1. IntroductionDengue fever (DF), a virus infection spread by mosquitoes, is most common in urban and suburban areas of tropical and subtropical climates. The incidence of dengue is increasing, and has reached an estimated 390 million annual cases, of which 96 million presents with clinical symptoms [1,2]. It is the most rapidly spreading vector-borne disease and is now endemic in over 100 countries, with half of the world’s population living in a dengue-risk area [3].Dengue epidemics are known to be related to fluctuations in temperature, precipitation, relative humidity (RH) and socioeconomic factors, such as urbanization and human movement [4,5,6,7,8,9,10,11,12]. Previous study has suggested that a 2 °C increase in temperature would simultaneously lengthen the lifespan of the mosquito and shorten the extrinsic incubation period of dengue virus, resulting in more infected mosquitoes for a longer period of time [13]. Recent research has found that diurnal temperature ranges are more important than average temperature when studying the development and transmission of dengue and malaria parasites [14,15,16]. Precipitation could provide water for breeding because Aedes spp. mosquitos prefer artificial containers around human-made environments [10]. However, mosquito survival is also reduced by extreme temperatures and large rainfall events [17].Taiwan is located in a subtropical and tropical region, with relatively high temperature and RH year-round, which is an ideal condition for mosquitoes that transmit dengue fever [18]. The primary mosquito vector of dengue virus in Taiwan is Aedes aegypti, followed by some transmission by Aedes albopictus [19]. Dengue fever is a communicable disease that must be reported, and a national web-based communicable diseases surveillance system has been established since 1997 [20]. According to the reports from Taiwan Centers for Disease Control, the main epidemic areas of dengue fever from 1998 to 2021 are located in southern Taiwan, including Kaohsiung City (45,658 indigenous dengue cases), Tainan City (26,580 indigenous dengue cases), and Pingtung County (2058 indigenous dengue cases) [20]. Over the last decade, two consecutive severe dengue epidemics occurred in 2014 and 2015 in southern Taiwan [21,22,23,24,25,26]. The death rate in 2015 dengue hemorrhagic fever (DHF)/dengue shock syndrome (DSS) cases was higher than in previous dengue epidemics or outbreaks [21]. Evidence-based study indicated that the majority of population in Taiwan born after 1940 is naïve to dengue virus and the prevalence of IgG antibody against dengue virus rises with age [27].Associations between temperature, RH, precipitation and dengue transmission have been mentioned in previous studies but there is no consistent conclusion have been found [18,28,29]. Several studies have been conducted to explore the association between climate variables and dengue fever with lagged effect during epidemic [6,30,31]. The time lagged effect could be caused by the lifespan of the mosquito and the incubation period of dengue virus [31].

Therefore, this study aims to link meteorological factors with weekly incidence rate of dengue fever to investigate the key determinants during epidemic and non-epidemic periods. We also try to explore the difference of time-lag effects of meteorological factors during dengue-epidemic (2014–2015) and non-epidemic periods (2007–2013 and 2016–2017) by statistical models.

4. Discussion

This study linked meteorological factors and epidemiological data to characterize the patterns in southern Taiwan from 2007 to 2017. We try to explore the difference and time-lag effect by meteorological factors during epidemic (2014–2015) and non-epidemic periods (2007–2013 and 2016–2017) by statistical models. Several aspects of dengue fever risk in southern Taiwan are revealed by our analysis. The major findings are as follows: (i) the effect of temperature variation may have a significant impact on dengue incidence rates than the other factors when compared with non-epidemic (2007–2013 and 2016–2017) and epidemic periods (2014–2015), (ii) the results show that 3–4 months lag in temperatures explained 54–56% (non-epidemic periods) and 22–29% (epidemic periods) of the variability in dengue incidences in two time periods in Tainan region, and (iii) For epidemic periods, the maximum temperature with 2-month lag (β = 0.8400, p < 0.001) and minimum temperature with 5-month lag (0.3832, p < 0.001); for non-epidemic periods minimum temperature with 3-month lag (0.1737, p < 0.001) and mean temperature with 2-month lag (2.6743, p < 0.001) had a positive effect on dengue incidence in Tainan and Kaohsiung, respectively.

According to the meta-analysis study [4], both temperature and precipitation increase the risk of dengue fever in tropical and subtropical regions. Temperature (RR = 1.13, 95% CI: 1.12, 1.15) and precipitation (RR = 1.01, 95% CI: 1.01, 1.01) were also found to have a statistically significant correlation with the risk of dengue fever [4]. Lee et al. [8] found that temperature and precipitation were positively associated with the occurrence of disease from 1994 to 2013 in Vietnam which was consistent with findings from the previous studies [37,38,39]. Huang et al. [40] found that the relative risk of dengue fever increased when the weekly average temperature was more than 15 °C at lagged weeks 5 to 18 in a study of meteorological factors affecting dengue fever in Kaohsiung in 2015. Yuan et al. [41] found that dengue transmission has a positive relationship with the minimum temperature predictors during the early summer while a negative relationship with all the maximum 24-h rainfall predictors during the early epidemic phase of dengue outbreaks. In our study, minimum and mean temperatures with time lag shown the significantly effects on dengue incidence in Tainan and Kaohsiung at non-epidemic periods (Table 2); However, in epidemic periods, maximum and minimum temperature shown the significantly factors (Table 2).In this study, temperature with time lag shown the different effects on dengue incidence in Tainan during the epidemic and non-epidemic periods; however, rainfall with time lag shown the significantly positive effect on dengue incidence in Kaohsiung city during the epidemic and non-epidemic periods (Table 2). Previous studies have also described the high temperatures and heavy rains in 2014 and 2015 in southern Taiwan [26,42]. Furthermore, one study mentioned the dengue outbreak in Kaohsiung city in 2014 might be due to the natural gas explosion incident on 1 August 2014 [42]. After the explosion, this situation created an ideal environment for Aedes mosquitoes [43,44]. The outbreak in 2015, Tainan city was the original division of the outbreak, and it had the highest prevalence rate of dengue. The reasons for this outbreak may include geographical location, high population density, and inadequate control of dengue outbreaks [26].The typhoon season from July to September has always been the main source of rainfall in Taiwan. During the most intense typhoons, a total of more than 1000 mm of rain has fallen in a few days. According to the Central Weather Bureau [34], different typhoon tracks affect the magnitude of rainfall in different regions of Taiwan. During the epidemic periods, Taiwan experienced two moderate typhoons and rain incidents occurred along with the southeast airflow. In 2014, southern Taiwan continued to fall from 7 August to 12 August. It is shown that the 24-h cumulative rainfall in Tainan’s Administrative Westport District and Anding District exceeded 350 mm, reaching super heavy rain. The accumulated rainfall in 24 h exceeded 200 mm, reaching the standard of heavy rain [45]. During typhoon season, the heavy rainfall causes a hot, wet environment with a high RH that is ideal for mosquito breeding [44]. Even though our research showed that the RH with time lag had different effects on dengue incidence in Tainan and Kaohsiung during the epidemic and non-epidemic periods.The main dengue control strategies in Taiwan are to eliminate vector breeding sources and effectively reduce vector (mosquito) density. Three-stage prevention strategy for dengue epidemic control have been implicated. Initial control measures include source reduction and control of vector populations. Secondary measures include disease surveillance and emergency/precautionary mechanisms, and tertiary prevention includes mortality control [32].A few limitations of this study warrant mention. First, this study only considers the major metrological factors, such as the weekly max., mean, and min. temperatures, rainfall intensity, and RH. The following variables were not considered: daily maximum rainfall, ELSO, population density, mosquito density, or human activity. With regards to the human populations, this study assumes a homogenous population even though exposure to mosquitoes and dengue risk are known to vary because of socioeconomic factors [46], race or ethnicity, and lifestyle choices such as time spent outdoors [47]. Moreover, the long-term investigation of the Breteau Index or Contain Index in Taiwan was also limited. The temporal and spatial distributions of the mosquito population or host population were not presented. 5. Conclusions

This study has identified the weekly maximum (max.), mean, and minimum (min.) temperatures (°C), relative humidity (%), and rainfall intensity (mm) as potential contributors, with max. temperature at lag 2 months and lag 5 months being the most significant, for dengue incidence in Tainan and Kaohsiung, respectively. Therefore, the results of this study suggest that vector control interventions in dengue epidemics are more effective at the early stage of the epidemic.