
==== Front
One Health
One Health
One Health
2352-7714
Elsevier

S2352-7714(24)00221-0
10.1016/j.onehlt.2024.100895
100895
Research Paper
Asymmetric association between meteorological factors and human infections with hemorrhagic fever with renal syndrome: A 16-year ecological trend study in Shaanxi, China
Xue Chenlu a
Zhang Bingjie a
Li Yanyan a
Li Xinxiao a
Xu Chunjie b
Wang Yongbin wybwho@163.com
a⁎
a Department of Epidemiology and Health Statistics, School of Public Health, The First Affiliated Hospital of Xinxiang Medical University, No. 601 Jinsui Road, Hongqi District, Xinxiang, Henan Province 453003, People's Republic of China
b Beijing Key Laboratory of Antimicrobial Agents/Laboratory of Pharmacology, Institute of Medicinal Biotechnology, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing 100050, China
⁎ Corresponding author. wybwho@163.com
13 9 2024
12 2024
13 9 2024
19 10089511 8 2023
11 9 2024
12 9 2024
© 2024 The Author(s)
2024
https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Objective

Hemorrhagic fever with renal syndrome (HFRS) continues to pose a significant threat to global health. This study aimed to investigate both the long- and short-term asymmetric impacts of variations in meteorological variables on HFRS.

Methods

The reported monthly HFRS incidence data from Shaanxi between 2004 and 2019, along with corresponding meteorological data, were collected to conduct an ecological trend analysis. Subsequently, the autoregressive distributed lag (ARDL) and nonlinear ARDL (NARDL) models were used to examine the long- and short-term asymmetric effects of climate variables on HFRS incidence.

Results

Overall, a reduction in HFRS incidence was observed in Shaanxi from 2004 to 2019, with an average annual percentage change of −0.498 % (95 %CI -13.247 % to 12.602 %). HFRS incidence peaked in December and reached its lowest point in March each year. A 1 mm increase in aggregate precipitation (AP) was associated with a 4.3 % rise in HFRS incidence, while a 1 mm decrease contributed to a 3.7 % increase, indicating a long-term asymmetric impact (Wald long-term asymmetry test [WLT] = 9.072, P = 0.003). In the short term, a 1 % decrease in mean relative humidity (MRH) led to a 5.7 % decline in HFRS incidence (Wald short-term asymmetry test [WSR] = 5.978, P = 0.015). Additionally, changes in meteorological variables showed varied effects: ΔMWV(+) at a 1-month lag had a significant positive short-term effect on HFRS; ΔMRH(+) at a 3-month lag, ΔAP(+) at a 2-month lag, ΔAP(−) at a 1-month lag, ΔASH(+) at a 1-month lag, and ΔASH(−) at a 3-month lag all exhibited strong negative short-term impacts on HFRS incidence.

Conclusions

Weather variability plays a significant role in influencing HFRS incidence, with both long- and short-term asymmetric and/or symmetric effects. Utilizing the NARDL model through a One Health lens offers promising opportunities for enhancing HFRS control measures.

Keywords

Hemorrhagic fever with renal syndrome
Meteorological factors
Asymmetric relationships
Nonlinear autoregressive distributed lag
Ecological trend study
Abbreviations

HFRS Hemorrhagic fever with renal syndrome

NARDL Nonlinear autoregressive distributed lag

ARDL Autoregressive distributed lag

HTNV Hantan virus

SEOV Seoul virus

MRH Mean relative humidity

AP Aggregate precipitation

MT Mean temperature

MWV Mean wind velocity

ASH Aggregate sunshine hours

AAPC Average annual percentage change

CI Confidence interval

VIF Variance inflation factor

ADF Augmented Dickey–Fuller

AIC Akaike information criterion

BIC Bayesian information criterion

HQ Hannan-Quinn

PACF Partial autocorrelation function

CUSUM Cumulative sum

CUSUM Cumulative sum of squares

WLT Wald long-term asymmetry

WST Wald short-term asymmetry
==== Body
pmc1 Introduction

Hemorrhagic fever with renal syndrome (HFRS) is a rodent-borne disease caused by hantaviruses. Human infection typically occurs through contact with contaminated droplets, inhalation of particulates, or direct contact with infected rodents and their excreta [1]. Once the virus enters the human bloodstream, it typically results in two distinct clinical presentations: Hantavirus Pulmonary Syndrome (HPS) and HFRS, which are determined by the specific hantavirus strain involved. HPS, primarily associated with Sin Nombre virus (SNV) in North America, manifests with symptoms such as fever, myalgia, and respiratory distress, often progressing rapidly and resulting in severe outcomes without timely medical intervention [1]. Conversely, HFRS, more common in Europe and Asia, is characterized by fever, hemorrhagic symptoms, and acute renal failure, primarily linked to the Hantaan virus (HTNV) [1,2]. In China, HTNV and Seoul virus (SEOV) are identified as the major causative agents of HFRS, with their primary natural reservoirs being Apodemus agrarius and Rattus norvegicus, respectively [3,4]. Globally, approximately 100,000 cases are reported annually, with over 90 % occurring in China, Korea, and Russia [5]. In China, HFRS is classified as a class B notifiable disease and poses a significant public health challenge [1]. From 1950 to 2014, there were 1,625,002 reported cases and 46,968 deaths in China, resulting in a death rate of 2.89 % [5]. Although the incidence of HFRS has fluctuated over recent decades, it remains one of the top nine communicable diseases in China [3]. The epidemiology of hantaviruses is largely influenced by the distribution of their reservoir hosts; however, these viruses are not uniformly present in all geographical areas inhabited by their hosts [6]. Recent studies indicate that the dual seasonal pattern of HFRS is linked to dominant hantavirus genotypes, with climate effects being more pronounced in HTNV-endemic regions compared to those dominated by SEOV [7]. This suggests that environmental factors and other potential influences should also be considered when studying the epidemiology of HFRS [6].

Research has shown that climate change can impact the epidemiology of hantavirus infections [1,6,8,9]; however, findings have been inconsistent. Most studies have employed linear models and failed to explore the long- and short-term asymmetric dynamic effects of meteorological variables on HFRS—meaning that an increase or decrease in meteorological variables may yield different effects—which is crucial for effective prevention strategies. Additionally, prior studies often overlooked the strong autocorrelations present in time series data, leading to potential overestimation. Recent research has demonstrated that the nonlinear autoregressive distributed lag (NARDL) model can address these gaps due to its several advantages [[9], [10], [11], [12]]: (1) It decomposes the effect of regressors into short- and long-term components, allowing for asymmetries in various combinations of short- and long-term dynamics; (2) it accommodates time series with different orders of integration; (3) it addresses endogenous relationships between variables; and (4) it automatically considers autocorrelations in time series analysis.

Shaanxi province in central China is located between longitudes 105°29′-111°15′E and latitudes 31°42′-39°35′N and experiences a continental monsoon climate. As of 2023, its population was approximately 39.56 million. In 2005, Shaanxi had the second highest incidence rates of HFRS after Heilongjiang [3]. From 2010 to 2012, it surpassed Heilongjiang to record the highest incidence rates of HFRS [3]. However, few studies have concentrated on the influence of meteorological factors on the long- and short-term transmission dynamics of HFRS in this region. Therefore, this study aimed to conduct a 16-year ecological trend analysis to investigate the long- and short-term asymmetric dynamic associations between meteorological factors and HFRS in Shaanxi using the NARDL model.

2 Material and methods

2.1 HFRS data

The monthly HFRS incidents in Shaanxi from 2004 to 2019 were collected from the Data-center of China Public Health Science (DCPHS). Population data was sourced from the Shaanxi Statistical Yearbook. All HFRS incidents were confirmed by authorized institutions and professionals following the diagnostic criteria for HFRS (http://www.nhc.gov.cn/wjw/s9491/wsbz.shtml).

2.2 Meteorological data

Daily meteorological variables, including mean relative humidity (MRH), aggregate precipitation (AP), mean temperature (MT), mean wind velocity (MWV), and aggregate sunshine hours (ASH) were provided by the National Meteorological Science Data Center (http://data.cma.cn/), and then these variables were collated as the monthly time series format.

2.3 Statistical analysis

During the statistical description, study variables were represented as mean ± standard deviation (x‾±s). The average annual percentage change (AAPC) with a 95 % confidence interval (CI) was computed to describe the epidemiological trend of HFRS [13]. Spearman's correlation was applied to test the correlation between meteorological factors and HFRS, with a correlation greater than 0.9 or a variance inflation factor (VIF) greater than 10 was indicating strong collinearity [14,15]. In cases of multicollinearity among variables, these variables were entered into different NARDL models alongside other meteorological drivers to investigate their effects on HFRS.

Autoregressive distributed lag (ARDL) model has been used to address issues related to autocorrelations and non-stationarity of key variables; details of this model have been provided in a prior study [16]. However, the ARDL model may yield biased results due to nonlinear and/or asymmetric impacts of meteorological factors on diseases [17]. The NARDL model was thus introduced to overcome the weakness. This approach allows for the investigation of both long- and short-term asymmetric dynamic effects [10,18]. In the presence of asymmetric impacts, the NARDL model can quantify the responses of HFRS to positive and negative changes in each of the meteorological factors by integrating the positive and negative partial sums of increments and decrements in these variables [10,18]. The NARDL analysis involves three steps [10,18,19]: First, investigation of the order of integration. The order of integration is not allowed to exceed one, although the NARDL model has relaxed this integration requirement. Besides, a pseudo regression may be generated by the non-stationary regressors. Thus, the augmented Dickey–Fuller (ADF) statistic was chosen to test the order of integration and stationarity in both independent and dependent variables [20]. Second, investigation of the long-term asymmetric cointegration. To determine whether a long-term asymmetric cointegration exists between regressors and dependent variables, the bounds test (F statistic) was applied [21]. If evidence suggests such a relationship is present, a Wald test is conducted to investigate short- and long-term asymmetries. Third, effect estimation. The positive and negative dynamic multiplier effects of regressors on the dependent variable are estimated accordingly.

Logarithmic transformation helps reduce the variation of dependent variables and better interpret the results, and thus the log(HFRS) was used in this study. The NARDL notation was below,ΔlogYt=a0+∑i=1piφilogYt−pi+∑i=0q1δ1i+xt−q1i++∑i=0q2δ1i−xt−q2i−+∑i=1p2p2iΔlogYt−p2i+∑i=0q3τ3i+Δxt−q3i+∑i=0q4τ4i−Δxt−q4i−+a1month+ϵt

where, Yt represents the HFRS cases, x signifies the meteorological factors such as MRH, AP, MT, MWV, and ASH, x+ and x− are the positive and negative partial sums of increases and decreases in each meteorological factor, respectively, p and q denote the optimal lag orders of the HFRS cases and meteorological variables, respectively, δ1i+ and δ1i− signify the long-term equilibrium parameters for the dependent variable, τ3i+ and τ3i− refer to the short-term parameters for the dependent variable, month represents the seasonal variables, Δ refers to the first-order difference.

In this study, the maximum lag orders were set at four months, reflecting an approximate 16-week incubation period from HFRS infection to the onset of symptoms [22]. The optimal lag orders were then determined using various criteria, including the Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), Hannan-Quinn (HQ) criterion, log-likelihood, and adjusted R2. To assess autocorrelation in the dependent variable, the partial autocorrelation function (PACF) was employed, which measures the correlation between current observations and past observations while controlling for other variables [23]. To account for seasonal effects, 11 monthly dummy variables were incorporated into the model. Furthermore, the stability of the NARDL model was evaluated using cumulative sum (CUSUM) and CUSUM of squares statistics [21]. All statistical analyses were conducted using EViews 10 (IHS, Inc. USA) and R 4.2.0 (R Development 164 Core Team, Vienna, Austria), with statistical significance defined as a two-sided P ≤ 0.05.

3 Results

3.1 Statistical description

During 2004–2019, a total of 26,431 cases of HFRS were reported in Shaanxi, averaging 138 cases per month and approximately 1652 cases annually. Overall, the incidence of HFRS demonstrated a declining trend over this period (AAPC = -0.498 %, 95 %CI -13.247 % to 12.602 %), with the highest peak occurring in 2012, when 3591 cases were reported (9.482 cases per 100,000 people). Following this peak, the incidence steadily decreased until 2016, when only 933 cases were recorded (2.408 cases per 100,000 people). The incidence of HFRS exhibited clear seasonal and periodic patterns, with peaks typically observed in December and troughs in March each year.

A summary of monthly HFRS cases alongside relevant meteorological factors is provided in Table 1. The mean values of MRH, AP, MT, MWV, and ASH were 64.74 ± 9.93 %, 54.10 ± 50.47 mm, 12.23 ± 9.08 °C, 1.97 ± 0.22 m/s, and 172.23 ± 42.30 h, respectively. As illustrated in Fig. 1, there appears to be a similar trend between HFRS and MT, MWV, and ASH. In contrast, HFRS incidence showed an inverse relationship with MRH and AP. Importantly, no correlations exceeded 0.9, nor did any VIF values surpass 10 among the variables, indicating a lack of strong collinearity (Table 1 and Fig. 2).Table 1 Summary for monthly HFRS cases and weather factors in Shanxi, 2004–2019.

Table 1Variable	Mean	S.D.	Min	P25	P50	P75	Max	VIF	
HFRS cases	137.90	177.91	9.00	43.75	74.00	145.50	1209.00	–	
MRH	64.74	9.93	42.24	58.14	63.89	72.00	84.76	5.97	
AP	54.10	50.47	0.07	11.87	40.76	83.78	251.57	4.13	
MT	12.23	9.08	−5.42	4.00	13.41	20.60	26.40	7.99	
MWV	1.97	0.22	1.50	1.81	1.95	2.13	2.58	2.52	
ASH	172.23	42.30	58.64	142.74	171.92	200.65	264.20	4.61	
HFRS cases, 1-month lag	–	–	–	–	–	–	–	1.58	
HFRS, hemorrhagic fever with renal syndrome; MRH, mean relative humidity; AP, aggregate precipitation; MT, mean temperature; MWV, mean wind velocity; ASH, aggregate sunshine hours, S.D., standard deviation; VIF, variance inflation factor.

Fig. 1 Time series graph suggesting the changing patterns of the weather variables and HFRS cases in Shanxi, 2004–2019 (HFRS, hemorrhagic fever with renal syndrome; MRH, mean relative humidity; AP, aggregate precipitation; MT, mean temperature; MWV, mean wind velocity; ASH, aggregate sunshine hours).

Fig. 1

Fig. 2 Spearman's correlation between variables. It was observed that there was no correlation greater than 0.9 between variables, indicating an absence of strong collinearity between variables (HFRS, hemorrhagic fever with renal syndrome; MRH, mean relative humidity; AP, aggregate precipitation; MT, mean temperature; MWV, mean wind velocity; ASH, aggregate sunshine hours).

Fig. 2

3.2 Development of the NARDL and ADRL models

The ADF test indicated that both the dependent and independent variables were non-stationary, as evidenced by the following P-values: log(HFRS) (P = 0.272), MT (P = 0.885), ASH (P = 0.387), MWV (P = 0.584), AP (P = 0.706), and MRH (P = 0.718). However, after differencing the data once, all variables became stationary (P < 0.001). The PACF revealed significant autocorrelation at delays of one and two months (Fig. S1). Furthermore, the bounds test yielded an F of 6.735 (which far exceeded the critical values [I0 = 1.82, I1 = 2.99]), confirming the presence of a long-term asymmetric cointegration relationship among the variables. Lastly, a wide range of NARDL models were developed (Table S1 and Fig. S2). Of the various candidates, we chose the NARDL(1, 4, 0, 4, 4, 0, 0, 2, 0, 4, 4) specification (in which the lag of log(HFRS) was one, lags of MRH(+) and MRH(−) were four and zero, respectively, lags of AP(+) and AP(−) were four, lags of MT(+) and MT(−) were zero, lags of MWV(+) and MWV(−) were two and zero, respectively) as the optimal model due to its lower AIC (0.372), BIC (1.132), and HQ (0.68), in conjunction with a higher adjusted R2 value of 0.927 and a log-likelihood of 9.211. As shown in Fig. 3, the resulting residuals from the CUSUM and CUSUM of squares tests remained within the 95 % CI, substantiating the stability of the NARDL model. Likewise, the ARDL(1, 4, 3, 4, 2, 4) model was identified as the best specification among the possible ARDL candidates (Table S2 and Fig. S3). Notably, the error metrics for NARDL—mean absolute error (MAE = 23.607) and root mean squared error (RMSE = 26.306)—were lower than those for ARDL (MAE = 24.986 and RMSE = 48.182). This suggests that the NARDL model provides a more accurate representation of the epidemic dynamics of HFRS by accounting for both long- and short-term asymmetries compared to the ARDL model (Fig. 4).Fig. 3 Stability test for the nonlinear autoregressive distributed lag model (NARDL). a. Cumulative sum (CUSUM) test, b. CUSUM of squares test. The CUSUM and CUSUM of squares were within 95 % confidence interval (CI) at different time confirmed the validity and stability of the NARDL.

Fig. 3

Fig. 4 Comparison of the forecasts under the nonlinear autoregressive distributed lag (NARDL) and autoregressive distributed lag (ARDL) models with the observed values. It seemed that the forecasts could be closer to the trend and seasonality under the NARDL model than under the ARDL model.

Fig. 4

3.3 The asymmetric and symmetric effects of meteorological factors on HFRS

Based on the findings presented in Table 2, there was a statistically significant positive long-term relationship between AP and HFRS. Specifically, a 1 mm increase in AP was associated with an approximate 4.3 % rise in HFRS, while a 1 mm decrease in AP resulted in an approximately 3.7 % increase in HFRS, indicating a cumulative effect on HFRS incidence. Conversely, the long-term coefficients for MRH(−) were statistically significant but negatively correlated with HFRS; a 1 % decrease in MRH corresponded to an approximate 5.7 % reduction in HFRS. Meanwhile, MT, MWV, and ASH did not show significant long-term coefficient. In terms of short-term dynamics, ΔMWV(+) at a 1-month lag exhibited a substantial positive effect on HFRS, with an increase of 1 m/s in MWV leading to an approximate 83.4 % rise in HFRS. On the other hand, ΔMRH(+) at a 3-month lag, ΔAP(+) and ΔAP(−) at a 1-month lag, along with ΔASH(+) at a 1-month lag and ΔASH(−) at a 3-month lag, demonstrated a more pronounced negative short-term effect on HFRS. Specifically, when MRH, AP, and ASH increased by 1 %, 1 mm, and 1 h, HFRS increased approximately by 2.1 %, 1.0 %, and 0.5 %, respectively. Conversely, decreases in AP and ASH by 1 mm and 1 h resulted in increases in HFRS of about 1.0 % and 0.6 %, respectively. The Wald test results listed in Table 3 indicated potential long-term asymmetric impacts of AP and MWV on HFRS, a finding further supported by the dynamic multiplier plots shown in Figs. 5a-5f, despite the long-term coefficient for MWV being non-significant. No long-term asymmetric relationships were observed for MRH, MT, and ASH. However, both MWV and MRH may exert short-term asymmetric effects on HFRS. For instance, as illustrated in Fig. 5d, HFRS incidence rises gradually with increases in AP(+). In the short term, AP(−) initially elevates the incidence of HFRS before subsequently decreasing it; yet over the long term, the relationship between AP(−) and HFRS turns negative. Ultimately, considering the combined effects of fluctuations in AP(+) and AP(−), the overall impact leads to an increase in HFRS incidence.Table 2 Long- and short-term effects by use of the optimal NARDL and ARDL.

Table 2NARDL	ARDL	
Variable	Coefficient	P	Variable	Coefficient	P	
Long-run effect	Long-run effect	
MRH(+)	−0.025	0.327	MRH	0.034	0.213	
MRH(−)	−0.057	0.040	AP	0.029	0.010	
AP(+)	0.043	<0.001	MT	−0.106	0.472	
AP(−)	0.037	<0.001	MWV	−0.316	0.781	
MT(+)	0.083	0.262	ASH	0.012	0.109	
MT(−)	0.069	0.365	Short-run effect	
MWV(+)	−1.442	0.223	∆MRH	−0.015	0.014	
MWV(−)	1.022	0.348	∆MRH, 1-month lag	−0.017	0.029	
ASH(+)	0.009	0.273	∆MRH, 2-month lag	−0.016	0.036	
ASH(−)	0.013	0.119	∆MRH, 3-month lag	−0.016	0.007	
Short-run effect	∆AP	−0.001	0.315	
∆MRH(+)	−0.016	0.074	∆AP, 1-month lag	−0.005	<0.001	
∆MRH(+), 1-month lag	−0.004	0.636	∆AP, 2-month lag	−0.003	0.004	
∆MRH(+), 2-month lag	−0.006	0.471	∆MT	0.020	0.297	
∆MRH(+), 3-month lag	−0.021	0.007	∆MT, 1-month lag	0.025	0.340	
∆AP(+)	0.001	0.314	∆MT, 2-month lag	0.027	0.241	
∆AP(+), 1-month lag	−0.012	<0.001	∆MT, 3-month lag	0.047	0.011	
∆AP(+), 2-month lag	−0.010	<0.001	∆MWV	0.324	0.076	
∆AP(+), 3-month lag	−0.003	0.087	∆MWV, 1-month lag	0.616	0.001	
∆AP(−)	−0.003	0.036	∆ASH	−0.003	0.017	
∆AP(−), 1-month lag	−0.010	<0.001	∆ASH, 1-month lag	−0.004	0.020	
∆AP(−), 2-month lag	−0.005	0.001	∆ASH, 2-month lag	−0.004	0.010	
∆AP(−), 3-month lag	−0.002	0.028	∆ASH, 3-month lag	−0.004	0.003	
∆MWV(+)	0.436	0.089				
∆MWV(+), 1-month lag	0.834	0.002				
∆ASH(+)	−0.002	0.202				
∆ASH(+), 1-month lag	−0.005	0.006				
∆ASH(+), 2-month lag	−0.004	0.005				
∆ASH(+), 3-month lag	−0.002	0.096				
∆ASH(−)	−0.003	0.045				
∆ASH(−), 1-month lag	−0.004	0.097				
∆ASH(−), 2-month lag	−0.005	0.013				
∆ASH(−), 3-month lag	−0.006	<0.001				
Note, adjustment for seasonality as dummy variable. NARDL, nonlinear autoregressive distributed lag; ARDL, autoregressive distributed lag; MRH, mean relative humidity; AP, aggregate precipitation; MT, mean temperature; MWV, mean wind velocity; ASH, aggregate sunshine hours, S.D., standard deviation; VIF, variance inflation factor.

Table 3 Wald test results for Long- and short-term asymmetries.

Table 3Variable	Long-term asymmetry	Short-term asymmetry	
WLT	P	WST	P	
MT	0.646	0.422	–	–	
AP	9.072	0.003	0.157	0.692	
ASH	2.015	0.156	1.528	0.216	
MWV	10.735	0.001	10.318	0.001	
MRH	3.067	0.080	5.978	0.015	
MRH, mean relative humidity; AP, aggregate precipitation; MT, mean temperature; MWV, mean wind velocity; ASH, aggregate sunshine hours, S.D., standard deviation; VIF, variance inflation factor; WLT, Wald long-term symmetry; WST, Wald short-term symmetry.

Fig. 5 Dynamic multiplier asymmetric effects of weather factors on hemorrhagic fever with renal syndrome (HFRS). a. multiplier graph for aggregate sunshine hours (ASH), b. multiplier graph for mean wind velocity (MWV), c. multiplier graph for mean temperature (MT), d. multiplier graph for aggregate precipitation (AP), e. multiplier graph for mean relative humidity (MRH).

Fig. 5

4 Discussion

The relationship between human health, animal health, and the environment has never been more crucial than in the contemporary context of global health challenges. The One Health approach, an integrative framework that promotes a collaborative effort across various sectors, is essential for addressing health issues that transcend human, animal, and environmental domains [24]. By understanding the ecological dynamics that contribute to the transmission of HFRS, stakeholders can implement comprehensive strategies to prevent outbreaks, educate communities, and enhance disease management protocols. This study found that, from a long-term perspective, AP had a significant positive nonlinear association with HFRS. In the short term, ΔMWV(+) at a 1-month lag was positively associated with HFRS, while ΔMRH(+) at a 3-month lag, ΔAP(+) and ΔAP(−) at a 1-month lag, and ΔASH(+) at a 1-month lag and ΔASH(−) at a 3-month lag exhibit a reverse association with HFRS. This is the only study to investigate both long- and short-term asymmetric impacts of meteorological factors on HFRS using the NARDL model in Shaanxi. The ARDL model is widely used for its capacity to estimate long- and short-term relationships among time series data while accommodating various orders of integration [16]. In the context of meteorological factors influencing HFRS, the ARDL model can reveal fundamental associations; however, it often operates under the assumption of linear impacts across the variables involved. This assumption may oversimplify the actual dynamics, as real-world relationships between meteorological variables and disease incidence may exhibit nonlinearity and asymmetry [10]. In contrast, the NARDL model expands upon the ARDL framework by allowing for nonlinear interactions among variables [10]. It enables researchers to differentiate the impact of positive and negative changes in explanatory variables, thus offering a more nuanced view of how meteorological factors affect the incidence of HFRS. By distinguishing between these dimensions, the NARDL model can reveal potentially critical insights into the underlying mechanisms driving the association between climate variables and disease outcomes. Our results also corroborate the lead time and both asymmetric and symmetric impacts of meteorological parameters on HFRS, highlighting the usefulness of the NARDL model in capturing the dynamic epidemic structure of HFRS incidence. These findings are helpful in estimating HFRS epidemics, providing sufficient lead time to develop targeted policies and implement effective public health interventions.

Our results revealed that overall a decreasing trend was observed in HFRS incidence, consistent with global and national epidemic patterns in China [3]. This decline can be attributed to a confluence of factors, including improved public health interventions, enhanced agricultural practices, increased public awareness, and ecological modifications [25]. As China continues to develop and implement comprehensive strategies for the prevention and control of HFRS, it is imperative that these efforts remain multifaceted, integrating education, community engagement, vaccination campaigns, and environmental management [25]. While the current trend is encouraging, maintaining vigilance is essential to preserve these gains and protect vulnerable populations from potential outbreaks of HFRS, as changing ecological conditions and patterns of human behavior could pose new challenges in the future [[25], [26], [27]]. Besides, our findings revealed a distinct seasonal profile in HFRS morbidity, with peaks in December and troughs in March, in alignment with the seasonality across China [4]. This pronounced seasonal profile may be closely linked to variations in rodent density and rainfall, both of which exhibit seasonal fluctuations [28].

An intriguing finding is that AP was one of the most important contributors to HFRS, exhibiting a notable positive long-term effect on HFRS, and it seems that an increase in AP has a stronger effect than a reduction, indicating an asymmetric relationship. Conversely, a significant negative short-term effect of AP was observed at 1-month, 2-month, and 3-month lags. Low-lying regions and wetlands with moist soil serve as ideal habitats, heightening the risk of HFRS transmission [29]. Moist and semi-moist soil is essential for the growth of vegetation and crops that either directly or indirectly provide sustenance for rodent hosts, resulting in larger rodent populations [30,31]. The densities of hosts, particularly hantavirus-positive hosts, were positively correlated with the risk of HFRS transmission [32]. Additionally, the previously mentioned significant negative short-term effect of AP at 1-month, 2-month, and 3-month lags suggests that excessive precipitation may disrupt the nests of host animals, thereby reducing the likelihood of rodent-rodent contact, rodent-human interaction, and subsequent virus transmission [12,29,33].

The second important finding is that MRH(−) had a negative long-term effect on HFRS, indicating an approximate 5.7 % decrease in HFRS incidence with a 1 % reduction in MRH. This aligns well with several studies that have reported a positive association between MRH and HFRS [17,30,34]. Moisture not only influences the growth of food sources that determine rodent population size, thereby affecting the HFRS transmission, but also directly influences rodent activity and hantavirus infectivity [35]. However, contrary to our conclusions, several studies reported a positive relationship between MRH and HFRS [22,36,37]. This discrepancy may be attributed, in part, to the different models employed in data analysis, variations in geographic regions, or the absence of autoregressive adjustments in the dependent variable.

The third important finding is that MWV had a positive short-term asymmetric impact on HFRS, and unfortunately the long-term coefficient was non-significant. This was consistent with the previous studies in Changchun, Shenyang, and Heilongjiang [12,38,39]. The short-term positive impact of wind on HFRS transmission involves two main mechanisms: increased human exposure to infectious materials and changes in rodent behavior [27,36,40]. First, high winds can aerosolize rodent urine and droppings, elevating the concentration of viral particles in the air and exposing humans during outdoor activities. Second, wind prompts rodents to seek shelter, but this behavior can drive them into areas with more human interaction, such as residential regions or agricultural fields. This increases the likelihood of zoonotic spillover events. Third, wind may also worsen drought conditions, affecting food supplies for rodents and forcing them closer to human settlements. The resultant increase in rodent-human interactions forms a nexus where the risks of HFRS transmission escalate. Lastly, Shaanxi is a windy province, strong winds can disrupt rodent populations, causing temporary population fluctuations as they disperse and settle in new areas, further heightening the potential for human encounters.

Our study documented negative short-term associations of ASH(+) and ASH(−) with HFRS, with significant association coefficients observed at delays of 1–3 months. However, the lack of evidence for asymmetry and the extremely small coefficient values indicate a very weak short-term relationship. Previous research has also reported negative associations between sunshine hours and HFRS incidence [33,39]. Sunshine hours and wind speed could influence crop yield, rodent reproduction, and vector density, which in turn may affect the likelihood of HFRS occurrence [17]. A deeper investigation into the plausible mechanism underlying the relationship between ASH and HFRS is warranted. In addition, researchers have identified both positive and negative associations between AT and HFRS incidence [4,34,41], while our study found no significant long-term or short-term relationships in this regard. Our study uses the NARDL model, which provides an insight into the link between variables by decomposing the effect of regressors into short- and long-term components—a capability that current models, aside from NARDL, do not provide. Consequently, we expect further validation efforts to be conducted in other areas.

Our research specifically focused on the asymmetric and/or symmetric impacts of variations in meteorological factors on HFRS incidence in both the long and short term. Prior study has emphasized the significance of considering the changes in population immunity, autocorrelations, possible lags and relationship patterns, and seasonality when performing time series analyses [42]. Although we did account for most of these factors, changes in population immunity could not be addressed due to a lack of data. Therefore, we are confident that our findings provide valid and reliable evidence that variations in meteorological factors play a crucial asymmetric and/or symmetric role in influencing HFRS incidence in both the long and short term. Nonetheless, our study has some limitations. First, under-reporting or under-diagnosis is an unavoidable issue in a passive monitoring system. Second, this ecological trend study does not permit an examination of individual-based relationships or the inference of causal effects. Third, daily or weekly data could offer deeper insights into temporal differences across years; however, their unavailability limits further investigation. Finally, we did not control for the effects of the unmeasured confounders (e.g., geographic and socioeconomic factors, population density, and host susceptibility).

5 Conclusion

Our study elucidates the significant long- and short-term asymmetric and symmetric contributions of AP, MWV, MRH, and ASH to the HFRS incidence through a 16-year ecological trend study. Given the implications of global climate change, it is imperative that meteorological variables are integrated into strategies for the control and prevention of HFRS. By understanding the ecological dynamics that facilitate the transmission of hantaviruses, stakeholders can implement comprehensive strategies for outbreak prevention, community education, and enhancement of disease management protocols. As an infectious disease continues to pose significant threats to global health, the necessity for multidisciplinary cooperation in understanding and mitigating these risks has never been more apparent. Embracing the One Health paradigm is not only critical for managing HFRS but is also essential for fortifying overall public health resilience against future zoonotic threats.

Ethics approval and consent to participate

This study protocol was approved by the study institutional review board of Xinxiang Medical University (No: XYLL-2019072). Because the DCPHS system shares the monthly number of HFRS cases anonymously and we cannot access any identifying information of the patients, and hence informed consent was waived.

Consent for publication

Not applicable.

Funding

This work was supported by Natural Science Foundation in Henan Province, the Key Scientific Research Project of Universities, and the Open Project Program of priority funding of The First Hospital of 10.13039/501100011841 Xinxiang Medical University (222300420265 , 21A330004 , and XZZX2022002 ).

CRediT authorship contribution statement

Chenlu Xue: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing. Bingjie Zhang: Data curation, Formal analysis, Investigation, Methodology, Software, Validation, Writing – original draft, Writing – review & editing. Yanyan Li: Formal analysis, Investigation, Software, Validation, Writing – original draft, Writing – review & editing. Xinxiao Li: Formal analysis, Investigation, Validation, Writing – review & editing. Chunjie Xu: Conceptualization, Data curation, Formal analysis, Investigation, Validation, Visualization, Writing – original draft, Writing – review & editing. Yongbin Wang: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A Supplementary data

Supplementary material: Model construction and diagnostic results of the ARDL and NARDL models.

Image 1

Data availability

All data for this work are presented in the results and conclusions or please contact the corresponding author on the reproducibility of this work.

Acknowledgments

We appreciated the DCPHS for sharing the HFRS morbidity data.

Appendix A Supplementary data to this article can be found online at https://doi.org/10.1016/j.onehlt.2024.100895.
==== Refs
References

1 Jiang H. Zheng X. Wang L. Du H. Wang P. Bai X. Hantavirus infection: a global zoonotic challenge Virol. Sin. 32 1 2017 32 43 28120221
2 Whitmer S.L.M. Whitesell A. Mobley M. Talundzic E. Shedroff E. Cossaboom C.M. Human Orthohantavirus disease prevalence and genotype distribution in the U.S., 2008-2020: a retrospective observational study. Lancet regional health Americas 37 2024 100836
3 Zou L.X. Chen M.J. Sun L. Haemorrhagic fever with renal syndrome: literature review and distribution analysis in China Int. J. Infect. Dis. 43 2016 95 100 26791541
4 Hansen A. Cameron S. Liu Q. Sun Y. Weinstein P. Williams C. Transmission of haemorrhagic fever with renal syndrome in China and the role of climate factors: a review Int. J. Infect. Dis. 33 2015 212 218 25704595
5 Jiang H. Du H. Wang L.M. Wang P.Z. Bai X.F. Hemorrhagic fever with renal syndrome: pathogenesis and clinical picture Front. Cell. Infect. Microbiol. 6 2016 1 26870699
6 Polat C. Erdin M. Kalkan S.O. Irmak S. Cetintas O. Colak F. Evaluating climate-dependent distribution of orthohantaviruses with monitoring wild rodents: one health perspective(dagger) Brazil. J. Microbiol.: [Publicat. Brazil. Soc. Microbiol.]. 2024 10.1007/s42770-024-01447-2
7 Lv C.L. Tian Y. Qiu Y. Xu Q. Chen J.J. Jiang B.G. Dual seasonal pattern for hemorrhagic fever with renal syndrome and its potential determinants in China Sci. Total Environ. 859 2 2023 160339
8 Wang Y. Duan Q. Pang B. Tian X. Ma J. Ma W. Assessing the relationship between climate variables and hemorrhagic fever with renal syndrome transmission in eastern China: a multi-cities time series study Transbound. Emerg. Dis. 2023 2023 5572334
9 Wang Y. Liang Z. Qing S. Xi Y. Xu C. Lin F. Asymmetric impact of climatic parameters on hemorrhagic fever with renal syndrome in Shandong using a nonlinear autoregressive distributed lag model Sci. Rep. 14 1 2024 9739 38679612
10 Shin Y. Yu B. Greenwoodnimmo M. Modelling asymmetric Cointegration and dynamic multipliers in a nonlinear ARDL framework Soc. Sci. Electron. Publ. 2014 281 314
11 Zheng Y. Zhou M. Wen F. Asymmetric effects of oil shocks on carbon allowance price: evidence from China Energy Econ. 97 2021 105183
12 Wang Y. Zhang B. Xue C. Zhou P. Dong X. Xu C. Long- and short-run asymmetric effects of meteorological parameters on hemorrhagic fever with renal syndrome in Heilongjiang: a population-based retrospective study Transbound. Emerg. Dis. 2024 1 2024
13 Clegg L.X. Hankey B.F. Tiwari R. Feuer E.J. Edwards B.K. Estimating average annual percent change in trend analysis Stat. Med. 28 29 2009 3670 3682 19856324
14 Mason C.H. Perreault W.D. Collinearity, power, and interpretation of multiple regression analysis J. Mark. Res. 28 3 1991 268 280
15 Alain F. Zuur Elena Ieno N. A protocol for data exploration to avoid common statistical problems Methods Ecol. Evol. 1 2010 3 14
16 He F. Hu Z.J. Zhang W.C. Cai L. Cai G.X. Aoyagi K. Construction and evaluation of two computational models for predicting the incidence of influenza in Nagasaki prefecture, Japan Sci. Rep. 7 1 2017 7192 28775299
17 Luo Y. Lv H. Yan H. Zhu C. Ai L. Li W. Meteorological change and hemorrhagic fever with renal syndrome epidemic in China, 2004-2018 Sci. Rep. 12 1 2022 20037 36414682
18 Fu L. Tu X. Liao J. Asymmetric impacts of climate policy uncertainty, investor sentiment on energy prices and renewable energy consumption: evidence from NARDL and wavelet coherence J. Environ. Manag. 367 2024 122057
19 Yuan Q. Cai H.H. Jiang Y. Khan N.U. Qamri G.M. The asymmetric effect of global value chain on environmental quality: implications for environmental management J. Environ. Manag. 365 2024 121470
20 Musbah H. Aly H.H. Little T.A. A proposed novel adaptive DC technique for non-stationary data removal Heliyon 9 2023 e13903
21 Gaies B. Nakhli M.S. Sahut J.M. Guesmi K. Is bitcoin rooted in confidence? – unraveling the determinants of globalized digital currencies Technol. Forecast. Soc. Chang. 172 2021 121038
22 Zhang R. Zhang N. Sun W. Lin H. Liu Y. Zhang T. Analysis of the effect of meteorological factors on hemorrhagic fever with renal syndrome in Taizhou City, China, 2008-2020 BMC Public Health 22 1 2022 1097 35650552
23 Zhang X. Zhang Q. Zhang G. Nie Z. Gui Z. Que H. A novel hybrid data-driven model for daily land surface temperature forecasting using long short-term memory neural network based on ensemble empirical mode decomposition Int. J. Environ. Res. Public Health 15 5 2018 1032 29883381
24 Rittem S. Plangsangmas T. Ruegg S.R. Ecosystem health appears neglected in the management of the human-macaque interface: a systematic review One Health (Amsterdam, Netherlands). 19 2024 100875
25 Wen B. Yang Z. Ren S. Fu T. Li R. Lu M. Spatial-temporal patterns and influencing factors for hemorrhagic fever with renal syndrome: a 16-year national surveillance analysis in China One Health (Amsterdam, Netherlands). 18 2024 100725
26 Liu S. Zhang T. A long-term retrospective analysis of the haemorrhagic fever with renal syndrome epidemic from 2005 to 2021 in Jiangxi Province, China Sci. Rep. 13 1 2023 2268 36755085
27 Liang W. Gu X. Li X. Zhang K. Wu K. Pang M. Mapping the epidemic changes and risks of hemorrhagic fever with renal syndrome in Shaanxi Province, China, 2005-2016 Sci. Rep. 8 1 2018 749 29335595
28 Tian H.Y. Yu P.B. Luis A.D. Bi P. Cazelles B. Laine M. Changes in rodent abundance and weather conditions potentially drive hemorrhagic fever with renal syndrome outbreaks in Xi’an, China, 2005-2012 PLoS Negl. Trop. Dis. 9 3 2015 e0003530
29 Sun W. Liu X. Li W. Mao Z. Sun J. Lu L. Effects and interaction of meteorological factors on hemorrhagic fever with renal syndrome incidence in Huludao City, northeastern China, 2007-2018 PLoS Negl. Trop. Dis. 15 3 2021 e0009217
30 Xiang J. Hansen A. Liu Q. Tong M.X. Liu X. Sun Y. Impact of meteorological factors on hemorrhagic fever with renal syndrome in 19 cities in China, 2005-2014 Sci. Total Environ. 636 2018 1249 1256 29913587
31 Zhang W.Y. Guo W.D. Fang L.Q. Li C.P. Bi P. Glass G.E. Climate variability and hemorrhagic fever with renal syndrome transmission in northeastern China Environ. Health Perspect. 118 7 2010 915 920.32 20142167
32 Jiang F. Wang L. Wang S. Zhu L. Dong L. Zhang Z. Meteorological factors affect the epidemiology of hemorrhagic fever with renal syndrome via altering the breeding and hantavirus-carrying states of rodents and mites: a 9 years’ longitudinal study Emerg. Microb. & Infect. 6 11 2017 e104
33 Wei Y. Wang Y. Li X. Qin P. Lu Y. Xu J. Meteorological factors and risk of hemorrhagic fever with renal syndrome in Guangzhou, southern China, 2006-2015 PLoS Negl. Trop. Dis. 12 6 2018 e0006604
34 Xiao H. Tian H.Y. Gao L.D. Liu H.N. Duan L.S. Basta N. Animal reservoir, natural and socioeconomic variations and the transmission of hemorrhagic fever with renal syndrome in Chenzhou, China, 2006-2010 PLoS Negl. Trop. Dis. 8 1 2014 e2615
35 Xiao H. Tian H.Y. Cazelles B. Li X.J. Tong S.L. Gao L.D. Atmospheric moisture variability and transmission of hemorrhagic fever with renal syndrome in Changsha City, mainland China, 1991-2010 PLoS Negl. Trop. Dis. 7 6 2013 e2260
36 Zheng L. Gao Q. Yu S. Chen Y. Shi Y. Sun M. Using empirical dynamic modeling to identify the impact of meteorological factors on hemorrhagic fever with renal syndrome in Weifang, northeastern China, from 2011 to 2020 PLoS Negl. Trop. Dis. 18 6 2024 e0012151
37 Luo Y. Zhang L. Xu Y. Kuai Q. Li W. Wu Y. Epidemic characteristics and meteorological risk factors of hemorrhagic fever with renal syndrome in 151 cities in China from 2015 to 2021: retrospective analysis JMIR Public Health Surveill. 10 2024 e52221
38 Liu X. Jiang B. Gu W. Liu Q. Temporal trend and climate factors of hemorrhagic fever with renal syndrome epidemic in Shenyang City, China BMC Infect. Dis. 11 2011 331 22133347
39 Zhao Q. Yang X. Liu H. Hu Y. He M. Huang B. Effects of climate factors on hemorrhagic fever with renal syndrome in Changchun, 2013 to 2017 Medicine (Baltimore) 98 9 2019 e14640
40 Tian H. Tie W.F. Li H. Hu X. Xie G.C. Du L.Y. Orthohantaviruses infections in humans and rodents in Baoji, China PLoS Negl. Trop. Dis. 14 10 2020 e0008778
41 Tian H. Yu P. Cazelles B. Xu L. Tan H. Yang J. Interannual cycles of Hantaan virus outbreaks at the human-animal interface in Central China are controlled by temperature and rainfall Proc. Natl. Acad. Sci. USA 114 30 2017 8041 8046 28696305
42 Imai C. Armstrong B. Chalabi Z. Mangtani P. Hashizume M. Time series regression model for infectious disease and weather Environ. Res. 142 2015 319 327 26188633
