Zika beyond the Americas: Travelers as sentinels of Zika virus transmission. A GeoSentinel analysis, 2012 to 2016.

Background Zika virus (ZIKV) was first isolated in Africa; decades later, caused large outbreaks in the Pacific, and is considered endemic in Asia. We aim to describe ZIKV disease epidemiology outside the Americas, the importance of travelers as sentinels of disease transmission, and discrepancies in travel advisories from major international health organizations. Methods and findings This descriptive analysis using GeoSentinel Surveillance Network records involves sixty-four travel and tropical medicine clinics in 29 countries. Ill returned travelers with a confirmed or probable diagnosis of ZIKV disease acquired in Africa, Asia and the Pacific seen between 1 January 2012 and 31 December 2016 are included, and the frequencies of demographic, trip, and diagnostic characteristics described. ZIKV was acquired in Asia (18), the Pacific (10) and Africa (1). For five countries (Indonesia, Philippines, Thailand, Vietnam, Cameroon), GeoSentinel patients were sentinel markers of recent Zika activity. Additionally, the first confirmed ZIKV infection acquired in Kiribati was reported to GeoSentinel (2015), and a probable case was reported from Timor Leste (April 2016), representing the only case known to date. Review of Zika situation updates from major international health authorities for country risk classifications shows heterogeneity in ZIKV country travel advisories. Conclusions Travelers are integral to the global spread of ZIKV, serving as sentinel markers of disease activity. Although GeoSentinel data are collected by specialized clinics and do not capture all imported cases, we show that surveillance of imported infections by returned travelers augments local surveillance system data regarding ZIKV epidemiology and can assist with risk categorization by international authorities. However, travel advisories are variable due to risk uncertainties.

Introduction Zika virus (ZIKV) made global headlines in 2015 when Brazil identified the first cases of confirmed transmission. The alarming association with congenital abnormalities triggered the World Health Organization (WHO) to declare a public health emergency of international concern on 1 February 2016 [1]. Whereas significant attention has focused on the Americas and travel-associated cases exported from the Americas, including exportation of cases reported to GeoSentinel [2], ZIKV was first isolated from a rhesus macaque in 1947 in the Zika Forest in Uganda [3,4] and was responsible for large outbreaks in 2007 in the Pacific Islands of Yap (Federated States of Micronesia) and in 2013 in French Polynesia [5,6]. Based on serological studies, ZIKV has been endemic in certain regions of Africa and of Asia for over 50 years [3,[7][8][9][10][11][12][13][14][15] and has been silently circulating in West Africa for over 20 years [16], although limitations in serological testing and potential cross-reactivity with other flaviviruses hinder reliable interpretation.
ZIKV is now widely distributed throughout the tropics and sub-tropics, with 75 countries and territories reporting evidence of vector-borne ZIKV transmission between 2015 and 2016 including: three in West and Central Africa (Cape Verde, Guinea-Bissau, Gabon), 13 in the Pacific (American Samoa, Fiji, Marshall Islands, Micronesia, Palau, Samoa, Tonga, New Caledonia, Cook Islands, French Polynesia, Papua New Guinea, Solomon Islands, Vanuatu), and 10 in Asia (Indonesia, Maldives, Thailand, Bangladesh, Singapore, Philippines, Vietnam, Cambodia, Malaysia, Lao People's Democratic Republic) [17]. In Africa and Asia-Pacific regions, an estimated 2Á6 billion people live in areas with competent mosquito vectors and climatic conditions are suitable for local transmission of ZIKV [18].
Despite recent advances in knowledge and research regarding ZIKV, a general lack of disease awareness prior to the current outbreak, misdiagnoses due to the clinical similarity of ZIKV infections with other infectious diseases, and limited access to diagnostic testing have resulted in many unresolved questions regarding the distribution and transmission risks outside the Americas. There is also uncertainty regarding the impact of different strains and the potential for obstetric and neurological complications, particularly in Asia and Africa. To date, all reported Pacific and American Zika isolates and associated neurologic or microcephaly complications have been attributed to the expansion of the Asian lineage [19][20][21], but cases of microcephaly have been investigated in Guinea-Bissau and Angola, where the current outbreak is caused by the African lineage [17]. While it is thought that Asian and African ZIKV strains differ in their abilities to infect cells of the central nervous system and to cause neurodevelopmental problems [22], residual uncertainties hinder accurate risk assessment and construction of clear, evidence-based recommendations. Nevertheless, international authorities must provide guidance as a key element of the outbreak response, including advice for international travelers for their own self-protection and in order to limit risks of secondary local transmission on return from travel to areas with the Aedes spp. vector.
To develop a broader understanding of the extent of ZIKV among international travelers from Africa, Asia and the Pacific and the role travelers play in identifying areas of endemicity, we analyzed ZIKV infections reported by the GeoSentinel Surveillance Network and juxtaposed these cases with a review of the published literature and web-based reports of ZIKV cases in select countries outside the Americas. We also examined the level and timing of advisories issued by four major international health authorities: WHO, European Center for Disease Control and Prevention (ECDC), US Centers for Disease Control and Prevention (CDC), and Public Health England's National Travel Health Network and Centre (NaTHNaC).

Methods
GeoSentinel is a global surveillance network established in 1995 (www.istm.org/geosentinel), currently comprising 64 specialized travel and tropical medicine clinics in 29 countries, the majority being in Europe, America, Australia or New Zealand. To be eligible for inclusion in the GeoSentinel database, patients must have crossed an international border and sought medical advice at a GeoSentinel clinic for a presumed travel-related illness. Data collected include: demographic information, travel data, reason for most recent travel, inpatient or outpatient status, travel duration, and time between the onset of symptoms and presentation at a GeoSentinel clinic. Diagnoses are taken from a list of >500 standardized diagnostic codes. Clinics collect de-identified data on ill travelers for whom they have provided clinical care using a standard form, and report these data to a centralized database [23]. The institutional review board officer at CDC's National Center for Emerging and Zoonotic Infectious Diseases classifies the data collection protocol as public health surveillance and not human subjects research.

Data extraction and disease diagnostic criteria
Data on patients diagnosed with ZIKV infection acquired in Africa, Asia or the Pacific (regions defined as per Harvey et al [23]) acquired between 1 January 2012 and 31 December 2016 were extracted from the GeoSentinel database. Sites contributing patient data provided supplemental information including clinical complications and results of diagnostic testing, which was based on local clinician preferences and availability of methods (serology, plaque reduction neutralization test [PRNT], and/or molecular testing including polymerase chain reaction [PCR] and sequencing). Using a modified version of the US Council of State and Territory Epidemiologists (CSTE) case definitions [24] (S1 Table), two clinicians (DHH, MPG) independently classified each patient as confirmed or probable, and discussed discrepancies to reach a final determination. We examined demographic information, itinerary details and reason for travel for each case; country of exposure was based on travel itinerary and relevant incubation periods, with cases excluded if the exposure country could not be ascertained. Geo-Sentinel cases were considered to be sentinel events (index cases) if no other known human case had been reported in web-based or published literature for at least 2 years prior. Data were managed in Microsoft Excel, with reporting of descriptive statistics.
PubMed, ProMED, International SOS reports, WHO situation reports, ECDC situation reports, the CDC website and online news reports were searched (minimum of two coauthors) using the search criteria "Zika" and "Africa or Asia or Pacific", or "Zika" and each individual country name in these three regions to identify locally acquired or exported cases. Data on the status and timing of international travel advisories were tabulated following a search of the relevant authorities' websites, including review of weekly situation reports where relevant [17,[25][26][27].

Results
The first case of ZIKV infection reported to GeoSentinel acquired outside the Americas occurred in a traveler in May 2012 (infection acquired in Indonesia). Since then, two subsequent cases were reported in 2013, five cases in 2014, three in 2015, and 18 cases in 2016 (total of 29 cases, Table 1). Eighteen cases were acquired in Asia, ten in the Pacific, and one in Africa (Fig 1). Twenty-two cases are considered confirmed based on PCR (16 cases), PRNT (3 cases) or serology (5 cases), and seven cases are considered probable. Sixteen cases occurred among males and thirteen among females, one of whom was in the second trimester of pregnancy at the time of exposure and has subsequently given birth to a healthy infant. The median age was 40 years (range: 20-66 years). Fifteen cases occurred among tourists, four among migrant workers, three were travelers visiting friends and relatives, three were missionary workers, two were business travelers, one occurred in a military personnel and one case in an expatriate. Where relevant, the median trip duration to country of exposure was 17 days (range: 6-335 days) and the median time from symptom onset to clinic presentation was 5 days (range: 1-49 days). Twenty-two patients were managed as outpatients. The diagnosis was made at GeoSentinel sites in Asia (8 cases), Western Europe (12 cases), North America (6 cases), Australia/ New Zealand (2 cases), and the Middle East (1 case).
Comparative temporal reporting of GeoSentinel cases with locally reported cases and exported non-GeoSentinel cases in travelers is shown (Figs 2-4, S2 Table). Eight GeoSentinel cases occurred in travelers who were diagnosed with ZIKV infection following a visit to an area early during a recognized outbreak (within first 4-6 weeks: one case in American Samoa  (Table 1) [28,30,32,[34][35][36].
• Sentinel case from Indonesia, exported by an Australian traveler in 2012 [32]. Apart from sero-survey data from the 1970s and 1980s (S2 Table), this was the first case from Indonesia (PCR-positive).
• Sentinel case from the Philippines in 2014, exported by a German traveler. This was the first Zika case in the Philippines since 2012 (not reported previously).  Table]).
• Sentinel case from Vietnam, exported by an Israeli traveler in December 2015 [36]. Apart from sero-survey data from the 1950s (S2 Table), this was the first case from Vietnam (PCRpositive).
• Only known case acquired in Cameroon since 2010, exported by a Belgian traveler in 2015 [28].
• First known case reported from Kiribati, exported by a traveler to New Zealand in 2015 (not previously reported).
• Only known case reported from Timor Leste, exported by a German traveler in April 2016 (not previously reported, albeit probable case only).
Comparisons of risk classifications and travel advisories issued by the four major international authorities are shown in Table 2, and are temporally associated with Zika cases as shown in Figs 2-4.

Discussion
Travelers are key to the global spread of ZIKV [37]. This report emphasizes the vital role travelers play as sentinels, the complementary information systematic traveler surveillance provides to local surveillance data, and the role travelers play in understanding the geographic extent of virus circulation in areas that may otherwise be undetected because of often-asymptomatic infections, limited surveillance, sub-optimal diagnostics, and co-circulation of other arboviruses [2,7,36,41]. When returning home to countries where competent vectors are present, travelers can introduce ZIKV into new areas, thereby also providing a source for propagation and circulation of infection among the local population [18,42]. The GeoSentinel ZIKV case acquired in Indonesia in May 2012 was among the first sporadic cases reported in a traveler [32,36]. In countries with lower access to reliable diagnostic testing, local surveillance systems may have limited capacity to detect endemic ZIKV activity, and patient assessment in well-resourced sites facilitates high-level diagnostic support from reference laboratories. Limited awareness of ZIKV may have resulted in earlier missed cases, but since then travelers have been involved in many additional importations globally, with exportations from Asia, the Pacific and Africa highlighted here since GeoSentinel cases from the Americas were recently reported [2].
The role travel-associated cases, including those reported by GeoSentinel, have played as indicator of local ZIKV infections has varied in different regions. ZIKV in South America, the Caribbean and the Pacific led to clinically apparent local outbreaks, with exported cases in travelers also reported [2]. GeoSentinel data cannot be used to infer risk of infection since there is non-uniformity in the likelihood of presenting to a specialized GeoSentinel clinic according to geographical and other factors, but as of the end of 2016, over 400 cases of ZIKV had been reported to GeoSentinel following travel to the Americas, compared to the 29 cases reported here from other regions. Comprehensive seroprevalence and strain cross-reactivity studies are lacking, but there appears to be lower epidemic potential for ZIKV in most African and Asian countries, with fewer recognized local cases and exported cases not infrequently triggering recognition of ZIKV circulation [16]. For Africa, exported cases in travelers have been reported from Senegal [43] and Cape Verde [33], in addition to the GeoSentinel case from Cameroon [28]. For Asia, we are aware of only a single country (Malaysia) that has reported exported cases but for which there is not a confirmed GeoSentinel case. Moreover, GeoSentinel cases served as the first identification of recent ZIKV activity in all Asian countries with known circulation except for the Maldives and Singapore, highlighting the importance of travelers in understanding ZIKV epidemiology. The lower epidemic potential particularly in Asia is presumed to be because of long-standing lower-level endemic ZIKV circulation, prior exposure to a broad range of flaviviruses and prior immunization with Japanese encephalitis vaccine which potentially provides cross-protection to ZIKV [44]. These factors which may have resulted in higher background ZIKV seroprevalence, greater levels of population-wide immunity, fewer clinical manifestations, and greater difficulties with serological confirmation [7].
During 2016, there were significant differences in risk categorizations and in interpretation of data by the four major international authorities, possibly underpinned by political factors, which created marked variability in country stratifications and non-uniformity in both timing and presence of travel advisories (Figs 2-4, S2 Table). While travel advisories generally presume that the presence of endemic disease and a partially immune population pose less risk to travelers than travel during a recognized outbreak, the impact on transmission risks to nonimmune visiting travelers from varying levels of herd immunity in local communities remains uncertain. In March 2017, country classifications for ZIKV epidemiology were modified by a working group formed by the WHO, CDC and ECDC in order to achieve greater harmonization, with resulting revisions leading to four main categories (Category 1: new or re-introduction and ongoing transmission; Category 2: past or ongoing transmission but not in the new or re-introduction phase; Category 3: interrupted transmission with potential for future    The minimum timeline for determining transition to an interrupted state is 12 months after the last confirmed case, and no cases identified in travelers. For countries with a high capacity for diagnostic testing, consistent timely reporting of diagnostic results, a comprehensive arboviral surveillance system and/or a temperate climate or island setting, the interruption of vector-borne transmission is defined as the absence of ZIKV infection 3 months after the last confirmed case.

Associated travel advisory for categories 1 and 2: WHO
advises pregnant women not to travel to areas with ongoing ZIKV outbreaks because of increased risk of microcephaly and other congenital malformations. Healthcare practitioners advising travelers should: i) Provide travelers to areas with ongoing ZIKV outbreaks with up-to-date advice on reducing infection risks (preventing mosquito bites and practicing safer sex); ii) Advise travelers to practice safer sex or consider abstinence for at least 6 months and not to donate blood for at least 1 month after return; iii) Advise pregnant women not to travel to areas with ongoing ZIKV outbreaks; iv) Advise pregnant women whose sexual partners live in or travel to areas with ongoing ZIKV outbreaks to ensure safer sexual practices or abstain from sex for the duration of their pregnancy

Categories 1 and 2 are associated with travel advisory:
Pregnant women should not to travel to these countries (Continued) transmission; Category 4: competent vector exists but no known past or current transmission, Table 2 [39]). However, each organization handles risk mitigation differently: CDC refers to the four country classification designations by WHO [45], but lists only two broad transmission risk levels: "risk present" and "risk absent" [25]. ECDC has the same four Zika virus epidemiology classifications as WHO, but differentiates the WHO Category 2 countries with current outbreaks, and also shows Category 4 subgroups [40]. NaTHNaC has four risk categories, with risk levels ranked into high, medium, low, or very low risk [27] ( In addition to uncertainties on ZIKV epidemiology, case definitions vary. The current CSTE definitions [24] are more detailed than WHO definitions [46], but two aspects of these definitions that merit attention. First, neither addresses anti-Zika IgG (likely because of assay unavailability in the United States), so in our modified case definitions we have added mention of IgG to both probable and confirmed categories (S1 Table). The more challenging limitation relates to interpretation of cases, such as the importation from Cameroon, for which dengue serology was negative (as were NS1 Ag and PCR), yet Zika IgM and confirmatory PRNT were positive. When routine tests for dengue serology are negative and Zika serology is positive, many centers would not perform PRNT. Furthermore, cases with positive PCR and negative serology are now recognized (such as the GeoSentinel case exported from Thailand to Italy, Table 1), leading to additional complexities with defining optimal testing algorithms. Updated guidelines for global definitions which take into account varying scenarios, optimal testing according to timing of presentation relative to exposure, and assay availability in different countries are needed.
Characteristics of ZIKV transmission can often be best determined in travelers returning from endemic areas, for which the period of exposure may be more easily determined than among local populations. Sexual transmission is best assessed in non-endemic settings where there are no competent mosquito vectors; notably the first detection of sexual transmission occurred in a traveler from Africa [43]. Additionally, investigation of returned travelers has ascertained the maximum reported period between symptom onset and sexual transmission (34-41 days) [47], the longest duration of shedding of infectious virus particles (69 days) [48], and the maximum duration of detection of Zika genomic sequences in semen (188 days) [49]. These data underpin current international guidelines recommendations for avoidance of pregnancy for at least 6 months after potential exposure for males and at least 8 weeks for females.
A limitation of this descriptive report is that surveillance data captured by GeoSentinel include only cases seen at specialized clinics, mostly in developed countries; they do not capture all imported cases and may not be representative of all global travelers. Moreover, Geo-Sentinel is unable to provide denominator data or absolute risk information. However, GeoSentinel's deliberate focus to collate data from highly specialized clinics underscores the utility of assessing disease in travelers as a mechanism for identifying geographic areas where emerging diseases may be occurring. Another limitation is that we cannot entirely exclude diagnosis or (unpublished) recognition of additional earlier ZIKV cases, thereby potentially overstating the value of the GeoSentinel reports as index cases. However, given our extensive review of published data and media sources, we nevertheless assert that GeoSentinel reports provide important additive case information.
In summary, many uncertainties in ZIKV epidemiology, transmission and risk of adverse outcomes remain, and consensus on diagnostic and definition issues is needed. We have described ZIKV cases among travelers visiting Africa, Asia and the Pacific, highlighted the integral role of travelers as sentinels of ZIKV infection, shown that GeoSentinel assists in the identification of these travelers, and have contextualized GeoSentinel cases with other reported (locally acquired or exported) cases. We have also demonstrated the differential impact of travelers as sentinels in Asia, which likely relates to the lower epidemic potential due to prior flavivirus circulation among the population of this region. Our findings also detail the basis for non-uniformity in travel advisories issued by the four major international authorities. The presence of variable travel advisories may create confusion for health practitioners and individual travelers when determining risks, and the small but non-zero potential for causing devastating fetal outcomes must be discussed. Traveler consultations need to account for overall ZIKV risk assessment, residual risk uncertainties and individual risk tolerance.

S1 Table. Definition of clinically suspected and confirmed cases used in the current series (adapted from US Council of State and Territory Epidemiologists Interim Zika virus Disease case definition [CSTE][2, 24]). (DOCX)
S2 Table. Reports of endemic Zika virus (ZIKV) cases, exported cases, and major travel authority notices during the height of the Zika outbreak (to Dec 2016).

(DOCX)
As corresponding author, I confirm that the authors had access to all the relevant data for this manuscript, take responsibility for the accuracy of the analysis, assisted with manuscript preparation approved the manuscript for submission and publication. I also confirm that all authors and agree to the copyright terms.