Michael Johansson and colleagues use wavelet analysis to show that there is limited evidence for a multiyear relationship between climate and dengue incidence in Puerto Rico, Mexico, and Thailand.
The mosquito-borne dengue viruses are a major public health problem throughout the tropical and subtropical regions of the world. Changes in temperature and precipitation have well-defined roles in the transmission cycle and may thus play a role in changing incidence levels. The El Niño Southern Oscillation (ENSO) is a multiyear climate driver of local temperature and precipitation worldwide. Previous studies have reported varying degrees of association between ENSO and dengue incidence.
We analyzed the relationship between ENSO, local weather, and dengue incidence in Puerto Rico, Mexico, and Thailand using wavelet analysis to identify time- and frequency-specific association. In Puerto Rico, ENSO was transiently associated with temperature and dengue incidence on multiyear scales. However, only local precipitation and not temperature was associated with dengue on multiyear scales. In Thailand, ENSO was associated with both temperature and precipitation. Although precipitation was associated with dengue incidence, the association was nonstationary and likely spurious. In Mexico, no association between any of the variables was observed on the multiyear scale.
The evidence for a relationship between ENSO, climate, and dengue incidence presented here is weak. While multiyear climate variability may play a role in endemic interannual dengue dynamics, we did not find evidence of a strong, consistent relationship in any of the study areas. The role of ENSO may be obscured by local climate heterogeneity, insufficient data, randomly coincident outbreaks, and other, potentially stronger, intrinsic factors regulating transmission dynamics.
Every year, as many as 50–100 million people become infected with one of four closely related dengue viruses through the bite of a female Aedes aegypti mosquito that has acquired the virus by feeding on infected human blood. Dengue is endemic (always present) in many tropical and subtropical countries but its incidence (the number of new cases in a population over a given time period) follows a seasonal pattern. This is because the abundance of Ae. aegypti is regulated by rainfall, which provides breeding sites and stimulates egg hatching, and by temperature, which influences the insects' survival and their rate of development and reproduction. Temperature also affects the mosquitoes' ability to transmit dengue virus—higher temperatures increase transmission rates. Although some people who become infected with dengue have no symptoms, many develop dengue fever, a severe, flu-like illness that lasts a few days. Other people—more than half a million a year—develop dengue hemorrhagic fever, a potentially fatal condition. There is no vaccine to prevent dengue and no specific treatment for the disease, but standard medical care can prevent most deaths from dengue.
As well as seasonal variations in the incidence of dengue, large dengue outbreaks (epidemics) occur every few years. To help with health care planning, public health officials would like a way to predict when these epidemics are likely to occur, but to develop such a system requires a good understanding of the factors that lead to major epidemics. Although variations in host–virus interactions (for example, changes in host immunity to dengue) almost certainly play an important role in the timing of dengue epidemics, interannual changes in temperature and rainfall could also be involved. One major cause of global interannual weather variation is the El Niño Southern Oscillation (ENSO), a climate cycle centered on the Pacific Ocean that repeats every 3–4 years. Previous studies have reported varying degrees of association between ENSO and dengue. In this study, the researchers reanalyze the relationship between ENSO, local weather, and dengue incidence in three dengue-endemic countries using “wavelet analysis.” This mathematical technique can separate the effects of seasonal weather variations on dengue incidence from those of interannual weather fluctuations.
The researchers retrieved data on the incidence of dengue fever and dengue hemorrhagic fever in Puerto Rico, Thailand and Mexico since the mid 1980s from national surveillance systems. They also collected historical weather data for each country and information on ENSO. They then used these data and wavelet analysis to investigate the relationship between ENSO, local weather, and dengue incidence in each country on the annual scale and on the multiyear scale. On the annual scale, temperature, rainfall, and dengue incidence were strongly associated in all three countries. On the multiyear scale, ENSO was associated with temperature and with dengue incidence in Puerto Rico, but only for part of the study period. Only local rainfall was associated with the incidence of dengue in that country. The lack of a direct path of association from ENSO to either weather variable to dengue incidence suggests that the ENSO–dengue association may be a spurious result. In Thailand, ENSO was associated with both temperature and rainfall, and rainfall was associated with dengue incidence. However, detailed analysis suggests that this latter association was also probably spurious. Finally, there was no association between any of the variables in Mexico on the multiyear scale.
Although these findings show a strong associations between both temperature and rainfall and dengue incidence on the annual scale in Puerto Rico, Thailand, and Mexico, they provide little evidence for any relationship between ENSO, climate, and dengue incidence. Multiyear climate variability may play a role in interannual variations in dengue incidence, the researchers suggest, but their study does not provide any evidence for a strong and consistent relationship between climate variability and dengue incidence. It is possible that the effects of ENSO on dengue incidence are being masked by local variations in weather or by stronger factors regulating disease transmission such as host–virus or host–vector interactions. Future studies into the relationship between dengue outbreaks and multiyear climate variability will need to include these and other factors. For now, however, information on ENSO cannot be used to design an early warning system for dengue outbreaks.
Please access these Web sites via the online version of this summary at http://dx.doi.org/10.1371/journal.pmed.1000168.
This study is further discussed in a PLoS Medicine Perspective by Pejman Rohani
The US Centers for Disease Control and Prevention provides detailed information about dengue fever and dengue hemorrhagic fever (in English and Spanish)
The World Health Organization provides information on dengue fever and dengue hemorrhagic fever around the world (in several languages)
Links to additional resources about dengue are provided by MedlinePlus (in English and Spanish)