USF Libraries
USF Digital Collections

Evaluation of the current State of Florida West Nile Surveillance Program as a predictor for control and prevention of h...

MISSING IMAGE

Material Information

Title:
Evaluation of the current State of Florida West Nile Surveillance Program as a predictor for control and prevention of human West Nile diseases
Physical Description:
Book
Language:
English
Creator:
Butler, Angela E
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla.
Publication Date:

Subjects

Subjects / Keywords:
avian
sentinel
surveillance
vector
West Nile
arboviruses
Dissertations, Academic -- Public Health -- Masters -- USF   ( lcsh )
Genre:
government publication (state, provincial, terriorial, dependent)   ( marcgt )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Summary:
ABSTRACT: West Nile is an important novel virus in the United States, having spread rapidly since it was first detected in New York in 1999. The Centers for Disease Control and Prevention as well as many State Health Departments, have mandated programs for surveillance of West Nile Virus activity. These programs incorporate many different aspects including existing arboserology programs with additional testing for West Nile Virus and new plans that incorporate active and passive surveillance methods. The objective of this study was to examine all aspects of the Florida West Nile surveillance program to determine if there was transmission in the animal systems prior to human cases. The predictive analyses were done using regional data graphs, spatial information, correlations and regression models.Data for sentinel chickens, bird necropsy and mosquito pool surveillance from participating counties in Florida were obtained from the State of Florida surveillance database. The human data was obtained from the State of Florida reportable disease database for each county whether participating in the state surveillance programs or not. Clinical cases were examined by demographics (gender and age) and an incidence rate was calculated to demonstrate the effects of disease. Specific statistical methods used included Pearson's coefficient correlation, Poisson distribution regression modeling to show if any of the surveillance systems were predictors for human disease. The incidence rate analysis for clinical cases showed clustering of cases in adjacent counties within a region where Florida's panhandle and adjacent counties northeast had the highest incidence. Florida's central and southern regions had moderate human incidence.This provides useful information in transmission geography for prevention and control measures. Demographic analysis showed that there were twice as many males than females diagnosed with West Nile in Florida, this was true across the groups as well. The highest number of cases was seen within the age group over 55 years of age for West Nile Neuroinvasive Disease and for West Nile Fever the highest number of cases was within the 36-54 age range. The temporal distribution was determined using graphical representations of all of the surveillance types and clinical cases. In order to include all relevant data, the temporality was set from week 20 to week 52. This study found that all of the surveillance types (dead birds, mosquitoes and sentinels) offered a specialized strength for predicting clinical cases. However, mosquitoes proved to be the least efficient out of the three surveillance systems.The regional and spatial analysis showed that positive dead birds and sentinels provided the coverage for the surveillance systems in the state. However, Pearson's correlation coefficient was low for sentinel surveillance; this may be due to higher participation showing West Nile Virus activity in areas (especially rural) that have no reported human cases. This analysis did show that West Nile is detected in mosquito pool samples before it is detected in the dead bird or sentinel surveillance systems which provides an earlier warning for human cases. The Poisson distribution regression model was only useful for the pooled years and 2003. These showed that mosquitoes, positive dead birds and sentinels were good predictors for clinical cases for the combined years and dead birds and sentinels were significant for 2003 as well.The recommendations based on the results from this study would be to continue all the current surveillance efforts but with the following enhancements: 1. Increase the coverage and consistency of submissions for all surveillance types. 2. Set standard levels of participation for all counties based on the regional analyses and populations at risk. 3. Create standardized approaches for sampling, shipping and submitting samples (especially for mosquito pool submissions) and require that participating counties adhere to these standards. 4. Only submit specific birds known to be especially susceptible to West Nile Virus (e.g. corvids). 5. Targeted prevention and education strategies for higher risk groups based on their potential levels of exposure.
Thesis:
Thesis (M.S.P.H.)--University of South Florida, 2004.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: World Wide Web browser and PDF reader.
System Details:
Mode of access: World Wide Web.
Statement of Responsibility:
by Angela E. Butler.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 185 pages.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001498286
oclc - 57724411
notis - AJU6891
usfldc doi - E14-SFE0000578
usfldc handle - e14.578
System ID:
SFS0025269:00001


This item is only available as the following downloads:


Full Text

PAGE 1

Evaluation Of The Current State Of Florid a West Nile Surveillance Program As A Predictor For Control And Prevention Of Human West Nile Diseases by Angela E. Butler A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Public Health Department of Epidemiology College of Public Health University of South Florida Co-Major Professor: Lillian M. Stark, Ph.D. Co-Major Professor: Aurora Sanchez-Anguiano, M.D., Ph.D. Getachew Dagne, Ph.D. Roger Sanderson, M.A. Date of Approval: November 19, 2004 Keywords: arboviruses, west nile, su rveillance, vector, sentinel, avian Copyright 2004 Angela E. Butler

PAGE 2

Acknowledgments I have had the privilege of working with the finest group of individuals including the College of Public Health faculty and students and the Arbosurveillance group of the Florida Department of Health, Tampa Branch Bureau of Laboratory, all of which helped immensely with this project. I am especially grateful to my committee for their guidance, assistance and patience through out this study. I would also like to thank Dr. Stark for her insight and her extensive knowledge about “everything”. I would like to thank Dr. Sanchez and Dr. Dagne for agreeing to fill positions on my committee on short notice. I want extend my appreciation to the following individuals that assisted me: Jazmine Mateus, Brenda White, Lisa Bowman and Jennifer Gemmer for their constant support and hard work at FDOH Tampa Bran ch Laboratory, Jerrold Scharninghausen for always being brutally honest, Christy Voakes and Brenda Brennan for paving the way in West Nile research at the CO PH and Maribel Casteneda & Rita Judge for all their efforts in the arboserology department at FDOH Tampa Branch Laboratory. I greatly appreciate Robin Krueger without whom this projec t would never have been completed. Most importantly I would like to thank my family the Boles, the Butlers, the Bakers and the Smiths who always encouraged me to continue my education.

PAGE 3

i Table of Contents T UList of TablesU T.....................................................................................................................iii T UList of FiguresU T....................................................................................................................iv T UList of Symbols and AbbreviationsU T.....................................................................................x T UAbstractU T..............................................................................................................................x i T UIntroductionU T.........................................................................................................................1 T UPublic Health SurveillanceU T..............................................................................................1 T UEvaluation of the Arbosurveillance System in FloridaU T...................................................4 T UDescription of Health Related EventU T..........................................................................4 T UDistributionU T.................................................................................................................4 T UTransmissionU T...............................................................................................................9 T UClinical PresentationU T.................................................................................................10 T UFlorida’s Surveillance SystemU T..................................................................................13 T UFlorida’s Arbosurveillance Response PlanU T...............................................................18 T UDefining the StakeholdersU T.........................................................................................21 T UPurpose and Objectives of SystemU T............................................................................22 T UEvaluation DesignU T.........................................................................................................22 T UMaterials and MethodsU T......................................................................................................24 T UAnalysisU T.........................................................................................................................26 T UResultsU T............................................................................................................................... 31 T UDemographic AnalysisU T..................................................................................................31 T UGenderU T.......................................................................................................................31 T UAge GroupsU T...............................................................................................................33 T UWest Nile Incidence RatesU T........................................................................................36 T USurveillanceU T...................................................................................................................39 T UTemporal DistributionU T...............................................................................................39 T UClinical CasesU T............................................................................................................42 T USentinel Chicken SurveillanceU T..................................................................................44 T UAvian SurveillanceU T....................................................................................................46 T UMosquito SurveillanceU T..............................................................................................53 T URegional SurveillanceU T...................................................................................................58 T UPanhandle RegionU T.....................................................................................................58 T UNorthern RegionU T........................................................................................................62 T UCentral RegionU T..........................................................................................................65 T USouthern RegionU T........................................................................................................68 T UClinical Cases by RegionU T..........................................................................................71 T USpatial AnalysisU T............................................................................................................73

PAGE 4

iiT UPositive Dead Birds and Clinical CasesU T....................................................................73 T UPositive Mosquito Pools and Clinical CasesU T.............................................................84 T USentinel Chicken Serconversion Rates and Clinical CasesU T......................................95 T UMultivariate Poisson Regression ModelU T.................................................................105 T UDiscussionU T.......................................................................................................................108 T UDemographic AnaylsisU T................................................................................................108 T UTemporal DistributionU T.................................................................................................109 T UPeak TransmissionU T......................................................................................................111 T URegional AnalysisU T.......................................................................................................114 T USpatial AnalysisU T..........................................................................................................118 T UPoisson Distribution Regression ModelU T......................................................................119 T UStudy Strengths and LimitationsU T.....................................................................................121 T UEvaluationU T...................................................................................................................122 T UReferences CitedU T.............................................................................................................124 T UBibliographyU T...................................................................................................................129 T UAppendicesU T......................................................................................................................132 T UAppendix I: Table of the Species of Birds that were found Positive for West Nile ……. Virus in the United States since 1999U T................................................134 T UAppendix II: Table of the Sp ecies of Mosquitoes that were found Positive for ……………… West Nile Virus in Mosquito Pools in the United States since 1999U T.142 T UAppendix III: Species Specific We st Nile Virus Reservoir Competence …….………… Index Values U T.....................................................................................143 T UAppendix VI: 2001 Data by CountyU T.........................................................................144 T UAppendix V: 2002 Data by CountyU T.........................................................................146 T UAppendix VI: 2003 Data by CountyU T.........................................................................148 T UAppendix VII: Number of Sentinel Chic ken Sites and Chickens per Site by YearU T...150 T UAppendix VIII: County Arbosurveillance TableU T........................................................151 T UAppendix IX: Regional Map of Florida Number ed by Location from East (Right) ………………. to West (Left) and North (Top) to South (Bottom)U T..........................153 T UAppendix X: Incidence Rate and Nu mber of Clinical Cases per County by YearU T154 T UAppendix XI: Graphs of Pooled Annual Aver ages (2001-2003) with Moving ………………... Averages for West Nile Surveillance Data Statewide and Regional ………………... per WeekU T.........................................................................................155 T UAppendix XII: PROC GENMOD Poisson Regression SAS OutputU T.......................165

PAGE 5

iii List of Tables T UTable 1.U T T UMonth of First Sera Submitted for Counties Participating in Sentinel Surveillance for 2001, 2002 and 2003 Organized by Region.U T....................117

PAGE 6

iv List of Figures T UFigure 1.U T T UThe Worldwide Geographic Distribu tion for the Serocomplex of the Family Flaviridae as of 2000.U T.....................................................................6 T UFigure 2a.U T T UWest Nile Virus Distribution acro ss the United States from 1999 -2002.U T..7 T UFigure 2b.U T T UWest Nile Virus Distribution ac ross the United States for 2003.U T...............8 T UFigure 3. U T T UThe Transmission Cycle for West Nile Virus.U T..........................................10 T UFigure 4. U T T UDiagram of the State of Florida’s Sp ecific Response Plan for Arbovirus Detection.U T..................................................................................................19 T UFigure 5.U T T UTotal Number of West Nile Cases and Gender for 2001-2003.U T.................32 T UFigure 6.U T Cumulative Number of WN Cases for Gender by Year............................32 T UFigure 7.U T T UWNND by Age Group for 2001-2003.U T.......................................................34 T UFigure 8.U T T UAge Comparison for WNND and WNF for 2003.U T......................................34 T UFigure 9.U T T UWNND by Gender and Age Group for 2001-2003.U T....................................35 T UFigure 10a.U T T UWest Nile Incidence Rate by County, Region and Location for 2001.U T.......37 T UFigure 10b.U T T UWest Nile Incidence Rate by County, Region and Location for 2002.U T.......37 T UFigure 10c.U T T UWest Nile Incidence Rate by County, Region and Location for 2003.U T.......38 T UFigure 11.U T T UWest Nile Clinical Statewide Incidence Rate for 2001 through 2003.U T........38 T UFigure 12. Sentinel Chicken Positive We st Nile Weekly Seroconversion Rate by Week for 2001 – 2003.U T.................................................................................40 T UFigure 13.U T T UWest Nile Clinical Cases by Week for 2001—2003.U T.................................40 T UFigure 14.U T T UTotal Number of West Nile Positive Dead Bi rds by Week for 2001 2003U T...............................................................................................41 T UFigure 15.U T T UTotal Number of West Nile Positive Mosquito Pools by Week for 2001 2003.U T..............................................................................................41

PAGE 7

vT UFigure 16.U T T UNumber of Clinical Cases of We st Nile per Week during peak Transmission Season.U T................................................................................43 T UFigure 17.U T T UBox-plot for the Average and Total Number of Clinical Cases for 2001, 2002 and 2003.U T..........................................................................................43 T UFigure 18.U T T USentinel Chicken Surveillance Rate during Transmission Season for 2001 2003.U T..............................................................................................45 T UFigure 19.U T T UBox-plot of Sentinel Surveillance rate for 2001, 2002 and 2003.U T............45 T UFigure 20a.U T T UTotal Number of West Nile Positi ve Dead Birds during Transmission Season for 2001 – 2003.U T............................................................................49 T UFigure 20b. U T T UDead Birds Percent Positive.U T.....................................................................49 T UFigure : 21a.U T T UThe Total Dead Birds Submitted and the Percent of Positive Dead Birds by Week for 2001.U T...........................................................................50 T UFigure : 21b.U T T UThe Total Dead Birds Submitted and the Percent of Positive Dead Birds by Week for 2002.U T..........................................................................50 T UFigure : 21c.U T T UThe Total Dead Birds Submitted and the Percent of Positive Dead Birds by Week for 2003.U T...........................................................................51 T UFigure 22.U T T UBox-plot the Number of Positive Dead Birds for 2001, 2002 and 2003.U T..........................................................................................................51 T UFigure 23.U T T UTotal Dead Birds Submitted for 2001 – 2003.U T..........................................52 T UFigure 24.U T T UBox-plot for Total Submitted Dead Birds for 2001, 2002 and 2003.U T.......52 T UFigure 25.U T T UTotal Positive Mosquito Pools during Tran smission Season for 2001 – 2003.U T..............................................................................................55 T UFigure 26.U T T UPercent of Positive Mosquito Pools during Transmission Season for 2001 – 2003.U T..............................................................................................55 T UFigure 27.U T T UBox-plot for Total Positive Mosquito Pools for 2001, 2002 and 2003.U T..........................................................................................................56 T UFigure : 28a.U T T UThe Total Positive Mosquito Pools and the Percent of Positive Mosquito Pools by Week for 2001.U T...........................................................56 T UFigure : 28b.U T T UThe Total Positive Mosquito Pools and the Percent of Positive Mosquito Pools by Week for 2002.U T...........................................................57

PAGE 8

viT UFigure : 28c.U T T UThe Total Positive Mosquito Pools and the Percent of Po sitive Mosquito Pools by Week for 2003.U T...........................................................57 T UFigure 29a.U T T UThe Number of Positive Avian, Sent inels and Mosquitoes with Human Cases for the Panhandle Region during 2001.U T..........................................60 T UFigure 29b.U T T UThe Number of Positive Avian, Sent inels and Mosquitoes with Human Cases for the Panhandle Region during 2002.U T..........................................60 T UFigure 29c.U T T UThe Number of Positive Avian, Sent inels and Mosquitoes with Human Cases for the Panhandle Region during 2003.U T..........................................61 T UFigure 30a.U T T UThe Number of Positive Avian, Sent inels and Mosquitoes with Human Cases for the Northern Region during 2001.U T............................................63 T UFigure 30b.U T T UThe Number of Positive Avian, Sent inels and Mosquitoes with Human Cases for the Northern Region during 2002.U T............................................63 T UFigure 30c.U T T UThe Number of Positive Avian, Sent inels and Mosquitoes with Human Cases for the Northern Region during 2003.U T............................................64 T UFigure 31a.U T T UThe Number of Positive Avian, Sent inels and Mosquitoes with Human Cases for the Central Region during 2001.U T...............................................66 T UFigure 31b.U T T UThe Number of Positive Avian, Sent inels and Mosquitoes with Human Cases for the Central Region during 2002.U T...............................................66 T UFigure 31c.U T T UThe Number of Positive Avian, Sent inels and Mosquitoes with Human Cases for the Central Region during 2003.U T...............................................67 T UFigure 32a.U T T UThe Number of Positive Avian, Sent inels and Mosquitoes with Human Cases for the Southern Region during 2001.U T............................................69 T UFigure 32b.U T T UThe Number of Positive Avian, Sent inels and Mosquitoes with Human Cases for the Southern Region during 2002.U T............................................69 T UFigure 32c.U T T UThe Number of Positive Avian, Sent inels and Mosquitoes with Human Cases for the Southern Region during 2002.U T............................................70 T UFigure 33.U T T UNumber of Clinical Ca ses by Region for 2001-2003.U T...............................72 T UFigure 34a. U T T UFlorida West Nile Cumulative Posi tive Dead Bird Distribution and Clinical Cases by County for 2001-2003.U T.................................................76 T UFigure 34b. U T T UScatter Plot for Correlation between the Number of Clinical Cases and Positive Dead Birds by Week for the Pooled Years 2001-2003.U T..............77T

PAGE 9

viiUFigure 34c. U UScatter Plot for Correlation between the Number of Clinical Cases and Positive Dead Birds by Week for the Pooled Years 2001-2003 excluding outlier observations.U..................................................................................77 UFigure 35a. U UFlorida West Nile Positive Dead Bi rd Distribution and Clinical Cases by County for 2001.U.......................................................................................78 UFigure 35b. U UScatter Plot for Correlation between the Number of Clinical Cases and Positive Dead Birds by Week for 2001.U....................................................79 UFigure 36a. U UFlorida West Nile Positive Dead Bi rd Distribution and Clinical Cases by County for 2002.U.......................................................................................80 UFigure 36b. U UScatter Plot for Correlation between the Number of Clinical Cases and Positive Dead Birds by Week for 2002.U....................................................81 UFigure 36c. U UScatter Plot for Correlation between the Number of Clinical Cases and Positive Dead Birds by Week for 2002 excluding outlier observations.U..81 UFigure 37a. U UFlorida West Nile Positive Dead Avia n Distribution and Clinical Cases by County for 2003.U.......................................................................................82 UFigure 37b. U UScatter Plot for Correlation between the Number of Clinical Cases and Positive Dead Birds by Week for 2003.U....................................................83 UFigure 37c. U UScatter Plot for Correlation between the Number of Clinical Cases and Positive Dead Birds by Week for 2003.U....................................................83 UFigure 38a. U UFlorida West Nile Positive Mosquito Pool Distribution and Clinical Cases by County for 2001-2003.U.........................................................................87 UFigure 38b. U UScatter Plot for Correlation between the Number of Clinical Cases and Positive Mosquito Pools by Week for the Pooled Years 2001 -2003.U......88 UFigure 38c. U UScatter Plot for Correlation between the Number of Clinical Cases and Positive Mosquito Pools by Week for the Pooled Years 2001 -2003 excluding outlier observations.U.................................................................88 UFigure 39a. U UFlorida West Nile Mosquito Pool Distribution and Clinical Cases by County for 2001.U.......................................................................................88 UFigure 39b. U UScatter Plot for Correlation between the Number of Clinical Cases and Positive Mosquito Pools by Week for 2001.U............................................90

PAGE 10

viiiUFigure 39c. U UScatter Plot for Correlation between the Number of Clinical Cases and Positive Mosquito Pools by Week for 2001 excluding outlier observations.U.............................................................................................90 UFigure 40a. U UFlorida West Nile Mosquito Pool Distribution and Clinical Cases by County for 2002.U.......................................................................................91 UFigure 40b. U UScatter Plot for Correlation between the Number of Clinical Cases and Positive Mosquito Pools by Week for 2001.U............................................92 UFigure 40c. U UScatter Plot for Correlation between the Number of Clinical Cases and Positive Mosquito Pools by Week for 2001 excluding outlier observations.U.............................................................................................92 UFigure 41a: U UFlorida West Nile Mosquito Pool Distribution and Clinical Cases by County for 2003.U.......................................................................................93 UFigure 41b. U UScatter Plot for Correlation between the Number of Clinical Cases and Positive Mosquito Pools by Week for 2003.U............................................94 UFigure 41c. U UScatter Plot for Correlation between the Number of Clinical Cases and Positive Mosquito Pools by Week for 2003 excluding outlier observations.U.............................................................................................94 UFigure 42a. U UFlorida West Nile Average Sentin el Chicken Seroconversion Rate Distribution and Clinical Cases by County for 2001-2003.U......................97 UFigure 42b. U UScatter Plot for Correlation between the Number of Clinical Cases and Sentinel Seroconversion Rates by Week for the Pooled Years 2001-2003.U................................................................................................98 UFigure 42c. U UScatter Plot for Correlation between the Number of Clinical Cases and Sentinel Seroconversion Rates by Week for the Pooled Years 2001-2003 excluding outlier observationsU................................................98 UFigure 43a. U UFlorida West Nile Average Sentin el Chicken Seroconversion Rate Distribution and Clinical Cases by County for 2001.U...............................99 UFigure 43b. U UScatter Plot for Correlation between the Number of Clinical Cases and Sentinel Seroconversion Rates by Week for 2001.U.................................100 UFigure 44a. U UFlorida West Nile Average Sentin el Chicken Seroconversion Rate Distribution and Clinical Cases by County for 2002.U.............................101 UFigure 44b. U UScatter Plot for Correlation between the Number of Clinical Cases and Sentinel Seroconversion Rates by Week for 2002.U.................................102

PAGE 11

ixUFigure 45a. U UFlorida West Nile Average Sentin el Chicken Seroconversion Rate Distribution and Clinical Cases by County for 2003.U.............................103 UFigure 45b. U UScatter Plot for Correlation between the Number of Clinical Cases and Sentinel Seroconversion Rates by Week for 2003.U.................................104 UFigure 45c. U UScatter Plot for Correlation between the Number of Clinical Cases and Sentinel Seroconversion Rates by Week for 2003 excluding outlier observations.U...........................................................................................104 UFigure 46: U UEpi Curve Comparing 2003 and 2004 Confirmed Human Cases.U..........111

PAGE 12

x List of Symbols and Abbreviations Symbol and Abbreviations Description % Percent BoEPC Bureau of Entomology and Pest Control BOL Bureau of Laboratories HIPAA Health Insurance Portability and Accountability Act of 1996 CDC Centers for Disease Control and Prevention CHD County Health Department CSF Cerebrospinal Fluid DACS Department of Agriculture and Consumer Services DAI Division of Animal Industry DEP Department of Environmental Protection FBE Florida Bureau of Epidemiology FDOH Florida Department of Health FWCC Florida Wildlife Conservation Commission HAI Hemagglutina tion Inhibition Assay IgG Immunoglobulin G IgM Immunoglobulin M IRB Institutional Review Board MAC-ELISA IgM Antibody Capture Enzyme-Linked Immunosorbent Assay PRNT Serum Neutralization Plaque Reduction Test WN West Nile WNF West Nile Fever WNND West Nile Neuroinvasive Disease WNV West Nile virus

PAGE 13

xi Evaluation of the Current State of Florida We st Nile Surveillance Program as a Predictor for Control and Prevention of Human West Nile Diseases Angela E. Butler Abstract West Nile is an important novel virus in the United States having spread rapidly since it was first detected in New York in 1999. The Centers for Disease Control and Prevention as well as, many State Health Departments have mandated programs for surveillance of West Nile Virus activity. These programs incorporate many different aspects including existing arboserology programs with additional testing for West Nile Virus and new plans that incorporate active and passive surveillance methods. The objective of this study was to examine all aspects of the Florida West Nile surveillance program to determine if there wa s transmission in the animal systems prior to human cases. The predictive analyses were done using regional data graphs, spatial information, correlations and regression models. Data for sentinel chickens, bird necr opsy and mosquito pool surveillance from participating counties in Florida were obtai ned from the State of Florida surveillance database. The human data was obtained from the State of Florida reportable disease database for each county whether participating in the state surveillance programs or not. Clinical cases were examined by demographics (gender and age) and an incidence rate

PAGE 14

xii was calculated to demonstrate the effects of disease. Specific statistical methods used included Pearson’s coefficient correlation, Po isson distribution regression modeling to show if any of the surveillance systems were predictors for human disease. The incidence rate analysis for clinical cases showed clustering of cases in adjacent counties with in a region where Florida’s panhandle and adjacent counties northeast had the highest incide nce. Florida’s central and so uthern regions had moderate human incidence. This provides useful information in transmission geography for prevention and control measures. Demographic an alysis showed that there were twice as many males then females diagnosed with West Nile in Florida, this was true across the groups as well. The highest number of cases was seen within the age group over 55 years of age for West Nile Neuroi nvasive Disease and for West Nile Fever the highest number of cases was within the 36-54 age range. The temporal distribution was determined using graphical representations of the all of the surveillance types and clinical cases. In order to include all relevant data the temporality was set from week 20 to w eek 52. This study found that all of the surveillance types (dead birds, mosquitoes a nd sentinels) offered a specialized strength for predicting clinical cases. However, mosquitoes proved to be the least efficient out of the three surveillance systems. The regional and spatial analysis showed that positive dead birds and sentinels provided the coverage for the surveillance systems in the state. However, Pearson’s correlation coefficient was low for sentinel surveillance; this may be due to higher participation showing West Nile Virus activity in areas (especially rural) that have no reported human cases. This analysis did show that West Nile is detected in mosquito pool samples before it is detected in the dead bird or sentinel surveillance

PAGE 15

xiii systems which provides an earlier warning for human cases. The Poisson distribution regression model was only useful for the pooled years and 2003. These showed that mosquitoes, positive dead birds and sentinels were good predictors for clinical cases for the combined years and dead birds and sen tinels were significant for 2003 as well. The recommendations based on the results from this study would be to cont inue all the current surveillance efforts but with the following enhancements: 1. Increase the coverage and consistency of submissions for all survei llance types. 2. Set standard levels of participation for all counties based on the re gional analyses and populations at risk. 3. Create standardized approaches for sampling, shipping and submitting samples (especially for mosquito pool submissions) and require that particip ating counties adhere to these standards. 4. Only submit specific bi rds known to be especially susceptible to West Nile Virus (e.g. corvids). 5. Targeted pr evention and education strategies for higher risk groups based on their potential levels of exposure.

PAGE 16

1 Introduction Public Health Surveillance Public health surveillance is defined as the continuous systematic collection, analysis, interpretation and di ssemination of data regarding a health-related event, which is used for public health actions to redu ce morbidity, mortality and improve the overall health of the population (CDC, 2001b). An effective surveillance program should incorporate five essential elements: 1) a precisely defined populat ion and disease case definition, 2) standardized data collecti on, consolidation and evaluation methods, 3) proper analysis tools to corre ctly interpret information, 4) a feedback system to disseminate information so that the target population is aware of pub lic health concerns and 5) timely response and implementation of changes to public he alth practices that reflect the information gained during surv eillance activities. The application of surveillance activities support case detection a nd public health interventions, provides an estimate of the impact of disease, define s the natural history of a health condition, determines the distribution and spread of illness, generates hypotheses and stimulates research, evaluates prevention and control measures and helps facilitate program planning (Teutsh, 2000). Perhaps the most important function of public health surveillance is outbreak detection – identifying an increase in the fr equency of a health-re lated event above the

PAGE 17

2 background occurrence of that event both timely and accurately (Broom, 2004). Early detection of outbreaks can be achieved by assu ring timely and complete receipt, review, and follow-up of disease case reports, respondi ng to slight indications that possible events of interest are occurring (i.e., lowe ring the threshold for investigating possible outbreaks, or using modeling tools to improve the predictive value of current programs to identify an outbreak at an ear lier stage), and by monitoring new types of data that may indicate an outbreak event earlier than curre nt surveillance data (Broom, 2004). In order to adequately identify or predict outbreak s, a baseline (background) threshold should be set to better recognize epidemics. An epidemic is relative to the fr equency of the disease within a defined population during a particular season of the year. Epidemics are usually characterized by an increase in cases showing two standard deviations above the mean. Consequently, one case of a disease normally absent or not previ ously recognized in a specified area may not be sufficient to signi fy an epidemic; where as two cases in the same area may be adequate (Last, 2001). By setting a threshold limit, levels of higher than normal activity can therefore be determ ined. This can help to provide a more accurate basis for implementing preventive and control measures. The most critical challenge to public h ealth surveillance is maintaining the efficacy of the program. In 1988, the Centers for Disease Control and Prevention (CDC) published Guidelines for Evaluating Surveillance Systems to promote the best use of public health resources through the developmen t of efficient and e ffective public health surveillance systems. These gui delines were updated in 2001 to address the need for a) the integration of surveillance and health information systems, b) the establishment of data standards, c) the electronic exchange of health data, and d) changes in the objectives

PAGE 18

3 of public health surveillance to facilitate th e response of public hea lth to emerging health threats. As a supplement to th ese publications, the CDC published Framework for Evaluating Public Health Surveillance Syst ems for Early Detection of Outbreaks in 2004 for the purposes of evaluating public heal th surveillance systems for their timely detection of outbreaks. The purpose of evaluati ng public health surveillance systems is to ensure that problems of public health im portance are being monitored efficiently and effectively. Public health surveillance syst ems should be evaluated periodically, to improve the quality, efficiency, and usefulness of the program and should focus on how well the system operates to meet its purpose and objectives. A public health surveillance system should emphasize the components that are most important fo r the objectives of the system. An evaluation of that system must therefore consider the same components (CDC, 2001c). In order to establish the re lative value of different approaches and improve early detection of out break efficacy, there must be a focused attention to the measurement of the performance of public health surveillance systems (Broom, 2004). The Updated Guidelines for Ev aluating Surveillance Systems and Framework for Evaluating Public Health Surveillance Syst ems for Early Detection of Outbreaks describe four main tasks involved in evaluating pub lic health surveillance systems for early outbreak detection. The first is a descripti on of the surveillance system being evaluated including: defining the stake holders (individuals or agencies who provide and use the information generated by the system), the pub lic health importance of the health-related event under surveillance (frequency, severity, preventabi lity and public interest), a description of the purpose and operation of the surveillance system (purpose and objectives of the system), planned uses of the data collected, h ealth-related event under

PAGE 19

4 surveillance with case definition, a response flow chart, and a description of the resources used to operate the surveillance system. Th e second task focuses on the evaluation design by determining the specific purpose of the ev aluation, considering what will be done with the information generated from the evaluati on, and specifying the que stions that will be answered by the evaluation. The third task ev aluates the performance of the surveillance system by indicating the level of usefulne ss and describing the system’s simplicity, flexibility, acceptability, and stability. Th e final task draws conclusions from the evaluation and recommends new uses and improvements to the system (CDC, 2001c; Broom, 2004). Evaluation of the Arbosurvei llance System in Florida Description of Health Related Event West Nile Virus (WNV) is a single-stranded RNA virus in the family Flaviviridae genus Flavivirus. It is an arbovirus (UAr Uthropod-UBo Urne-UVirus U) primarily transmitted by mosquitoes. The virus can infect a wide range of hosts including humans, birds and horses. WNV has also been isolated in other mammals and alligators, however, little is known about the a ssociated symptomology for th ese infections (Stark, 2003). Distribution West Nile Virus has been characterized with worldwide outbreaks every few years with the temporal distribution primarily during late summer and fall. It was first identified in the West Nile district of Uganda in 1937 (Huhn, 2003), but it wasn’t until 1957, during an outbreak in Israel that it was recognized as a cause of severe human meningoencephalitis (Chowers, 2000). Descriptions of disease symptoms in Israel can be

PAGE 20

5 found dating back to the early 1940’s. This connection linking WNV and Israel may be due to the fact that many bird species have migratory patterns w ith their fly way zone from Europe to Africa through Israel (Leffkowitz, 1942). In the 1960’s WNV was associated with equine illnesses in Fr ance and Egypt (Murgue, 2000; Halouzka, 1999) and began to threaten the United States in 1999 (Komar, 1999; Nasci, 1999; Nash, 2001). Worldwide distribution is limited by ecological patterns supporting the WNV transmission cycle. These limiting factors incl ude temperature, prec ipitation levels and vegetation all of which influence the vector and host relationship (Gubler, 2001). Figure 1 shows the worldwide distribution for flaviv iruses as of 2000 where WNV was primarily found in parts of Africa, Europe and parts of Asia with the exception of New York in the United States.

PAGE 21

6 Figure 1. The Worldwide Geographic Distribution for the Serocomplex of the Family Flaviridae as of 2000. This map shows West Nile Virus prim arily in Africa and Europe however the New York outbreak is indicated. *Map source: CDC website. Available from URL http://www.cdc.gov/ncidod/dvbid/westnile/map.htm West Nile Virus emerged as a public hea lth threat in the United States in 1999 with an outbreak in New York (CDC, 1999; Komar, 1999; Nasci, 1999; Nash, 1999), and has since spread throughout North America. Figure 2a shows the distribution of West

PAGE 22

7 Nile Virus activity as of 2002 and Figure 2b shows additional activity from 2003. Confirmed cases of WN disease in humans has been reported in 45 states and the District of Columbia (CDC, 2003b) with evidence of en zootic activity (natural transmission cycle between mosquitoes and avian hosts) in 28 states (Huhn, 2003). Figure 2a. West Nile Virus Distribution across the United States from 1999-2002. This map shows the spread by year fro m the first appearance of WN in New York through 2002. The states colored white have had no WN activity as of December 2002. *Map source: CDC website. Available from URL http://www.cdc.gov/ncidod/dvbid/we stnile/surv&control03Maps99_02.htm

PAGE 23

8 Figure 2b. West Nile Virus Distribution across the United States for 2003. This map shows the distribution of human WN disease as well as bird, animal and mosquito infections. The states colored white have not had any WN activity as of December 2003. *Map source: CDC website. Available from URL http://www.cdc.gov/ncidod/dvbid/we stnile/surv&control03Maps.htm In Florida, WNV was first isolated in July of 2001 in a crow from Jefferson County located in the panhandle region (Bl ackmore, 2001). As of December, 2003 there have been a total 140 laboratory confirmed cases of WN disease in the human population across 44 of Florida’s 67 counties (Stark, 2003).

PAGE 24

9 Transmission Certain wild avian populations are co nsidered the primary hosts for WNV maintaining a natural enzootic cycle with mo squitoes. The passerines (song birds) have been found to be the most competent primary hosts by supporting a viremia high enough to infect feeding mosquitoes; however some of these species do not have long enough sustained life. The most competent pr imary host which supports both a high enough viremia and a long enough sustained life is most likely the House Finch (Komar, 2003). The specific species, their associated viremi a and sustained life span are shown in Appendix III. Like other viruses in its fa mily, WNV infections in humans and other animals (including some bird species) are cons idered to be dead-end or incidental hosts because viremia is not sufficient to suppor t transmission (Hadler, 2000; Huhn, 2003). The primary enzootic cycle is maintained by the presence of the vector and the host and epizootic cycle requires interact ion with the bridge vector mo squito and an “incidental” human host. The mosquitoes that are part of the cycle tend to be ornithophilic (e.g. Culex pipiens ) (Tyler, 2001; CDC, 2002). There have been several documented cases where person-to-person transmission has occurred, how ever, through organ transplants, blood transfusion and breast-feeding (Mitka, 2003). Human cases of WNV infection are generally low in number until daily rainfall patterns stimulate mosquito vectors to become more active (FDEH, 2003). It should be noted however, that a hallmark of outbreaks located in the United States had significant mortality within population of corvid species (crow and blue jay) servi ng as amplifying hosts (Eidson et al., 2000a; Bernard, 2000; Eidson et al., 2000b). However, corvids tend to die rapidly after they are

PAGE 25

10 infected with WNV leading to a debate on whether they can truly be the primary amplifying host. A detailed picture of the tr ansmission cycles is depicted in Figure 3. Figure 3. The Transmission Cycle for West Nile Virus. The diagram shows the vector/bird primary enzootic cycle and transmission to “incidental” hosts by a bridge vector mosquito. The incidental hosts include horses, humans and other animals. *Diagram used with permission from David Klemm. Clinical Presentation Following transmission of the virus by th e mosquito vector, WNV multiplies in the host’s circulatory system and may cross the blood-brain barrier to reach the central nervous system (CNS). The virus generally requires an incubati on period of 3-14 days and symptoms can last from 3 to 6 days (CDC, 2001a; Campbell, 2002). Infection of the brain interferes with normal CNS function and begins to cause inflammation of the tissues surrounding the brain (Sejvar, 2003). Th e severity of the infection depends on the host’s immune response to viral replication. People over 50 years of age are more likely to develop sever symptoms associated with West Nile infection, but severe disease can occur in individuals of any age (Campbell, 2002; Petersen, 2002). It is unknown whether

PAGE 26

11 immunocompromised persons are at an increas ed risk for WNV associated infections (Peterson, 2003). Most WNV infections are sub-clinical with only about 20% of infected individuals developing a mild form of di sease termed WNF. Du e to under-reporting, a complete clinical picture of WNF’s effects on the population of the United States has not yet been determined, but it is characterized as a febrile illness of sudden onset with generalized flu-like symptoms including mala ise, anorexia, nausea, vomiting, eye pain, headache, myalgia, rash and lymphadenopathy (CDC, 2001a). A more severe neurological disease ma y develop in approximately 1 out 150 people infected with West Nile. West Nile Ne uro-invasive Disease (W NND) is classified as a meningeoencephalitis (inflammation of the brain and surrounding membranes) and occurs most often in patients of advanced age (> 50 years old). Recent outbreaks have described a hallmark fever, severe muscle weakness ranging to flaccid paralysis (rare), ataxia, gastrointestinal upset, changes in mental status and death. More severe neurological presentations have included seizures, myelitis, cranial nerve deterioration, optic neuritis, and polyradiculitis. Rarely, a maculopapular or morbilliform rash forming on the neck, trunk, arms or legs will deve lop with severe disease (CDC, 2001a). Treatment of WNV associated illnesses, whether mild or severe, is typically supportive. Often, the patient will be hos pitalized and given intravenous fluids, respiratory support and antibiotics for the pr evention of secondary infections (CDC, 2001a). The Centers for Disease Control and Prevention’s case definition for WNND includes probable and confirmed cases. A proba ble case is determined by the presence of

PAGE 27

12 encephalitis or meningitis during arboviral transmission season and serology showing an elevated titer of virus-specific serum antibodie s (less than a two-fold increase) or serum IgM or IgG antibodies detected by antibody-captu re EIA with no available results for the same or later specimen of virus-specific serum antibodies. A probable case will be considered a confirmed case when tests sero logically confirm a four fold increase in virus-specific antibodies betw een acute and convalescent se ra usually collected two weeks apart (CDC, 2003a; Marfin, 2001) Based on this case definition a confirmed case of WNND may be diagnosed in Florida after one of following laboratory criteria has been met: a fourfold or greater change in WNV-sp ecific serum antibody titer, or the isolation of WNV from or demonstrati on of West Nile viral antigen or genomic sequences in tissue, blood, cerebrospinal fluid (CSF), or other body fluid, or the presence of specific IgM antibody by enzyme immunoassay (EIA) antibody capture in CSF or serum with confirmation by IgG-EIA or another se rologic assay (e.g. neutralization or hemagglutination inhibition (HAI)). Both acute and convalescent sera from reported and suspected cases should be drawn since cross -reactivity between WNV and other closely related flaviviruses can occur if the patie nt has recently been vaccinated against a flavivirus (e.g., Yellow Fever) or infected with another flavivirus and may present a false positive West Nile MAC-ELISA result (FDEH, 2003). West Nile Neuro-invasive Disease is a mandatory stat e reportable disease under Section 381.0031.of the Florida Statute. Theref ore, all probable and confirmed cases of WNND must be reported to the Flor ida Department of Health (FDOH). Diagnosis of WNND usually begins when a patient presents w ith an encephalitis or meningitis of unknown origin. Upon rece ipt of this evidence an immediate

PAGE 28

13 epidemiological investigati on for arboviral infection s hould be conducted. The Human Case Investigation Guidelines set forth in Surveillance and Control of Selected Arthropod-borne Diseases in Florida 2003 recommends conducting a case interview which includes the age of the patient, history of mosquito bites within 14 days prior to the onset of symptoms, travel and activity history that would increase the risk of an arboviral illness, an environmental investigation to de termine the risk of mosquito activity and local enzootic transmission or other re gional human cases. Neurological symptoms caused by arboviruses mimic symptoms of mo st other CNS infections. Therefore, appropriate specimens should be collected and sent to the Florida De partment of Health (FDOH) Bureau of Laboratories (BOL) for a confirmed diagnosis. Florida’s Surveillance System A comprehensive surveillance program for arboviruses should consist of monitoring for increases in arboviral seroconve rsion rates in sentinel chickens, weather patterns, the presence of vect or and amplification host specie s (based on species infection rates), and the incidence of huma n and animal disease (FDEH, 2003). Florida’s existing arbosurveillance plan involves several distinct monitoring systems including the vectors, a wide variet y of hosts and captive sentinels. The hosts include serosurveillance of sen tinel chicken flocks, wild-avi an sero-survey and detection of the virus through molecular and isolati on methods for the vectors and laboratory submitted dead avian specimen. Veterinary surv eillance is passive and detection is also indicated by molecular methods and isola tion from the tissue of submitted animals. Serological testing for veterina ry surveillance is also done at the Kissimmee Department of Agriculture and Consum er Services (DACS) lab.

PAGE 29

14 The entire surveillance plan encompa sses several laboratory methods for determining the presence of WNV: Hemagglutination Inhi bition Assays (HAI), IgM Antibody Capture Enzyme-Linked Immunosorbent Assay (ELISA), Serum Neutralization-Plaque Reduction Assay (SNP R), Reverse Transcriptase-Polymerase Chain Reaction (RT-PCR) and virus isolation in cell culture. These collectively are the most common detection methods used in WN surveillance. The HIA is used to detect antibodies for a specified virus (i.e. flaviviruses) in sera from a wide variety of animals (most comm only humans, chickens and horses). HAI is used as an initial screening assay for sentinel surveillance because it can efficiently test a large number of samples from diverse speci es. Confirmations for sentinel chicken positives or reactive samples are assayed using the IgM Antibody Capture ELISA which is more specific than the HAI test. Confir mations for non-chicken samples are confirmed with SNPR. Animal tissues are tested for the presen ce of West Nile viral RNA using the RTPCR and virus culture for isolation. This is utilized for WNV detection in avian surveillance and mosquito surveillance to det ect virus activity from submitting counties. Viral amplification and transmission in the environment can be deduced using the arboviral surveillance system for detection, which may also indicate risks of human disease. Due to the complex nature of the arboviral transmission cycles, multiple surveillance and detection methods are necessary for an accurate risk assessment. This is achieved by providing specific th reshold levels, indicator parameters and predictive models, all of which may vary by season and region. The surveillance system in Florida

PAGE 30

15 consists chiefly of avian mortality and morb idity, sentinel chicken flocks, and vector testing. These samples are submitted to the Fl orida Department of Health, Tampa Branch Bureau of Laboratories, but program complian ce is reliant on indivi dual county Mosquito Control Districts and County Hea lth Departments. Equine surv eillance is also included as a component of West Nile surveillance. Te sting is generally done by DACS however; it is only performed for suspected cases. Some Mosquito Control di stricts do their own mosquito pool testing using traditional Poly merase Chain Reaction (PCR), Vectests and Ramp tests. The data from these tests are not reported to the Department of Health; therefore, outcomes are not included for prevention and control measures mandated by the state. Bird mortality surveillance is used to detect West Nile activity within a particular geographical area. Through reporting and testin g of dead birds, mortality patterns among avian species (particularly corvids) may be uti lized to generate a spatial analysis to help determine a specific geographical area at ri sk for human cases of WNV associated disease (Eidson, 2000a). This is only useful however, when looking at mortality among non-migratory bird species. Laboratory testing of dead bird tissues is necessary for an accurate picture of vira l presence in a specific geographi cal area. Collection, shipping and laboratory activities are time consuming and co stly; however, posing a major challenge to this type of surveillance. A recent study on the economic impact of WN disease in Louisiana during 2002 showed an overall co st of 20.l million dollars from overall medical expenses for 342 cases (Armineh, 2004). Another area of concern when using this data to predict the risk of human dis ease is the mobility of individual birds. For example, geographic analyses can be skewed when a dead bird is submitted from an area

PAGE 31

16 other than where it was initially infected. Spat ial analysis from reports of dead birds tends to work better in urban areas of Florida (e.g., North Florida) wh ere dead bird s are correlated with human population dens ity, as opposed to rural areas. Serosurveillance of captive sentinel fl ocks reduces these concerns. Ideally, a sentinel flock would be sus ceptible to infection but resistant to the disease thus minimizing their contribution to the transmissi on cycle. Sentinels should develop a rapid immune response and seroconvert before the disease can be detected within the human population. Lastly, they must be easily mainta ined with minimal health risks to their handlers (Komar, 2001). Florida uses chickens as their sentinel species since they maintain most of the attributes of an ideal sentinel. They have provided a good model for local activity of arboviruses. Chickens are pl aced in cages on sites located throughout the state to allow natural transmission between th e vector and the host. The sites are chosen based either on the geographical availability or because of past activity in an area. The overall goal is to indicate WN activity within a geographi cal area through seroconversion rates of the sentinel chickens. Several mammal models have been use d, ranging from equines and canines to several different species of rodents (Kom ar, 2001). Theoretically, sentinel mammals would be a better representation of the epiz ootic transmission cycle rather than the enzootic cycle among birds. Therefore, captive mammals used as sentinels may be a better gauge for the risk of human disease. In Florida horses were in itially looked at as a sentinel mammal but it proved too costly to maintain these herds. Mosquito Control Districts and some pr ivate special taxing districts are the controlling entities of the county mosquito control in which th ey are located, and

PAGE 32

17 participation in surveillance activities is voluntary. While some counties participate regularly in surveillance activities most do not. Florida has not had much success with mosquito surveillance however; some states (e.g., Colorado) with higher viral activity and submittal rates have had more success with this type of surveillance. If areas in Florida were to have an increase in tran smission or more submissions during peak periods of activity for the affected geographical areas, mosquitoes may show better predictive value for forecasting trends in clinical West Nile cases. Several collection techniques are compa tible with laboratory testing; those available to the mosquito control distri cts include CDC traps (with or without CoP2P), Gravid traps, ABC light traps (with CoP2P), MM-X traps (with COP2P), Lardcan and Mosquito Magnet traps. These traps may be placed anywhere WNV transmission is suspected. Additional trap sites can be added as needed, and should be considered in areas of past viral activity or where avian reservoirs or se ntinel flocks are locat ed. Male mosquitoes do not take blood meals and are incapable of transmitting WNV, therefore, mosquito pools sent to the laboratory for testing should consist of non-fed (because it would be impossible to differentiate whether the virus originated with the mosquito or the blood meal), gravid, female mosquitoes only (FDEH, 2004). Collection of mosquito pools by the county mosquito control districts are logged by collection site, date and speci es of mosquito then shipped to the laboratory for testing with RT-PCR and viral isolati on for the presence of WNV. Florida also uses passive veterinary survei llance for the detection of West Nile in other animals. This type of surveillance requ ires that veterinarian s send tissue, blood or CSF from suspected animal cases of West Nile to the laboratory for testing. The tissues

PAGE 33

18 are assayed through molecular testing and isol ation by the same protocols as avian and mosquito samples. Unfortunately, data collec ted from passive surveillance efforts is traditionally unreliable due to poor compliance thus, does not provide useful information for use as a predictive model in human cases. Florida’s Arbosurveillance Response Plan The state of Florida has set up a response plan for Mosquito-B orne (arboviruses) Diseases (figure 4) based on the continuous re sults from the arbosurveillance efforts. The plan has four tiers: Background Activity, Mo squito-Borne Illness Advisory, MosquitoBorne Illness Alert, and Mo squito-Borne Illness Threat Each tier includes specific criteria, geographical areas and response effo rts. Since transmission generally occurs locally, the response plan is proposed for the affected counties or regions and is not intended as a statewide re sponse plan (FDEH, 2004).

PAGE 34

19 Figure 4. Diagram of the State of Florida’s Speci fic Response Plan for Arbovirus Detection. This includes different advisories fo r a specific percentage increase among the surveillance systems. Dead Birds Symptomatic + E p idemiolo g ical Evidence Laboratory CSF or Sera Mac-ELISA i + + Stop + + Monitor for significant changes in # of Dead Birds reporte d Stop + Stop Dead Bird Web Database Reporting Mosquito Borne Illness Advisory 10% increase in Sentinel Chicken Seroconversions 10% increase in Corvid Mortality 10% increase in the Minimal Infection Rate (MIR) of Vector Mosquitoes Two or more Confirmed Equine Cases (increases above background activity) County Pools Laboratory Testing Isolation & Detection Laboratory Testing Virus Isolation & PCR Detection Sentinel Chicken Surveillance Laboratory HAI testing Weekly Sentinel Chicken Sera Mosquito Surveillance Clinical Case Surveillance Stop Avian Surveillance Stop Mosquito-Borne Illness Threat Potential for Widespread Distribution of a large number of Human Cases (Declared by the St ate Health Officer) Mosquito-Borne Illness Alert A confirmed human case 50% increase in Sentinel Chicken Serconversions in a county or flock 50% increase in corvid mortality ( increase above advisor y) ELISA,SN Confirm

PAGE 35

20 Background activity occurs when the pe rcentage of positives within the surveillance system does not exceed historical levels for that region. The response for background activity would be continued regul ar surveillance efforts (FDEH, 2004). A Mosquito-Borne Illness Advisory, the second response tier, will be declared when the surveillance in a particular geogra phic area reveals a 10% in crease in sentinel seroconversions, a 10% increase in corvid mortality, a 10% incr ease in the minimal infection rate (MIR) of vector mosquitoes all above normal background activity or two or more confirmed equine cases during two c onsecutive weeks. This would indicate a potential increase in viral act ivity thus increasing the risk of human infections. The overall response should include continued surveillance activ ities with the addition of public information announcements and health care provider advisories (FEDH, 2004). The third tier, Mosquito-Borne Illness Alert, is initiated with a confirmed clinical case or 50% increase in sentin el seroconversions within a county or flock, or a 50% increase in corvid mortality. A mosquito-bor ne illness alert response should include the continuation of the previous responses for background activity and Mosquito-Borne Illness Advisory while increasing mosquito control measures on the local level. The last tier, Mosquito-Borne Illness Threat, will be invoke d if there is a potential for widespread clinical disease associated w ith arboviral infection a nd will be declared by the State Health O fficer (FDEH, 2004). In addition to Florida’s Response Plan for Mosquito-Borne Diseases, the state maintains an ongoing program for the preventi on of transmission to humans. The existing arboviral prevention campaign in Florida includ es education on the “5 D’s of Prevention:

PAGE 36

21 Dusk and Dawn, Dress, DEET, and Drai nage” and community intervention through medical alerts issued when surveillance sy stems indicate increas es arboviral activity. Currently, no vaccine is available for the prevention of WNV, but clinical trials are in progress (WHO, 2004). The CDC also has a national program called “Fight the Bite” which is based on the same strategi es as Florida’s prevention campaign (CDC, 2004). Defining the Stakeholders The State of Florida defines its primary stakeholders as “interagency partners” taken from both the state and local levels. Th ese partners are responsible for coordinating the dissemination of information to the pr oper parties. DACS Bureau of Entomology and Pest Control (BoEPC) is respons ible for notifying all mosqui to control agencies for the affected counties and DACS Division of An imal Industry (DAI) will notify animal industry organizations and veterinarians. The Florida Wildlife Conservation Commission (FWCC) and the Department of Environmen tal Protection (DEP) will notify regional biologists and wildlife rehabi litators. The FDOH, BOL will notify the County Health Departments (CHD), the Department of Comm unity and Environmental Health, CDC and sample submitters. The FDOH and CHDs are also responsible for the release of all public information regarding health alerts to physicians and hospitals and recommended precautions, while the local mosquito contro l agencies (or BoEPC if none exist) are responsible for the release of information regarding mosqu ito control activities (FDEH, 2003).

PAGE 37

22 Purpose and Objectives of System The Surveillance and Control of Selected Arthropod-borne Diseases in Florida 2003 published by the Florida Department of He alth, Division of Environmental Health establishes guidelines for de tecting and monitoring arthr opod-borne diseases such as West Nile Neuro-invasive Disease (WNND) and WNF to minimize the risk of human infection. Its purpose is two-fold. First, to identify the system functions, such as surveillance and data management activities, that monitor se ntinel chicken flocks, wild bird and mosquito populations. The second is to propose prompt and effective control methods. Such a program would eventually al low the State of Flor ida to determine the likelihood of the location and tim e of possible arboviral transm ission to humans prior to an outbreak event, allowing time to implement the necessary preventive procedures. The data collected from these surveillan ce activities should be used to further characterize the threat of ar thropod-borne diseases in Fl orida and the United States, provide adequate prevention and control of arthropod-borne diseases during peak transmission periods, and improve the overa ll health and well being of the general population in these areas. Evaluation Design The unique environmental and demographic conditions found in Florida create an increased risk of exposure and a potential in creased risk of contracting the diseases associated with West Nile Infection. Th e State of Florida has an extensive ongoing surveillance program for the detection of arboviruses. The overall outcome of each disease monitoring activity included in the surveillanc e program should either

PAGE 38

23 independently or collectively act as a warn ing system to prevent human WNV Disease via various medical alerts, increased mosquito control and heightened public awareness through educational efforts. Early detection of outbreak conditions through surveillance could become an important public health control and preventi on strategy for human illness associated with West Nile virus. This study evaluated the current arbosurv eillance programs for the detection of WNND and WNF cases in the state of Florida. While evaluating the data collected from these surveillance programs, this study attempted to develop a model for the early detection of human West Nile infections. Questions that this evaluation answers include: will the data collected by Florida’s Arth ropod-borne Disease Surveillance Program provide the appropriate information necessary for the generation of a model to predict human West Nile virus infection, which dis ease monitoring effort or combination of efforts included within the evaluation yielde d the most significant data for use with the model, and will this model provide an accu rate and early warning of human disease associated with West Nile infection? The information generated from this ev aluation was used strictly to determine whether a model for predicting human disease is possible and whethe r that model could serve as an accurate predictor of human cases.

PAGE 39

24 Materials and Methods Four datasets were compiled for this evaluation: human, sent inel chicken, dead birds and mosquito and are summari zed for each year (2001, 2002 and 2003) in Appendices IV, V and VI respectively. Th e Florida Department of Health (FDOH), Division of Environmental Health, provided the information contained in the human dataset. Sentinel chicken, dead birds and mosquito data were all obtained through the FDOH, Bureau of Laboratories. The human dataset consists of all confirmed WNND a nd WNF cases reported to the FDOH from 2001 to 2003. The year 2003 marked the first time that WNF was reported along with WNND. Age, gender, date of onset of symptoms and the reporting county were the only variables in this dataset, which confor ms to all HIPPA regulations. Ethical clearance was obtained from th e Institutional Review Board (IRB). Counties participating in the sentinel chicken surveillance program (Appendix VII) submit weekly serum samples taken from chickens located in various geographical sites around the state. Th e levels of participation among c ounties vary from year to year with some counties increasing th e number of sites and others decrease sites. Only sera confirmed positive between the years 2001—2003 were included in this evaluation’s dataset. To evaluate temporality, each repor ted result date was a ssigned a week number provided by the Microsoft Excel formula WEEKNUM. A weekly seroconversion rate was then calculated for the chicken population using the following equation: Total # of positive chicken sera by county per week Total # of chickens bled by county per week

PAGE 40

25 Two methods of collecting dead bird s information are described in the Surveillance and Control of Selected Ar thropod-borne Diseases in Florida 2003 ; dead bird web based reporting and laboratory testi ng of dead birds. The web based reporting system uses the internet as an interface for citizens or Count y Health Department officials to report dead birds. This reporting system could not be included in this dataset because information was not available for all year s included in this evaluation. Counties participating in the dead bird surveillance program are liste d in Appendix VIII. Variables included in the dead bird data set consisted of specimen collection date, laboratory result and county of collection. Dead birds laborat ory confirmed positive for West Nile virus collected from 2001 through 2003 were assigned a week based on collection date according to the WEEKNUM Microsoft Excel formula. The mosquito dataset included species sp ecific mosquito pool results collected by mosquito control agencies from counties submitting to the FDOH, BOL. The counties participating are listed in the table shown in Appendix VIII. The dataset for this study included all laboratory confirmed positive mo squito pools collected from the year 2001 through 2003. Mosquito Control District testi ng was not included in the data analyzed. The only variables used in analysis were co llection date, laboratory result and county of collection. These were queried from the F DOH, BOL’s Microsoft Access Database for 2002 and 2003; for 2001 the variables were queried from a Microsoft Excel Database. The collection date was used to generate a week number using the Microsoft Excel formula WEEKNUM. Data from the various sour ces were excluded based on certa in criteria. The data in each surveillance system were included if the collection date, laboratory result and

PAGE 41

26 county of collection were available. If one or more of these variables were missing, the data was omitted from the final dataset. The population of each county was acqui red through Florida CHARTS for the years 2001, 2002 and 2003 for the purpose of ca lculating incidence rates. Florida CHARTS provides an estimate of the total popul ation and is classified by age, race and gender. Only the estimated total populati ons for counties with confirmed WNND or WNF reports were used. Analysis Descriptive and summary statistical met hods were used for examining the four datasets. Microsoft Excel was used to generate graphical representati ons of distributions for clinical demographics, confirmed hu man case reports, sentinel chicken positive conversion rates, positive bi rds, positive mosquito pools and dead bird reports. The clinical demographics examined included age and gender. In order to adequately represent the distributio n of age it was divided into four categories <18 years, 19 to 36 years, 37-54 years and >55 years. Incidence was calculated separa tely for all counties with established clinical cases for each year (2001, 2002 or 2003). The calcula tion was computed according to the given incidence equation: Number of New Cases Specified Population The number of new cases is the total c onfirmed WNND and WNF case reports for the associated county by year. The specified popul ation is the estimated population of the county where a confirmed case was reported du ring the indicated year. The counties were

PAGE 42

27 organized by region and assigned a specific number indicating location to adjacent counties (Appendix IX). The statewide incide nce was computed by adding the total new cases from each county and dividing it by the population of each county with new cases. The temporal distribution was determined using week numbers generated from the collection date of dead bi rds and mosquito pools, reported date for sentinel chicken results and date for start of symptoms fo r clinical cases. The summary statistics calculated for each surveillance type and human cases were used to determine the starting and ending week for the West Nile transmi ssion season. Outliers were defined as data points outside the calculated transmission seas on. In order to show these distributions were not due to chance alone, a moving aver age analysis was done, plotting the average for each surveillance type with the calculate d moving average by region using Microsoft Excel. The average for each year’s survei llance systems was calculated by pooling the data by week, for each year and dividing by the total number of years observed (three). Comparative summary statistics were gene rated for each surveillance system and clinical cases by Analyse-it version 1.71 a Microsoft Excel add-in program using a generated box-plot. Box-plots graphically show the central location and scatter/dispersion of the observations of the samples. Refer to figure 17 on page 43 for an example showing a box-plot. The blue line series shows para metric statistics where the blue diamond represents the mean and the confidence interval around the mean. The notched blue lines show the parametric percen tile range. The notched box shows the median, lower and upper quartiles, and confidence interval around the median. The dotted-line connects the nearest observations within 1.5 IQRs (inte r-quartile ranges) of the lower and upper quartiles. The red crosses (+) and circles (o) in dicate possible outliers observations more

PAGE 43

28 than 1.5 IQRs (near outliers) and 3.0 IQRs (far outliers) from the qua rtiles (Analyse-it, 2003). The mosquito pool and dead bird data se ts were graphed using columns and lines with a primary and secondary axis (y and z) by the total number of submitted specimen and the percent positive using Microsoft Excel The percent positive was calculated using the following equation: Number of Positive per Week Total Number of Specimen Submitted per Week Descriptive statistics were used to represent distributi ons for the Florida regions (Panhandle, North, Central and South) depi cting the overall arbosurveillance outcome preceding and during the time period that included clinical cases. These were accomplished by pooling county data into regions based on physiographic climate differences shown in Appendix IX (Day, 1996). The surveillance data included the number of sentinel, dead bird and mosquito positives and the total reported dead birds. These were all graphed together using colu mns and lines with a primary and secondary axis (y and z) where the columns represente d the number of positive mosquito pools and sentinels with the number of dead bird pos itives were represented by the line using Microsoft Excel. The numbers of clinical cas es were represented by arrows showing the week of the onset of symptoms. Spatial analysis was accomplished us ing ArcView 3.3 (Environment Systems Research Institute, Inc, Redlands, CA, USA) for cartographical images. The shape file for

PAGE 44

29 Florida with county boundaries was obtaine d from Environment Systems Research Institute, Inc (ESRI) files for the United St ates. Each type of surveillance (sentinel chicken, dead birds submitted, mosquito and total dead birds reported) for 2001, 2002 and 2003 were independently examined w ith confirmed WNND and WNF reports. A correlation using Pearson’s correlation coefficient was used to determine an association between the number of clinical cases and each surveillance type. The correlation graph and values were generated through the Microsoft Excel add-in Analyse-it. Analytical statistic methods were perfo rmed using SAS (Statistical Software Systems v8.0, SAS Institute Inc., Cary, NC USA) for multivariate Poisson Regression (PROC GENMOD) analysis of the datasets. Th is regression model was used because the response variable was an incide nce rate. The Poisson regression is a member of a class of generalized linear models, which is an extension of traditional linear models that allows the mean of a population to depend on a lin ear predictor through a nonlinear link function and allows the response probability distribut ion to be any member of an exponential family of distributions (Neter, 1996; Arge sti, 1996; Stokes, 2000). The PROC GENMOD of SAS can fit a wide range of generalized linear models. The following SAS statements use PROC GENMOD to fit the Poisson regression: log(BiB)= log( pBiB) + B0B + B1B (surveillance system) (where log( pBiB) is used as the offset in the calculations: offset=log of the population for each region by year, surveillance system=positive dead bird rate, positive mosquito pool rate or sentinel seroconversion rate) when considering the effects of region: log(BiB)= log( pBiB) + B0B + B1B (central) + B2B (north) + B3B (panhandle) + B4B (south) + B5B (surveillance system) (where log( pBiB) is used as the offset in the calculations: offset=log of the population for each region by year, B4B (south) was the reference region therefore=0.00, su rveillance system=positive dead bird rate, positive mosquito pool rate or sentinel seroconversion rate)

PAGE 45

30 A combined year analysis for 2001, 2002 and 2003 classified by region for all surveillance systems was used to identify the best model for the early detection of West Nile associated human illness. The surveillance types were set as parameters and modeled collectively and independe ntly (due to differences in participation levels) with and without regional data. The incidence rate (calculation shown previously) which based on number of confirmed WNND and WNF th e case reports was set as the response variable for the variable of sentinel chicke n seroconversion rates, percent of positive dead birds, and percent of positive mosquito pool s per region by week reported by the FDOH. Calculations for percent positives were done using the previously shown equation. These parameters and the response variable were analyzed independently in a combined model and by individual year (2001, 2002 and 2003) annually. There was not consistent participation in all the surveillance types by the counties therefore; a collective model of all the parameters was not used.

PAGE 46

31 Results Demographic Analysis Gender The graph in Figure 5 illust rates the number of cases of West Nile Infection for 2001-2003. The graph shows the number of males is approximately twice as high as the number of females with West Nile. The graph for individual years and numb er of West Nile cases by gender is represented in Figure 6. This graph exhibits an overall higher trend of diagnosed males (RP2P = 0.5535) than females and the number of West Nile infections in males decreased from 2001 to 2002 but had a sharp increase in 2003. The number of females with West Nile disease shows a slight increase from 2001 to 2002 but remains constant from 2002 to 2003. The overall trend for females is c onsistent over the three years with an RP2P =0.75. The regression trend line equation for male s and females were y = 9x + 10.333 and y = 2x + 2.6667 respectively (male slope=9 and female slope=2).

PAGE 47

32 Figure 5. Total Number of West N ile Cases and Gender for 2001-2003. The number of males is much greater th an the number of females infected with WNND. The magenta bar repres ents the number of female WN clinical cases and the blue bar repres ents the number of male WN clinical cases. 0 20 40 60 80 100 120 FemaleMaleNumber of West Nile Cases Figure 6. Cumulative Number of WN Cases for Gender by Year. The male cases are shown in blue, females in magenta. The black line represents the fitted regression line for the total number of cases for each gender (males and females) over all years. R2 = 0.75 R2 = 0.5535 0 5 10 15 20 25 30 35 40 45200120022003YearNumber of Cases Female Male

PAGE 48

33 Age Groups Age group distributions for 2001, 2002 and 2003 were combined and are shown in Figure 7. Among the age groups the greatest number of WNND is seen in the population over 55 with the lowest number under 18 years old. There is a perfect correlation (Pearson’s Correlation Coefficien t = 1.00 with a p-valu e of 0.0034) between the age groups and the number of cases of severe disease. This shows that as age increases (by age group) the number of WN ND cases also increases. Prior to 2003, primarily hospitalized severe cases were diagnosed by la boratory testing. Both WNND and WNF are reported during 2003. Therefore, the 2003 data was further separated by infection type: WNND and WNF. The graphical representation for specific infection type and age group indicates that WNF is more prominent among 36 to 54 year olds whereas WNND occurs more often in the older age group. When gender and age are grouped together (F igure 9) the data further revealed the increase in the number of clinical cases as age increases regardless of gender, but the gender trend noted in Figure 5 remains constant.

PAGE 49

34 Figure 7. WNND by Age Group for 2001-2003. The red bars represent the number of WN clinical cases for each age group. 0 10 20 30 40 50 60 <1818 to 3536 to 54>55Age GroupNumber of Clinical Cases Figure 8. Age Comparison for WNND and WNF for 2003. This graph shows the age groups separated by WNND (red bar) and WNF (blue bar). 0 5 10 15 20 25 30 35 40 <1818 to 3536 to 54>55<1818 to 3536 to 54>55Number of Clinical Cases West Nile Fever West Nile Neuroinvasive Disease Age GroupAge Group

PAGE 50

35 Figure 9. WNND by Gender and Age Group for 2001-2003. This graph shows the number of ma les (blue bars) and the females (magenta bars) for all age groups with WNND. 0 5 10 15 20 25 30 35 40 45 50 FMFMFMFM <1818 to 3536 to 54>55 Total Number

PAGE 51

36 West Nile Incidence Rates Figures 10a through 10c de pict incidence rates for counties with West Nile associated clinical cases for 2001-03. The graphs were also grouped by region and generally ordered by location from west (left) to east (right) and north (top) to south (bottom) shown in Appendix XI. The highest incidence rates were seen in Madison, Baker, and Gulf counties resp ectively, and the lowe st reported case rates were seen in Palm Beach and Dade counties. These gra phs show a clustering of WN disease by location. Comparison of counties located adjacen t to one another, shows higher incidence counties tend to be located together. This pattern shows the highe st incidence in the panhandle and adjacent areas; minimal activity in most of the eastern peninsular counties; moderate in the southernmost counties. Stat ewide incidence rates increased linearly from year to year with an RP2P = 0.9907 as shown in Figure 11. A ppendix XI shows the raw data for the number and incidence of clin ical cases in each county by year.

PAGE 52

37 Figure 10a. West Nile Incidence Rate by County, Region and Location for 2001. The x-axis represents the countie s listed in legend by color. 0 2 4 6 8 10 12Incidence per 100,000 Washington Leon Jefferson Madison Duval Putnam Marion Sarasota Monroe Palm Beach Panhandle North South Central Figure 10b. West Nile Incidence Rate by County, Region and Location for 2002. The x-axis represents the countie s listed in legend by color. 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5Incidence per 100,000 Escambia Santa Rosa Jackson Alachua Baker Clay Duval Marion Sumter Lake Orange Brevard Polk Highlands Hillsborough Manatee Sarasota Lee Collier Dade Palm Beach Panhandle North South Central

PAGE 53

38 Figure 10c. West Nile Incidence Rate by County, Region and Location for 2003. The x-axis represents the count ies listed by color in legend. Figure 11. West Nile Clinical Statew ide Incidence Rate for 2001 through 2003. This graph shows the incidence for each consecutive year represented by the blue diamonds. The black line is th e fitted regression line for the three years. y = 0.1415x 0.0763 R2 = 0.9907 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4200120022003Incidence per 100,000 0 2 4 6 8 10 12 14 16 18 20 Incidence per 100,000 Escambia Santa Rosa Okaloosa Holmes Washington Bay Calhoun Gul f Lafayette Suwannee A lachua Union Duval St Johns Marion Citrus Pasco Sumte r Volusia Seminole DeSoto Sarasota Lee Collie r Monroe Dade Broward Palm Beach Panhandle North South Central

PAGE 54

39 Surveillance Temporal Distribution Figure 12 shows the peak transmission peri od for total number of clinical cases, which begins at week 25 and ends at w eek 50. The transmission period for sentinel chicken positives (Figure 13) was determined by pooling the totals of sentinel chicken positives by week for each year. The peak transmission period begins at the 24PthP week and ends at the 52PndP week. Outliers (from week 1 through week 5) are present in the data, but were excluded from the analysis because they were probably residual positives from the previous year. The tempor al graph of the number of pos itive dead birds (Figure 14) indicates a peak transmission period from week 24 to week 53 and mosquito pools beginning at the 20PthP week and ending in the 50PthP week (Figure 15). An overall transmission period beginning at week 20 a nd ending with week 53 created by combining all four sets of surveillan ce data. The line graphs plotted for the averages of each surveillance type for the combined years (2001-2003) with the calculated moving averages shown in Appendix XI indicates co mparatively similar systematic patterns, indicating that the data ’s distribution is not due to chance alone.

PAGE 55

40 Figure 12. Sentinel Chicken Positive West Nile Weekly Seroconversion Rate by Week for 2001 – 2003. 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.091 3 5 7 9 1 1 13 1 5 17 1 9 21 23 25 2 7 29 3 1 33 35 37 3 9 41 43 45 47 49 5 1WeekPositive Seroconversion Rate 2001 2002 2003 Peak Detection outliers Figure 13. West Nile Clinical Cases by Week for 2001—2003. 0 2 4 6 8 10 121 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51WeekNumber of Clinical Cases 2001 2002 2003

PAGE 56

41 Figure 14. Total Number of West Nile Po sitive Dead Birds by Week for 2001 – 2003. 0 20 40 60 80 100 120 1401 3 5 7 9 11 13 1 5 17 1 9 21 23 25 27 29 3 1 33 3 5 37 39 41 43 45 4 7 49 5 1WeekNumber Positive 2001 2002 2003 Peak Detection Figure 15. Total Number of West Nile Po sitive Mosquito Pools by Week for 2001 –2003. 0 2 4 6 8 10 12 14 16 18 201 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51WeekNumber Positive 2001 2002 2003 PkDtti Peak Detection

PAGE 57

42 Clinical Cases Figure 16 represents the number of clinic al cases per week an d separated by year. The data reveals a different timing of peak incidence of cases each year. The number of cases does not show consistent a patte rn but occur sporadically within the transmission season from year to year. However, the cumulative weeks for the individual years show a dist ribution similar to a bell sh aped curve. During 2001, the highest weekly number of confirmed hu man cases of WNV was two, occurring in weeks 28, 36 and 40. In 2002 a high of four cases occurred at weeks 39 and 44. Eleven cases was the highest number of week ly clinical cases, and occurred during weeks 35, 38 and 39 during 2003. The total cases of West Nile Dis ease in humans in 2001 there were 12, in 2002 there were 35 and in 2003 there were 92. The overall distribution of the number of diagnosed cases is shown as a box plot in Figure 17. For these samples the box plot demonstrates an increase in both the average and total numbers of clinical cases by week fr om 2001 to 2003. Far outliers are seen in 2001 and 2003 with near outliers observed in each year.

PAGE 58

43 Figure 16. Number of Clinical Cases of West Nile per Week during Peak Transmission Season. 0 2 4 6 8 10 1220 22 24 2 6 28 30 32 34 3 6 38 40 42 44 4 6 48 50 52WeekNumber of Clinical Cases 2001 2002 2003 Figure 17. Box-plot for the Average a nd Total Number of Clinical Cases for 2001, 2002 and 2003. This graph shows a distinct increas e from 2001 to 2003 in the average and total numbers of cases by week per year. -4 -2 0 2 4 6 8 10 12 2001 Clinical Cases2002 Clinical Cases 2003 Clinical CasesNumber of Clinical Cases

PAGE 59

44 Sentinel Chicken Surveillance There were a total of 204 sentinel sites with a total of 2,128 sentinel chickens monitored during 2001 of which 202 seroconvert ed for WN antibodies (Stark, 2001). In 2002 there were a total of 202 se ntinel sites with 1,093 WN sero conversions out of a total of 3,356 sentinel chickens (Stark, 2002). Th ere were 289 sentinel sites with 4,361 total chickens monitored showing 1,343 WN positive seroconversions during 2003 (Stark, 2003). The graph depicted in figure 18 for sentin el chicken serosurvei llance has several different peak weeks for each year. The highe st seroconversion rate for 2001, seen at week 46, was 0.0235 and was significantly lowe r than observed in the following two years. The highest rate for 2002 was 0.0822 at week 44, with other weeks of notably higher rates at weeks 38, 39, 41 and 43. In 2003 the maximum seroconversion rate was 0.1127 during week 47, and comparably high rate s by week also occurred at weeks 33, 34, 35, 37 and 38. The overall distribution of the seroconve rsion rates of the sentinel chicken surveillance data by week is shown in the box pl ot (figure 19). This illustrates an increase in the average and total rates of seroconve rsion for each consecutive year from 2001 to 2003. There are near outlier observation for the seroconversion ra te in 2001 and 2003 with far outliers in 2001 only. Ther e were no observed outliers in 2002.

PAGE 60

45 Figure 18. Sentinel Chicken Surveillance Rate during Transmission Season for 2001 2003. 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.0920 22 24 26 28 3 0 3 2 34 36 38 40 42 44 46 4 8 5 0 52WeekPositive Seroconversion Rate 2001 2002 2003 Figure 19. Box-plot of Sentinel Su rveillance rate for 2001, 2002 and 2003. This graph shows a distinct increas e from 2001 to 2003 in the average and total seroconversion rate per year. -0.02 0 0.02 0.04 0.06 0.08 2001 Sentinel Seroconversion Rate 2002 Sentinel Serconversion Rate 2003 Sentinel Seroconversion RateSeroconversion Rate

PAGE 61

46 Avian Surveillance There were a total of 7,675 total dead av ian samples submitted with 1,106 testing positive for WNV (Stark, 2001). During 2002 th ere were a total of 4,020 of which 439 tested positive for WNV (Stark, 2002). There were 2,320 dead birds submitted during 2003 with 486 testing positive for WNV (Stark, 2003). Figure 20a, statewide total for positiv e dead avian specimen, shows some consistent curves for the peak period during 2001, 2002 and 2003. The largest numbers of cases occurred from weeks 32 to 42 each year. The highest number of dead bird positives in a week for 2001 was 126 which was greater than the numbers observed in 2002 and 2003. The greatest number of positives for 2002 was 36, at week 32, other notably high numbers were seen during weeks 41 and 35. In 2003 the greatest number of positives was 52, observed during week 32, with high numbers also occurring from week 33 to week 35. Figure 20b shows the observed percent of positive dead avian specimen out of the total number submitted. The highest rate obs erved for the all the weeks in 2001, 2002 and 2003 was week 24 in 2001 which was 100% positive out of the submitted birds (1 submitted and 1 positive). The overall rates were highest in 2003 with the highest rate at 0.6 during week 36. The 2001 and 2002 rates had si milar observed rates with the highest in 2001 (excluding week 23) 0.25 during week 52 and 0.28 during week 29. Figures 21 a, b and c show the total number of birds submitted with percent of positive samples for each year (2001, 2002 and 2003). In 2001 (figure 21a), the percent increases variably with an increase in to tal birds submitted except for the first positive (week 24). During 2002 shown in figure 21b the graph shows sporadic increases in the

PAGE 62

47 percent positive however, the overall increase is greater during w eeks with a larger number of submissions. Excluding the first observation in 2001 the percent positives from 2001 and 2002 have similar overall patterns with highest percents at 19% and 22% respectively. The graph for 2003 (figure 21c) sh ows a more consistent relationship with the observed temporal distributi on and exhibits incr eased percent positives with amplified dead bird submitting. All three years had a si gnificant positive correlation between the percent of positive dead birds and the tota l number of submitted (p-value < 0.0001). The specific Pearson’s correlation coefficient for each year was 0.79 for 2001, 0.70 for 2002 and 0.76 for 2003. The box plot shown in Figure 22 for the numbers of positive dead birds does not follow the previous distribu tions seen with clinical cases and sentinel chicken serosurveillance. The box plot shows that average number of positive dead birds by week decrease from 2001 to 2002 and increase from 2002 to 2003. The total number of positives indicated by the plots follow the same trend however, near outliers are seen in 2001 and 2002 with far outliers observed in 2001. There were not outlier observations in 2003. The total submitted dead avian specimen shown in figure 23 for the entire year indicates the submissions of dead birds begi n to increase at approximately week 24. The highest numbers of submissions for all thr ee years were observed between week 31 and 41. The a highest total submitted dead birds for the three years were 732 in 2001, 254 in 2002 and 117 in 2003. The box plot in figure 24 for total subm itted dead birds decreased from 2001 to 2003. The overall numbers submitted are significantly higher in 2001 and decline

PAGE 63

48 dramatically over 2002 and 2003. For the total dead birds submitted the near outliers are seen in all three years with one far outlier observation for 2002.

PAGE 64

49 Figure 20a. Total Number of West Nile Po sitive Dead Birds during Transmission Season for 2001 – 2003. This graph shows the highest numbers of positives during 2001. The number of positives per week was higher in 2003 compared to 2002. 0 20 40 60 80 100 120 1402 0 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52WeekNumber Positive 2001 2002 2003 Figure 20b. Dead Birds Percent Positive. Percent positive was calculated by dividing the number of WN positive dead birds by the total number submitted. 0 0.2 0.4 0.6 0.8 1 1.220 22 24 26 28 30 32 34 36 38 40 42 4 4 4 6 48 50 52WeekRate of Dead Avian Positives 2001 2002 2003

PAGE 65

50 Figure : 21a. The Total Dead Birds Submi tted and the Percent of Positive Dead Birds by Week for 2001. Percent positive was calculated by dividing the number of WN positive dead birds by the total number submitted. 0 100 200 300 400 500 600 700 8001 3 5 7 9 1 1 13 1 5 1 7 19 2 1 2 3 25 2 7 29 3 1 3 3 35 3 7 3 9 4 1 4 3 45 4 7 4 9 51WeeksTotal Number Dead Birds Submitted0 0.2 0.4 0.6 0.8 1 1.2Percent Positive Total Submitted Dead Birds Percent Positive Figure : 21b. The Total Dead Birds Submi tted and the Percent of Positive Dead Birds by Week for 2002. Percent positive was calculated by dividing the number of WN positive dead birds by the total number submitted. 0 50 100 150 200 250 3001 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 4 1 43 45 47 49 5 1WeeksTotal Number Dead Birds Submitted0 0.05 0.1 0.15 0.2 0.25Percent Positive Total Submitted Dead Birds Percent Positive

PAGE 66

51 Figure : 21c. The Total Dead Birds Submi tted and the Percent of Positive Dead Birds by Week for 2003. Percent positive was calculated by dividing the number of WN positive dead birds by the total number submitted. 0 20 40 60 80 100 120 1401 3 5 7 9 11 1 3 15 17 1 9 21 23 25 27 29 31 3 3 3 5 37 39 41 43 45 47 4 9 51WeeksTotal Number Dead Birds Submitted0 0.1 0.2 0.3 0.4 0.5 0.6Percent Positive Total Submitted Dead Birds Percent Positive Figure 22. Box-plot the Number of Positive Dead Birds for 2001, 2002 and 2003. -40 -20 0 20 40 60 80 100 120 140 2001 Positive Dead Birds2002 Positive Dead Brids2003 Positive Dead Birds Number of Positive Dead Birds

PAGE 67

52 Figure 23. Total Dead Birds Submitted for 2001 – 2003. 0 100 200 300 400 500 600 700 8001 3 5 7 9 11 1 3 1 5 17 19 2 1 2 3 2 5 27 2 9 3 1 3 3 35 3 7 3 9 41 4 3 4 5 4 7 4 9 5 1 5 3WeekTotal Birds Submitted 2001 2002 2003 Peak Transmission Season Figure 24. Box-plot for Total Subm itted Dead Birds for 2001, 2002 and 2003. -200 -100 0 100 200 300 400 500 600 700 800 2001 Total Dead Birds2002 Total Dead Birds2003 Total Dead BirdsTotal Dead Birds

PAGE 68

53 Mosquito Surveillance In 2001 there were 1,378 submitted mosquito pools of which ten tested positive for WNV (Stark, 2001). The total number of mosquito pools submitted in 2002 was 3,886 with 25 testing positive for WNV (Star k, 2002). There were 6,292 mosquito pools submitted during 2003 with 42 positive for WNV (Stark, 2003). The histogram pictured in Figure 25 for positive mosquito pools has the lowest number of positives among all the surveillanc e systems. The collective distribution for 2001, 2002 and 2003 is similar with the curves showing a range for the peak positive activity from week 29 to week 35. The highest number of mosquito pool positives for 2001 seen at week 39 was two. The highest number of positives for 2002 was four at weeks 31 and 32 with three positives during 29 and 35. Th e data for 2003 was notably greater than the previous years with 18 tota l positives at week 32 and high numbers of positives during week 31 and 33. Figure 26 shows the observed percent of positive mosquito pools out of the total number of mosquito pools submitted for 2001-2003. The highest percent positives observed during 2001 was 6.7% at week 40, 2002 was 3.5% at week 43 and 2003 was 6.5% at week 32. The overall rates were higher in 2001 and 2003. The overall distribution for the years observed were spor adic, exhibiting no consistent patterns. Figures 28 a, b and c show the total number of mosquito pools submitted with the percent positive for each year (2001, 2002 and 2003). The graph for 2001 (figure 28a) shows that the percent appears sporadic in dicating that there is no correlation with number submitted (Pearson’s Correlation Coefficient=0.31 and p-value=0.097). The 2002 distribution of percent positive mosquito pools and total submitted shown in figure 28b

PAGE 69

54 has the best correlation (Pearson’s Correla tion Coefficient=0.37 and p-value=0.0087) of the three years. The 2003 distribution has an overall higher number submitted mosquito pools compared to the previous two years. Pearson’s correlation coefficient for 2003 was 0.38 with a significant p-value of 0.0415. The highest number of submitted mosquito pools was 181 during weeks 28 and 46 in 2001. The highest in 2002 was 268 during week 32 and 221 during week 25 in 2003. The ove rall data for mosquito pool positives shown in the box plot (figure 27) illustrates an in crease in the average and total numbers of positives detected per week for 2001, 2002 and 2003. There were near outlier observations in 2002 and 2003 with far outliers in 2001 and 2003.

PAGE 70

55 Figure 25. Total Positive Mosquito Pools during Transmission Season for 2001 – 2003 0 2 4 6 8 10 12 14 16 1820 22 24 2 6 28 30 3 2 34 3 6 38 40 42 44 46 48 5 0 52WeekNumber Positive Mosquito Pools 2001 2003 2004 Figure 26. Percent of Positive Mosquito Pools during Transmission Season for 2001 – 2003. 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.072 0 22 2 4 2 6 2 8 3 0 32 3 4 3 6 38 4 0 4 2 4 4 4 6 48 5 0 5 2WeekPercent Positive Mosquito Pools 2001 2002 2003

PAGE 71

56 Figure 27. Box-plot for Total Positive Mosquito Pools for 2001, 2002 and 2003. -5 0 5 10 15 20 2001 Positive Mosquitoes 2002 Positive Mosquitoes 2003 Positive Mosquitoes Number of Positve Mosquitoes Figure : 28a. The Total Positive Mosqui to Pools and the Percent of Positive Mosquito Pools by Week for 2001. Percent positive was calculated by dividing the number of WN positive mosquito pools by the total number submitted. 0 0.01 0.02 0.03 0.04 0.05 0.06 0.071 3 5 7 9 11 13 1 5 17 19 2 1 2 3 25 27 29 31 3 3 35 37 3 9 4 1 43 4 5 4 7 49 51WeeksTotal Number Mosquito Pools Submitted0 20 40 60 80 100 120 140 160 180 200Percent Positive Percent Positive Total Submitted Mosquito Pools

PAGE 72

57 Figure : 28b. The Total Positive Mosqui to Pools and the Percent of Positive Mosquito Pools by Week for 2002. Percent positive was calculated by dividing the number of WN positive mosquito pools by the total number submitted. 0 50 100 150 200 250 3001 3 5 7 9 1 1 13 15 1 7 19 21 2 3 2 5 2 7 2 9 3 1 3 3 35 37 3 9 41 43 4 5 4 7 49 5 1 5 3WeeksTotal Number Mosquito Pools Submitted0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04Percent Positive Total Submitted Mosquito Pools Percent Positive Figure : 28c. The Total Positive Mosqui to Pools and the Percent of Positive Mosquito Pools by Week for 2003. Percent positive was calculated by dividing the number of WN positive mosquito pools by the total number submitted. 0 50 100 150 200 2501 3 5 7 9 11 1 3 1 5 1 7 1 9 21 23 2 5 2 7 2 9 31 33 3 5 3 7 39 4 1 43 45 47 4 9 5 1 53WeeksTotal Number Mosquito Pools Submitted0 0.01 0.02 0.03 0.04 0.05 0.06 0.07Percent Positive Total Submitted Mosquito Pools Percent Positive

PAGE 73

58 Regional Surveillance Panhandle Region To show results within similar physiogra phic climates, Florida was divided into four geographic regions: Panhandle, Northern, Central and Southern regions. Figures 29a, 29b and 29c correspond to the Panhandle Region, while Figures 30-32 a, b and c correspond to the Northern, Central and Sout hern regions respec tively. The combined year average and the moving average plots for each region are shown in Appendix XI. The graph for each region (Panhandle, North, Central and South) and average positive outcome pooled for 2001-2003 for Humans, Dead Birds, Mosiquito Pools and Sentinel Chickens by week each show similar pa tterns compared to the calculated moving average. This relationship would indicate that the distributions seen within each region for the surveillance types are most likely not due to chance. Figure 29a depicts positive mosquito pool s, dead birds and sentinels for the Panhandle region during 2001 with human cases pi n-pointed. There were two clusters of human cases. Positive dead birds and mosquito pool first occurred about four weeks prior to the first human case and the first increase in positive sentinel chickens was seen two weeks prior to the first case. A second peak in positive mosquito pools and dead birds occurred two to three weeks pr ior to the second cluster of human cases. Nevertheless, in the panhandle, the peak for positive sentinels be gan approximately one week after the last confirmed human case. Figure 29b shows the regional graph of the 2002 Panhandle region indicates two clusters of human cases approximately se ven weeks apart, each with corresponding

PAGE 74

59 increases in the number of sentinel and dead bird positives before the initial human case occurred. However, the mosquito surveillance program did not detect any positives for this region during 2002. This is consistent with the current mosquito trends where there has been little or no detection for West Nile from submitted samples. Figures 29c illustrates similar positive distri butions in the dead bird and sentinel chicken flock populations in the panhandle region for 2003. There were a few mosquito pool positives detected, but these numbers are small (week 28 had 1 positive pool and week 32 had 13 positive pools) and did not co rrespond to the distribution trends. There were a large number of clinical cases ( 31) spread throughout the entire transmission period making individual analys is of separate peaks within the period impossible. The graph also shows a lower total number of dead bird positives for the surveillance systems than the previous years, but a much higher to tal for the number of se ntinel positive. The first clinical WN case occurred one week prior to the detecti on of virus or antibodies in sentinels, dead birds or mosquitoes. With the exception of mosquitoes, the virus was detected at an increasing rate within the w eeks following the first clinical case; however, dead birds had a bimodal distribution with a decrease in the number positive midway through the transmission season.

PAGE 75

60 Figure 29a. The Number of Positive Avian, Sentinels and Mosquitoes with human cases for the Panhandle Region during 2001. The two axes, y (number of positive mosquito pools) and z(number of positive dead avian) have different maximums, minimums and scales for each. The breaks in the lines for pos itive avian represent no specimen submitted for that week. 0 2 4 6 8 10 12 14 16 2426283032343638404244464850 WeekNumber Positive Mosquito & Sentinel0 10 20 30 40 50 60Number Positive Avian Positive Mosquito Positive Sentinel Positive Avian Human CaseHuman Cases Figure 29b. The Number of Positive Avian, Sentinels and Mosquitoes with human cases for the Panhandle Region during 2002. The two axes, y (number of positive mosquito pools) and z(number of positive dead avian) have different maximums, minimums and scales for each. The breaks in the lines for pos itive avian represent no specimen submitted for that week. 0 0.5 1 1.5 2 2.5 3 3.5 242628303234363840424446485052 WeekNumber of Positive Mosquitoes and Sentinels0 5 10 15 20 25Number of Positive Avian Positive Mosqiutoes Positive Sentinels Positivel Avian Human CasesHuman Cases

PAGE 76

61 Figure 29c. The Number of Positive Avian, Sentinels and Mosquitoes with human cases for the Panhandle Region during 2003. The two axes, y (number of positive mosquito pools) and z(number of positive dead avian) have different maximums, minimums and scales for each. The breaks in the lines for pos itive avian represent no specimen submitted for that week. 0 5 10 15 20 25 30 35 242628303234363840424446485052 WeekNumber of Positive Mosquitoes and Sentinels0 5 10 15 20 25 30 35 40 45 50Number of Positive Avian Positive Mosquitoes Positive Sentinels Positive Avian Human CaseHuman CasesHuman Cases Human Cases

PAGE 77

62 Northern Region The Northern region of Florida for the year 2001 (Figure 30a) shows a corresponding increase in the rate of West Nile positive sentinels and the number dead bird with the first three (of four) clinical cases. There wa s one case detected after the sentinel and dead bird positives peaked. There were no positive mosquito pools for this region during 2001. In 2002 (Figure 30b), there were positive sentinel outcomes before the first human case and numbers continued to increa se throughout the transmission period with three peaks each approximately one week prio r to each of the clusters of human cases. Dead bird positives were detected in low numbers before the first human case with peaks prior to or during the separate clusters of human cases. Mo squito pools provided an early indication of human cases for this period as well, peaking approximately two weeks after the first clinical case and four weeks prio r to the second case. The graph for the northern region during 2003 (Figure 30c) shows an increase in the number of confirmed human cases. Ther efore, specific peak evaluation for the surveillance systems within the time period for human transmission was not possible. Nevertheless, positive dead bi rds and sentinels were observe d before the initial human case, and increased and decreased with the occurrence of clinical cases. No mosquito pools were positive during this period for northern Florida.

PAGE 78

63 Figure 30a. The Number of Positive Avian, Sentinels and Mosquitoes with human cases for the Northern Region during 2001. The two axes, y (number of positive mosquito pools) and z (number of positive dead avian) have different maximums, minimums and scales for each. 0 5 10 15 20 25 242628303234363840424446485052 WeekNumber of Positive Mosquitoes & Sentinels0 10 20 30 40 50 60 70 80Number of Avian Positive Mosquitoes Positive Sentinel Positive Avian Human Case Human CaseHuman CaseHuman Case Figure 30b. The Number of Positive Avian, Sentinels and Mosquitoes with human cases for the Northern Region during 2002. The two axes, y (number of positive mosquito pools) and z(number of positive dead avian) have different maximums, minimums and scales for each. 0 5 10 15 20 25 30 35 40 242628303234363840424446485052 WeekNumber of Positive Mosquitoes & Sentinels0 2 4 6 8 10 12 14 16 18 20Number of Positive Avian Postive Mosquitoes Positive Sentinels Positive Avian Human Cases Human Cases Human CaseHuman Cases

PAGE 79

64 Figure 30c. The Number of Positive Avian, Sentinels and Mosquitoes with human cases for the Northern Region during 2003. The two axes, y (number of positive mosquito pools) and z(number of positive dead avian) have different maximums, minimums and scales for each. The breaks in the lines for pos itive avian represent no specimen submitted for that week. 0 5 10 15 20 25 30 35 242628303234363840424446485052 WeekNumber of Positive Mosquitoes and Sentinels0 1 2 3 4 5 6 7 8 9Number of Positive Avian Positive Mosquitoes Positive Sentinels Positive Avian Human CaseHuman CasesHuman Cases Human Cases

PAGE 80

65 Central Region The distribution of the data from 2001 for the central Florida region figure 31a demonstrates an increase simultaneously with and following the first and only clinical case diagnosed. Positive dead birds were det ected the same week as the human case and increased after. One sentinel positive was obs erved three weeks after the clinical case. There were two positive mosquito pools de tected 15 weeks after the human case. Evaluation of the graph for the centra l region in 2002 (figure 31b) shows a different picture than 2001. There was a spik e in the number of positive mosquito pools and with an increase in the number of positives among the sentinel population approximately eight weeks before the first human case was diagnosed. There was a small increase in positive dead birds four week before the first case. All three surveillance systems showed a spike in numbers four to eight weeks before the second cluster of clinical cases. However, mosquito pools had only one positive result. The number of dead bird positives peaked five weeks before the start of the second group of cases whereas the sentinels had a significant eleva tion at week 41, a week before the next cluster of clinical cases. Figure 31c for the 2003 central Florida region shows an increase in the number of positive sentinel chickens starting four week s before the first human case and positives began tapering off after the second case a nd began increasing again before the third human case. The numbers of dead bird positives were extremely low for this year directly corresponding with the overall numbers of birds submitted for 2003. Even with the small numbers, there were positives detected before the first clinical case. Two mosquito pools were positives, one and two weeks before the last human case.

PAGE 81

66 Figure 31a. The Number of Positive Avian, Sentinels and Mosquitoes with human cases for the Central region during 2001. The two axes, y (number of positive mosquito pools) and z(number of positive dead avian) have different maximums, minimums and scales for each. The breaks in the lines for pos itive avian represent no specimen submitted for that week. 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 242628303234363840424446485052 WeekNumber of Positive Mosquitoes and Sentinels0 1 2 3 4 5 6Number of Positive Avian Positive Mosquitoes Positve Sentinels Positive Avian Human Case Figure 31b. The Number of Positive Avian, Sentinels and Mosquitoes with human cases for the Central Region during 2002. The two axes, y (number of positive mosquito pools) and z(number of positive dead avian) have different maximums, minimums and scales for each. 0 5 10 15 20 25 30 35 40 45 50 242628303234363840424446485052 WeekNumber of Positive Mosquitoes and Sentinels0 1 2 3 4 5 6 7 8 9 10Number of Positive Avian Positive Mosquitoes Positive Sentinels Positive Avian Human CasesHuman CasesHuman Case

PAGE 82

67 Figure 31c. The Number of Positive Avian, Sentinels and Mosquitoes with human cases for the Central Region during 2003. The two axes, y (number of positive mosquito pools) and z(number of positive dead avian) have different maximums, minimums and scales for each. The breaks in the lines for pos itive avian represent no specimen submitted for that week. 0 5 10 15 20 25 30 35 242628303234363840424446485052 WeekNumber of Positive Mosquitoes and Sentinels0 0.5 1 1.5 2 2.5Number of Positive Avian Positive Mosquitoes Positive Sentinels Positive Avian Human CaseHuman CaseHuman Case

PAGE 83

68 Southern Region The graph for the southern region of Flor ida in 2001 is shown in figure 32a. This shows the initial clinical case before any su rveillance positives were detected. However, two weeks prior to the second case positive WN birds were detected and one week prior WN was detected in mosquito pools. Sen tinels showed an increase in WN antibody detection nine weeks before the second human case. The distribution for Florida’s southern re gion in 2002 is illustrated in figure 32b showed a small increase in positive sentinels and dead birds before the first human case was diagnosed. An increased detection of West Nile was seen in sentinel and dead bird positives two to three weeks before the second and third cases were seen. There were two mosquito pool positives detected f our weeks before the third case. Figure 32c, the graph of the southern re gion for 2003, shows a distribution similar to the previous two years. There were a fe w positive sentinels and three positive dead birds detected before the first case. They both increased over a period coinciding with diagnosed cases and decreased after the last clinical case. There were a few mosquito pools that tested positive around the same w eek of the last huma n case and ten weeks after the last case.

PAGE 84

69 Figure 32a. The Number of Positive Avian, Sentinels and Mosquitoes with human cases for the Southern Region during 2001. The two axes, y (number of positive mosquito pools) and z(number of positive dead avian) have different maximums, minimums and scales for each. 0 0.5 1 1.5 2 2.5 3 3.5 242628303234363840424446485052 WeekNumber of Positive Mosquitoes & Sentinels0 2 4 6 8 10 12 14 16 18Number of Positive Avian Positive Mosquitoes Positive Sentinels Positive Avian Human Case Human Case Figure 32b. The Number of Positive Avian, Sentinels and Mosquitoes with human cases for the Southern Region during 2002. The two axes, y (number of positive mosquito pools) and z (number of positive dead avian) have different maximums, minimums and scales for each. 0 5 10 15 20 25 30 35 242628303234363840424446485052 WeekNumber of Positive Mosquitoes and Sentinels0 0.5 1 1.5 2 2.5 3 3.5 4 4.5Number of Positive Avian Positive Mosquitoes Positive Sentinels Positive Avian Human CaseHuman CaseHuman Case

PAGE 85

70 Figure 32c. The Number of Positive Avian, Sentinels and Mosquitoes with human cases for the Southern Region during 2002. The two axes, y (number of positive mosquito pools) and z (number of positive dead avian) have different maximums, minimums and scales for each. The breaks in the lines for pos itive avian represent no specimen submitted for that week. 0 5 10 15 20 25 30 35 40 242628303234363840424446485052 WeekNumber of Positive Mosquitoes and Sentinels0 1 2 3 4 5 6 7 8 9 10Number of Positive Avian Positive Mosquitoes Positive Sentinels Positive Avian Human CasesHuman CasesHuman Cases

PAGE 86

71 Clinical Cases by Region The number of clinical cases by region fi gure shows the epicenter in the northern region for 2001 and 2002 and the panhandle region for 2003. The graph also shows the central region had the least number of cases in 2001 and 2003 with lowest number of cases for 2002 in the southern region. All re gions showed an increase in number of human cases each year, with the exception of the central region which had a decrease in clinical cases for 2003.

PAGE 87

72 Figure 33. Number of Clinical Cases by Region for 2001-2003. Regions are organized north to sout h and east to west. Each year is indicated in the legend by a different color. 0 5 1 0 15 20 25 30 35Panhandle NorthCentralSouthNumber of Clinical Case s 2001 2002 2003

PAGE 88

73 Spatial Analysis Positive Dead Birds and Clinical Cases The overall trend for dead bird positives appears to coincide with the number of clinical cases diagnosed per county. The counties that show ed the highest number of positive dead birds also had the highest number of clinical cases. Figure 34a shows the highest total number of clinical cases in Escambia and Bay counties as 19 and 14 respectively. These counties also had the hi ghest range of positive dead birds between 120-206 total. Other counties of interest for th e collective year spat ial analysis includes Duval, Dade, Santa Rosa, Okaloosa and Ma rion which all have higher numbers of clinical cases and positive d ead birds than the other count ies observed. The correlation plot (figure 34b) shows a significant positive co rrelation between the number of clinical cases and the number of positive dead birds by county for the three years (Pearson’s correlation coefficient=0.76 and p-value<0.0001) There were apparent outliers in the scatter plot which may have influenced the f it of the line. A scatter plot excluding these observations is shown in figure 34c. The ou tlier observations were determined by a boxplot which indicated two separate weeks w ith the highest number of clinical cases numbers (14 and 19). The scatter plot showed less of a positive correlation then what was seen in the plot incl uding the outlier data. Figures 35a 36a, and 37a show the dead bird distribution an d diagnosed clinical cases by county for 2001, 2002 and 2003 whereas figures 35b, 36b and 37b show the scatter plot correlation between the number of positive dead birds and clinical cases by county for each year. In 2001 (Figure 35a) th ere were insufficient number of clinical cases to accurately evaluate the correlations (figure 35b) with positive dead birds. The

PAGE 89

74 Pearson’s coefficient correlation was 0.23 with a p-value of 0.0651. However, 2002 and 2003 show similar trends to the combined ye ar map and correlati on. Box-plot analysis did not indicate any outlier observations. In 2002 depicted in Figure 36a there are two counties (Escambia and Marion) which show a correlation between the numbers of posi tive dead birds and the number of clinical cases. Escambia County had the highest numbe rs of human cases (seven) and positive dead birds (in the range from 35-121). Mari on County had the second highest number of human cases but shared the range for positive dead avian specimen (20-35) with several other counties that had fewer clinical cas es. The Pearson’s correlation coefficient was 0.92 with a p-value of < 0.0001 showing that th ere is a significant positive correlation with the increase in the number of positive d ead birds with an increase in the number of clinical cases. Outliers were present in the data determined by box-plot analysis. These observations were at three different weeks spec ifically excluded was the data points that had seven clinical cases with 121 positive dead birds and three clinical cases with 34 positive dead birds. The scatter plot excluding these observations is shown in figure 36c. The scatter plot showed a posit ive correlation however, it was lower then the scatter plot including the outliers. The map for 2003 shown in figure 37 a has four counties which showed corresponding increases with the range of numbers of positive dead birds and human cases. Figure 37b indicates ther e is a significant correlation between the number positive dead birds and diagnosed clin ical cases (Pearson’s correl ation coefficient= 0.91 and pvalue < 0.0001). The highest coun ties for clinical cases are Bay and Escambia with 14 and 12 clinical cases however, these were considered outlie r observations in the box-plot

PAGE 90

75 calculations. The scatter plot excluding th ese data are shown in Figure 37c which indicates a smaller positive co rrelation between clinical cas es and positive dead birds. These two counties also had the most pos itive dead birds ranging from 46-103. This corresponds to the total combin ed year data. Okaloosa and Dade counties had eight and six clinical cases respectively with the tota l number of dead birds ranging between 17 and 45. Broward County also had corresponding numbe rs with number of positive dead avian specimen (ranging from 3-5) a nd clinical cases (four). Seve ral counties had one, two or zero human cases where the total number of positive dead birds ranged from 0-2. Gulf, Lee and Lafayette counties submitted no dead birds but reported clinical cases.

PAGE 91

76 Figure 34a. Florida West Nile Cumulati ve Positive Dead Bird Distribution and Clinical Cases by County for 2001-2003. Intensity of color indicates numbers of West Nile positive birds detected. The areas designated without color si gnify counties that did not submit any dead birds for testing. 1 7 3 6 4 2 8 14 19 1 2 3 1 2 4 1 11 2 1 1 1 1 1 8 1 3 1 2 1 1 2 1 1 4 4 3 2 1 3 1 11Number of Positive Avian 0 12 13 28 29 57 58 119 120 206 No Submitted Specimen 0 200Miles N E W S# Represents the Number of Clinically Diagnosed Cases of West Nile

PAGE 92

77 Figure 34b. Scatter Plot for Correlation between the Number of Clinical Cases and Positive Dead Birds by W eek for the Pooled Years 2001-2003. The black line represents the fitted regression line for the number of clinical cases and positive dead avian. 0 50 100 150 200 25005101520Number of Clinical Cases Number of Positive Dead Avian Figure 34c. Scatter Plot for Correlation between the Number of Clinical Cases and Positive Dead Birds by W eek for the Pooled Years 2001-2003 excluding outlier observations. The black line represents the fitted regression line for the number of clinical cases and positive dead avian. 0 20 40 60 80 100 120 140024681012Number of Clinical Cases Number of Positive Dead Avian

PAGE 93

78 Figure 35a. Florida West Nile Positive De ad Bird Distribution and Clinical Cases by County for 2001. Intensity of color indicates numbers of West Nile positive birds detected. The areas designated without color si gnify counties that did not submit any dead birds for testing. 1 1 1 1 1 2 1 1Number of Positive Avian 0 6 7 15 16 29 30 64 65 109 No Submitted Specimen 0 200Miles N E W S# Represents the Number of Clinically Diagnosed Cases of West Nile

PAGE 94

79 Figure 35b. Scatter Plot for Correlation between the Number of Clinical Cases and Positive Dead Birds by Week for 2001. The black line represents the fitted regression line for the number of clinical cases and positive dead avian. 0 20 40 60 80 100 12000.511.522.5Number of Clinical Cases Number of Positive Dead Avian

PAGE 95

80 Figure 36a. Florida West Nile Positive De ad Bird Distribution and Clinical Cases by County for 2002. Intensity of color indicates numbers of West Nile positive birds detected. The areas designated without color si gnify counties that did not submit any dead birds for testing. 1 7 1 1 1 1 1 3 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 2 7Number of Positive Avian 0 3 4 9 10 19 20 34 35 121 No Submitted Specimen 0 200Miles N E W S# Represents the Number of Clinically Diagnosed Cases of West Nile

PAGE 96

81 Figure 36b. Scatter Plot for Correlation between the Number of Clinical Cases and Positive Dead Birds by Week for 2002. The black line represents the fitted regression line for the number of clinical cases and positive dead avian. 0 20 40 60 80 100 120 140012345678Number of Clinical Cases Number of Positive Dead Avian Figure 36c. Scatter Plot for Correlation between the Number of Clinical Cases and Positive Dead Birds by Week for 2002 excluding outlier observations. The black line represents the fitted regression line for the number of clinical cases and positive dead avian. 0 5 10 15 20 25 30 35 40 45 500123456789Number of Clinical Cases Number of Positive Dead Avian

PAGE 97

82 Figure 37a. Florida West Nile Positive Dead Avian Distribution and Clinical Cases by County for 2003. Intensity of color indicates numbers of West Nile positive birds detected. The areas designated without color si gnify counties that did not submit any dead birds for testing. 11181 2 2 4 1 1 1 1 1 6 1 2 1 2 2 1 4 6 1 2 3 1 12 10 14Number of Positive Avian 0 2 3 5 6 16 17 45 46 103 No Submitted Specimen 0 200Miles N E W S# Represents the Number of Clinically Diagnosed Cases of West Nile

PAGE 98

83 Figure 37b. Scatter Plot for Correlation between the Number of Clinical Cases and Positive Dead Birds by Week for 2003. The black line represents the fitted regression line for the number of clinical cases and positive dead avian. 0 20 40 60 80 100 1200246810121416Number of Clinical Cases Number of Positive Dead Avian Figure 37c. Scatter Plot for Correlation between the Number of Clinical Cases and Positive Dead Birds by Week for 2003 excluding outlier observations. The black line represents the fitted regression line for the number of clinical cases and positive dead avian. 0 5 10 15 20 25 30 35 40 45 50024681012Number of Clinical Cases Number of Positive Dead Avian

PAGE 99

84 Positive Mosquito Pools and Clinical Cases Mosquito surveillance was perfor med in 23 counties during 2001—2003. The overall distribution of positive mosquito pool s did not correlate with the number of clinical cases. However, there were some si gnificant associations for a few individual counties. The counties participating in the st ate mosquito surveillance program are listed in the table found in Appendix VIII. The combined year data for West Nile positive mosquito pools (figure 38a) showed two counties (Escambia and Gulf) with positive mosquito pools ranging from 1 to 3 that also had clinical cases. However, tw elve out of the 23 pa rticipating counties had no WNV positive mosquito pools, though some of these counties (Lee, Dade and Gulf) had more than four human cases each. Seve ral counties including St. Johns, Volusia, Pinellas, Collier, Monroe and Dade also had high levels of positive mosquito pools (1832) with a low number of clinical cases (l ess than four). The correlation scatter plot (figure 38b) for the combined years showed an inverse association between the number of positive mosquito pools and the number of clinical cases (Pearson’s correlation coefficient = -0.17 with a p-value = 0.4399). This indicates that there is no linear relationship with the number of positive mos quito pools and clinical cases by county for the combined years. The outliers determined by the box-plots were the observations with 11, 14 and 19 clinical cases. The scatter plot shown figure 38c for the cumulative years excluding these outliers show s a similar negative corre lation as compared to 38b. The individual year maps for 2001, 2002 and 2003 positive mosquito pool distribution and diagnosed clin ical cases by county are shown in figures 39a, 40a and 41a. During 2001 (Figure 39a) there were not enough clinical cases to accurately evaluate

PAGE 100

85 the possible trends between clinical cases and positive mosquito pools. The scatter plot (figure 39b) shows a positiv e correlation however, it is not significant (Pearson’s correlation coefficient = 0.49, p-value=0.0908). The outliers determined by the box-plots were the observations with two clinical cas es. The scatter plot (figure 39c) for the cumulative years excluding thes e outliers shows a negative correlation as opposed to the positive trend seen in figure 39b. The distribution between West Nile positive mosquito pools and clinical cases (figure 40a) does not appear to be associated in 2002. This is also shown in Figure 40b with correlation graph between positive mosquitoes and clinical cases indicating a negative slope trend line. The Pearson’s correlation coefficient is -0.50 w ith a p-value of 0.6667 therefore, no significant correlation is seen between number of positive mosquito pools and clinical cases. This means the numbers of clinical cases from the participating counties does not correspond with the number of positive mosquito pools. The outliers for 2002 were the observation with seven clin ical cases (calculate d by box-plot). The scatter plot (figure 40c) is a be tter fit for the data then th e plot including the outliers however, there is still a negative correlation. Figure 41a shows an association in Es cambia County for 2003 West Nile positive mosquito pool distribution and number of clinical cases. The specific numbers for Escambia County were 12 diagnos ed clinical cases with th ree positive mosquito pools. Although the overall correl ation shown in figure 41b is not significant (p-value=.8593 for the combined counties) there are a greater number of positive mosquitoes with a greater number of clinical cases for Escambia Count y. There were 19 participating counties out of which only four had positive mosquito pool s. The outliers calculated by the box-plot

PAGE 101

86 for 2003 clinical cases and mosquito positives were observations with 10 and 12 clinical cases. The scatter plot excluding these observa tions is shown in figure 41c which showed a similar positive correlation to the pl ot including the outliers (figure 41b).

PAGE 102

87 Figure 38a. Florida West Nile Positive Mo squito Pool Distribution and Clinical Cases by County for 2001-2003. Intensity of color indicates numbers of West Nile positive mosquito pools detected. The areas designated withou t color signify coun ties that did not submit any mosquito pools for testing. 1 7 3 6 4 2 8 14 19 1 2 3 1 2 4 1 11 2 1 1 1 1 1 8 1 3 1 2 1 1 2 1 1 4 4 3 3 2 1 11 1Number of Positive Mosquito Pools 0 1 3 4 7 8 17 18 32 No Submitted Specimen 0 200Miles N E W S# Represents the Number of Clinically Diagnosed Cases of West Nile

PAGE 103

88 Figure 38b. Scatter Plot for Correlation between the Number of Clinical Cases and Positive Mosquito Pools by Week for the Pooled Years 2001 2003. The black line represents the fitted regression line for the number of clinical cases and positive dead avian. -5 0 5 10 15 20 25 30 3502468101214161820Number of Clinical Cases Number of Positive Mosquito Pools Figure 38c. Scatter Plot for Correlation between the Number of Clinical Cases and Positive Mosquito Pools by W eek for the Pooled Years 2001 -2003 excluding outlier observations. The black line represents the fitted regression line for the number of clinical cases and positive dead avian. 0 5 10 15 20 25 30 350123456789Number of Clinical Cases Number of Positive Mosquito Pools

PAGE 104

89 Figure 39a. Florida West Nile Mosquito Pool Distributio n and Clinical Cases by County for 2001. Intensity of color indicates numbers of West Nile positive mosquito pools detected. The areas designated withou t color signify coun ties that did not submit any mosquito pools for testing. 1 1 1 1 1 2 1 1Number of Positive Mosquito Pools 0 1 2 3 6 No Submitted Specimen 0 200Miles N E W S# Represents the Number of Clinically Diagnosed Cases of West Nile

PAGE 105

90 Figure 39b. Scatter Plot for Correlation between the Number of Clinical Cases and Positive Mosquito Pools by Week for 2001. The black line represents the fitted regression line for the number of clinical cases and positive mosquito pools. 0 1 2 3 4 5 6 700.511.522.5Number of Clinical Cases Number of Positive Mosquito Pools Figure 39c. Scatter Plot for Correlation between the Number of Clinical Cases and Positive Mosquito Pools by Week for 2001 excluding outlier observations. The black line represents the fitted regression line for the number of clinical cases and positive mosquito pools. 0 0.5 1 1.5 2 2.500.20.40.60.811.2Number of Clinical Cases Number of Positive Mosquito Pools

PAGE 106

91 Figure 40a. Florida West Nile Mosquito Pool Distributio n and Clinical Cases by County for 2002. Intensity of color indicates numbers of West Nile positive mosquito pools detected. The areas designated withou t color signify coun ties that did not submit any mosquito pools for testing. 1 7 1 1 1 1 1 3 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 2 7Number of Positive Mosquito Pools 0 1 2 3 7 No Submitted Specimen 0 200Miles N E W S# Represents the Number of Clinically Diagnosed Cases of West Nile

PAGE 107

92 Figure 40b. Scatter Plot for Correlation between the Number of Clinical Cases and Positive Mosquito Pools by Week for 2002. The black line represents the fitted regression line for the number of clinical cases and positive mosquito pools. -2 0 2 4 6 8 10 12 14012345678Number of Clinical Cases Number of Positive Mosquito Pools Figure 40c. Scatter Plot for Correlation between the Number of Clinical Cases and Positive Mosquito Pools by Week for 2002 excluding outlier observations. The black line represents the fitted regression line for the number of clinical cases and positive mosquito pools. -4 -2 0 2 4 6 8 10 12 1400.511.522.533.5Number of Clinical Cases Number of Positive Mosquito Pools

PAGE 108

93 Figure 41a: Florida West N ile Mosquito Pool Distribut ion and Clinical Cases by County for 2003. Intensity of color indicates numbers of West Nile positive mosquito pools detected. The areas designated withou t color signify coun ties that did not submit any mosquito pools for testing. 1181 2 2 4 1 1 1 1 1 6 1 2 1 2 2 1 4 6 1 2 3 1 12 10 14Number of Positive Mosquito Pools 0 1 3 4 13 14 30 No Submitted Specimen 0 200Miles N E W S# Represents the Number of Clinically Diagnosed Cases of West Nile

PAGE 109

94 Figure 41b. Scatter Plot for Correlation between the Number of Clinical Cases and Positive Mosquito P ools by Week for 2003. The black line represents the fitted regression line for the number of clinical cases and positive mosquito pools. -5 0 5 10 15 20 25 30 350246810121416Number of Clinical Cases Number of Positive Mosquito Pools Figure 41c. Scatter Plot for Correlation between the Number of Clinical Cases and Positive Mosquito Pools by Week for 2003 excluding outlier observations. The black line represents the fitted regression line for the number of clinical cases and positive mosquito pools. -10 -5 0 5 10 15 20 25 30 350246810121416Number of Clinical Cases Number of Positive Mosquito Pools

PAGE 110

95 Sentinel Chicken Seroconversion Rates and Clinical Cases The cartographical represen tation of the serosurveill ance positive seroconversion rates for the sentinel chickens and observed clinical cases for the participating counties during 2001—2003 is shown in Figure 42a. Th ere were six counties that showed a greater number of clinical cas es with higher sentinel chic ken seroconversions to West Nile antibody positive. This distribution wa s present in Bay, Duval, Collier, Lee and Marion with the most clinical cases in Ba y County. Notably, each of these counties had a seroconversion rate between 0.043 and 0.062 for the overall three years. The combined year correlation scatter plot (figure 42b) showed no signifi cant association between the sentinel seroconversion rate a nd the number of clinical cases with a Pearson’s correlation coefficient of -0.09 and a p-value of 0.5891. Fi gure 42c shows the cumulative year scatter plot excluding outliers (determined by box-pl ot to be observations with 8, 11 or 14 clinical cases). The data showed a negative correlation similar to 42b which included the outliers. The maps for 2001, 2002 and 2003 sentinel chicken seroconversion rates and diagnosed clinical cases are shown in figures 43a, 44a and 45a. The 2001 map shows (Figure 43a) similar patterns to the previ ous 2001 maps for mosquitoes and bird surveillance. However, there were not enough human cases to accurately evaluate the possible trends between clinical cases and sent inel seroconversion rate s. The scatter plot (figure 43b) for 2001 data by week shows no correlation with a P earson’s correlation coefficient equal to 0.21 and a p-vale of 0.2648. Figure 43a shows the distribution of the annual sentinel sero conversion rate and clinical cases for 2002 by count y. The individual counties with the highest numbers of

PAGE 111

96 clinical cases, Escambia and Marion County, were not partic ipating in the sentinel program in 2002. The counties participating in the sentinel surveillance program in 2002 had low numbers of clinical cas es (1 or 2); subsequently no conclusions on trends could be drawn from this map. The overall correlation (figure 43b) shows no significant association (Pearson’s correlation coe fficient r = 0.12 and p-value = 0.7046). Figure 45a, the geographical distribution of the data for the annual West Nile sentinel chicken seroconversions and pos itive WNV human cases during 2003, showed a corresponding increase in the observed seroc onversion rates and number of clinical cases in a three of the participa ting counties: Bay, Lee and Duva l. Bay County had the most clinical cases, 14, with a sen tinel chicken seroconversion of 0.0559. Lee County had only three clinical cases but had sentinel sero conversion rate of 0. 0813 which fell in the highest rate range (0.062 to 0.125) shown on the map. There we re six diagnosed clinical cases in Duval County, with a seroconvers ion rate of 0.0478, which was in the same sentinel seroconversion rate range as Bay County. The ot her counties either did not participate in sentinel surv eillance or they did not have considerable trends. The 2003 scatter plot (figure 45b) shows no correlation be tween the number of clinical cases and the sentinel seroconversion ra te (Pearson’s correlation coe fficient r = 0.22 and p-value = 0.5637). Figure 45c shows the scatter plot ex cluding the observations with 14 clinical cases (outlier). This plot shows a negative correlation as opposed to the correlation shown in figure 45b including the outlier observati ons which had no correlation between sentinel seroconversion rates and clinical cases.

PAGE 112

97 Figure 42a. Florida West Nile Average Sentinel Chicken Seroconversion Rate Distribution and Clinical Cases by County for 2001-2003. Intensity of color indicates the rate of West Nile serconversion. The areas designated without co lor signify counties that di d not submit any sentinel chicken sera for testing. 1 7 3 6 4 2 8 14 19 1 2 3 1 2 4 1 11 2 1 1 1 1 1 8 1 3 1 2 1 1 2 1 1 4 4 3 2 1 3 1 11Sentinel Chicken Positve Conversion Rate 0 0.01 0.01 0.023 0.023 0.043 0.043 0.062 0.062 0.125 No Submitted Specimen 0 200Miles N E W S# Represents the Number of Clinically Diagnosed Cases of West Nile

PAGE 113

98 Figure 42b. Scatter Plot for Correlation between the Number of Clinical Cases and Sentinel Seroconversion Rates by Week for the Pooled Years 2001-2003. The black line represents the fitted regression line for the number of clinical cases and sentin el seroconversion rate. 0 0.02 0.04 0.06 0.08 0.1 0.12 0.140246810121416Number of Clinical Cases Sentinel Seroconversion Rate Figure 42c. Scatter Plot for Correlation between the Number of Clinical Cases and Sentinel Seroconversion Rates by Week for the Pooled Years 2001-2003 excluding outlier observations. The black line represents the fitted regression line for the number of clinical cases and sentin el seroconversion rate. 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14012345678Number of Clinical Cases Sentinel Seroconversion Rate

PAGE 114

99 Figure 43a. Florida West Nile Average Sentinel Chicken Seroconversion Rate Distribution and Clinical Cases by County for 2001. Intensity of color indicates the rate of West Nile serconversion. The areas designated without co lor signify counties that di d not submit any sentinel chicken sera for testing. 1 1 1 1 1 2 1 1Sentinel Chicken Positive Conversion Rate 0 0.005 0.005 0.015 0.015 0.057 0.057 0.125 No Submitted Specimen 0 200Miles N E W S# Represents the Number of Clinically Diagnosed Cases of West Nile

PAGE 115

100 Figure 43b. Scatter Plot for Correlation between the Number of Clinical Cases and Sentinel Seroconversion Rates by Week for 2001. The black line represents the fitted regression line for the number of clinical cases and sentin el seroconversion rate. 0 0.02 0.04 0.06 0.08 0.1 0.12 0.1400.20.40.60.811.2Number of Clinical Cases Sentinel Seroconversion Rate

PAGE 116

101 Figure 44a. Florida West Nile Average Sentinel Chicken Seroconversion Rate Distribution and Clinical Cases by County for 2002. Intensity of color indicates the rate of West Nile serconversion. The areas designated without co lor signify counties that di d not submit any sentinel chicken sera for testing. 11 7 1 1 1 1 1 3 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 2 7Sentinel Chicken Postive Conversion Rate 0 0.004 0.004 0.019 0.019 0.041 0.041 0.06 0.06 0.157 No Submitted Specimen 0 200Miles N E W S# Represents the Number of Clinically Diagnosed Cases of West Nile

PAGE 117

102 Figure 44b. Scatter Plot for Correlation between the Number of Clinical Cases and Sentinel Seroconversion Rates by Week for 2002. The black line represents the fitted regression line for the number of clinical cases and sentin el seroconversion rate. 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.1800.511.522.533.5Number of Clinical Cases Sentinel Seroconversion Rate

PAGE 118

103 Figure 45a. Florida West Nile Average Sentinel Chicken Seroconversion Rate Distribution and Clinical Cases by County for 2003. Intensity of color indicates the rate of West Nile serconversion. The areas designated without co lor signify counties that di d not submit any sentinel chicken sera for testing. 1181 2 2 4 1 1 1 1 1 6 1 2 1 2 2 1 4 6 1 2 3 1 12 10 14Sentinel Chicken Positive Conversion Rate 0 0.014 0.014 0.038 0.038 0.062 0.062 0.094 No Submitted Specimen 0 200Miles N E W S# Represents the Number of Clinically Diagnosed Cases of West Nile

PAGE 119

104 Figure 45b. Scatter Plot for Correlation between the Number of Clinical Cases and Sentinel Seroconversion Rates by Week for 2003. The black line represents the fitted regression line for the number of clinical cases and sentin el seroconversion rate. 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10246810121416Number of Clinical Cases Sentinel Seroconversion Rate Figure 45c. Scatter Plot for Correlation between the Number of Clinical Cases and Sentinel Seroconversion Rates by Week for 2003 excluding outlier observations. The black line represents the fitted regression line for the number of clinical cases and sentin el seroconversion rate. 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1024681012Number of Clinical Cases Sentinel Seroconversion Rate

PAGE 120

105 Multivariate Poisson Regression Model The multivariate and univariate Poisson distribution regression model results are summarized in Appendix XII. The level of significance for all models was set at =0.10. The data were divided by region and week and annual (weeks 1-52) data was used for analysis on the cumulative and individual year s. The response variable was set as the incidence of human cases and the explanatory variables were the rate of positives for each surveillance system with and without region in the model. Region was based on four levels: panhandle, north, central and south. The annual individual year multivariate Poisson distribution for the incidence of clinical cases for 2001 – 2003 was separately calculate d for each surveillance type considering region and not considering region in th e model. For 2001 and 2002 there were no significant p-values for both models with or without region. Ther efore, none of the surveillance systems for these years were good predictors for clinic al cases. The values from this analysis can be found in Appendi x XII. The overall model equation for all Poisson calculations for the cumulative years was: log(BiB)= log( pBiB) + B0B + B1B (surveillance system) + t (year) log(BiB)= log( pBiB) + B0B + B1B (central) + B2B (north) + B3B (panhandle) + B4B (south, reference region) + B5B (surveillance system) + t (year) The specific value estimates are listed in Appendix XII. The overall model for the pooled ye ars 2001, 2002 and 2003 indicated each of the surveillance types (dead birds, mosquitoes and sentinels) include d in separate models or together predicted clini cal cases with significant p-va lues (p < 0.10). The parameter estimates for all variables are shown in Appendix XII.

PAGE 121

106 The Poisson distribution for 2003 showed that positive dead birds and sentinels were the best predictors whereas mosquitoes were insignificant as predictors (values in Appendix XII). The model for the positiv e avian rate had a p-value of 0.0002. The predictive model equation for clinical case incidence and avian positives is log =-8.9099 + 6.5038XB1B (where log=incidence of clinical cases, XB1B=avian positive rate). The clinical case incidence and sentinel rate model had a p-value of 0.0117 and the equation was log =-7.9687 + 12.1318XB1B (where log=incidence of clinical cases, XB1B=sentinel positive rate). Table 1 lists the significant p-va lues and equations for these models. The adjusted model including all surveill ance types (dead birds, sentinels, and mosquitoes) was shown to be significant pr edictor during 2003. No other models for the individual years (2001 and 2002) had significant p-values. The model equation for 2003 for all surveillance types is: log =-8.9078 + -6.0103XB1B+ 10.4488XB2B+ -13.0034XB3B (log=incidence of clinical cases, XB1B= avian positive rate, XB2B=mosquito positive rate, XB3B=sentinel seroconversion rate). Sentinel rate was the only individual significant predictor for 2001 when the effect of region was considered. All other models for 2001 and 2002 were not considered good predictors (p>0.10) for human cases. The multivariate Poisson regression an alysis for the annual model for 2003 including region indicated that it had an effect with all the surveillance type rates. The panhandle region was significant for all thr ee surveillance types. The p-values for sentinel and dead avian rates were 0.1257 and 0.016 respectively. These values were higher then the p-values without region include d, where sentinels were significant in the

PAGE 122

107 previous model but not in this model. The panhandle p-value for the sentinel rate model was 0.0350 and 0.0217 for the avian positive rate model. The model equation for sentinel positives is not given since it was insignificant, the values can be found in Appendix XII. The equation for clinical case incidence and positive avian ra tes with region was log =-10.0786 + -1.4089XB1B+ 0.2240XB2B+ 3.4915XB3B+ 0.000XB4B+4.9847XB5B (log=incidence of clinical cases, XB1B= central region, XB2B=northern region, XB3B=panhandle region XB4B=southern region XB5B=avian positive rate). Mosquito pool rates had a lower p-value (0.2860) with region in the model where the p-value for the panhandle was 0.0217 compared to the other regions. No equation model is given because the mosquito positive rate was not si gnificant, the specific values are given in Appendix XII and table 2 lists the signi ficant model p-values and equations.

PAGE 123

108 Discussion The frequency of diagnosed West Nile In fections in Florida is currently low; however, the risk of future epidemics remains unknown. Research has shown that approximately 80% of West Nile infections in the areas it is present are unreported because of the mild symptoms or asymptomatic infections (CDC, 2001a). Florida’s age demographic, which includes a sizeable population over 50 years of age and warm climate, may stimulate an increase in th e number of confirmed WNND and WNF cases in the future. Since the morbidity and mortality of West Nile infection increases when neurological symptoms are present, early de tection of WNND is an important prevention strategy to help reduce transmission. Demographic Analysis Based on gender demographic analysis, men were twice as likely as women to be diagnosed with WNND or WNF during the da ta collection period. There has not been documented association between gender and West Nile disease. Therefore, it is possible since surveillance data for th is population is very limited (3 years), the observed data trend may be due to chance. Further surveillan ce data will be needed to confirm whether men are, in fact, more likely to be dia gnosed with WNND or WNF. If this trend continues, compounding risk factors such as a greater likelihood of men participating in outdoor activities during transmission times (i.e. dusk and dawn) and a lower compliance with personal mosquito protection, the 5 D’s of Prevention need to be considered (19). Further analysis of gender also indicated th e numbers of males were notably higher then females across each age grouping. The demographic analysis for age shows the number

PAGE 124

109 of human WNND cases increased with age by group (> 55 years old was the highest age group). WNF, however, was more frequent among individuals 31 to 49 years old. This corresponds with the Center for Disease Contro l and Prevention’s a ssessment of age as a risk factor (CDC, 2001a). Examination of the county incidence rate s for laboratory confirmed cases of WNND and WNF shows an increase in reported cases each year. The annual incidence rates show a linear increase over time. Florid a will quickly develop a high disease burden if this incidence tre nd continues over the next few year s. Basic counts by county show the same trend, but do not take into considerati on that reporting of m ild disease (WNF) may underestimate the problem for the more rura l areas of Florida. The state has large geographical regions with low population dens ities due to large agricultural land uses. Levels of incidence provide less biased data than numbers of cases per county because they are based on population. However, county medical alerts are based in part on the number of diagnosed cases per county. The numbe r of clinical cases is also important to consider for control and prevention Temporal Distribution The weekly temporal distribution of c onfirmed West Nile events was similar across each surveillance type (sentinel chicke n, dead avian and mos quito) with almost no events occurring from January to May. Anal ysis showed a transmission season beginning at approximately week 20 (late May to early June) and ending at week 52 (late December). The peak detection periods varied for each surveillance type, but when the focus was narrowed to look at only the tran smission season, each system had a distinct peak period. A few outliers were observed outside the transmission season. These were

PAGE 125

110 not included as part of the temporal dist ribution analysis because they occurred far outside the observed curve of the peak tran smission period. These events may be random, carried over from the previous year’s transm ission season or part of normal background activity and thus do not play a role in dete rmining the temporal transmission to humans. Since West Nile virus is new to Florida, hist orical data is not yet available to determine normal thresholds of virus activity. Data shou ld therefore continue to be collected in order to properly determine the baseline for WNV activity in Florida. The distribution of confir med WNND and WNF cases by week of onset did not appear to have a distinct peak period however, all of th e reported cases had the same temporal distribution as other positive surveillance events. The lack of a clear peak range with in the transmission season may be a di rect reflection on the low numbers of cases observed over the three year pe riod. It is interesting to not e that the numbe r of reported cases has tripled with each successive year. This implies this trend will remain unknown without further data and investigation. This implicates that Florida will likely see an increase in the overall burden of disease in the future. Continuation of surveillance programs would allow for better case finding and more testing avai lability leading to more conclusive results and an improved abil ity to predict the risk of disease. The number of West Nile cases in Florida for 2004 has been lower then the previous years. Figure 43 shows the epi curve clinical cases of West Nile Disease where the number of cases have decreased during 2004 as compared to 2003. This may be due to herd immunity to West Nile or better co ntrol measures (e.g. mosquito control).

PAGE 126

111 Figure 46: Epi Curve Comparing 2003 and 2004 Confirmed Human Cases. 0 2 4 6 8 10 12 242526272829303132333435363738394041424344454647 CDC WeekNumber of Confirmed Cases Sum of 2003 Sum of 2004 http://www.doh.state.fl.us/Environment/h see/arbo/Epi_Curves/epicurve10_29_04.pdf Peak Transmission A mean of the weeks with the highest number of cases of confirmed WNND and WNF for all three years (2001-2003) was the 37PthP week (mid-September). By combining the weeks with the greatest num ber of clinical cases from each year, human cases would be expected between the 28PthP and 44PthP week. The mean transmission peaks varied for each surveillance type with sentinel chicken surveillance showing a mean peak in the 40PthP week (early November), dead bird surveillance showing a mean peak in the 36PthP week (early September), and mosquito surveillance showing a mean p eak in week 33 (mid July). Collectively the surveillance

PAGE 127

112 types have a mean transmission peak in week 36 with a range from week 29-46. The transmission range provided an indication th at prevention and control efforts for West Nile and other arboviruses during and leadi ng up to this time period would be most effective. Dead bird and mosquito surveillance m ean peaks were one and five weeks, respectively, prior to the m ean peak in reported human cas es. Conversely, the sentinel mean peak was three weeks after the mean peak in reported human cases. There was a substantial increase in the number of sentinel chicken sites in 2002 and 2003 which may indicate the data from th e years considered (2001, 2002 and 2003) are not comparable. The peak of sentinel seroconvers ions is not as good an indica tion of risk as when positive outcomes first begin to appear and how qui ckly the outcomes increase. Ascertaining when separate surveillance systems peak in relation to when confirmed WNND and WNF reports peak is an indication of which survei llance system’s data may be the most useful in a model for the early detection of West Nile associated illnesses in Florida. Unfortunately, more data will be needed to verify that dead bird and mosquito surveillance consistently peak prior to human cases, since participation in these surveillance activities also differed greatly be tween counties and years. Over one half of the human cases from 2001 and 2002 occurred in areas with no sentinel surveillance. In order to better utilize the sentinel surveill ance program a baseline assessment over the next few years would help show increase s above the normal background activity. Since WNV is new in Florida, the activity we have seen may not be normal therefore, the next years are imperative in es tablishing this threshold.

PAGE 128

113 Since 2001 mosquito pool submissions a nd sentinel sites have increased in number by over 40% each year. This has provided better coverage in the counties participating in these surveillance types. Th e dead bird submissions have drastically decreased by almost 40% per year. The first ye ar of WN detection (2001) had most of the dead bird submissions after the first dead crow was found positive, there were over 7,000 birds submitted in a six month period. Co mparing the percent positive and number positive for dead birds showed a positive correl ation with number of submitted dead birds and the percent positive. During 2001 there was greater number of positives then in the subsequent years (2002 & 2003) but the per cent positive was lower. This means the excessive numbers of dead birds that were submitted may have diluted the results. The percent positive in 2003 was higher then the othe r two years, this may be due to the lower numbers or the species of dead birds submitted for testing. Over the last few years trends with particular bird species more prone to WN infection (e.g. corvids) have become well known resulting in some counties submitting on ly these species. Evaluating the percent WN positive and the total submitted dead birds for the three years observed suggests that 2003 was a comparable model, having the hi ghest percent among th e years even though the least number were submitted. This indicates the methods used during 2003 may be the best in order to show more specific re sults and keep the associated costs at a minimum. Mosquito surveillance, WN percent positive was more consistent over the three years with 2001 and 2003 having the highest ra tes. However, there was no significant correlation for 2001 probably due to the lower numbers of submitted mosquito pools. The highest rate for mosquito positives was 0.07 s howing a smaller range then the dead birds

PAGE 129

114 where the highest rate excluding the first po sitive specimen (1.0), was 0.60. These results suggest that an increase in mosquito pool s ubmissions would show better detection rates with out diluting the outcome. Mosquito control responses to sentinel seroconversions could possibly lead to a reduction or elimination of the number of hu man cases of WN disease. Consequently, these proactive efforts could influence survei llance data by stimulating an increase in sampling and submissions which might reduce the percentage of positive mosquito pools and dead birds. Regional Analysis Based on the physiographic climate, Florid a’s 67 counties were separated into four regions: panhandle, nor th, central and south (D ay, 1996). The cartographic representation is shown in Appendix IX. Each region’s surveillance data for dead avian, mosquito and sentinel chickens was examined independently to control for the bias of differing participation levels. The data sets were evalua ted based on their predictive characteristics for human West Nile dis ease among regions. A list of the counties grouped by region with surveillance participa tion levels is located in Appendix VIII. In general, interpretation of the region al surveillance data indicated that dead avian and sentinel chicken surveillance have the best predictive qualities. The graphs which show an increase in the number of sen tinel seroconversions and positive dead birds before human cases were detected for each region (Fig. 29a,b; 30b,c; 31b,c; 32b,c). There were incidents where the initial clinical case was seen before an increase in surveillance activity. This may be due to clinical cases observed early in the season or lack of participation from the region at the time. Th e latter may be an anomaly associated with

PAGE 130

115 the submission of dead birds because of medi a attention given to West Nile whenever a new clinical case is diagnosed. Surveillance data collected in 2001 from all three systems proved to be too unreliable to evaluate, limitations from this year may be due to increased sample submissions after reported cases of WNND a nd WNF. Some regions showed human case foci initially in areas wi thout surveillance programs. Mosquito surveillance data proved unreli able for use as a predictor for human cases because of low detection numbers of WNV as well. This may be based on several factors including the number of mosquitoes needed fo r testing, locat ion, collection methods, and storage conditions. All of these is sues can individually or collectively have an adverse affect on the samples leading to the inability to detect the virus. Florida’s southern region a ppears to have the worst predictive qualities of the four regions. The explanation may lie with the fact that the southern re gion continually began surveillance sample submission later than the other regions. Add itionally the counties within the region were inconsistent with th e level of participati on for each surveillance type. Excluding mosquito data, the best region for predicting c linical cases from surveillance data in Florida was the panhandle. This is also seen in Appendix VIII which shows the levels of participat ion by counties within each region. The regional graphs showed several peak s for sentinel positives within the distribution for some of the years examine d. These may be an anomaly caused by the replacement of birds with in a flock after the positive birds are removed. The first week the chicken is bled is used as the baseline sera. The birds are then placed with exposed flock for weekly serosurveillance. The new sent inel bird may seroconvert after this time.

PAGE 131

116 Thus, there is a 2 week lag between rem oving seroconverted sentinels and adding new ones which may account the decrease in positives during this time period. Since sentinel surveillance is controlled by the submitting counties it is important to consider the point during the year at wh ich counties begin their surveillance. Table 1 lists all the counties participating in sen tinel surveillance and the month they start submitting sentinel sera. Counties that pa rticipated during 2001, 2002 and 2003 began the surveillance before clinical cases were seen with the exception of DeSoto, Santa Rosa and Washington counties during 2002. These counties were not consistent in submitting their sera in a timely manner allowing br eaks in weeks before submitting the next specimen.

PAGE 132

117 Table 1. Month of First Sera Submitted for Counties Participating in Sentinel Surveillance for 2001, 2002 and 2003 Organized by Region. County 2001 20022003County 20012002 2003 Panhandle Central Bay May Jan Jan Brevard May Feb Mar Jackson None NoneJun D esoto Aug Sept None Jefferson None NoneJun HillsboroughJan Jan Jan Leon May May Apr Indian River May Jan Jan Santa Rosa None Sept Jan Manatee Jan Jan Jan Walton Jun Jan Jan Okeechobee Jul Jul Jul Washington Jul Aug Jun Osceola Feb Feb Jan North Pinellas May Jan Jan Alachua Apr Apr Apr Polk Jul None None Citrus May Jun May Sarasota Jul Jan Jan Duval Jun May Apr St. Lucie May May Jan Flagler Apr Apr Mar South Marion None NoneJun Charlotte May May Jan Nassau None NoneMay Collier Jun May Apr Orange Jan Jan Jan Dade None None Mar Pasco Jul Jan Jan Hendry Jun Jun May Putnam Jun Jul Apr Lee Jan Jan Jan Seminole May Jan Apr Martin Jun Jun Mar St. John's Jun May Jun Palm Beach Apr Jan Jan Suwannee Jul NoneNone Volusia May Jan Jan Limitations with this type of analysis in clude a lack of a representative sample for mosquitoes and dead birds. Raw numbers were used instead of rates in order to show all of the surveillance types in one graph. Th ere was no way to correctly estimate the numbers needed for the denominator for each region, county or statewide by week. This is important to consider because samp le submissions by county range depending on several factors. Urban areas generally submit more dead birds, increased media attention also stimulates submissions; but rural areas may not have the available funds to support

PAGE 133

118 these increases. Therefore, the surveillance data may ove r or underestimate the actual disease burden within individual regions. Spatial Analysis The best overall indicator from the spatial analysis of clinical cases was positive dead birds which had a positive correlation showing an increase in both diagnosed cases of West Nile infection and positive dead birds. This is especially noticeable in Escambia and Bay Counties which had the highest num bers of human WN cases during 2002 and 2003. Some counties had lower submissions of avian which may account for the lack of correlation with clinical cases There may also be an association between the number of dead bird submissions and urba n population centers. It is more probable for dead birds to be found and submitted in urban areas then in rural areas. Sentinel serosurveillance also had some value; unfortunately, the counties with the most human cases did not participate in the program. There was a negative correlation for the cumulative years, 2001 and 2002. The correlation for 2003 was positive with all the data considered but negative when outli ers were excluded. A better correlated model may be ascertained if all the counties in the state participated. Anot her explanation for the negative correlation may be associated with increased control measures after sentinels seroconvert therefore, preventing cases. Th is would show an increase in sentinel seroconversion and a decrease in clinical cases. The mosquito pool surveillance data set wa s not a useful indicator for clinical cases for any of the years observed. Ther e were not enough positive mosquito pools to accurately show correlations between the clin ical cases and positive mosquitoes. There may be a need for better collecting methods or increasing the number of mosquitoes

PAGE 134

119 submitted to facilitate a improved detection of West Nile virus. Another explanation for the negative mosquito correlations, as show n in the sentinel population, could be associated with increased control measures after a positive mosquito pool is detected. Spatial analysis does not allow for a se quential time period to be evaluated therefore, the temporality for casual inference may introduce bias into the models. This is especially important to consider when us ing the surveillance systems as a predictor because detection of WN viru s activity may not necessarily come before the clinical cases. This type of analysis is useful for obtaining an historical overview, comparing geographic locations and establishing overall participation in surveillance by county or region. The correlations showed outlier observa tions for all of the years and the cumulative years for positive dead bird and mosquito pool surveillance. The sentinel serconversion and clinical case correlations showed outliers for the cumulative and 2003 analysis. Excluding these obser vations resulted in better fitting regression lines. However, the overall correlation was similar to the original with exception of sentinel serconversion rates and clini cal cases during 2003 which chan ged from no correlation to a negative correlation. This effect may be important to consider because the outlying observation could skew the overall correlation. Poisson Distribution Regression Model Due to the response variable (incidence of human cases) being a rate and a rare event, the regression model was done using the Poisson distribution in order to accurately show the predictive values for the survei llance systems. The data was organized by region using cumulative (2001, 2002 and 2003) and weekly observations. Regression

PAGE 135

120 analysis showed that each surveillance type showed different value as a predictive model for human disease. The actual outcome from the regression analysis was similar to the expected for positive dead birds and sentinels which were shown to be good predictors for clinical cases. Also expected was the outcome for Mosquito surveillance which was not an overall good predictor of human disease. The analysis for 2001 and 2002 did not show any of the surveillance systems as predictors (p-values >0.10), except for 2001 with region included, this may due to the lack of human cases which resulted in very small rates for incidence. This phenomenon may also be a consequence of increased submi ssions of specimen after the first clinical case because of the media attent ion especially in the first year (2001) of WN detection. The adjusted model including the data fr om all surveillance types showed that positive dead bird surveillance was the best predictor for human cases in 2003. The cumulative year adjusted model was similar to the unvariate analysis These results from the 2003 data indicate that positive dead birds were the best predicto r for clinical cases. The effect of region in the model was s een in cumulative analysis for mosquito and sentinel surveillance, a nd the dead bird surveillance in 2003. However, mosquitoes were improved as a predictor for the individua l year analysis for human cases with region considered was mosquito pools. The data al so showed in 2001 sentinels were better predictors when region was adjusted for the region having the most effect comparatively was the panhandle. These results may be associated with confounding because of a greater number of clinical cases or number of positives (mosquitoes-2003 and sentinels2001) in the panhandle. This would be adjusted for by including region.

PAGE 136

121 Study Strengths and Limitations This study’s major strength was indica ting where improvements to the overall West Nile surveillance system could be impl emented to enhance public health. It also indicated that the current surveillance syst em has been successful in predicting human disease in the areas that cons istently participate. The use of separate analyses for each surveillance type show ed that the dead bird, sentinel and mosquito pool surveillance projects were multifaceted. This means that while one system may not indicate a correlation, others may. It also provides a basi s for future studies on the natural history of West Nile Virus. There were several limitations encoun tered in this study. The data proved unreliable in showing a representative ra te because there was no available population information for mosquitoes and birds. The rates used for positive birds may have over or under estimated West Nile activity depending on the county, because submissions may have been larger in the urban areas a nd smaller in rural areas of Florida. Mosquito pool surveillance data may have been biased because the submissions were sporadic and were mostly based on the individual county inte rests; therefore the mosquito pools submitted may not have been for surveillance purposes only. Mosquito specimens have many problems as a surveilla nce program including the need for large submission number in order to detect We st Nile and specific sample handling and shipping procedures which are not mandated in any form. Human surveillance for 2001 and 2002 had low incidence rates and were not reliable for predictive models. Another problem was due to the varied participation levels

PAGE 137

122 within the Florida regions and counties. This included counties with human cases that did not use any of the surveillance systems. Specific problems were observed with in the sentinel surveillance program. The overall data were reliable however; Sant a Rosa and Washington counties were not consistent in their participation. Santa Rosa county submitted sera from six chickens at three separate times during 2002 and four separate times during 2003 each was about three to four weeks apart. Washington County did the same for 2003. This does not constitute participation in the sentinel surveillance program. Evaluation An overall evaluation of the performance of the Arbovirus surveillance system for WNV shows that each surveillance type (dead bird, mosquito and sentinel chickens) provides important information about virus transmission in the environment. Positive mosquito trends tend to provide an earlier warning system when adequate sampling is performed. Positive dead birds appear to be better predictors for human cases and sentinel chicken serosurveillance indicates activity within a specific geographic area. These systems are costly to maintain however, c ontinuation will help to lower the disease burden and risk of disease resulting in signi ficantly decreased health care expenditures. The programs initiated for education by the Flor ida Department of Health and the Center for Disease Control and Prevention are also us eful tools to keep up public awareness and decrease the number of possible cases. The su rveillance program has proven to be stable and successful since Florida has been usi ng these surveillance t echniques for other endemic arboviruses for over 30 years.

PAGE 138

123 Nevertheless, this evaluation illuminates several areas in which enhancements can be made to help show even better corre lations and predictive values for the overall system. Increase the coverage and consistenc y of submissions for all surveillance types. Set standard levels of participati on for all counties based on the regional analyses and populations at risk. Create standardized approaches for sampling, shipping and submitting samples (especially for mosquito pool submissi ons) and require that participating counties adhere to these standards. Only submit specific birds known to be especially susceptible to West Nile Virus (e.g. corvids). Targeted prevention and education stra tegies for higher risk groups based on their potential levels of exposure.

PAGE 139

124 References Cited Agresti A. (1996). An Introduction to Categorical Data Analysis New York: John Wiley & Sons, Inc. Armineh Z, Meltzer MI, Ratar d, R et al (2004). West Nile Economic Impact, Louisiana, 2002. Emerging Infectious Diseases 10(10): 1736-1744. Bernard KA, Maffei JG, Jones SA, et al. (2000) West Nile Virus Infection in Birds and Mosquitoes, New York State, 2000. Emerging Infectious Diseases, 7 (4), 679 685. Blackmore CM, Stark LM, Jeter WC, et al. (2 001). Surveillance Results from the First West Nile Virus Transmission Season in Florida, 2001. American Journal of Tropical Medicine and Hygiene, 69 (2), 141 150. Broom C, Beuehler JW, Gr esham L, et al. (2004). Framework for Evaluating Public Health Surveillance Systems for Early Detection of Outbreaks Campbell GL, Marfin AA, Lanciotti RS, et al. (2002). West Nile Virus Review. The Lancet Infectious Diseases, 2 519 529. Centers for Disease Control and Preven tion. (1988). Guidelines for Evaluating Surveillance Systems. Morbidity and Mortality Weekly Report, 37(S-5):1-18. Centers for Disease Control and Preven tion. (1993). Guidelines for Arbovirus Surveillance Programs in the Unite d States. Available from URL H Uhttp://www.cdc.gov/nci dod/dvbid/arbor/arboguid.pdf U H Centers for Disease Control and Prevention. (1 999). Outbreak of West Nile-Like Viral Encephalitis New York, 1999. Morbidity and Mortality Weekly Report, 48 (38), 845 848. Centers for Disease Contro l and Prevention. (2001a). Laboratory Diagnosis of West Nile Virus Infections Available from URL: H Uhttp://www.cdc.gov/ncidod/dvbid/west nile/clinicians/clindesc.htm#lab U H Centers for Disease Control and Prevention. (2001b). Overview of Public Health Surveillance. Available from URL: H Uhttp://www.cdc.gov/epo/dphsi/phs/overview.htm U H Centers for Disease Control and Prevention. (2 001c). Updated Guidelines for Evaluating Public Health Surveillance Systems. Morbidity and Mortality Weekly Report, 50 (13), 1 35. Centers for Disease Control and Prevention. (2002). West Nile Virus: Background and

PAGE 140

125 Ecology. Available from URL: H Uhttp://www.cdc.gov/ncidod/ dvbid/westnile/background.htm U H Centers for Disease Control and Prevention. (2003). Epidemic/Epizootic West Nile Virus in the United States: Revised Guid elines for Surveillance, Prevention, and Control. Available from URL: H Uhttp://www.cdc.gov/ncidod/dvbid/westnile /resources/wnv-guidelines-aug-2003.pdf U H Centers for Disease Control and Prevention. (2 003c). Surveillance fo r Acute InsecticideRelated Illness Associated with Mos quito Control Efforts—Nine States, 19992002. Morbidity and Mortal ity Weekly Report, 52(27):629-634. Centers for Disease Contro l and Prevention. (2004). Fight the Bite Available from URL H Uhttp://www.cdc.gov/ncidod/dvbid /westnile/preve ntion_info.htm U H Chowers, MY, Lang R, Nassar F, et al. (2000). Clinical Characteristic s of the West Nile Fever Outbreak, Israel, 2000. Emerging Infectious Diseases, 7 (4), 675 678. Day JF and Stark LM. (2000). Frequency of Saint Louis Encephalitis Virus in Humans from Florida, USA: 1990 1999. Journal of Medical Entomology, 37 (4), 626 633. Eidson M, Miller J, Kramer J, et al. (2000a ). Dead Crow Densities and Human Cases of West Nile Virus, New York State, 2000. Emerging Infectious Diseases, 7 (4), 662 664. Eidson M, Kramer L, Stone W, et al. ( 2000b). Dead Bird Surveillance as an Early Warning System for West Nile Virus. Emerging Infectious Diseases, 7 (4), 631 635. Eidson M, Komar L, Sorhage F, et al. (1999) Crow Death as a Sentinal Survelliance System for West Nile Virus in the Northeastern United States, 1999. Emerging Infectious Diseases, 7 (4), 615 620. Florida Bureau of Environmental Health a nd Florida Department of Health. (2004). Surveillance of Selected Arthropod-born e Diseases in Florida, 2004 Guidebook. Available from URL H Uhttp://www.doh.state.fl.us/Environmen t/hsee/arbo/pdf_files/2004_ArboGuide.pdf U H Giladi M, Metzer-Cotter E, Martin DA, et al (1999). West Nile Encephalitis in Israel, 1999 The New York Connection. Emerging Infectious Diseases, 7 (4), 659 661. Gubler DJ. (2001). Human Arbovi rus Infections Worldwide. Annals of the New York

PAGE 141

126 Academy of Sciences, 951 (1), 13 24. Gubler DJ and Roehrig JT (1998). In Topley and Wilson’s Microbiology and Microbial Infections, Vol 1, Virolo gy. Mahy, BWJ, Collier, L, eds. Arboviruses (Togaviridae and Flaviviridae). Oxford University Press, New York, 29:579600. Guptill SC, Julina KG, Campbell GL, et al. (2003). Early-Season Avian Deaths from West Nile Virus as a Warning of Human Infection. Emerging Infectious Diseases, 9 (4), 483 484. Halperin W, Baker EL, and Monson RR. (1992). Public Health Surveillance. John Wiley & Sons, New York, USA. Huhn G, Sevar JJ, Montgomery SP, et al. ( 2003). West Nile Virus in the United States: An Update on an Emerging Infectious Disease. American Family Physician, 68 (4), 653 660. Johnson DJ, Ostlund EN, Pedersen D, et al. ( 2000). Detection of North American West Nile Virus in Animal Tissue by a Reve rse Transcription-Nested Polymerase Chain Reaction Assay. Emerging Infectious Diseases, 7 (4), 739 741. Komar N, Panella NA, Burns JE et al. (2001) Serologic Evidence for West Nile Virus Infection in Birds in the New York City Vicinity During an Outbreak in 1999. Emerging Infectious Diseases, 7 (4), 621 625. Komar N, Panella NA, Boyce E. (1999). Expos ure of Domestic Mammals to West Nile during an Outbreak of Human Encephalitis, New York City, 1999. Emerging Infectious Diseases, 7 (4), 736 738. Komar N (2003), Experimental Infection of North American Birds with the New York Strain of West Nile Virus. Emerging Infectious Diseases, 9(3): 311-322. Langevin, SA, Bunning M, Davis B, et al. (2001) Experimental Infection of Chickens as Candidate Sentinels for West Nile Virus. Emerging Infectious Diseases, 7(4):726-729. Leffkowitz M (1942). On the Frequent Appearan ce of an Unclear Infectious Disease, 22: 3-4. Marfin AA and Gubler DJ. (2001) West Nile Encephalitis: An Emerging Disease in the United States. Emerging Infections, 33 1713 1719. Murgue B, Murri S, Zientara S, et al. (2000). West Nile Outbreak in Horses in Southern France, 2000: The Return after 35 Years. Emerging Infectious Diseases, 7 (4), 692 696. Nasci, RS, White DJ, Stirling H, et al. (1999). West Nile Virus Isolates from Mosquitoes in New York and New Jersey, 1999. Emerging Infectious Diseases, 7 (4), 626

PAGE 142

127 630. Nash D, MostF Fine A, et al. (2001). The Ou tbreak of West Nile Virus Infection in the New York City Area in 1999. The New England Journal of Medicine, 344 (24), 1807 1814. Neter J, Kutner MH, Nachtsheim CJ, et al. (1996), Applied Linear Statistical Models Fourth Edition, McGraw-Hill Companies, Inc., New York, USA. Petersen LR and Marfin AA. (2002). West Nile Virus: A Primer for the Clinician. Annals of Internal Medicine, 137 (3), 173 179. Petersen LR, Marfin AA, and Gubl er DJ. (2003). West Nile Virus. Journal of American Medical Association, 290 (4), 524 528. SAS Institute Inc (1999). SAS/STAT User’s Guide, Version 8, Cary, NC: SAS Institute Inc. Sejvar JJ, Haddad MB, Tierney BC, et al (2003). Neurologic Ma nifestations and Outcome of West Nile Virus Infection. Journal of American Medical Association, 290 (4), 511 514. Shroyer DA. (2001). The Collection of Mo squitoes for Arbovirus Assay: Some Fundamental Considerations. Wing Beats of the Florida Mosquito Control Association 12(2), 4-5, 18 and 25. Stark LM and Kazanis D. (2001). Arbovirus Surveillance: Annual Summary Report 2001 Tampa: Florida Department of Health Tampa Laboratory. Stark LM and Amuso P. (2002). Annual Report (Virology): 2002 Tampa: Florida Department of Health Tampa Laboratory. Stark LM and Amuso P. (2003). Annual Report (Virology): 2003 Tampa: Florida Department of Health Tampa Laboratory. Stokes ME, Davis CS, and Koch GG. (2000). Categorical Data Analysis Using The SAS System Cary, North Carolina: SAS Institute Inc. Swayne DE, Beck JR, Smith CS, et al. (2000) Fatal Encephalitis and Myocarditis in Young Domestic Geese ( Anser anser domesticus ) Caused by West Nile Virus. Emerging Infectious Diseases, 7 (4), 751 753. Tabachnick, Walter J. (2002). Florida's Plans for Controlling West Nile Virus Gainesville. Teutsh SM and Churchill RE, (Ed.). (2000). Principles and Practice of Public Health

PAGE 143

128 Surveillance New York: Oxford University Press, Inc. Trock SC, Meade BJ, Glaser AL et al. (2000). West Nile Virus Outbreak Among Horses in New York State, 1999 and 2000. Emerging Infectious Diseases, 7 (4), 745 747. Tyler KL. (2001). West Nile Vi rus Encephalitis in America. The New England Journal of Medicine, 344 (24), 1858 1859. World Health Organization (2004), West Nile Virus, available from URL Uhttp://www.who.int/vaccine_research/ documents/new_vaccines/en/index6.html

PAGE 144

129 Bibliography Barker CM, Reisen WK, and Kramer VL. ( 2003). California State Mosquito-Borne Virus Surveillance and Response Plan: A Retr ospective Evaluation Using Conditional Simulations. American Journal of Tropica l Medicine and Hygiene, 68 (5), 508 518. Benjamini E, Coico R, and Sunshine G. (2000). Immunology: A Short Course, 4PthP ed. John Wiley & Sons, New York, USA. Bigler B. (1999). Eastern Equine Encephalitis : Department of Health, Bureau of Epidemiology. Available from URL: H Uhttp://www.doh.state.fl.us/disease_ctrl/epi/htopics/reports/eeeprs.pdf U H Bigler B. (1999). St. Louis Encephalitis : Department of Health, Bureau of Epidemiology. Available from URL: H Uhttp://www.doh.state.fl.us/disease_ctr l/epi/htopics/reports/slepres2.pdf U H Calisher CH, Fremount HN, Vesley WL, et al. (1986c). Relevance of Detection of Immunoglobulin M Antibody Response in Bi rds Used for Arbovirus Surveillance. Journal of Clinical Microbiology, 24(5):770-774. Casals, J. (1957). The Arthropodborne Group of Animal Viruses. Tr NY Acad Sc 19:219-235. Casals J. and Brown LV. (1954). Hemagglu tination with Arthropod-borne Viruses. Journal Exp Med, 99:429-449. Centers for Disease Control and Prevention. (2 001). Fact Sheet:West Nile Virus (WNV) Infection: Information for Clinicians. Available from URL: H Uhttp://www.cdc.gov/ncidod/dvbid/westnile /resources/fact_sh eet_clinician.htm U H Centers for Disease Contro l and Prevention. (2004). TWest Nile Virus: What You Need To Know, CDC Fact Sheet. Available from URL: H Uhttp://www.cdc.gov/ncidod/dvbid/ westnile/wnv_factsheet.htm U H Channock RM and Sabin AB. (1954b). The He magglutinin of West Nile Virus: Recovery, properties and antigenic relationships. Journal of Immunology, 73:352362. Clarke DH, and Casals J. (1958). Technique s for Hemagglutination and Hemaglutinationinhibition with Arthropod-borne Viruses. American Journal of Tropical Medicine and Hygiene, 7:561-573. Collins C, and Blackmore C. (2003). Mosquito-Borne Disease Summary through the Week Ending December 29, 2003. Florida Depa rtment of Healt h. Available from URL H Uhttp://www.doh.state.fl.us/Environm ent/hsee/arbo/data/2003/03-12-29.pdf U H

PAGE 145

130 Cody RP and Smith JK. (1997). Applied Statistics and the SAS Programming Language Cary, North Carolina: Prentice-Hall, Inc. Day JF. (2001). Predicting St. Louis Encephalitis Virus Epidemics: Lessons from Recent, and Not So Recent, Outbreaks. Annual Review of Entomology, 46 111 138. Day JF and Stark LM (1996). Transmission patt erns of St. Louis encephalitis and Eastern Collins C, Conti L, Blackmore C, et al. (2002). Florida Arboviral Activity Summary Delwiche LD and Slaughter SJ. (1998). The Little SAS Book: A Primer, Second Edition Cary, North Carolina: SAS Institute Inc. Ebel GD, Dupuis AP, Ngo K, et al. (2000). Par tial Genetic Characteri zation of West Nile Virus Strains, New York State, 2000. Emerging Infectious Diseases, 7 (4), 650 653. Florida Bureau of Environmental Health. ( 2003). Florida Arbovira l Encephalitis and West Nile Virus Information: Reporting of Denominator and Numerator Data for Arbovirus Surveillance. Available from URL H Uhttp://www.doh.state.fl.us/Environment/hsee/arbo/denom_numer.htm U H Florida Bureau of Environmental Health and Florida Department of Health. (1999). Mosquito-borne Disease in Flor ida. Available from URL H Uhttp://www.doh.state.fl.us/Environmen t/hsee/arbo/pdf_file s/arbo_flyer.pdf U H Goddard LB, Roth AE, Reisen WK, et al (2002). Vector Competence of California Mosquitoes for West Nile Virus. Emerging Infectious Diseases, 8 (12), 1385 1391. Hadler J, Nelson R, McCarthy T, et al. (2000). West Nile Virus Surveillance in Connecticut in 2000: An Intense Epizootic without High Risk for Severe Human Disease. Emerging Infectious Diseases, 7 (4), 636 642. Henderson, JR, Karabatsos N, Bourke ATC, et al (1962). A Survey For Arthropod-Borne Viruses in South-Central Florida. American Journal of Tropical Medical Hygiene, Andreadis TG, Anderson JF, and Vossbrinck CR. (2000). Mosquito Surveillance West Nile Virus in Connect icut, 2000: Isolation from Culex pipiens Cx. restuans Cx. salinariusm and, Culiseta melanura Emerging Infectious Diseases, 7 (4), 670 674. Johnson AJ, Langevin S, Wolff KL, et al. (2003). Detection of Anti-West Nile Virus Immunoglobulin M in Chicken Serum by an Enzyme-Linked Immunosorbent Assay. Journal of Clinical Microbiology, 41(5):2002-2007. Kulasekera VL, Kramer L, Nasci RS, et al (2000). West Nile Virus Infection in Mosquitoes, Birds, Horses, and Humans in Staten Island, New York, 2000. Emerging Infectious Diseases, 7 (4), 722 725.

PAGE 146

131 Marfin AA, Petersen LR, Eidson M, et al. (2000). Widespread West Nile Virus Activity, Eastern United States, 2000. Emerging Infectious Diseases, 7 (4), 730 735. Mitka M. (2003). As West Nile Virus S eason Heats Up, Blood Safety Testing Lags Behind. Journal of American Me dical Association, 289 (18), 2341 2342. Mukhopadhyay S, Kim B, Chipman PR, et al. (2003). Structure of West Nile Virus. Science, 302 248. Nasci RS, Savage HM, White DJ, et al. ( 2000). West Nile Virus in Overwintering Culex Mosquitoes, New York City, 2000. Emerging Infectious Diseases, 7 (4), 742 744. Ostlund EN, Crom RL, Pedersen DD, et al (2000). Equine West Nile Encephalitis, United States. Emerging Infectious Diseases, 7 (4), 665 669. Panella NA, Kerst AJ, Laniciotti RS, et al. (2000). Comparativ e West Nile Virus Detection in Organs of Natura lly Infected American Crows ( Corvus brachyrhynchos ). Emerging Infectious Diseases, 7 (4), 754 755. Rappole JH, Derrickson SR, and Hubalek Z. (20 00). Migratory Birds and Spread of West Nile Virus in the Western Hemisphere. Emerging Infectious Diseases, 6 (4), 319 328. Scherret JH, Poidinger M, Mackenzie JS, et al. (2000). The Relationships between West Nile and Kunjin Viruses. Emerging Infectious Diseases, 7 (4), 697 705. Shaman J, Day JF, Stieglitz M, et al. (2004) Seasonal Forecast of St. Louis Encephalitis Virus Transmission, Florida. Emerging Infectious Diseases, 10 (5), 802 809. 1:800-810 White DJ, Kramer LD, Backenson B, et al. (2000). Mosquito Surveillance and Polymerase Chain Reaction Detection of West Nile Virus, New York State. Emerging Infectious Diseases, 7 (4), 643 649.

PAGE 147

132 Appendices

PAGE 148

133

PAGE 149

134 Appendix I: Table of the Species of Birds that were found Positive for West Nile Virus in the United States since 1999 Bird Species Common Name Native/Exotic/Captive 1 Abyssinian Ground-Hornbill Exotic-captive 2 Acorn Woodpecker Native 3 American Coot Native 4 *American Crow Native 5 American Dipper Native 6 American Flamingo Exotic-captive 7 American Goldfinch Native 8 *American Kestrel Native 9 *American Robin Native 10 American White Pelican Native 11 Anna's Hummingbird Native 12 Bald Eagle Native 13 *Baltimore Oriole Native 14 *Bank Swallow Native 15 Barn Owl Native 16 Barn Swallow Native 17 Barred Owl Native 18 Belted Kingfisher Native 19 Black Phoebe Native 20 Black Skimmer Native 21 Black Vulture Native 22 Black-billed Magpie Native 23 Black-capped Chickadee Native 24 Black-capped Lory Exotic-captive 25 Black-chinned Sparrow Native 26 Black-crowned Night Heron Native 27 Black-footed Penguin Exotic-captive 28 *Black-headed Grosbeak Native 29 Blackpoll Warbler Native 30 Black-throated Blue Warbler Native 31 Black-whiskered Vireo Native 32 *Blue Jay Native 33 Blue-crowned Conure Exotic-captive 34 Blue-eared Pheasant Exotic-captive 35 Blue-streaked Lory Exotic-captive 36 Blythe's Tragopan Exotic-captive 37 *Boat-tailed Grackle Native 38 Bobolink Native

PAGE 150

135 Appendix I: (Continued) Bird Species Common Name Native/Exotic/Captive 39 Boreal Owl Native-captive 40 Brewer's Blackbird Native 41 *Broad-winged Hawk Native 42 Bronze-winged Duck Exotic-captive 43 *Brown Thrasher Native 44 *Brown-headed Cowbird Native 45 Budgerigar Introduced (captive) 46 Bufflehead Native-captive 47 Burrowing Owl Native 48 Cactus Wren Native 49 California Gull Native 50 California Quail Unknown 51 California Towhee Native 52 *Canada Goose Native 53 *Canada Warbler Native 54 Canary-winged Parakeet Exotic-captive 55 Canvasback Native 56 *Carolina Chickadee Native 57 *Carolina Wren Native 58 *Cassin's Finch Native 59 Cedar Waxwing Native 60 Chihuahuan Raven Native 61 Chilean Flamingo Exotic-captive 62 Chimney Swift Native 63 Chinese Goose Exotic-captive 64 Chukar Introduced-captive 65 Cinnamon Teal Native-captive 66 Clark's Grebe Native 67 Clark's Nutcracker Native-captive 68 Cliff Swallow Native 69 *Cockatiel Exotic-captive 70 Cockatoo Exotic-captive 71 Common Canary Exotic-captive 72 Common Goldeneye Native-captive 73 *Common Grackle Native 74 Common Ground-Dove Native 75 Common Loon Native 76 Common Merganser Native-captive 77 Common Moorhen Native

PAGE 151

136 Appendix I: (Continued) Bird Species Common Name Native/Exotic/Captive 78 Common Nighthawk Native 79 Common Peafowl Exotic-captive 80 Common Raven Native 81 *Common Yellowthroat Native 82 *Cooper's Hawk Native 83 Crimson Rosella Exotic-captive 84 Dark-eyed Junco Native 85 Dickcissel Native 86 Domestic Chicken Exotic-captive 87 Double-crested Cormorant Native 88 Downy Woodpecker Native 89 Dusky Lory Exotic-captive 90 *Eastern Bluebird Native 91 Eastern Kingbird Native 92 *Eastern Phoebe Native 93 *Eastern Screech-Owl Native 94 Eastern Towhee Native 95 Elegant Crested Tinamou Exotic-captive 96 Elf Owl Unknown 97 Emperor Goose Native-captive 98 Emu Exotic-captive 99 *Eurasian Collared-Dove Introduced 100 Eurasian Jay Exotic-captive 101 Eurasian Wigeon Native-captive 102 European Goldfinch Exotic-captive 103 *European Starling Introduced 104 Evening Grosbeak Native 105 Ferruginous Hawk Native 106 Field Sparrow Native 107 *Fish Crow Native 108 Flammulated Owl Native-Captive 109 Fox Sparrow Native 110 Gila Woodpecker Native 111 *Golden Eagle Native 112 Gouldian Finch Exotic-captive 113 *Gray Catbird Native 114 Gray-cheeked Thrush Native 115 Great Black-backed Gull Native 116 Great Blue Heron Native

PAGE 152

137 Appendix I: (Continued) Bird Species Common Name Native/Exotic/Captive 117 Great Crested Flycatcher Native 118 Great Egret Native 119 Great Gray Owl Native-captive 120 *Great Horned Owl Native 121 Greater Prairie-Chicken Native 122 Greater Sage-Grouse Native 123 Greater Scaup Native-captive 124 Greater White-fronted Goose Native 125 Great-tailed Grackle Native 126 *Green Heron Native 127 Guanay Cormorant Exotic-captive 128 Gyrfalcon Native-Captive 129 Hairy Woodpecker Native 130 Hammond's Flycatcher Native 131 Harris' Hawk Native-captive 132 Hawaiian Goose (Nene) Exotic-captive 133 *Hermit Thrush Native 134 Herring Gull Native 135 Hooded Crow Exotic-captive 136 Hooded Merganser Native-Captive 137 Hooded Oriole Native 138 *Hooded Warbler Native 139 *House Finch Native 140 *House Sparrow Introduced 141 *House Wren Native 142 Humboldt Penguin Exotic-captive 143 Impeyan Pheasant Exotic-captive 144 Inca Dove Native 145 Inca Tern Exotic-captive 146 Kentucky Warbler Native 147 Killdeer Native 148 Lark Sparrow Native 149 *Laughing Gull Native 150 Least Bittern Native 151 Lesser Goldfinch Native 152 Lesser Nighthawk Native 153 Lesser Scaup Native-captive 154 Lewis' Woodpecker Native 155 *Limpkin Native

PAGE 153

138 Appendix I: (Continued) Bird Species Common Name Native/Exotic/Captive 156 *Loggerhead Shrike Native 157 Long-eared Owl Native 158 Macaw Exotic-captive 159 *Mallard Native 160 *Merlin Native 161 Mexican Jay Native 162 Micronesian Kingfisher Exotic-captive 163 Mississippi Kite Native 164 Monal Pheasant Exotic-captive 165 *Mottled Duck Native 166 Mountain Bluebird Native 167 Mountain Chickadee Native 168 Mountain Quail Native 169 *Mourning Dove Native 170 Muscovy Duck Exotic 171 Mute Swan Introduced 172 Nashville Warbler Native 173 Northern Bobwhite Native 174 *Northern Cardinal Native 175 Northern Flicker Native 176 Northern Goshawk Native 177 Northern Harrier Native 178 Northern Hawk-Owl Native-captive 179 *Northern Mockingbird Native 180 Northern Parula Native 181 Northern Saw-whet Owl Native 182 Northern Waterthrush Native 183 Olive-sided Flycatcher Native 184 *Orange-crowned Warbler Native 185 Orchard Oriole Native 186 Osprey Native 187 Ovenbird Native 188 Pacific Parrotlet Exotic-captive 189 Pacific-slope Flycatcher Native 190 Palm Tanager Exotic-captive 191 Peregrine Falcon Native 192 Pied-billed Grebe Native 193 Pine Siskin Native 194 Pinyon Jay Native

PAGE 154

139 Appendix I: (Continued) Bird Species Common Name Native/Exotic/Captive 195 Piping Plover Native 196 Prairie Falcon Native-captive 197 Puna Teal Exotic-captive 198 *Purple Finch Native 199 *Purple Gallinule Native 200 Purple Martin Native 201 Pygmy Nuthatch Native 202 Rainbow Lorikeet Exotic-captive 203 Red Crossbill Native 204 Red Lory Exotic-captive 205 Red-bellied Woodpecker Native 206 Red-breasted Goose Exotic-captive 207 Red-breasted Sapsucker Native 208 Red-crowned Parrot Exotic-captive 209 Red-eyed Vireo Native 210 Red-headed Woodpecker Native 211 *Red-shouldered Hawk Native 212 *Red-tailed Hawk Native 213 *Red-winged Blackbird Native 214 Ring-billed Gull Native 215 Ring-necked Pheasant Introduced 216 *Rock Dove Introduced 217 Rose-breasted Grosbeak Native 218 Rough-legged Hawk Native-captive 219 Ruby-throated Hummingbird Native 220 Ruddy Duck Native 221 Ruddy Turnstone Native 222 Ruffed Grouse Native 223 Rufous Hummingbird Native 224 Rusty Blackbird Native 225 Sandhill Crane Native 226 Satyr Tragopan Exotic-captive 227 Savannah Sparrow Native 228 Scarlet Ibis In troduced(captive) 229 Scarlet Tanager Native 230 Scissor-tailed Flycatcher Native 231 Sharp-shinned Hawk Native 232 Short-eared Owl Native 233 Smew Exotic-captive

PAGE 155

140 Appendix I: (Continued) Bird Species Common Name Native/Exotic/Captive 234 Snowy Owl Native-captive 235 Society Finch Exotic-captive 236 Song Sparrow Native 237 Spotted Owl Native-captive 238 Spotted Towhee Native 239 Steller's Jay Native 240 Swainson's Hawk Native 241 *Swainson's Thrush Native 243 Swallow-tailed Kite Native 244 Tawny Owl Exotic-captive 245 Thick-billed Parrot Exotic-captive 246 Townsend's Warbler Native 247 Traill's Flycatcher Native 248 Tree Swallow Native 249 *Tufted Titmouse Native 250 Tundra Swan Native-captive 251 Turkey Vulture Native 252 Varied Thrush Native 253 Varied Tit Exotic-captive 254 Veery Native 255 Violet-necked Lorikeet Exotic-captive 256 Virginia Rail Native 257 Warbling Vireo Native 258 Wedge-tail Eagle Exotic-captive 259 Western Bluebird Native 260 Western Kingbird Native 261 Western Meadowlark Native 262 Western Sandpiper Native 263 Western Screech-Owl Native 264 Western Scrub-Jay Native 265 Western Tanager Native 266 White-breasted Nuthatch Native 267 *White-crowned Pigeon Native 268 White-crowned Sparrow Native 269 White-tailed Kite Unknown 270 White-winged Dove Native 271 Wild Turkey Native 272 Wilson's Warbler Native 273 Winter Wren Native

PAGE 156

141 Appendix I: (Continued) Bird Species Common Name Native/Exotic/Captive 274 *Wood Duck Native 275 Wood Thrush Native 276 Yellow Warbler Native 277 *Yellow-bellied Sapsucker Native 278 *Yellow-billed Cuckoo Native 279 Yellow-billed Duck Exotic-captive 280 Yellow-billed Magpie Native 281 Yellow-crowned Night-Heron Native 282 *Yellow-rumped Warbler Native 283 *Zebra Finch Exotic-captive 284 Zenaida Dove Exotic-captive *Found positive in Florida Adapted from the CDC compiled list of We st Nile Virus Positive Bird Species. Available from URL: Uhttp://www.cdc.gov/ncidod/dvbid/ westnile/birdspecies.htm

PAGE 157

142 Appendix II: Table of the Species of Mosquito es that were found Positive for West Nile Virus in Mosquito Pools in the United States since 1999 Mosquito Species Aedes albopictus Aedes aegypti Aedes vexans Aedes Aedes cinereus Anopheles barberi *Anopheles atropos *Anopheles crucians/bradleyi Anopheles punctipennis Anopheles quadrimaculatus Anopheles Anopheles walkeri Coquillettidia Coquillettidia perturbans Culiseta Culiseta inornata *Culiseta melanura *Culex erraticus *Culex nigripalpus Culex pipiens Culex quinquefasciatus Culex restuans *Culex salinarius Culex Culex tarsalis Culex territans Deinocerites *Deinocerites cancer Ochlerotatus atropalpus *Ochlerotatus atlanticus/tormentor Ochlerotatus canadensis Ochlerotatus cantator Ochlerotatus dorsalis Ochlerotatus f itchii *Ochlerotatus infirmatus Ochlerotatus j aponicus Ochlerotatus provocans Ochlerotatus sollicitans Ochlerotatus sticticus Ochlerotatus stimulans Ochlerotatus *Ochlerotatus taeniorhynchus Ochlerotatus triseriatus Ochlerotatus trivittatus Orthopodom y ia Orthopodomyia signifera Psorophora ciliata Psorophora columbiae Psorophora ferox Psorophora TPsorophora howardiiT Uranotaenia Uranotaenia sapphirina *Found positive in Florida Table adapted from CDC List of Species of West Nile Positive Mosquito Pools. Available from URL Uhttp://www.cdc.gov/ncidod/dvbid/we stnile/mosquitoSpecies.htm

PAGE 158

143 Appendix III: West Nile virus reservoir comp etence index values for 25 species of birds Mean Days Mean PeakU Species n Infectious* Viremia ** c i *** UBlue jay 2 4 12.3 2.4 Common grackle 6 3 9.4 1.0 House sparrow 6 3 8.9 0.9 House finch 2 6 8.8 0.8 American robin 2 3 8.5 0.6 Red-wing. blackbird 3 3 8.1 0.5 Mallard 2 3 6.7 0.3 European starling 6 2 6.0 0.1 Canada goose 3 0 4.7 0 American coot 1 0 4.6 0 Rock dove 6 0 4.3 0 Chicken 16 0 3.2 0 Ring-neck Pheasant 3 0 2.7 0 *Infectious viremia = log 5 or great er per ml serum; ** log pfu/ml serum *** c i = susceptibility mean infe ctiousness days infectious *Table adapted from (Komar, 2003).

PAGE 159

144 Appendix VI: 2001 data by County County Region Clinical Cases Avian Positive Mosquito Positive Sentinel Rate Sentinel Positive Total Avian Alachua North 058 0.024120 341 Baker North 02 4 Bay Panhandle 08410.030514 458 Bradford North 019 63 Brevard Central 03 0.00321 337 Broward South 014 185 Calhoun Panhandle 08 16 Charlotte South 08 0.00000 95 Citrus North 01500.02713 142 Clay North 051 239 Collier South 07 0.02635 203 Columbia North 029 50 Dade South 021 393 De Soto Central 00 0.00000 28 Dixie North 09 35 Duval North 110300.092432 574 Escambia Panhandle 09 215 Flagler North 02 0.00000 122 Franklin Panhandle 04 21 Gadsden Panhandle 027 52 Gilchrist North 011 37 Glades South 01 4 Gulf Panhandle 015 56 Hamilton North 014 31 Hardee Central 00 2 Hendry South 02 0.00000 159 Hernando North 08 127 Highlands Central 02 47 Hillsborough Central 02 0.00252 220 Holmes Panhandle 0120 34 Indian River Central 00 0.00403 35 Jackson Panhandle 012 96 Jefferson Panhandle 1150 26 Lafayette North 08 15 Lake North 014 143 Lee South 02 0.00254 11 Leon Panhandle 1109 0.050850 305 Levy North 023 94 Liberty Panhandle 09 17 Madison North 2101 35 Manatee Central 04 0.00000 87 Marion North 141 173 Martin South 03 0.03478 37 Monroe South 2186 165 Nassau North 024 96 Okaloosa Panhandle 0250 406 Okeechobee Central 01 0.08802 19 Orange North 03 0.00143 260 Osceola Central 01 0.00212 42

PAGE 160

145 Appendix IV: (Continued) County Region Clinical Cases Avian Positive Mosquito Positive Sentinel Rate Sentinel Positive Total Avian Palm Beach South 18 0.00222 204 Pasco North 021 0.00271 246 Pinellas Central 0120.00792 203 Polk Central 06 0.01521 60 Putnam North 112 0.03188 82 Santa Rosa Panhandle 0140 153 Sarasota Central 1400.00000 165 Seminole North 04 0.00151 200 St. Johns North 00 0.057420 32 St. Lucie Central 02 0.00480 37 Sumter North 01 35 Suwannee North 047 0.12503 92 Taylor North 013 29 Union North 08 32 Volusia North 0300.00000 76 Wakulla Panhandle 064 108 Walton Panhandle 08 0.00695 82 Washington Panhandle 2171 44

PAGE 161

146 Appendix V: 2002 data by County County Region Clinical Cases Positive Avians Positive Mosquitoes Sentinel Rate Sentinel Positive Total_Avian Alachua North 217 0.017127 196 Baker North 10 0 Bay Panhandle 06 0.00255 225 Bradford North 01 13 Brevard Central 11 0.041060 86 Broward South 01 137 Calhoun Panhandle 02 8 Charlotte South 01 0.035234 20 Citrus North 1520.031824 70 Clay North 13 73 Collier South 11200.039035 147 Columbia North 02 35 Dade South 11 162 De Soto Central 01 0.02363 14 Dixie North 01 18 Duval North 10 0.046319 36 Escambia Panhandle 71210 297 Flagler North 06 0.060316 20 Franklin Panhandle 00 3 Gadsden Panhandle 00 1 Gilchrist North 05 22 Glades South 00 7 Gulf Panhandle 00 3 Hamilton North 00 15 Hardee Central 02 3 Hendry South 01 0.00000 19 Hernando North 06 99 Highlands Central 14 21 Hillsborough Central 18 0.016344 89 Holmes Panhandle 03 11 Indian River Central 00 0.016939 4 Jackson Panhandle 134 138 Jefferson Panhandle 00 7 Lafayette North Lake North 134 56 Lee South 12 0.045261 13 Leon Panhandle 02 0.00388 83 Levy North 08 44 Liberty Panhandle Madison North 00 24 Manatee Central 10 0.035050 2 Marion North 334 96 Martin South 00 0.046831 25 Monroe South 001 58 Nassau North 00 25 Okaloosa Panhandle 050 144 Okeechobee Central 02 0.040715 31 Orange North 1910.0129128 118 Osceola Central 00 0.031144 21

PAGE 162

147 Appendix V: (Continued) County Region Clinical Cases Positive Avians Positive Mosquitoes Sentinel Rate Sentinel Positive Total_Avian Palm Beach South 1720.010851 277 Pasco North 011 0.014029 81 Pinellas Central 0620.013326 43 Polk Central 10 20 Putnam North 07 0.156922 59 Santa Rosa Panhandle 11100.00000 98 Sarasota Central 21900.012577 276 Seminole North 05 0.013720 88 St. Johns North 00130.086255 1 St. Lucie Central 01 0.011312 39 Sumter North 15 32 Suwannee North 00 5 Taylor North 00 12 Union North 01 10 Volusia North 0570.034871 42 Wakulla Panhandle 00 3 Walton Panhandle 0300.019319 43 Washington Panhandle 0300.00000 31

PAGE 163

148 Appendix VI: 2003 data by County County Region Clinical Cases Avian Positive Mosquito Positive Sentinel Rate Sentinel Positive Total Avian Alachua North 11200.014123 79 Baker North 00 1 Bay Panhandle 14103 0.055948 147 Bradford North 05 14 Brevard Central 10 0.028528 1 Broward South 450 33 Calhoun Panhandle 28 14 Charlotte South 00 0.061742 2 Citrus North 2200.023723 32 Clay North 03 43 Collier South 23130.056459 4 Columbia North 05 13 Dade South 616 0.00369 217 De Soto Central 10 3 Dixie North 00 13 Duval North 6200.047827 20 Escambia Panhandle 12763 233 Flagler North 0 0.02487 Franklin Panhandle 12 6 Gadsden Panhandle Gilchrist North 010 7 Glades South 01 1 Gulf Panhandle 4 Hamilton North 01 11 Hardee Central Hendry South 0 0.091453 Hernando North 01 65 Highlands Central 00 1 Hillsborough Central 11 0.024740 47 Holmes Panhandle 25 13 Indian River Central 00 0.019147 4 Jackson Panhandle 032 0.022820 123 Jefferson Panhandle 02 0.094419 14 Lafayette North 1 Lake North 00 2 Lee South 3 0.0814116 Leon Panhandle 08 0.018245 84 Levy North 0100 25 Liberty Panhandle 02 8 Madison North 04 29 Manatee Central 0 0.021464 Marion North 2200.024819 45 Martin South 01 0.033624 9 Monroe South 1210 52 Nassau North 01 0.024928 14 Okaloosa Panhandle 8450 192 Okeechobee Central 03 0.038114 15 Orange North 00 0.010368 49 Osceola Central 01 0.024524 13

PAGE 164

149 Appendix VI: (Continued) County Region Clinical Cases Avian Positive Mosquito Positive Sentinel Rate Sentinel Positive Total Avian Palm Beach South 09300.0281113 167 Pasco North 01 0.00316 20 Pinellas Central 0100.008015 47 Polk Central 00 10 Putnam North 00 0.082941 1 Santa Rosa Panhandle 102000.00000 86 Sarasota Central 1100.020168 41 Seminole North 20 0.00585 37 St. Johns North 1 00.041750 St. Lucie Central 01 0.010012 12 Sumter North 00 3 Suwannee North 10 1 Taylor North 08 13 Union North 12 12 Volusia North 1000.008518 3 Wakulla Panhandle 03 10 Walton Panhandle 1500.0261173 39 Washington Panhandle 1800.00000 21

PAGE 165

150 Appendix VII: Number of Sentinel Chicke n Sites and Chickens per Site by Year County 200120022003 Alachua 766 Bay 444 Brevard 101010 Charlotte 569 Citrus 777 Collier 669 Desoto 122 Dade n/an/a6 Duval 688 Flagler 445 Hendry 225 Jackson n/an/a6 Jefferson n/an/a5 Okeechobee 333 Orange 121213 Orange/Reedy 878 Osceola 8912 Palm Beach 81010 Pasco 566 Pinellas 888 Polk 988 Hillsborough 7811 Indian River 889 Lee 181818 Leon 999 Manatee 91115 Marion n/an/a6 Martin 558 Nassau n/an/a7 Putnam 999 Sarasota 101012 Seminole 556 St. Johns 8810 St. Lucie 556 Suwannee 1n/an/a Santa Rosa n/a** Volusia 61017 Walton (North) n/an/a8 Walton (South) 51516 Washington 033 n/a=not included for that year

PAGE 166

151 Appendix VIII: County Arbosurveillance Table 2001 2002 2003 County Sentinel Avian Mosquito Sentinel Avi an Mosquito Sentinel Avian Mosquito Panhandle Bay Calhoun Escambia Franklin Gadsden Gulf Holmes Jackson Jefferson Leon Okaloosa Santa Rosa Wakulla Walton Washington North Alachua Baker Bradford Citrus Clay Columbia Dixie Duval Flagler Gilchrist Hamilton Hernando Lafayette Lake Levy Liberty Madison Marion Nassau Orange Pasco Putnam Seminole

PAGE 167

152 Appendix VIII: (Continued) 2001 2002 2003 County Sentinel Avian Mosquito Sentinel Avi an Mosquito Sentinel Avian Mosquito St Johns Sumter Suwannee Taylor Union Volusia Central Brevard Desoto Hardee Highlands Hillsboroug h Indian River Manatee Okeechobee Osceola Pinellas Polk Sarasota St Lucie South Broward Charlotte Collier Dade Glades Hendry Lee Martin Monroe Palm Beach participation in the surveillance system

PAGE 168

153 Appendix IX: Regional Map of Fl orida Numbered by Location from East (Right) to West (Left) and North (Top) to South (Bottom) ESCAMBIA SANTA ROSA OKALOOSA WALTON HOLMES WASHINGTON BAY JACKSON CALHOUN GULF LIBERTY FRANKLIN GADSDEN LEON WAKULLA JEFFERSON TAYLOR MADISON HAMILTON SUWANNEE LAFAYETTE DIXIE LEVY GILCHRIST COLUMBIA ALACHUA UNION BRADFRD NASSAU MARION CITRUS HERNADO PASCO SUMTER LAKE PUTNAM CLAY BAKER DUVAL ST JOHNS FLAGLER VOLUSIA SEMINOLE ORANGE BREVARD OSCEOLA POLK HILLSBOROUGH PINELLAS MANATEE HARDEE DESOTO CHARLOTTE LEE GLADES HENDRY COLLIER MONROE DADE BROWARD PALM BEACH MARTIN ST LUCIE OKEECHOBEE INDIAN RIVER HIGHLANDS SARASOTA3 1 4 2 121110 9 8 7 6 5 22 23 21 20 19 17 18 16 14 15 13 24 25 26 27 282930 37 36 35 34 33 32 31 47 46 45 44 43 42 41 40 39 38 57 56 55 54 53 52 51 50 49 48 67 66 65 64 63 62 61 60 59 58URegion Panhandle Central North South

PAGE 169

154 Appendix X: Incidence Rate and Number of Clinical Cases per County by Year County Region Location 2001 Incidence 2002 Incidence 2003 Incidence 2001 Count 2002 Count 2003 Count Escambia Panhandle 1 2.302.30 77 Santa Rosa Panhandle 2 0.792.31 13 Okaloosa Panhandle 3 2.75 5 Holmes Panhandle 5 10.53 2 Washington Panhandle 64.65 4.551 1 Bay Panhandle 7 5.15 8 Jackson Panhandle 8 2.08 1 Calhoun Panhandle 9 14.83 2 Gulf Panhandle 10 19.14 3 Leon Panhandle 140.41 1 Jefferson Panhandle 167.63 1 Madison North 1710.59 2 Lafayette North 20 13.53 1 Suwannee North 21 2.74 1 Alachua North 26 0.870.43 21 Union North 27 7.26 1 Baker North 28 4.33 1 Clay North 30 0.64 1 Duval North 310.130.120.241 12 St Johns North 33 0.71 1 Putnam North 351.41 1 Marion North 360.381.100.351 31 Citrus North 37 0.790.79 11 Pasco North 39 0.26 1 Sumter North 40 1.612.31 1 Lake North 41 0.43 1 Volusia North 42 0.21 1 Seminole North 43 0.25 1 Orange North 44 0.10 1 Brevard Central 45 0.20 1 Polk Central 50 0.20 1 Highlands Central 51 1.12 1 DeSoto Central 52 2.95 1 Hillsborough Central 54 0.09 1 Manatee Central 56 0.36 1 Sarasota Central 570.300.590.291 21 Lee South 59 0.210.40 12 Collier South 60 0.340.68 12 Monroe South 612.47 1.242 1 Dade South 62 0.040.30 17 Broward South 63 0.18 3 Palm Beach South 660.090.080.081 11

PAGE 170

155 Appendix XI: Graphs of Pooled Annual Av erages (2001-2003) with Moving Averages for West Nile Surveillance Data Statewide and Regional per Week Plot of the Combined Year (2001-2003) Av erage and the Moving Average Models for the Number of Clinical Cases in Florida. 0 1 2 3 4 5 61 3 5 7 9 11 13 1 5 17 1 9 21 23 25 27 29 31 33 35 37 39 41 43 4 5 47 4 9 51WeekNumber of Clinical Cases Actual Moving Average Plot of the Combined Year (2001) Aver age and the Moving Average Models for the Number of Positive Dead Birds in Florida. 0 10 20 30 40 50 601 3 5 7 9 11 13 15 17 19 21 23 25 27 29 3 1 3 3 35 3 7 39 41 43 45 47 49 51WeekNumber of Positive Dead Birds Actual Moving Average

PAGE 171

156 Appendix XI: (Continued) Plot of the Combined Year (2002) Aver age and the Moving Average Models for the Sentinel Seroconversion Rate in Florida. 0.00 0.01 0.02 0.03 0.04 0.05 0.061 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 3 5 37 39 41 43 45 47 49 51WeekSentinel Seroconversion Rate Actual Moving Average Plot of the Combined Year (2003) Aver age and the Moving Average Models for the Number of Positive Mosquito Pools in Florida. 0 1 2 3 4 5 6 7 81 3 5 7 9 11 13 15 17 19 21 23 25 27 29 3 1 33 3 5 3 7 39 41 43 45 47 49 51WeekNumber of Positive Mosquito Pools Actual Moving Average

PAGE 172

157 Appendix XI: (Continued) Plot of the Combined Year (2001-2003) Av erage and the Moving Average Models for the Number of Clinical Cases in the Panhandle Region of Florida. 0 0.5 1 1.5 2 2.5 3 3.51 3 5 7 9 11 13 1 5 17 1 9 21 23 25 27 29 31 33 35 37 39 41 43 45 4 7 49 51WeekNumber of Clinical Cases Actual Moving Average Plot of the Combined Year (2001) Aver age and the Moving Average Models for the Number of Positive Dead Birds in the Panhandle Region of Florida. 0 5 10 15 20 25 30 351 3 5 7 9 11 13 15 17 19 21 23 25 27 29 3 1 3 3 35 3 7 39 41 43 45 47 49 51WeekNumber of Positive Dead Birds Actual Moving Average

PAGE 173

158 Appendix XI: (Continued) Plot of the Combined Year (2002) Aver age and the Moving Average Models for the Sentinel Seroconversion Rate in the Panhandle Region of Florida. 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.141 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 3 5 37 39 41 43 45 47 49 51WeekSentinel Seroconversion Rate Actual Moving Average Plot of the Combined Year (2003) Aver age and the Moving Average Models for the Number of Positive Mosquito Pools in the Panhandle Region of Florida. 0 0.2 0.4 0.6 0.8 1 1.21 3 5 7 9 11 13 15 17 19 21 23 25 27 29 3 1 33 3 5 37 39 41 43 45 47 49 51WeekNumber of Positive Mosquito Pools Actual Moving Average

PAGE 174

159 Appendix XI: (Continued) Plot of the Combined Year (2001-2003) Av erage and the Moving Average Models for the Number of Clinical Cases in the Northern Region of Florida. 0 0.2 0.4 0.6 0.8 1 1.2 1.41 3 5 7 9 11 13 1 5 17 1 9 21 23 25 27 29 31 33 35 37 39 41 43 45 4 7 49 51WeekNumber of Clinical Cases Actual Moving Average Plot of the Combined Year (2001) Averag e and the Moving Average Models for the Number of Positive Dead Birds in the Northern Region of Florida. 0 5 10 15 20 25 30 351 3 5 7 9 11 13 15 17 19 21 23 25 27 29 3 1 3 3 35 3 7 39 41 43 45 47 49 51WeekNumber of Positive Dead Birds Actual Moving Average

PAGE 175

160 Appendix XI: (Continued) Plot of the Combined Year (2002) Aver age and the Moving Average Models for the Sentinel Seroconversion Rate in the Northern Region of Florida. 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.091 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 3 5 37 39 41 43 45 47 49 51WeekSentinel Seroconversion Rate Actual Moving Average Plot of the Combined Year (2003) Aver age and the Moving Average Models for the Number of Positive Mosquito Pools in the Northern Region of Florida. 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.61 3 5 7 9 11 13 15 17 19 21 23 25 27 29 3 1 33 3 5 37 39 41 43 45 47 49 51WeekNumber of Positive Mosquito Pools Actual Moving Average

PAGE 176

161 Appendix XI: (Continued) Plot of the Combined Year (2001-2003) Av erage and the Moving Average Models for the Number of Clinical Cases in the Central Region of Florida. 0 0.2 0.4 0.6 0.8 1 1.2 Week Actual Moving Average Plot of the Combined Year (2001) Averag e and the Moving Average Models for the Number of Positive Dead Birds in the Central Region of Florida. 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 51 3 5 7 9 11 13 15 17 19 21 23 25 27 29 3 1 33 3 5 37 39 41 43 45 47 49 51WeekNumber of Positive Dead Birds Actual Moving Average

PAGE 177

162 Appendix XI: (Continued) Plot of the Combined Year (2002) Averag e and the Moving Average Models for the Sentinel Seroconversion Rate in the Central Region of Florida. 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.071 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 3 5 37 39 41 43 45 47 49 51WeekSentinel Seroconversion Rate Actual Moving Average Plot of the Combined Year (2003) Averag e and the Moving Average Models for the Number of Positive Mosquito Pools in the Central Region of Florida. 0 0.2 0.4 0.6 0.8 1 1.21 3 5 7 9 11 13 15 17 19 21 23 25 27 29 3 1 33 3 5 37 39 41 43 45 47 49 51WeekNumber of Positive Mosquito Pools Actual Moving Average

PAGE 178

163 Appendix XI: (Continued) Plot of the Combined Year (2001-2003) Av erage and the Moving Average Models for the Number of Clinical Cases in the Southern Region of Florida. 0 0.2 0.4 0.6 0.8 1 1.21 3 5 7 9 11 13 1 5 17 1 9 21 23 25 27 29 31 33 35 37 39 41 43 45 4 7 49 51WeekNumber of Clinical Cases Actual Moving Average Plot of the Combined Year (2001) Averag e and the Moving Average Models for the Number of Positive Dead Birds in the Southern Region of Florida. 0 1 2 3 4 5 6 71 3 5 7 9 11 13 15 17 19 21 23 25 27 29 3 1 33 3 5 3 7 39 41 43 45 47 49 51WeekNumber of Positive Dead Birds Actual Moving Average

PAGE 179

164 Appendix XI: (Continued) Plot of the Combined Year (2002) Averag e and the Moving Average Models for the Sentinel Seroconversion Rate in the Southern Region of Florida. 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.091 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 3 5 37 39 41 43 45 47 49 51WeekSentinel Seroconversion Rate Actual Moving Average Plot of the Combined Year (2003) Averag e and the Moving Average Models for the Number of Positive Mosquito Pools in the Southern Region of Florida. 0 1 2 3 4 5 6 7 8 9 101 3 5 7 9 11 13 15 17 19 21 23 25 27 29 3 1 3 3 35 3 7 39 41 43 45 47 49 51WeekNumber of Positive Mosquito Pools Actual Moving Average

PAGE 180

165 Appendix XII: PROC GENMOD Po isson Regression SAS Output Poisson Regression Output for Pooled Years (2001-2003) Human Incidence, Avian, Mosquito with Sentinel Rates including Region Parameter DF Estimate StdErr Lower WaldCL Upper WaldCL ChiSq Prob ChiSq Intercept 1 -55.7542 3.9755 -63.5460 -47.9623 196.68 < 0.0001 Region: Central 1 -0.8253 0.4436 -1.6948 0.0440 3.46 0.0628 Region: North 1 -2.1877 0.3994 -2.9706 -1.4048 30.00 < 0.0001 Region:Panhandle 1 -3.4272 0.4176 -4.2457 -2.6086 67.34 < 0.0001 Region: South 0 0.0000 0.0000 0.0000 0.0000 Year 1 23.5276 1.3590 20.8640 26.1911 299.73 < 0.0001 Avian Rate 1 -0.0133 0.0035 -0.0201 -0.0065 15.08 < 0.0001 Mosquito Rate 1 0.0393 0.0052 0.0291 0.0495 57.42 < 0.0001 Sentinel Rate 1 -0.0172 0.0012 -0.0195 -0.0148 211.90 < 0.0001 Scale 0 1 0 1 1 Human Incidence with Avian Rates including Region Intercept 1 -61.0452 2.1902 -65.3380 -56.7524 776.82 < 0.0001 Region: Central 1 -8.9956 0.3446 -9.6710 -8.3201 681.38 < 0.0001 Region: North 1 -7.0472 0.3177 -7.6699 -6.4244 491.95 < 0.0001 Region:Panhandle 1 -6.5669 0.3933 -7.3377 -5.7960 278.82 < 0.0001 Region: South 0 0.0000 0.0000 0.0000 0.0000 Year 1 26.8621 0.7923 25.3092 28.4150 1149.48 < 0.0001 Avian Rate 1 -0.0496 0.0034 -0.0564 -0.0429 207.97 < 0.0001 Scale 0 1 0 1 1 Human Incidence with Mosquito Rates including Region Intercept 1 -48.5445 3.9168 -56.2214 -40.8677 153.61 < 0.0001 Region: Central 1 -2.2198 0.2308 -2.6722 -1.7675 92.53 < 0.0001 Region: North 1 -1.0844 0.1567 -1.3914 -0.7773 47.92 < 0.0001 Region:Panhandle 1 -1.8177 0.1127 -2.0385 -1.5969 260.38 < 0.0001 Region: South 0 0.0000 0.0000 0.0000 0.0000 Year 1 19.8966 1.3111 17.3270 22.4663 230.31 < 0.0001 Mosquito Rate 1 0.0870 0.0048 0.0775 0.0965 322.38 < 0.0001 Scale 0 1 0 1 1 Human Incidence with Sentinel Rates including Region Intercept 1 -52.8139 2.0459 -56.8239 -48.8039 666.36 < 0.0001 Region: Central 1 0.3472 0.1920 -0.0290 0.7236 3.27 0.0706 Region: North 1 -0.9973 0.1293 -1.2507 -0.7438 59.50 < 0.0001 Region:Panhandle 1 -1.4530 0.1093 -1.6673 -1.2388 176.74 < 0.0001 Region: South 0 0.0000 0.0000 0.0000 0.0000 Year 1 22.3450 0.6838 21.0048 23.6853 1067.78 < 0.0001 Sentinel Rate 1 -0.0223 0.0010 -0.0242 -0.0204 535.95 < 0.0001 Scale 0 1 0 1 1 Human Incidence, Avian, Mosquito with Sentinel Rates Intercept 1 -51.8830 3.9198 -59.5656 -44.2003 175.20 < 0.0001 Year 1 21.4351 1.3125 18.8627 24.0075 266.73 < 0.0001 Avian Rate 1 0.0099 0.0008 0.0083 0.0116 139.60 < 0.0001 Mosquito Rate 1 0.0120 0.0044 0.0035 0.0206 7.60 0.0058 Sentinel Rate 1 -0.0184 0.0011 -0.0206 -0.0162 264.47 < 0.0001 Scale 0 1 0 1 1

PAGE 181

166 Appendix XII: (Continued) Poisson Regression Output for Pooled Years (2001-2003) continued Human Incidence with Avian Rates Intercept 1 -45.2244 2.0356 -49.2142 -41.2347 493.58 < 0.0001 Year 1 18.4753 0.6813 17.1399 19.8107 735.28 < 0.0001 Avian Rate 1 0.0158 0.0015 0.0129 0.0187 114.06 < 0.0001 Scale 0 1 0 1 1 Human Incidence with Mosquito Rates Intercept 1 -51.0595 2.0316 -55.0414 -47.0777 631.64 < 0.0001 Year 1 21.3938 0.6783 20.0643 22.7233 994.66 < 0.0001 Mosquito Rate 1 -0.0190 0.0007 -0.0204 -0.0176 682.60 < 0.0001 Scale 0 1 0 1 1 Human Incidence with Sentinel Rates Intercept 1 -51.0595 2.0316 -55.0414 -47.0777 631.64 < 0.0001 Year 1 21.3938 0.6783 20.0643 22.7233 994.66 < 0.0001 Sentinel Rate 1 -0.0190 0.0007 -0.0204 -0.0176 682.60 < 0.0001 Scale 0 1 0 1 1 Poisson Regression Output for 2001 Parameter DF Estimate StdErr Lower WaldCL Upper WaldCL ChiSq Prob ChiSq Human Incidence with Avian Rates including Region Intercept 1 -10.9622 4.6423 -20.0610 -1.8634 5.58 0.0182 Region: Central 1 -0.3554 7.9922 -16.0198 15.3090 0.00 0.9645 Region: North 1 1.2820 5.4996 -9.4970 12.0611 0.05 0.8157 Region:Panhandle 1 3.8569 5.0179 -5.9779 13.6917 0.59 0.4421 Region: South 0 0.0000 0.0000 0.0000 0.0000 Avian Rate 1 -0.7181 11.5965 -23.4468 22.0105 0.00 0.9506 Scale 0 1 0 1 1 Human Incidence with Mosquito Rates including Region Intercept 1 -14.8761 42.2150 -97.6159 67.8637 0.12 0.7245 Region: Central 1 -20.9219 0.00 1.0000 Region: North 1 5.2308 41.6165 -76.3359 86.7976 0.02 0.9000 Region:Panhandle 1 -19.4291 0.00 1.0000 Region: South 0 0.0000 0.0000 0.0000 0.0000 Year 1 36.4068 213.1301 -381.3205 454.1341 0.03 0.8644 Mosquito Rate 1 1 0 1 1 Scale 0 -14.8761 42.2150 -97.6159 67.8637 0.12 0.7245 Human Incidence with Sentinel Rates including Region Intercept 1 -10.9524 4.5698 -19.9091 -1.9957 5.74 0.0165 Region: Central 1 -0.4591 7.9502 -16.0412 15.1230 0.00 0.9539 Region: North 1 1.3641 5.5100 -9.4352 12.1636 0.06 0.8045 Region:Panhandle 1 4.0172 4.9531 -5.6907 13.7251 0.66 0.4173 Region: South 0 0.0000 0.0000 0.0000 0.0000 Sentinel Rate 1 -10.9524 4.5698 -19.9091 -1.9957 5.74 0.0165 Scale 0 1 0 1 1

PAGE 182

167 Appendix XII: (Continued) Poisson Regression Output for 2001 (continued) Human Incidence, Avian, Mosquito and Sentinel Rates Intercept 1 -11.5302 13.0079 -37.0252 13.9648 0.79 0.3754 Avian Rate 1 -2.1936 77.5583 -154.2050 149.8177 0.00 0.9774 Mosquito Rate 1 19.8565 51.0767 -80.2518 119.9649 0.15 0.6975 Sentinel Rate 1 13.7760 133.2162 -247.3228 274.8750 0.01 0.9176 Scale 0 1 0 1 1 Human Incidence and Avian Rates Intercept 1 -9.9580 2.2934 -14.4529 -5.4630 18.85 < 0.0001 Avian Rate 1 2.7880 12.3590 -21.4352 27.0113 0.05 0.8215 Scale 0 1 0 1 1 Human Incidence and Mosquito Rates Intercept 1 -11.4575 7.0314 -25.2388 2.2388 2.66 0.1032 Mosquito Rate 1 19.1985 46.1163 -71.1878 109.5849 .17 0.6772 Scale 0 1 0 1 1 Human Incidence and Sentinel Rates Intercept 1 -9.6803 1.5899 -12.7965 -6.5641 37.07 < 0.0001 Sentinel Rate 1 10.3426 39.6098 -67.2911 87.9765 0.07 0.7940 Scale 0 1 0 1 1 Poisson Regression Output for 2002 Human Incidence, Avian, Mosquito and Sentinel Rates with Region Parameter DF Estimate StdErr Lower WaldCL Upper WaldCL ChiSq Prob ChiSq Intercept 1 -11.2618 4.4117 -19.9085 -2.6151 6.52 0.0107 Region: Central 1 1.9449 5.4732 -8.7823 12.6722 0.13 0.7223 Region: North 1 2.2267 5.5352 -8.6219 13.0755 0.16 0.6875 Region:Panhandle 1 4.2285 5.9490 -7.4312 15.8883 0.51 0.4772 Region: South 0 0.0000 0.0000 0.0000 0.0000 Avian Rate 1 -0.6241 22.2509 -44.2351 42.9867 0.00 0.9776 Mosquito Rate 1 -5.4626 44.7942 -93.2575 82.3323 0.01 0.9029 Sentinel Rate 1 15.5899 50.7628 -83.9032 115.0830 0.09 0.7588 Scale 0 1 0 1 1 Human Incidence and Avian Rates with Region Intercept 1 -10.9989 3.9928 -18.8246 -3.1732 7.59 0.0059 Region: Central 1 1.1745 4.7427 -8.1210 10.4700 0.06 0.8044 Region: North 1 1.2400 4.5145 -7.6082 10.0884 0.08 0.7836 Region:Panhandle 1 3.6515 4.2325 -4.6440 11.9470 0.74 0.3883 Region: South 0 0.0000 0.0000 0.0000 0.0000 Avian Rate 1 5.5874 8.0656 -10.2207 21.3957 0.48 0.4885 Scale 0 1 0 1 1

PAGE 183

168 Appendix XII: (Continued) Poisson Regression Output for 2002 (continued) Human Incidence and Mosquito Rates with Region Intercept 1 -10.8668 3.9867 -18.6805 -3.0530 7.43 0.0064 Region: Central 1 2.1625 5.1709 -7.9722 12.2973 0.17 0.6758 Region: North 1 2.6208 4.6136 -6.4215 11.6632 0.32 0.5700 Region:Panhandle 1 3.5959 5.3417 -6.8735 14.0654 0.45 0.5008 Region: South 0 0.0000 0.0000 0.0000 0.0000 Mosquito Rate 1 -2.4708 50.0514 -100.5697 95.6281 0.00 0.9606 Scale 0 1 0 1 1 Human Incidence and Sentinel Rates with Region Intercept 1 -10.9294 4.0002 -18.7697 -3.0891 7.46 0.0063 Region: Central 1 1.3624 4.6990 -7.8474 10.5722 0.08 0.7719 Region: North 1 1.6111 4.4629 -7.1360 10.3582 0.13 0.7181 Region:Panhandle 1 4.0616 4.1588 -4.0895 12.2128 0.95 0.3288 Region: South 0 0.0000 0.0000 0.0000 0.0000 Sentinel Rate 1 3.6538 14.0376 -23.8593 31.1671 0.07 0.7946 Scale 0 1 0 1 1 Human Incidence, Avian, Mosquito and Sentinel Rates Intercept 1 -10.1635 2.6733 -15.4032 -4.9239 14.45 0.0001 Avian Rate 1 6.2190 15.6806 -24.5144 36.9525 0.16 0.6917 Mosquito Rate 1 -2.5149 41.4530 -83.7613 78.7314 0.00 0.9516 Sentinel Rate 1 11.0237 44.0186 -75.2509 97.2985 0.06 0.8023 Scale 0 1 0 1 1 Human Incidence and Avian Rates Intercept 1 -9.4846 1.3123 -12.0567 -6.9126 52.24 < 0.0001 Avian Rate 1 8.0642 8.1358 -7.8816 24.0100 0.98 0.3216 Scale 0 1 0 1 1 Human Incidence and Mosquito Rates Intercept 1 -9.3075 1.5203 -12.2871 -6.3279 37.48 < 0.0001 Mosquito Rate 1 6.6712 36.7346 -65.3271 78.6697 0.03 0.8559 Scale 0 1 0 1 1 Human Incidence and Sentinel Rates Intercept 1 -8.9743 1.0904 -11.1115 -6.8372 67.74 < 0.0001 Sentinel Rate 1 4.0253 23.5314 -42.0952 50.1459 0.03 0.8642 Scale 0 1 0 1 1

PAGE 184

169 Appendix XII: (Continued) Poisson Regression Output for 2003 Human Incidence, Avian, Mosquito and Sentinel Rates with Region Parameter DF Estimate StdErr Lower WaldCL Upper WaldCL ChiSq Prob ChiSq Intercept 1 -10.3056 2.2724 -14.7594 -5.8517 20.57 < 0.0001 Region: Central 1 -21.7511 0.00 0.9999 Region: North 1 0.5112 2.8788 -5.1312 6.1536 0.03 0.8591 Region:Panhandle 1 3.6543 2.2918 -0.8375 8.1462 2.54 0.1108 Region: South 0 0.0000 0.0000 0.0000 0.0000 Avian Rate 1 4.5416 2.3229 -0.0110 9.0944 3.82 0.0506 Mosquito Rate 1 -1.7200 22.1394 -45.1125 41.6724 0.01 0.9381 Sentinel Rate 1 6.6392 5.7041 -4.5406 17.8190 1.35 0.2445 Scale 0 1 0 1 1 Human Incidence and Avian Rates with Region Intercept 1 -10.0786 2.0605 -14.1172 -6.0401 23.93 < 0.0001 Region: Central 1 -1.4089 5.1189 -11.4418 8.6239 0.08 0.7831 Region: North 1 0.2240 2.5970 -4.8659 5.3140 0.01 0.9313 Region:Panhandle 1 3.4915 2.1207 -0.6650 7.6481 2.71 0.0997 Region: South 0 0.0000 0.0000 0.0000 0.0000 Avian Rate 1 4.9847 1.9757 1.1124 8.8570 6.37 0.0116 Scale 0 1 0 1 1 Human Incidence and Mosquito Rates with Region Intercept 1 -9.8365 2.0987 -13.9500 -5.7231 21.97 < 0.0001 Region: Central 1 -20.6474 0.00 0.9999 Region: North 1 1.6266 2.6749 -3.6160 6.8693 0.37 0.5431 Region:Panhandle 1 4.8434 2.1096 0.7087 8.9781 5.27 0.0217 Region: South 0 0.0000 0.0000 0.0000 0.0000 Mosquito Rate 1 18.3062 17.1565 -15.3199 51.9325 1.14 0.2860 Scale 0 1 0 1 1 Human Incidence and Sentinel Rates with Region Intercept 1 -9.7131 2.0263 -13.6845 -5.7416 22.98 < 0.0001 Region: Central 1 -0.8502 4.3390 -9.3544 7.6540 0.04 0.8447 Region: North 1 0.7632 2.5849 -4.3030 5.8295 0.09 0.7678 Region:Panhandle 1 4.3712 2.0737 0.3068 8.4356 4.44 0.0350 Region: South 0 0.0000 0.0000 0.0000 0.0000 Sentinel Rate 1 6.1294 4.0030 -1.7164 13.9752 2.34 0.1257 Scale 0 1 0 1 1 Human Incidence, Avian, Mosquito and Sentinel Rates Intercept 1 -8.9078 1.0333 -10.9331 -6.8826 74.32 < 0.0001 Avian Rate 1 6.0103 1.8263 2.4308 9.5899 10.83 0.0010 Mosquito Rate 1 10.4488 7.6957 -4.6343 25.5320 1.84 0.1745 Sentinel Rate 1 -13.0034 15.8515 -44.0717 18.0648 0.67 0.4120 Scale 0 1 0 1 1 Human Incidence and Avian Rates Intercept 1 -8.9099 0.8709 -10.6168 -7.2030 104.67 < 0.0001 Avian Rate 1 6.5038 1.7403 3.0928 9.9147 13.97 0.0002 Scale 0 1 0 1 1

PAGE 185

170 Appendix XII: (Continued) Poisson Regression Output for 2003 (continued) Human Incidence and Mosquito Rates Intercept 1 -7.1707 0.4898 -8.1307 -6.2107 214.34 < 0.0001 Mosquito Rate 1 9.8792 14.9771 -19.4752 39.2337 0.44 0.5095 Scale 0 1 0 1 1 Human Incidence and Sentinel Rates Intercept 1 -7.9687 0.5599 -9.0661 -6.8713 202.56 < 0.0001 Sentinel Rate 1 12.1318 4.8137 2.6971 21.5666 6.35 0.0117 Scale 0 1 0 1 1


xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001498286
003 fts
006 m||||e|||d||||||||
007 cr mnu|||uuuuu
008 041209s2004 flua sbm s000|0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0000578
035
(OCoLC)57724411
9
AJU6891
b SE
040
FHM
c FHM
1 100
Butler, Angela E.
0 245
Evaluation of the current State of Florida West Nile Surveillance Program as a predictor for control and prevention of human West Nile diseases
h [electronic resource] /
by Angela E. Butler.
260
[Tampa, Fla.] :
University of South Florida,
2004.
502
Thesis (M.S.P.H.)--University of South Florida, 2004.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
538
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
500
Title from PDF of title page.
Document formatted into pages; contains 185 pages.
520
ABSTRACT: West Nile is an important novel virus in the United States, having spread rapidly since it was first detected in New York in 1999. The Centers for Disease Control and Prevention as well as many State Health Departments, have mandated programs for surveillance of West Nile Virus activity. These programs incorporate many different aspects including existing arboserology programs with additional testing for West Nile Virus and new plans that incorporate active and passive surveillance methods. The objective of this study was to examine all aspects of the Florida West Nile surveillance program to determine if there was transmission in the animal systems prior to human cases. The predictive analyses were done using regional data graphs, spatial information, correlations and regression models.Data for sentinel chickens, bird necropsy and mosquito pool surveillance from participating counties in Florida were obtained from the State of Florida surveillance database. The human data was obtained from the State of Florida reportable disease database for each county whether participating in the state surveillance programs or not. Clinical cases were examined by demographics (gender and age) and an incidence rate was calculated to demonstrate the effects of disease. Specific statistical methods used included Pearson's coefficient correlation, Poisson distribution regression modeling to show if any of the surveillance systems were predictors for human disease. The incidence rate analysis for clinical cases showed clustering of cases in adjacent counties within a region where Florida's panhandle and adjacent counties northeast had the highest incidence. Florida's central and southern regions had moderate human incidence.This provides useful information in transmission geography for prevention and control measures. Demographic analysis showed that there were twice as many males than females diagnosed with West Nile in Florida, this was true across the groups as well. The highest number of cases was seen within the age group over 55 years of age for West Nile Neuroinvasive Disease and for West Nile Fever the highest number of cases was within the 36-54 age range. The temporal distribution was determined using graphical representations of all of the surveillance types and clinical cases. In order to include all relevant data, the temporality was set from week 20 to week 52. This study found that all of the surveillance types (dead birds, mosquitoes and sentinels) offered a specialized strength for predicting clinical cases. However, mosquitoes proved to be the least efficient out of the three surveillance systems.The regional and spatial analysis showed that positive dead birds and sentinels provided the coverage for the surveillance systems in the state. However, Pearson's correlation coefficient was low for sentinel surveillance; this may be due to higher participation showing West Nile Virus activity in areas (especially rural) that have no reported human cases. This analysis did show that West Nile is detected in mosquito pool samples before it is detected in the dead bird or sentinel surveillance systems which provides an earlier warning for human cases. The Poisson distribution regression model was only useful for the pooled years and 2003. These showed that mosquitoes, positive dead birds and sentinels were good predictors for clinical cases for the combined years and dead birds and sentinels were significant for 2003 as well.The recommendations based on the results from this study would be to continue all the current surveillance efforts but with the following enhancements: 1. Increase the coverage and consistency of submissions for all surveillance types. 2. Set standard levels of participation for all counties based on the regional analyses and populations at risk. 3. Create standardized approaches for sampling, shipping and submitting samples (especially for mosquito pool submissions) and require that participating counties adhere to these standards. 4. Only submit specific birds known to be especially susceptible to West Nile Virus (e.g. corvids). 5. Targeted prevention and education strategies for higher risk groups based on their potential levels of exposure.
590
Co-adviser: Lillian M. Stark.
Co-adviser: Aurora Sanchez-Anguiano.
653
avian.
sentinel.
surveillance.
vector.
West Nile.
arboviruses.
690
Dissertations, Academic
z USF
x Public Health
Masters.
090
RA425 (ONLINE)
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.578