key: cord-0320198-pnr3rtr1
authors: Petrone, M. E.; Earnest, R.; Lourenco, J.; Kraemer, M. U. G.; Paulino-Ramirez, R.; Grubaugh, N. D.; Tapia, L.
title: Asynchronicity of endemic and emerging mosquito-borne disease outbreaks in the Dominican Republic
date: 2020-06-20
journal: nan
DOI: 10.1101/2020.06.17.20133975
sha: ada2f96f8b34b6f76c684ae7f84ed00cf74f5bcc
doc_id: 320198
cord_uid: pnr3rtr1

Mosquito-borne viruses pose a perpetual public health threat to countries and territories in the Carribean due to the region's tropical climate and seasonal reception of international tourists. Outbreaks of the emerging viruses chikungunya and Zika in 2014 and 2016, respectively, demonstrated the rapidity with which these viruses can spread between islands. At the same time, the number of reported dengue fever cases, caused by the endemic dengue virus, has steadily climbed over the past decade, and a large dengue outbreak that began sweeping through this region in 2019 continues in 2020. Sustainable disease and mosquito control measures are urgently needed to quell virus transmission in the long term and prevent future outbreaks from occurring. To improve upon current surveillance methods, we analyzed temporal and spatial patterns of chikungunya, Zika, and dengue outbreaks reported in the Dominican Republic between 2012 and 2018. The viruses that caused these outbreaks are transmitted by Aedes mosquitoes, which are sensitive to seasonal climatological variability. In this study, we evaluated whether climate and the spatio-temporal dynamics of past dengue outbreaks could inform when and where future emerging disease outbreaks might occur. We found that the temporal and spatial distribution of emerging disease outbreaks did not conform to those of seasonal dengue outbreaks. Rather, the former occurred when climatological conditions were suboptimal for Aedes activity. Provincial dengue attack rates did not correspond to those of emerging diseases. Our study also provides evidence for under-reporting of dengue cases, especially following the 2016 Zika outbreak. We advocate for the implementation of a sustainable and long-term surveillance system to monitor the spread of known mosquito-borne viruses and to identify emerging threats before they cause outbreaks. Specifically, we recommend the use of febrile illness incidence, case fatality rates, and serosurveys during inter-outbreak periods to better understand rates of transmission and asymptomatic infection.

Emerging and endemic mosquito-borne viruses are a constant public health concern in the Carribean (Cao-36 Lormeau . In addition to this threat, dengue virus is endemic to 45 many Caribbean countries and territories and has caused outbreaks with increased frequency over the past 46 decade. Large outbreaks, which began in 2019 but have continued through 2020 (Pan-American Health 47

Organization (PAHO), 2020a), extend an alarming trend of a rising number of dengue cases reported 48 annually in the Americas in recent years (PAHO, 2014) . However, the danger of these viruses and the 49 diseases they cause lie not only in their debilitating and sometimes life-threatening symptoms. The 50 unpredictability of when and where a new outbreak will occur precludes preparedness. Outbreak response 51 strategies are inherently reactionary and, in their transience, disrupt public health systems when new 52

initiatives are introduced and subsequently phased out. This is especially problematic in resource-limited 53 settings where the strategic allocation of resources should be prioritized to maximize the impact of disease 54 control efforts. New approaches centered around sustainable, long-term surveillance are needed to curtail 55 the potential for future public health crises in the Caribbean. 56

One such approach is the use of climate data to evaluate the risk of viral spread. The transmission of 57 mosquito-borne viruses, both emerging and endemic, should adhere to similar temporal and spatial patterns 58

such that the dynamics of past outbreaks can inform those of future outbreaks. Many flaviviruses, like 59 dengue and Zika viruses, and alphaviruses, like chikungunya virus, are transmitted by Aedes mosquitoes, 60

which are sensitive to climatological variability (Kraemer et 2016). Therefore, we considered whether, while climate may be a useful indicator for future endemic virus 68 outbreaks, other factors including population size, demographics, and the timing of introduction should be 69 considered when developing strategies to prevent future emerging disease outbreaks. 70

To answer these questions in the context of the Caribbean, we analyzed dengue, chikungunya, and Zika 71 cases reported daily in the Dominican Republic between 2012 and 2018. We found that emerging disease 72 outbreaks (chikungunya and Zika) occurred earlier in the year than dengue outbreaks, and the timing and 73 location of introductions of emerging viruses impacted when and where corresponding outbreaks occurred. 74

Moreover, the spread of chikungunya and Zika viruses was tolerant to sub-optimal climates for transmission 75 by Aedes mosquitoes, likely due to the large size of the susceptible human population. Predicted mosquito 76 abundance was similarly uninformative for the spatial distribution of emerging disease attack rates and force 77 of infection. Moreover, provincial-level dengue attack rates were consistent between dengue outbreaks, but 78 they did not correspond to local attack rates of chikungunya and Zika. Taken together, we demonstrate that 79 dengue virus may not always be an appropriate model to prepare for future emerging mosquito-borne 80 disease outbreaks. Instead, a sustainable and long-term mosquito-borne disease surveillance system is 81 needed to facilitate proactive responses to emerging threats and to track the continued spread of known 82 diseases including dengue, chikungunya, and Zika. We specifically propose the use of two indicators, 83

incidence of febrile illness cases and dengue case fatality rates, to monitor surveillance performance and 84 identify potential emerging threats. 85 delineated this period into six seasons (Seasons 1-6), each beginning in April of every year between 2012 90 -2018, coincident with the start of the rainy season. Three of the five outbreaks were caused by dengue 91 virus (Seasons 1, 2, and 4; Fig. 1b Because the implementation of national-level reporting of chikungunya and Zika cases influenced when 130 these outbreaks were detected (Fig 1c, arrows) , we calculated the number of weeks elapsed between the 131 first reported case of each outbreak and the peak number of cases within provinces (Fig. 1d) . We assumed 132 that climatological factors did not vary widely between provinces during a given season. we did not expect transmission rates to vary widely between provinces, provincial outbreaks that peaked 137 soon after the reporting system was implemented would indicate that transmission in those provinces 138

preceded the establishment of this system. 139

Our analysis identified one such instance. The majority of provincial chikungunya and Zika outbreaks peaked 140 within 12 and 25 weeks, and the mean time to peak across provinces did not significantly differ (Fig. 1d,e) .

This was in contrast to the substantial heterogeneity of the timing of the three dengue outbreaks. The lack 142

of uniformity across these outbreaks may have been due to a slower rate of spread between regions 143 because of a pre-existing and spatially heterogeneous herd immunity landscape. Five provinces, four of 144 which are located in the south-western part of the country, reported peak numbers of Zika cases in January 145

and February of 2016 (Fig. 1d) . For the reasons stated above and because Zika virus had been circulating 146 in the Americas for at least two years before it was reported in the Dominican Republic (Faria et al., 2016, 147 2017), this observation suggested that Zika virus was already circulating in those provinces before the 148 national reporting system was implemented at the beginning of January. We hypothesized that early Zika 149 virus transmission could have occurred in the south-western region because of the region's proximity to 150

Haiti. In particular, this region shares a border with Haiti, is connected to the Haitian capital Port-au-Prince 151 via a major roadway, and is home to a large binational market (Fig. 1e , map, star). A temporal comparison 152 of the Zika outbreaks reported in this region, the rest of the country, and Haiti revealed that the regional 153 outbreak was more consistent with the 2016 Zika outbreak reported in Haiti (Fig. 1e, inset) . These findings 154

indicate that the Dominican Republic experienced two geographically and temporally distinct Zika outbreaks. 155

From reported case counts alone we cannot conclude whether Zika virus was introduced into the Dominican 156

Republic from Haiti or vice versa; however, it is clear that binational coordination is an essential component 157 of local mosquito-borne disease control because the viruses these vectors transmit do not recognize 158 international borders. 159 160 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted June 20, 2020. . https://doi.org/10.1101/2020.06.17.20133975 doi: medRxiv preprint 

Initial outbreaks of emerging mosquito-borne diseases can occur during periods of sub-optimal 175 climatological conditions 176

We considered two possible explanations for the asynchronicity we observed between outbreaks of the 177 emerging and endemic viruses ( Fig. 1) .

(1) Either climate patterns differed between seasons and the 178 emerging outbreaks preceded those of dengue due to seasonal stochasticity, or (2) the spread of emerging 179 viruses was less sensitive to climatological factors compared to dengue virus. By analyzing daily 180 climatological data collected over the duration of our study period, we found support for the latter (Fig. 2) . 181

To investigate the relationship between climate and case incidence, we used temperature and humidity time 182 series data to estimate the mosquito-borne transmission potential throughout our study period (Fig. 2) . For 183 this analysis, we used Index P, a metric that is calculated by incorporating climate and entomological data 184 into a Bayesian framework to estimate the transmission potential of individual female mosquitoes (Obolski 185 et al., 2019). We calculated transmission potential for each week of our study period using data reported in 186 six of thirty-two provinces. We selected these six provinces based on their representativeness of the 187 country's sub-climates and data availability. On average, transmission potential fluctuated seasonally, rising 188

between April and November of each year, coincident with the rainy season, and fell shortly thereafter ( Fig.  189 2). When we compared the temporal dynamics of transmission potential to reported disease incidence, we 190

found that the number of emerging disease cases reported weekly peaked before transmission potential 191 had reached a seasonal maximum for both outbreaks (chikungunya and Zika), while the number of dengue 192 cases reported weekly peaked after this point for all three dengue outbreaks (Fig. 2a) . Similar climatological 193 patterns persisted on the provincial level ( Fig. 2b) . Some provinces experienced clear seasonal fluctuations 194 in transmission potential, whereas seasonal peaks and troughs were less defined in Distrito Nacional and 195

María Trinidad Sánchez, which have low to moderately humid sub-climates. Despite this variation, the timing 196 of provincial outbreaks conformed to the trend observed on the national level: dengue outbreaks peaked 197 after a period characterized by high transmission potential, just as transmission potential was beginning to 198 wane; in contrast emerging disease outbreaks occurred concurrently with increasing transmission potential. 199

Moreover, we cannot attribute climatic anomalies to the decline in reported dengue cases in Seasons 3, 5, 200 and 6. 201 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 20, 2020. Variability in seasonal weather patterns and vectorial capacity did not account for differences in the timing 214 of emerging disease outbreaks (Fig. 2) . Rather, we hypothesized that the patterns observed for dengue 215 were likely to have been influenced by a pre-existing and spatially heterogeneous herd immunity landscape. 216

This would mandate that transmission potential must remain high for an extended period before epidemic 217 growth is achieved each season. 218

To test the hypothesis that local susceptibility influenced the timing of the outbreaks, we examined the 219 relationship between the size of the susceptible human population and each outbreak's doubling time (Fig.  220 3). We estimated that the median estimated effective reproduction number (Reff) for the three dengue 221 outbreaks was between 1.3 and 1.4, while those of the chikungunya and Zika outbreaks ranged from 1.6 222 to 2.45 (Fig. 3a) . the third dengue outbreak, consistent with increasing levels of herd immunity from the two previous 228

outbreaks. In our dataset, reported dengue cases were primarily in the 0-15 age group, indicating older age 229 groups had high levels of pre-existing immunity and were not susceptible to disease (Table 1; Fig. 3b ). In 230 contrast, we speculated that the entire population was susceptible to chikungunya and Zika, allowing these 231 viruses to spread despite sub-optimal weather conditions, facilitating an earlier outbreak peak, and affecting 232 a much wider range of ages (Table 1; Fig. 3b ). 233 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 20, 2020. . Given that the timing of emerging disease outbreaks (chikungunya and Zika) did not conform to that of 242 dengue outbreaks (Fig. 1,2) , we suspected that the spatial distribution of dengue cases would be equally 243 uninformative for chikungunya and Zika outbreaks. Specifically, we hypothesized that the relative burden of 244 dengue within provinces during individual outbreaks would correlate with the burden of dengue during 245 subsequent outbreaks but not with the relative burden of emerging disease cases. 246

To address this question, we measured the age-and sex-adjusted attack rates by province for each of the 247 outbreaks. We found that the attack rates for individual provinces across outbreaks were well correlated 248 between dengue outbreaks and between chikungunya and Zika outbreaks, but not across the endemic and 249 emerging viruses (Fig. 4a, b) . 250

Next, we investigated the role of climate and land-use in perpetuating this trend, reasoning that larger 251 mosquito populations would facilitate higher attack rates (Kraemer, et al., 2015 ; Rodriguez-Barraquer, Salje, 252 et al., 2019). To this end, we compared an Aedes aegypti suitability score (AaS; Fig. 4c ), a metric that 253

incorporates ecological variables not included in our transmission potential estimates such as vegetation 254 levels and precipitation, to population-adjusted attack rates of disease (Fig. S1, Fig. 4a) . Also unlike our 255 transmission potential estimates, AaS is an aggregated estimate of suitability per month in any given year. 256

Interestingly, AaS was not well correlated with the adjusted provincial attack rates or with mean age of 257 infection (Fig. S1) . Moreover, provinces in which the highest levels of mosquito abundance were predicted 258 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 20, 2020. . https://doi.org/10.1101/2020.06.17.20133975 doi: medRxiv preprint did not consistently report the highest burden of disease (Fig. 4a,c) . One reason for these inconsistencies 259

is that AaS relies on historical climate data from 1970 to 2000 to estimate mosquito abundance. Therefore, 260 our estimates do not capture anomalous climatological events such as the hurricane that made landfall in 261 the Dominican Republic in October 2012. AaS also assumes a high level of stability in land use and minimal 262 urbanization since the thirty year period from which the estimates were made. Regardless of the true 263 underlying reason for these discrepancies, the spatial distributions between endemic (dengue) and 264

emerging ( 

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 20, 2020. to the temporal and spatial patterns of endemic disease outbreaks in the Dominican Republic (Figs. 1-4) .

While similar inconsistencies in outbreak dynamics have been observed elsewhere in the Americas (Fig.  278 S2) (Lourenço et al., 2017; Faria et al., 2016a) , it is difficult to discern if our findings are due to differences 279 in epidemiology or in underlying reporting biases. Therefore, we used general patterns of febrile illness and 280 clinical characteristics to show that reporting biases did not influence our previous conclusions (Fig. 5) . 281

Because the vast majority of cases in our dataset reported fever independent of disease (Table 1) , we 282

hypothesized that febrile illness incidence should reflect disease incidence reported during our study period. 283

When we examined the number of febrile illness cases reported in the Dominican Republic per 284 epidemiological week per season, we found that temporal trends in febrile illness cases in Season 3 (2014-285 2015) and Season 5 (2016-2017; Fig. 5a ) were consistent with those we observed in our chikungunya and 286

Zika case data, respectively (Fig. 1a) . In Season 4 (2015-2016, dengue), febrile illness cases peaked in 287

week 44, coincident with the peak of the dengue outbreak (Fig. 1a) . To ensure that the spatial relationships 288

we identified in Figure 4 were not purely the product of differential reporting practices, we compared 289

population-adjusted attack rates of mosquito-borne disease and febrile illness reported by individual 290 provinces during each season and found no significant correlation (Fig. S3) . Equally pronounced seasonal 291 peaks of reported febrile illness were not observed in Season 1 and Season 2 despite the known dengue 292 outbreaks that occurred within that period. We speculate that changes made in reporting protocols for febrile 293 illness cases during the chikungunya outbreak in Season 3 prompted a sustained increase in national 294

reporting thereafter (Ministerio de Salud Pública, 2014a). 295

We then investigated why reported febrile illness cases peaked seasonally in Seasons 5 and 6 despite an 296 apparent decline in coincident dengue cases by comparing case-fatality rates (CFRs) of dengue cases 297 across seasons. Specifically, we considered the possibility that this decline was due to under-reporting of 298 dengue cases during the interim periods between dengue outbreaks (Seasons 3, 5, and 6). The average 299 CFR during these seasons was 1.87% (range: 1.60%-2.40%), while the average CFR of reported cases 300 during the three large dengue outbreaks in Seasons 1, 2, and 4 was 0.8% (range: 0.48%-1.20%) (Fig. 5b) .

A wide range of dengue CFRs is reported in the literature, with a mean of 1.62% (95% CI: 0.64-4.02%) 302 (Low et al., 2018) , and without serosurveillance data, we are unable to rule out the possibility that the 303 circulation of different dengue serotypes resulted in differences in seasonal CFRs. However, if we assume 304 that fatality among cases during the third dengue outbreak (Season 4) were well reported, dengue cases 305 were under-reported by 33% in Season 3 and Season 5, and by 2-fold in Season 6. These are likely 306

conservative estimates as the current surveillance system is passive and therefore does not capture 307 asymptomatic cases. As a result, Seasons 3, 5, and 6 may represent periods characterized by a large 308 number of mild cases that were not captured by the surveillance system. It is also likely that the number of 309 deaths due to chikungunya was substantially under-reported; however, our calculation of 0.19% 310 corresponds to a corrected, post-hoc estimate of this rate (0.15%) for the Dominican Republic (Freitas et  311 al., 2018). 312 313 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 20, 2020. . https://doi.org/10.1101/2020.06.17.20133975 doi: medRxiv preprint Maintaining a sustainable surveillance system is critical for preventing the silent transmission of viruses that 334

can fuel large outbreaks. Other countries in the Americas reported subsequent outbreaks of chikungunya 335

and Zika after their initial outbreaks. We cannot conclusively determine whether the Dominican Republic 336 experienced a similar pattern because surveillance data for chikungunya and Zika are not available for 337 seasons following the initial outbreaks of these diseases. Elucidating whether dengue, chikungunya, and 338

Zika are co-circulating in the country will be critical for triaging and providing appropriate clinical care to 339 patients who present with febrile illness ( . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 20, 2020. . understanding the transmission patterns of viruses and developing a unified, international plan to combat 352 them before they cause an outbreak will help mitigate the potential for such an event. There are a few important limitations to our study. First, our dataset included chikungunya and Zika case 374 data from the initial wave of each disease, and we cannot therefore compare temporal and spatial dynamics 375 of these diseases across seasons. After these initial outbreaks, diagnostic testing for these diseases has 376 largely ceased. While the number of cases of these diseases reported in the Dominican Republic has 377 declined to zero the true burden of disease is unknown. Future studies should investigate whether these 378

viruses have continued to circulate undetected in the country and whether their spatiotemporal dynamics 379

have since synchronized with that of dengue virus. Second, the reporting system for suspected chikungunya 380 cases differed from that used for suspected Zika cases. During the chikungunya outbreak, most febrile 381 illnesses cases without apparent cause were initially classified as suspected chikungunya cases. For this 382 reason, the number of cases reported by the Pan-American Health Organization (PAHO) and the Ministry 383 of Health was significantly larger than those which we have reported here (Ministerio de Salud Pública, 384 2014b). Our chikungunya case data contains a disproportionate number of children in the <1 year age group, 385

indicating that the dissemination of diagnostic testing may have been skewed towards high-risk groups. 386

Third, our findings demonstrate that an epidemiological relationship existed between the Dominican 387

Republic and Haiti during the Zika epidemic in 2016, but we cannot determine the directionality of cross-388 border virus movement without virus genomic data. However, given that this relationship exists and that 389 mosquitoes do not recognize political boundaries, it can be assumed that bi-directional spillover of mosquito-390 borne diseases will occur in the future unless appropriate bi-national surveillance and control measures are 391

implemented. Finally, our analysis primarily focused on virus transmission by Aedes aegypti mosquitoes, 392

but it is possible that other mosquito vectors contributed to the propagation of the outbreaks we investigated. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 20, 2020. Time to outbreak peak 442 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 20, 2020. . https://doi.org/10.1101/2020.06.17.20133975 doi: medRxiv preprint We calculated the time to peak for provincial outbreaks by first identifying the epidemiological week in which 443 the first case of the national outbreak was reported (chikungunya: EW 6, Zika: EW 1, dengue (3): EW 14) 444

and then counting the number of weeks elapsed until each of the 32 provinces reported the maximum 445 number of cases for the corresponding outbreak. The mean time to peak for each outbreak was compared 446

using an unpaired T test implemented in Prism v8.4.

Transmission potential 448 We calculated weekly transmission potential (Index P) with the Bayesian approach developed by Obolski et 449 al. (Obolski et al., 2019) . Briefly, we extracted daily average temperature and relative humidity for 6 cities 450 as described above. Rarely, temperature or humidity were not available for a given day. In these cases, we 451 averaged the respective variable from the same date across the remaining 6 years. We did this for 14 days 452

for Barahona and Puerto Plata, 3 days for Santiago, and 1 day for María Trinidad Sánchez. 453

We then used the R package MVSE and entomological and epidemiological priors documented in the 454 literature (Table S1 ) to calculate daily transmission potential, which we then aggregated by week (Lourenço, 455 2019). We found that the model was reasonably robust to a range of priors for each of the parameters and 456 therefore elected to use short human incubation and infectious period estimates to inform the model. We 457 We used a range of values for latent and infectious periods that were well reported in the literature to 468 calculate f, the proportion of the serial interval in the latent period, and v, the serial interval (Table S2) . We 469 obtained r, the epidemic growth rate, by extracting the slope of our model after fitting it to our case data. All 470 calculations were done using R v4.0.0. 471 472

Adjusted attack rates and linear regression 473

Province-level attack rates by outbreak were age-and sex-adjusted using the direct standardization method, 474

with the national population as the reference population. 475 476 We calculated Pearson's R correlation coefficient for province-level attack rates between various pairs of 477 outbreaks (Fig. 4a,b) . An outlier analysis showed that the removal of outliers did not substantially affect the 478 size or significance of the correlation coefficients, so we included all data points. 479 480

Aedes aegypti suitability score (AaS) 481

AaS was calculated for each month using both climate and land-use data on a 5km x 5 km scale collected 482 from 1970 and 2000 and compiled by WorldClim as described by Kraemer et al. (Kraemer, et al., 2015) . 483

Monthly suitability scores were extracted and averaged by province in Rv4.0.0. 484

ACKNOWLEDGEMENTS 486 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 20, 2020. . We thank C. Vogels, A. Brito, J. Fauver, C. Kalinich, I. Ott, S. Lapidus, K. Gangavarapu, J. Pack, and S. 487

Taylor for feedback and/or assistance with the methodology. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted June 20, 2020. . https://doi.org/10.1101/2020.06.17.20133975 doi: medRxiv preprint

Epidemiology of Chikungunya Virus in Bahia, Brazil

Zika virus in the Americas: Early 547 epidemiological and genetic findings

Establishment and cryptic transmission of Zika virus in Brazil and the Americas

Xenosurveillance to Detect Human Bacteria, Parasites, and Viruses in Mosquito Bloodmeals

Xenosurveillance reflects traditional sampling techniques for the identification of 557 human pathogens: A comparative study in West Africa

EPIDEMIOLOGY. Countering the Zika epidemic in Latin America

Excess mortality profile during the Asian 563 genotype chikungunya epidemic in the Dominican Republic

Space-time dynamics of a triple epidemic: dengue, chikungunya 566 and Zika clusters in the city of Rio de Janeiro

Mayaro virus in Latin America and the Caribbean

Spontaneous Abortion Associated with Zika Virus Infection 572 and Persistent Viremia

A survey 574 of tire-breeding mosquitoes (Diptera: Culicidae)

Micro-environmental features associated to container-dwelling mosquitoes (Diptera: Culicidae) in an urban cemetery 578 of the Dominican Republic

Prior dengue virus infection and risk of Zika: A pediatric 581 cohort in Nicaragua

Genomic Insights into Zika Virus Emergence and Spread

Tracking virus 585 outbreaks in the twenty-first century

Travel Surveillance and Genomics Uncover a Hidden Zika Outbreak during the Waning 589

Xenosurveillance: a novel mosquito-based approach for examining the human-pathogen 592 landscape

Reappearance of chikungunya, formerly called dengue

Dengue, West Nile virus, chikungunya, Zika-and now Mayaro?

Adult survivorship of the dengue mosquito Aedes aegypti varies seasonally in central Vietnam. PLoS 604 Neglected Tropical Diseases

Estimating dengue transmission intensity from sero-prevalence 606 surveys in multiple countries

Infection Pattern of Mayaro Virus in Aedes aegypti 609 (Diptera: Culicidae) and Transmission Potential of the Virus in Mixed Infections With Chikungunya Virus

Neutralizing antibody titers against dengue virus 612 correlate with protection from symptomatic infection in a longitudinal cohort

West Nile Virus Survey of Birds and Mosquitoes in 616 the Dominican Republic

Past and future spread of the arbovirus vectors Aedes aegypti and Aedes albopictus

The global distribution of the arbovirus vectors Aedes aegypti and Ae. 625 albopictus. eLife, 4, e08347

The global compendium of Aedes aegypti and Ae. albopictus occurrence. Scientific Data, 2, 150035

Chikungunya 631 virus adapts to tiger mosquito via evolutionary convergence: a sign of things to come?

Molecular Virologic and Clinical 635 Characteristics of a Chikungunya Fever Outbreak in La Romana

Mayaro Virus in Child with Acute Febrile Illness

Times to key events in Zika virus infection and implications for blood donation: a systematic review. 642 Bulletin of the World Health Organization

Oral susceptibility of Singapore Aedes (Stegomyia) aegypti (Linnaeus) to

Zika virus

Chikungunya Virus Disease among Travelers-United States

Transmission dynamics and control of severe acute respiratory syndrome

Vectorial capacity of Aedes aegypti: effects of 651 temperature and implications for global dengue epidemic potential

Reviewing estimates of the basic 654 reproduction number for dengue, Zika and chikungunya across global climate zones

Experimental 657 transmission of Mayaro virus by Aedes aegypti

Mayaro virus distribution in South America

Epidemiological and ecological determinants of Zika virus 663 transmission in an urban setting

Estimator of climate-driven, mosquito-borne viral suitability index

Global dengue death before and after the new World

Health Organization 2009 case classification: A systematic review and meta-regression analysis

Estimating a feasible serial interval range for Zika fever

Going Viral 2019: Zika, Chikungunya, and Dengue. Dermatologic 673

Islands as Hotspots for Emerging Mosquito-Borne 675 Viruses: A One-Health Perspective

Zika virus evolution and spread in the Americas

Chikungunya and Dengue Virus Infections Among United States Community Service Volunteers 683 Returning from the Dominican Republic

Plan de preparación y respuesta frente a brotes de Fiebre Chikungunya

República Dominicana

Detecting the impact of 692 temperature on transmission of Zika, dengue, and chikungunya using mechanistic models

Unexpected outbreaks of arbovirus 695 infections: lessons learned from the Pacific and tropical America. The Lancet Infectious Diseases

Transmission potential of Zika virus infection in the South 699

IJID: Official Publication of the International Society for Infectious

MVSE : An R-package that 702 estimates a climate-driven mosquito-borne viral suitability index

Five-fold increase in dengue cases in the Americas over the past decade

Suspected dengue cases by epidemiological week for countries and 708 territories of the Americas

Dengue serotypes by year for countries and territories of the Americas

Pan-American Health Organization (PAHO) (2020c)

Pan-American Health Organization (PAHO) (2020d). PLISA Health Information Platform for the Americas

Vector competence of Aedes aegypti

Aedes albopictus, and Culex quinquefasciatus mosquitoes for Mayaro virus

The decline of dengue in the Americas in 2017: discussion of multiple hypotheses

Measuring Mosquito-borne Viral Suitability in Myanmar and Implications for 728

Local Zika Virus Transmission

Chikungunya in the region of the Americas. A challenge 731 for rheumatologists and health care systems

Cases of chikungunya virus infection 735 in travellers returning to Spain from Haiti or Dominican Republic

Influence of herd immunity in the cyclical 739 nature of arboviruses. Current Opinion in Virology

Does immunity after Zika virus infection cross-protect against 742 dengue? The Lancet

A comparative analysis of Chikungunya and Zika transmission

Impact of preexisting dengue immunity on Zika virus 748 emergence in a dengue endemic region

Opportunities for improved surveillance and control of dengue from 750 age-specific case data

Chikungunya 752 infection in the general population and in patients with rheumatoid arthritis on biological therapy

Chikungunya fever: an epidemiological review of a re-emerging infectious 755 disease

Dispersal and other population parameters of Aedes aegypti in an African village and their 758 possible significance in epidemiology of vector-borne diseases

Estimates of Population Size, Dispersal, and Longevity of Domestic Aedes 761 aegypti aegypti (Diptera: Culicidae) by Mark-Release-Recapture in the Village of Shauri Moyo in Eastern Kenya

Effects of infection history on dengue virus infection and 765 pathogenicity

Dominican Republic: Human Development Indicators

Two Chikungunya isolates from the outbreak of La Reunion

Ocean) exhibit different patterns of infection in the mosquito, Aedes albopictus

Arbovirus coinfection and co-773 transmission: A neglected public health concern?

Present and future arboviral threats

A Tale 778 of Two Flaviviruses: A Seroepidemiological Study of Dengue Virus and West Nile Virus Transmission in the Ouest

Transmission potential of Mayaro virus in Florida Aedes aegypti and Aedes 782 albopictus mosquitoes

Impact of temperature on the extrinsic incubation period of Zika 784 virus in Aedes aegypti

Aedes (Stegomyia) albopictus (Skuse): a potential vector 786 of Zika virus in Singapore

World Health Organization (WHO) Zika virus: Fact Sheet

A study of biting habits of Aedes aegypti in

Clinical manifestations of chikungunya among university professors and 792 staff in Santo Domingo, the Dominican Republic