key: cord-103179-naynznc1
authors: Simon, L. M.; Rangel, T. F.
title: Are temperature suitability and socioeconomic factors reliable predictors of dengue transmission in Brazil?
date: 2020-11-04
journal: nan
DOI: 10.1101/2020.11.02.20224444
sha: 
doc_id: 103179
cord_uid: naynznc1

Background Dengue disease is an ongoing problem, especially in tropical countries. Like many other vector-borne diseases, the spread of dengue is driven by a myriad of climate and socioeconomic factors. Over recent years, mechanistic approaches have predicted areas of dengue risk according to the temperature effect on mosquitos' lifespan and incubation period shaping their persistence and competence in transmission. Within developing countries such as Brazil, heterogeneities on socioeconomic factors are expected to create variable conditions for dengue transmission by its main vectors. However, both the relative role of socioeconomic aspects and its association with the temperature effect in determining the effective dengue prevalence are poorly understood. Methodology/Principal findings Here we gathered essential socioeconomic factors comprising demography, infrastructure, and urbanization over 5570 municipalities across Brazil and evaluated their relative effect on dengue prevalence jointly with a previously predicted temperature suitability for transmission. Using a simultaneous autoregressive approach (SAR), we showed that the variability in the prevalence of dengue cases across Brazil is highly explained by the combined effect of climate and socio-economic factors. Moreover, the temperature effect on transmission potential might be a better proxy at some dengue epidemy seasons but the socioeconomic factors are tightly linked with the recent increase of the dengue prevalence over Brazil. Conclusions/Significance In a large and heterogeneous country such as Brazil recognizing the drivers of transmission by mosquitoes is a fundamental issue to effectively predict and combat tropical neglected diseases as dengue. Ultimately, it indicates that not considering socioeconomic factors in disease transmission predictions might compromise efficient strategies of surveillance. Our study indicates that sanitation, urbanization, and GDP are regional indicators that should be considered along with temperature suitability for dengue transmission, setting a good starting point to effective vector-borne disease control.

12 Background 13

Dengue disease is an ongoing problem, especially in tropical countries. Like many 14 other vector-borne diseases, the spread of dengue is driven by a myriad of climate and 15 socioeconomic factors. Over recent years, mechanistic approaches have predicted areas 16 of dengue risk according to the temperature effect on mosquitos' lifespan and 17 incubation period shaping their persistence and competence in transmission. Within 18 developing countries such as Brazil, heterogeneities on socioeconomic factors are 19 expected to create variable conditions for dengue transmission by its main vectors. 20

However, both the relative role of socioeconomic aspects and its association with the 21 temperature effect in determining the effective dengue prevalence are poorly 22 understood. 23

Here we gathered essential socioeconomic factors comprising demography, 25 infrastructure, and urbanization over 5570 municipalities across Brazil and evaluated 26 their relative effect on dengue prevalence jointly with a previously predicted 27 temperature suitability for transmission. Using a simultaneous autoregressive approach 28 (SAR), we showed that the variability in the prevalence of dengue cases across Brazil is 29 highly explained by the combined effect of climate and socio-economic factors. 30

Moreover, the temperature effect on transmission potential might be a better proxy at 31 some dengue epidemy seasons but the socioeconomic factors are tightly linked with the 32 recent increase of the dengue prevalence over Brazil. 33

In a large and heterogeneous country such as Brazil recognizing the drivers of 35 transmission by mosquitoes is a fundamental issue to effectively predict and combat 36 tropical neglected diseases as dengue. Ultimately, it indicates that not considering 37 socioeconomic factors in disease transmission predictions might compromise efficient 38 strategies of surveillance. Our study indicates that sanitation, urbanization, and GDP are 39 regional indicators that should be considered along with temperature suitability for 40 dengue transmission, setting a good starting point to effective vector-borne disease 41 control. 42 43 AUTHOR SUMMARY: 44 Dengue, a disease transmitted by mosquitoes, is a great problem in countries 45

where the climate is predominantly hot and wet. Researchers know that temperature 46 plays an important role in mosquitoes' ability to transmits diseases. Usually, 47 temperature alone is a good explanation for why dengue occurs in certain regions that 48 have stable warm temperatures. Here we show that, in addition to the role of 49 temperature on dengue spread, large urban areas with sanitation infrastructure and 50 health assistance also prelude dengue cases prevalence. We highlight that dengue 51 surveillance should consider socioeconomic regional differences. For instance, greater 52 urban centers might be the focus of the dengue burden, where the presence of medical 53 assistance and sanitation seems not to avoid the increase in disease cases. Conversely, 54 less urbanized regions with suitable temperature for dengue transmission might require 55 distinct actions for the disease combat. 56 57 . CC-BY-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 4, 2020. ; https://doi.org/10.1101/2020.11.02.20224444 doi: medRxiv preprint

The presence and prevalence of many infectious diseases have clear geographic 59

structures. These health threats vary from country-to-country and cause the loss of 60 millions of lives annually (1,2). Identifying patterns and drivers of infectious diseases 61 spread has become a fundamental concern on disease ecology (3). For instance, 62 understanding why some regions have a higher charge in diseases and pathogens 63 richness than others might help to identify hotspots for infections outbreak (2). 64

However, a multitude of factors determines infectious disease geographical distribution 65 and potential outbreaks, spanning from socioeconomic (e.g. urbanization; population 66 density) to environmental (e.g. temperature; precipitation) and biotic (e.g. vectors 67 competition) aspects (1,4). Acknowledging the variation over space in these drivers of 68 infectious disease is increasing among ecologists in an attempt to identify regions of 69 outbreak potential, once the disease dynamic is as tightly linked with exogenous factors 70 as it is with endogenous mechanisms (5,6). 71

Dengue, a mosquito-borne infectious disease, is a global public health concern. 72 The incidence of dengue has increased thirty-fold over the last five decades, and it is 73 estimated that approximately one hundred million new infections occur annually (7). In 74 the Americas, the disease is present in almost all countries with great prevalence (8), 75 where rapid urban expansion led to ideal environmental conditions for dengue to spread 76 (9). Nonetheless, dengue is still considered a neglected tropical disease (10). The 77 geographic distribution of dengue vectors and the probability of virus transmission to 78 human hosts are likewise driven by the ecological role of temperature (9,11). Despite its 79 pervasiveness, the effect of ecological variables like temperature and precipitation 80 might be outweighed by the influence of socio-economic aspects on dengue 81 . CC-BY-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 4, 2020. ; https://doi.org/10. 1101 /2020 reproduction and development (16, 20) . In this sense, some infectious diseases prevail 106 in countries where the characteristics of both demography and infrastructure create 107 favoring conditions for transmission outcomes (12). 108

Distinct approaches have been used to address the presence and prevalence of 109 infectious diseases, such as mechanistic (i.e., process-based) and statistical models 110 a multi-model climate-driven approach has also been proposed to forecast Aedes-borne 118 diseases and support surveillance operations (25). Albeit integrating many transmission-119 related factors might turn intractable in a process-based procedure, the absence of key 120 drivers still brings uncertainty to the estimated transmission potential (24). 121

Although dengue is present in almost all tropical and subtropical countries (7), 122

Brazil has experienced a higher-than-expected number of infection cases in the last 123 century (8,10). Since the 80's, the reintroduction of dengue in the country has led to its 124 rampant geographic expansion (26). Initially the presence of dengue virus was more 125 intense in large urban centers, but since 90's it has spread to small towns and 126 countryside. In the 2000s, the dengue vectors (i.e., Aedes aegypti and Aedes albopictus) 127

were already present in 72% of Brazilian municipalities, dramatically increasing the 128 disease cases and overloading the Brazilian health system (27). Although pervasive 129 . CC-BY-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 4, 2020. ; https://doi.org/10. 1101 /2020 across Brazil, the prevalence of dengue cases varies widely among the urban centers 130 making it difficult for dengue season preparedness especially in conjunction with other 131 infectious disease outbreak (27, 28) . 132

In this paper we evaluate the relative impact of socioeconomic conditions and 133 temperature suitability over the spatial pattern of dengue fever prevalence across Brazil. 134

We used a previously estimated temperature suitability for dengue transmission (23) 135 and 7 socio-economic variables to study the prevalence of dengue disease in 5570 136 municipalities across Brazil. We also seek to understand the fit between the estimated 137 temperature suitability for transmission and the effective prevalence of dengue in the 138 last years. We predict that in a highly heterogeneous country, such as Brazil, 139 socioeconomic factors are the greater source of high levels of dengue prevalence. In 140

Brazil, regions with the highest dengue prevalence are not those with highest estimated 141 temperature suitability for transmission, although suitability is still a good indicator of 142 the disease occurrence. 143 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 4, 2020. ; https://doi.org/10.1101/2020.11.02.20224444 doi: medRxiv preprint clinical and epidemiological evidence, and are carried out by a local health surveillance 152 team (30). 153

(ii) Temperature suitability 154

Here we used the simulated dengue transmission suitability maps by Brady et al. 155 (23) as a predictor of dengue presence and prevalence in Brazil, from which we extracted 156 the raster information regarding each municipality. Brady et al. (23) , estimated the 157 dengue transmission suitability from the temperature influence on mosquitoes' (Ae. 158 aegypti and Ae. albopictus) survivorship and extrinsic incubation period (i.e., EIP). The 159 EIP represents the period between mosquito biting an infected host and being able to 160 transmit the virus after processing the pathogen into the gut (i.e., become infectious) 161 (31). Brady's et al. (23) mechanistic model considered the dynamic between the EIP and 162 the adult vector survival -both temperature-dependent -over the basic reproductive 163 number (i.e., R0) (see (32,33)). The model outcome was then combined with a spatially 164 explicit temperature data, producing predictive maps of vectors' suitability range on the 165 persistence and competence of dengue transmission (23). 166

For each Brazilian municipality, we likewise gathered important socio-economic 168 predictors to the distribution and prevalence of mosquito-borne diseases (see Table A  169 in S1 Appendix), which were: human population density, urbanization, population size, is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 4, 2020. ; https://doi.org/10.1101/2020.11.02.20224444 doi: medRxiv preprint political-administrative extension comprises 5570 municipalities, which were all 174 included. 175

Epidemiological studies suggest that regions with high population density and 176 great urbanization favor the increase of dengue cases by facilitating vectors' mobility 177 and reproduction (20,36). Here we estimated the population density as the ratio among 178 population size and the area of each municipality. We accounted for the population size 179 as the total number of people within each municipality, following the IBGE classification. 180

To access the proportion of urbanization within municipalities, we estimated the ratio 181 between the urbanized area (maps based on satellite images (34)) and the whole 182 political-administrative extent from each municipality. Also, to account for the effect of 183 medical diagnosis, notification and local investments, we used the number of people 184 assisted by educational and health assistance in each municipality (16, 27) . Finally, to 185 represent economic development, we also considered GDP (log scale) and the presence 186 of the basic sanitation system (i.e., sewage treatment and/or rainfall water 187 management) (see Table A in S1 Appendix). 188

We employed linear correlations among the predictor variables to assess their 190 collinearity. In a stepwise procedure, we evaluated the non-independence between the 191 predictors by measuring the Variance Inflation Factor (VIF) among them and set apart 192 the most redundant variables. We started with a full model and iterated the procedure 193 until all the predictors had a VIF lower than 10 (37). According to this procedure the 194 variables population size and education are the most collinear and were therefore 195 removed from analyses. Although important to disease transmission in general (Table A  196 . CC-BY-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 4, 2020. ; https://doi.org/10. 1101 /2020 in S1 Appendix), the population size has a confounding relation with GDP in Brazil as a 197 consequence of internal migration patterns to economically developed areas (38). 198 Because our analysis is spatially explicit, we accounted for the spatial 199 autocorrelation that could potentially inflate Type-I errors of statistical inferences (39). is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 4, 2020.

Dengue cases are unevenly distributed across Brazil, both in occurrence and 221 prevalence (i.e., number of cases) (Fig 1) . Over the last years, most dengue cases showed 222 to be concentrated into the Southeast and Midwest regions of Brazil, but were also less 223 frequently present in the North and Northeast. From 2007-2014 (Fig 1a) there were 224 fewer recorded dengue cases when compared with a more recent epidemic period 225 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 4, 2020. ; https://doi.org/10.1101/2020.11.02.20224444 doi: medRxiv preprint circles corresponds to the magnitude of dengue prevalence (i.e., the mean number of cases). The intensity 233 of purple indicates the mean temperature suitability for the dengue transmission by Aedes spp. vectors 234 (23).

The graphical comparison between the distribution of real dengue cases (Fig. 1,  236 red circles) and the estimated temperature suitability for potential dengue transmission 237 ( Fig. 1, purple shades) The autoregressive model revealed that the relative importance of the socio-248 economic factors and estimated temperature suitability for dengue transmission in 249

Brazil (Table 1) is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 4, 2020. ; https://doi.org/10.1101/2020.11.02.20224444 doi: medRxiv preprint transmission also showed higher explanatory power for the distribution of dengue cases 257 in 2007-2014 (z = 14.825) than in 2015-2016 (z = 7.145 

Human population density was not a significant explanatory factor for the 261 number of recorded dengue cases in both periods, neither was its interaction with 262 estimated temperature suitability for dengue transmission. However, urbanization, a 263 proxy for the expansion of human-modified areas, was found to be an important 264 . CC-BY-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 4, 2020. ;

predictor of the dengue cases from 2007 to 2014 (Table 1) . Both urbanization itself and 265 its interaction with the temperature suitability for dengue transmission were significant 266 in those previous years, but less relevant in the more recent period of 2015-2016. The 267 coefficients of determination (R 2 ) of the SARerror models varied between periods, 268 suggesting that the same socio-economic variables and the temperature suitability for 269 dengue transmission have higher explanatory power during years of lower transmission 270 (R 2 = 0.53) than in periods of an increased outbreak (R 2 = 0.49). The SAR model had lower 271

AICs than its Ordinary Least Squares (OLS) counterpart (Table 1) . 272

Spatial autocorrelation was successfully controlled by the SAR model (Fig A in S2  273 Appendix). The residuals of the models revealed a narrow variation on the range of 274 values and there were no marked spatial patterns of residuals across Brazil (Fig 2) . In the 275 southern region the residuals are very small in many municipalities, indicating that 276 model predictions were accurate in these areas. In contrast, some municipalities in the 277 Amazon region had negative residuals (overestimated number of dengue cases), 278 whereas some municipalities in central and southeast Brazil had positive residuals 279 (underestimated number of dengue cases) (Fig 2) . 280 . CC-BY-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 4, 2020. ; https://doi.org/10.1101/2020.11.02.20224444 doi: medRxiv preprint 

The higher prevalence of dengue over Brazil compared to other countries has 285 intrigued researchers for decades, revealing that singular factors might regulate the 286 transmission of this infectious disease within particular countries (27). This constant 287 reemergence and maintenance of a high number of dengue cases in Brazil remain 288 unclear, which is justified due to the complex nature of the biological features of virus 289 (e.g., serotypes, virulence), host (e.g., immune system) and vectors (e.g., ambient 290 suitability, reproduction rates) (45). Therefore, the magnitude of the dengue incidence 291 fluctuates according to a myriad of environmental factors. Our results showed that, 292 although temperature suitability for transmission is a good indicator of dengue 293 . CC-BY-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 4, 2020. ; https://doi.org/10.1101/2020.11.02.20224444 doi: medRxiv preprint occurrence, the socio-economic characteristics are the fundamental determinants of 294 the spatial patterns in dengue prevalence in Brazil. 295 Similar to other tropical nations, Brazil is a heterogeneous country that has 296 undergone substantial urban growth in recent decades. This rapid urban expansion 297 along with favorable climatic conditions creates an ideal scenario for the establishment 298 and spread of critical infectious diseases, especially those carried by mosquitoes (22). 299

However, dengue occurrence and number of cases differ substantially among regions 300 within Brazil. Thus, predicting the potential transmission of mosquito-borne infectious 301 disease requires incorporating environmental and socio-economic heterogeneities 302 among regions, especially in countries where an endemic scenario is well established 303 (27) . Although the temperature is a physical factor known to affect the physiology of 304 mosquitoes and its capacity as disease vectors (18) is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 4, 2020. ; https://doi.org/10.1101 https://doi.org/10. /2020 temperature scenarios. Under lower spatial and temporal scales, the relationship 318 between temperature and multiple other drivers, such as urbanization, GDP, and 319 sanitation, should be more appropriate to describe patterns of disease transmission 320 (16, 18) . For instance, we showed that Brady's et al. model is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 4, 2020. ; https://doi.org/10.1101/2020.11.02.20224444 doi: medRxiv preprint sanitation system and GDP with the prevalence of dengue, indicating that their presence 342 by itself may not attest to the benefits of economic wealth in reducing the dengue 343 disease cases in Brazil. 344

The demography in urban environments is thought to be an important driver of 345 dengue prevalence (50). For instance, a temporal analysis of the dengue outbreak in 346

Singapore found that the population demography is the main driver for the dengue 347 increase in the last years (51). This encounter is usually accurate given the expected 348 mixed contact between hosts and vectors, where the pure increase in individuals density 349 is expected to increase the contact rates between hosts and vectors (32). However, after 350 controlling for other covariates, our model did not find a substantial effect of 351 demography on the prevalence of dengue cases across Brazil. Although demography 352 might be a good proxy for the number of susceptible individuals, in dengue-endemic 353 countries such as Brazil the serotypes immunization is likely to reduce this proportion 354 (52). Still, the population density may have great importance at the local scale (e.g., 355 among neighborhoods), once it increases the probability of vector contact with hosts 356 when the virus is installed (53). The greater relation of municipalities' GDP with dengue 357 prevalence could also be an indication of the interchange between higher population, 358 immunization, and different serotypes circulating, once Brazilian cities with greater 359 income grown faster by historically being attractive for migration (54). 360

There is no doubt that the burden of dengue is heavier in some regions than 361 others. In Brazil, where dengue cases greatly vary across space and time, we highlight 362 that the combined effect of climate and socio-economic factors drive the dengue 363 occurrence and prevalence. Still, due to the lack of reliable reports on dengue 364 prevalence, most predictive models overemphasize the role of temperature on dengue 365 . CC-BY-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 4, 2020. ; https://doi.org/10. 1101 /2020 transmission in large scales (22). By accounting for the effect of socio-economic drivers 366 in an extremely heterogeneous country, we showed that dengue prevalence is 367 explained not only by the temperature suitability for transmission but also by social and 368 economic factors. Highly urbanized centers with great income were found to be 369 epicenters of dengue transmission in Brazil, aligned with other infectious diseases (55). 370

Consequently, projections of dengue risk areas over actual or future climatic conditions 371 should include socioeconomic covariates to make predictions reliable for decision 372 making regarding a vector-borne tropical neglected disease such as dengue, especially 373 when considering that dengue season might come when another infectious disease 374 epidemic is already overloading the health system (e.g. SARS-CoV-2). Here we 375 emphasize the need of considering social, economic, and cultural differences between 376

Brazilian regions for effective decision making. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 4, 2020. ; https://doi.org/10. 1101 /2020 Supporting information 531 532 S1 Appendix. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 4, 2020. ; https://doi.org/10. 1101 /2020 

Disease Ecology, Biodiversity, and the 390 Latitudinal Gradient in Income. Hochberg ME, editor

Ecology Drives the Worldwide 393 Distribution of Human Diseases

The neglected geography 395 of human pathogens and diseases

Global drivers of human pathogen 397 richness and prevalence

399 Relationship between altitude and intensity of malaria transmission in the Usambara 400 Mountains, Tanzania

Temperature modulates dengue virus epidemic growth rates through its effects on 403 reproduction numbers and generation intervals

WHO. World Health Organization

Dengue -PAHO/WHO | Pan American Health Organization

Dengue reborn: widespread resurgence of a resilient vector

The 412 global distribution and burden of dengue

414 Modelling adult Aedes aegypti and Aedes albopictus survival at different temperatures in 415 laboratory and field settings

Lifestyle Limits Transmission of Dengue Virus. Emerg Infect Dis

Joint effects of climate variability and 422 socioecological factors on dengue transmission: epidemiological evidence

The ecology of climate change and infectious diseases

Review of the Factors Modulating Dengue Transmission

Global spread and persistence of dengue

Traits, Behavior, and Disease Transmission

Systematic variation in the temperature 434 dependence of physiological and ecological traits

Regulation and Stability of Host-Parasite Population 437 Interactions: I. Regulatory Processes

et 439 al. Climate, environmental and socio-economic change: weighing up the balance in 440 vector-borne disease transmission

443 The many projected futures of dengue

Global temperature constraints on Aedes aegypti and Ae. albopictus persistence and 446 competence for dengue virus transmission

Consensus and conflict 448 among ecological forecasts of Zika virus outbreaks in the United States

AeDES: a next-generation monitoring and forecasting system for 452 environmental suitability of Aedes-borne disease transmission. Sci Rep

Expansion of the dengue transmission area in Brazil: the 455 role of climate and cities

Dengue: twenty-five years 457 since reemergence in Brazil

São Paulo urban heat islands have a higher incidence of dengue than 460 other urban areas

Dengue 465 and dengue hemorrhagic fever, Brazil

The Incubation Periods of Dengue Viruses. Vasilakis N, 468 editor

Infectious Diseases of Humans: Dynamics and Control. 470 OUP Oxford; 1992

Modelling the global constraints of temperature on transmission of Plasmodium 473 falciparum and P. vivax. Parasit Vectors

A 479 global assembly of adult female mosquito mark-release-recapture data to inform the 480 control of mosquito-borne pathogens

A Caution Regarding Rules of Thumb for Variance Inflation Factors

Internal Migration and Economic Development in Brazil

Spatial autocorrelation and the selection of simultaneous 487 autoregressive models

489 Spatial Dependence: Weighting Schemes

Nagelkerke NJD. A Note on a General Definition of the Coefficient of 495 Determination

The global distribution of the arbovirus vectors Aedes aegypti and Ae. albopictus. 502 eLife

Climate change 504 could shift disease burden from malaria to arboviruses in Africa. Lancet Planet Health

Local and global effects of climate on 507 dengue transmission in Puerto Rico

Potential 509 Distribution of Dengue Fever Under Scenarios of Climate Change and Economic 510 Development

512 Population density, water supply, and the risk of dengue fever in Vietnam: cohort study 513 and spatial analysis

Social Behavior and Demography Drive Patterns of Fine-Scale Dengue Transmission in 521

Endemic Areas of Colombia. Paul R, editor. PLOS ONE

History, epidemiology and diagnostics 524 of dengue in the American and Brazilian contexts: a review

Epidemiological and clinical characteristics of the COVID-19 epidemic in Brazil. Nat 528 Hum Behav