Behavioral, climatic, and environmental risk factors for Zika and Chikungunya virus infections in Rio de Janeiro, Brazil, 2015-16
Behavioral, climatic, and environmental risk factors for Zika and Chikungunya virus infections in Rio de Janeiro, Brazil, 2015-16
Trevon L. Fuller 0 1 2
Guilherme Calvet 0 2
Camila Genaro Estevam 0 2
Jussara Rafael Angelo 0 2 3
Gbenga J. Abiodun 0 2
Umme-Aiman Halai 0 2
Bianca De Santis 0 2
Patricia Carvalho Sequeira 0 2
Eliane Machado Araujo 0 2
Simone Alves Sampaio 0 2
Marco Cesar Lima de MendoncË a 0 2
Allison Fabri 0 2
Rita Maria Ribeiro 0 2
Ryan Harrigan 0 1 2
Thomas B. Smith 0 1 2
Claudia Raja Gabaglia 0 2
PatrÂõcia Brasil 0 2
Ana Maria Bispo de Filippis 0 2
Karin Nielsen-Saines 0 2
0 Current address: County of Los Angeles Department of Public Health , Los Angeles, California , United States of America
1 Institute of the Environment and Sustainability, University of California Los Angeles , Los Angeles , California, United States of America, 2 Instituto Nacional de Infectologia Evandro Chagas , FundacËão Oswaldo Cruz, Rio de Janeiro , Brazil , 3 Universidade Estadual de São Paulo , Rio Claro, São Paulo , Brazil
2 Editor: Kevin K. ArieÈn , Instituut voor Tropische Geneeskunde , BELGIUM
3 Escola Nacional de SauÂ de PuÂ blica, FundacËão Oswaldo Cruz , Rio de Janeiro , Brazil , 5 Foundation for Professional Development , Pretoria, Gauteng , South Africa , 6 David Geffen UCLA School of Medicine, Los Angeles, California, United States of America, 7 Laboratorio de Referência de Flavivirus, Instituto Oswaldo Cruz , FundacËão Oswaldo Cruz, Rio de Janeiro , Brazil , 8 Department of Ecology and Evolutionary Biology, University of California Los Angeles , Los Angeles , California, United States of America, 9 Biomedical Research Institute of Southern California , Oceanside, California , United States of America
The burden of arboviruses in the Americas is high and may result in long-term sequelae with infants disabled by Zika virus infection (ZIKV) and arthritis caused by infection with Chikungunya virus (CHIKV). We aimed to identify environmental drivers of arbovirus epidemics to predict where the next epidemics will occur and prioritize municipalities for vector control and eventual vaccination. We screened sera and urine samples (n = 10,459) from residents of 48 municipalities in the state of Rio de Janeiro for CHIKV, dengue virus (DENV), and ZIKV by molecular PCR diagnostics. Further, we assessed the spatial pattern of arbovirus incidence at the municipal and neighborhood scales and the timing of epidemics and major rainfall events. Lab-confirmed cases included 1,717 infections with ZIKV (43.8%) and 2,170 with CHIKV (55.4%) and only 29 (<1%) with DENV. ZIKV incidence was greater in neighborhoods with little access to municipal water infrastructure (r = -0.47, p = 1.2x10-8). CHIKV incidence was weakly correlated with urbanization (r = 0.2, p = 0.02). Rains began in October 2015 and were followed one month later by the largest wave of ZIKV epidemic. ZIKV cases markedly declined in February 2016, which coincided with the start of a CHIKV out-
Data Availability Statement: All relevant data are
within the paper and its Supporting Information
Funding: This work was partially supported by: the
European Union's Horizon 2020 research and
innovation program (Zika action grant agreement
no. 734857 and Zika plan grant agreement no.
734584), FundacËão de Amparo à Pesquisa do
Estado do Rio de Janeiro grant no. E-18/2015TXB,
National Council for Scientific and Technological
break. Rainfall predicted ZIKV and CHIKV with a lead time of 3 weeks each time. The
association between rainfall and epidemics reflects vector ecology as the larval stages of Aedes
aegypti require pools of water to develop. The temporal dynamics of ZIKV and CHIKV may
Development of Brazil, Oswaldo Cruz Foundation/
General Coordination of Public Health Laboratories/
Ministry of Health, National Science Foundation
grant number 1243524, a University of California
Global Health Institute grant, National Institutes of
Health grants AI28697, 1R21AI129534-01, and
D43TW009343, and by the Pan American Health
Organization Small Grants Programme for
Research on the Zika Virus Outbreak in the
be explained by the shorter incubation period of the viruses in the mosquito vector; 2 days
for CHIKV versus 10 days for ZIKV.
The burden of arboviral disease in the Americas is high and increasing. It includes infants
permanently disabled by infection with the Zika virus (ZIKV) in Brazil [1±3] and persistent,
incapacitating arthritis caused by infection with Chikungunya virus (CHIKV) [
]. The past decade
has seen a substantial increase in the burden of arboviruses driven by factors such as the
proliferation of mosquito breeding sites in cities and range expansions of ZIKV and CHIKV from
Africa and Asia to Oceania and the Americas [5±9]. It is predicted that with climate the ranges
of Aedes mosquitoes that are vectors of CHIKV, DENV, and ZIKV will expand in South
Although a variety of studies have investigated the introduction of a single arbovirus into a
naïve population, an integrative analysis of more than one arbovirus has the potential to yield
insights about interactions among the viruses. A salient example is the state of Rio de Janeiro,
Brazil, which has experienced recent epidemics of four arboviruses transmitted by the
mosquito Aedes aegypti: CHIKV, dengue (DENV), Yellow Fever virus, and ZIKV. ZIKV is of
particular concern because in pregnant women infection can result in fetal abnormalities
including microcephaly [
]. In non-pregnant adults approximately 1% of ZIKV infections
result in Guillain-BarreÂ Syndrome (GBS), an inflammatory disorder that causes acute flaccid
]. Like ZIKV, CHIKV is a novel arbovirus in Rio de Janeiro with the first
locally transmitted cases reported in 2015 [
As a variety of ecological and economic factors could contribute to arbovirus epidemics in
Rio de Janeiro including sanitation infrastructure and pools of standing water that become
mosquito breeding sites, an in-depth study is needed to assess the importance of these factors.
To understand arboviruses in an integrative fashion, the objectives of the present study were
to 1) identify drivers of CHIKV and ZIKV epidemics. Although we included DENV in the
analysis because it is endemic in Rio de Janeiro [
], the focus of our study was CHIKV and
ZIKV as they are novel arboviruses in the region and caused epidemics during the study
period; 2) predict where the next epidemics may occur; and 3) prioritize areas for vector
control and eventual vaccination. We screened suspected cases by polymerase chain reaction
(PCR), and clinical criteria and analyzed the timing, geographic locations, and socio-economic
and infrastructure-related characteristics of confirmed cases.
Materials and methods
We used an ecological study design to collect serum and urine samples from symptomatic
patients suspected to have CHIKV, DENV, or ZIKV in the state of Rio de Janeiro (n = 10,459).
Samples were collected at outpatient clinics and hospitals in 48 municipalities across the state of
Rio de Janeiro from January 2015 to October 2016 (S1 Table). None of the patients had a history
of travel outside the state. Patients were offered the opportunity to participate in the study if they
were suspected to have an arbovirus based on clinical signs and symptoms such as acute febrile
illness with mosquito exposure. After providing written informed consent, participants filled out
an information sheet with their age, sex, and the municipality and neighborhood where they
lived. Human subject research was approved by the Evandro Chagas National Institute of
Infectious Diseases, Oswaldo Cruz Foundation (Ethics Approval CAAE0026.0.009.000±07). Sera and
2 / 15
urine were tested for ZIKV by real-time RT±PCR with the QuantiTect Probe kit [
] and for
DENV using the CDC RT-PCR Assay . Sera and urine were tested for CHIKV by qRT-PCR
] or by enzyme-linked immunosorbent assay (ELISA) to detect IgM antibodies using the
Euroimmun kit [
]. A confirmed case was anyone positive for ZIKV by qRT-PCR, positive
for DENV by qRT-PCR, or positive for CHIKV either by qRT-PCR or by ELISA.
If both sera and urine samples from the same individual were positive by lab assays, to
avoid duplicating cases, only one positive sample was retained for climatic and spatial
modeling. Furthermore, if a patient was positive for CHIKV by RT-PCR and IgM, we only included
the date of the RT-PCR positive sample in the database, because the date of the PCR positive is
closer to the acute phase of the patient's illness. Finally, if a patient tested negative for CHIKV
by PCR but was positive by IgM, the IgM sample was included in the database of positive
Screening was conducted at the Brazilian Ministry of Health Regional Reference Laboratory
for CHIKV, DENV, ZIKV, Yellow Fever, and West Nile virusÐInstituto Oswaldo Cruz
(hereafter LABFLA). No Yellow Fever virus or West Nile virus screening was carried out for this
analysis. LABFLA is a centralized reference service that receives samples from all areas of the
state of Rio de Janeiro (http://www.fiocruz.br/ioclabs/cgi/cgilua.exe/sys/start.htm?sid=58).
Due to its status as a reference center, during the ZIKV and CHIKV epidemic, samples from
across the state were submitted to LABFLA. Sample collection was first coordinated at the
state of Rio de Janeiro public health lab, city health departments in the municipalities of
Campos dos Goytacazes and NiteroÂi and three research institutions: the Evandro Chagas Institute,
the Fernandes Figueira Institute, and the National School of Public Health. In addition, we
collected samples at the Oswaldo Cruz Foundation clinic in Manguinhos, which is a community
with 30,000 inhabitants in the northern district of the city of Rio de Janeiro. Patients were
referred to the clinic from other clinics or sought care there as they lived nearby due to clinical
suspicion of disease. After collection at the aforementioned institutions (Table 1), the samples
were sent to LABFLA for screening.
In Rio de Janeiro the first locally acquired ZIKV infections were detected in May 2015,
however LABFLA conducted a retrospective study of samples that were collected beginning in
January 2015, and tested negative for DENV by PCR. The retrospective study detected ZIKV
by RT-PCR in February 2015.
Previous studies have compared the diagnosis of ZIKV in Rio de Janeiro based on clinical
features versus laboratory tests. The results indicated that the symptoms that best discriminate
ZIKV from DENV and CHIKV are pruritus and conjunctival hyperemia [
]. We analyze
labconfirmed cases of CHIKV, DENV, and ZIKV. The use of lab-confirmed cases reduces the
3 / 15
geographic scope of the analysis somewhat to the extent that six municipalities in the state of
Rio de Janeiro reported clinical cases of CHIKV, DENV, or ZIKV but not lab-confirmed cases.
However, the municipalities that reported clinical cases but not lab-confirmed cases are all
small towns (average population: 30,000 inhabitants) that do not appear to have experienced
large outbreaks of CHIKV, DENV, and ZIKV.
Climatic, geographic, and infrastructural analysis
As cases of DENV were extremely rare, we investigated how climate and infrastructure affected
the timing and geographic pattern of incidence of ZIKV and CHIKV. The scale of the analysis
was the state of Rio de Janeiro (area: 44,000 km2), which has a warm temperate climate in
mountainous areas of the interior and an equatorial climate along the coast [
]. The larval
stages of Aedes aegypti require shallow pools of water to develop [
]. After a rainstorm, such
pools will be more abundant. In light of this, we hypothesized that after a delay of two to three
weeks following a major rain event there will be increased mosquito abundance, and
heightened vector transmission of ZIKV. To test this hypothesis, we obtained rainfall data from
March 2015 to December 2016 from six weather stations operated by the National Institute of
Further, we investigated socio-economic risk factors for ZIKV by analyzing data on
sanitation in the city of Rio de Janeiro. Previous studies in Rio de Janeiro have shown that lack of
access to municipal water prompts households to hoard water in barrels that become infested
with Aedes aegypti larvae, leading to increased vector abundance and DENV transmission
]. To assess whether water infrastructure also affects ZIKV risk, we compared the incidence
of ZIKV per neighborhood with the percentage of households serviced by the municipal
system. Water data were obtained from the GeoOpenData portal of the Rio de Janeiro Mayor's
Office. Water storage is linked to socio-economic status in Brazil insofar as having running
water in the household is correlated with income such that low-income households are less
likely to be connected to water infrastructure and more likely to store water in improvised
]. We calculated Pearson's product moment correlation coefficient between
CHIKV and ZIKV incidence and variables representing the level of urbanization at the
neighborhood scale and the availability of infrastructure such as access to the municipal water
We also tested whether there was a correlation between the ZIKV incidence in a
municipality and mosquito density. Our analysis used data from the 2016 Rapid Assessment of Aedes
aegypti Infestation of the State Health Department of Rio de Janeiro, which is a standardized
insect survey carried out at the municipal scale. During the survey, trained observers made
note of the presence of mosquito larvae and containers where mosquitos can oviposit such as
tires filled with water [
We developed a simulation model to explore what-if scenarios that might explain the timing
of CHIKV and ZIKV epidemics in Rio de Janeiro (S2 Table). The model was based on
differential equations like MacDonald-Ross models used to simulate malaria transmission [30±32].
Each arbovirus was modeled independently of the other using the epidemic model (Fig 1). The
human population was divided into three compartments: individuals susceptible to infection,
those currently infected, and the recovered. The model simplified the transmission cycle by
omitting sexual transmission. The mosquito population was divided into susceptible, exposed
but not infectious, and infected compartments. We assumed that due to the short lifespan of
the vector, it would not recover from the infection. Humans can become infected by being
4 / 15
Fig 1. Compartmental model used to simulate CHIKV and ZIKV epidemics.
bitten by an infected mosquito. Further, mosquitoes can become infected by biting an infected
human. In both the human and mosquito populations, new susceptibles are born into the
population and some susceptibles experience mortality due to causes other than the arbovirus.
The simulation model included parameters such as the incubation period of the viruses in
the mosquito. As it was not possible to measure these parameters in Rio de Janeiro, we used
values from the literature (S3 and S4 Tables). The model was implemented in MatLab using a
fourth order Runge-Kutta scheme.
Of the 10,459 patients screened for CHIKV, DENV, and ZIKV, 6,543 were negative for the
three arboviruses. There were 3,887 lab-confirmed cases of CHIKV and ZIKV including 1,717
of ZIKV (44.2%), and 2,170 of CHIKV (55.8%) (Table 1). ZIKV positive samples were detected
from February 2015 to May 2016 and CHIKV positive samples from September 2015 to
October 2016. Compared to the other two viruses, there were very few lab-confirmed cases of
DENV (n = 29). As DENV was very rare, we did not analyze the timing or spatial pattern of
Timing of rainfall, CHIKV, and ZIKV cases
From mid-September to early October 2015, a series of large rainstorms occurred, which were
followed 3±4 weeks later by the beginning of the largest outbreak of ZIKV, which began in
October and continued to December 2015 (Fig 2). The largest wave of the ZIKV epidemic
ended in the first quarter of 2016 (Fig 2). The greatest number of cases of ZIKV was reported
in January 2016, after which there was a steady decline in cases. Cases of CHIKV were lowest
in January and increased steadily every month during the first quarter of 2016. In late February
5 / 15
Fig 2. Confirmed cases of ZIKV and CHIKV per week in the state of Rio de Janeiro, March 2015 to May 2016 (LABFLA data set).
and early March 2016, major rains occurred, after which cases of CHIKV increased, peaking
in April. We did not observe a coupling between temperature or relative humidity and ZIKV
cases (S3 and S4 Figs).
According to the LABFLA database, the number of cases of CHIKV in February to May
2016 was greater than the number of cases of ZIKV.
Socio-economic and geographic characteristics of lab-confirmed cases of ZIKV and CHIKV
ZIKV incidence was inversely proportional to the percentage of households connected to
municipal water infrastructure in the city of Rio de Janeiro (r = -0.47, t = -6.1, df = 130, p = 1.2
x 10−8, Fig 3A). There was no relationship between mosquito density and the incidence of
ZIKV (r = 0.12, t = 0.899, df = 57, p = 0.37) or CHIKV (r = 0.035, t = 0.267, df = 57, p = 0.79).
CHIKV incidence increased with the percent of urbanized land in each neighborhood (r = 0.2,
t = 2.3, df = 130, p = 0.02, Fig 3B). Although significantly more individuals aged 40 or younger
were tested than those aged 40 or older, CHIKV incidence was greater in individuals 40 years
of age or older (p< 2.2 × 10−16, Table 2). ZIKV incidence was highest in individuals 20±39
years of age (4.81 cases/100,000 people). However, as ZIKV sampling was biased toward
women who were pregnant, it is not possible to drawn robust conclusions about incidence in
different age groups. The surveillance data also provided insights about the spatial pattern of
ZIKV incidence. There was geographic overlap between health regions with high incidence of
ZIKV and CHIKV, which could have led to competition of the viruses in Aedes aegypti (Fig 4).
We compared the timing of the observed epidemics to the timing of epidemics simulated
using the epidemic model. Our analysis indicated that the incubation period of the viruses in
the mosquito was an important parameter for determining the timing of ZIKV and CHIKV
epidemics. The incubation period of CHIKV in Aedes aegypti is 2±4 days whereas that of
ZIKV is at least 10 days (see refs. in S3 Table). When we parameterized the model with these
settings, CHIKV spread more rapidly and replaced ZIKV in the simulated mosquito
population. As CHIKV had a shorter incubation period than ZIKV in the mosquitoes in the
simulations, it was transmitted more frequently to humans than was ZIKV, which was associated
with a decline in the number of human cases of ZIKV and an increase in human cases of
6 / 15
Fig 3. Effect of infrastructure on CHIKV and ZIKV incidence in the city of Rio de Janeiro. Each point represents one neighborhood in the city of Rio de
Janeiro. Incidence was defined as the number of lab-confirmed cases per 10,000 inhabitants. (A) ZIKV incidence is greater in neighborhoods with little
access to municipal water supplies in the city of Rio de Janeiro. (B) CHIKV incidence increases with the percentage of urbanized land in the neighborhood.
CHIKV. Over time, the number of cases of CHIKV also declined in the simulated human
population as individuals recovered from the infection (Fig 5).
The CHIKV and ZIKV epidemics in Rio de Janeiro illustrate that integrating information
across viruses and climatic and socioeconomic variables reveals perspectives that would not
have been possible by examining one virus at a time. ZIKV did not become notifiable until
February 2016. If a decision-maker only had access to data from 2016, he or she would have
missed the association between rainfall and ZIKV in the last quarter of 2015. In addition, our
analysis provides insights about climatic and ecological factors associated with the start and
the end of the ZIKV epidemic, which could be useful for predicting and controlling future
arbovirus outbreaks. We found that heavy rainfall precedes cases by three weeks and is a
predictor of potential outbreaks. Rainstorms in early October 2015 likely would have increased
mosquito breeding sites and the abundance of adult mosquitoes after a lag of a few weeks. The
rainfall-associated increase in vector abundance may have triggered the largest ZIKV outbreak,
which began in late October 2015.
LABFLA lab-confirmed cases
CHIKV (%) ZIKV (%)
59 (2.7) 22 (1.3)
106 (4.9) 212 (12.3)
577 (26.6) 1319 (76.8)
903 (41.6) 137 (8)
525 (24.2) 27 (1.6)
7 / 15
Fig 4. Geographic pattern of ZIKV and CHIKV incidence. (A) High correlation between the incidence of
ZIKV and CHIKV in the Metro 2 health region, which comprises the eastern half of the metropolitan area of the
city of Rio de Janeiro. This geographic overlap between the viruses could have led to competition in Aedes
aegypti. The correlation is moderate but lower in the Metro 1 region. The Metro 1 and 2 regions represent 80%
of the state's population and 90% of the samples in this analysis. The correlation is also high in the North and
Coastal regions, whereas in other health regions, ZIKV and CHIKV appear not to be correlated. However, the
lower sample sizes outside the metropolitan area make it difficult to draw robust conclusions about the
correlation in these regions. (B) Relative proportion of ZIKV and CHIKV. ZIKV dominates over CHIKV in the
Rio de Janeiro metropolitan area and the Coastal and North health regions. (C) Human population density.
8 / 15
Fig 5. Model in which CHIKV outcompetes ZIKV is supported by the data. In the observed
epidemiological data (Fig 2), the epidemic of ZIKV occurred before that of CHIKV. In our simulations, CHIKV
spread more quickly than ZIKV in the mosquito, and the outbreak of ZIKV was followed by an outbreak of
In Rio de Janeiro, ZIKV and CHIKV both circulated at low prevalence in the first half of
2015, but did not trigger large epidemics. The largest wave of the ZIKV epidemic occurred in
the fourth quarter of 2015 followed by the CHIKV epidemic in the first quarter of 2016, which
leads to the question of why CHIKV did not spread before or at the same time as ZIKV.
Modeling studies have shown that when the prevalence of an infectious disease is low, as was
the case for ZIKV and CHIKV in the first half of 2015, the disease may go extinct due to
demographic stochasticity. Whether the disease goes extinct or triggers an epidemic depends on
factors such as the infectious period. Modeling studies indicate that if the infectious period is
highly variable, there is a greater risk that the disease will go extinct [
]. The infectious period
of CHIKV appears to be highly variable: while in many patients the illness resolves in ten days,
approximately 50% of patients may remain symptomatic with arthralgia for up to one year
]. The variable infectious period of CHIKV may have led to its extinction due to
demographic stochasticity in 2015, when the largest wave of the ZIKV epidemic occurred.
According to this scenario, CHIKV may have subsequently been reintroduced and spread in Rio de
Janeiro causing an epidemic in 2016.
Modeling studies indicate that when two arboviruses are present in a human population,
one virus will generally drive the other to extinction, and which one will persist is determined
by factors such as the number of mosquitoes that are initially infected with each virus [
The shorter incubation period of CHIKV may have made it a superior competitor that was
able to spread more quickly than ZIKV in the mosquito population. Other factors that could
9 / 15
have influenced the timing of simulated epidemics include the number of mosquitoes and
humans assumed to be susceptible to or infected with the two viruses at the beginning of the
Areas of high CHIKV and ZIKV incidence were not merely areas where there is always
high mosquito density. Instead, it appears to be levels of urbanization and access to municipal
water that contributed significantly to the CHIKV and ZIKV epidemics in the city of Rio de
Janeiro. Socio-economic status (SES) may affect arbovirus risk as people with lower SES may
have lifestyle factors such as living in more crowded conditions that increase arbovirus risk
]. More broadly, at the continental scale environmental factors like altitude may also
influence ZIKV risk; for instance, risk appears to decrease with altitude, as high elevations are
ecologically unsuitable for the mosquito . CHIKV incidence was weakly correlated with
the percentage of urbanized land per neighborhood but not water infrastructure. The two
viruses may have been associated with distinct environmental factors because of they differ
with respect to their rates of spread in the population. The basic reproduction number R0,
which is the estimated number of new cases generated by an infected individual, is estimated
to be 4 for CHIKV , but only 2 for ZIKV [
]. Due to its more rapid rate of spread,
CHIKV may be associated with different environmental factors than ZIKV the incidence of
which was lower in neighborhoods without water infrastructure.
Our finding that rainfall appears to precede ZIKV and CHIKV epidemics suggests that an
early warning system based on weather that predicts these outbreaks a few weeks in advance
would provide policy-makers and clinicians a warning to prepare countermeasures, which
could lead to improved prognoses for ZIKV patients. For example, GBS is treated with
intravenous immunoglobulin or plasmapheresis [40±42]. Physicians in tropical countries may not
have immunoglobulin on hand, but a weather-based early warning system could give them a
lead-time of a few weeks to gain access to the treatment. Our geographic analysis found that
health regions in the Rio de Janeiro metropolitan area and the North health region of the state
reported the highest ZIKV incidence and could be vulnerable to future outbreaks. The
vulnerable health regions identified here could be prioritized for strengthening the capacity of the
public health system to respond to ZIKV.
The analysis identified cohorts with high incidence of arbovirus infection, which could help
decision-makers prioritize human populations for educational campaigns and outreach.
Confirmed ZIKV cases were almost all women, which could be because women are more likely
than men to seek treatment, because health authorities allocated more resources to screening
samples from pregnant women due to the risk of fetal microcephaly, or because there is
maleto-female sexual transmission of ZIKV that resulted in higher incidence in women [
The rationale for our analysis of ZIKV and CHIKV incidence by age was that this was that
incidence of DENV increases with the percent of women and persons older than 60 in the
population, possibly because these populations have greater exposure to Aedes aegypti within the
26, 45, 46
]. Since CHIKV, DENV, and ZIKV are all transmitted by the same vector,
we hypothesized that the incidence of ZIKV and CHIKV would be higher in women and older
age groups. The incidence of CHIKV was higher in persons older than 40, which is similar to
the incidence by age group in Suriname . This pattern could arise because middle-aged
and elderly individuals are more apt to seek healthcare, have greater exposure to Aedes
mosquitoes because they spend more time indoors without air conditioning in developing
countries, or because their general health is poorer making them more susceptible to virus
Our analysis was subject to limitations that may restrict the generalizability of our findings
to other geographic regions. Since ZIKV is a novel pathogen, we only had samples for a
twoyear period. It is possible that the climatic and environmental drivers of ZIKV in 2015 and
10 / 15
2016 may not be important in future years, which would limit the extent to which rainfall
could be used to predict epidemics in other countries. Our results suggest that it is not merely
the occurrence of rainfall that triggers ZIKV epidemics. For example, there were rainstorms in
late December 2015 but ZIKV began declining during this period. Future work should
investigate how the intensity and duration of rainstorms affect mosquito abundance. For instance,
extremely intense rainstorms that produce a high volume of precipitation in a few hours may
flush larvae leading to decreased vector abundance and arbovirus transmission [
One hypothesis for the decline in ZIKV disease cases in the first quarter of 2016 was that
CHIKV and ZIKV competed within the vector Aedes aegypti, and CHIKV was the superior
competitor and spread more quickly. An alternative hypothesis is that ZIKV infections
decreased because the State Health Department implemented effective mosquito abatement
programs in early 2016 that reduced the transmission of the virus, however then one should
not have seen a subsequent rise in CHIKV cases which is dependent on the same vector. Yet
another hypothesis is that ZIKV disease cases declined because the population became infected
and acquired immunity. This hypothesis could be tested by measuring ZIKV seroprevalence.
Seroprevalence surveys in the Pacific Islands found that ZIKV infected 1% of the population of
New Caledonia and 12% of the population of French Polynesia [
]. As ZIKV seroprevalence
studies have not yet been carried out in Brazil [
], the population immunity hypothesis
awaits confirmation in future studies. An additional shortcoming of the analysis is that our
geographic analysis assumes that an individual was infected in the municipality where he or
she resides. As Aedes mosquitoes are diurnal, it is possible that individuals are bitten not in the
household but in the workplace, which could be located in a different municipality.
Nevertheless, our data are robust in the sense that our identification of ZIKV cases was
strictly based on molecular assays, which provides a definitive diagnosis. Identification of
ZIKV cases based on serology or clinical findings can result in false positive results due to
serologic cross-reactivity between ZIKV and prior existing DENV antibodies in patients residing
in endemic areas. Diagnosis based on clinical symptoms can also misclassify cases of ZIKV,
DENV or CHIKV as the diseases have similar clinical presentations and tend to co-circulate in
endemic areas. The WHO's ZIKV Strategic Plan states that controlling arboviruses requires
mapping their social and environmental drivers [
]. Our findings can contribute to such
efforts in the state of Rio de Janeiro by showing that rainfall predicts arbovirus epidemics and
by identifying the cohorts and geographic regions with the highest incidence. It is plausible
that in the coming years other arboviruses will expand their ranges from Africa and Asia to the
Americas like CHIKV and ZIKV. The development of accurate predictive models and
surveillance data analysis approaches is essential if we are to prevent the next novel virus from
causing a major public health disaster like ZIKV in Brazil.
S1 Fig. Timeline of ZIKV and CHIKV epidemics and surveillance in the state of Rio de
S2 Fig. Incidence of lab-confirmed ZIKV and CHIKV cases.
S3 Fig. Timing of temperature and lab-confirmed cases of ZIKV (LABFLA dataset)
January 2015-July 2016.
11 / 15
S4 Fig. Timing of relative humidity and lab-confirmed cases of ZIKV (LABFLA dataset)
January 2015-July 2016.
S1 Table. Incidence of ZIKV and CHIKV infection by municipality in the state of Rio de
S2 Table. Formulation of the epidemic model.
S3 Table. ZIKV and CHIKV incubation period in Aedes mosquitoes.
S4 Table. Settings used in the epidemic model simulations.
Carolina Cardoso dos Santos, Aline da Silva Santos, CÂõntia Damasceno, Leda Maria dos
Santos, and JoseÂ Farias Filho for processing samples.
Conceptualization: Trevon L. Fuller, Guilherme Calvet, Camila Genaro Estevam, Jussara
Rafael Angelo, Umme-Aiman Halai, Ryan Harrigan, Thomas B. Smith, PatrÂõcia Brasil,
Data curation: Guilherme Calvet, Camila Genaro Estevam, PatrÂõcia Brasil, Karin
Formal analysis: Trevon L. Fuller, Camila Genaro Estevam, Gbenga J. Abiodun, Ryan
Harrigan, Karin Nielsen-Saines.
Funding acquisition: Karin Nielsen-Saines.
Investigation: Camila Genaro Estevam, Jussara Rafael Angelo, Ryan Harrigan, Thomas B.
Smith, Claudia Raja Gabaglia, PatrÂõcia Brasil, Karin Nielsen-Saines.
Methodology: Trevon L. Fuller, Camila Genaro Estevam, Jussara Rafael Angelo, Gbenga J.
Abiodun, Bianca De Santis, Patricia Carvalho Sequeira, Eliane Machado Araujo, Simone
Alves Sampaio, Marco Cesar Lima de MendoncËa, Allison Fabri, Rita Maria Ribeiro, Ryan
Harrigan, Thomas B. Smith, PatrÂõcia Brasil, Ana Maria Bispo de Filippis, Karin
Project administration: Karin Nielsen-Saines.
Software: Gbenga J. Abiodun.
Supervision: Thomas B. Smith, PatrÂõcia Brasil, Ana Maria Bispo de Filippis, Karin
Visualization: Camila Genaro Estevam, Gbenga J. Abiodun, Ryan Harrigan, Karin
Writing ± original draft: Trevon L. Fuller, Camila Genaro Estevam.
12 / 15
Writing ± review & editing: Trevon L. Fuller, Guilherme Calvet, Camila Genaro Estevam,
Jussara Rafael Angelo, Gbenga J. Abiodun, Umme-Aiman Halai, Ryan Harrigan, Thomas B.
Smith, Claudia Raja Gabaglia, PatrÂõcia Brasil, Ana Maria Bispo de Filippis, Karin
13 / 15
2008; 14(8):1232±9. Epub 2008/08/06. https://doi.org/10.3201/eid1408.080287 PMID: 18680646;
PubMed Central PMCID: PMCPMC2600394.
14 / 15
1. MinisteÂrio da SauÂde . Monitoramento integrado de alteracËões no crescimento e desenvolvimento relacionadas à infeccËão pelo vÂõrus Zika e outras etiologias infecciosas , ateÂ a Semana EpidemioloÂgica 16 / 2017 . BrasÂõlia: Secretaria de VigilaÃncia em SauÂde 2017 .
2. Rodrigues LC . Microcephaly and Zika virus infection . The Lancet . 2016 ; 387 : 2070 ± 1 . https://doi.org/10. 1016/S0140- 6736 ( 16 ) 00742 -X PMID: 26993880 .
3. Wang L , Valderramos SG , Wu A , Ouyang S , Li C , Brasil P , et al. From mosquitos to humans: genetic evolution of Zika virus . Cell Host & Microbe . 2016 ; 19 ( 5 ): 561 ±5. https://doi.org/10.1016/j.chom. 2016 . 04 .006 PMID: 27091703
4. Fredericks AC , Fernandez-Sesma A . The burden of dengue and Chikungunya worldwide: implications for the southern United States and California . Ann Glob Health. 2014 ; 80 ( 6 ): 466 ± 75 . https://doi.org/10. 1016/j.aogh. 2015 . 02 .006 PMID: 25960096
5. Vos T , Barber RM , Bell B , Bertozzi-Villa A , Biryukov S , Bolliger I , et al. Global, regional, and national incidence, prevalence, and years lived with disability for 301 acute and chronic diseases and injuries in 188 countries, 1990 ±2013: a systematic analysis for the Global Burden of Disease Study 2013 . Lancet. 2015 ; 386 ( 9995 ): 743 ± 800 . https://doi.org/10.1016/S0140- 6736 ( 15 ) 60692 - 4 PMID: 26063472
6. Bhatt S , Gething PW , Brady OJ , Messina JP , Farlow AW , Moyes CL , et al. The global distribution and burden of dengue . Nature . 2013 ; 496 ( 7446 ): 504 ±7. https://doi.org/10.1038/nature12060 PMID: 23563266
7. Roth A , Mercier A , Lepers C , Hoy D , Duituturaga S , Benyon E , et al. Concurrent outbreaks of dengue, Chikungunya and Zika virus infectionsÐan unprecedented epidemic wave of mosquito-borne viruses in the Pacific , 2012 ± 2014 . Eurosurveillance. 2014 ; 19 ( 41 ): 20929 . https://doi.org/10.2807/ 1560 - 7917 . ES2014 . 19 .41.20929 PMID: 25345518
8. Weaver SC . Arrival of Chikungunya virus in the New World: prospects for spread and impact on public health . PLoS Neglected Tropical Diseases . 2014 ; 8 ( 6 ):e0002921. https://doi.org/10.1371/journal.pntd. 0002921 PMID: 24967777
9. Faye O , Freire CCM , Iamarino A , Faye O , de Oliveira JVC , Diallo M , et al. Molecular evolution of Zika virus during its emergence in the 20th Century . PLoS Neglected Tropical Diseases . 2014 ; 8 ( 1 ):e2636. https://doi.org/10.1371/journal.pntd. 0002636 PMID: 24421913
10. Nunes de Lima-Camara T , Alves Honorio N. Climate change and its effect on urban mosquitoes in South America . In: Dhang P, editor. Climate Change Impacts on Urban Pests . Oxford, UK: CABI; 2016 . p. 127 ± 40 .
11. Brasil P , Pereira JP , Moreira ME , Nogueira RMR , Damasceno L , Wakimoto M , et al. Zika virus infection in pregnant women in Rio de Janeiro. New England Journal of Medicine . 2016 ; 375 ( 24 ): 2321 ± 34 . https://doi.org/10.1056/NEJMoa1602412 PMID: 26943629
12. Paixao ES , Barreto F , Teixeira MG , Costa MC , Rodrigues LC . History, epidemiology, and clinical manifestations of Zika: a systematic review . American Journal of Public Health . 2016 ; 106 ( 4 ): 606 ± 12 . https://doi.org/10.2105/AJPH. 2016 .303112 PMID: 26959260
13. Krauer F , Riesen M , Reveiz L , Oladapo OT , Martinez-Vega R , Porgo TV , et al. Zika virus infection as a cause of congenital brain abnormalities and Guillain-Barre Syndrome: systematic review . PLoS Medicine . 2017 ; 14 ( 1 ):e1002203. https://doi.org/10.1371/journal.pmed. 1002203 PMID: 28045901
14. Dos Santos T , Rodriguez A , Almiron M , Sanhueza A , Ramon P , de Oliveira WK , et al. Zika and the Guillain-Barre SyndromeÐcase series from seven countries . New England Journal of Medicine . 2016;October 20 , 2016 : 1598 ±9. https://doi.org/10.1056/NEJMc1609015 PMID: 27579558
15. City Health Department of Rio de Janeiro. Cases of Chikungunya by month of symptom onset . http:// www.rio.rj.gov.br/web/sms/exibeConteudo?id= 4769664 . Rio de Janeiro: Mayor's Office of Rio de Janeiro; 2015 .
16. Ueleres Braga J , Bressan C , Razal Dalvi AP , Calvet GA , Paiva Dumas R , Rodrigues N , et al. Accuracy of Zika virus disease case definition during simultaneous Dengue and Chikungunya epidemics . Plos One . 2017 ; 12 ( 6 ):e0179725. https://doi.org/10.1371/journal.pone. 0179725 PMID: 28650987
17. Lanciotti RS , Kosoy OL , Laven JJ , Velez JO , Lambert AJ , Johnson AJ , et al. Genetic and serologic properties of Zika virus associated with an epidemic, Yap State , Micronesia, 2007 . Emerg Infect Dis.
18. Duffy MR , Chen T-H , Hancock WT , Powers AM , Kool JL , Lanciotti RS , et al. Zika virus outbreak on Yap Island, Federated States of Micronesia. New England Journal of Medicine . 2009 ; 360 ( 24 ): 2536 ± 43 . https://doi.org/10.1056/NEJMoa0805715 PMID: 19516034
19. Santiago GA , Vergne E , Quiles Y , Cosme J , Vazquez J , Medina JF , et al. Analytical and clinical performance of the CDC real time RT-PCR assay for detection and typing of dengue virus . PLoS Negl Trop Dis . 2013 ; 7 ( 7 ): e2311 . Epub 2013 /07/23. https://doi.org/10.1371/journal.pntd.0002311 PMID: 23875046; PubMed Central PMCID : PMCPMC3708876 .
20. Lanciotti RS , Kosoy OL , Laven JJ , Panella AJ , Velez JO , Lambert AJ , et al. Chikungunya virus in US travelers returning from India , 2006 . Emerg Infect Dis . 2007 ; 13 ( 5 ): 764 ± 7 . Epub 2007/06/08. https://doi. org/10.3201/eid1305.070015 PMID: 17553261; PubMed Central PMCID : PMCPMC2738459 .
21. Brasil P , Calvet GA , Siqueira AM , Wakimoto M , de Sequeira PC , Nobre A , et al. Zika Virus outbreak in Rio de Janeiro, Brazil: clinical characterization, epidemiological and virological aspects . PLoS Neglected Tropical Diseases . 2016 ; 10 ( 4 ):e0004636. https://doi.org/10.1371/journal.pntd. 0004636 PMID: 27070912
22. Litzba N , Schuffenecker I , Zeller H , Drosten C , Emmerich P , Charrel R , et al. Evaluation of the first commercial Chikungunya virus indirect immunofluorescence test . J Virol Methods . 2008 ; 149 ( 1 ): 175 ±9. https://doi.org/10.1016/j.jviromet. 2008 . 01 .004 PMID: 18294706
23. Ross J. Geografia do Brasil. São Paulo: Universidade de São Paulo; 2005 .
24. Chouin-Carneiro T , Vega-Rua A , Vazeille M , Yebakima A , Girod R , Goindin D , et al. Differential susceptibilities of Aedes aegypti and Aedes albopictus from the Americas to Zika Virus . PLoS Neglected Tropical Diseases . 2016 ; 10 ( 3 ):e0004543. https://doi.org/10.1371/journal.pntd. 0004543 PMID: 26938868
25. INMET. Banco de Dados MeteoroloÂgicos para Ensino e Pesquisa (BDMEP). BrasÂõlia: Instituto Nacional de Meteorologia (INMET); 2017 .
26. Flauzino RF , Souza-Santos R , Magalhães de Oliveira R. Socio-environmental indicators for dengue surveillance in the local level . SauÂde Soc São Paulo . 2011 ; 20 ( 1 ): 225 ± 40 . https://doi.org/10.1590/ S0104-12902011000100023
27. Battiston D , Cruces G , Lopez-Calva LF , Lugo MA , Santos ME . Income and beyond: multidimensional poverty in six Latin American countries . Soc Indic Res . 2013 ; 112 ( 2 ): 291 ± 314 . https://doi.org/10.1007/ s11205-013-0249-3
28. de Oliveira AF , Leite ID , Valente JG . Global burden of diarrheal disease attributable to the water supply and sanitation system in the State of Minas Gerais , Brazil: 2005 . Cienc SauÂde Coletiva. 2015 ; 20 ( 4 ): 1027 ± 36 . https://doi.org/10.1590/ 1413 - 81232015204 .00372014 PMID: 25923615
29. MinisteÂrio da SauÂde. Levantamento RaÂpido de ÂIndices para Aedes aegyptiÐLIRAaÐpara VigilaÃncia EntomoloÂgica do Aedes aegypti no Brasil. Metodologia para AvaliacËão dos ÂIndices de Breteau e Predial e Tipo de Recipientes. BrasÂõlia: MinisteÂrio da SauÂde; 2013 .
30. Keeling M , Rohani P . Modeling Infectious Diseases in Humans and Animals . Princeton, New Jersey: Princeton University Press; 2008 .
31. Keeling MJ , Danon L . Mathematical modelling of infectious diseases . British Medical Bulletin . 2009 ; 92 ( 1 ): 33 ±42 https://doi.org/10.1093/bmb/ldp038 PMID: 19855103
32. Smith DL , Battle KE , Hay SI , Barker CM , Scott TW , McKenzie FE . Ross, Macdonald, and a theory for the dynamics and control of mosquito-transmitted pathogens . PLoS Pathogens . 2012 ; 8:e1002588 . PMCID: PMC3320609 https://doi.org/10.1371/journal.ppat. 1002588 PMID: 22496640
33. Rodriguez-Morales AJ , Restrepo-Posada VM , Acevedo-Escalante N , Rodriguez-Munoz ED , ValenciaMarin M , Castrillon-Spitia JD , et al. Impaired quality of life after chikungunya virus infection: a 12-month follow-up study of its chronic inflammatory rheumatism in La Virginia , Risaralda, Colombia. Rheumatol Int . 2017 ; 37 ( 10 ): 1757 ±8. https://doi.org/10.1007/s00296-017 -3795-1 PMID: 28828632
34. Feng ZL , Velasco Hernandez JX . Competitive exclusion in a vector-host model for the Dengue fever . Journal of Mathematical Biology . 1997 ; 35 ( 5 ): 523 ± 44 . https://doi.org/10.1007/s002850050064 PMID: 9145954
35. Vincenti-Gonzalez MF , Grillet ME , Velasco-Salas ZI , Lizarazo EF , Amarista MA , Sierra GM , et al. Spatial analysis of dengue seroprevalence and modeling of transmission risk factors in a dengue hyperendemic city of Venezuela . Plos Neglected Tropical Diseases . 2017 ; 11 ( 1 ): 21 . https://doi.org/10.1371/ journal.pntd. 0005317 PMID: 28114342
36. LaBeaud AD. Why arboviruses can be neglected tropical diseases . PLoS Neglected Tropical Diseases . 2008 ; 2 ( 6 ):e247. https://doi.org/10.1371/journal.pntd.0000247 PMID: 18575597 Watts AG , Miniota J , Joseph HA , Brady OJ , Kraemer MUG , Grills AW , et al. Elevation as a proxy for mosquito-borne Zika virus transmission in the Americas . PLoS ONE . 2017 ; 12 ( 5 ):e0178211. https://doi. org/10.1371/journal.pone. 0178211 PMID: 28542540
38. Yakob L , Clements ACA . A mathematical model of Chikungunya dynamics and control: the major epidemic on ReÂunion Island . PLoS One . 2013 ; 8 ( 3 ):e57448. https://doi.org/10.1371/journal.pone. 0057448 PMID: 23554860
39. Villela DAM , Bastos LS , Carvalho LM , Cruz OG , Gomes MFC , Durovni B , et al. Zika in Rio de Janeiro: Assesment of basic reproduction number and comparison with dengue outbreaks . Epidemiology and Infection . 2017 ; 145 : 1649 ± 57 . https://doi.org/10.1017/S0950268817000358 PMID: 28240195
40. Kazatchkine MD , Kaveri SV . Advances in immunology: immunomodulation of autoimmune and inflammatory diseases with intravenous immune globulin . New England Journal of Medicine . 2001 ; 345 ( 10 ): 747 ± 55 . https://doi.org/10.1056/NEJMra993360 PMID: 11547745
41. Vandermeche FGA , Schmitz PIM . A randomized trial comparing intravenous immune globulin and plasma-exchange in Guillain-Barre Syndrome . New England Journal of Medicine . 1992 ; 326 ( 17 ): 1123 ± 9. https://doi.org/10.1056/NEJM199204233261705 PMID: 1552913
42. Hughes RAC , Cornblath DR . Guillain-Barre syndrome . Lancet . 2005 ; 366 ( 9497 ): 1653 ± 66 . https://doi. org/10.1016/S0140- 6736 ( 05 ) 67665 - 9 PMID: 16271648
43. CodecËo Coelho F , Durovni B , Saraceni V , Lemos C , Torres Codeco C , Camargo S , et al. Higher incidence of Zika in adult women than adult men in Rio de Janeiro suggests a significant contribution of sexual transmission from men to women . International Journal of Infectious Diseases . 2016 ; 51 : 128 ± 32 . https://doi.org/10.1016/j.ijid. 2016 . 08 .023 PMID: 27664930
44. Foy BD , Kobylinski KC , Foy JLC , Blitvich BJ , da Rosa AT , Haddow AD , et al. Probable non-vectorborne transmission of Zika virus , Colorado, USA. Emerging Infectious Diseases. 2011 ; 17 ( 5 ): 880 ±2. https://doi.org/10.3201/eid1705.101939 PMID: 21529401
45. Flauzino RF , Souza-Santos R , Barcellos C , Gracie R , Magalhães MA , Magalhães R . Spatial heterogeneity of dengue fever in local studies, city of NiteroÂi, Southeastern Brazil . Revista de SauÂde PuÂblica. 2009 ; 43 : 1035 ± 43 . https://doi.org/10.1590/S0034-89102009005000064 PMID: 19967254
46. Flauzino RF , Souza-Santos R , Magalhães MA , Magalhães R. Dengue , geoprocessamento e indicadores socioeconoÃmicos e ambientais: um estudo de revisão. Panamerican Journal of Public Health . 2009 ; 25 ( 5 ): 456 ± 61 . https://doi.org/10.1590/S1020-49892009000500012
47. van Genderen FT , Krishnadath I , Sno R , Grunberg MG , Zijlmans W , Adhin MR . First Chikungunya outbreak in Suriname; clinical and epidemiological features . PloS Neglected Tropical Diseases . 2016 ; 10 ( 4 ): 18 . https://doi.org/10.1371/journal.pntd. 0004625 PMID: 27082985
48. Tompkins AM , Ermert V. A regional-scale, high resolution dynamical malaria model that accounts for population density, climate and surface hydrology . Malar J . 2013 ; 12 : 24 . https://doi.org/10.1186/ 1475 - 2875-12-65
49. Musso D , Gubler DJ. Zika Virus . Clinical Microbiology Reviews . 2016 ; 29 ( 3 ): 487 ± 524 . https://doi.org/ 10.1128/CMR.00072-15 PMID: 27029595
50. Jaenisch T , Rosenberger KD , Brito C , Brady O , Brasil P , Marques ETA . Risk of microcephaly after Zika virus infection in Brazil, 2015 to 2016 . Bulletin of the World Health Organization. 2017 ; 95 ( 3 ): 191 ±8. https://doi.org/10.2471/BLT.16.178608 PMID: 28250532
51. Brito CA , Cordeiro MT . One year after the Zika virus outbreak in Brazil: from hypotheses to evidence . Revista da Sociedade Brasileira de Medicina Tropical . 2016 ; 49 : 537 ± 43 . https://doi.org/10.1590/ 0037 - 8682- 0328 -2016 PMID: 27812646
52. World Health Organization. Zika Strategic Response Plan: Updated 30 June 2016 . Genebra: WHO; 2016 . 50 p.