Introduction
Scrub typhus, also known as tsutsugamushi disease, is a naturally occurring febrile illness caused by the gram-negative bacterium
Orientia tsutsugamushi [
1]. The disease is transmitted to humans through the bite of the larval (chigger) stage of
Leptotrombidium mites (Acari: Trombiculidae) [
2], which act as vectors when infected with the pathogen. Wild rodents serve as both the natural and reservoir hosts of
O. tsutsugamushi, while humans are considered accidental hosts [
3]. The incubation period for scrub typhus is generally 10 to 12 days, although it can range from 6 to 21 days [
4]. Clinical manifestations often include fever, headache, myalgia, rash, eschar formation, generalized lymphadenopathy, and transient hearing loss; however, severe complications such as acute encephalitis syndrome and multiorgan failure can occur [
5]. To date, no effective vaccine is commercially available to prevent or control scrub typhus [
6]. Prompt diagnosis and treatment (most commonly doxycycline) are crucial to reducing complications and mortality [
7].
Scrub typhus is widely prevalent within the “Tsutsugamushi Triangle,” extending north to northern Japan and eastern Russia, south to northern Australia, and west to Pakistan and Afghanistan [
6]. Recent reports indicate spread beyond its traditional geographic boundaries, with evidence of wider distribution in Africa, the Middle East, and South America than previously recognized [
8,
9], highlighting the potential for global expansion. Among the 5 countries that classify scrub typhus as a notifiable infectious disease (Japan, mainland China, the Republic of Korea, Taiwan, and Thailand), the Republic of Korea reported the highest incidence rate (IR) as of 2016 [
10]. Eight species of
Leptotrombidium mites known to carry
O. tsutsugamushi are found in the Republic of Korea [
11]. Among them,
Leptotrombidium scutellare is the dominant species, accounting for 37.7% of all identified mites, and is primarily distributed in the southern regions, including Gyeongnam, Jeonnam, and Jeonbuk [
12,
13].
Leptotrombidium pallidum, the second most prevalent species, accounts for 11.5% and is more commonly found in the central regions, such as Gyeonggi, Gangwon, and Chungbuk [
12,
13]. However, recent surveillance data from the Korea Disease Control and Prevention Agency (KDCA) indicate a northward expansion of
L. scutellare [
12,
13], suggesting emerging risk in regions previously considered low-incidence areas.
As poikilothermic organisms, chigger mites are highly sensitive to climatic factors, which influence their ecology, reproductive capacity, and population dynamics [
14,
15]. A spatial regression study conducted in tropical India reported that a 1 °C increase in mean temperature (Tmean) was associated with an 18.8% decrease in monthly scrub typhus cases, whereas a 1% increase in mean relative humidity (RH) corresponded to a 7.6% increase [
16]. Similarly, in a study from subtropical China applying a spatial lag model, Tmean and mean wind speed were identified as the most influential weather factors determining the risk of scrub typhus [
1]. A study conducted in the Republic of Korea reported peak cumulative risks at 18.7 °C for temperature (relative risk [RR], 9.73; 95% confidence interval [CI], 5.54–17.10) and 162.0 mm for precipitation (RR, 1.87; 95% CI, 1.02–3.83), relative to baselines of 9.7 °C and 2.8 mm, respectively [
2]. However, these studies did not incorporate more recent data and relied on analyses aggregated across the entire study period, limiting their capacity to capture how the impact of climate on scrub typhus incidence may have changed over time, particularly in the context of ongoing climate change.
Therefore, this study aimed to comprehensively describe the epidemiological trends of scrub typhus in the Republic of Korea using data from 2001 to 2024 and to evaluate the relationship between meteorological factors and disease occurrence using a distributed lag non-linear model (DLNM). In doing so, we sought to explore how climatic changes may be influencing the transmission dynamics of scrub typhus in the Republic of Korea.
Materials and Methods
National Surveillance
In the Republic of Korea, scrub typhus is classified as a third-grade notifiable infectious disease and must be reported within 24 hours. This disease has been included in the National Notifiable Disease Surveillance System since 2001. The KDCA established a centralized electronic reporting system, the Integrated Disease Health Management System (IS) in 2005; by 2007, all case notifications had transitioned to a web-based platform. Since 2015, an automated infectious disease reporting support system has been disseminated by integrating electronic medical record systems in healthcare institutions with the IS. In 2024, the KDCA reorganized the IS into the Integrated Disease Prevention Information System to support faster and more efficient quarantine operations; however, the fundamental structure of the existing infectious disease system remained unchanged.
Case Definition
The KDCA revised the reporting and diagnostic criteria for scrub typhus on April 30, 2019. First, clinically suspected cases without laboratory test results were excluded from the reporting requirements [
17]. Second, whereas a single antibody titer of immunoglobulin (Ig) G ≥1:256 or IgM ≥1:16 had been considered positive, the diagnostic criterion was revised to a fourfold or greater increase in antibody titers [
18]. This change was designed to improve diagnostic accuracy by distinguishing between past and recent infections [
18]. Lastly, the previous non-confirmatory criterion, “detection of specific antibodies,” was revised to “detection of specific antibodies in a blood specimen” [
17]. Therefore, a confirmed case was defined as a patient presenting with compatible clinical symptoms and either (1) isolation and identification of
O. tsutsugamushi from specimens (e.g., blood, tissue, or eschar), (2) a fourfold or greater increase in antibody titer between acute and convalescent sera, or (3) detection of
O. tsutsugamushi DNA by polymerase chain reaction from specimens (e.g., blood, tissue, or eschar). A probable case was defined as a patient presenting with compatible clinical symptoms and epidemiological exposure (e.g., outdoor activity or chigger exposure within the 10 days before symptom onset), along with a positive result from either an immunochromatographic assay or a single immunofluorescence assay [
13].
Study Design, Period, and Population
We conducted a retrospective time-series study of monthly scrub typhus incidence and meteorological factors in the Republic of Korea from 2001 to 2024. Of a total of 149,303 reported cases, 149,289 were included in the final study population, consisting of 87,995 probable cases and 61,294 confirmed cases. Fourteen suspected cases were excluded to ensure data quality and representativeness. Mortality data related to scrub typhus have been available since 2011, with a cumulative total of 174 deaths reported as of 2024. To reduce variability and ensure data homogeneity, the study period was divided into 4 phases based on major changes in surveillance systems, climate patterns, and case definitions: phase I (2001–2006), phase II (2007–2012), phase III (2013–2018), and phase IV (2019–2024).
Data Sources and Collection
To examine the incidence of scrub typhus in the Republic of Korea, we used monthly scrub typhus case counts from 2001 to 2024 and mortality data from 2011 to 2024 for 17 administrative regions, as published by the KDCA Infectious Disease Portal (
https://dportal.kdca.go.kr). Because the portal data do not clearly distinguish between suspected and probable cases, we additionally reviewed records through the KDCA IS. Cases without a recorded diagnosis date were classified as suspected cases. Cumulative IR and annual age-standardized incidence rates (ASIR) for each region were calculated using mid-year population estimates obtained from the Korean Statistical Information Service of Statistics Korea (
https://kosis.kr). For spatial analysis of the geographic distribution of scrub typhus, administrative boundary shapefiles were obtained from the Statistical Geographic Information Service of Statistics Korea (
https://sgis.kostat.go.kr). To explore the lagged and cumulative effects of meteorological factors on scrub typhus incidence, meteorological data were acquired from the Korea Meteorological Administration’s Open Data Portal (
https://data.kma.go.kr). We used multisite statistics from at least 2 automated synoptic or disaster weather observation stations per region. The monthly meteorological variables included Tmean, cumulative precipitation, mean RH, mean wind velocity (WV), and total sunshine hours (SH).
Data Analysis
A descriptive analysis of scrub typhus cases and meteorological factors was performed using Microsoft Excel 2019 (Microsoft Corp.). Temporal trends were analyzed with Joinpoint Regression Software (ver. 5.3.0.0; National Cancer Institute). Spearman correlation analysis and time-series regression analysis with DLNM were conducted using R ver. 4.4.3 in RStudio (R Studio Inc.). All statistical tests were 2-sided, and p-values of less than 0.05 were considered to indicate statistical significance.
Descriptive analysis
Annual national incidence and mortality trends for scrub typhus were visualized using frequency analyses. Demographic characteristics of cases were summarized as percentages.
Monthly meteorological data from the 17 administrative regions were aggregated by phase, and descriptive statistics—including mean, standard deviation, percentiles (P10, P25, P50, P75, and P90), minimum, and maximum—were calculated for each climatic variable to characterize the nationwide distribution during each phase.
Joinpoint regression analysis
To identify changes and parallel trends, Joinpoint analysis was conducted using ASIR per 100,000 population, calculated by the direct standardization method using the World Health Organization World Standard Population for 2000–2025. Because the analysis yielded zero joinpoints, the average annual percent change (AAPC) can be interpreted in the same way as the annual percent change. An upward trend was indicated by a positive AAPC with the lower bound of its 95% CI greater than zero.
Correlation analysis
To examine the relationship between meteorological variables and scrub typhus incidence, Spearman correlation analysis was used due to the non-normal distribution of the data and the presence of non-linear relationships. Multicollinearity among meteorological variables was assessed using variance inflation factors, all of which were below 5, indicating no significant multicollinearity [
19]. Given that scrub typhus cases in the Republic of Korea typically peak in autumn (October–November), seasonality was addressed by pairing observations by calendar month, and Spearman's rank correlation coefficients were calculated.
Time-series regression analysis
To evaluate the lagged and cumulative effects of meteorological exposure on scrub typhus incidence, we implemented a DLNM framework using the “dlnm” package in R, built on a generalized linear model. DLNM is a flexible approach that captures both non-linear and delayed effects of predictors on an outcome over time [
20]. In this study, meteorological factors (e.g., Tmean) were treated as independent variables, and monthly scrub typhus incidence represented the dependent variable. Both single-variable and multi-variable DLNM models were employed to assess the effects of individual meteorological variables. The single-variable models incorporated seasonal and temporal trends, incidence quantile groups, and 1-month lagged incidence to adjust for potential confounders [
21]. In the multi-variable models, all meteorological variables and covariates—including time trends, seasonality, incidence quantiles, and prior-month incidence—were adjusted simultaneously [
21]. The median value of each meteorological variable was set as the reference exposure. Low- and high-risk exposure levels were defined by P10 and P90, respectively, which also served as thresholds for estimating both lagged and cumulative RR.
Prior to model fitting, we defined a 2-dimensional function space using a cross-basis function. Overdispersion was evaluated by phase for each meteorological variable; where dispersion exceeded 1, the quasi-Poisson distribution was adopted. Precipitation was excluded due to skewed distribution, zero inflation, and failure to meet model convergence criteria within the P10–P90 range.
We tested lag periods ranging from 3 to 7 months and selected, for each phase, the lag duration that yielded the lowest quasi-Bayesian information criterion in the multi-variable DLNM model [
21], ensuring optimal model fit and capturing phase-specific temporal dynamics. While the overall model (2001–2024) adopted a 5-month lag, the optimal lags for the phase-specific models were 4, 7, 5, and 6 months for phases I–IV, respectively. Natural cubic splines of time with 7 degrees of freedom (df) per year were used to control for confounders by capturing long-term trends and seasonality [
22]. The optimal df for the natural spline functions of the meteorological variables and their lag-response relationships was determined by comparing quasi-Akaike information criterion (QAIC) values across models with df from 3 to 5; a lower QAIC indicates a better fit [
22]. In the final model, the lag dimension was set to 4 df for phase I and 3 df for phases II–IV. The B-spline function for the exposure dimension was fixed at 3 df, based on previous research [
23].
Knots for the lag dimension were placed at equally spaced values on a logarithmic scale using the logknots() function. Lagged and cumulative RR values at the P10 and P90 exposure levels were estimated and visualized using contour plots and exposure-response curves to illustrate both the temporal patterns of delayed effects and the magnitude of cumulative effects.
IRB/IACUC Approval
This study was reviewed and approved by the Institutional Review Board of the KDCA (No: KDCA-2025-04-04-PE-01) and was performed in accordance with the principles of the Declaration of Helsinki. The requirement for informed consent was waived because the study retrospectively collected de-identified data.
Results
Overall National Incidence and Case Fatality Trends
A total of 149,289 scrub typhus cases were reported in the Republic of Korea from 2001 to 2024, with an average annual IR of 12.4 per 100,000 population. From 2001 to 2017, the annual number of cases and the IR displayed a gradual increase, with intermittent stepwise changes. Since 2018, the total number of cases has declined, with a particularly sharp decrease in confirmed cases observed in 2019 (
Figure 1A). From 2011 to 2024, a total of 174 deaths were reported, corresponding to a case fatality rate of 2.5%. Despite declining incidence in phase Ⅳ, both the number of deaths and the case fatality rate reached their highest levels in 2022 and 2023, with case fatality exceeding 0.3% (
Figure 1B). The AAPC of ASIR decreased from 26.79% per year (95% CI, −2.46% to 63.84%;
p=0.079) in phase I to −6.72% (95% CI, −25.32% to 16.36%;
p=0.439) in phase Ⅲ, representing a gradual slowdown in growth. However, the AAPC in phase Ⅳ was 4.32% (95% CI, −1.83% to 10.68%;
p=0.174), indicating a transition to a rising trend (
Figure 2).
Demographic and Regional Features
The total number of scrub typhus cases was 149,289, including 23,915, 36,500, 56,309, and 32,565 in phase I, phase II, phase Ⅲ, and phase Ⅳ, respectively. Overall, a higher proportion of cases was reported among female individuals (61.8%), although this proportion decreased over time. By age, the proportion of cases among those in their 70s increased from 20.5% in phase I to 28.6% in phase IV, whereas those aged ≥80 years showed a marked increase from 5.2% in phase I to 20.5% in phase IV. Regionally, the highest proportion was observed in Gyeongnam (16.1%), which increased from 12.1% in phase I to 20.6% in phase IV (
Table 1).
Geographical and Temporal Trends
The geographical distribution of scrub typhus in the Republic of Korea showed that IRs per 100,000 population were largely concentrated in the southern, southeastern, and southwestern regions. Jeonnam recorded the highest incidence, with 78.7 cases per 100,000 population in phase Ⅲ and 52.2 in phase Ⅳ (
Figure 3). Based on ASIR, which adjusts for differences in regional age structure, significant increases in AAPC were observed in specific regions across each phase. In phase I, Gyeonggi showed a substantial increase of 43.96% (95% CI, 14.06%–81.21%;
p<0.001). In phases I and II, Jeju experienced significant increases of 36.86% (95% CI, 13.02%–66.19%;
p=0.002) and 30.95% (95% CI, 3.82%–64.26%;
p=0.014), respectively. In phase Ⅳ, increases were observed in Gwangju (14.83%; 95% CI, 1.76%–29.58%;
p=0.028), Chungnam (11.95%; 95% CI, 4.4%–20.1%;
p<0.001) and Seoul (5.33%; 95% CI, 0.37%–10.52%;
p=0.031) (
Tables S1,
S2).
Distribution of Meteorological Factors
Based on mean values across the entire study period, Tmean increased from 13.1 to 14.0 °C, RH increased from 69.7% to 72.9%, and WV decreased from 2.3 to 2.0 m/s. In phase III, precipitation was the lowest of all phases, with a monthly mean of 98.5 mm and a maximum of 676.2 mm [
24] (
Table S3). These temporal trends in nationwide meteorological factors were similarly observed across all administrative regions (
Tables S4,
S5).
Correlation Analysis
Spearman correlation analysis revealed that the incidence of scrub typhus per 100,000 population was significantly positively correlated with Tmean, RH, and SH and negatively correlated with WV. In phase IV, unlike in previous phases, no notable correlations were observed with other climatic variables, except for Tmean (
rs=0.146,
p<0.001) and RH (
rs=0.191,
p<0.001). Among all climatic variables, RH displayed the strongest significant correlation across all phases, particularly in phase III (
rs=0.339,
p<0.001) (
Table 2).
Time-series Regression Analyses
Lag-specific risks of meteorological variables
Using DLNM, we observed significant non-linear exposure-response relationships between scrub typhus incidence and meteorological factors, with lag periods varying by phase: 4 months in phase I, 7 months in phase II, 5 months in phase III, and 6 months in phase IV.
Using the median value of each meteorological variable as the reference, the single-variable models showed that RH exhibited the highest RR in phases I and III, whereas Tmean displayed the maximum RR in phases II and IV. In phase I, RH of 54.1% was associated with increased risk with a 4-month lag (RR, 1.37; 95% CI, 1.20–1.56), whereas in phase III, RH of 87% showed a stronger association with no lag (RR, 1.95; 95% CI, 1.73–2.21). Tmean exhibited the highest RR values in phases II and IV with no lag. Specifically, Tmean of 6.4 °C was associated with increased risk in phase II (RR, 1.62; 95% CI, 1.29–2.03) and phase IV (RR, 1.98; 95% CI, 1.53–2.58) (
Figures S1A–
S4A,
Tables S6,
S7).
In the multi-variable models, Tmean consistently demonstrated the highest RR with notable lag effects from phases II to III; for example, in phase III, the highest RR was 1.79 (95% CI, 1.25–2.57) at 25.2 °C with a 3-month lag. RH and WV remained significant predictors with elevated RR across all phases. RH increased from 83.1% to 87% over time, peaking in phase III with an RR of 1.46 (95% CI, 1.25–1.71). Moreover, RH-associated risk tended to manifest immediately, with no observable lag. In contrast, WV gradually declined from 3.4 to 1.3 m/s over the study period, accompanied by a decreasing RR trend; however, a consistent lag effect of approximately 3 to 5 months for WV was observed in all phases. Regarding SH, in phase II, the 10th percentile (130.3 hours) was associated with a maximum RR of 1.19 (95% CI, 1.05–1.34) with a 7-month lag. In comparison, phase III showed an immediate increase in RR to 1.15 (95% CI, 1.06–1.25) from the 75th to 90th percentiles (256 hours) (
Figures 4A–
7A,
Tables 3,
4).
Cumulative RRs of meteorological variables
We present cumulative exposure-response relationships between meteorological variables and scrub typhus incidence, showing non-linear patterns across phases.
In the single-variable models, certain meteorological factors were associated with increased risk in phases Ⅰ–Ⅲ, whereas no significant weather factors were identified in phase IV (
Figures S1B–
S4B,
Table S8).
In the multi-variable models, phase II showed significant risks for 3 variables, excluding Tmean, with RH at 84% (reference, 71%) exhibiting the highest risk (RR, 2.67; 95% CI, 1.64–4.36). In phase III, all variables except RH were significantly associated with increased risk, and Tmean at 25.2 °C (reference, 14.3 °C) displayed the highest RR (RR, 5.86; 95% CI, 2.56–13.42). In contrast, during phase IV, only RH at 58% (reference, 74%) was significantly associated with disease risk (RR, 1.68; 95% CI, 1.29–2.20) (
Figures 4B–
7B,
Table 5).
Risk-enhancing exposure ranges
To identify specific meteorological conditions associated with increased disease incidence, we analyzed risk-enhancing exposure ranges. In the single-variable models, several meteorological variables were associated with increased risk (
Table S9); however, in the multi-variable models, a clear difference emerged between phases I–III and phase IV. Specifically, during phase III, significant associations were observed for Tmean ranging from 15.2 to 25.2 °C (reference, 14.3 °C), RH between 74% and 81% (reference, 73%), WV at 2.4 m/s (reference, 2.0 m/s), and SH between 214.0 and 257.0 hours (reference, 213.2 hours). In phase IV, by contrast, increased incidence was significantly associated only with RH levels between 58% and 73% (reference, 74%) (
Table 6).
Discussion
This study is the most comprehensive to date to evaluate the non-linear, delayed, and cumulative associations between meteorological variables and scrub typhus incidence in the Republic of Korea, spanning more than 2 decades in the context of accelerating climate change.
From 2001 to 2017, national incidence steadily increased, with a turning point in 2018, just before the 2019 revisions to case definitions. Historically, scrub typhus was concentrated in southern rural areas and predominantly affected older women. However, our findings highlight a shift; incidence has increased among older adults (≥70 years) and in central and urban regions, while the proportion of female cases has declined. These changes may reflect the combined effects of climatic, ecological, and sociodemographic factors on disease dynamics [
25].
These temporal and demographic changes were paralleled by shifts in meteorological influences, with Tmean and RH emerging as key contributors to disease risk, although correlation strengths were modest—likely due to lagged effects [
26]. In some tropical species of
Leptotrombidium, the entire life cycle can be completed within 2–3 months, yielding at least 2 generations per year [
27]. By contrast, in temperate regions, only a single generation per year is typically possible due to climatic constraints, consistent with previous research [
28]. Our findings demonstrated an overall lagged effect of approximately 5 months (ranging from 4 to 7 months across phases) between meteorological exposure and scrub typhus incidence during 2001–2024. Given the sharp increase in cases between October and November, this pattern suggests a stronger association with climatic conditions occurring in spring to early summer. This contrasts with previous studies emphasizing summer conditions [
29].
In the multi-variable models, phase III showed a peak RR of 1.79 at 25.2 ℃ with a 3 month lag, and the cumulative risk increased to 5.86. Incidence rose significantly within the 15.2–25.2 °C range. This period followed the “warming hiatus” and was characterized by a rapid rise in temperature [
30,
31], possibly promoting mite survival and reproduction and accelerating their life cycle [
32]. In contrast, phase IV exhibited weaker associations and longer lag periods, with only RH at 58% showing a significant cumulative effect. The loss of statistical significance for Tmean in phase IV suggests that this pattern may reflect a new equilibrium in vector–host–climate interactions, potentially after exceeding critical thermal thresholds [
33]. However, this attenuation does not preclude indirect or interacting effects with other climatic variables and non-climatic drivers. Non-climatic factors may include delayed ecological adaptation of vector populations, changes in agricultural practices that alter human–vector contact patterns, and modifications in surveillance systems. Meanwhile, the renewed positive trend observed in phase IV may have been influenced by mite adaptations to climatic extremes, potentially enabling habitat expansion and improved fitness [
34–
36], as well as anthropogenic factors such as land-use change, urbanization, and shifts in human mobility and behavior [
37].
Environmental factors may affect the transmission of scrub typhus through 3 primary mechanisms: the development of vector populations, changes in host ecology, and patterns of human exposure [
38].
Temperature plays a critical role in both the life cycle of
Leptotrombidium mites and outdoor human behavior, making it a key determinant of the transmission dynamics of mite-borne diseases [
39]. Mite activity is highly temperature-dependent: adult mites resume activity in spring when temperatures exceed 10 °C [
25], and oviposition increases in early summer as temperatures reach 20–30 °C [
40]. A temperature of approximately 23±1 °C has been reported to provide optimal hatching conditions for chigger mite eggs [
41]. Consistent with previous studies, increased scrub typhus incidence was observed at temperatures below 25.2 °C during phases I–III in both single-variable and multi-variable models. However, in phase IV, no temperature range was significantly associated with increased incidence. The relationship between temperature and scrub typhus incidence generally follows a rise-then-decline pattern [
42]. Compared with phase I, the average Tmean in phase IV was approximately 1 °C higher, possibly reflecting the accelerating effects of global warming. Once temperatures exceed certain thresholds, egg hatchability may decrease [
27,
41]. Moreover, extremely high temperatures may reduce outdoor human activity, thereby lowering the risk of exposure to infected chigger mites [
42]. Collectively, these factors could suggest an attenuation of temperature effects in recent years.
Environmental humidity may be an important factor influencing the prevalence and geographic distribution of chigger mites and may also affect the reproduction of rodent reservoir hosts [
43,
44]. A study in Chile reported that chigger mites tend to thrive when RH exceeds 50%, while their activity appears to decrease markedly when humidity rises above 75%–80% or falls below 50% [
44]. In our study, although the RR associated with humidity generally declined over time, consistent lagged and cumulative effects were observed through phase III, particularly under high-humidity conditions (P90). However, in phase IV, a statistically significant increase in scrub typhus incidence was observed only within the moderate humidity range of 58% to 73%. These findings suggest that RH continues to impact disease risk, with the peak-risk range potentially shifting toward more moderate levels. From a behavioral perspective, higher humidity conditions could influence human activities, such as agricultural work, which in turn might increase the likelihood of exposure to infected chigger mites.
WV influences scrub typhus transmission through effects on both vector–host interactions and environmental dispersal. Strong winds may disrupt larval attachment [
45] and reduce outdoor human activity, lowering transmission risk [
46]. Conversely, moderate winds associated with forested areas can aid the passive spread of chiggers, potentially increasing exposure [
1]. Although most previous studies have reported positive correlations between wind speed and scrub typhus incidence [
47–
51], with a Korean study noting a significant association with a 5–6-month lag [
50], our findings differ. In this study, WV showed a negative correlation with incidence, with lag effects ranging from 3 to 5 months. The single-variable models revealed elevated RR at low to moderate WV in earlier phases (e.g., 1.5 m/s in phase II and 2.4 m/s in phase III). However, in phase IV, no significant cumulative effect or specific risk-enhancing range was identified. This weakening association may reflect changing ecological dynamics or behavioral adaptations, indicating a need for further investigation into the evolving role of wind in disease ecology.
Chigger mites have a thin cuticle and a low capacity for water retention, making them prone to dehydration when exposed to direct sunlight for extended periods [
27]. Consequently, they prefer humid, shaded environments and tend to be more active in the early morning and late afternoon. With climate change contributing to longer sunshine duration and more frequent heat waves, sunlight exposure may become increasingly important for mite survival. In our study, SH was associated with both lagged and cumulative risks in phases I–III, but no statistically significant effect was observed in phase IV. Although this suggests that light conditions may influence the ecological activity cycles of vectors and hosts, further studies are needed, as variations in seasonality and increasingly unpredictable weather linked to climate change may have obscured these effects.
When considered alongside the results for temperature, humidity, wind, and sunshine, these findings indicate that the climatic drivers of scrub typhus risk have become more variable and phase-dependent in recent years. These shifts support the need for dynamic, climate-responsive risk assessment frameworks capable of adapting to changes in the relative importance and timing of different meteorological factors.
This study has several limitations. First, infectious diseases such as scrub typhus should be understood from a macro-level perspective. The sustained increase in case fatality from 2020 to 2023 is presumed to be partly attributable to coronavirus disease 2019, which may have influenced scrub typhus incidence through reduced healthcare utilization, altered healthcare-seeking behavior, and restricted community mobility. These observations highlight the need for long-term assessments of the burden of scrub typhus and underscore the importance of public health campaigns aimed at improving awareness and prevention among older and high-risk populations. Second, because climate change is a gradual and continuous process, robust analysis of meteorological trends generally requires at least 10 years of data. In this study, however, the observation period was divided into 6-year phases to explore temporal dynamics; therefore, our findings should be re-evaluated using longer observation windows to more accurately capture long-term effects. Lastly, although scrub typhus incidence may be influenced by both geographic and socioeconomic factors, this study considered only meteorological variables. Future research should incorporate additional covariates—such as land-use change, the Normalized Difference Vegetation Index, population density, and income levels—to better adjust for potential confounding.
