American Journal of Climate Change
Vol.06 No.01(2017), Article ID:75278,51 pages
10.4236/ajcc.2017.61010

Impacts of Increasing Temperature on the Future Incidence of West Nile Neuroinvasive Disease in the United States

Anna Belova1, David Mills2*, Ronald Hall2, Alexis St. Juliana2, Allison Crimmins3, Chris Barker4, Russell Jones2

1Cognistx, Pittsburgh, PA, USA

2Abt Associates, Boulder, CO, USA

3Climate Change Division, U.S. Environmental Protection Agency, Washington, DC, USA

4School of Veterinary Medicine, University of California, Davis, CA, USA

Copyright © 2017 by authors and Scientific Research Publishing Inc.

This work is licensed under the Creative Commons Attribution International License (CC BY 4.0).

http://creativecommons.org/licenses/by/4.0/

Received: December 15, 2016; Accepted: March 28, 2017; Published: March 31, 2017

ABSTRACT

Multiple studies have identified links between climate and West Nile virus disease since the virus arrived in North America. Here we sought to extend these results by developing a Health Impact Function (HIF) to generate county-level estimates of the expected annual number of West Nile neuroinvasive disease (WNND) cases based on the county’s historical WNND incidence, annual average temperature, and population size. To better understand the potential impact of projected temperature change on WNND risk, we used the HIF to project the change in expected annual number of WNND cases attributable to changing temperatures by 2050 and by 2090 using data from five global climate models under two representative concentration pathways (RCP4.5 and RCP8.5). To estimate the costs of anticipated changes, as well as to enable comparisons with other public health impacts, projected WNND cases were allocated to nonfatal and fatal outcomes, then monetized using a cost-of-illness estimate and the U.S. Environmental Protection Agency’s value of a statistical life, respectively. We found that projected future temperature and population changes could increase the expected annual number of WNND cases to ≈2000 - 2200 cases by 2050 and to ≈2700 - 4300 cases by 2090, from a baseline of 970 cases. Holding population constant at future levels while varying temperature from a 1995 baseline, we estimated projected temperature change alone is responsible for ≈590 and ≈960 incremental WNND cases in 2050 and 2090 (respectively) under the RCP4.5 scenario, and ≈820 and ≈2500 cases in 2050 and 2090 (respectively) for the RCP8.5 scenario, with substantial regional variation. The monetized impact of these temperature-attributable incremental cases is estimated at $0.5 billion in 2050 and $1.0 billion in 2090 under the RCP4.5 scenario, and $0.7 billion in 2050 and $2.6 billion in 2090 under the RCP8.5 scenario (undiscounted 2015 U.S. dollars).

Keywords:

Human Health, Climate Change, Temperature, West Nile Virus, West Nile Neuroinvasive Disease, Economic Impacts

1. Introduction

West Nile virus (WNV) is the most widely distributed arthropod-borne virus in the world and the leading cause of arthropod-borne viral disease in the United States [1] [2] . WNV’s ability to exploit new ecological niches is exemplified by its rapid spread across the Western Hemisphere: after being first detected in the Western Hemisphere in 1999, it was present in much of the Americas by 2005 [3] [4] . The virus is now endemic throughout most of the continental United States, being transmitted between passerine birds and several species of mosquitoes in the genus Culex, with incidental infection of humans during periods of high transmission [1] [5] .

WNV disease is classified as a nationally notifiable health outcome; accordingly, state health agencies are responsible for reporting cases to the Centers for Disease Control and Prevention (CDC) [6] [7] . West Nile disease cases can be distinguished by severity of the patient’s symptoms [6] [7] . Milder cases may produce symptoms (e.g., fever, headache, rash, vomiting) that are indistinguishable from other illnesses [6] [8] , raising questions about the reporting accuracy for these milder WNV expressions because of potential under-reporting and misclassification. In contrast, cases of West Nile neuroinvasive disease (WNND), which occur for less than 1% of people infected with the disease, affect the brain or cause neurologic dysfunction and typically result in a patient’s hospitalization [6] [9] . Because it is unlikely that these WNV patients could or would avoid hospitalization given the severity of their symptoms, there is more certainty in summaries of WNND cases [7] [10] .

Climate change has the potential to alter the geographic distributions of WNV and its vectors (e.g., [11] [12] [13] [14] ). WNV disease outbreaks have been associated with climate variables, including temperature and precipitation, in a number of studies [10] [13] [15] [16] [17] [18] [19] . While the nature and strength of the observed associations have varied in these studies according to the region and lag-times among other factors, above-normal temperatures have been among the most consistent predictors of outbreaks, due in part to the acceleration of viral incubation in mosquitoes and increased mosquito reproduction rates at higher temperatures [19] [20] [21] [22] [23] .

This analysis was undertaken as part of the U.S. Environmental Protection Agency’s (USEPA’s) Climate Change Impacts and Risk Analysis (CIRA) project [24] . CIRA focuses on quantifying the degree to which global greenhouse gas (GHG) mitigation and climate adaptation may reduce climate change-related risks and damages in the United States compared to futures with little or no action across multiple sectors (e.g., human health, infrastructure, water resources). The CIRA framework is designed to enable comparisons of impacts across space, time, and sectors by combining existing quantitative relationship estimates with a consistent set of socioeconomic and climate projection data. This analysis expands CIRA’s sectoral coverage to the health impacts of climate-sensitive, vector-borne disease. Given the WNV reporting accuracy concerns, we focused on relationships that could be used to quantify the future incidence of WNND. Like WNV incidence, WNND incidence has been previously linked to several climate variables, including temperature [10] [18] , precipitation [10] [18] , and drought [18] . We incorporated and expand on the relationship between temperature and the probability of above-average WNND incidence by region developed in [10] to quantify future cases and economic impacts under two climate scenarios.

We generated county-level estimates of the expected annual WNND incidence rate for 2050 and 2090 using temperature data from five global climate models (GCMs) under two representative concentration pathways (RCP4.5 and RCP8.5). We then combined these results with projections of county-level populations to calculate the potential number of WNND cases for 2050 and 2090. To isolate the impact of projected temperature changes, we computed the change in the expected number of WNND cases holding populations constant. Finally, we monetized these climate-attributable WNND effects to express the impact in dollars so that they can be more readily compared with other sectoral impact estimates within the CIRA framework [24] .

2. Materials and Methods

We designed and implemented a health impact assessment model to estimate the effect of projected temperature changes on the future number of WNND cases. Section 2.1 describes the development of the Health Impact Function (HIF), which relates temperature to the expected annual number of WNND cases. Section 2.2 summarizes the data and approach used to project the change in expected annual number of WNND cases in the United States. Section 2.3 describes our approach to monetizing temperature-attributable changes in the number of WNND cases.

2.1. Linking Temperature and Expected Annual Number of WNND Cases in the U.S. Population

The HIF was developed based on the approach, as well as the environmental, WNND case, and population data for 2004-2012, used in [10] . Specifically, we obtained estimates of the model used in [10] that linked a county’s standardized annual temperature to the probability that its year-specific, standardized WNND incidence rate (IR) would exceed a z-score value of 0.5. The model allowed for regional heterogeneity in the effect of temperature on the probability of elevated WNND IR by estimating these relationships separately for 10 climate regions. Notably, these relationships were not originally statistically significant in three of the regions. Because this model involved county-level standardization of temperature and WNND IR, we parameterized the HIF separately for each county with reported cases.

We used four analytical steps to specify a county-specific HIF. First, we used observed county-specific average annual temperature data for 2004-2012, corresponding to years of elevated county-level WNND data, to develop annual average-temperature standardization formulas for each county. Second, we developed region-specific relationships that convert standardized temperature values to the probability that the standardized WNND IR would exceed 0.5. Third, we used historical WNND IR data to compute a county-specific high incidence rate threshold (HIRT), one that corresponded to the standardized WNND IR of 0.5. Fourth, we specified a functional relationship that linked the estimated probability of the WNND IR to exceed a county-specific HIRT and the expected number of WNND cases per person per year, under the assumption that the WNND county-level counts are generated by a Poisson process. Numerical optimization techniques were used to solve for the expected county-level number of WNND cases per person per year, based on the HIRT and the temperature-dependent HIRT exceedance probability. Additional details on this method are presented in Appendix 1.

2.2. Projecting Change in the Expected Annual Number of WNND Cases in the United States

For consistency with the CIRA project modeling framework [23] , we projected the potential change in the expected annual number of WNND cases in the United States between a baseline climate year of 1995 and two future reporting years of 2050 and 2090. The expected annual number of WNND cases for each of these three climate periods was estimated using 20 years of modeled climate data around the reporting year (i.e., 1986-2005 for 1995, 2040-2059 for 2050, and 2080-2099 for 2090). We used the county-level HIFs to integrate the reporting year-specific annual average temperature data and population size estimates.

We obtained future temperature projections from a subset of five GCMs from the full suite of the fifth Coupled Model Intercomparison Project (CMIP5; [25] ): CCSM4, GISS-E2-R, CanESM2, HadGEM2-ES, and MIROC5. These models reflect a large range of variability in climate outcomes observed across the entire CMIP5 ensemble. Each GCM was paired with two RCPs that captured a range of plausible emissions futures. The RCPs, originally developed for the Intergovern- mental Panel on Climate Change’s Fifth Assessment Report, are identified by their approximate total radiative forcing in the year 2100, relative to 1750: 8.5 W/m2 (RCP8.5) and 4.5 W/m2 (RCP4.5). RCP8.5 reflects a future with continued high emissions growth with limited efforts to reduce GHGs, whereas RCP4.5 re- presents a future under a global GHG mitigation regimen. These combinations of GCMs and RCPs, selected for use in the CIRA project and the fourth National Climate Assessment [26] , are used here to support integration and comparison of our results with other impact estimates. Appendix 2 provides additional details regarding the GCM selection process; an overview of the selected models; and processes for producing the relevant, county-level annual temperature measures. Appendix 2 also describes the modeled baseline climate dataset for the years 1986-2005, designated for use with the GCM projections.

All-age, county-level population projections were obtained from the Integrated Climate and Land Use Scenarios (ICLUS) v2.0 [27] for 2010, 2050, and 2090. The 2010 population estimates were used with the modeled baseline climate period for 1986-2005 to provide a more recent representation of the population. The choice to incorporate the ICLUS population projections was also made for consistency with CIRA methods where impact estimates are sensitive to po- pulation estimates.

Projections of WNND cases were created separately for each county and GCM/RCP combination for 1995 (baseline year), and two future reporting years, 2050 and 2090, by applying the county-specific HIF to the 20 annual average temperature estimates for each time period, and multiplying the per-person level of expected annual number of WNND cases by the corresponding county-level population estimate from the ICLUS v2.0 data. Given the uncertainty in choosing a single year to represent temperature conditions in baseline and future years, we calculated the change in the expected annual number of WNND cases for each of the 400 possible combinations of 20 baseline and 20 future years for 2050 and 2090. From initial county-level estimates, we separately computed results by state, region, and nationally for each GCM/RCP using the results of the 400 possible combinations of baseline and future years to define the potential distribution in our average results.

2.3. Monetizing Temperature-Related WNND Cases

We monetized future changes in the expected annual number of WNND cases attributable to rising global temperatures to reflect the potential benefits of climate mitigation to future generations from avoiding these health effects. To isolate the impact of projected changes in temperature, we calculated projected changes in the expected annual number of WNND cases while holding population sizes constant at their future values.

The appropriate economic value per WNND case depends on the case disposition with respect to the patient’s survival (i.e., nonfatal or fatal). For nonfatal outcomes, [28] reported the mean reimbursement for incurred hospital charges for subsets of WNND patients distinguished by their syndromes. Sixty-two patients in this group were determined to have conditions consistent with CDC’s clinical criteria for WNND, including diagnoses of meningitis, encephalitis, or acute flaccid paralysis [8] . The weighted mean hospital reimbursement for these 62 patients was $41,391 after adjusting the original study values using a government price index [29] (values in our paper are in undiscounted 2015 U.S. dollars unless stated otherwise). These hospitalization costs do not account for lost productivity during the hospitalization, related follow-up outpatient costs, or pain and suffering associated with the episode [28] . Thus, this represents a conservative estimate of the value of a nonfatal WNND case. We monetized fatal WNND cases using the following year-appropriate value of a statistical life (VSL) estimates: $12,436,623 for 2050 and $15,182,273 for 2090 [24] [30] .

Cohort studies and national summaries of WNND cases provide information to allocate WNND cases to fatal and nonfatal outcomes (e.g., [2] [28] ). The [28]] study reported 6 of the 62 patients (9.7%) with conditions consistent with WNND died during their initial hospitalization. The mortality rate in this sample contrasts with a 6.5% mortality rate reported in the national summary of 2014 WNND cases, reflecting 87 deaths from among 1347 WNND cases [2] . We applied the lower national 2014 WNND mortality rate to allocate projected WNND cases to fatal and nonfatal outcome categories.

3. Results

3.1. Projected Temperature Increases

Approximately half of all U.S. counties reported at least one WNND case between 2004 and 2012. Differences in future temperatures from 2004 to 2012 are a major contributor to the modeled future expected annual number of WNND cases. Figure 1 summarizes the number of years out of the 20 years modeled for each future reporting period (i.e., 2050 and 2090), in which future temperatures

Figure 1. Summary of years with a substantial difference in projected future average temperature compared to average temperature observed during 2004-2012, among U.S. counties with at least one reported WNND case during 2004-2012.

represent a substantial difference from the baseline observed mean temperature. In the figure, a substantial difference is defined using a z-score threshold value of 0.5, when projected temperatures from a GCM are compared to mean observed temperatures for 2004-2012. Counts of years in Figure 1 reflect results averaged across the five GCMs.

The averaged results across GCMs for 2050 under both RCP4.5 and RCP8.5, and in 2009 under RCP4.5, show that few counties are projected to have more than 4 out of 20 years during which projected temperatures are substantially higher compared to the observed 2004-2012 average temperature, using the 0.5 z-score threshold. However, results for 2090 under RCP8.5 stand out in contrast: even after averaging across the five GCMs, many counties are projected to have four or more future years in which temperatures are substantially different compared to the observed 2004-2012 average temperature. In particular, the results for 2090 under RCP8.5 identify a number of areas (e.g., Gulf of Mexico coast, South Florida, San Francisco Bay) where substantial annual average temperature increases are projected to occur in more than 10 of the possible 20 years. Figure 1 also indicates that there are counties in which future annual average temperatures are not substantially different from the observed 2004-2012 average temperature. However, this does not mean there is no observed temperature change in these counties. Instead, this is a reflection of our incorporating the z-score threshold to identify relatively large temperature changes. In this regard, what is particularly noticeable is the increased frequency in counties over time and the nearly complete lack of counties in 2090 under the RCP8.5 scenario, in which this threshold is not exceeded.

3.2. Projected WNND Cases

Table 1 summarizes our estimates of the expected annual number of WNND cases by climate region and RCP. Each entry in the table is an average of results over the GCMs for the calendar years corresponding to either the baseline period or one of two future climate periods. Table 1 also includes results combining the baseline climate data with future populations to support our economic analyses (results in columns 5 and 6).

Comparing the results in columns 3 and 4 in Table 1 with those in column 2 show that the expected annual average number of WNND cases increase across all climate regions in both future time periods and for both RCPs. Specifically, the results show an expected increase in WNND cases across the nation from nearly 1000 cases in the baseline period to approximately 2000 by 2050 and 2700 by 2090 under RCP4.5, and to approximately 2200 by 2050 and 4,300 by 2090 under RCP8.5. Collectively, this suggests more than a doubling of the anticipated number of annual cases by mid-century relative to the baseline (under either RCP), and a near tripling to quadrupling of the number of annual cases by late- century. Consistent with the temperature differences presented in Figure 1, the largest increase in cases in Table 1 are seen for 2090 under RCP8.5. The nearly 4,300 WNND cases for 2090 under RCP8.5 represent an increase of more than

Table 1. Projections of the expected annual number of WNND cases averaged across GCMs for RCP4.5 and RCP8.5.

a. Regions where [10] did not report a statistically significant result. Totals may not sum due to rounding. b. Totals may not sum due to rounding.

2000 cases from estimates for 2050 (under either RCP) and an increase of nearly 1500 cases relative to 2090 estimates under RCP4.5.

Columns 5-6 in Table 1 provide results combining baseline climates with pro- jected future populations to enable comparisons that isolate the relative importance of the projected temperature changes. These results do not vary by RCP because of the use of the baseline temperature data. Comparing the results in column 3 to those in column 5 shows the impact of temperature changes by 2050 on the expected annual number of WNND cases by holding the affected population constant at its 2050 value. Likewise, comparing results in column 4 to the results in column 6 shows impacts of projected temperature changes by 2090, holding the affected population constant at its 2090 value. The resulting differences in the cases from these comparisons are used to monetize the impact of the projected temperature-related changes on the expected annual number of WNND cases.

In their research, [10] did not find statistically significant associations between temperature and the WNND incidence rate in the Southwest, West, and Northwest regions. Therefore, Table 1 also provides a second set of projected national total case estimates that exclude the projected results for these regions. Removing these regions leads to projections of annual WNND cases of approximately 550 cases in the baseline period, with nearly 800 and 1300 additional WNND cases in 2050 and 2090, respectively, under RCP4.5, and roughly 1000 and 2900 additional cases in 2050 and 2090, respectively, under RCP8.5, while allowing for projected changes in population.

Figure 2 summarizes the results in columns 2, 3, and 4 of Table 1, which reflect impacts of temperature and population over time. Consistent with Figure 1, the estimates of the expected annual average number of WNND cases for 2090 under RCP8.5 in Figure 2 differ considerably from the other estimates. While all regions show increases in the future expected annual number of cases, the results for the Southeast are the most striking: the number of cases grows from fewer than 20 in the baseline to more than 640 in 2090 under RCP8.5. Appendix 3 provides detailed state-level projections (mean and distribution) of expected annual number of WNND cases across all time periods, GCMs, and RCPs, using the projected 2050 population for 2050 estimates and the projected 2090 population for 2090 estimates.

Table 2 summarizes the estimated temperature-related increases in the expected annual number of WNND cases in the United States for 2050 and 2090, along with the potential economic benefits of avoiding these additional cases. In 2050, the monetized impacts of temperature on the expected annual number of WNND cases are approximately $0.5 billion (under RCP4.5) and $0.7 billion (under RCP8.5), across all U.S. regions. In 2090 these impacts increase to $1.0 billion (under RCP4.5) to $2.6 billion (under RCP8.5). These estimates are driven almost entirely by the underlying VSL used to monetize projected fatal WNND cases, as it is nearly three orders of magnitude larger than the value for nonfatal WNND cases. Removing cases from regions where [10] ’s relationships

Figure 2. Projected regional WNND cases by time period and RCP. Populations are consistent with the representative year, and results are averaged over GCMs and calendar year.

were not statistically significant reduces the estimated monetized impacts only slightly.

3.3. Discussion

We projected approximately 590 additional WNND cases per year due to temperature increases by 2050 under the RCP4.5 scenario, with a monetized impact of nearly $0.5 billion. This represents an increase of approximately 40% relative to the annual number of WNND cases expected under baseline temperatures for a 2050 population, and is the most conservative estimated increase in WNND

Table 2. Monetized impact of temperature-related increases in expected annual number of WNND cases for 2050 and 2090 under RCP4.5 and RCP8.5.

a. Case counts are in addition to the 1425 cases projected using the 1995 climate with a 2050 population and the 1736 cases projected using the 1995 climate with a 2090 population (see Table 1). b. Average of results across GCMs and modeled climate years. c. Values are U.S. dollars in year 2015 dollars and are not discounted. d. Excludes three regions (Southwest, West, and Northwest) where temperature-WNND relationships were not statistically significant in [10] .

incidence we modeled. Projected temperature changes for 2050 under the RCP8.5 scenario result in roughly 820 additional WNND cases, or a 60% increase relative to the number of WNND cases expected under baseline temperatures. By 2090, the temperature-related additional WNND incidence is estimated at approximately 960 cases (under RCP4.5) and 2500 cases (under RCP8.5), repre- senting respective increases of roughly 60% and 150%, respectively, relative to the number of WNND cases expected under baseline temperatures for the 2090 population.

There are a limited number of studies that provide a direct basis for comparison with our results. The [18] study reported a near doubling of cases from a baseline estimate while evaluating an ensemble of models using the RCP8.5 scenario for a period centered roughly around 2043. While there are significant differences in approach, the [18] results are consistent with our estimates (60% increase by 2050). Projected increases in the number of cases in our results are also within the bounds of year-to-year changes observed in recent history. For example, national totals for reported WNND cases increased from 486 in 2011 to 2873 in 2012 [7] . This suggests that our modeling reflects changes in projected cases on the order of those currently seen in outbreak years. A limitation to our method and the presentation of our results for future years is the emphasis on presenting impacts averaged over 20 years of observations across 5 climate models for a given RCP, which produced outcomes that muted the signal from particularly severe (i.e., outbreak) years in these future samples. However, there is nothing in our work to suggest a diminished potential for future WNND outbreaks. The plausibility of our results is also indirectly supported by research (e.g., [13] ), concluding that climate change will increase the habitat suitable to support WNV.

A clear limit to our modeled relationship is that we only account for projected changes in temperature, one of a number of factors that can influence WNV incidence [2] [14] [15] [16] [17] . Our modeling could be enhanced by adopting a framework that accounts for both temperature and precipitation. Other factors that may be important for modeling WNND incidence, but could not be accounted for within the scope of this study, include changes in land use characteristics may affect bird, mosquito, and human distributions. However, consistent with the complexity of the WNND transmission cycle, there remains uncertainty in how these climate-sensitive factors may interact, often at different timescales, to affect WNND incidence. By highlighting these issues with respect to precipitation [18] , hypotheses can be described, including, for example, how increased precipitation will increase mosquito abundance by creating breeding habitat or limit it by washing out existing suitable habitat. A related issue is that our modeling does not account for potential shifts in the suitability of habitat, which would support the expansion of WNND into counties excluded from our modeling on the basis of not having any reported cases from 2004 to 2012. The limits of this restriction are highlighted by research (e.g., [13] ) that projects an increase in the habitat suitable for WNV over the 21st century.

Our modeling is also constrained by the difficulty in predicting how human behavior may respond to a changing climate. By extrapolating current statistical associations into the future, we assume that future human behavior patterns and resulting mosquito-biting exposure will vary with temperature in the same way they currently do. As regional temperatures increase, along with possible behavioral changes or modifications to housing and lifestyles, this may be an increasingly tenuous assumption. Similarly, our analysis assumes that the effects of interventions (e.g., mosquito control and public outreach regarding personal protection from mosquito biting) are captured by the original temperature-WNND incidence relationships and that the nature of those relationships will not change over time.

The projected temperature-related increases in the U.S. incidence of WNND are noteworthy considering the number of additional WNND cases, the severity of associated health impacts, and the magnitude of these increases relative to the projected baseline. The monetary impact that ranges from the hundreds of millions to billions of dollars per year, depending on the evaluated future reporting period and scenario, provides additional context for these results. At these levels, incorporating the projected impacts to WNND with other similar impact estimates could make a difference in potential future benefit-cost analyses of the risks and impacts of climate change and proposed mitigation strategies. These differences in monetized impacts between the RCPs over time also highlight some of the benefits that could be realized by adopting strategies consistent with the RCP4.5 scenario, which could mitigate the extent and pace of future climate change. With respect to WNND, these benefits could be equivalent to avoiding hundreds of WNND cases annually by the middle to end of the century.

Acknowledgements

The authors thank the Centers for Disease Control and Prevention staff and ArboNET for granting us access to data on aggregate annual cases of WNND at the county level. The project also greatly benefited from the comments and insight shared with the authors by staff from the Centers for Disease Control and Prevention, particularly Micah Hahn and Rebecca Eisen. USEPA’s Climate Change Division funded this research through contract EP-BPA-12-H-0024 with Abt Associates Inc. The views expressed in this document are solely those of the authors and do not necessarily reflect those of their affiliated institutions, including USEPA.

Cite this paper

Belova, A., Mills, D., Hall, R., Juliana, A.S., Crimmins, A., Barker, C. and Jones, R. (2017) Impacts of Increasing Temperature on the Future Incidence of West Nile Neuroinvasive Di- sease in the United States. American Journal of Climate Change, 6, 166-216. https://doi.org/10.4236/ajcc.2017.61010

References

  1. 1. Kramer, L.D., Styer, L.M. and Ebel, G.D. (2008) A Global Perspective on the Epidemiology of West Nile Virus. Annual Review of Entomology, 53, 61-81.
    https://doi.org/10.1146/annurev.ento.53.103106.093258

  2. 2. Lindsey, N.P., Lehman, J.A., Staples, E. and Fischer, M. (2015) West Nile Virus and Other Nationally Notifiable Arboviral Diseases—United States, 2014. Morbidity and Mortality Weekly Report, 64, 929-934.
    https://doi.org/10.15585/mmwr.mm6434a1

  3. 3. Gubler, D.J. (2007) The Continuing Spread of West Nile Virus in the Western Hemisphere. Clin Infect Dis, 45, 1039-1046.
    https://doi.org/10.1086/521911

  4. 4. Hayes, E.B. and Gubler, D.J. (2006) West Nile Virus: Epidemiology and Clinical Features of an Emerging Epidemic in the United States. Annual Review of Medicine, 57, 181-194.
    https://doi.org/10.1146/annurev.med.57.121304.131418

  5. 5. Reisen, W.K. (2013) Ecology of West Nile Virus in North America. Viruses, 5, 2079-2105.
    https://doi.org/10.3390/v5092079

  6. 6. CDC (2015) West Nile Virus: General Questions about West Nile Virus. Centers for Disease Control and Prevention.
    https://www.cdc.gov/westnile/faq/genquestions.html

  7. 7. CDC (2015) West Nile Virus: Surveillance Resources. Centers for Disease Control and Prevention.
    https://www.cdc.gov/westnile/resourcepages/survresources.html

  8. 8. CDC (2016) National Notifiable Diseases Surveillance System (NNDSS): Arboviral Diseases, Neuroinvasive and Non-Neuroinvasive 2015 Case Definition. Centers for Disease Control and Prevention.
    https://wwwn.cdc.gov/nndss/conditions/arboviral-diseases-neuroinvasive-and-non-neuroinvasive/case-definition/2015/

  9. 9. Beard, C.B., Eisen, R.J., Barker, C.M., Garofalo, J.F., Hahn, M., Hayden, M., Monaghan, A.J., Ogden, N.H. and Schramm, P.J. (2016) Chapter 5: Vector-Borne Diseases. In: Crimmins, A., Balbus, J., Gamble, J.L., Beard, C.B., Bell, J.E., Dodgen, D., Eisen, R.J., Fann, N., Hawkins, M.D., Herring, S.C., Jantarasami, L., Mills, D.M., Saha, S., Sarofim, M.C., Trtanj, J. and Ziska, L., Eds., The Impacts of Climate Change on Human Health in the United States: A Scientific Assessment, U.S. Global Change Research Program, Washington DC, 312 p.
    https://doi.org/10.7930/J0R49NQX

  10. 10. Hahn, M.B., Monaghan, A.J., Hayden, M.H., Eisen, R.J., Delorey, M.J., Lindsey, N.P., Nasci, R.S. and Fischer, M. (2015) Meteorological Conditions Associated with Increased Incidence of West Nile Virus Disease in the United States, 2004-2012. Am J Trop Med Hyg, 92, 13-14.

  11. 11. Gubler, D.J., Reiter, P., Ebi, K.L., Yap, W., Nasci, R. and Patz, J.A. (2001) Climate Variability and Change in the United States: Potential Impacts on Vector- and Rodent-Borne Diseases. Environmental Health Perspectives, 109, 223-233.
    https://doi.org/10.2307/3435012

  12. 12. USGCRP (2016) The Impacts of Climate Change on Human Health in the United States: A Scientific Assessment. In: Crimmins, A., Balbus, J., Gamble, J.L., Beard, C.B., Bell, J.E., Dodgen, D., Eisen, R.J., Fann, N., Hawkins, M.D., Herring, S.C., Jantarasami, L., Mills, D.M., Saha, S., Sarofim, M.C., Trtanj, J. and Ziska, L., Eds., U.S. Global Change Research Program, Washington, DC, 312.

  13. 13. Harrigan, R.J., Thomassen, H.A., Buermann, W. and Smith, T.B. (2014) A Continental Risk Assessment of West Nile Virus under Climate Change. Global Change Biology, 20, 2417-2425.
    https://doi.org/10.1111/gcb.12534

  14. 14. Hoover, K.C. and Barker, C.M. (2016) West Nile Virus, Climate Change, and Circumpolar Vulnerability. WIREs Climate Change, 7, 283-300.
    https://doi.org/10.1002/wcc.382

  15. 15. Chung, W.M., Buseman, C.M., Joyner, S.N., Hughes, S.M., Fomby, T.B., Luby, J.P. and Haley, R.W. (2013) The 2012 West Nile Encephalitis Epidemic in Dallas, Texas. JAMA, 310, 297-307.
    https://doi.org/10.1001/jama.2013.8267

  16. 16. Manore, C.A., Davis, J., Christofferson, R.C., Wesson, D., Hyman, J.M. and Mores, C.N. (2014) Towards an Early Warning System for Forecasting Human West Nile Virus Incidence. PLoS Currents, 6.
    https://doi.org/10.1371/currents.outbreaks.f0b3978230599a56830ce30cb9ce0500

  17. 17. Wimberly, M.C., Lamsal, A., Giacomo, P. and Chuang, T.-W. (2014) Regional Variation of Climatic Influences on West Nile Virus Outbreaks in the United States. The American Journal of Tropical Medicine and Hygiene, 91, 677-684.
    https://doi.org/10.4269/ajtmh.14-0239

  18. 18. Paull, S.H., Horton, D.E., Ashfaq, M., Rastogi, D., Kramer, L.D., Diffenbaugh, N.S. and Kilpatrick, A.M. (2017) Drought and Immunity Determine the Intensity of West Nile Virus Epidemics and Climate Change Impacts. Proceedings of the Royal Society B, 284, Article ID: 20162708.
    https://doi.org/10.1098/rspb.2016.2078

  19. 19. Soverow, J.E., Wellenius, G.A., Fisman, D.N. and Mittleman, M.A. (2009) Infectious Disease in a Warming World: How Weather Influenced West Nile Virus in the United States (2001-2005). Environmental Health Perspectives, 117, 1049-1052.
    https://doi.org/10.1289/ehp.0800487

  20. 20. Danforth, M.E., Reisen, W.K. and Barker, C.M. (2016) The Impact of Cycling Temperature on the Transmission of West Nile Virus. Journal of Medical Entomology, 53, 681-686.
    https://doi.org/10.1093/jme/tjw013

  21. 21. Harbison, J.E., Metzger, M.E., Walton, W.E. and Hu, R. (2009) Evaluation of Factors for Rapid Development of Culex quinquefasciatus in Belowground Stormwater Treatment Devices. Journal of Vector Ecology, 34, 182-190.

  22. 22. Hartley, D.M., Barker, C.M., Le Menach, A., Niu, T., Gaff, H.D. and Reisen, W.K. (2012) Effects of Temperature on Emergence and Seasonality of West Nile Virus in California. The American Journal of Tropical Medicine and Hygiene, 86, 884-894.
    https://doi.org/10.4269/ajtmh.2012.11-0342

  23. 23. Reisen, W.K., Fang, Y. and Martinez, V.M. (2006) Effects of Temperature on the Transmission of West Nile Virus by Culex tarsalis (Diptera: Culicidae). Journal of Medical Entomology, 43, 309-317.
    https://doi.org/10.1093/jmedent/43.2.309

  24. 24. U.S. EPA (2015) Climate Change in the United States: Benefits of Global Action. U.S. Environmental Protection Agency, Washington DC.
    https://www.epa.gov/cira

  25. 25. Taylor, K.E., Stouffer, R.J. and Meehl, G.A. (2012) An Overview of CMIP5 and the Experiment Design. Bulletin of the American Meteorological Society, 485-498.
    https://doi.org/10.1175/BAMS-D-11-00094.1

  26. 26. USGCRP (2017) Fourth National Climate Assessment.
    http://www.globalchange.gov/nca4

  27. 27. U.S. EPA (2016) Updates to the Demographic and Spatial Allocation Models to Produce Integrated Climate and Land Use Scenarios (ICLUS) (Version 2) (External Review Draft). EPA/600/R-14/324. U.S. Environmental Protection Agency, Washington DC.

  28. 28. Staples, J.E., Shankar, M.B., Senjar, J.J., Meltzer, M.I. and Fischer, M. (2014) Initial and Long-Term Costs of Patients Hospitalized with West Nile Virus Disease. The American Journal of Tropical Medicine and Hygiene, 90, 402-409.
    https://doi.org/10.4269/ajtmh.13-0206

  29. 29. BEA (2016) Table 1.1.9. Implicit Price Deflators for Gross Domestic Product. Bureau of Economic Analysis.
    http://bea.gov/iTable/iTable.cfm?reqid=9&step=3&isuri=1&903=13#reqid=9&step=3&isuri=1&904=2000&903=13&906=a&905=2016&910=x&911=0

  30. 30. U.S. EPA (2016) Valuing Mortality Risk Reductions for Policy: A Meta-Analytic Approach. U.S. Environmental Protection Agency, Washington DC.
    https://yosemite.epa.gov/sab/sabproduct.nsf/0/0CA9E925C9A702F285257F380050C842/$File/VSL%20white%20paper_final_020516.pdf

  31. 31. R Core Team (2016) R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria.
    https://www.R-project.org/

  32. 32. Gent, P.R., Danabasoglu, G., Donner, L.J., Holland, M.M., Hunke, E., Jayne, S., Lawrence, D., Neale, R.B., Rasch, P.J., Vertenstein, M. and Worley, P.H. (2011) The Community Climate System Model Version 4. Journal of Climate, 24, 4973-4991.
    https://doi.org/10.1175/2011JCLI4083.1

  33. 33. Neale, R.B., Richter, J., Park, S., Lauritzen, P.H., Vavrus, S.J., Rasch, P. and Zhang, M. (2013) The Mean Climate of the Community Atmosphere Model (CAM4) in Forced SST and Fully Coupled Experiments. Journal of Climate, 26, 5150-5168.
    https://doi.org/10.1175/JCLI-D-12-00236.1

  34. 34. Schmidt, G.A., Ruedy, R., Hansen, J.E., Aleinov, I., Bell, N., Bauer, M., Bauer, S., Cairns, B., Canuto, V., Cheng, Y. and Del Genio, A. (2006) Present-Day Atmospheric Simulations Using GISS Model: Comparison to in Situ, Satellite, and Reanalysis Data. Journal of Climate, 19, 153-192.
    https://doi.org/10.1175/JCLI3612.1

  35. 35. Von Salzen, K., Scinocca, J.F., McFarlane, N.A., Li, J., Cole, J.N., Plummer, D., Verseghy, D., Reader, M.C., Ma, X., Lazare, M. and Solheim, L. (2013) The Canadian Fourth Generation Atmospheric Global Climate Model (CanAM4). Part I: Representation of Physical Processes. Atmosphere-Ocean, 51, 104-125.
    https://doi.org/10.1080/07055900.2012.755610

  36. 36. Davies, T., Cullen, M.J., Malcolm, A.J., Mawson, M.H., Staniforth, A., White, A.A. and Wood, N. (2005) A New Dynamical Core for the Met Office’s Global and Regional Modelling of the Atmosphere. Quarterly Journal of the Royal Meteorological Society, 131, 1759-1782.
    https://doi.org/10.1256/qj.04.101

  37. 37. Collins, W.J., Bellouin, N., Doutriaux-Boucher, M., Gedney, N., Halloran, P., Hinton, T., Hughes, J., Jones, C.D., Joshi, M., Liddicoat, S. and Martin, G. (2011) Development and Evaluation of an Earthsystem Model—HadGEM2. Geoscientific Model Development, 4, 1051-1075.
    https://doi.org/10.5194/gmd-4-1051-2011

  38. 38. Watanabe, M., Suzuki, T., O’ishi, R., Komuro, Y., Watanabe, S., Emori, S., Takemura, T., Chikira, M., Ogura, T., Sekiguchi, M. and Takata, K. (2010) Improved Climate Simulation by MIROC5: Mean States, Variability, and Climate Sensitivity. Journal of Climate, 23, 6312-6335.
    https://doi.org/10.1175/2010JCLI3679.1

  39. 39. Sanderson, B., Knutti, R. and Caldwell, P. (2015) A Representative Democracy to Reduce Interdependency in a Multimodel Ensemble. Journal of Climate, 28, 5171-5194.
    https://doi.org/10.1175/JCLI-D-14-00362.1

  40. 40. Sanderson, B., Knutti, R. and Caldwell, P. (2015) Addressing Interdependency in a Multi-Model Ensemble by Interpolation of Model Properties. Journal of Climate, 28, 5150-5170.
    https://doi.org/10.1175/JCLI-D-14-00361.1

  41. 41. Pierce, D.W., Cayan, D.R. and Thrasher, B.L. (2014) Statistical Downscaling Using Localized Constructed Analogs (LOCA). Journal of Hydrometeorology, 15, 2558-2585.
    https://doi.org/10.1175/JHM-D-14-0082.1

  42. 42. Pierce, D.W., Cayan, D.R., Maurer, E.P., Abatzoglou, J.T. and Hegewisch, K.C. (2015) Improved Bias Correction Techniques for Hydrological Simulations of Climate Change. Journal of Hydrometeorology, 16, 2421-2442.
    https://doi.org/10.1175/JHM-D-14-0236.1

  43. 43. USBR, NCAR, USGS, LLNL, SCU, Climate Analytics Group, CIRES, Climate Central, USACE and Scripps (2016) Downscaled CMIP3 and CMIP5 Climate and Hydrology Projections—Addendum. Release of Downscaled CMIP5 Climate Projections (LOCA) and Comparison with Preceding Information. U.S. Bureau of Reclamation, National Center for Atmospheric Research, Lawrence Livermore National Library, Santa Clara University, Climate Analytics Group, Cooperative Institute for Research in Environmental Sciences, Climate Central, U.S. Army Corps of Engineers, and Scripps Institution of Oceanography.
    http://gdo-dcp.ucllnl.org/downscaled_cmip_projections/

  44. 44. Livneh, B., Bohn, T., Pierce, D., Munoz-Arriola, F., Nijssen, B., Vose, R., Cayan, D. and Brekke, L. (2015) A Spatially Comprehensive, Hydrometeorological Data Set for Mexico, the U.S., and Southern Canada 1950-2013. Scientific Data, 2, Article Number: 150042.
    https://doi.org/10.1038/sdata.2015.42

Appendix 1: Development of the Health Impact Function to Link Temperature and Expected Annual Number of West Nile Neuroinvasive Disease Cases in the U.S. Population

To develop the Health Impact Function (HIF), we first re-estimated the model from [10] using the same information as the authors.

Following this approach, we estimated a logistic regression model that linked a county’s standardized annual average temperature to the probability that the county’s standardized West Nile neuroinvasive disease (WNND) case incidence rate (IR) would exceed 0.5:

(A1.1)

where:

ciy is the number of WNND cases observed in county i during year y;

Niy is the size of the population in county i during year y;

is the WNND IR in county i during year y;

mi is the average WNND IR for county i during 2004-2012;

si is the standard deviation of WNND IR for county i during 2004-2012;

is the standardized WNND IR in county i during year y;

tiy is the annual average temperature in county i during year y;

ai is the average tiy in county i during 2004-2012;

di is the standard deviation of tiy in county i during 2004-2012;

is the standardized annual average temperature in county i during year y;

and are National Oceanic and Atmospheric Administration region- specific coefficient estimates.

Because of the spatial variability in standardization parameters for WNND IR (i.e., mi and si) and annual temperature (i.e., ai and di), we implemented HIF calculations at the county level. Below we describe these calculations for county i and a new annual average temperature value,.

Step 1: Standardize annual average temperature using county-level parameters ai and di as follows:

(A1.2)

Step 2: Estimate probability of a high WNND IR using model coefficient estimates for the climate region that contain county i (i.e., and) and the standardized annual average temperature:

(A1.3)

Step 3: Estimate the county-specific high incidence rate threshold (HIRT): A WNND IR that corresponds to the standardized WNND IR of 0.5:

(A1.4)

(A1.5)

(A1.6)

Step 4: Estimate the expected person-level number annual of WNND cases using the estimates developed in Steps 2 - 3, and assuming that the county- level population is a known fixed value of 1 and that the county-level counts of WNND cases, Ci, is a Poisson-distributed random variable:

(A1.7)

(A1.8)

(A1.9)

(A1.10)

where is the incomplete gamma function, is the floor function, and is the cumulative Poisson density function with an expected mean rate of. We solve Equation (A1.10) for numerically, for each average annual temperature value evaluated in the county using the R base package function optim () [31] . To obtain estimates of the expected annual number of WNND cases in the county, we multiply by the appropriate county population size estimate.

Appendix 2: Rationale for Selection of Climate Models and Process for Generating Meteorological Variables

The selection of a subset of global climate models (GCMs) was necessary due to computational, time, and resource constraints. As such, five GCMs were chosen (Table A2.1) to ensure that the subset captures a large range of the variability in climate outcomes observed across the entire ensemble from the fifth phase of the Coupled Model Inter comparison Project (CMIP5; [25] ).

Variability in Climate Outcomes

While many different metrics could be used in this type of comparison, a logical approach was to compare the projections from CMIP5 GCMs for annual and seasonal temperature and precipitation. While these averaged metrics may not be perfect substitutes for comparing extreme weather effects, the relationship should be sufficiently strong for selecting climate models from the broader ensemble.

The following scatter plots show the variability across the CMIP5 ensemble for projected changes (2071-2100 compared to 1976-2005 baseline) in annual and seasonal (primarily summertime) temperature and precipitation.

As shown in Figures A2.1-A2.3, the five selected GCMs (CanESM2, CCSM4, GISS-E2-R, HadGEM2-ES, and MIROC5) cover a large range of variability across the entire ensemble in terms of annual and seasonal temperature and precipitation. This selection also balances the range alongside considerations of model

Table A2.1. Overview of selected GCMs.

LOCA: Localized Constructed Analogs. SNAP: Scenarios Network for Alaska and Arctic Planning.

Figure A2.1. Variability of projected annual temperature and precipitation change across the CMIP5 ensemble for the contiguous United States.

Figure A2.2. Variability of projected summertime temperature and precipitation change across the CMIP5 ensemble for the contiguous United States.

Figure A2.3. Variability of projected wintertime temperature and precipitation change across the CMIP5 ensemble for the contiguous United States.

independence, broader usage by the scientific community, and skill at reproducing observed climate. [39] [40] provide analysis of both models’ skill at the global scale and independence of underlying code. These criteria were considered in the selection process. Note that a number of GCMs in the scatterplots contain multiple initializations which are designated with numbers in subscript. The dashed lines in the plots represent the median value for each axis.

To provide localized climate projections and to bias correct the projections to improve consistency with the historical period, we used the LOCA dataset [41]] [42] [43] . The LOCA projections, which are derived from the CMIP5 ensemble outputs, are the primary dataset being used in the forthcoming Climate Science Special Report of the U.S. Global Change Research Program’s Fourth National Climate Assessment. The LOCA downscaled dataset provides daily maximum and minimum temperatures (Tmin and Tmax), and daily precipitation values at 1/16-degree resolution from 2006 to 2100. For each climate scenario, we calculated an average daily change factor for temperature and precipitation at each grid cell by comparing 20 years of LOCA projections centered on 2050 and 2090 to an historical 1/16-degree gridded dataset from the 1986-2005 period [44] . We calculated these daily change factors as a spatial average of nine 1/16-degree LOCA grid cells (3 × 3 window) surrounding each location.

We calculated annual average temperature by first averaging daily model-projected changes in Tmin and Tmax to produce a daily average temperature. Annual averages, in a grid cell, were calculated as the average of all daily temperatures over the course of the West Nile neuroinvasive disease case year as defined in [10] (October-September). A county-level value was calculated by averaging the values for all grid cells that intersected a county boundary.

Elsewhere in the manuscript, results from the selected GCMs are referenced with the following abbreviations:

・ CanESM2 (can),

・ CCSM4 (ccs),

・ GISS-E2-R (gis),

・ HadGEM2-ES (had),

・ MIROC5 (mir).

Appendix 3: Detailed State-Level Projections of the Expected Annual Number of West Nile Neuroinvasive Disease Cases across All Time Periods, Global Climate Models, and Representative Concentration Pathways ―With Constant and Varying Population Sizes

Results in the following tables (Tables A3.1-A3.5) for the specific global climate models (GCMs) are presented using the following abbreviations for the full model names:

・ CanESM2 (can),

・ CCSM4 (ccs),

・ GISS-E2-R (gis),

・ HadGEM2-ES (had),

・ MIROC5 (mir).

The mean value reflects the average of modeled results for the given combination of model, representative concentration pathway (RCP), and population while the 2.5% Quantile and 97.5% Quantile values reflect the values from different points in the same distribution of the modeled results.

Table A3.1. Detailed state-level projections of the expected annual number of West Nile Neuroinvasive disease cases across all global climate models for the 2010 population and baseline climate period.

Table A3.2. Detailed state-level projections of the expected annual number of West Nile neuroinvasive disease cases across all global climate models for the 2050 population and 2050 climate period.

Table A3.3. Detailed state-level projections of the expected annual number of West Nile neuroinvasive disease cases across all global climate models for the 2090 population and 2090 climate period.

Table A3.4. Detailed state-level projections of the expected annual number of West Nile neuroinvasive disease cases across all global climate models for the 2010 population and 2050 climate period.

Table A3.5. Detailed state-level projections of the expected annual number of West Nile neuroinvasive disease cases across all global climate models for the 2010 population and 2090 climate period.