those aged 5\u0026ndash;9 years, the time-series design indicated a significant increase in risk, whereas the case-crossover design did not show a significant association. The analyses stratified by age group, along with the corresponding figures, are presented in the supplementary material.\u0026nbsp;\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eIn this study, we evaluated the association between ambient temperature and KD case counts using two complementary statistical approaches: a time-stratified case-crossover design and a time-series regression analysis. Each method has distinct strengths, with the case-crossover design offering efficient control for individual-level fixed confounders and time-invariant characteristics, while the time-series framework provides a population-level perspective with flexible adjustment for long-term and seasonal trends. By combining these approaches, we sought to maximize their respective advantages and compensate for their limitations. Although simple inspection of monthly seasonality did not reveal a clear relationship with temperature, both analyses consistently demonstrated that high ambient temperature was associated with an increased risk of KD after a delayed period. Specifically, at lag days 14\u0026ndash;21, exposure to the 97.5th percentile of daily mean temperature (29.7 \u0026deg;C) compared with the reference of 25 \u0026deg;C was significantly associated with elevated ORs and RRs.\u003c/p\u003e\n\u003cp\u003eThis pattern was most prominent in children aged 1\u0026ndash;4 years, highlighting a particularly susceptible subgroup within the pediatric population. Among children aged five years and older, the time-series analysis revealed a significant increase in KD risk at temperatures exceeding 25 \u0026deg;C, whereas this association was not observed in the case-crossover design. This inconsistency highlights the need for further investigations with larger sample sizes and refined study designs that incorporate multiple potential confounders. In contrast, no significant temperature effect was detected among infants younger than one year, which may be explained by their limited outdoor activity and the tendency for this age group to remain in air-conditioned indoor environments\u003csup\u003e20\u003c/sup\u003e that reduce actual exposure during periods of elevated ambient temperature. Moreover, infants under one year typically remain at home under parental care, whereas children beyond the first year of life often enter daycare or other group settings\u003csup\u003e21, 22\u003c/sup\u003e, where increased social contact and institutional childcare may modify patterns of infectious disease exposure.\u003c/p\u003e\n\u003cp\u003eThe use of two complementary statistical approaches in this study is of particular methodological importance. When exposure is common to all subjects on a given day, the case-crossover and time-series frameworks are grounded in closely related statistical principles and, under appropriate control for temporal confounding, can yield comparable effect estimates\u0026mdash;particularly when a time-stratified referent selection (same day of week within the same month and year) is used to avoid overlap bias. However, estimates may diverge depending on referent selection in the case-crossover design and on the degrees of freedom chosen for spline functions to adjust long-term and seasonal trends in the time-series approach\u003csup\u003e23\u003c/sup\u003e. The case-crossover design evaluates short-term environmental effects by comparing each subject to him-/herself, thereby removing confounding by time-invariant individual characteristics\u003csup\u003e24\u003c/sup\u003e. In this study, referent days were selected as the same day of week within the same calendar month and year as the event day\u0026mdash;a strategy that controls for seasonality, long-term trends, and day-of-week effects and generally yields a balanced distribution of referents around the case day\u003csup\u003e25\u003c/sup\u003e. A practical limitation is that, unless explicitly modeled, short-term fluctuations in healthcare utilization related to holidays and adjacent days may influence case ascertainment in the case-crossover setting. In parallel, the time-series framework provides population-level inference with high statistical efficiency by using all daily cases; calendar time was modeled with natural cubic splines to address long-term and seasonal trends \u003csup\u003e26\u003c/sup\u003e, and indicators for holidays and adjacent days were included to capture short-term utilization shifts. Nonetheless, results in the time-series approach can be sensitive to the selection of degrees of freedom and other modeling choices\u003csup\u003e23\u003c/sup\u003e. By combining both approaches, we aimed to capitalize on their strengths while mitigating their respective limitations.\u003c/p\u003e\n\u003cp\u003eThe incorporation of distributed lag nonlinear models (DLNM) into both case-crossover and time-series frameworks represents a key methodological strength of this study. DLNM allow simultaneous modeling of nonlinear exposure\u0026ndash;response relationships and complex lag structures, thereby enabling a realistic characterization of delayed environmental effects\u003csup\u003e27\u003c/sup\u003e. Previous epidemiological investigations have reported significant associations between ambient temperature and KD incidence, suggesting that extreme temperature conditions may act as risk\u003csup\u003e9, 10, 28\u003c/sup\u003e. Moreover, Kawasaki disease exhibits marked seasonality across countries modifiers\u003csup\u003e28, 29\u003c/sup\u003e a pattern plausibly linked to temperature fluctuations and the epidemic dynamics of infectious diseases. Although some studies, including those from Japan, have identified significant associations within shorter lags of 0\u0026ndash;5 days\u003csup\u003e8, 10\u003c/sup\u003e, the etiological hypothesis that KD may be preceded by infections\u003csup\u003e30\u003c/sup\u003e and the diagnostic requirement of at least five days of fever\u003csup\u003e31\u003c/sup\u003e,\u0026nbsp;indicate that longer lag periods should be considered. In this study, we explicitly extended the lag structure up to 21 days and employed reduced-basis models to examine cumulative effects within specific intervals. This approach revealed significant associations during lag days 14\u0026ndash;21, highlighting a critical window of vulnerability. Unlike previous studies that focused on shorter lags, our findings emphasize that elevated temperatures may influence KD occurrence even over comparatively long lag periods.\u003c/p\u003e\n\u003cp\u003eAn additional strength of this study lies in the use of a comprehensive claims database derived from the Korean National Health Insurance system, which covers virtually the entire population\u003csup\u003e32, 33\u003c/sup\u003e. Case definition required both an ICD-10 diagnosis code for KD and administration of IVIG, thereby enhancing specificity. By restricting the analysis to residents of Seoul, we were able to achieve near-complete case ascertainment within a well-defined urban population. The compact geographic area of Seoul reduced variation in individual exposure relative to the single meteorological monitoring site, while the city\u0026rsquo;s high population density ensured a sufficiently large sample size for robust analysis. Although data derived from a single country or region may have limitations in terms of generalizability, they provide valuable information for estimating the health and economic burden of temperature within that setting. In particular, the near-complete data from Seoul enabled direct estimation of the health impact of rising temperatures in an urban Korean population, offering a strong basis for projecting risks in other metropolitan areas of Korea with similar demographic and environmental characteristics.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eSeveral limitations should be considered. First, changes in the diagnostic criteria for KD in 2017\u003csup\u003e31, 34\u003c/sup\u003e. may have affected temporal patterns of case identification. Such changes influence only the outcome and not the exposure, and their potential impact on short-term associations is likely to have been minimized by our methodological approaches, namely the time-stratified case-crossover design and the adjustment for seasonality and long-term trends using natural splines in the time-series analysis. We did not adjust for other daily environmental exposures such as air pollution. However, these factors are more plausibly mediators, as they are influenced by temperature and subsequently affect KD risk, rather than independent confounders, since it is unlikely that they directly alter variations in ambient temperature. Under modern epidemiologic theory, a true confounder is defined as a factor associated with both the exposure and the outcome\u003csup\u003e35\u003c/sup\u003e. In contrast, variables that affect only the outcome (e.g., changes in diagnostic criteria) or operate as mediators (e.g., air pollutants) influenced by the exposure do not bias the estimated effect size or direction of the exposure\u0026ndash;outcome association. Therefore, while not adjusting for these factors may have led to some overestimation of the direct effect of temperature, it is unlikely to have materially affected our estimation of the overall impact attributable to temperature. Additionally, because we assigned the same daily ambient temperature exposure to all cases occurring on a given day, we were unable to take advantage of higher-resolution exposure data to assess individual-level variation, which would have maximized the strengths of the case-crossover design. This limitation also reflects the inability to account for personal-level exposures, such as time spent indoors or the use of cooling and heating systems, thereby restricting prediction of temperature-related risks at the individual level. Nevertheless, the approach provides valuable insights into the broader epidemiological and public health implications of rising ambient temperatures at the regional scale.\u003c/p\u003e"},{"header":"Conclusions","content":"\u003cp\u003eThis study provides novel evidence that elevated ambient temperature is associated with an increased risk of Kawasaki disease, with risks rising above the reference temperature of 25 \u0026deg;C, particularly during lag days 14\u0026ndash;21 and most prominently among children aged 1\u0026ndash;4 years. These findings highlight the potential impact of rising temperatures on pediatric inflammatory diseases and underscore the importance of incorporating climate factors into public health preparedness and policy planning, especially in metropolitan settings such as Seoul.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgment\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was supported by the Dongguk University, College of Medicine Research Fund of 2025.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDisclosure\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare no conflicts of interest.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthical statements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study was approved by the Institutional Review Board of Dongsan Medical Center (DSMC IRB No. 2025-02-010). As only non-identifiable data were used, the requirement for informed consent was waived.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contribution\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eJung Hee Hong: Conceptualization, Investigation, Validation, Writing- Reviewing and Editing\u003c/p\u003e\n\u003cp\u003eKiook Baek: Visualization, Methodology, Data curation, Writing- Original draft preparation.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDeclaration of generative AI in scientific writing\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDuring the preparation of this work the authors used ChatGPT in order to English translation and polishing. After using this tool/service, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eNewburger JW, Takahashi M, Burns JC. Kawasaki disease. \u003cem\u003eJournal of the American College of Cardiology. \u003c/em\u003e2016;67(14):1738-1749.\u003c/li\u003e\n\u003cli\u003eUehara R, Belay ED. Epidemiology of Kawasaki Disease in Asia, Europe, and the United States. \u003cem\u003eJournal of Epidemiology. \u003c/em\u003e2012;22(2):79-85.\u003c/li\u003e\n\u003cli\u003eLegg S. IPCC, 2021: Climate change 2021-the physical science basis. \u003cem\u003eInteraction. \u003c/em\u003e2021;49(4):44-45.\u003c/li\u003e\n\u003cli\u003eLafferty KD. The ecology of climate change and infectious diseases. \u003cem\u003eEcology. \u003c/em\u003e2009;90(4):888-900.\u003c/li\u003e\n\u003cli\u003eRohr JR, Cohen JM. Understanding how temperature shifts could impact infectious disease. \u003cem\u003ePLoS biology. \u003c/em\u003e2020;18(11):e3000938.\u003c/li\u003e\n\u003cli\u003eNawa N, Nishimura H, Fushimi K, Fujiwara T. Association between heat exposure and intussusception in children in Japan from 2011 to 2022. \u003cem\u003ePediatr Res. \u003c/em\u003e2025:1-5.\u003c/li\u003e\n\u003cli\u003eBaek K, Park C. Analyzing the dynamics of complicated and uncomplicated appendicitis during the COVID-19 pandemic in Seoul, Korea: a multifaceted time series approach. \u003cem\u003eEpidemiology and Health. \u003c/em\u003e2024;46:e2024081.\u003c/li\u003e\n\u003cli\u003eNawa N, Nishimura H, Fushimi K, Fujiwara T. Association between heat exposure and Kawasaki disease: A time-stratified case-crossover study. \u003cem\u003eEnviron Res. \u003c/em\u003e2024;263:120231.\u003c/li\u003e\n\u003cli\u003eCelis-Seposo AK, Madaniyazi L, Seposo X, Hashizume M, Yoshida LM, Toizumi M. Incidence and seasonality of Kawasaki disease in children in the Philippines, and its association with ambient air temperature. \u003cem\u003eFrontiers in Pediatrics. \u003c/em\u003e2024;12:1358638.\u003c/li\u003e\n\u003cli\u003eFujii F, Egami N, Inoue M, Koga H. Weather condition, air pollutants, and epidemics as factors that potentially influence the development of Kawasaki disease. \u003cem\u003eSci Total Environ. \u003c/em\u003e2020;741:140469.\u003c/li\u003e\n\u003cli\u003eLin Z, Meng X, Chen R\u003cem\u003e, et al.\u003c/em\u003e Ambient air pollution, temperature and Kawasaki disease in Shanghai, China. \u003cem\u003eChemosphere. \u003c/em\u003e2017;186:817-822.\u003c/li\u003e\n\u003cli\u003eYin Q, Zhang Y, Liu Y. Investigation on Thermal Comfort and Thermal Adaptive Behaviors of Rural Residents in Suibin Town, China, in Summer. \u003cem\u003eSustainability. \u003c/em\u003e2023;15(8):6630.\u003c/li\u003e\n\u003cli\u003eKingma BR, Frijns AJ, Schellen L, van Marken Lichtenbelt WD. Beyond the classic thermoneutral zone: Including thermal comfort. \u003cem\u003eTemperature (Austin). \u003c/em\u003eJul-Sep 2014;1(2):142-149.\u003c/li\u003e\n\u003cli\u003eMora R, Bean R. Thermal comfort: Designing for people. \u003cem\u003eASHRAE J. \u003c/em\u003e2018;60(2):40-46.\u003c/li\u003e\n\u003cli\u003eYang Y, Li X, Wang S\u003cem\u003e, et al.\u003c/em\u003e Assessing the impact of temperature on acute exacerbation of chronic obstructive pulmonary disease hospitalizations in residents of Panzhihua City: a multi-districts study using a distributed lag non-linear model. \u003cem\u003eBMC Public Health. \u003c/em\u003e2024;24(1):2151.\u003c/li\u003e\n\u003cli\u003eGao S, Yang T, Zhang X\u003cem\u003e, et al.\u003c/em\u003e A longitudinal study on the effect of extreme temperature on non-accidental deaths in Hulunbuir City based on DLNM model. \u003cem\u003eInt Arch Occup Environ Health. \u003c/em\u003e2023;96(7):1009-1014.\u003c/li\u003e\n\u003cli\u003ePeng RD, Dominici F, Louis TA. Model Choice in Time Series Studies of Air Pollution and Mortality. \u003cem\u003eJournal of the Royal Statistical Society Series A: Statistics in Society. \u003c/em\u003e2006;169(2):179-203.\u003c/li\u003e\n\u003cli\u003eBhaskaran K, Gasparrini A, Hajat S, Smeeth L, Armstrong B. Time series regression studies in environmental epidemiology. \u003cem\u003eInt J Epidemiol. \u003c/em\u003eAug 2013;42(4):1187-1195.\u003c/li\u003e\n\u003cli\u003eDominici F, Samet JM, Zeger SL. Combining evidence on air pollution and daily mortality from the 20 largest US cities: a hierarchical modelling strategy. \u003cem\u003eJournal of the Royal Statistical Society Series A: Statistics in Society. \u003c/em\u003e2000;163(3):263-302.\u003c/li\u003e\n\u003cli\u003eYoon H, Seo J, Kim T\u003cem\u003e, et al.\u003c/em\u003e Development of Korean exposure factors for children in Korea. 2017.\u003c/li\u003e\n\u003cli\u003eAhn J, Shin N. The use of child care center for infants of dual-working families in Korea. \u003cem\u003eChildren and Youth Services Review. \u003c/em\u003e2013;35(9):1510-1519.\u003c/li\u003e\n\u003cli\u003eYun-Jin B. ECEC Statistics of KOREA: Recent trends of Services, Enrollment, and Workforce. 2023.\u003c/li\u003e\n\u003cli\u003eLu Y, Zeger SL. On the equivalence of case-crossover and time series methods in environmental epidemiology. \u003cem\u003eBiostatistics. \u003c/em\u003e2007;8(2):337-344.\u003c/li\u003e\n\u003cli\u003eMaclure M. The case-crossover design: a method for studying transient effects on the risk of acute events. \u003cem\u003eAmerican journal of epidemiology. \u003c/em\u003e1991;133(2):144-153.\u003c/li\u003e\n\u003cli\u003eSzyszkowicz M. Case-Crossover Method with a Short Time-Window. \u003cem\u003eInt J Environ Res Public Health. \u003c/em\u003eDec 27 2019;17(1).\u003c/li\u003e\n\u003cli\u003eKim H, Lee J-T, Fong KC, Bell ML. Alternative adjustment for seasonality and long-term time-trend in time-series analysis for long-term environmental exposures and disease counts. \u003cem\u003eBMC Medical Research Methodology. \u003c/em\u003e2021;21(1):2.\u003c/li\u003e\n\u003cli\u003eGasparrini A. Distributed lag linear and non-linear models in R: the package dlnm. \u003cem\u003eJournal of statistical software. \u003c/em\u003e2011;43:1-20.\u003c/li\u003e\n\u003cli\u003eLin M-T, Wu M-H. The global epidemiology of Kawasaki disease: review and future perspectives. \u003cem\u003eGlobal cardiology science \u0026amp; practice. \u003c/em\u003e2017;2017(3):e201720.\u003c/li\u003e\n\u003cli\u003eBurns JC, Cayan DR, Tong G\u003cem\u003e, et al.\u003c/em\u003e Seasonality and temporal clustering of Kawasaki syndrome. \u003cem\u003eEpidemiology. \u003c/em\u003e2005;16(2):220-225.\u003c/li\u003e\n\u003cli\u003eBurgner D, Harnden A. Kawasaki disease: what is the epidemiology telling us about the etiology? \u003cem\u003eInternational journal of infectious diseases. \u003c/em\u003e2005;9(4):185-194.\u003c/li\u003e\n\u003cli\u003eMcCrindle BW, Rowley AH, Newburger JW\u003cem\u003e, et al.\u003c/em\u003e Diagnosis, Treatment, and Long-Term Management of Kawasaki Disease: A Scientific Statement for Health Professionals From the American Heart Association. \u003cem\u003eCirculation. \u003c/em\u003eApr 25 2017;135(17):e927-e999.\u003c/li\u003e\n\u003cli\u003eKim L, Kim J-A, Kim S. A guide for the utilization of health insurance review and assessment service national patient samples. \u003cem\u003eEpidemiology and health. \u003c/em\u003e2014;36:e2014008.\u003c/li\u003e\n\u003cli\u003eCheol Seong S, Kim Y-Y, Khang Y-H\u003cem\u003e, et al.\u003c/em\u003e Data resource profile: the national health information database of the National Health Insurance Service in South Korea. \u003cem\u003eInternational journal of epidemiology. \u003c/em\u003e2017;46(3):799-800.\u003c/li\u003e\n\u003cli\u003eKobayashi T, Ayusawa M, Suzuki H\u003cem\u003e, et al.\u003c/em\u003e Revision of diagnostic guidelines for Kawasaki disease. \u003cem\u003ePediatrics international. \u003c/em\u003e2020;62(10).\u003c/li\u003e\n\u003cli\u003eTennant PW, Murray EJ, Arnold KF\u003cem\u003e, et al.\u003c/em\u003e Use of directed acyclic graphs (DAGs) to identify confounders in applied health research: review and recommendations. \u003cem\u003eInt J Epidemiol. \u003c/em\u003e2021;50(2):620-632.\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003eTable 1. General characteristics of participants\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"277\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 79px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003eNumber\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 72px;\"\u003e\n \u003cp\u003e(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 79px;\"\u003e\n \u003cp\u003eTotal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e22,727\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 72px;\"\u003e\n \u003cp\u003e(100)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 79px;\"\u003e\n \u003cp\u003eSex\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 127px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 72px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 79px;\"\u003e\n \u003cp\u003e\u0026nbsp; Male\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e13,077\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 72px;\"\u003e\n \u003cp\u003e(64.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 79px;\"\u003e\n \u003cp\u003e\u0026nbsp; Female\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e9,650\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 72px;\"\u003e\n \u003cp\u003e(35.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 79px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 127px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 72px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 79px;\"\u003e\n \u003cp\u003eAge\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e2.35 \u0026plusmn; 1.70\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 72px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 79px;\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026lt;1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e4,968\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 72px;\"\u003e\n \u003cp\u003e(23.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 79px;\"\u003e\n \u003cp\u003e\u0026nbsp; 1\u0026ndash;4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e14,917\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 72px;\"\u003e\n \u003cp\u003e(64.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 79px;\"\u003e\n \u003cp\u003e\u0026nbsp; 5\u0026ndash;9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e2,842\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 72px;\"\u003e\n \u003cp\u003e(11.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" style=\"width: 205px;\"\u003e\n \u003cp\u003eDiagnosed year\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 72px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 79px;\"\u003e\n \u003cp\u003e\u0026nbsp; 2012\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e2,747\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 72px;\"\u003e\n \u003cp\u003e(12.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 79px;\"\u003e\n \u003cp\u003e\u0026nbsp; 2013\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e3,063\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 72px;\"\u003e\n \u003cp\u003e(13.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 79px;\"\u003e\n \u003cp\u003e\u0026nbsp; 2014\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e3,129\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 72px;\"\u003e\n \u003cp\u003e(13.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 79px;\"\u003e\n \u003cp\u003e\u0026nbsp; 2015\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e2,956\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 72px;\"\u003e\n \u003cp\u003e(13.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 79px;\"\u003e\n \u003cp\u003e\u0026nbsp; 2016\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e2,874\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 72px;\"\u003e\n \u003cp\u003e(12.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 79px;\"\u003e\n \u003cp\u003e\u0026nbsp; 2017\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e2,591\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 72px;\"\u003e\n \u003cp\u003e(11.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 79px;\"\u003e\n \u003cp\u003e\u0026nbsp; 2018\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e2,792\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 72px;\"\u003e\n \u003cp\u003e(12.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 79px;\"\u003e\n \u003cp\u003e\u0026nbsp; 2019\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e2,575\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 72px;\"\u003e\n \u003cp\u003e(11.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eTable 2. Effect estimates at the 2.5 and 97.5 percentiles of temperature compared with 25 \u0026deg;C in the case-crossover and time-series models\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"853\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 196px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd colspan=\"4\" style=\"width: 657px;\"\u003e\n \u003cp\u003eCumulative lags (days)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 196px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 329px;\"\u003e\n \u003cp\u003e0\u0026ndash;21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 329px;\"\u003e\n \u003cp\u003e14\u0026ndash;21\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 196px;\"\u003e\n \u003cp\u003eTemperature (reference: 25\u0026deg;C)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 159px;\"\u003e\n \u003cp\u003e\u0026minus;6.9 \u0026deg;C (2.5 percentile)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 170px;\"\u003e\n \u003cp\u003e29.7 \u0026deg;C (97.5 percentile)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 159px;\"\u003e\n \u003cp\u003e\u0026minus;6.9 \u0026deg;C (2.5 percentile)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 170px;\"\u003e\n \u003cp\u003e29.7 \u0026deg;C (97.5 percentile)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 196px;\"\u003e\n \u003cp\u003eCase-crossover (OR, 95% CI)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 159px;\"\u003e\n \u003cp\u003e1.10 (0.86, 1.42)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 170px;\"\u003e\n \u003cp\u003e1.08 (0.92, 1.27)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 159px;\"\u003e\n \u003cp\u003e1.03 (0.84, 1.25)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 170px;\"\u003e\n \u003cp\u003e1.13 (1.03, 1.23)\u003csup\u003e\u0026nbsp;*\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 196px;\"\u003e\n \u003cp\u003eTime-series (RR, 95% CI)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 159px;\"\u003e\n \u003cp\u003e0.52 (0.30, 0.89)\u003csup\u003e\u0026nbsp;*\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 170px;\"\u003e\n \u003cp\u003e1.18 (1.05, 1.32)\u003csup\u003e\u0026nbsp;*\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 159px;\"\u003e\n \u003cp\u003e0.84 (0.64, 1.11)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 170px;\"\u003e\n \u003cp\u003e1.14 (1.05, 1.25)\u003csup\u003e\u0026nbsp;*\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eOR, Odds ratio; RR, Rate ratio\u003c/p\u003e\n\u003cp\u003e\u003csup\u003e*\u003c/sup\u003e: statistically significant (\u0026lt; 0.05)\u003c/p\u003e\n\u003cp\u003eTo obtain effect estimates at specific points, non-linear modeling of lag and exposure variables was performed using a DLNM framework.\u003c/p\u003e\n\u003cp\u003eIn the case-crossover design, control days were selected as other days of the same month, whereas in the time-series analysis, calendar time was adjusted using a natural spline with 7 degrees of freedom per year based on daily counts.\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Kawasaki disease, Ambient temperature, Climate change, Case-crossover study, Time-series analysis","lastPublishedDoi":"10.21203/rs.3.rs-7559001/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7559001/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e\u003cp\u003eKawasaki disease (KD) is an acute vasculitis of childhood and the leading cause of acquired heart disease in developed countries. The impact of ambient temperature on KD incidence remains unclear. We analyzed nationwide claims data for children aged 0\u0026ndash;9 years in Seoul between 2012 and 2019. Associations between ambient temperature and KD were examined using a time-stratified case-crossover design and quasi-Poisson time-series regression within a distributed lag non-linear model framework (lags up to 21 days). Reduced-basis models summarized cumulative effects, and analyses were stratified by age. A total of 28,866 KD cases were identified. In the case-crossover analysis, the cumulative effect over lag 0\u0026ndash;21 days was not significant. However, at lag 14\u0026ndash;21 days, risk increased relatively linearly above 25\u0026deg;C, with an OR of 1.13 (95% CI: 1.03\u0026ndash;1.23) at 29.7\u0026deg;C (97.5th percentile) compared with 25\u0026deg;C, while no significant decrease was observed at \u0026minus;\u0026thinsp;6.9\u0026deg;C (2.5th percentile). In the time-series analysis, risk decreased below 25\u0026deg;C and increased above 25\u0026deg;C over lag 0\u0026ndash;21 days, with a more pronounced cumulative effect at lag 14\u0026ndash;21 days (RR 1.14, 95% CI: 1.05\u0026ndash;1.25 at 29.7\u0026deg;C). Age-stratified analyses showed consistent results in children aged 1\u0026ndash;4 years, while associations were less clear in \u0026lt;\u0026thinsp;1 and 5\u0026ndash;9 years. High ambient temperatures were associated with a delayed increase in KD risk, particularly among children aged 1\u0026ndash;4 years, highlighting the potential influence of climate change on pediatric health.\u003c/p\u003e","manuscriptTitle":"Ambient Temperature and Kawasaki Disease in Seoul, Korea: Evidence from Nationwide HIRA Data Using Case-Crossover and Time-Series Analyses","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-09-17 18:46:57","doi":"10.21203/rs.3.rs-7559001/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"c90d9bcb-e5af-488d-b4fc-848521932198","owner":[],"postedDate":"September 17th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-12-07T21:48:42+00:00","versionOfRecord":[],"versionCreatedAt":"2025-09-17 18:46:57","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7559001","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7559001","identity":"rs-7559001","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}