Background
Dengue fever is a common mosquito-borne viral disease of humans transmitted by the bite of an infected female adult mosquito namely the
Aedes aegypti as the primary vector and
Aedes albopictus as the secondary vector.
Aedes aegypti is a holometabolous insect with a life cycle consisting of four stages, namely: egg, instars, pupa and adult [
1]. It is characterized with a round or globular head structure, a white flat scales in the middle vertex and a slender, black, long and cylindrical proboscis [
2]. Dengue can cause fatal complications such as dengue haemorrhagic fever (DHF) and dengue shock syndrome [
3] . Dengue virus belongs to the single-stranded RNA virus of the Flaviviridae family that has four viral serotypes namely; DEN-1, DEN-2, DEN-3, DEN-4 [
4]. The adult female
Aedes aegypti, the primary vector is a small, black mosquito with white markings around its body. Frequently, mosquito vectors lay its eggs in places within or near human dwellings; with only female adult mosquitoes transmitting the dengue virus [
3,
5].
Dengue remains a major public health concern in tropical and subtropical areas [
3]. Over the last 50 years, dengue incidence has increased by 30-fold and around 2.5 billion people live in areas where dengue is endemic. Moreover, an estimated 50–200 million cases of dengue infections occur annually in the world [
6]. The spread of dengue may be partly due to the increase of international travel, unplanned urbanization, rapid increase in population growth, lack of effective vector management, climate change and extreme weather events, and poor socio-economic status [
7‐
9].
Dengue disease has risen in an alarming state in the Philippines in recent years. From January 1st to August 6th of 2016, the Philippines’ Department of Health (DOH) reported an estimated suspected dengue cases of 84,085 in the country, which is 15.8% higher compared to the same period of last year in 2015 with only 72,627 reported cases; out of this, 372 resulted to death [
10]. Out of the 10 Association of South East Asian Nations (ASEAN), the Philippines ranked fourth for having the highest number of dengue cases as of 2012 [
11]. This alarming rate is partly due to several factors such as environmental degradation, climatic condition, lack of clean water supply, inappropriate waste disposal and management, rapid urbanization, increasing population, and poor mosquito surveillance and control system all contributed to the increasing number of dengue cases in the country [
12].
The increasing dengue incidence worldwide is caused by several factors and one of them, which is our primary focus, are the meteorological factors. Change in these factors is believed to influence people’s health through the spread of vector-borne diseases [
8]. For example, meteorological factors such as temperature, rainfall, and humidity influence the life stages of female adult
Aedes mosquitoes. A warm temperature is important to adult mosquitos’ behavior and maturation, especially the larval development rate is shortened [
13,
14]. In addition, rainfall provides plenty of breeding sites for mosquito vectors such as puddles, while humidity affects the adult mosquitoes’ survival and biting frequency [
15].
Many countries have conducted studies on the relationship between meteorological factors and dengue cases. For instance, in the temperature and dengue relationship studies, different lagged effect was observed. An increase in RR of dengue was reported to be related with an increase in minimum and maximum temperature by up to 1 to 2-month lag period in Brazil [
16] and a lag of 1 month for maximum temperature in Mexico [
17]. Meanwhile, a longer lag of up to 3 to 4 months was observed in Australia [
18] and Barbados [
19]. Furthermore, cumulative rainfall and dengue have also been found to have a varying lag effect; such as a 2-week lag in Mexico [
20], a 4-week lag in Thailand [
21], and a 10-week lag in Taiwan [
22]. The differences on the effects of weather on dengue incidences could be due to the different variations of the amount of rainfall or the range of temperatures in the different regions with respect to their geographical locations [
23].
Various approaches have been utilized to estimate the risks and the associated delayed risks of various local meteorological variables on dengue incidence, with a variety of linear [
24] and non-linear models [
25]. Recent methodological advancements have resulted to the utilization of a distributed lag nonlinear model (DLNM) [
26], taking into account the bi-dimensionality of the risks, exposure and lag components, evident in previous studies [
27‐
30]. Further methodological specifications of DLNM are extensively discussed elsewhere [
26,
31,
32]. Here, we aim to elucidate the effects of the local meteorological variables, one of the significant driving forces on dengue transmission [
33]. This will enable us to determine the period of high risk of dengue infection since we hypothesized that, local meteorological factors may have an impact on the dengue incidences in Davao Region, Philippines. Clarifying the effects of these meteorological factors on dengue could provide an insight into the seasonal mechanisms of the disease thereby rendering information to understand the complex relationship between meteorological factors and health. Moreover, this study could provide valuable information to health officers for a more effective management of the disease for its control and prevention.
Discussion
In this study, we have observed significant but varying non-stationary periodicities between the local meteorological variables and that of dengue incidence. Further analysis indicated the effects of the varying levels of meteorological variables on dengue incidence. Findings from our study can be utilized for an integrated dengue early warning system, relevant for disease control and management.
In Fig.
4, we have observed significant, but varying inter-annual periodicities of dengue incidence with respect to the specific meteorological variables of interest. In particular, there are consistent longer significant bands from the 16th to 32nd week, for all the three local variables, which coincides with the summer month of April and gradually transitions towards the rainy month of August. There are also small pockets of significant mild periodicities in the 4th to 16th week across the years amongst the three meteorological variables, however, we have observed a significant inter-annual periodicity (2010–2012) from the 6th week to the 15th week, dry season, apparent only with average temperature. Though some studies have noted that mosquito activity would be high in the rainy periods, there are indications that even in dry season mosquito activity may be heightened. Wai et al. [
51] observed that vector breeding was enhanced in the dry season in the Philippines due to the presence of water storage, conducive for mosquito growth and development. Tsuzuki et al. [
52] also noted the potential of dengue transmission even in hot-dry periods in Nha Trang, Viet Nam, with similarly possible linkage to unchecked and left out indoor water containers.
Beyond the non-stationary relationship observed in the wavelet coherence patterns, we further investigated the impact of local meteorological variables on the dengue morbidity, with the primary focus on identifying the lag (weeks) in which dengue cases has occurred. Based on DLNM, there was a positive association between rainfall, average temperature and dew point with dengue cases. We found that, a moderate amount of rainfall has resulted with increasing RR and gradually decreased as the amount of rainfall increased.
Our findings are consistent with previous studies, whereby a high occurrence of dengue in the few weeks were also observed after moderate rainfall. Ehelepola et al. [
53] have found that regular rain favors an increase in dengue but not heavy rain. Similarly, Sarfraz [
54] noted that heavy rainfalls may flush away the eggs, larvae and pupae of dengue mosquitoes, which could have consequently affected the mosquito abundance [
55]. On the other hand, we have observed that moderate amount of rainfall (20–30 mm) was related to higher dengue incidences in Davao Region. Eastin et al. [
56] and Hii et al. [
5] noted that light to moderate rainfall can increase the usage of water containers, which are conducive breeding sites for the mosquito.
According to DOH, Davao City which is located near Davao del Sur and a highly urbanized area in Davao Region, has always had the highest number of dengue, comprising around 70% of the cases [
57] while the rest are from the neighboring provinces. Dengue is usually higher in highly urbanized areas like cities, where there is overcrowding and poor environmental waste management [
58]. In 2013, there was a spike in dengue incidence in Davao City, which was assumed to be linked to the lack of cleanliness drive, unpredictable weather conditions and floods in the area [
59] despite the increasing exertion of the DOH and the local health units in Davao Region on the implementation of dengue vector control programs throughout the years. One of the initiated clean-up activities by the DOH is the 4 o’clock habit, which entails the search and destruction of possible mosquito breeding sites, usually done in four in the afternoon [
59].
Average temperature at 26 °C has resulted to an increased RR, while higher temperature from 27 °C to 31 °C has lower RR. This was suggestive of the established fact that mosquito development has an optimum range of 25 to 27 °C [
60]. This optimum temperature for mosquito strongly enhanced the development from larva to adult, the biting frequency in humans, and the extrinsic incubation period of dengue virus in the mosquito. The decrease in the RR of temperatures from 27 °C to 31 °C is indicative that higher temperature above the optimum range for the
Aedes mosquito development brings about a protective effect on dengue transmission [
61], as observed in Fig.
5. Higher temperatures may have a negative effect on adult life span of mosquitoes, thereby affecting consequent transmission [
62]. In particular, reduced vector competence and activity may result from an increased temperature [
63].
Beyond a dew point temperature of 25.3 °C, risks were apparently increasing. Similar observations of significant relationship by dew point on dengue incidence was observed in Brazil [
64]. Mechanisms of how dew point affects dengue incidence maybe related to that of the mechanisms posed by humidity. Mathematically and theoretically, dew point and humidity have a nearly linear relationship [
65]. Taking this into account, we can observe an identical J-shaped pattern of absolute humidity and RR in a study done in Singapore [
29] compared to Fig.
4. High humidity favors an increased longevity of adult mosquitoes as well as the shortening of viral incubation period, thereby allowing an increased transmission intensity [
66].
We acknowledge that under-estimation of the burden poses one of the few challenges in establishing the robustness of the effects estimates in this study. However, even after applying an EF of 7, effects estimates as well as the risk curve remained the same, making the estimates robust (in Additional file
5: Figure S4). Furthermore, a recent study in the Philippines noted a high proportion of laboratory confirmed cases (86.1%) from the suspected cases, which thereby indicate that accuracy of clinical diagnosis at admission [
67].
This is the very first study in the Philippines which extensively described the association between local meteorological variables and dengue incidences. This study could help improve the dengue surveillance system in the country by taking into account the underlying mechanisms which could be framed in the context of non-stationary relationship as well as the candidate threshold levels, for better dengue prediction. Furthermore, we also acknowledge some limitations in this study. First, the weekly dengue data used in this study were notified suspected dengue cases from clinics and hospitals and are not laboratory confirmed. Second, we did not take into account the mosquito density, population immunity, age classification, social behavior, and socioeconomic conditions for these data are unavailable.
Acknowledgements
We would also like to thank the National Oceanic and Atmospheric Administration for the meteorological data, as well as the Department of Health-Davao Region (Region XI) regional director Dr. Abdullah B. Dumama Jr., for generously providing the weekly dengue data report. Without their generosity, this study wouldn’t be possible.