Yellow fever (YF) is a disease caused by yellow fever virus (YFV), a member of the Flaviviridae family. The disease is considered endemic in parts of Africa and South America. Due to its varied clinical presentation and limited surveillance, yellow fever is under-reported: annually, approximately 5,000 cases are reported to WHO (World Health Organization) from Africa and 300 cases from South America but true incidence is believed to be 10–50 fold higher [1]. The natural transmission cycle of YF involves tree-hole breeding mosquitoes and a wide array of monkeys, apes and marmosets [1]. While primate species in Africa rarely develop fatal disease following YFV infection [2], several species of New World monkeys in the Americas are susceptible to severe, fatal YF disease. In the Americas, deaths of susceptible non-human primates are sentinel events that may indicate presence of YFV in a specific geographic location or environment and surveillance for such events is an important tool to prevent human disease [3]. The goal of surveillance for deaths among non-human primates is to provide an early warning of risk of YFV transmission to humans, for rapid implementation of vaccination and prevention strategies. However, there have been few reports on how epizootic surveillance has been used to inform YF vaccination in human populations.

In Brazil, YF caused major urban epidemics and was considered a public health scourge in Brazilian cities until successful vector control and human vaccination strategies eliminated urban transmission in 1942 [4]. However, the existence of a sylvatic cycle of YFV transmission involving non-human primates was described in the 1930s [5]. Since that time, sporadic human cases and outbreaks have resulted from human activity in endemic areas and expansion of viral transmission to previously unaffected regions. The coastal area of Brazil is generally considered free of YF, including states on the northern, southeastern and southern coasts [2]. Mainly in rural areas, the sylvatic cycle has been maintained by transmission among non-human primates and mosquito vectors. Some New World monkeys, e.g. howler monkeys (genus Alouatta) are extremely susceptible to YFV [6] and develop fatal disease, similar to humans [1]. There are records of epizootics characterized by substantial mortality in howler monkeys populations due to this disease [7]–[13].

Surveillance for YF detection in non-human primates was initiated in the state of Rio Grande do Sul, Brazil in 1999, initially as a passive strategy. Active surveillance began in 2001 following detection of YFV in the northwestern part of the state (bordering Argentina) [11], [14], [15] and vaccination of residents in surrounding municipalities. Between December, 2002 and September, 2008, no evidence was found of YFV circulation or of immunity to YFV in blood specimens from more than 200 captured howler monkeys of Alouatta species or evidence of YFV in more than 2,000 mosquitoes (including Haemagogus leucocelaenus) [16]. Seven years after active surveillance began, a rapidly-spreading YF outbreak in the state between October, 2008 and June, 2009 [7], [17]. The coincident YF epizootic in non-human primates was the largest ever documented, with more than two-thirds of specimens from non-human primates testing positive for YFV infection [7], [17].

The YF outbreak provided the opportunity to evaluate surveillance for epizootic events involving deaths of non-human primates. We assessed timeliness of notification of non-human primate deaths in relation to the occurrence of confirmed human cases, vaccine coverage and timing of YF vaccination in municipalities in the state of Rio Grande do Sul during the YF outbreak.

Prior to the beginning of the outbreak in October, 2008, vaccination against YF was recommended in 52 municipalities, with a total population of 531,163. By June, 2009, after eight months of YFV spread, YF vaccination was recommended in 293 municipalities, with a total population of 6.9 million inhabitants (Table 1).

Of the 293 municipalities in which yellow fever vaccination was recommended by July, 2009, 67 (23%) were classified as “affected areas” (based on confirmed YFV circulation during the outbreak period) and 226 (77%) as “expanded areas” (Figure 2). Of these, 61 (21%) municipalities reported monkey deaths without laboratory evidence of YFV circulation and 165 (56%) were adjacent to municipalities with YFV circulation but had no reports of monkey deaths during the outbreak period. In 26 additional municipalities, epizootics were reported but laboratory and environmental investigations did not suggest YFV circulation and vaccination was not recommended.

Between November 30, 2008 and July 5, 2009, vaccination against YF was initiated in 241 municipalities (Figure 3). Among 67 municipalities with confirmed YFV circulation, implementation of YF vaccine recommendations was considered timely in 39 (58%): YF vaccine was recommended before the outbreak in 23 municipalities (11 with vaccination since 2001) while in 16, recommendations were implemented within 2 weeks of the first report of epizootic activity. In 28 (42%) of 67 municipalities with confirmed YFV circulation, vaccine recommendations were implemented more than 2 weeks (range 15 to 138 days) after first reported epizootic activity.

During the 9-month period, a total of 118 suspected human cases of YF were reported from 59 municipalities: 21 (18%) were confirmed based on laboratory and epidemiologic evidence, while YF was discarded in 97 (82%) without laboratory evidence of YF infection. Case-fatality among confirmed YF cases was 43% (9 of 21). None of the confirmed cases had documented history of YF vaccination. All confirmed human cases had been exposed to forested areas and, consequently, to the presence of sylvatic vectors, including Haemagogus leucocelaenus. Enhanced surveillance did not identify the presence of other mosquito vectors, including H. spegazzinii, H. janthinomys or Sabethes sp. Specimens from human cases were collected an average of 10 days (range, 2 to 17 days) after reported onset of symptoms and YFV infection was confirmed by laboratory testing from 13 to 53 days after symptom onset.

Nine municipalities were identified as probable locations of infection for 21 confirmed human YF cases. Of these, deaths of non-human primates were reported in eight (89%) municipalities 12 days to two months prior to onset of illness in persons with confirmed YF (Table 2). In two municipalities in the northwestern part of the state (each the probable location of infection for one human case) in which vaccination had been recommended since 2001, one case occurred in a recently arrived resident and one in a visitor to the area. Although in both cases family members reported being vaccinated, neither case patient was aware of YF vaccine recommendations in the area. In four other municipalities in the northwestern part of the state, YF vaccination was recommended within 7 days of the first report of deaths among non-human primates, before results of laboratory testing confirmed YFV epizootics. In two municipalities in the central part of the state (Santa Cruz do Sul and Vera Cruz), probable locations of infection for 11 human cases, vaccination began more than two months after deaths of non-human primates were first reported. Specimens collected in the municipality following initial investigations tested negative for the presence of YFV (Table 2).

By the end of February, 2009, more than 1.6 million doses of YF vaccine had been administered in municipalities in which vaccine was recommended, with a total population of 1.8 million. Rural areas were prioritized for vaccination. By July, 2009, more than 4.2 million doses had been administered, but the total population in municipalities with YF vaccine recommendation had jumped to 6.9 million, resulting in lower estimates of administrative coverage. Estimated coverage in more populous urban areas lagged behind coverage in rural areas. By the end of 2010, administrative coverage of YF vaccine in the area where vaccination was recommended was greater than 80%. Of 67 municipalities with confirmed YF epizootics during the 2008–09 outbreak, 41 (69%) reported administrative vaccination coverage of 90% or greater, versus 124 (55%) of 226 municipalities without confirmed YF epizootics during the outbreak.

Our analysis of the vaccination response to a YF outbreak in southern Brazil in 2008–2009 highlights challenges of using epizootic surveillance as an early warning system for YF in humans. Delay in detection or confirmation of YF epizootics may have resulted in late initiation of vaccination strategies in several municipalities with documented YFV transmission. Reporting of suspected epizootics and vaccination activities occurred most rapidly in areas where viral circulation had been identified during epizootic activity in 2001 and 2002, where many professionals had received training on YF surveillance. However, the rapid spread of YFV to areas previously free of the virus [7], where YF vaccination was not recommended, overwhelmed public health services, affecting the timeliness of the sampling and laboratory confirmation of YFV circulation. Even when specimens were obtained, laboratory confirmation by reference laboratories of the presence of YFV required a minimum of 13 days. Immunization strategies take time to reach high coverage, and vaccine-induced immunity develops over 7–14 days [1]. Early in an outbreak, information from epizootic surveillance may be most useful to identify populations at highest risk of YFV exposure to prioritize vaccination and other prevention strategies, including personal protection.

In two municipalities that accounted for more than 50% of human cases, detection of YFV in non-human primates was delayed. Despite collection of specimens from non-human primates prior to occurrence of human cases, vaccination was not initiated until later specimens tested positive for YFV. Delayed initiation of vaccination in these municipalities where YF vaccine had not been recommended previously may have contributed to higher numbers of people infected in this area. Prevention of human YF cases requires immediate risk communication and implementation of vaccine recommendations after the first reports of epizootic activity when YF is suspected. Based on epizootic surveillance, YF vaccination was recommended in 165 municipalities in “expanded areas” without any reports of dead monkey sightings, as well as 61 municipalities in which epizootic events were not confirmed to be yellow fever. By defining outlying areas as “at risk”, authorities did not wait to confirm YFV circulation before initiating vaccination activities.

Investigation of reported monkey deaths during the outbreak provided evidence of extensive YF epizootics in non-human primates [7]. In six months, YFV circulation in monkeys spread more than 600 km from west to east, spreading from 2 municipalities with confirmed circulation of YFV to 67 municipalities [7], [17]. Prior to the epidemic, active surveillance for the presence of YFV and other arboviruses among howler monkeys (Alouatta sp.) identified no immunity to YFV in this population [16]. However, active surveillance identified YFV-antibodies in howler monkey captured in November, 2009, several months after the outbreak period (unpublished data). Data from the investigations of epizootic activity during this outbreak provided new information about YFV transmission in susceptible howler monkey populations. Further study is needed to understand how YFV is spread between troops of howler monkeys in areas susceptible to YFV circulation and how environmental factors influence YFV occurrence and spread.

In 2011, following the epidemic, the Rio Grande do Sul state health department in coordination with the Brazilian Ministry of Health recommended YF vaccination in an additional 169 municipalities (Figure 3), leaving only 34 (7%) of 496 municipalities in the state without vaccine recommendation. The decision to extend YF vaccination to previously unaffected municipalities was based on the rapid spread of YFV circulation in 2008/2009, environmental characteristics, population movements and discussions with YF experts and scientific advisory committees. Better communication of YF recommendations to travelers are also needed; two of the human cases were probably infected with YF virus in areas in which vaccination had been recommended since 2001, yet neither was aware of vaccine recommendations.

The purpose of surveillance for YFV infection in non-human primates is to prevent YF cases and deaths among humans. In Rio Grande do Sul, surveillance in non-human primates was unable to predict the 2008–2009 outbreak or the rapid spread of YFV transmission, although it detected YFV circulation in non-human-primates in October, 2008 before human cases occurred. Also, active surveillance prior to the outbreak identified a susceptible population of non-human primates. During the outbreak, surveillance for epizootic activity allowed health authorities to monitor virus spread and informed vaccination strategies. In addition, surveillance for epizootics in non-human primates may have averted human cases by alerting the public to the risk of YF and intensifying vaccination efforts, although the number of human cases prevented through vaccination is unknown. Unfortunately, we had different levels of response according to the preparedness of municipalities and the familiarity of the various regions of the state with this type of emergency. The epidemic of 2008–2009 hit areas of the state in which we had no information on any previous occurrence of YF. Active surveillance for YFV infection in non-human primates in the state also developed capacity by training over 300 health professionals from Rio Grande do Sul and other Brazilian states, leading to improved reporting and investigation of epizootic events. Continued monitoring and evaluation of surveillance for YF in non-human primates in Brazil is essential to improve performance. Communication strategies to inform persons at risk of exposure to sylvatic yellow fever may also encourage reporting of epizootic events, to improve rapid detection and response to future YF outbreaks.