Assessing the Impact of Climate Factors on Dengue Transmissions in Puerto Rico

Dengue fever, also known as "bone crusher's disease," affects 50-100 million people worldwide; this year, 2.5 billion people are living in at risk areas as opposed to only as 66,000 in 1980. In the United States, dengue fever has been found in Texas, the Virgin Islands, and Puerto Rico. The symptoms range from asymptomatic to a 20% mortality rate. Unfortunately, dengue can be clinically confused with other diseases such as influenza, and no specific medical treatment or vaccine is currently available. As a result, dengue is a serious disease for the at risk public.

The dengue virus is transmitted in the Caribbean primarily through the mosquito vector Aedes aegypti. Its habitat is primarily in the tropics; however, global warming studies show that the mosquito's range is expanding around the world. Elevated temperatures can also shorten both larval development of Aedes aegytpi, and the period for dengue virus replication (extrinsic incubation period) in the mosquito. Both cases lead to increased transmission of dengue in the at risk areas. Therefore, to better understand dengue transmission and epidemics, a thorough analysis of climatic factors that affect the Aedes aegypti population and life cycle was performed.

This study focuses on understanding the dengue epidemics in Puerto Rico in the 1990s. The factors to be investigated include relative humidity, precipitation, air temperature, sea-surface temperature (SST), and tropical cyclones. The general hypothesis is that high air temperatures and SSTs will produce high dengue transmissions; high precipitation, relative humidity, and strong tropical cyclones following dry periods will have the same effect. Each of the aforementioned climate factors was correlated with dengue outbreaks on a weekly basis. Theoretically, precipitation factors into mosquito populations. However, in most of Puerto Rico, people keep many water-filled artificial containers that serve as mosquito breeding sites. This makes the mosquito population independent of precipitation.

Relative humidity showed slightly better relationships with dengue, but still generally weak most likely, because humidity shows little variability in a tropical environment and would therefore have minimal impact on the Aedes aegypti lifecycle. In contrast, SST and dengue showed strong, positive, and exponential correlations. To better understand this relationship between SST and dengue outbreaks, statistical tests were performed in the statistical program R. For 1994, the nonlinear graph was transformed to show a linear relationship using the log function for the dependent variable. The new linear relationship was analyzed using linear regression. The model had an r2 of 0.66, which was very strong for this type of analysis. The residual plot had no distinct pattern, and the t-statistic was significant at the 0.1% level. The 1998 data, however, could not be successfully analyzed with one model, so a threshold analysis by a regression tree was done. The SST threshold was found to be around 29„aC. Next, weekly SST values over an eleven-year interval was plotted to see how SSTs varied between epidemic years and non-epidemic year. SSTs in 1994 were not much higher than SSTs from non-epidemic years. SSTs in 1998 were high compared to the other ten years, but that is because it was a La Nina year. In general, 1994 had the bigger epidemic but 1998 had the more extreme SST. Hence, higher SSTs may not necessarily result in more dengue outbreaks. Instead, the observed relationship in both years may be more of a seasonal phenomenon.

Temperature and dengue outbreaks do not have a strong correlation which agrees with a past study that claimed an increase in average global temperatures would cause the largest change in the Aedes aegypti population and dengue epidemic potential to be in temperate regions. The tropical and subtropical regions would experience little or no change. Since Puerto Rico is in the tropics, it is understandable to see that increased temperatures do not have a strong effect on dengue transmissions.

Air temperature thresholds for dengue outbreaks were evaluated using the regression tree in R. The thresholds were approximately 30„aC and 32„aC. These thresholds and the SST threshold agree with past work that claimed the highest Aedes aegypti survival rate was between 20 and 30„aC.

Finally, a qualitative look at tropical cyclones occurrence before and during the two epidemic years showed that there were no tropical cyclones near Puerto Rico in 1994 when the largest epidemic in the 1990s occurred. During the 1998 epidemic, Hurricane Georges hit Puerto Rico right after the height of the dengue outbreak. Dengue reporting was briefly interrupted and then returned to above average levels. In this case, for uncertain reasons, Geroges produced no significant effects on dengue outbreaks. These data for two epidemics indicate that tropical storms do not have a predictable impact on dengue cases.

From the results of this study, SST is the only climate factor to show strong a correlation with dengue transmissions. However, even that relationship varied between 1994 and 1998. This could be because only two years were taken into consideration. Furthermore, a combination of climate factors could have affected dengue transmissions that were not examined in this study.

Aside from climate factors, social factors could be of equal or more importance in predicting dengue outbreaks in Puerto Rico. Although detailed information were not available, a preliminary investigation into some of the social factors involved in the large 1994 outbreak pointed to decay in public health infrastructures and increased air-travel. In 1994, about 20 million people left the United States and traveled to places where dengue is endemic. They could have been returning to U.S. territory with the virus and spreading it. Tourism is a likely factor in determining the number of dengue outbreaks. In general, without more in-depth research on social factors, it is difficult to determine how great of a role they play, but it is evident that both social and climate factors, especially SST, can contribute to the increase in dengue transmissions.