Large-scale environmental forces influence infectious diseases

Large-scale environmental forces influence infectious diseases. This is clearly the case in the Caribbean and Gulf of Mexico (Chretien et al. 2015, Dobson 2009). Variability of specific environmental factors affects dengue fever occurrence and water quality of recreational beaches (Chowell and Sanchez 2006, Pednekar et al. 2005). Thus, it should be possible to develop better management, disease surveillance, and mitigation strategies by understanding the variability of environmental forces and their influence on public-health related issues. In this dissertation, I examined these problems in more detail in the northwest coast of the State of Yucatan, Mexico, and near San Juan, Puerto Rico, USA.
2.1 Environmental and demographic factors influence on vector-borne diseases
Human populations in the Caribbean Sea and the Gulf of Mexico have seen an increase in the incidence of vector-borne diseases. Dengue fever cases have increased especially since the 1970s (Dick et al. 2012, Laureano-Rosario et al. 2017, Mendez-Lazaro et al. 2014). This increase is in part due to the adaptation of the mosquito, Aedes aegypti, to live in urban areas (Gratz 1991, Gubler 2002). Previous studies have shown the influence of specific environmental and demographic factors on the occurrence of dengue fever cases in places like Yucatan State, Mexico and San Juan, Puerto Rico (Colon-Gonzalez et al. 2011, Colon-Gonzalez et al. 2013, Mendez-Lazaro et al. 2014). Furthermore, local environmental factors and population behavior play a key role in the epidemiology and phenology of dengue fever (Eastin et al. 2014). Consequently, the understanding of the local variability of environmental factors is important to understand their influence on dengue fever occurrences.
Dengue fever is mostly transmitted by Aedes aegypti, a mosquito found around tropical and subtropical areas (Gubler 2002). These mosquitoes use water containers (natural and artificial) to develop, being precipitation and temperature the main promoters of their development (Brady et al. 2013, Campbell-Lendrum et al. 2015, Descloux et al. 2012, Johansson et al. 2009). Warmer temperatures decrease mosquito development time, increasing mosquito egg production, hatching, and density (Dickerson 2007). Furthermore, increased temperatures lead to higher metabolic activity, which promotes more mosquito biting (by female mosquitoes) due to energetic demands (Paaijmans et al. 2013). Both Mexico and Puerto Rico have reported Aedes albopictus as another vector for dengue fever (Dantes et al. 2014, Dick et al. 2012, Mendez-Lazaro et al. 2014, Stramer et al. 2012). Dengue has four serotypes (DENV-1, DENV-2, DENV-3, and DENV-4; Halstead 1988), which have been reported in both Mexico and Puerto Rico. More recently, studies have shown the emergence of sylvatic dengue 5 (DENV-5; Joob and Wiwanitkit 2016, Mustafa et al. 2015). Peaks in dengue cases usually take place after a shift from one serotype to another, since during this time the population would only be partially immune to the other serotypes (Gubler and Clark 1995, Rothman 2004). Relevant epidemiological studies in Yucatan and Puerto Rico have focused on understanding where Aedes aegypti’s larvae are found (e.g., schools, households) and how the disease is transmitted (Baak-Baak et al. 2014a, Baak-Baak et al. 2014b, Garcia-Rejon et al. 2008, Garcia-Rejon et al. 2011). In both tropical locations, dengue fever coincides with periods of higher precipitation, higher SST, higher mean sea level, and higher minimum air temperature along the coast.
Climatic variations are expected to influence the ecology and geographic distribution of vector-borne diseases. Studies have shown how vectors that transmit malaria (i.e., Anopheles spp.) have been found in higher altitudes in Africa due to warmer temperatures (Afrane et al. 2007, Afrane et al. 2012, Harvell et al. 2002). Similarly, studies have documented both increases and re-occurrences of vector-borne diseases in Europe due to recent warmer conditions (Medlock and Leach 2015). Nevertheless, these are also affected by human activities such as population movement, farming, dams, and changes in irrigations systems. Therefore, some of these climatic effects might be masked by human activities, including human population movement across the world, leading to further spreading and increasing incidence rates (Campbell-Lendrum et al. 2015).
Modelling dengue fever in endemic areas is important to better mitigate and manage these occurrences. The present work was driven by the hypothesis that variability and trends in environmental factors (e.g., precipitation, temperatures, and humidity) are primary drivers of dengue fever incidence, and that including satellite-derived SST improves dengue fever incidence rate predictions. The objective was to help improve epidemiological surveillance through the combination of oceanographic, meteorological, and long-term epidemiological data.
2.2 The influence of environmental factors on fecal indicator bacteria and recreational water quality
Water quality is a major concern to coastal communities due to the potential for exposure to pathogens in beaches downstream of watersheds with sources of fecal contamination (Garcia-Montiel et al. 2014, Pruss 1998, Soderberg 2012). Wastewater discharges are point sources. Other sources include septic tanks and open sewers that discharge directly to river streams. Likewise, resuspension of bacteria by winds and waves, and stormwater discharges are potential non-point sources of fecal contamination in coastal areas (Cordero et al. 2012, Quiñones 2012, Rochelle-Newall et al. 2015).
Fecal indicator bacteria (FIB) are used by the United States Environmental Protection Agency (U.S. EPA) to identify poor recreational water quality. Out of these FIB, culturable enterococci are commonly used in fresh and marine waters (U.S. EPA 2012). The U.S. EPA established the 2012 Recreational Water Quality Criteria (RWQC), where these culturable enterococci cannot exceed the geometric mean of 35 colony forming units (CFU) per 100 mL. This represents 36 illnesses per 1,000 primary contact recreators (U.S. EPA 2012). This value was modified in 2014 to the Beach Action Value (BAV) of 70 CFU/100 mL based on specific criteria for conducting research (U.S. EPA 2014). These guidelines were adopted by the Environmental Quality Board of Puerto Rico (PREQB). In Puerto Rico, the PREQB assesses bathing water quality at beaches throughout the island every two weeks, and if concentrations exceed those values set by the U.S. EPA (i.e., BAV of 70 CFU/100 mL; PREQB 2016), they issue beach advisories. These data are openly available but are only used for issuing public warnings.
FIB variability has been associated with environmental forces in both subtropical and tropical regions (Aranda et al. 2016, Lamparelli et al. 2015, Viau et al. 2011, Wright et al. 2011). These studies have shown how specific environmental factors (e.g., precipitation, turbidity, temperatures) influence higher or lower FIB concentrations in marine and fresh waters (Byappanahalli et al. 2010, He and He 2008, Nevers and Whitman 2005). Therefore, a series of statistical models (e.g., linear and multiple regression models) were used to better understand variability of culturable enterococci concentrations. This was guided by the hypothesis that changes in culturable enterococci concentration in surface waters at Escambron Beach (Puerto Rico) were related to variations of environmental factors (e.g., SST, turbidity, precipitation). The main objective was to improve early warnings for FIB and health risks.
2.3 Predicting vector-borne diseases and recreational water quality with Artificial Neural Networks
Predictive models can help improve management and mitigation of health-related matters (de Brauwere et al. 2014, Gonzalez and Noble 2014, Gubler 2010, Tabachnick 2010). In this dissertation, a nonlinear model was used to evaluate prediction of dengue fever outbreaks in endemic areas, as well as exceedances of FIB in tropical areas.
Modelling can help predict and understand the epidemiology of dengue fever in endemic areas (Medeiros et al. 2011, Racloz et al. 2012). Likewise, recreational water quality modelling helps protect humans from potential exposure to specific FIB (Colford et al. 2007, Pruss 1998). For example, some studies have applied Monte Carlo and support vector machine to predict dengue fever cases (Husin et al. 2008, Wu et al. 2008). Similarly, nonlinear modelling using ANNs, decisions trees, and Monte Carlo approaches helped model water quality (Jiang et al. 2013, Lin et al. 2008) and have supported beach management (Mavani et al. 2014, Zhang et al. 1998, Thoe et al. 2014).
Predicting health-related matters is a management goal. These ANN models do not assume functional relationships between predictor factors (e.g., environmental factors) and target variable (e.g., dengue fever, culturable enterococci concentration), thus they can identify nonlinear, complex relationships (Zhang et al. 1998). ANN models were applied to predict dengue fever outbreak occurrences in Mexico and Puerto Rico for specific population segments (i.e., population younger than 24 years and those younger than 5 years and older than 65 years). These ANNs models were also applied to predict culturable enterococci concentration exceedance in surface waters at Escambron Beach in Puerto Rico. The objective was to help management and mitigation of these two health-related matters.