This is an uncorrected proof.
Local vector control and public health agencies in California use the California Mosquito-Borne Virus Surveillance and Response Plan to monitor and evaluate West Nile virus (WNV) activity and guide responses to reduce the burden of WNV disease. All available data from environmental surveillance, such as the abundance and WNV infection rates in Culex tarsalis and the Culex pipiens complex mosquitoes, the numbers of dead birds, seroconversions in sentinel chickens, and ambient air temperatures, are fed into a formula to estimate the risk level and associated risk of human infections. In many other areas of the US, the vector index, based only on vector mosquito abundance and infection rates, is used by vector control programs to estimate the risk of human WNV transmission. We built models to determine the association between risk level and the number of reported symptomatic human disease cases with onset in the following three weeks to identify the essential components of the risk level and to compare California’s risk estimates to vector index. Risk level calculations based on Cx. tarsalis and Cx. pipiens complex levels were significantly associated with increased human risk, particularly when accounting for vector control area and population, and were better predictors than using vector index. Including all potential environmental components created an effective tool to estimate the risk of WNV transmission to humans in California.
The California Mosquito-Borne Virus Surveillance and Response Plan was designed to monitor and evaluate West Nile virus activity in the state and guide vector control and public health responses to reduce the burden of West Nile virus disease. It uses a formula to estimate the risk of human infections based on available environmental surveillance data: the abundance and infection rates in Culex tarsalis and the Culex pipiens complex mosquitoes, the numbers of dead birds, seroconversions in sentinel chickens, and/or ambient air temperatures. We evaluated the effectiveness of this risk level system and compared it to the vector index, a simpler risk estimate based only on mosquito abundance and infection rates, used by many vector control agencies inside the U.S. We found that that the California risk levels were significantly associated with increased reported human disease incidence, particularly when all potential environmental components were included, and better fit the data than vector index.
Citation: Danforth ME, Snyder RE, Lonstrup ETN, Barker CM, Kramer VL (2022) Evaluation of the effectiveness of the california mosquito-borne virus surveillance & response plan, 2009–2018. PLoS Negl Trop Dis 16(5): e0010375. https://doi.org/10.1371/journal.pntd.0010375
Editor: Jason L. Rasgon, The Pennsylvania State University, UNITED STATES
Received: August 31, 2021; Accepted: March 30, 2022; Published: May 9, 2022
This is an open access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.
Data Availability: Mosquito, chicken, and dead wild bird surveillance data can be requested for specific research purposes by submitting a request with details of the proposed project via VectorSurv [https://vectorsurv.org]. Limited, de-identified human surveillance data are publicly available via the California Health and Human Services Open Data Portal: https://data.chhs.ca.gov/dataset/west-nile-virus-cases-2006-present.
Funding: VLK acknowledges support from the Emerging Infections Program Cooperative Agreement (grant CK17-1701) from the U.S. Centers for Disease Control and Prevention. CMB acknowledges support from the Pacific Southwest Regional Center of Excellence for Vector-Borne Diseases funded by the U.S. Centers for Disease Control and Prevention (Cooperative Agreement U01CK000517; https://www.cdc.gov). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
West Nile virus (WNV) is a flavivirus that is the leading cause of mosquito-borne disease in the United States [1,2]. Enzootic transmission of WNV is maintained in a cycle involving wild birds and mosquitoes but there is spillover that can result in human infections [3–5]. The primary mosquito vectors in California are Culex tarsalis and the Cx. pipiens complex [4,6], namely Cx. pipiens and Cx. quinquefasciatus and their hybrids. Cx. tarsalis is found primarily in rural, agricultural areas of the state whereas Cx. pipiens complex is found in more urban areas , with Cx. pipiens in the north, Cx. quinquefasciatus in the south, and both species, as well as an interspecific hybrid, found in the middle . Birds in the order Passeriformes are the primary zoonotic hosts of WNV, and some species, particularly those in the family Corvidae, are regularly infected and found dead from WNV [9,10]. The complex ecology of this disease is heavily influenced by temperature, with higher temperatures resulting in increased WNV transmission due to increased mosquito abundance and a faster extrinsic incubation period [11–14].
While the majority of WNV infections are asymptomatic, 20–30% of infections result in febrile illness, with fewer than 1% of cases developing West Nile neuroinvasive disease (WNND), which has a case fatality-rate close to 10% [3,15–17]. Diagnosis and reporting of neuroinvasive disease are more complete than milder non-neuroinvasive disease due to the severity of disease [18,19]. From 2009 through 2018, 21,869 WNV disease cases were reported in the US, including 12,835 neuroinvasive cases and 1,199 deaths . California reported 4,035 (18%) of all US cases, more than any other state, and disproportionate to its 12% of the nation’s population [21,22]. The majority of California’s cases have occurred in Southern California and the Central Valley [21,23].
The California Department of Public Health (CDPH), local mosquito and vector control agencies, and academic institutions in California developed an organized, comprehensive environmental surveillance program to monitor mosquito-borne viruses, beginning with mosquito testing in 1969 . From its inception, the program has focused on arboviruses endemic to California, initially western equine encephalitis virus (WEEV) and St. Louis encephalitis virus (SLEV), and more recently WNV, which was first detected in the state in 2003 . Vector control agencies conduct routine environmental surveillance by trapping mosquitoes, identifying and tallying the captures, and testing the primary disease vectors for the endemic viruses of WNV, SLEV, and WEEV . Many agencies also collect and test dead wild birds for WNV . Because chickens develop antibodies to WNV, SLEV, and WEEV after exposure but are dead end hosts and do not otherwise show signs of infection , many agencies maintain sentinel chicken flocks from late spring through the fall to monitor arbovirus activity in their region. Ambient air temperature, obtained from NASA’s North American Land Data Assimilation System [28,29], is also monitored. The California Mosquito-Borne Virus Surveillance and Response Plan (“Response Plan”)  provides guidelines for vector control agencies to estimate risk using these surveillance elements . Each is assigned a weight and averaged across an area to provide an overall risk level to estimate the risk of WNV transmission in that particular area at any given time and guide response decisions. The “Response Plan” has been updated annually since its inception and includes a detailed description of the methodology used to calculate transmission risk . The quantitative risk assessment is integrated into California’s Vectorborne Disease Surveillance (VectorSurv) system, an online data management system that supports vector-borne disease surveillance and control. VectorSurv enables local vector control agencies to enter their surveillance data and automatically estimates risk levels for any spatial area with adequate surveillance . In comparison, many vector control programs outside of California rely on the vector index alone, which is a simpler measure based solely on vector abundance and mosquito infection rate, to estimate risk of human infection and trigger additional interventions [33,34].
Although previous analyses have considered the association between the Response Plan system and environmental arbovirus activity [31,35], there has never been a study examining how the risk levels are associated with reported human disease. This analysis seeks to evaluate the performance of the Response Plan for predicting human incidence of WNV disease for the years 2009 through 2018. Also, the system has been evaluated only as a whole, including all available environmental surveillance elements, so we evaluated how well human incidence was predicted when one or more environmental factors were excluded from the analysis. Finally, we sought to compare the predictive performance of the Response Plan, which incorporates many surveillance elements, compared to that of the simpler vector index, which is based only on entomological surveillance.
Analysis of human surveillance data is a routine public health activity and therefore exempt from Institutional Review Board review and approval (Project 2020-072-CDPH).
Human cases of WNV disease in California residents that meet the Council of State and Territorial Epidemiologists’ case definition  are reported to CDPH and subsequently to the United States Centers for Disease Control and Prevention because WNV disease is a nationally notifiable condition . Asymptomatic WNV infections (such as those identified via screening of blood donors) do not fulfill the case definition and were excluded from this analysis. Other inclusion criteria for human WNV disease included disease onset between April 1 and December 31 for the years 2009 through 2018, residence in a county in the Central Valley or Southern California, and residence within the jurisdictional boundaries of a vector control agency that has at least 20,000 residents within its service area (Fig 1).
County data courtesy of U.S. Census Bureau TIGER/line spatial files, 2016 (https://www.census.gov/geographies/mapping-files/time-series/geo/tiger-line-file.2016.html). Vector control agency data courtesy of Mosquito and Vector Control Association of California web map, 2018, (sources: ESRI, USGS, NOAA, TomTom, U.S. Department of Commerce, U.S. Census Bureau, NPS). (https://www.arcgis.com/apps/mapviewer/index.html?webmap=604a0fe9f2b74e98a53b53d192b2ac67).
Environmental surveillance data
Environmental surveillance data, originating from specimens collected by vector control agencies, are stored in an online data management system, the California Vectorborne Disease Surveillance Gateway (VectorSurv). Data for this project were obtained via data request #000032. We included the data elements described below for the period April 1-December 31 of each year from 2009 through 2018.
Local vector control agencies collected adult mosquitoes using a variety of traps, sorted them by species into pools of ≤50 females, and froze samples at -80°C before arbovirus testing. Mosquito abundance was calculated as the mean count of a particular species per trap-night, then divided by the prior 5-year average abundance for the same location. The University of California Davis Arbovirus Virus Research and Training (DART) laboratory conducted most arbovirus testing, although in recent years some vector control agencies began in-house testing programs. Vector control agencies that conducted their own testing passed annual proficiency panels developed by DART and reviewed by CDPH. All mosquito pools were screened for WNV, SLEV, and WEEV primarily by a multiplex real-time (TaqMan) reverse transcriptase-polymerase chain reaction (RT-qPCR) or singleplex RT-qPCR, although a few agencies used rapid commercial antigen-capture assay (RAMP, Rapid Analyte Measurement Platform, Response Biomedical Co., Vancouver, BC, Canada). Mosquito infection rates were calculated per 1,000 mosquitoes. CDPH was notified of dead birds by the public via a WNV call center, an online reporting system, and reports that came through vector control agencies. Carcasses that fulfilled collection criteria and were in acceptable condition were collected by vector control agencies and submitted to DART or tested in-house. Tissue samples and oral swabs were tested by RT-qPCR, RAMP, or VecTOR/VecTest (Medical Analysis Systems, Inc., Camarillo, CA). Dead birds were reported as the number of positive birds in a broad or specific region during the prior 3-month period. Sentinel chickens were obtained each spring by participating vector control agencies and tested to ensure that they had no antibodies for WNV, SLEV, or WEEV. Flocks of 6 to 10 chickens were established at locations specified by vector control agencies and bled biweekly by comb prick from April through November, then tested for WNV antibodies at CDPH. Chicken surveillance was reported as the number of seroconversions within a flock in a broad or specific region. The temperature data were obtained from NASA’s North American Land Data Assimilation System [28,29].
West-Nile virus risk assessment
Each environmental variable was assigned an ordinal number from 1 through 5 based on the observations within a 2-week window, as defined in the risk assessment model in the California Mosquito-Borne Virus Surveillance and Response Plan (Table 1) . The biweekly interval was reset and recalculated every week (e.g. the value from week 19 represented data from weeks 18 and 19, while the week 20 value represented data from weeks 19 and 20). Mosquito abundance and infection rate were calculated separately for Cx. tarsalis and Cx. pipiens complex mosquitoes. For each 2-week interval within a given geographic area that had observations for temperature and at least one other environmental component, all available environmental risk values were then averaged to calculate the overall risk level for Cx. tarsalis and Cx. pipiens complex mosquitoes, ranging between 1 and 5. Risk levels between 1.0 and 2.5 are considered Normal Season, 2.6 to 4.0 Emergency Planning, and 4.1 to 5.0 Epidemic Conditions.
Data were analyzed in R Statistical Software version 4.0.2 , using package glmmTMB version 22.214.171.124  and package DHARMa version 0.4.4 . Because several vector control agencies span multiple counties, all data were aggregated spatially within each unique combination of county and agency, hereafter called vector control areas (VCAs). Each VCA operates independently, making decisions on the extent of their surveillance, which could impact when and where WNV was detected. The outcome, measured as the number of human cases of WNV disease that occurred in a 3-week period within each VCA, was analyzed by negative binomial regression. To compare the predictive value of the full suite of environmental surveillance elements versus reduced subsets typical of local programs with fewer resources, overall risk levels were also calculated by systematically excluding one or more components from calculations. Vector index was calculated by multiplying the mean number of mosquitoes captured per trap night by the mosquito infection rate per 1,000 mosquitoes, by species, for the VCA. Other elements included in all models were: 1) random intercepts for each VCA to allow for variation in baseline WNV disease incidence and differences in surveillance and control practices among VCAs, and 2) the log of the 2017 population  as an offset.
To determine how accurately the Response Plan predicts human cases, models were stratified by Cx. tarsalis and Cx. pipiens complex. To determine which components were essential to the Response Plan, data were used for time periods and locations which had all five risk components from agencies that consistently operate and collect data for sentinel chicken and dead bird programs. Risk levels were compared to the vector index using data from time periods and locations which had both abundance and infection rate data for both Cx. tarsalis and Cx. pipiens complex.
All models were compared using the Akaike information criterion (AIC), where the lowest AIC value indicates the best-fitting model, though AIC differences ≤2 are not considered significant . The best-fitting model for research question was then analyzed for spatial autocorrelation by Moran’s I, using distance measurements based on the centroid for each VCA. To reduce the influence of multiple time points on spatial autocorrelation, all longitude values were shifted by 10,000 km each biweekly interval.
During the study period from 2009 through 2018, 4,123 cases of WNV disease in humans were reported in California. Of those, 3,614 (88%) cases met the inclusion criteria of residing in a Central Valley or Southern California county, within a vector control agency that serves ≥20,000 people, and with symptom onset between April 1 and December 31. There were 15,210 biweekly time periods within VCAs with sufficient environmental surveillance data to calculate risk levels. Observed disease incidence per 100,000 individuals generally increased as the observed overall Cx. tarsalis and Cx. pipiens complex risk levels increased (Figs 2 and 3).
Records from 36 VCAs had data for both Cx. tarsalis and Cx. pipiens complex mosquitoes, resulting in 12,364 observations (S1 Table). When comparing models, including population as an offset and agency as a random effect produced the models with the lowest AIC for predicting human disease incidence using both the Cx. tarsalis and Cx. pipiens complex risk levels (Tables 2 and S2). When controlling for human population and VCA, the incidence of human disease increased 274% for every unit increase in Cx. pipiens complex overall risk level, a risk ratio of 3.74 (95% CI: 3.58–3.90) and 224% for every unit increase in Cx. tarsalis overall risk level, a risk ratio of 3.24 (95% CI: 3.11–3.38). Though models using the Cx. pipiens complex risk level had lower AIC values than those with Cx. tarsalis risk level, their 95% CI had nearly perfect congruence (Fig 4). Both models had significant distance-based autocorrelation (p<0.01). All subsequent comparisons were made using models that included an environmental predictor (risk level or vector index), population as an offset, and VCA as a random effect.
Seven VCAs consistently operated sentinel chicken and dead bird surveillance programs during this time period; these vector control agencies covered all or most of Butte, Los Angeles, Placer, Sacramento, Sutter, Yolo, and Yuba counties. Collectively, these agencies had 1,176 observations of risk levels when all five environmental components were available, including abundance and infection data for both Culex species. Risk levels for Cx. tarsalis and Cx. pipiens complex calculated using all available components were on average better than risk levels that omitted components (Tables 3 and S3). Interestingly, the Cx. tarsalis risk level was a better predictor of human incidence when mosquito infection rate was omitted. Similarly, predicting human incidence by Cx. pipiens complex response level was slightly improved by omitting mosquito abundance. While AIC was slightly lower than the full model for Cx. pipiens complex when only temperature and mosquito infection rate were included, the difference was not significant. Although response level calculations that omitted one variable (i.e., drop out models) were technically stronger than the response levels that included all components for both Cx. tarsalis and Cx. pipiens complex based on AIC, the 95% CI for the full and drop out models did not differ significantly in their ability to predict human incidence (Figs 5 and 6). The best two models for Cx. tarsalis and Cx. pipiens complex had significant distance-based autocorrelation (p<0.01).
Sufficient data were available to calculate vector index values from 34 VCAs, yielding observed values for 6,600 biweekly periods, the same time scale as response levels. Most observations (56.4%) of vector index for Cx. pipiens complex were 0, with a mean of 77.2 (range: 0–25,954), whereas 65.1% of observations of vector index for Cx. tarsalis were 0, with a mean of 40.5 (range: 0–2,709). To reduce the influence of outliers, the data was subset to vector index observations ≤1,000, which left 98% of observations for both Cx. pipiens complex (1,528/1,561) and Cx. tarsalis (1,307/1,330). Models based on a response level had a lower AIC than the associated model based on vector index for the same species. The AIC of the regression model based on Cx. pipiens complex vector index was 1,336 higher than that which was based on response level, while the ΔAIC for Cx. tarsalis vector index model was 1,421 higher than the response level model, indicating that the response level was a better predictor of human disease incidence overall compared to vector index (Figs 7 and 8 and S4 Table). All four models had significant distance-based autocorrelation by the Moran’s I test (p<0.01).
We found that the current California arbovirus surveillance system was positively associated with reported human WNV disease incidence and that relationship was improved after accounting for population size and baseline variation among local vector control areas (VCAs). The latter element encompasses geographic, ecological, and administrative differences across various regions of California. There were no significant differences in predicted human disease incidence when using environmental indicators based on Cx. tarsalis or Cx. pipiens complex vectors. Although there were models that omitted specific elements that more accurately predicted human disease incidence than those that included all possible environmental elements, namely omitting Cx. pipiens complex mosquito abundance or Cx. tarsalis mosquito infection, there was no significant difference in model fit when comparing these to the full model. Including all possible environmental elements in Response Plan risk calculations predicted human WNV disease incidence and also simplifies both data entry and response decisions for vector control agencies. The Response Plan improved predictions of human incidence compared to the vector index for both Cx. tarsalis and Cx. pipiens complex.
There have been numerous studies investigating WNV risk in California, seeking to identify important environmental predictors. Hartley et. al. successfully used two components from the California Vectorborne Disease Surveillance System, temperature and mosquito abundance, to predict a third component, sentinel chicken seroconversions, as a measure of virus transmission . The Dynamic Continuous-Area Space-Time (DYCAST) system used dead bird reports to identify areas at high risk of WNV transmission to humans, with an 81% sensitivity and 91% specificity during an epidemic year, though predictive performance declined in non-epidemic years . Kwan et al. compared how well the Response Plan, vector index, and DYCAST predicted human WNV cases in Los Angeles, California and found that the including surveillance data from bird, mosquito, and temperature sources led to the best risk assessment model for human WNV cases .
When identifying important components of the Response Plan, we used data only from VCAs that had robust dead bird and sentinel chicken programs and had available data monitoring all five possible environmental components. Since data for all Response Plan elements are not collected by most VCAs, only temperature and one other environmental parameter are required to estimate risk level. However, our models showed the additional value when agencies were able to incorporate dead birds and sentinel chickens because they improved predictions of human disease risk. Whenever possible, VCAs should consider including these components, as well as mosquito and temperature data, when feasible in their arbovirus surveillance system. A previous analysis found that in areas with susceptible bird species, surveillance systems based on dead birds reported by the public along with strategic mosquito and sentinel flock monitoring was effective and efficient in estimating risk in terms of both cost and effect .
The vector index is used in many US states for estimating risk of human disease caused by WNV and other arboviruses [33,34,47,48]. Our models showed a significant positive relationship between human incidence and vector index (P < 0.001), when accounting for population and VCA. However, models that used California’s risk level system had a better fit, indicating that they better explained the variability in WNV disease incidence. This is likely attributable to the additional elements included in the Response Plan: temperature, which directly influences viral replication and the rate of transmission, and dead birds and sentinel chickens, which are effective indicators of virus transmission. This study’s primary focus was on the predictive performance of the California response plan’s risk model, including the full range of surveillance elements or subsets thereof. We compared the California risk model’s estimates to the vector index alone as a typical basis for vector control action thresholds, and future studies would be useful to evaluate other ways to synthesize available environmental and surveillance data to predict human disease risk.
One limitation of this study is that our estimates of human disease incidence were based on reported cases of WNV; however, WNV is a vastly underreported disease. WNND is most likely to be diagnosed and reported, due to its severity of disease, but it is estimated that fewer than 5% of non-neuroinvasive disease cases are diagnosed and reported [2,19]. For every neuroinvasive disease case, there are an estimated 30–70 non-neuroinvasive cases, which means the true annual incidence of WNV in California is between 19 and 45 cases per 100,000 people , considerably higher than the observed incidence of 1.1 cases per 100,000 during this time-period . WNV reporting is also likely influenced by additional factors such as access to healthcare and socio-economic status, which could lead to underreporting from some areas. In addition, this analysis focused exclusively on the Central Valley and Southern California, areas with high levels of endemic WNV transmission and correspondingly higher burdens of WNV disease. Excluding areas with lower levels of WNV activity eliminated some data at the lower end of the risk levels. However, by extending the study period to slightly before and after the season when most human WNV infections occur, we have accounted for the full range of conditions under which human cases were likely to occur and those during which transmission risk was low. Human infectious disease cases are reported based on the jurisdiction within which the case-patients reside and does not take into consideration the possibility that these individuals may spend considerable amounts of time outside that jurisdiction for work and/or recreation.
There was spatial autocorrelation within each of our models which limits the interpretability of our study. Public health surveillance data is not randomly collected; VCAs monitor areas where there have been previously high levels of virus activity. In addition, VCAS may increase mosquito trapping and testing in areas where there have been other indicators of viral activity, including dead birds, human cases, and even equine cases. VCAs are often within close proximity to one another and have similar ecologies, so the risk within one VCA can influence the risk within another.
This study aimed to compare the predictive performance of the Response Plan’s risk assessment and the vector index in a relative sense to identify the surveillance components that were most predictive of disease incidence. Based on these analyses, it is not possible to declare that the incidence predicted by a given model is “accurate,” only that it was more accurate than other models considered. Further study would be needed to evaluate predictive performance of the Response Plan and vector index using operationally relevant measures of accuracy, such as the sensitivity or specificity of various action thresholds that could be used as triggers for public health or vector control actions. All models treated human disease risk versus environmental surveillance factors as a simple logarithmic relationship: as response level increased, so did human risk. However, our observed data shows a slight but consistent decline in human incidence following time periods with the highest response level for both Cx. tarsalis and Cx. pipiens complex. This may be due to activities conducted by local vector control agencies in response to these elevated levels which are not accounted for in this analysis, such as the use of adult mosquito control products (adulticides) at the highest response levels, thus lowering human disease risk. As there are no treatments or vaccines available for WNV [49,50], the use of adulticides is one the strongest tools available to combat human disease. During a season with epidemic WNV disease and environmental conditions, Sacramento-Yolo Mosquito and Vector Control spent $700,000 to provide emergency vector control, including adulticiding, covering the county’s 2,570 km2 . With an estimated cost of $33,000 to treat each WNV neuroinvasive disease case, it was more cost effective to adulticide than treat patients if only 15 cases were prevented . More recent estimates from California found a median charge for treating a WNND case of $142,321, making adulticiding an even more cost-effective tool for mitigation of human disease .
Building a WNV disease surveillance program that includes the monitoring of vertebrate hosts and temperature, in addition to mosquito abundance and infection rates, yields a robust system for predicting when and where human cases are most likely to occur. The VectorSurv system then automatically analyzes the data and generates risk levels that are easily interpretable, and each state has the ability to adjust risk thresholds for each environmental surveillance component based on their own analysis . This automation of the Response Plan’s risk assessment allows California’s local vector control and public health agencies to allocate resources optimally to reduce human disease risk, including public education on personal protection and enhanced mosquito surveillance and control efforts.
S1 Table. Counties and vector control agencies included in these analyses.
S2 Table. Comparison of coefficients from predictive models of human WNV disease occurrence, CA, 2009–2018, ranked by Akaike Information Criterion (AIC).
Lower AIC values indicate better fit.
S3 Table. Comparisons of coefficients based on models for different calculations for risk level, ranked by Akaike Information (AIC).
The authors thank the California local mosquito and vector control agencies who collected and submitted environmental samples and provided financial support to testing laboratories, as well as the local public health jurisdictions who investigated and reported human WNV disease cases. We acknowledge the contributions of staff at the Davis Arbovirus Research and Training laboratory at UC Davis who conducted testing and maintain VectorSurv, particularly Ying Fang, Sandra Garcia, and Jody Simpson. The authors also appreciate Tina Feiszli, Leslie Foss, Kerry Padgett, and the other members of the CDPH Vector-Borne Disease Section, who conduct testing, staff the WNV call center, and provide extensive knowledge about WNV in California.
- 1. Rosenberg R, Lindsey NP, Fischer M, Gregory CJ, Hinckley AF, Mead PS, et al. Vital signs: trends in reported vectorborne disease cases—United States and Territories, 2004–2016. MMWR Morb Mortal Wkly Rep. 2018;67(17):496. pmid:29723166
- 2. McDonald E, Martin SW, Landry K, Gould CV, Lehman J, Fischer M, et al. West Nile virus and Other domestic nationally notifiable arboviral diseases—United States, 2018. MMWR Morb Mortal Wkly Rep. 2019;68(31):673–8. pmid:31393865
- 3. Petersen L, Carson P, Biggerstaff B, Custer B, Borchardt S, Busch M. Estimated cumulative incidence of West Nile virus infection in US adults, 1999–2010. Epidemiol. Infect. 2013;141(3):591–5. pmid:22640592
- 4. Reisen WK. Ecology of West Nile virus in North America. Viruses. 2013;5(9):2079–105. pmid:24008376; PubMed Central PMCID: PMC3798891.
- 5. Rochlin I, Faraji A, Healy K, Andreadis TG. West Nile virus mosquito vectors in North America. J Med Entomol. 2019. Epub 2019/09/25. pmid:31549725.
- 6. Goddard LB, Roth AE, Reisen WK, Scott TW. Vector competence of California mosquitoes for West Nile virus. Emerg Infect Dis. 2002;8(12):1385–91. pmid:12498652; PubMed Central PMCID: PMC2738502.
- 7. Bohart RM, Washino RK. Mosquitoes of California. third ed. Berkeley, California: University of California Division of Agricultural Sciences; 1978.
- 8. Kothera L, Nelms BM, Reisen WK, Savage HM. Population genetic and admixture analyses of Culex pipiens complex (Diptera: Culicidae) populations in California, United States. Am. J. Trop. Med. Hyg. 2013;89(6):1154. pmid:23958909
- 9. Reisen WK, Barker CM, Carney R, Lothrop HD, Wheeler SS, Wilson JL, et al. Role of corvids in epidemiology of West Nile virus in southern California. J Med Entomol. 2014;43(2):356–67.
- 10. Foss L, Padgett K, Reisen WK, Kjemtrup A, Ogawa J, Kramer V. West Nile Virus-related trends in avian mortality in California, USA, 2003–12. J Wildl Dis. 2015;51(3):576–88. pmid:25919466.
- 11. Reisen WK, Fang Y, Martinez VM. Effects of temperature on the transmission of west nile virus by Culex tarsalis (Diptera: Culicidae). J Med Entomol. 2006;43(2):309–17. pmid:16619616.
- 12. Danforth ME, Reisen WK, Barker CM. Extrinsic incubation rate is not accelerated in recent California strains of West Nile virus in Culex tarsalis (Diptera: Culicidae). J Med Entomol. 2015;52(5):1083–9. Epub 2015/09/04. pmid:26336222; PubMed Central PMCID: PMC4574603.
- 13. Danforth ME, Reisen WK, Barker CM. The impact of cycling temperature on the transmission of West Nile virus. J Med Entomol. 2016;53(3):681–6. Epub 2016/03/31. pmid:27026160; PubMed Central PMCID: PMC5853677.
- 14. Ciota AT, Matacchiero AC, Kilpatrick AM, Kramer LD. The effect of temperature on life history traits of Culex mosquitoes. J Med Entomol. 2014;51(1):55–62. pmid:24605453
- 15. Mostashari F, Bunning ML, Kitsutani PT, Singer DA, Nash D, Cooper MJ, et al. Epidemic West Nile encephalitis, New York, 1999: results of a household-based seroepidemiological survey. Lancet. 2001;358(9278):261–4. pmid:11498211.
- 16. Sejvar JJ, Marfin AA. Manifestations of West Nile neuroinvasive disease. Rev Med Virol. 2006;16(4):209–24. pmid:16906589.
- 17. Sejvar JJ. The long-term outcomes of human West Nile virus infection. Clin Infect Dis. 2007;44(12):1617–24. pmid:17516407.
- 18. Lindsey NP, Kuhn S, Campbell GL, Hayes EB. West Nile virus neuroinvasive disease incidence in the United States, 2002–2006. Vector-Borne Zoonotic Dis. 2008;8(1):35–40. pmid:18237264
- 19. Lindsey NP, Staples JE, Lehman JA, Fischer M, Centers for Disease C, Prevention. Surveillance for human West Nile virus disease—United States, 1999–2008. MMWR Surveill Summ. 2010;59(2):1–17. pmid:20360671.
- 20. Centers for Disease Control and Prevention (CDC). West Nile virus disease cases and deaths reported to CDC by year and clinical presentation, 1999–2018. 2020. Available from: https://www.cdc.gov/westnile/statsmaps/cumMapsData.html#three.
- 21. Snyder RE, Feiszli T, Foss L, Messenger S, Fang Y, Barker CM, et al. West Nile virus in California, 2003–2018: A persistent threat. PLoS Negl Trop Dis. 2020;14(11):e0008841. Epub 2020/11/19. pmid:33206634.
- 22. California Department of Public Health Vector-Borne Disease Section. Reported incidence of West Nile virus—California, 2019. In: California Department of Public Health, editor. 2019.
- 23. Danforth ME, Fischer M, Snyder RE, Lindsey NP, Martin SW, Kramer VL. Characterizing areas with increased burden of West Nile virus disease in California, 2009–2018. Vector-Borne Zoonotic Dis. 2021. pmid:34077676
- 24. Reeves WC, Milby MM, Reisen WK. 1990. Development of a statewide arbovirus surveillance program and models of vector populations and virus transmission. Pp: 431–458. In: Reeves W.C. (ed) Epidemiology and control of mosquito-borne arboviruses in California, 1983–1987. Sacramento CA. Calif. Mosq. Vector Control Assoc. Inc.
- 25. Reisen W, Lothrop H, Chiles R, Madon M, Cossen C, Woods L, et al. West Nile virus in California. Emerg Infect Dis. 2004;10(8):1369–78. pmid:15496236; PubMed Central PMCID: PMC3320391.
- 26. California Department of Public Health, Mosquito and Vector Control Association of California, University of California. California Mosquito-Borne Virus Surveillance & Response Plan. In: Health CDoP, editor. Sacramento, CA: Vector-Borne Disease Section; 2019.
- 27. Patiris PJ, Oceguera LF 3rd, Peck GW, Chiles RE, Reisen WK, Hanson CV. Serologic diagnosis of West Nile and St. Louis encephalitis virus infections in domestic chickens. Am J Trop Med Hyg. 2008;78(3):434–41. pmid:18337340.
- 28. Xia Y, Ek M, Wei H, Meng J. Comparative analysis of relationships between NLDAS-2 forcings and model outputs. Hydrol Process. 2012;26(3):467–74.
- 29. Xia Y, Mitchell K, Ek M, Cosgrove B, Sheffield J, Luo L, et al. Continental-scale water and energy flux analysis and validation for North American Land Data Assimilation System project phase 2 (NLDAS-2): 2. Validation of model-simulated streamflow. J Geophys Res Atmos. 2012;117(D3).
- 30. California Department of Public Health, Mosquito & Vector Control Association of California, University of California. California Mosquito-Borne Virus Surveillance & Response Plan. In: California Department of Public Health, editor. Sacramento, CA2021.
- 31. Barker CM, Reisen WK, Kramer VL. California state Mosquito-Borne Virus Surveillance and Response Plan: a retrospective evaluation using conditional simulations. Am J Trop Med Hyg. 2003;68(5):508–18. pmid:12812335.
- 32. California Vectorborne Disease Surveillance Gateway [Internet]. 2021. Available from: https://vectorsurv.org.
- 33. Jones RC, Weaver KN, Smith S, Blanco C, Flores C, Gibbs K, et al. Use of the vector index and geographic information system to prospectively inform West Nile virus interventions. J Am Mosq Control Assoc. 2011;27(3):315–9. pmid:22017098
- 34. Chung WM, Buseman CM, Joyner SN, Hughes SM, Fomby TB, Luby JP, et al. The 2012 West Nile encephalitis epidemic in Dallas, Texas. JAMA. 2013;310(3):297–307. pmid:23860988
- 35. Barker CM, Park BK, Melton FS, Eldridge BF, Kramer VL, Reisen WK, editors. 2007 Year-in-Review: integration of NASA’s meteorological data into the California response network. Proceedings and Papers of the Mosquito and Vector Control Association of California; 2008.
- 36. Epidemiologists CoSaT. CSTE Position Statement 14-ID-04: Arboviral Diseases, Neuroinvasvie and Non-Neuorinvasvie 2015 Case Definition. 2014.
- 37. National Notifiable Diseases Surveillance System. In: Centers for Disease Control and Prevention, editor. 2020.
- 38. R Core Team. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; 2020.
- 39. Brooks ME, Kristensen K, van Benthem KJ, Magnusson A, Berg CW, Nielsen A, et al. glmmTMB balances speed and flexibility among packages for zero-inflated generalized linear mixed modeling. The R Journal. 2017;9(2):378–400.
- 40. Hartig F, Lohse L. DHARMa: residual diagnostics for hierarchical (multi-level / mixed) regression models. https://CRAN.R-project.org/package=DHARMa
- 41. California MaVCAo. Mosquito and Vector Control Association of California: 2017 Yearbook. 2017.
- 42. Anderson DR, Burnham KP. Avoiding pitfalls when using information-theoretic methods. J Wildl Manage. 2002:912–8.
- 43. Hartley DM, Barker CM, Le Menach A, Niu T, Gaff HD, Reisen WK. Effects of temperature on emergence and seasonality of West Nile virus in California. Am J Trop Med Hyg. 2012; 86(5):884–94. pmid:22556092
- 44. Carney RM, Ahearn SC, McConchie A, Glasner C, Jean C, Barker C, et al. Early warning system for West Nile virus risk areas, California, USA. Emerg Infect Dis. 2011;17(8):1445–54 pmid:21801622
- 45. Kwan JL, Park BK, Carpenter TE, Ngo V, Civen R, Reisen WK. Comparison of enzootic risk measures for predicting West Nile disease, Los Angeles, California, USA, 2004–2010. Emerg Infect Dis. 2012;18(8):1298–306 pmid:22840314
- 46. Healy JM, Reisen WK, Kramer VL, Fischer M, Lindsey NP, Nasci RS, et al. Comparison of the efficiency and cost of West Nile virus surveillance methods in California. Vector Borne Zoonotic Dis. 2015;15(2):147–55. Epub 2015/02/24. pmid:25700046; PubMed Central PMCID: PMC4340646.
- 47. Bolling BG, Barker CM, Moore CG, Pape WJ, Eisen L. Seasonal patterns for entomological measures of risk for exposure to Culex vectors and West Nile virus in relation to human disease cases in northeastern Colorado. J Med Entomol. 2009;46(6):1519–31. pmid:19960707
- 48. Colborn JM, Smith KA, Townsend J, Damian D, Nasci RS, Mutebi J-P. West Nile virus outbreak in Phoenix, Arizona—2010: entomological observations and epidemiological correlations. J Am Mosq Control Assoc. 2013;29(2):123. pmid:23923326
- 49. Beasley DW. Vaccines and immunotherapeutics for the prevention and treatment of infections with West Nile virus. Immunotherapy. 2011;3(2):269–85. pmid:21322763
- 50. Kaiser JA, Barrett AD. Twenty years of progress toward West Nile virus vaccine development. Viruses. 2019;11(9):823. pmid:31491885
- 51. Carney RM, Husted S, Jean C, Glaser C, Kramer V. Efficacy of aerial spraying of mosquito adulticide in reducing incidence of West Nile virus, California, 2005. Emerg Infect Dis. 2008;14(5):747–54. pmid:18439356; PubMed Central PMCID: PMC2600250.
- 52. Barber LM, Schleier JJ, 3rd, Peterson RK. Economic cost analysis of West Nile virus outbreak, Sacramento County, California, USA, 2005. Emerg Infect Dis. 2010;16(3):480–6. Epub 2010/03/06. pmid:20202424; PubMed Central PMCID: PMC3322011.
- 53. Snyder RE, Cooksey G, Kramer V, Jain S, Vugia DJ. West Nile Virus-associated ospitalizations, California, USA, 2004–2017. Clin Infect Dis. 2020. pmid:32525967
- 54. VectorSurv Development Team. VectorSurv—Vectorborne Disease Surveillance System. 2020. Available from: https://vectorsurv.org/.