Mosquito surveillance and polymerase chain reaction detection of West Nile virus, New York State.

West Nile (WN) virus was detected in the metropolitan New York City (NYC) area during the summer and fall of 1999. Sixty-two human cases, 7 fatal, were documented. The New York State Department of Health initiated a departmental effort to implement a statewide mosquito and virus surveillance system. During the 2000 arbovirus surveillance season, we collected 317,676 mosquitoes, submitted 9,952 pools for virus testing, and detected 363 WN virus-positive pools by polymerase chain reaction (PCR). Eight species of mosquitoes were found infected. Our mosquito surveillance system complemented other surveillance systems in the state to identify relative risk for human exposure to WN virus. PCR WN virus-positive mosquitoes were detected in NYC and six counties in the lower Hudson River Valley and metropolitan NYC area. Collective surveillance activities suggest that WN virus can disperse throughout the state and may impact local health jurisdictions in the state in future years.

West Nile (WN) virus was detected in the metropolitan New York City (NYC) area during the summer and fall of 1999. Sixty-two human cases, 7 fatal, were documented. The New York State Department of Health initiated a departmental effort to implement a statewide mosquito and virus surveillance system. During the 2000 arbovirus surveillance season, we collected 317,676 mosquitoes, submitted 9,952 pools for virus testing, and detected 363 WN virus-positive pools by polymerase chain reaction (PCR). Eight species of mosquitoes were found infected. Our mosquito surveillance system complemented other surveillance systems in the state to identify relative risk for human exposure to WN virus. PCR WN virus-positive mosquitoes were detected in NYC and six counties in the lower Hudson River Valley and metropolitan NYC area. Collective surveillance activities suggest that WN virus can disperse throughout the state and may impact local health jurisdictions in the state in future years.

West Nile Virus
primarily in the Culex genus. LHDs conducted early season larval surveillance and habitat descriptions, initiated habitat management and larval control in habitats where immature mosquitoes were present (focusing on container breeding, storm drain, or catch basin habitats), and developed a local database of these habitats. Adult mosquito surveillance activities could start at any time in the individual counties, and mosquito pools could be submitted to the NYSDOH Arbovirus Research Laboratory beginning the first week of June. The NYSDOH provided CO 2 -baited CDC miniature light traps (14) and gravid mosquito traps (15) to LHDs to conduct adult mosquito surveillance. Anticipating a total statewide weekly submission of 400 mosquito pools (up to 50 mosquitoes/pool), we provided LHDs with a weekly pool allocation and scheduled day of submission for all mosquito pools to be tested for arboviruses.
The department also provided other surveillance supplies that LHDs would need to conduct initial mosquito surveillance activities or to enhance existing programs. We divided the surveillance equipment among the counties on the basis of 1) 1999 surveillance data and recognized geographic distribution of WN virus and 2) the anticipated geographic distribution of WN virus in the metropolitan NYC and lower Hudson Valley area during 2000. We also allocated pools for eastern equine encephalitis surveillance conducted annually in Long Island and central New York. Mosquito pool allocations for 2000 were divided among the LHDs in NYC and the rest of the state on the basis of human population density and distance from the 1999 epicenter. Weekly pool allocations ranged from 90 pools for NYC, 45 pools for each county on Long Island, 40 pools for Westchester County, and 10-15 pools for counties further upstate that were conducting adult mosquito surveillance.
Mosquitoes submitted for virus testing arrived at the Arbovirus Research Laboratory on a weekly basis; some counties submitted specimens midweek and others at the end of the week to split the initial 400-pool weekly load. Mosquito pools contained specimens from a single species (or combined species such as Cx. pipiens and Cx. restuans) collected at a single site during the week. In some cases where numbers were below the target of 50 specimens per pool, collections from 2 sequential weeks or from closely adjacent sites were combined. Although the NYSDOH initially requested separation, it is unlikely that all counties effectively removed all blood-fed or gravid females from weekly collections. Therefore, some pools positive by polymerase chain reaction (PCR) may have contained blood-fed or gravid mosquitoes. Mosquito pools submitted to the Arbovirus Research Laboratory were tested for viral RNA by reverse transcription (RT)-PCR techniques following established protocols and reported as described (16). Briefly, mosquitoes were tested by TaqMan RT-PCR with two primer-probe sets. Virus isolation attempts in Vero cell culture followed the detection of PCRpositive pools, as described (16).

State Analysis of Surveillance Data
Surveillance data related to adult mosquito populations, adult mosquito pool submissions for virus testing, and laboratory test results (as well as bird, animal, and human data) were entered into the NYSDOH Health Information Network, a secure statewide health agency communication network. These surveillance data were maintained on a daily basis and available to NYSDOH and the contributing LHD. Local agency data were held confidential for 24 hours before all other LHDs were provided summary information.
All LHDs were responsible for the entry of mosquito surveillance data in the Health Information Network, as described. All surveillance mechanisms (mosquito, bird, animal, and human) provided data for this secure database. Program and laboratory review of the surveillance database, as well as laboratory entry of virus detection results, provided a current picture of mosquito populations and a virus infection registry of statewide data. When used fully and properly by the LHDs, mosquito population dynamics, rates, and location of arbovirus activity could be derived from regular review of this comprehensive database. All analyses conducted for this report were developed through the summary of the information extracted from that network.

Mosquito Surveillance
Counties conducting adult mosquito surveillance during 2000 along with a summary of their general surveillance efforts are listed in Table 1. Although 26 counties collected and submitted pools of mosquitoes for virus testing (Figure 1), PCR-positive mosquitoes were found only in the five NYC boroughs, both counties on Long Island, and four counties in the lower Hudson River Valley. Figure 2 provides the weekly numbers of mosquitoes collected by genus through the weeks of the season, with a peak of 1,200 pools submitted the week of August 5. Laboratory protocols were initially established at a maximum of 100 mosquitoes per pool, but that figure was reduced to 50 mosquitoes per pool in May to enhance the ability to detect virus.

Virus Surveillance in Mosquitoes
Quantified summary data related to virus distribution by county and by mosquito species are provided in Table 2.

West Nile Virus
We also investigated the contribution of species minimum infection ratios (MIRs) per 1,000 mosquitoes throughout the season. MIRs (based on PCR-positive tests) were evaluated by individual pool sizes for each species and by analysis of weekly and seasonal aggregated species data within and among LHDs. Figures 4 and 5 illustrate the weekly MIRs by species for each genus (Aedes, Ochlerotatus, and Culex) through the adult mosquito surveillance season in counties that had evidence of human or equine cases (8).
(These figures represent the recapitulation of results after all laboratory data were completed.) WN virus infection (confirmed by PCR) in Culex spp. began in early July and continued through the summer, concurrent with human case onsets; weekly MIRs ranged from 1 to 3.5 per 1,000, with a subsequent peak in early October. Although 80 to 100 pools of Aedes or Ochlerotatus were submitted during June and July, the PCR WN virus-positive Oc. japonicus in early July resulted in relatively high genus-specific MIRs. A second peak of elevated Aedes and Ochlerotatus MIRs occurred in late August and September, concurrent with an observed peak of equine cases. Further review of these data from counties where human or equine cases occurred, using MIR and multifactorial analyses, is required.
By the end of the 2000 surveillance season in New York, PCR WN virus-positive specimens were detected throughout the state (Figure 6). While human disease was restricted to four NYC boroughs (Staten Island, Brooklyn, Manhattan, and Queens), PCR evidence of WN virus infection in mosquitoes was detected in all five NYC boroughs, four adjacent metropolitan counties, and two counties further upstate. Evidence of infection in avian hosts, however, was detected throughout the state (except for Chenango County) (17). Despite the submission of 2,660 pools of primary vector species from 17 upstate counties outside the immediate metropolitan area, we detected PCR evidence of WN virus only in Orange (two pools of Oc. japonicus and one each of Cx. pipiens and Cx. pipiens/restuans) and Columbia (one pool of Oc. japonicus) counties. However, MIRs generally reflected the pattern of human cases and dead crow sightings per square mile (17) in three categories: 1) a combination of high number of human cases, dead crow sightings per square mile, and an overall seasonal MIR of 5.27/1,000 tested in Staten Island; 2) a small number of human cases, moderate number of dead crow sightings per square mile, and seasonal MIRs ranging from 0.18 to 2.36/1,000 for counties in and near NYC; and 3) no human cases, low number of dead crow sightings per square mile, and seasonal MIRs ranging from 0 to 0.86/1,000 for upstate counties ( Figure 6). Documented WN virus transmission to humans and horses was limited to a much smaller geographic area of New York State, and, over the last two seasons, has included only NYC, Long Island, Westchester County, and Orange County (roughly 50 miles north of NYC).

Discussion
Review of specific local human and mosquito population data will be addressed in separate publications. Our work summarizes statewide surveillance components and identifies some potential flaws due to the geographic aggregation of data. In addition, surveillance may not have been uniformly applied across all municipalities. However, review of these initial surveillance data will lead investigators to design research projects that will better enable the public health

West Nile Virus
community to answer questions that continue to arise (e.g., regarding virus transmission, host pathogenicity, immunology, interventions, sociologic aspects, and vaccination or treatment issues).
There may be some risk of faulty interpretation in a retrospective review of statewide field and laboratory data. In addition, we have not fully evaluated the potential effect of mosquito control operations on overall mosquito populations. However, these data provide necessary baseline information related to virus ecology and infection dynamics. The detection of MIRs associated with individual mosquito species may help public health agencies and academic institutions to focus on specific species with high levels of vectorial capacity (18).
The surveillance system in New York State during 2000 led to the submission of >300,000 mosquitoes in almost 10,000 pools for arbovirus screening; most were derived from LHDs that had not previously conducted mosquito or arbovirus surveillance. These efforts required regular internal discussions on laboratory capacity, steps that could be taken to improve it, and proper use of laboratory capacity to answer both routine surveillance and research demands on limited resources.
Although there is an inherent procedural lag between mosquito collection and the availability of laboratory results, this initial analysis supports the natural presence of WN virus PCR-positive Culex, Aedes, and Ochlerotatus mosquitoes 2 weeks before onset of human illness. In addition, peak Culex infection during the period of prominent human infection appears to be associated with a peak of PCR WN virus-positive Aedes and Ochlerotatus immediately before and during the onsets of equine cases, which occurred several weeks after the peak of human onsets. However, laboratory confirmation of positivity for all virus surveillance mechanisms (mosquitoes, birds, humans, and horses) usually lagged behind specimen collection or onset dates by several weeks. In addition, infection dates for human and horse cases are likely to precede onset dates by several days to weeks, depending on the individual incubation periods.
Interpretation of the influence of trap type on yield of PCR-positive pools is complicated by the potential for detecting virus that may have been present in the host on which individual mosquitoes may have fed. Ideally, blood-fed or gravid mosquitoes should be separated from unfed females before laboratory testing. Clearly, however, a blood-fed female mosquito one week will become a host-seeking female a week or two later, and full understanding of the transmission dynamics of WN virus may require this separation, along with specific determination of host meals. Data indicate that agencies with limited budgets may be able to identify the local presence and natural circulation of WN virus more efficiently by focusing on gravid trap-based programs.
Initial analyses of MIRs associated with individual mosquito species indicate that certain species, especially Cx. salinarius, Cx. pipiens, Oc. japonicus and Oc. triseriatus, may play major roles in the natural transmission of WN virus among animal hosts. Several of these species (or species complexes) may be involved with an as-yet-unknown reservoir host capable of maintaining the virus in the absence of overt host pathogenicity. Note, however, that no infectious virus was isolated in approximately half of the mosquito pools tested with detectable RNA. WN virus RNA was detected in five pools of Oc. japonicus, for example, but no infectious virus was isolated from these mosquitoes (19). The level of infectious virus may have been too low to be detected and may have increased with further extrinsic incubation, complicating the interpretation of MIRs.
The interrelationships, ecologic niches, and host-meal preferences, for example, of the various Culex species may play a major role in the entire ecology of the virus now that it has been introduced to a completely naive hemisphere. In addition, the MIRs of Culex and other species associated with the apparently bimodal human and equine case distribution may point to a complex mosquito vector relationship not only among the various Culex species, especially Cx. salinarius and Cx. pipiens and perhaps Cx. restuans, but also with other more prominent mammal-feeding species such as Ae. vexans, Oc. triseriatus, and Oc. japonicus. What will happen as this virus continues to spread into other areas of North or Central America where aggressive and more numerous mammalfeeding species exist? How will human disease risks increase if Oc. sollicitans, Cx. tarsalis, Ae. aegypti, or Anopheles species become competent vectors under normal field conditions? Are non-mosquito vectors (such as ticks, mites, or black flies) contributing to the total ecologic cycle (e.g., transmission and overwintering) of this virus in nature? Intervention programs must be established that can account for answers to these questions and maintain our original goals of minimizing the risks for human and other animal infections. Given the recognized ability of WN virus to survive in hibernating mosquitoes in the northeastern United States (20) and the current distribution of the virus in birds, mammals, and mosquitoes along the Atlantic seaboard, public health agencies in affected and neighboring states should be prepared to address the emergence of this virus in their jurisdictions in future years.
Another major goal that public health agencies must face is developing a more timely virus surveillance system. This system should allow detection of the virus in mosquito or bird host systems in sufficient time to permit an appropriate intervention to minimize or prevent further mammal infection. Our current surveillance and laboratory testing systems have inherent time lags that may preclude our ability to intervene against infected adult mosquitoes on a timely basis, before humans are exposed.