Assessing the risk of autochthonous yellow
fever transmission in Lazio, central Italy
, Giorgio Guzzetta
, Federico FilipponiID
, Angelo Solimini
, Alessandra della Torre
, Roberto RosàID
*, Stefano Merler
1Department of Biodiversity and Molecular Ecology, Research and Innovation Centre, Fondazione Edmund
Mach, San Michele all’Adige, Trento, Italy, 2Center for Information Technology, Fondazione Bruno Kessler,
Trento, Italy, 3Epilab-JRU, FEM-FBK Joint Research Unit, Trento, Italy, 4Department of Public Health and
Infectious Diseases, Laboratory affiliated to Istituto Pasteur Italia–Fondazione Cenci Bolognetti, Sapienza
University of Rome, Rome, Italy
☯These authors contributed equally to this work.
Yellow fever virus (YFV) causes a highly lethal mosquito-borne disease that has recently
reemerged after 30 years of low incidence due to vaccination campaigns. Large epidemics
occurred in Angola and Democratic Republic of the Congo (DRC) in 2016 through 2017
(overall about 1,000 confirmed cases and 140 deaths) and in Brazil in 2017 through 2018
(2,037 confirmed cases and 674 deaths) [1,2]. The high international connectivity of Brazil
raises concern about the potential spread of disease to other countries by infected travelers [3,
4]; this possibility was confirmed during spring 2018, with the notification of six infected trav-
elers from five European countries, two of which had a fatal outcome . Recent laboratory
experiments suggest that European populations of Aedes albopictus may be competent for
transmission of YFV , and therefore large areas highly infested by this species in Mediterra-
nean countries are potentially exposed to the risk of outbreaks .
Here, we provide a quantitative assessment of the risk of YFV transmission in Lazio, the
central Italian region where the metropolitan city of Rome is located and where the largest
arboviral outbreak in continental Europe occurred in summer 2017 . To do so, we adapted
a stochastic transmission model, previously developed to assess the transmission risk of chi-
kungunya virus (CHIKV)  in the same area, to account for relevant epidemiological dynam-
ics of YFV, using existing field data on A.albopictus abundance  and biting rate on
The assessment of YFV transmission risk was carried out on 18 sampling sites placed along a
70 km transect representing a wide range of ecological landscapes, from low–human-popula-
tion density areas (coastal and rural sites) to highly urbanized areas (metropolitan city of
Rome). The site-specific vector abundances over time were characterized by calibrating a mos-
quito population model against observed mosquito captures , taking as input the average
daily temperature . The average vector density between July and September was estimated
to range between 154 and 4,866 female mosquitoes/ha, whereas the human density within
sampling sites ranged from 5 to 267 inhabitants/ha. The selected sites covered a wide range of
the vector-to-host ratio (4 to 138 female mosquitoes per inhabitant; see S1 Fig).
PLOS Neglected Tropical Diseases | https://doi.org/10.1371/journal.pntd.0006970 January 10, 2019 1 / 6
Citation: Manica M, Guzzetta G, Filipponi F,
Solimini A, Caputo B, della Torre A, et al. (2019)
Assessing the risk of autochthonous yellow fever
transmission in Lazio, central Italy. PLoS Negl Trop
Dis 13(1): e0006970. https://doi.org/10.1371/
Editor: Christopher M. Barker, University of
California, Davis, UNITED STATES
Published: January 10, 2019
Copyright: ©2019 Manica et al. This is an open
access article distributed under the terms of the
Creative Commons Attribution License, which
permits unrestricted use, distribution, and
reproduction in any medium, provided the original
author and source are credited.
Funding: The authors received no specific funding
for this work.
Competing interests: The authors have declared
that no competing interests exist
The stochastic transmission model previously developed for CHIKV  was adapted to model
YFV transmission. Briefly, the overall model includes a temperature-driven model providing
the abundance of A.albopictus and was calibrated to mosquito-capture data, coupled with a
disease-transmission model, and was informed with available estimates on epidemiological
parameters for YFV and was initialized with a single imported infection in a fully susceptible
population. At the end of a patient’s incubation period, we sampled the occurrence of clinical
symptoms from a binomial distribution. In baseline simulations, we assumed that only symp-
tomatic patients (including both mild and severe cases) are able to transmit the virus. Fatal
outcome was modeled as a binomial process at the end of the infectious period for symptom-
atic individuals, given that severe symptoms develop in a small fraction of patients only after
the end of the viremic period . The outcome of the importation of a single infected case
was evaluated at different times between May 1st and November 15th. To account for both sto-
chastic effects and the uncertainty in epidemiological parameters, we evaluated model simula-
tions by repeatedly sampling parameters from known distributions. The total number of
simulations for each study site was set to 30,000 (i.e., about 1,000 per week of importation).
Full details on the model’s structure are described in the S1 File.
We estimated the basic reproduction numbers over time, the probability of occurrence of
an outbreak, and the expected number of yellow fever (YF) cases and fatalities, under the
assumption of no disease control (either by vaccination or by vector management
The basic reproduction number R
represents the average number of secondary cases trans-
mitted by a typical infector during the entire period of infectivity; a value larger than 1 implies
potential for a sustained epidemic in the population. R
was computed for different times of
importation from standard equations [14,15], after adjusting to consider only symptomatic
cases (see S2 File). R
never exceeded the epidemic threshold in urban sites. However, some
coastal and rural sites had an average value of R
of about 2.1 at the peak, with an epidemic sea-
son (i.e., the time span over which R
exceeds 1) that extended from mid-July to the end of
Depending on the site, the average probability of autochthonous symptomatic YF cases
throughout the study period ranged between 1.0% and 11.8% (Fig 1); large outbreaks, here
defined as those involving more than 50 symptomatic cases, were unlikely (less than 0.65% in
all sites). The risk of YFV autochthonous transmission was uneven during the mosquito breed-
ing season but remained relatively stable between mid-July and mid-September in all sites.
However, the probability of occurrence of a large outbreak was higher for importations occur-
ring in mid-July, due to the broader time window during which transmission was possible.
Coastal and rural sites, characterized by high vector-to-host ratios, had a higher probability of
autochthonous transmission (up to 33.1%) and large outbreaks (up to 9.0%) compared to met-
ropolitan sites (maximum probability of autochthonous transmission: 25.9%; maximum prob-
ability of large outbreaks: 4.0%). The average probability of observing at least one death ranged
between 7.1% and 12.3% and that of observing more than 10 deaths was below 0.5% for all
sites; however, peak probabilities of observing at least one death reached 27.6% for importa-
tions occurring in mid-July in coastal and rural sites (see S2 Fig).
Given the observed importation of YFV in Europe via infected travelers  and the laboratory
competence of European A.albopictus populations to transmit this virus , we quantified the
risk of YFV transmission for Lazio, Italy, a region that was affected by a CHIKV outbreak
PLOS Neglected Tropical Diseases | https://doi.org/10.1371/journal.pntd.0006970 January 10, 2019 2 / 6
during the summer of 2017 . We showed that, given one imported case in urban areas, the
risk of transmission is generally low and limited to sporadic cases. However, for some coastal
and rural sites there is a nonnegligible potential for large outbreaks, especially if importation
Fig 1. Autochthonous symptomatic YF cases. Probability of autochthonous symptomatic YF cases estimated by the model in 18 sites in Lazio region (Italy),
conditional to the introduction of a single imported case at different times of the year and disaggregated by number of secondary cases. Font colors for site
IDs represent the geographic classification of the site. Blue: coastal, red: urban, green: rural. YF, yellow fever.
PLOS Neglected Tropical Diseases | https://doi.org/10.1371/journal.pntd.0006970 January 10, 2019 3 / 6
occurs during the second half of July. In practice, given the severity of disease caused by YFV,
it is likely that ongoing local transmission will be promptly identified and limited by integrated
control measures (not included in our model); nonetheless, this result confirms the impor-
tance of early outbreak detection capacity . As previously shown for CHIKV , the higher
risk in coastal and rural sites is related to lower human population densities, which tend to
increase the vector-to-host ratio. One of the rural sites with highest estimated YFV transmis-
sion risk is Fiumicino, where the largest Italian international airport is located. This increases
the chance of presence of potentially infected travelers compared to other rural areas. It is rele-
vant to note that our estimates of vector abundance and vector mortality rates are based on
temperatures recorded during 2017. In addition, a precise quantification of risks is subject to
many uncertainties on epidemiological parameters of YFV transmission in European A.albo-
pictus mosquitoes. These uncertainties include, but are not limited to, their dependence on
temperature, potential variability across different strains of the virus, and deviations between
laboratory measures and actual conditions in the field.
Another open question concerns the potential infectiousness of asymptomatic cases, which
is difficult to prove as they are mostly identified a posteriori using serological investigations.
Relaxing the assumption that asymptomatic individuals do not transmit and assuming in the
extreme opposite case that they transmit at the same rate as symptomatic patients, the esti-
mated risks would be much higher. For example, the peak probability of local transmission
would exceed 60% in some metropolitan sites and 80% in coastal and rural ones.
Recent estimates suggest that two YF cases were imported in Italy in 2017 ; the number
of imported cases for 2018 might be slightly larger due to a higher incidence of infection in
Brazil in 2018 (by about 50%) . Imported cases are more likely to arrive during the Brazilian
summer, which corresponds to the European winter, when mosquito populations are not
active, and in crowded urban areas with lower vector-to-host ratios. This largely reduces the
likelihood of an importation in Italy at a time and site of favorable conditions for transmission.
Nonetheless, YF cases were confirmed in Brazil throughout June through September 2017, so
the possibility of importation during the European summer is not to be completely discarded.
Furthermore, many other areas of the world are at risk of YFV outbreaks because of their
international connectivity and insufficient vaccination coverage : the possible expansion of
YFV to countries with year-round (rather than seasonal) transmission might significantly
increase the chances of importation in Europe during the summer in the near future.
Finally, we note that the estimated risk of locally transmitted symptomatic cases of YFV in
the considered study area is in the same order of magnitude of dengue (see S3 File). Although
introductions of dengue virus (DENV) are currently much more frequent (and indeed local
transmission of DENV has repeatedly been detected in France and Croatia ), the higher
severity of YF and the intrinsic stochasticity by which cases arrive over time and space suggest
that the risk of local YFV transmission should not be neglected.
Overall, the present work reveals a low, but nonnegligible risk of YFV transmission in Euro-
pean areas characterized by substantial A.albopictus infestation and medium-to-low human
density. Considering the severity of YF, this result highlights the need for public health author-
ities to ensure early diagnosis (not trivial since YF has not been reported in Italy since the 19th
century ), prompt notification of infected cases and swift responses targeting mosquito
populations through vector control interventions and the human population via reactive vacci-
nation campaigns. The recent unexpected rise of YF has caused a worldwide shortage in vac-
cine stockpiles, which has led to the adoption of fractional dosing immunization during the
PLOS Neglected Tropical Diseases | https://doi.org/10.1371/journal.pntd.0006970 January 10, 2019 4 / 6
recent Brazilian outbreaks . Although this approach seems to have been effective in con-
trolling the epidemics locally, many questions remain open . More generally, the availabil-
ity of YFV vaccine stockpiles at the global scale may be an important challenge to outbreak
control in the future. For these reasons, public health authorities might consider preventive
general-purpose risk reduction measures such as larviciding, given their demonstrated cost-
effectiveness in simultaneously preventing outbreaks of different arboviruses, even in areas
with limited risks .
S1 Fig. Map of the study sites. Location of the 18 sites for which mosquito abundance esti-
mates were available. The study sites are located along a 70 km transect encompassing the met-
ropolitan city of Rome, Lazio region, Italy. Four sticky traps were placed within each site and
weekly mosquito collection lasted from July to November 2012 . The area of circles repre-
sents the estimated peak vector-to-host ratio of each site, averaged across the period July
through September. Dark grey areas indicate human density higher than 10 inhabitants/ha.
Base layers elaborated from ISTAT data (https://www.istat.it). Spatial data processing and map
layout generation were done using QGIS (https://www.qgis.org). ISTAT, Istituto Nazionbale
di Statistica; QGIS, Quantum Geographic Information System.
S2 Fig. YF fatal cases. Probability of fatal outcome due to autochthonous YF transmission
estimated by the model in 18 sites in Lazio region (Italy), conditional to the introduction of a
single imported case at different times of the year and disaggregated by the number of expected
deaths. YF, yellow fever.
S1 File. YF model description. YF, yellow fever.
S2 File. YF basic reproductive number (R
). YF, yellow fever.
S3 File. Results for dengue transmission.
1. World Health Organization (WHO). Emergencies Yellow fever outbreak Angola, Democratic Republic
of the Congo and Uganda 2016–2017. Available from: http://www.who.int/emergencies/yellow-fever/
en/. [cited 2018 June 6].
2. Ministerio da Saude, Brazil. Monitoramento do Perı´odo Sazonal da Febre Amarela. Brasil– 2017/2018.
Informe no. 26, 2017/2018. Available from: http://portalarquivos2.saude.gov.br/images/pdf/2018/maio/
18/Informe-FA-26.pdf. [cited 2018 June 6].
3. Dorigatti I, Hamlet A, Aguas R, Cattarino L, Cori A, Donnelly CA, et al. International risk of yellow fever
spread from the ongoing outbreak in Brazil, December 2016 to May 2017. Euro Surveillance. 2017; 22
(28): pii = 30572. https://doi.org/10.2807/1560-7917.ES.2017.22.28.30572 PMID: 28749337
4. Brent SE, Watts A, Cetron M, German M, Kraemer MUG, Bogoch II, et al. International travel between
global urban centres vulnerable to yellow fever transmission. Bull World Health Organ. 2018; 96(5):
343–354B. https://doi.org/10.2471/BLT.17.205658 PMID: 29875519
5. Oliosi E, Corcos SE, Barroso PF, Bleibtreu A, Grard G, De Filippis BAM, et al. Yellow fever in two
unvaccinated French tourists to Brazil, January and March, 2018. Euro Surveillance. 2018; 23(21): pii =
1800240. https://doi.org/10.2807/1560-7917.ES.2018.23.21.1800240 PMID: 29845927
PLOS Neglected Tropical Diseases | https://doi.org/10.1371/journal.pntd.0006970 January 10, 2019 5 / 6
6. Amraoui F, Vazeille M, Failloux AB. French Aedes albopictus are able to transmit yellow fever virus.
Euro Surveillance. 2016; 21(39): pii = 30361. https://doi.org/10.2807/1560-7917.ES.2016.21.39.30361
7. European Centre for Disease Prevention and Control and European Food Safety Authority. Mosquito
maps [internet]. Stockholm: ECDC; 2018. Available from: https://ecdc.europa.eu/en/disease-vectors/
surveillance-and-disease-data/mosquito-maps. [cited 2018 Aug 8].
8. Istituto Superiore di Sanita (ISS). Italia: focolai autoctoni di infezione da virus chikungunya. [Italy: auto-
chtonous cases of chikungunya virus]. Rome: ISS; 27 Oct 2017. Italian. Available from: http://www.
salute.gov.it/portale/temi/documenti/chikungunya/bollettino_chikungunya_ULTIMO.pdf. [cited 2018
9. Manica M, Guzzetta G, Poletti P, Filipponi F, Solimini A, Caputo B, et al. Transmission dynamics of the
ongoing chikungunya outbreak in Central Italy: from coastal areas to the metropolitan city of Rome,
summer 2017. Euro Surveillance. 2017; 22(44): pii = 17–00685. https://doi.org/10.2807/1560-7917.ES.
10. Manica M, Filipponi F, D’Alessandro A, Screti A, Neteler M, RosàR, et al. Spatial and Temporal Hot
Spots of Aedes albopictus Abundance inside and outside a South European Metropolitan Area. PLoS
Negl Trop Dis. 2016; 10(6):e0004758. https://doi.org/10.1371/journal.pntd.0004758 PMID: 27333276
11. Manica M, RosàR, Della Torre A, Caputo B. From eggs to bites: do ovitrap data provide reliable esti-
mates of Aedes albopictus biting females? PeerJ. 2017; 5:e2998. https://doi.org/10.7717/peerj.2998
12. Regione Lazio, Ufficio Idrografico e Mareografico. Available from: http://www.idrografico.roma.it/annali/
. [cited 20 Mar 2018].
13. WHO. Managing epidemics: key facts about major deadly diseases. Geneva: World Health Organiza-
tion; 2018. Licence: CC BY-NC-SA 3.0 IGO. ISBN 978-92-4-156553-0 (Yellow Fever—pg 94)
14. Poletti P, Messeri G, Ajelli M, Vallorani R, Rizzo C, Merler S. Transmission potential of chikungunya
virus and control measures: the case of Italy. PLoS ONE. 2011; 6(5):e18860. https://doi.org/10.1371/
journal.pone.0018860 PMID: 21559329
15. Guzzetta G, Montarsi F, Baldacchino FA, Metz M, Capelli G, Rizzoli A, et al. Potential risk of dengue
and chikungunya outbreaks in northern Italy based on a population model of Aedes albopictus (Diptera:
Culicidae). PLoS Negl Trop Dis. 2016; 10(6):e0004762. https://doi.org/10.1371/journal.pntd.0004762
16. Domingo C, Ellerbrok H, Koopmans M, Nitsche A, Leitmeyer K, Charrel RN, Reusken CB. Need for
additional capacity and improved capability for molecular detection of yellow fever virus in European
Expert Laboratories: External Quality Assessment, March 2018. Eurosurveillance. 2018; 23
17. Rezza G. Dengue and chikungunya: long-distance spread and outbreaks in naïve areas. Pathogens
and global health. 2014 Dec 1; 108(8):349–55. https://doi.org/10.1179/2047773214Y.0000000163
18. Bres PLJ. A century of progress in combating yellow fever. Bull World Health Organ 1986; 64: 775–
786. PMID: 3549030
19. Vannice K, Wilder-Smith A, Hombach J. Fractional-Dose Yellow Fever Vaccination—Advancing the
Evidence Base. New England Journal of Medicine. 2018 Jul 11.
20. Guzzetta G, Trentini F, Poletti P, Baldacchino FA, Montarsi F, Capelli G, et al. Effectiveness and eco-
nomic assessment of routine larviciding for prevention of chikungunya and dengue in temperate urban
settings in Europe. PLoS Negl Trop Dis. 2017; 11(9):e0005918. https://doi.org/10.1371/journal.pntd.
0005918 PMID: 28892499
PLOS Neglected Tropical Diseases | https://doi.org/10.1371/journal.pntd.0006970 January 10, 2019 6 / 6