ABSTRACT Title of dissertation: ASSESSING KNOWLEDGE, ATTITUDES, AND BEHAVIORS TOWARD WEST NILE VIRUS PREVENTION AMONG ADULTS ≥ 60 YEARS OLD IN MARYLAND: AN APPLICATION OF THE HEALTH BELIEF MODEL Kimberly Carole Mitchell, Doctor of Philosophy, 2014 Dissertation directed by: Associate Professor Donna E. Howard Department of Behavioral and Community Health West Nile Virus (WNV) is a mosquito-borne virus and the leading cause of arboviral (arthropod-borne) disease in the U.S. While most WNV cases are asymptomatic, 20% of infected people develop WNV fever and < 1% develop severe neurologic disease. Individuals over 50 years old are at greatest risk of severe disease and death. Dramatic increases in WNV activity in 2012 underscored its unpredictable nature and highlighted concerns for adverse effects on older adults. It is important to understand factors that influence this population’s engagement in WNV prevention. This study analyzed data collected by the Maryland Department of Health and Mental Hygiene (DHMH) via cross-sectional survey to identify barriers to WNV prevention among adults ≥60 years of age. Subjects were recruited via stratified random sample of 1,700 households from counties with ≥ two WNV cases, enrolling 211 Maryland adults ≥ 60 years old. Six constructs of the Health Belief Model (HBM)-- perceived susceptibility, perceived severity, perceived benefits, perceived barriers, cues to action, and self-efficacy--were examined to assess how they predicted attitudes and behaviors toward WNV prevention. Univariate, bivariate and multivariate analyses examined the utility of the HBM for explaining WNV preventive behaviors in Maryland adults. Multivariate logistic regression models tested 36 hypotheses examining associations between HBM constructs and six outcomes: (1) avoiding the outdoors at dusk and dawn, (2) dressing in long- sleeved shirts and long pants when outdoors, (3) using insect repellent on exposed skin, (4) draining standing water from objects around one’s property, (5) acceptance of a WNV vaccine, and (6) support for community mosquito control programs. Findings showed high WNV knowledge and awareness but low perceptions of personal risk for WNV infection. Perceived susceptibility to WNV predicted use of insect repellent, draining of standing water from objects around the home, and acceptance of a WNV vaccine; perceived benefits were associated with draining standing water and support for mosquito control programs. Feelings of worry about WNV may inform future WNV interventions and risk communication to older adults. Findings have implications for theory-based research, which could probe applications of the HBM and other theories in understanding WNV attitudes and behaviors. ASSESSING KNOWLEDGE, ATTITUDES, AND BEHAVIORS TOWARD WEST NILE VIRUS PREVENTION AMONG ADULTS ≥ 60 YEARS OLD IN MARYLAND: AN APPLICATION OF THE HEALTH BELIEF MODEL by Kimberly Carole Mitchell Dissertation submitted to the Faculty of the Graduate School of the University of Maryland, College Park in partial fulfillment of the requirements for the degree of Doctor of Philosophy 2014 Advisory Committee Associate Professor Donna E. Howard, Chair Associate Professor Sharon M. Desmond Assistant Professor Mary A. Garza Associate Professor Kerry M. Green Associate Professor Amy R. Sapkota © Copyright by Kimberly Carole Mitchell 2014 ii DEDICATION I first dedicate this accomplishment to God, for it is only by His grace that I have prevailed. This dissertation is also dedicated to my parents, whose unconditional love and support have seen me through this PhD journey and so many challenges before it. Because of you both I am a better person every day. I must also share a special dedication to my mother, who unfailingly invested her time and energy in helping me through the dissertation process. Mom, you have quite literally been with me every step of the way. From lending your editing expertise on my tables and text, to offering a shoulder to lean on in my moments of exhaustion and frustration, you have been my teammate and biggest cheerleader. I share this victory with you, Mom. Thank you. iii ACKNOWLEDGMENTS There are many individuals whom I wish to acknowledge for their vital role in assisting me as I navigated the dissertation process. To Dr. Donna Howard, my advisor, mentor, ally, and committee chair: you have been a tremendous source of inspiration, guidance, comfort, and support. I cannot tell you how much I valued our meetings at Starbucks, our e-mail conversations, and most of all your constant willingness to give of your time and your knowledge to shepherd me along the way. I will always be grateful for your warmth and caring and I look forward to continuing our relationship after graduation. To my dissertation committee members: Drs., Desmond, Green, Garza, and Sapkota, thank you for agreeing to serve on my committee and for not hesitating to share your expertise, encouragement, and guidance whenever they were needed. To Dr. Katherine Feldman and Ms. Patricia Ryan, Maryland Department of Health and Mental Hygiene: thank you for graciously allowing me to use the Emerging Infections Program arboviral disease project for my dissertation research and for sharing your input and expertise along the way. Katherine: a special thanks for consistently granting my leave requests to tend to school obligations, even when it pained you to not have me around handling rabies issues! To Mary Armolt: thank you for being a good friend and colleague and for being willing to share lots of helpful document formatting tips during our after-hours sessions. To all my CZVBD and DHMH friends: thank you for believing in me and cheering me on. You make our workplace an enjoyable and welcome place to be. To my wonderful friends, without whom I could not have survived the journey: My thanks to you all for understanding my sacrifice and for not complaining when I had to put school first. You rallied behind me, pushed me to move forward, and buoyed me with love and laughter. I am grateful to you for having faith in me and lifting my spirits when I felt inadequate to the task. To my Douglas Memorial church family: thank you for your words of encouragement and most of all your prayers. I have relied often on the comfort of your kindness and fellowship and feel blessed to be part of such a loving community. A special thanks to my fellow dance ministry members: I have missed you all more than words can say, but I will rejoin you soon! iv Last, but by no means least, a heartfelt thank you to my family, both immediate and extended, but especially to Mom, Dad, and Damani. Your love sustains me, especially in the hardest of times. I thank God for you and I hope I have made you proud. v Table of Contents Table of Contents ............................................................................................................................ v LIST OF FIGURES AND TABLES............................................................................................ viii CHAPTER 1: INTRODUCTION ................................................................................................... 1 1.1 Statement of Problem ........................................................................................... 1 1.1.1 History........................................................................................................... 1 1.1.2 Epidemiology and Virology .......................................................................... 2 1.1.3 West Nile Virus in Maryland ........................................................................ 3 1.2 West Nile Virus Prevention Measures ................................................................. 4 1.3 Theory Applied to WNV Prevention Behaviors .................................................. 5 1.3.1 Existing Studies Applying HBM to WNV Prevention ................................. 6 1.4 Rationale for Study............................................................................................... 8 1.4.1 Background and Pilot Study ......................................................................... 8 1.4.2 Purpose and Rationale................................................................................... 9 1.5 Research Question and Hypotheses ................................................................... 12 1.6 Summary ............................................................................................................ 15 1.7 Definition of Terms ............................................................................................ 16 CHAPTER 2: LITERATURE REVIEW ..................................................................................... 17 2.1 Introduction and Purpose ................................................................................... 17 2.2 West Nile Virus Overview ................................................................................. 18 2.2.1 History and Geographic Distribution .......................................................... 18 2.2.2 Epidemiology and Emergence in the U.S. and Maryland ........................... 19 2.2.3 West Nile Virus: Disease Classification ..................................................... 21 2.2.4 Transmission ............................................................................................... 21 2.2.5 Clinical Presentation ................................................................................... 22 2.2.6 Populations at Risk for WNV ..................................................................... 23 2.2.7 Prevention Behaviors .................................................................................. 25 2.3 Health Belief Model ........................................................................................... 26 2.3.1 Definition and Origins of Model................................................................. 26 2.3.2 Health Belief Model Structure .................................................................... 27 vi 2.3.3 Measurement of HBM Constructs .............................................................. 31 2.3.4 Strengths and Weaknesses of HBM ............................................................ 32 2.3.5 Applications of the Health Belief Model .................................................... 37 2.3.6 Health Belief Model and Communicable Diseases .................................... 37 2.4 Health Belief Model and West Nile Virus Prevention ....................................... 37 2.5 Central Themes in West Nile Virus Prevention Behavior Literature................. 39 2.5.1 Strengths and Weaknesses of Existing Studies of WNV Protective Behavior 54 2.6 Conclusions ........................................................................................................ 55 CHAPTER 3: METHODOLOGY ................................................................................................ 57 3.1 Introduction ........................................................................................................ 57 3.2 Population Description ....................................................................................... 57 3.3 Sampling Procedures .......................................................................................... 65 3.4 Theoretical Framework ...................................................................................... 69 3.5 Study Design ...................................................................................................... 71 3.6 Data Collection ................................................................................................... 79 3.7 Variables and Measurement ............................................................................... 82 3.7.1 Independent Variables: Health Belief Model Constructs ........................... 83 3.7.2 Modifying Factors ....................................................................................... 90 3.7.3 Control Variables ........................................................................................ 90 3.7.4 Other Demographic Factors ........................................................................ 91 3.7.5 Knowledge .................................................................................................. 93 3.7.6 Dependent Variables ................................................................................... 94 3.8 Data Analysis Plan ............................................................................................. 97 3.9 Testing of Hypotheses ........................................................................................ 99 3.10 Data Management ............................................................................................ 103 CHAPTER 4: RESULTS ........................................................................................................... 105 4.1 Introduction ...................................................................................................... 105 4.2 Description of Sample ...................................................................................... 105 4.2.1 Demographic Characteristics .................................................................... 105 4.2.2 Behavioral Outcomes ................................................................................ 108 vii 4.2.3 Health Belief Model Constructs................................................................ 109 4.2.4 Knowledge ................................................................................................ 113 4.2.5 Summary ................................................................................................... 114 4.3 Bivariate Analysis ............................................................................................ 114 4.4 Multivariate Analysis ....................................................................................... 126 4.4.1 Summary of Bivariate Analysis Results ................................................... 126 4.4.2 Selection of Control Variables for Multivariate Analysis ........................ 127 4.4.3 Building Logistic Regression Models ....................................................... 127 4.4.4 Results of Multivariate Logistic Regression Analysis .............................. 129 4.4.5 Summary of Findings ................................................................................ 142 CHAPTER 5: DISCUSSION ...................................................................................................... 147 5.1 Overview .......................................................................................................... 147 5.2 Summary of Study and Key Findings .............................................................. 147 5.2.1 Key Findings ............................................................................................. 148 5.3 Hypothesis Testing Results .............................................................................. 150 5.4 Influence of Demographic Characteristics and Knowledge ............................. 163 5.5 Qualitative Analysis of Risk Perceptions from DHMH Study ........................ 164 5.6 Implications of Findings................................................................................... 167 5.7 Implications for Public Health Practice ........................................................... 170 5.8 Implications for Theory .................................................................................... 171 5.9 Study Limitations ............................................................................................. 176 5.10 Future Research Directions .............................................................................. 178 5.11 Summary and Conclusions ............................................................................... 180 APPENDIX A: INTRODUCTORY LETTER ........................................................................... 181 APPENDIX B: RESPONSE POSTCARD ................................................................................ 182 APPENDIX C: WEST NILE VIRUS KNOWLEDGE, ATTITUDES, AND BEHAVIOR QUESTIONNAIRE .................................................................................................................... 183 APPENDIX D: PROJECT INFORMATION SHEET ............................................................... 198 APPENDIX E: UMCP IRB APPROVAL LETTER .................................................................. 200 REFERENCES ........................................................................................................................... 202 viii LIST OF FIGURES AND TABLES Figure 1. Average annual incidence of West Nile Virus neuroinvasive disease reported to CDC by age group, 1999-2013 ..................................................................................................... 12 Figure 2. Health Belief Model ...................................................................................................... 32 Figure 3. Diagram of Sampling and Data Collection Procedures ................................................. 65 Figure 4. Conceptual Model of Health Belief Model Applied to WNV Prevention .................... 70 Table 1. Summary of Studies on West Nile Virus Knowledge, Attitudes, and Behavior ............ 40 Table 2. Summary of Studies Using the HBM to Examine WNV Personal Protective Behaviors ...................................................................................................................................... 53 Table 3. Population of Maryland Residents ≥ 60 Years Old by Zip Code and Age Category ........................................................................................................................................ 59 Table 4. Research Question and Hypotheses ................................................................................ 60 Table 5. Summary of Variables, Operational Definitions, and Survey Questions ....................... 74 Table 6. Comparison of Study Participants and Non-participants ................................................ 82 Table 7. Operationalization of Hypotheses ................................................................................... 89 Table 8. Demographic Characteristics ....................................................................................... 106 Table 9. Personal Protective Behavior Outcomes...................................................................... 108 Table 10. Attitudes Toward WNV Prevention Programs and Services ..................................... 109 Table 11. Health Belief Model Constructs ................................................................................ 112 Table 12. Knowledge of West Nile Virus Risk and Transmission ............................................ 113 Table 13. Spearman Rank Correlations ...................................................................................... 115 Table 14. Demographic Characteristics, Knowledge, and Avoiding the Outdoors at Dusk and Dawn .................................................................................................................................... 116 Table 15. HBM Constructs and Avoiding the Outdoors at Dusk and Dawn .............................. 117 Table 16. Demographic Characteristics, Knowledge, and Dress in Long Sleeved Shirts and Pants ..................................................................................................................................... 118 Table 17. HBM Constructs and Dress in Long Sleeved Shirts and Pants .................................. 118 Table 18. Demographic Characteristics, Knowledge, and Use of Insect Repellent in the Last 90 Days ............................................................................................................................... 119 Table 19. HBM Constructs and Use of Insect Repellent in the Last 90 Days ............................ 120 Table 20. Demographic Characteristics, Knowledge, and Drain Standing Water ..................... 121 Table 21. HBM Constructs and Drain Standing Water .............................................................. 122 Table 22. Demographic Characteristics, Knowledge, and Willingness to Accept a WNV Vaccine ....................................................................................................................................... 123 Table 23. HBM Constructs and Willingness to Accept a WNV Vaccine .................................. 123 Table 24. Demographic Characteristics, WNV Knowledge, and Support for Mosquito Control Program.......................................................................................................................... 124 Table 25. HBM Constructs and Support for Mosquito Control Program ................................... 125 ix Table 26. Logistic Regression Models for HBM Constructs and Avoidance of the Outdoors (Hypotheses 1, 7, 13, 19, 25, and 31)*........................................................................ 130 Table 27. Logistic Regression Models for HBM Constructs and Dressing in Long- Sleeved Shirts and Long Pants (Hypotheses 2, 8, 14, 20, 26, and 32)*...................................... 132 Table 28. Logistic Regression Models for HBM Constructs and Use of Insect Repellent in the Last 90 Days (Hypotheses 3, 9, 15, 21, 27, 33)* .............................................................. 134 Table 29. Logistic Regression Models for HBM Constructs and Draining Standing Water (Hypotheses 4, 10, 16, 22, 28, and 34)* ..................................................................................... 137 Table 30. Logistic Regression Models for HBM Constructs and Willingness to Accept a WNV Vaccine (Hypotheses 5, 11, 17, 23, 29, 35)*.................................................................... 138 Table 31. Logistic Regression Models for HBM Constructs and Support for Community Mosquito Control Programs (Hypotheses 6, 12, 18, 24, 30, 36)* .............................................. 141 Table 32. Summary of Hypothesis Test Results, Arranged by HBM Construct ....................... 143 1 CHAPTER 1: INTRODUCTION 1.1 Statement of Problem West Nile Virus (WNV) is an arboviral (arthropod-borne) virus transmitted to humans by infected mosquitoes. It is currently the leading cause of domestically- acquired arboviral disease in the U.S., and is among the leading sources of severe neuroinvasive disease (Lindsey, Lehman, Staples, & Fischer, 2014). As such, WNV represents a significant public health problem nationwide. Due to the lack of an available human WNV vaccine, prevention of WNV disease relies largely on individual and community-level practice of personal protective behaviors (Lindsey et al., 2014). In the face of declining federal support for vector-borne disease surveillance and control efforts, a need exists to increase awareness of prevention measures and identify new strategies to reach out to individuals most at risk of severe WNV disease by promoting those behavioral measures. This study examined the utility of the Health Belief Model at predicting knowledge, attitudes, and behaviors of Maryland adults at high risk of severe neurologic disease from WNV. 1.1.1 History West Nile Virus is a mosquito-borne virus that was first isolated from a female patient in Uganda in 1937, and has historically been found in Africa, Asia, and the Middle East (Smithburn et al., 1940; Hayes et al., 2005). In the last decade, WNV has established a significant presence in the United States and is currently the leading cause of arboviral (arthropod-borne) disease in the U.S. (Murray et al., 2010). Since it was first documented in North America in 1999, significant human WNV epidemics were 2 identified in 2002 and 2003 as the virus spread westward across the U.S. (O’Leary et al., 2004). Since then, WNV activity in humans, birds, mosquitoes, and veterinary species has been documented in every state except Hawaii, Alaska, and Oregon (CDC Q&A, 2012). 1.1.2 Epidemiology and Virology West Nile virus is a Flavivirus, belonging to the same family of viruses as Yellow Fever, dengue hemorrhagic fever, and Japanese encephalitis (Hayes et al., 2005). It circulates through a primary transmission cycle between a mosquito vector and an avian reservoir, with horses, humans, and other vertebrates as incidental hosts (Murray et al., 2010). Although rare, a small number of human WNV cases transmitted via organ donation, blood transfusion, and intrauterine transmission have been reported (Hayes et al., 2005). While most WNV cases are asymptomatic, approximately one in every 150 infected people develop severe neurologic disease and 20% of all infected people develop WNV fever (Hayes et al., 2005). In the U.S., case fatality rates for WNV range from 3- 15% and are highest among individuals over 50 years of age (CDC COCA 2012). Individuals in this age group who become infected with WNV are at greatest risk of developing severe neuroinvasive disease, in the form of encephalitis, aseptic meningitis, or poliomyelitis (CDC COCA 2012). Thus, they are considered a high risk group for WNV. Following the early WNV epidemics from 1999-2003, WNV activity throughout the US declined considerably. By the end of 2011, the national incidence of WNV neuroinvasive disease was 0.16 per 100,000 population, down from approximately 1.0 3 per 100,000 in 2003 (CDC COCA, 2012). However, in the summer of 2012, the disease made a dramatic resurgence in several states. Of the nearly 4,000 human WNV cases reported nationwide by the end of the first week of October, nearly 80% of those cases were reported from ten states: Texas, Louisiana, California, Illinois, Michigan, Mississippi, Nebraska, Colorado, Oklahoma, and South Dakota; and, over one-third of all cases documented nationwide were reported from Texas (CDC, 2012). In 2012, national health officials documented the largest WNV epidemic since the first epidemic in 2002, with 5,387 human WNV cases and 243 fatalities reported nationwide by December 11, 2012, the highest number of cases reported by that week since 2003 (CDC, 2012). The increased WNV activity in 2012 illustrates the unpredictable and transient nature of the virus, and its often deleterious health effects on older adults are of grave public health concern (Bitto et al., 2005; Petersen et al. 2012). 1.1.3 West Nile Virus in Maryland Although WNV activity in Maryland in 2012 did not increase in the same dramatic fashion as occurred in other states, the state did report its largest number of WNV human cases since the peak year of 2003, in which 73 human cases were reported statewide (Maryland Arbovirus Surveillance Results, 2012). By the end of 2012, 47 human WNV cases had been reported in Maryland, including four fatalities. Maryland health officials continue to monitor the disease in human and animal populations through rigorous surveillance efforts including adherence to standard case definitions and diagnostic procedures and thorough laboratory testing, but health officials have been hampered by reductions in federal funding for arboviral disease surveillance, which limits 4 their capacity to maintain enhanced passive human surveillance and to sustain long-term mosquito surveillance. The unexpected resurgence in West Nile disease activity has returned WNV to the forefront of public interest, prompting even greater scrutiny by state and local health officials, and bringing with it increased media attention and greater public concern about the risk the disease poses to Maryland residents (CDC Q&A, 2012). This heightened awareness highlights a need to better understand barriers to WNV prevention among Maryland residents, particularly among those at highest risk of severe disease. 1.2 West Nile Virus Prevention Measures Public health professionals have identified specific health behaviors that aid in WNV prevention. These are referred to as personal protective behaviors (PPBs) and include using insect repellent (containing an EPA-registered ingredient such as N,N- diethyl-meta-toluamide or DEET) on exposed skin when outside, dressing in long- sleeved shirts and long pants when outdoors, avoiding outdoor activities during peak mosquito feeding hours (dusk and dawn), repairing damaged window screens, and draining or emptying areas of standing water (which can be potential mosquito breeding sites) around personal property (Zielinski-Gutierrez et al., 2003; CDC 2008). Although community level mosquito control programs that use larvicides and/or adulticides to kill juvenile and adult mosquitoes are also an important part of WNV prevention, individual WNV prevention behaviors remain vital to combating WNV infection, particularly among high-risk individuals. Individuals’ knowledge, perceptions, and behaviors with respect to WNV are often measured via survey. Knowledge of WNV is typically measured by asking 5 respondents about the basics of WNV: how it is transmitted, which groups are at risk, and what prevention measures are available. Beliefs or perceptions are commonly measured by asking about perceived risk of getting infected with WNV, awareness of others who have gotten it, and opinions about recommended prevention steps. Behaviors are measured by asking about actions respondents took during a given time period to protect themselves and to avoid mosquito bites. PPBs are generally measured via a series of questionnaire items designed to capture the frequency with which people practice them (or the reasons why they don’t) as well as their perceptions and opinions about the quality and effectiveness of such behaviors (Aquino et al., 2004; Loeb et al., 2005). 1.3 Theory Applied to WNV Prevention Behaviors A small body of public health research has examined behavioral barriers to WNV prevention by evaluating PPB measures and some have incorporated a theoretical framework as the basis for their evaluation of WNV preventive behavior (Aquino et al., 2004; Bitto et al., 2005; Gujral et al., 2007). The theory that has been most commonly applied to the examination of health behaviors for WNV and other mosquito-borne disease prevention is the Health Belief Model (HBM). The HBM, a value-expectancy theory developed by federal public health officials in the 1950s, is a widely recognized theoretical model which posits that an individual will take steps to prevent, screen for, or control disease according to six constructs: perceived susceptibility, perceived severity, perceived benefits, perceived barriers, cues to action, and self-efficacy (Glanz et al., 2008). With respect to WNV, perceived susceptibility refers to the belief that one is vulnerable to getting WNV infection as well as knowledge of the impact of WNV disease, perceived severity indicates the extent to which a person believes WNV to be a 6 serious illness that can cause morbidity and death, cues to action refer to specific triggers that prompt a person to engage in PPBs, and self-efficacy is the belief that one is capable of avoiding WNV infection through enactment of PPBs. 1.3.1 Existing Studies Applying HBM to WNV Prevention In recent years, some researchers have applied the HBM to the study of behaviors concerning WNV and other mosquito-borne diseases (Aquino et al., 2004; Bitto et al., 2005; Herrington et al., 2003). These researchers have used the Health Belief Model as a theoretical framework to guide the development of survey instruments designed to capture WNV prevention behaviors. Prior to the first national WNV outbreak in 1999, Herrington and colleagues conducted a nationally-based cross-sectional survey of 1,500 US adults plus an oversample of an additional 250 adults in six states to establish a baseline of their knowledge, attitudes, and behaviors toward WNV and other mosquito-borne diseases prior to that first outbreak. Their instrument measured respondents’ knowledge of mosquitoes and arboviral encephalitis, perceptions of the severity of mosquito-borne disease, perceptions of their susceptibility to such illnesses, and beliefs about the effectiveness of recommended actions/behaviors to prevent mosquito-borne infections (Herrington et al., 2003). It was implemented via a computer-assisted telephone interview system (Herrington et al., 2003). Results revealed that being concerned about mosquito bites, perceiving insect repellents as effective and not harmful to health, and preferring to remain indoors during the late afternoon and early evening were all significant predictors of behaviors to prevent mosquito bites. 7 Similarly, in 2003, Aquino et al. conducted a survey of WNV PPBs among adult residents of British Columbia using the Health Belief Model as the theoretical framework. The study instrument measured the frequency of reported PPBs as well as selected constructs from the HBM, including perceived susceptibility, perceived severity, barriers and benefits to action, and cues to action. The authors found that the major barriers to practicing WNV PPBs were perception of DEET as an environmental and health hazard, time needed to drain standing water, and desire to engage in outdoor leisure activities during peak mosquito feeding times (Aquino et al., 2004). Finally, Bitto and colleagues used the HBM to design and implement a tailored health education program among a sample of adults 50 years old and over in Monroe County, PA (Bitto et al.). This included administration of a cross-sectional survey instrument adapted from another state’s health department. Survey items were framed around all six HBM constructs. Perceived susceptibility and perceived severity were examined collectively as perceived threat, perceived benefits referred to positive outcomes associated with taking steps to prevent WNV, and perceived barriers were identified as factors or obstacles that hinder actions to prevent WNV infection. Responses revealed that selected HBM constructs were highly relevant to this group, particularly perceived severity and perceived susceptibility to WNV infection (Bitto et al., 2005). The above behavioral survey studies reveal the usefulness of incorporating the HBM into studies of WNV prevention behaviors. These and other studies of WNV prevention behavior are described in detail in the Literature Review. 8 1.4 Rationale for Study 1.4.1 Background and Pilot Study Overall, little is known about the knowledge, attitudes, and behaviors of high-risk adults towards WNV prevention. Accordingly in 2012, the Maryland Department of Health and Mental Hygiene (DHMH) initiated a cross-sectional study with the aim of identifying barriers to WNV prevention among adults ≥ 60 years old. While national recommendations for WNV prevention have denoted adults at least 50 years of age as at highest risk for WNV disease, for purposes of the DHMH study, the focus was on adults 60 years old and older in an effort to target those individuals who may encounter more barriers (physical, mental, and financial) when engaging in WNV prevention behaviors and to reach an accessible sample. People aged 60 years and over are more likely to have retired from work and to spend time at home or outdoors engaged in leisure-time activities that might bring them in contact with mosquitoes. Staff at the Maryland Department of Health and Mental Hygiene, led by the Chief of Rabies and Vector-borne Diseases, who served as the primary investigator (KCM), coordinated the DHMH study including its conceptualization, design, and development from January to July 2012. The primary investigator of the DHMH Barriers study developed the original survey instrument, which included sections on Knowledge, Attitudes toward Prevention, and Demographic Factors, and worked with other DHMH epidemiologists to refine, pilot test, and administer it from August through December 9 2012. The implementation process included data collection and data management, along with training of interviewers and coding of questionnaire items. The study, entitled: “Assessment of Barriers to West Nile Virus Prevention among Adults at Least 60 Years Old in Maryland”, (the Maryland WNV Barriers Study) used a proportionate random sample of 1700 households stratified by zip code from counties with two or more previous WNV cases. From this sampling frame, 211 individuals meeting eligibility criteria were enrolled. Eligibility criteria included age 60 years of age or older, English-speaking, and living independently (not in a long term care facility, nursing home, or other institution). Telephone interviews were conducted with all enrollees by trained interviewers. The survey instrument incorporated elements of existing surveys of WNV prevention behavior and used the HBM as a theoretical framework. It asked questions about the respondents’ knowledge and awareness of WNV, use of PPBs, and perceptions of the effectiveness of PPBs, insect repellents, and other products (e.g. WNV vaccine). The purpose of the study was to identify barriers to WNV prevention among Maryland residents 60 years old and older given their elevated risk for severe WNV disease. Survey administration concluded in December 2012. 1.4.2 Purpose and Rationale The purpose of this dissertation study was to evaluate the utility of the Health Belief Model for predicting WNV knowledge, attitudes, and practices among adults 60 years of age and over in Maryland, using a proportionate random sample of adults from the Maryland WNV Barriers Study. This study involved primary analysis of a subset of data from the larger DHMH WNV Barriers study, and examined six HBM constructs: perceived susceptibility, perceived severity, perceived benefits, perceived barriers, cues 10 to action, and self-efficacy to determine their individual effectiveness at predicting participants’ perceptions and behavior. Although the target sample size was 256 participants, 211 were ultimately enrolled, which was believed to yield sufficient statistical power for testing of study hypotheses. The rationale for undertaking this dissertation study was three-fold. First, WNV made a dramatic resurgence in 2012, and it has been described as “the adolescent that never grows up,” due to its sporadic occurrence and dramatically unpredictable nature; moreover, such fluctuations in WNV incidence are expected to continue in the future (Petersen et al., 2012). Because WNV activity is shaped by a multitude of ecologic factors, including climate, vector biology, host migration, and human behavior patterns, it is impossible to fully predict the timing or intensity of future occurrences (Petersen et al., 2012). Given this uncertainty, and given the absence of an available human WNV vaccine, personal prevention measures are even more important to prevent human WNV disease during these epidemics. As there is limited availability of theory-based studies examining such WNV prevention behavior in the literature, a study demonstrating the utility of the Health Belief Model to explain or predict WNV prevention behavior among high-risk Maryland residents could be of considerable value. Second, adults 60 years of age and over have been identified as among those at highest risk for developing severe WNV disease in the form of encephalitis, aseptic meningitis, meningoencephalitis, or WNV poliomyelitis, a form of acute flaccid paralysis (CDC COCA, 2012). Given that this group is most vulnerable to severe illness and death from WNV, and that there is a rapidly growing senior population in the US and in Maryland, a pressing need exists to understand the beliefs and motivations that hamper or 11 encourage this population to engage in WNV preventive behaviors. Figure 1 below provides a breakdown of the average annual incidence of WNV neuroinvasive disease by age group. It clearly illustrates that the majority of WNV cases involving severe neurologic illness occur among adults ≥ 50 years of age (CDC). Third, previously developed theory-based instruments for measuring WNV prevention behaviors have not been validated for adults in this high-risk group in Maryland. A dissertation by Yerby (2007) remains the only validation study of WNV beliefs, attitudes, and behaviors, and it studied a sample of women aged 19 and older living in West Alabama (Yerby, 2007). It would be useful to conduct an assessment of a theoretical model for predicting WNV protective behaviors in adults aged 60 years old and older in Maryland. Use of a theory-based instrument is relevant because it provides a conceptual or a guiding framework for understanding, explaining and/or predicting health protective behaviors. Given the ambiguities and challenges associated with determining relationships between HBM constructs, particularly for cross-sectional designs, this study did not attempt to capture such mediating relationships (Glanz, 2008; Carpenter, 2010). This dissertation may further expand the field of vector-borne disease research by providing additional support for a widely-recognized theoretical model to effectively predict WNV prevention behaviors in populations at risk. Findings from this research can be used in the development of targeted interventions and media campaigns for prevention of WNV exposure among high risk Maryland residents. 12 Source: ArboNET, Arboviral Diseases Branch, Centers for Disease Control and Prevention Figure 1. Average annual incidence of West Nile Virus neuroinvasive disease reported to CDC by age group, 1999-2013 1.5 Research Question and Hypotheses Research Question: Is the Health Belief Model a useful theoretical framework for predicting perceptions and behavior toward West Nile virus prevention among Maryland adults 60 years of age and older? The following 36 hypotheses were used to test the six HBM constructs against each of the following six outcomes: (1) avoidance of the outdoors at prime mosquito feeding hours of dusk and dawn, (2) dressing in long shirts and pants when outdoors, (3) use of insect repellent in the previous 90 days, (4) draining standing water from objects around the property, (5) willingness to accept a WNV vaccine if one were available, and (6) support for community mosquito control programs. Hypotheses 1-6: Perceived susceptibility to WNV disease 1. Individuals with high perceived susceptibility to WNV infection are more likely to avoid going outdoors during mosquito feeding hours of dusk and dawn than those with low perceived susceptibility. 13 2. Individuals with high perceived susceptibility to WNV infection are more likely to dress in long-sleeved shirts and long pants to avoid mosquito bites than those with low perceived susceptibility. 3. Individuals with high perceived susceptibility to WNV infection are more likely to have used insect repellent in the last 90 days than those with low perceived susceptibility. 4. Individuals with high perceived susceptibility to WNV infection are more likely to drain standing water from objects around their property than those with low perceived susceptibility. 5. Individuals with high perceived susceptibility to WNV infection are more likely to accept a human WNV vaccine if one were available than those with low perceived susceptibility. 6. Individuals with high perceived susceptibility to WNV infection are more likely to support mosquito control programs in their community than those with low perceived susceptibility. Hypotheses 7-12: Perceived severity of WNV disease 7. Individuals who perceive WNV as causing serious illness are more likely to avoid going outdoors during mosquito feeding hours of dusk and dawn. 8. Individuals who perceive WNV as causing serious illness are more likely to dress in long-sleeved shirts and long pants to avoid mosquito bites. 9. Individuals who perceive WNV as causing serious illness are more likely to have used insect repellent in the last 90 days. 10. Individuals who perceive WNV as causing serious illness are more likely to drain standing water from objects around their property that collect water. 11. Individuals who perceive WNV as causing serious illness are more likely to accept a human WNV vaccine if one were available. 12. Individuals who perceive WNV as causing serious illness are more likely to support mosquito control programs in their community. Hypotheses 13-18: Perceived benefits of practicing WNV protective behaviors 13. Individuals who perceive benefits in practicing WNV protective behaviors are more likely to avoid going outdoors during mosquito feeding hours of dusk and dawn. 14. Individuals who perceive benefits in practicing WNV protective behaviors are more likely to dress in long-sleeved shirts and long pants to avoid mosquito bites. 15. Individuals who perceive benefits in practicing WNV protective behaviors are more likely to have used insect repellent in the last 90 days. 16. Individuals who perceive benefits in practicing WNV protective behaviors are more likely to drain standing water from objects around their property that collect water. 17. Individuals who perceive benefits in practicing WNV protective behaviors are more likely to accept a human WNV vaccine if one were available. 18. Individuals who perceive benefits in practicing WNV protective behaviors are more likely to support mosquito control programs in their community. 14 Hypotheses 19-24: Perceived barriers to practicing WNV protective behaviors 19. Individuals with high perceived barriers to practicing WNV protective behaviors are less likely to avoid going outdoors during mosquito feeding hours of dusk and dawn than those with low perceived barriers. 20. Individuals with high perceived barriers to practicing WNV protective behaviors are less likely to dress in long-sleeved shirts and long pants to avoid mosquito bites than those with low perceived barriers. 21. Individuals with high perceived barriers to practicing WNV protective behaviors are less likely to have used insect repellent in the last 90 days than those with low perceived barriers. 22. Individuals with high perceived barriers to practicing WNV protective behaviors are less likely to drain standing water from objects around their property than those with low perceived barriers. 23. Individuals with high perceived barriers to practicing WNV protective behaviors are less likely to accept a human WNV vaccine if one were available than those with low perceived barriers. 24. Individuals with high perceived barriers to practicing WNV protective behaviors are less likely to support mosquito control programs in their community than those with low perceived barriers. Hypotheses 25-30: Cues to action for preventing WNV infection 25. Individuals who receive cues to action for preventing WNV infection are more likely to avoid going outdoors during mosquito feeding hours of dusk and dawn than those who do not receive cues to action. 26. Individuals who receive cues to action for preventing WNV infection are more likely to dress in long-sleeved shirts and long pants than those who do not receive cues to action. 27. Individuals who receive cues to action for preventing WNV infection are more likely to have used insect repellent in the last 90 days than those who do not receive cues to action. 28. Individuals who receive cues to action for preventing WNV infection are more likely to drain standing water from objects around their property than those who do not receive cues to action. 29. Individuals who receive cues to action for preventing WNV infection are more likely to accept a human WNV vaccine if one were available than those who do not receive cues to action. 30. Individuals who receive cues to action for preventing WNV infection are more likely to support mosquito control programs in their community than those who do not receive cues to action. 15 Hypotheses 31-36: Self-efficacy for preventing WNV infection 31. Individuals with high self-efficacy for preventing WNV infection are more likely to avoid going outdoors during mosquito feeding hours of dusk and dawn than those with low self-efficacy. 32. Individuals with high self-efficacy for preventing WNV infection are more likely to dress in long-sleeved shirts and long pants to avoid mosquito bites than those with low self-efficacy. 33. Individuals with high self-efficacy for preventing WNV infection are more likely to have used insect repellent in the last 90 days than those with low self-efficacy. 34. Individuals with high self-efficacy for preventing WNV infection are more likely to drain standing water from objects around their property than those with low self-efficacy. 35. Individuals with high self-efficacy for preventing WNV infection are more likely to accept a human WNV vaccine if one were available than those with low self- efficacy. 36. Individuals with high self-efficacy for preventing WNV infection\are more likely to support mosquito control programs in their community than those with low self-efficacy. 1.6 Summary Chapter 1 provided an overview of West Nile virus, including its origins, pathogenesis, and epidemiology in the United States and Maryland. It outlined the purpose and rationale for this study, which examined the Health Belief Model as a theoretical framework for explaining WNV prevention behaviors in a sample of high-risk adults ages 60 and over in Maryland. By conducting a primary analysis of data collected during a cross-sectional study by the Maryland Department of Health and Mental Hygiene, this study examined the application of six constructs of the HBM for predicting WNV prevention behavior among adults 60 years of age and older. This study has implications for the development of future behavioral interventions to reduce the WNV disease burden in this population. 16 1.7 Definition of Terms Acute Flaccid Paralysis: Any condition caused by the weakening or loss of muscle tone Arbovirus: Arthropod-borne virus; any of a group of viruses transmitted by mosquitoes, ticks, and other arthropods Encephalitis: An inflammation of the brain Flavivirus: Any virus belonging to the family Flaviridae, which includes Yellow Fever, Dengue hemorrhagic fever, and St. Louis encephalitis Meningitis: Inflammation of the membranes surrounding the brain and spinal cord Meningoencephalitis: Inflammation of both the brain and the membranes surrounding it Neuroinvasive Disease: Any illness affecting the central nervous system Non-neuroinvasive Disease: Illness that causes physical symptoms but does not involve the nervous system Poliomyelitis: Inflammation of the spinal cord; typically characterized by paralysis Vector: an intermediate organism in indirect transmission that carries the agent from a reservoir (source) to a susceptible host West Nile Virus: A virus belonging to the same group that includes Japanese encephalitis and St. Louis encephalitis and causes clinical illness affecting the nervous system 17 CHAPTER 2: LITERATURE REVIEW 2.1 Introduction and Purpose The purpose of this study was to examine the utility of the Health Belief Model for predicting WNV knowledge, attitudes, and behaviors among adults 60 years of age and older in Maryland. This review of the literature includes an overview of West Nile virus, including its history, epidemiology, prevention, surveillance, and public health impact in the United States and Maryland. This is followed by a description of the history, constructs, and applications of the Health Belief Model (HBM). It examines key themes in the existing literature on the HBM as it applies to WNV prevention behavior. The primary objective of the review is to provide a better understanding of the application of constructs from the Health Belief Model to West Nile virus prevention behaviors. Examples are offered of the salient, but limited, literature on WNV protective behaviors, with emphasis on those studies that used the HBM as a theoretical framework. Emphasis is placed on the anticipated contribution that this study will make to the fields of both infectious disease epidemiology and health promotion. Methods for Literature Search The following databases were searched during this review: Academic Search Premier, Medline, PubMed, CINAHL, ERIC, PsychInfo, and Social Science Abstracts. Key search terms included combinations of the following: West Nile virus, mosquito, vector-borne disease, knowledge, perceptions, beliefs, attitudes, behavior, and Health Belief Model. Inclusion criteria were: (1) English language, (2) research based in North America (US or Canada), and (3) measurement of perceptions, attitudes, and/or behaviors toward West Nile Virus and/or other mosquito-borne diseases. Both quantitative and 18 qualitative studies were included, and a variety of study designs, including cross- sectional surveys, focus groups, and randomized controlled trials. Of the 20 articles identified that examined knowledge, attitudes, and/or practices related to WNV prevention, only seven applied a theoretical model or framework, which in each instance was the HBM. The extant literature was reviewed to compare approaches to examination of WNV attitudes and behavior and to examine overall assessment of WNV knowledge, perceptions, and protective behavior. 2.2 West Nile Virus Overview 2.2.1 History and Geographic Distribution West Nile virus (WNV) is a mosquito-borne virus that was first isolated from a febrile female patient in the West Nile province of Uganda in 1937 (Murray et al., 2011). As such, it is one of the earliest recognized arboviruses (arthropod-borne viruses) and belongs to the genus Flavivirus of the family Flaviridae (Murray et al., 2011). The virus is maintained in a transmission cycle between mosquito vectors and bird reservoir hosts with humans, horses, and other mammals as incidental (dead-end) hosts. Originally documented in the Old World throughout Africa and Eurasia (including India, the Middle East, and various regions in Europe, Asia, and Russia) the virus rapidly spread to the Americas, eventually emerging in the Western Hemisphere (Artsob et al., 2009). Following its initial emergence in North America in 1999, WNV spread rapidly westward from its origin in New York throughout the other East Coast states including Maryland and subsequently across the U.S. as well as northward into Canada (Artsob et al., 2009). 19 2.2.2 Epidemiology and Emergence in the U.S. and Maryland 2.2.2.1 WNV in United States WNV was first introduced in the Western hemisphere in 1999 and in 2002 caused the largest epidemic of human neurological disease in the U.S., with over 4100 human cases (both neuroinvasive and non-neuroinvasive disease) reported in 39 states and the District of Columbia (Zielinski-Gutierriez, 2003). Federal health officials documented the peak incidence of WNV neuroinvasive disease in the years 2002-2003, at 1.02 per 100,000 U.S. population (DeBiasi, 2011). Following several subsequent years of more moderate but ongoing WNV activity nationwide, the Centers for Disease Control and Prevention (CDC) announced in summer 2012 that increases in WNV activity have been reported in some parts of the country. A media advisory in fall 2012 reported that selected states, including Texas, Oklahoma, and Mississippi, had been experiencing significantly greater human WNV activity that year. 2.2.2.2 WNV in Maryland West Nile virus is one of three arboviruses of public health significance in Maryland, along with LaCrosse encephalitis virus (LAC) and St. Louis encephalitis virus (SLE). Since the introduction of West Nile virus (WNV) into the U.S. in 1999, Maryland’s experience with WNV and other arboviruses has been similar to that of many other regions of the country. In 2001, Maryland reported its first human WNV activity, with six human cases including three fatalities. During the peak year of 2003, 73 human cases with eight fatalities were reported. Human arboviral activity has declined in subsequent years, yet continues to pose a threat to Maryland residents. In 2010, 23 human cases were reported, along with eight WNV-positive mosquito pools, one WNV 20 equine case and two human cases of LaCrosse encephalitis. In 2011, 19 human WNV cases, including one fatality, were identified in six Maryland jurisdictions, in addition to arboviral activity in mosquitoes, birds, and horses. A single human case of St. Louis encephalitis was identified that year as well. A dramatic upsurge in WNV activity was seen in Maryland and throughout the U.S. in 2012. By the close of the 2012 arboviral surveillance season on October 31, 46 human WNV cases and four fatalities had been reported in the state. Although Maryland was among a handful of states that experienced elevated WNV activity in 2012, other states in the south, west, and Central Plains accounted for the majority of WNV activity, including Texas, Louisiana, California, and Mississippi, which were among the 13 states that accounted for 80% of all WNV cases that year (CDC). Each year in Maryland, the appearance of the first human WNV case serves as a pivotal reminder that the disease remains a public health threat and that Maryland residents must continue to take precautions. State health officials continue to monitor these arboviral diseases and to assess their impact on Maryland communities and to determine barriers for reducing risk. Maryland is a small, densely-populated state in the mid-Atlantic region of the U.S., and its varied population, topography, and climate present numerous public health opportunities and challenges. The state’s population, as reported by the 2010 U.S. census, is 5.8 million, and this represents a 9% increase since the 2000 U.S. census. The majority of Maryland's population is concentrated in the cities and suburbs surrounding Maryland’s most populous city, Baltimore, and the nation’s capital, Washington, District of Columbia. The eastern, southern, and western portions of the state are more rural, ranging from the Atlantic Coastal Plain in the east where Maryland’s easternmost county, 21 Worcester, meets the Atlantic Ocean, to the mountainous region in Maryland’s westernmost counties: Garrett, Allegany, Washington and Frederick. Surveillance for WNV and other arboviruses is coordinated through a partnership between state officials in the departments of Health and Mental Hygiene, Agriculture, and Natural Resources in collaboration with local health departments and the Department of Defense. 2.2.3 West Nile Virus: Disease Classification Belonging to the Japanese encephalitis virus serocomplex along with LAC and SLE, WNV can cause neuroinvasive disease (disease involving the brain and central nervous system) in humans, in the form of encephalitis (swelling of the brain), meningitis (swelling of the membranes surrounding the brain and spinal cord), meningoencephalitis, or acute flaccid paralysis (weakness or paralysis and reduced muscle tone). Approximately 20% of infected individuals develop a milder WNV Fever, a non- neuroinvasive form of disease characterized by flu-like symptoms including fever, headache, body aches, tiredness, and possible skin rash. The epidemiology of WNV incidence in human populations is seasonal in accordance with the life cycle of the mosquito vectors, usually from early June through late October. Although the most common mode of transmission is via the bite of an infected mosquito, rare instances of transmission through organ donation, blood transfusion, and intrauterine transmission have been previously documented (Hayes et al., 2005). 2.2.4 Transmission West Nile Virus is primarily transmitted through the bite of an infected mosquito. Mosquitoes become infected with WNV when they feed on infected birds, and can then spread the virus to humans and other animals (CDC, 2008). Most infected birds survive, 22 but some, particularly American Crows and Blue jays, will sicken and die from the infection. Horses and humans are considered dead-end hosts of the virus, because they do not develop a high level of virus in their blood, and therefore cannot pass the virus on to another biting mosquito (CDC, 2008). Several mosquito vectors have been implicated in the spread of WNV both nationwide and in Maryland. As described in the 2013 CDC Guidelines for Management of Arboviruses, WNV has been detected in 65 different mosquito species nationwide (CDC 2013). Mosquitoes belonging to the Culex species have been found to be the primary vector of WNV in the U.S., depending on the geographic region. In the northeast, the Culex pipiens, Culex restuans, and Culex salinarius species are the most common WNV vectors (Molaei, 2006). Culex mosquitoes typically feed (take blood meals) during the hours of dusk and dawn (Molaei, 2006), and serve as the basis for recommendations that people avoid going outdoors during those times, as they are considered the prime mosquito feeding hours. 2.2.5 Clinical Presentation Most WNV-infected individuals remain asymptomatic (~80%), while another 20% typically develop WNV Fever (WNF), a milder, self-limiting illness whose hallmarks include fever, headache, muscle aches, and gastrointestinal upset (DeBiasi, 2011). While WNF cases originally appeared to pose minimal disease burden to infected individuals, recent research has revealed that a small but growing number of such cases are resulting in death, particularly among the elderly (Sejvar, 2011). Although only a small percentage of infected individuals develop neuroinvasive disease (<1%), the effects can be severe. While most such West Nile neuroinvasive disease (WNND) cases present 23 with encephalitis or aseptic meningitis, approximately 5-15% present with acute flaccid paralysis, a syndrome resembling poliomyelitis with symptoms that range from single extremity weakness to paralysis or quadriplegia (Murray et al., 2011). Serological surveys have been conducted to determine the prevalence of WNV antibodies among human populations in areas of documented WNV transmission in the US (Artsob et al., 2011). Previous serologic surveys have noted that for every case of WNV neuroinvasive disease, there are approximately 140 WNV infections, suggesting that the impact of WNV outbreaks may extend beyond those cases detected by surveillance systems (DeBiasi, 2011). The CDC collaborated with state health officials to establish a passive surveillance system, ArboNet, through which human WNV cases are reported nationally (Sejvar et al., 2011). Through this system, state health departments routinely monitor arboviral activity in their respective states, reporting cases according to established public health guidelines that classify neuroinvasive and nonneuroinvasive disease according to specific clinical and laboratory criteria which must be met in order to classify a Confirmed or Probable WNV or other arboviral case (CDC NNDSS, 2011). 2.2.6 Populations at Risk for WNV Any person living in an area where mosquitoes are active is at risk for WNV infection. Individuals who spend significant amounts of time outdoors, where they are more likely to come into contact with mosquitoes are at higher risk of infection than those who largely remain indoors. However, age remains by far the most critical risk factor for developing severe WNV disease. According to the CDC, persons 50 years of age and older have been identified as a high-risk group for WNV disease, as they are more vulnerable to diseases of the central nervous system, resulting in encephalitis, 24 aseptic meningitis, or meningoencephalitis (CDC COCA, 2012). A total of 123 non-fatal human WNV cases were reported in the U.S. from 1999-2001, of which 60% were over 60 years old; while the same age group accounted for 75% of the fatal cases (Elliott et al., 2008). Similarly, in the U.S. in 2004, people aged 50 and older accounted for over 60 percent of severe West Nile cases and 95 percent of deaths due to WNV (CDC COCA, 2012). Earlier studies have suggested that despite its high fatality rate, particularly among elderly individuals, that those who survive WNV disease tend to have favorable outcomes (Berner et al., 2003), but more recent research has suggested otherwise (Lindsey et al., 2012). Since human arboviral activity was first detected in Maryland in 2001, there have been 21 WNV fatalities and all were among persons over 50 years of age (range 54-93 years; mean: 72.5 years) (Maryland Arbovirus Surveillance Summary Report, 2011). The impact of WNV infection in older adults has been well-established (Berner et al., 2002; Sejvar et al., 2011). Prior research has noted the value of assessing WNV risks, perceptions, and behavior in elderly populations, but only a few studies have focused exclusively on this group. LaBeaud et al. compared pediatric and adult behaviors relevant to WNV exposure, but looked broadly at all adults 18 years of age and older rather than examining risk and behavior in the over 50 population (LaBeaud et al., 2007). A need exists to better understand the behavioral risks common to this high-risk age group and to identify impediments to prevention so that tailored interventions can be developed to promote increased prevention measures and reduce WNV disease and mortality in this population. 25 While national recommendations have generally targeted adults at least 50 years of age, for purposes of this study, the focus was on adults 60 years old and older in an effort to target those individuals who may encounter more barriers (physical, mental, financial, or otherwise) when engaging in WNV prevention behaviors. Specifically, people aged 60 years and over are more likely to have retired from work and to spend time at home or outdoors engaged in leisure-time activities that might bring them in contact with mosquitoes. People in this age group may also be more subject to financial limitations of a fixed income (social security) that could limit their ability to support local mosquito control programs or to purchase tools that may aid in mosquito bite prevention. 2.2.7 Prevention Behaviors In light of the ongoing public health threat posed by WNV throughout the country, effort has been made to identify WNV behavioral risk factors and effective prevention measures. In the absence of an available human WNV vaccine and given the sometimes undetectable nature of WNV infection, personal behaviors have been identified as important for preventing WNV disease. Key prevention measures are recommended to lower the risk of WNV infection. These key individual personal protective behaviors are commonly referred to as the 4 D’s: DEET (wearing insect repellent containing DEET), Dress (wearing long sleeves and long pants when outdoors), Drain (draining objects that may collect water in which mosquitoes lay eggs), and Dusk to dawn (avoiding the outdoors during prime mosquito feeding hours of dusk and dawn) (Gujral et al., 2007). Other prevention measures include source reduction measures such as eliminating routes of indoor exposure by repairing damaged window and door screens and cleaning gutters where water may collect. Prior research has examined the frequency 26 with which individuals perform the above preventive behaviors and tested the association of those behaviors with a reduction in WNV infection among various populations (Loeb et al., 2005; Gujral et al., 2007). Findings from these studies suggest that engaging in these preventive steps can significantly reduce the risk of WNV infection. In addition, community prevention efforts in the form of mosquito control programs further supplement individual protective behaviors to reduce WNV activity. These community prevention methods generally involve widespread mosquito control programs that reduce the number of mosquitoes and seek to eliminate mosquito habitats. Such methods include adulticiding (applying pesticides to kill adult mosquitoes) and larviciding (products used to kill immature mosquitoes before they reach adulthood) (CDC, 2008). 2.3 Health Belief Model 2.3.1 Definition and Origins of Model The Health Belief Model (HBM) was created for the purpose of explaining an individual’s engagement in a particular health behavior and provides a theoretical framework for describing individual actions according to perceptions of the risks and benefits associated with the target behavior. Originally developed in the 1950s by US Public Health Service officials to explain people’s lack of participation in free tuberculosis screening programs and other preventive health services, the HBM has since evolved in response to the ever-changing needs of the public health community (Hochbaum, 1958; Rosenstock 1960, 1974). It has been widely applied within three domains of health behavior: 1) preventive health behaviors, including health protective and risk reducing behaviors; 2) sick role behaviors, including compliance; and, 3) clinical or medical care seeking behaviors (Janz & Becker, 1984). Although the model has been 27 modified and expanded over the years since its inception, to adapt to various public health concerns, it traces its origin to two psychosocial learning theories: Stimulus- Response Theory and Cognitive Theory (Champion & Skinner, 2008). Proponents of the former believed an individual’s behavior was driven by events or reinforcements that accrue after the behavior has been enacted, while proponents of the latter maintained that learning was a function of the subjective value of an outcome and the probability or expectation that certain actions would lead to that outcome (Champion and Skinner, 2008). Thus, the HBM is considered a value-expectancy theory. As such, the model has been formulated to depict the value that an individual assigns to an outcome associated with his/her expectation that performing a particular behavior will result in that outcome. Specifically, the likelihood of a person engaging in a preventive behavior is a function of a subjective process of weighing the costs and benefits of taking the action. Those costs and benefits are direct determinants of an individual’s actions, which prompt adoption of a behavior (Reid et al., 2011). In short, the model posits that a person will take steps to avoid illness and reduce his/her risk of disease with the expectation of a positive health outcome (Janz & Becker, 1984). 2.3.2 Health Belief Model Structure The original Health Belief Model consists of four core constructs: perceived susceptibility (whether a person considers him/herself to be at risk for the disease), perceived severity (how serious a person believes the disease to be), perceived benefits (one’s belief in the efficacy of the recommended action to reduce the risk or seriousness of an adverse outcome), and perceived barriers (one’s opinion of the tangible and psychological costs of the recommended action). Perceived susceptibility and perceived 28 severity together form perceived threat, or the motivation to take action (Murray-Johnson et al., 2006). This combined mechanism is believed to provide a person with the force or energy to act on a behavior (Janz and Becker, 1984). Barriers refer to a person’s assessment or appraisal of potential adverse consequences that may occur as a result of taking a health action (Mattson et al., 1999). As described by Mattson et al., perceived barriers can be physical, psychological, financial, or emotional (Mattson et al., 1999). Benefits refer to the extent to which a person considers certain health behaviors to be effective or beneficial at reducing risk or impact of the disease (Wong et al., 2005). Several studies have suggested that among all the HBM constructs, perceived barriers are the most powerful predictor of health behavior (Mahoney et al., 1995; Atkinson et al., 2009; Chin et al., 2012). Similar to how severity and susceptibility combine to form perceived threat, perceived benefits and perceived barriers are collectively regarded as composing a person’s outcome expectancy, or likelihood of taking action (Michel et al., 2011; Murray-Johnson et al., 2006). When combined, these two constructs act in an additive fashion, with barriers subtracted from benefits to yield a net behavioral effect, making it possible to estimate a person’s likelihood of engaging in a specific action (Murray-Johnson, 2006; Carpenter, 2010). Cues to action were also included in the original HBM for the purpose of activating behavior. The cues serve as triggers or stimuli that prompt persons to engage in a specific health behavior. They can be internal (e.g. symptoms of illness) or external (e.g. disease in family members or friends or media health outreach) to the individual and are designed to create awareness of the health threat as well as to stimulate a readiness to act against it (Mattson et al., 1999; Atkinson et al., 2009). Recent articles in the 29 literature have noted that cues to action remain the least researched and least developed of all the HBM constructs and as such are often addressed in only a cursory fashion (Carpenter, 2010; Mattson et al., 1999). As a result, such cues are rarely addressed in intervention research that examines the application of the HBM to specific health behaviors (Mattson et al., 1999; Carpenter, 2010). Despite the insufficient research on cues to action, they have been demonstrated to serve as valuable motivators to action for such behaviors as Severe Acute Respiratory Syndrome (SARS) prevention efforts, diabetes foot exam practice, and use of birth control (Chin et al., 2012; Wong et al., 2005; Wang et al., 2006). Self-efficacy (confidence in one’s ability to perform a given behavior) was added to the HBM in 1988 by Rosenstock, Strecher, and Becker to address the need for individuals to feel competent to surmount adverse health habits such as smoking and overeating (Rosenstock et al., 1988). The authors were careful to make a distinction between efficacy expectations (the belief that one can successfully execute the necessary behavior to yield an outcome) and outcome expectations (the conviction that a specific behavior will likely lead to a particular outcome) so as to eliminate any confusion over such different perceptions and clarify that the former involves a judgment of one’s own abilities (Rosenstock et al., 1988). The self-efficacy construct arose from Bandura’s Self-Efficacy Theory, which was introduced in 1977 and defined a person’s perceived efficacy as a situational awareness of his/her own abilities to act in certain situations (van der Bijl, 2001). The construct has been described as arising from four key information sources: (1) personal accomplishments, (2) vicarious experiences, (3) verbal persuasion, and (4) physiological 30 information (van der Bijl, 2001). As such, it reflects a person’s belief in his/her ability to surmount a specific task or situation, by making judgments about their capacity to learn a new behavior and/or curtail a damaging one (AbuSabha et al., 1997). The construct of self-efficacy typically applies to habitual or repeated behaviors, such as smoking, physical activity, and dietary behaviors (Brewer and Rimer, 2008). It has been shown to have high specificity and effectiveness at predicting smoking cessation, weight control, and exercise (AbuSabha et al., 1997). Testing of the self-efficacy construct can be complex and requires measurement according to the three central dimensions of magnitude, strength, and generality (AbuSabha et al., 1997, van der Bijl 2001). Self-efficacy has been demonstrated to be a strong predictor of various behaviors across several different behavior theories/models, including Social Cognitive Theory, the Theory of Reasoned Action, Theory of Planned Behavior, and the HBM (AbuSabha et al., 1997). It has been applied to numerous health behaviors, including physical activity, dietary habits, sexual risk behaviors, and breast self-examination, among others (Pinto and Floyd, 2008; Lin et al., 2005; Norman et al., 2005). In addition to the main HBM constructs described above, the HBM has been conceptualized to contain a series of modifying factors that facilitate the process of assessing perceived threat and determining the likelihood of action. These modifying factors interact with an individual’s perceptions about the focal health condition or disease and include: demographic variables such as socioeconomic status, gender, age, race, and ethnicity; psychological variables such as personality type; and, structural variables such as prior experience with disease and knowledge of disease (Rosenstock, 31 1974; Green and Glanz, 1995; Champion and Skinner, 2008; Painter et al., 2008, Hall, 2011). Specifically, the modifying factors are intended to serve as either mediators or moderators of the relationship between selected HBM constructs and the likelihood of an individual taking a particular action (Rosenstock, 1974). As described by Hall et al. (2011), the constructs emphasize such modifiable factors in order to facilitate tailored interventions to promote healthy behaviors. The mediating role of these modifying factors is important because it allows for more robust analysis of the direct effects of the core constructs by placing them in the context of such demographic variables as age, gender, income, and race (Fulton et al., 1991). Figure 2 provides an illustration of the HBM. 2.3.3 Measurement of HBM Constructs Measurement of each of the HBM constructs can prove challenging and complex. Conventionally, perceived susceptibility and perceived severity are measured in combination as perceived threat from a specific disease or health condition. This measurement involves development of a combinatorial model based on the product of the two constructs: Perceived susceptibility x Perceived Severity = Perceived Threat (Glanz, 2008). Perceived threat has been demonstrated to be a consistent predictor of certain behaviors such as infection control, influenza immunization and cancer screening (Wong et al., 2005; Wolf et al., 2008; Flood et al., 2010; Tavafian, 2012). According to Glanz et al. (2008), an elevated state of (perceived) severity is needed in order for perceived susceptibility to become a strong predictor of behavior. Within the context of high perceived severity, the authors explain, perceived susceptibility will become a stronger predictor of an individual’s intention to engage in a 32 specific health-related behavior, such as WNV prevention measures (Champion and Skinner, 2008). This refers to a relationship between the two constructs wherein an individual’s perception of his/her vulnerability to contracting a disease is impacted by the perception of its seriousness (Galloway, 2003). The collective perceived threat is described as the degree of the overall health threat combined with the person’s confidence that the target behavior will ameliorate that threat (Galloway, 2003). Source: Glanz; Champion and Skinner (2008) Figure 2. Health Belief Model 2.3.4 Strengths and Weaknesses of HBM Strengths Given its history as one of the oldest conceptual models of health behavior, the HBM has been the subject of quite a few studies critiquing its effectiveness at explaining and/or predicting individual health behaviors. Early reviews of the model, such as Janz and Becker’s (1984) review of 46 studies involving the HBM, provided considerable 33 empirical support for the model, concluding that it is highly useful for examining preventive behavior (actions taken to avoid illness) over sick-role behavior (actions taken to restore health after diagnosis of a medical condition) and that its constructs represent valuable contributors to the prediction and explanation of individual health behaviors. More recent reviews have supported the assertion that perceived barriers are, in fact, one of the strongest predictors of behavior. A key contribution of the HBM to the behavioral sciences is its focus on personal perceptions as key constructs and determinants of individual behavior (Champion and Skinner, 2008; Elder et al., 1999; Walker, 1999). Specifically, the main strength of the HBM has been identified as its “common sense operationalization” that focuses on beliefs pertaining to decisions about individual health behaviors (Conner, 2010). The HBM constructs operationalize the manner in which individuals incorporate their knowledge, beliefs, and attitudes toward a behavior into a cost/benefit analysis of the pros and cons of engaging in it (Conner, 2010; NCI, 2005). Additional strengths of the HBM include its diverse applicability to various ethnic groups and its utility for explaining or predicting habitual behaviors such as mammography, medication adherence, or substance use (Conner, 2010; Pasick et al., 2008). Although it has been argued that a valid assessment of the effectiveness of the HBM constructs in predicting behavior can only be achieved through use of longitudinal studies, its validation has been inferred from cross-sectional study designs. Specifically, Carpenter noted in his review that there has been dissent among reviewers over the years regarding the capacity of the model to be used for examination of cross-sectional data (Carpenter, 2010). He noted that Rosenstock expressed concern about the validity of 34 applying the HBM to cross-sectional data, due to the problem of subjects’ perceptions changing over time and the potential to yield overestimates of relationships among variables. Janz and Becker (1984) argued the opposite, namely that cross-sectional associations would be weaker than those found in longitudinal studies. Weaknesses Despite its advantages, the HBM has been noted to have significant limitations as well. First, it assumes that people’s actions are under volitional control, which may not always be the case. That is, the model operates most effectively when based on the assumption that a person’s actions are habitual and are entirely voluntary (due to an individual’s choice) such as condom use, exercise, and mammography screening (Wong et al., 2005; Calvocoressi et al., 2004; Deavenport et al., 2011 ). Conversely, addictive behaviors triggered by a physiological addiction such as substance abuse and smoking are less readily explained by the HBM as a whole, and may only be accurately predicted by select constructs from the HBM and other behavioral models (Bonar et al., 2011; Schofield et al. 2007). Second, although the model has been demonstrated to have good explanatory power for certain behaviors, its effectiveness tends to vary depending on the specific health threat (Greene et al., 2003). For example, susceptibility, severity and cues to action have all been suggested to be poor predictors of condom use, but strong predictors of influenza vaccination (Reid et al., 2011; Flood et al., 2010). Third, the model fails to distinguish between first time and repeat behaviors. In addition, a number of challenges have emerged in the effort to develop suitable scales for measurement of HBM concepts. These scale development efforts have been 35 plagued by such problems as inconsistency with theory, measurement error, and failure to account for the temporality of relationships (Glanz, 2008). Some studies have indicated that the HBM has very limited utility for explaining the variance in certain populations for selected behaviors, such as condom use and other sexual practices among college students (Lollis et al., 1997) and that it has limited predictive power overall (Yarbrough et al., 2001). Galloway suggested that the HBM is based on the premise that behavior is prompted by a desire to avoid disease or illness, unlike other models such as the Health Promotion Model, which promotes enjoyment or satisfaction with one’s health status (Galloway, 2003). Beyond the assertion that it is a model of avoidance behavior, he further added that the HBM is “merely an articulation of variables affecting behavior,” implying that the model lacks cohesion as a viable tool to predict behavior (Galloway, 2003). Roden echoed this sentiment in her critique of the HBM, stating that the model is limited by its “unclear construct and relationship development” (Roden, 2004). It has been further suggested that the HBM has little or no capacity for predicting long-term health behaviors (Jones et al., 2013). Furthermore, a number of studies have suggested that the HBM explains only a small proportion of the total variance in an analysis of behavior, implying that it is not a useful model for predicting behaviors (Lollis et al., 1997). To that end, Mattson et al. have proposed re-conceptualizing the HBM to place the cues to action at the center, such that they mediate the other constructs (Mattson et al., 1999). As noted earlier, there is also disagreement among researchers as to which HBM construct is the most effective predictor of behavior, as this often varies by disease 36 outcome (Greene & Brinn, 2003). For example, while several studies have suggested that perceived barriers are the most powerful predictor of behavior (Mahoney et al., 1995; Atkinson et al., 2009), Swaim et al. pointed out that this is only true for “one-shot” behaviors such as vaccination and screening exams, while for behaviors practiced over the long term, like dietary habits, dental hygiene and exercise, self-efficacy emerges as the best predictor (Swaim et al., 2008; DeBate et al., 2006). Further support for that assertion was articulated by Fulton and colleagues, who stated that the model’s utility was unclear for long-term behaviors that involve lifestyle changes (Fulton et al., 1991). Conversely, Mahoney et al. have asserted that perceived barriers remain the most useful predictor of behavior, followed in order by benefits, susceptibility, and severity (Mahoney et al., 1995). Despite some research studies which have suggested that perceived susceptibility and perceived severity are both among the strongest predictors of preventive health behaviors, the latter has sometimes been shown to be inconsistent at predicting some health behaviors (Yarbrough et al., 2001). Despite these limitations, the HBM remains a widely used theory. It is an appropriate theoretical framework for development of a survey instrument to examine West Nile Virus prevention behaviors in older adults in Maryland. As previously mentioned, it has been applied in other research on WNV protective behaviors. Moreover the HBM, as will be described subsequently, has demonstrated utility in explaining individuals’ behaviors for a variety of other infectious diseases. 37 2.3.5 Applications of the Health Belief Model Following its original use in the context of evaluating TB screening behaviors, the model has subsequently been applied to a variety of health behaviors including mammography screening, vaccination compliance, adherence to medication regimens, and engagement in safe sex behaviors (Janz and Becker, 1984; Glanz, 2008). The HBM has also been used to explain behavior with respect to a number of chronic diseases, such as cancer (including breast, cervical, and prostate cancer), and related screening behaviors (Calvocoressi et al., 2004; McQueen et al., 2010; Deavenport et al., 2011). 2.3.6 Health Belief Model and Communicable Diseases The HBM has also been used as a theoretical framework to guide interventions and exploratory focus group studies for various communicable diseases, including Dengue Hemorrhagic Fever (Phuanukoonon et al., 2006), human papillomavirus (HPV) (Krawczyk et al., 2012; Reiter et al., 2009), Influenza A (Flu), MRSA, rotavirus (Morin et al., 2012), vancomycin-resistant enterococcus (Curry and Cole, 2001) and assorted zoonoses (diseases transmitted between vertebrate animals and humans) (Bosch et al., 2010). This suggests that the model may have utility in explaining individual behaviors related to risk and transmission of these and other communicable diseases. 2.4 Health Belief Model and West Nile Virus Prevention Although the HBM has been applied to the study of preventive behaviors for selected communicable diseases, its use in arboviral disease research has been limited. A small body of social science research has considered the application of health behavior theories and models to WNV prevention, but there remains a paucity of research that 38 explores the application of such theoretical constructs to the prevention of WNV and other mosquito-borne diseases. A limited number of studies have examined individuals’ attitudes, perceptions, and/or knowledge about WNV disease perception, and a small proportion of those have applied a theoretical framework to that analysis. With respect to WNV prevention, the Health Belief Model can be constructed as follows:  Perceived susceptibility to West Nile Virus refers to a person’s belief that s/he is vulnerable to WNV infection; this can be a key component of his/her intention to take action to prevent it. This can be challenging to measure however, as not all members of the target population (adults over 60 years of age) are aware of the risks associated with WNV infection.  Perceived severity of West Nile Virus refers to how serious an individual considers WNV to be; they may or may not recognize that it can cause severe neurologic disease resulting in significant morbidity and/or death. This is particularly relevant for persons aged 50 and over as they are at greatest risk of severe illness and death from WNV infection.  Perceived benefits of WNV protective behaviors can include a variety of positive outcomes associated with engaging in the preventative behaviors; such benefits can include avoidance of infection, illness, and hospitalization as well as greater confidence in ability to protect ones’ self from the disease.  Perceived barriers to WNV protective behaviors refer to any factor that impedes a person from engaging in WNV protective behaviors; barriers can include financial limitations that prevent a person, lack of education about the appropriate 39 repellents to use or how to use them, and/or physical limitations that may prevent a person from regularly draining areas of standing water on their property.  Cues to action are particular events or messages that prompt an individual to engage in personal protective behaviors; they can include learning of an acquaintance that has been diagnosed with WNV disease, hearing public health alerts about WNV on the news, or receiving information about it from a health care provider.  Self-efficacy for WNV prevention refers to the extent to which a person feels confident that s/he is capable of performing the steps needed to avoid WNV infection (such as using insect repellent, dressing appropriately when outdoors, avoiding the outdoors during mosquito feeding hours, and routinely emptying items containing standing water). 2.5 Central Themes in West Nile Virus Prevention Behavior Literature Some key themes have emerged in the existing literature regarding assessment of people’s knowledge, attitudes, and behaviors toward WNV prevention. Some studies incorporated a theoretical/conceptual framework while others did not. Some of the earliest studies examined individuals’ knowledge, attitudes, and behaviors related to WNV, either immediately preceding or immediately following two of the largest WNV epidemics nationwide in 2002 and 2003. These studies took the form of retrospective analyses or cross-sectional designs and included serosurveys to determine exposure to WNV as well as assessment of people’s attitudes and behaviors toward the virus. Given the widespread impact of WNV outbreaks affecting all of North America, researchers in 40 both the U.S. and Canada began to analyze the interrelationships between people’s knowledge, attitudes and their behaviors to reduce their risk of WNV infection. A discussion of these studies and their main findings are summarized in Table 1 and described in more detail below. Table 1. Summary of Studies on West Nile Virus Knowledge, Attitudes, and Behavior Year Name of study Authors Study design Theoretical model Constructs Findings 2001 WNV Serosurvey and Assessment of Personal Prevention Efforts in Area with Intense Epizootic Activity, 2000 McCarthy et al. Household survey and serosampling of CT residents aged 12 and older None N/A High WNV awareness; female gender and high perceived risk associated with practicing 2 or more PPBs; adults >50 less likely to practice PPBs 2003 Knowledge, Attitudes, and Behaviors about WNV— Connecticut, 2002 Adams et al. Cross- sectional; telephone survey; CT residents aged ≥ 18 years None N/A People over 50 more likely than younger adults to always use at least 1 PPB. Use of PPBs highly associated w/ female gender, age ≥ 65, and upper income 2003 Pre-WNV Outbreak: Perceptions and Practices Herrington et al. Cross- sectional; nationally representative survey Health Belief Model Perceived susceptibility; perceived severity Perceived susceptibility to repellent toxicity outweighed vulnerability to disease 2003 Current Knowledge, Practices, and Attitudes of Miami-Dade Residents about WNV Disease and its Prevention Leguen et al. Cross- sectional CATI survey of adults 18 and over in Miami-Dade County None N/A Most respondents familiar with WNV; nearly half use insect repellent; 20% believe WNV not a serious health threat 41 Year Name of study Authors Study design Theoretical model Constructs Findings 2004 WNV in British Columbia Aquino et al. Cross- sectional; telephone survey Health Belief Model Perceived susceptibility, severity, barriers, benefits, cues to action Perceived hazard of DEET, time needed to drain standing water are major barriers 2004 WNV: Knowledge, Behaviors, Attitudes Among Seniors, Denton County, Texas Henry et al. Focus groups and face-to- face interviews with Denton County residents aged 50 and older None N/A Most respondents knowledgeable; lack of familiarity with symptoms; majority recognized that WNV can be fatal; few practice PPBs 2005 Evaluation of West Nile Virus Education Campaign Averett et al. Cross- sectional; telephone survey of KS residents and media survey of news outlets None N/A Most respondents knowledgeable about WNV risk; mass media and word of mouth most successful methods of WNV info dissemination 2005 What Seniors Say About WNV: The Threat is Not Over Bitto et al. Cross- sectional survey and qualitative evaluation of educational intervention Health Belief Model Perceived susceptibility, severity, barriers, self- efficacy, cues to action Most seniors motivated to use PPBs; Lack of self-efficacy and perceived barriers were common among seniors who had problems using insect repellent 2005 Heeding the Message? Determinants of Risk Behaviours for West Nile Virus Wilson et al. Cross- sectional telephone survey None N/A Nearly ¼ of respondents had not heard of WNV and almost half thought it was an unimportant health issue; fewer urban residents used DEET; nearly 1/3 do not routinely drain standing water 42 Year Name of study Authors Study design Theoretical model Constructs Findings 2006 The Effect of Health Communicatio n on a Statewide WNV Public Health Education Campaign Fox et al. Cross- sectional; telephone survey of adult KS residents None N/A Men more likely respond to radio ads, while women prefer brochures and literature; Internet is effective tool; older adults fixed window screens more often than younger adults 2006 Seroprevalenc e of WNV in Saskatchewan ’s Five Hills Health Region Schellenberg et al. Cross- sectional phone survey & seroprevalence study; adults 18 years and over None N/A Good knowledge of WNV and prevention methods; odds of being WNV+ were 6x higher for rural residents 2007 Behavioral Risks for WNV Disease, Northern Colorado, 2003 Gujral et al. Cross- sectional telephone survey of adults 18 and over None N/A Fort Collins residents more likely than Loveland residents to use DEET; older persons seldom use DEET; those with low income seldom practice PPBs; those not worried about WNV rarely practice PPBs 2007 Exposure to WNV during the 2002 Epidemic in Cuyahoga County, Ohio: A Comparison of Pediatric and Adult Behaviors LaBeaud et al. Cross- sectional household survey None N/A Children had higher outdoor exposure than adults; both children and adults educated about WNV; subjects more worried about WNV infection than pesticide use 2007 The Validation of a WNV Survey Based Yerby Qualitative Health Belief Model Perceived susceptibility, severity, benefits, Modified survey instrument validated; all 43 Year Name of study Authors Study design Theoretical model Constructs Findings on the Health Belief Model barriers, cues to action, self-efficacy constructs except Perceived Barriers to Repellent use predicted personal protective behavior 2008 West Nile Virus: The Buzz on Ottawa Residents’ Awareness, Attitudes, and Practices Elliott et al. Cross- sectional telephone survey None N/A WNV awareness and worry very high; female gender and concern for WNV predicted practice of PPBs 2008 Self-reported Protective Behaviour Against WNV Among Pregnant Women in Toronto Kiehn et al. Cross- sectional survey of pregnant clients at outpatient obstetric clinics None N/A High awareness of WNV and transmission; more pregnant women practiced 2 or more PPBs than general population; major concerns about hazardous effects of insect repellent on baby; worry about WNV infection during pregnancy 2009 Knowledge, Perceptions, and Practices: Mosquito- borne Disease Transmission in Southwest Virginia Butterworth Cross- sectional; in- person survey Health Belief Model Perceived susceptibility, severity, benefits, and barriers Gender predicted knowledge; perceived hazards of repellent significant barrier; removing standing water best way to avoid mosquito bites 2009 Public Health Responses to Elmieh Cross sectional; Health Belief Perceived susceptibility, WNV risk perception 44 Year Name of study Authors Study design Theoretical model Constructs Findings WNV: The Role of Risk Perceptions and Behavioral Uncertainty in Risk Communicatio n and Policy telephone survey Model severity, benefits, barriers, cues to action, self-efficacy varied with age, gender, and household location; laypeople more willing to trade-off risk than experts 2009 Effect of WNV Perceptions and Knowledge on Protective Behavior and Mosquito Breeding in Residential Yards in Upstate New York Tuiten et al. Cross- sectional KAP survey and entomological survey via property inspection None N/A Most respondents knowledgeable about WNV transmission and risk groups; fear of repellent was major barrier; no association between WNV knowledge and mosquito production on property 2011 Public Attitudes, Knowledge and Practices on West Nile Virus Eichler Cross- sectional telephone survey Health Belief Model Perceived susceptibility, severity, benefits, barriers, cues to action, self-efficacy WNV knowledge and concern did not predict PPB practice; older age predicted increased PPB use U.S. Studies of WNV Protective Behavior Herrington et al. (2003) administered a nationally representative survey instrument to a primary sample of 1500 adults plus an oversample of an additional 250 adults in six states to describe the prevalence of adults’ perceptions and practices prior to the first outbreak of WNV in 1999. The survey instrument used selected constructs from the HBM (perceived susceptibility and perceived severity) and selected other behavioral variables from other theories. They found that one-third of all respondents perceived use of insect repellent on skin, staying indoors during peak mosquito hours, and wearing long 45 sleeves and long pants as the most effective measures for preventing WNV (Herrington et al., 2003). Results also suggested that, for many individuals, the perceived susceptibility to toxicity from insect repellent outweighed their perceived susceptibility to WNV encephalitis. However, concern about being bitten by mosquitoes (perceived susceptibility to mosquito bites) was the most significant predictor of actions to prevent such bites (Herrington et al., 2003). In another one of the earliest studies conducted immediately after the 2002 WNV outbreak in the US, Connecticut state health officials, in collaboration with CDC, administered a random-digit-dialed telephone survey to Connecticut residents aged ≥ 18 years old (Adams et al., 2003). Staff at the Connecticut Department of Public Health modified the state’s existing Behavioral Risk Factor Surveillance System (BRFSS) survey to include items asking about individuals’ knowledge, awareness, attitudes and behaviors toward WNV prevention. They identified significant associations between female gender and practice of two or more personal protective behaviors (PPBs) as well as between worry about getting WNV infection and practice of PPBs. Findings suggested that awareness and knowledge of the severity of WNV were high among elderly individuals (Adams et al., 2003). Most notably, respondents aged 50 years of age and older were more likely to always practice at least one personal protective behavior than younger respondents (aged 18-49 years old) (Adams et al., 2003). Their results suggest that adults in the high-risk (over 50) age group are acutely aware of their own susceptibility to WNV disease. Gujral et al. (2007) administered a telephone survey to Colorado residents in two adjacent cities, following the 2003 WNV outbreak, to assess their personal protective 46 behaviors. They determined that greater exposure to the outdoors during mosquito feeding times and lower use of insect repellent both contributed to the incidence of WNV neuroinvasive disease (Gujral et al., 2007). They also found that older adults and those not concerned about WNV infection were less likely to use insect repellent containing DEET. However, perceived risk (threat) of WNV disease was determined to be a consistent predictor across all age groups. Like many studies, this one was somewhat limited by both recall and reporting bias, as it was administered several months after a WNV outbreak, but such impacts were considered minimal (Gujral et al. 2007). LaBeaud and colleagues (2007) conducted a survey to compare WNV knowledge and behaviors between adult and pediatric populations in Cuyahoga County, Ohio during the 2002 WNV epidemic. A questionnaire was administered to residents of randomly- selected households during summer 2002. Seropositivity of participants was measured through blood sampling. While both populations were well informed about WNV, television, rather than health care providers, served as participants’ primary means for obtaining WNV education. Furthermore, children tended to spend more time outdoors than adults and were far less likely to wear protective clothing such as long sleeves and long pants when outside. As a result, children tended to experience more mosquito exposure. Children’s attitudes toward WNV risk were essentially similar to those of their parents (LaBeaud et al., 2007). Only one study has exclusively examined attitudes and behaviors toward WNV among adults over 50 years of age. Bitto and colleagues (2005) developed and implemented a tailored health education intervention among senior citizens in Monroe County, PA in an effort to educate this high-risk group and to counter widespread myths 47 and misperceptions about WNV disease in the general population. The authors partnered with local health officials and senior clubs to conduct a series of training sessions and group discussions and to administer a descriptive survey that captured respondents’ demographic information as well as their knowledge, attitudes, and practices regarding WNV prevention (Bitto et al. 2005). They determined that several of the HBM constructs were predictive of WNV protective behavior in elderly populations; namely perceived threat (susceptibility and severity) and perceived benefits and barriers, all of which were associated with reduced WNV risk behavior (Bitto et al., 2005). In 2009, Tuiten et al. conducted the first study that combined a Knowledge, Attitudes, and Practice (KAP) survey with an entomological survey of residential mosquito breeding sites. As with other studies, most respondents demonstrated acceptable knowledge of the risk factors, transmission, and prevention measures for WNV. Fear of the odor and toxicity of insect repellents was again found to be a significant perceived barrier to their use. Two research studies specifically sought to assess the impact and effectiveness of WNV public education campaigns on selected communities. Both Averett et al. (2005) and Fox et al. (2006) conducted surveys to assess Kansas residents’ knowledge, attitudes, and behaviors following a statewide WNV education campaign. While both studies revealed that the majority of respondents were familiar with WNV, its risk groups, and modes of transmission, perceived risk of acquiring WNV disease remained surprisingly low (Averett et al., 2005). 48 Canadian Studies of WNV Protective Behavior Aquino et al. (2004) developed a questionnaire based on the HBM and administered it to 309 Canadian residents to examine determinants of engaging in WNV protective behaviors. They found that the most significant barrier to engaging in such behavior was the belief that DEET is a major health and environmental hazard. They also identified knowledge deficiencies among respondents, particularly in regards to the groups at greatest risk of severe WNV disease (adults over 50 years old) and determined that increasing perceived susceptibility and perceived severity to match actual susceptibility and severity could benefit this population (Aquino et al., 2004). Finally, it was concluded that dissemination of WNV prevention information via TV and other media increased respondents’ odds of practicing protective behaviors more frequently. Like the Herrington (2003) study, the Aquino study also found that potential health hazards of insect repellent containing DEET were a significant barrier to practicing that and other protective behaviors. Likewise, Elliott et al. (2008) conducted a telephone survey of Ontario residents in spring 2003 to assess their uptake of public health messaging following the 2002 WNV outbreak. They discovered that despite high levels of awareness of WNV disease risk that actual practice of PPBs was surprisingly low, suggesting a need to revise risk communication strategies to reinforce the severity and public health impact of WNV disease. In another study of Canadian populations, Schellenberg et al. (2006) conducted a cross-sectional prevalence study to assess seroprevalence as well as knowledge, attitudes, and behaviors among residents in an area of Saskatchewan that reported the highest 49 number of WNV human cases per capita in summer 2003. Results revealed an overall good knowledge of methods of prevention and transmission of WNV. Also, age disparities regarding select behavioral practices were detected, i.e., persons over 60 years old were less likely to use repellent containing DEET as compared to their younger counterparts. Following the discovery of alternate yet rare modes of WNV transmission (e.g., organ donation, blood transfusion, breastfeeding, and intrauterine transmission), a number of researchers undertook studies of these special populations. In 2005, a single case of WNV during pregnancy with long-term fetal sequelae was noted (Kiehn et al., 2008). In 2006, Kiehn et al. conducted a cross-sectional survey of WNV protective behaviors among a sample of pregnant women in Toronto and found that the majority of them practiced protective behaviors to avoid mosquito bites (Kiehn et al., 2008). However, similar to respondents in other behavioral studies, they expressed considerable concern about the safety of using insect repellents (while pregnant). Doctoral Dissertation Research and Theses examining WNV Prevention Behavior In addition to the published, peer-reviewed studies of WNV health behaviors described above, a number of graduate student researchers conducting thesis or dissertation work have also examined the application of the HBM to the development and implementation of survey instruments for the purpose of assessing individuals’ knowledge, perceptions and behaviors relative to prevention of WNV and other mosquito-borne diseases. One of the earliest such research studies completed during doctoral study was a validation of a WNV survey instrument. In her 2007 dissertation, Yerby modified an 50 existing Canadian WNV survey instrument that was based on the HBM and validated it among a sample of women in the Southeastern states. Specifically, the survey captured West Nile virus knowledge, attitudes, and personal protective behaviors (PPB) among women 19 years old and over living in West Alabama in an effort to determine if the HBM constructs predicted their practice of PPBs, as well as to gather information about that group’s knowledge, attitudes, and practices surrounding WNV and to formulate educational programs for the future. The multi-stage validation process involved a five- step process that included expert panel review, face-to-face interviews, focus groups, pilot testing of the instrument, and statistical analysis (Yerby, 2007). Her efforts ultimately resulted in the creation of an instrument that was judged to be both valid and reliable for assessing the perceptions and WNV prevention behaviors of women (Yerby, 2007). Yerby found that all construct factors, with the exception of Perceived Barriers to Repellent Use were predictive of personal protective behaviors. Her dissertation resulted in the first formally validated theory-based survey instrument for capturing WNV prevention behaviors. In contrast to Yerby’s study, which focused on women 19 years old and over in southern Alabama, this study was the first to examine the WNV preventive behaviors of high-risk adults (men and women) in Maryland. In 2009, Butterworth conducted a cross-sectional survey study to examine the relationship between selected demographic and socioeconomic variables and individual perceptions of mosquito-borne diseases in two Virginia counties (Butterworth, 2009). This research was conducted within a disease ecology framework, using the HBM as a theoretical framework, and applying principles of medical geography to analyze human- mosquito interactions and consider the impact of behaviors on disease prevention 51 (Butterworth, 2009). Study findings suggested that concerns about susceptibility to and severity of mosquito-borne disease varied significantly by gender and other factors, with women demonstrating greater perceived susceptibility than men and stronger belief that such diseases require medical attention. Lack of time was the most significant barrier preventing respondents from emptying areas of standing water around their properties. In addition, a perception of physical hazards associated with insect repellent use emerged as a major barrier to their use. A number of potential benefits from having greater accessibility to WNV prevention information online were also identified, and suggestions made to incorporate this into the Virginia Health Department web pages. While Butterworth’s study focused on a disease ecology framework to examine WNV behaviors in Virginia residents, it did not consider the specific perceptions and barriers experienced by older adults in that region. Accordingly, this dissertation examined knowledge, attitudes and behaviors toward WNV prevention among a specific sample of adults 60 years of age and older in Maryland, who are at greatest risk of severe disease. In the same year, Negar Elmieh (2009) examined risk perceptions, risk communication, and behavioral uncertainties related to WNV prevention and control among residents of British Columbia, Alberta, and Manitoba, Canada. The HBM was used as a theoretical framework to frame some of the research questions, which sought to determine which health beliefs influenced respondents’ engagement in recommended WNV risk reduction behaviors and how those behaviors varied according to potential WNV exposure and demographic variables. Perceived barriers were significantly associated with all outcomes except that of wearing protective clothing (long sleeves and long pants) (Elmieh, 2009). Specifically, study participants were less likely to use 52 window screens, avoid the outdoors during dusk and dawn, or initiate source reduction behaviors (draining items containing standing water). Of particular interest was the finding regarding perceived susceptibility to WNV infection. Respondents’ perception of their own risk of acquiring the disease was driven not only by the actions they chose to take (or not take) but also by their immediate environment: they felt varying amounts of vulnerability to getting infected at home versus outside during seasonal recreational activities. Elmieh inferred from this finding that future WNV educational materials could be tailored to address such concerns by specifically targeting each area (creating flyers that promote use of window and door screens to prevent mosquito exposure at home and bulletin boards and billboards posted at public parks and other public outdoor venues). Similarly, Eichler (2011) developed an original survey to examine public attitudes, knowledge, and behavior practices toward WNV in Delaware County, PA. Results revealed high levels of knowledge of personal protective behaviors (PPBs) among the sample and an apparent association between perceived susceptibility and PPB use as well as between age and PPB use and education and pesticide use (Eichler, 2011). The Eichler study was broad in that it sampled a population of adults of all ages without focusing on those adults at highest risk of severe WNV disease. In contrast, this study specifically examined beliefs, perceptions, and behaviors of high-risk adults over 60 years old in Maryland. Only seven of the above studies specifically incorporated use of the HBM to study WNV PPBs in different samples/populations. Table 2 below summarizes those studies in greater detail, and includes the HBM constructs examined, study findings, strengths and weaknesses, and application to public health practice. 53 Table 2. Summary of Studies Using the HBM to Examine WNV Personal Protective Behaviors Title Authors Population Study design Constructs Findings Strengths & Weaknesses Practice Application Pre-WNV Outbreak: Perceptions and Practices Herringt on, 2003 Nationwide sample of 1500 adults; oversample of 250 residents in NE states Cross- sectional; nationally representati ve survey Perceived susceptibilit y; perceived severity Perceived susceptibilit y to repellent toxicity outweighed vulnerabilit y to disease Nationally representative sample; self- report data; cross-sectional design (lack of temporality) Development of future longitudinal studies of WNV behavior in MD residents WNV in British Columbia Aquino et al., 2004 Adult residents of British Columbia Cross- sectional; telephone survey Perceived susceptibilit y, severity, barriers, benefits, cues to action Perceived hazard of DEET, time to drain standing water are major barriers Tailored questionnaire designed for study; self- report data; no info on non- respondents Basis for design of educational interventions to promote practice of 4 D’s in Maryland What Seniors Say About WNV: The Threat is Not Over Bitto et al., 2005 Adults ≥ 50 yrs old Cross- sectional survey and qualitative evaluation of educational intervention Perceived susceptibilit y, severity, barriers, self- efficacy, cues to action Most seniors motivated to use PPBs; Lack of self- efficacy and perceived barriers were common among seniors who had problems using insect repellent Epidemiologic triad as conceptual model; qualitative design Basis for development of interventions to increase personal protective behaviors among adults > 50 yo and older The Validation of a WNV Survey Based on the Health Belief Model Yerby, 2007 Women ≥19 yo in Alabama Qualitative Perceived susceptibilit y, severity, benefits, barriers, cues to action, self- efficacy Modified survey instrument validated; all constructs except Perceived Barriers to Repellent use predicted personal protective behavior First validated instrument for measuring WNV preventive behaviors; Convenience sample; limited to Southeastern US; Model for validation study of MD WNV Barriers Study instrument Knowledge, Perceptions, and Practices: Mosquito- borne Disease Transmission in Southwest Virginia Butterwo rth, 2009 Adult residents of two counties in Southwest Virginia Cross- sectional; in-person survey; short survey administere d in-person at local post offices Perceived susceptibilit y, severity, benefits, and barriers Gender predicted knowledge; perceived hazards of repellent major barrier; removing standing Strengths: Surveys administered during height of mosquito season Weaknesses: Social desirability bias during survey Provide useful guide for local health departments under financial constraints, because findings 54 Title Authors Population Study design Constructs Findings Strengths & Weaknesses Practice Application water best way to avoid mosquito bites administration; recall bias emphasize individual preventative actions Public Health Responses to WNV: The Role of Risk Perceptions and Behavioral Uncertainty in Risk Communicati on and Policy Elmieh, 2009 Adult Canadian residents Cross sectional; online, mail, and telephone surveys of laypeople and health professional s Perceived susceptibilit y, severity, benefits, barriers, cues to action, self- efficacy WNV risk perception varied with age, gender, and household location; laypeople more willing to trade-off risk than experts Actionable recommendatio ns made according to type of planned intervention; Lengthy survey instrument; exploratory design Guide for development of risk communicati on to target risk groups Public Attitudes, Knowledge and Practices on West Nile Virus Eichler, 2011 Adult residents of Delaware County, PA Cross- sectional telephone survey Perceived susceptibilit y, severity, benefits, barriers, cues to action, self- efficacy WNV knowledge and concern did not predict PPB practice; older age predicted increased PPB use Anonymous, self- administered survey; wide variety of venues for survey administration; convenience sample; predominantly female sample Guide for revision of WNV educational literature to address specific knowledge gaps, particularly among high- risk adults 2.5.1 Strengths and Weaknesses of Existing Studies of WNV Protective Behavior In summary, the above studies have consisted largely of cross-sectional study designs, along with large representative samples, randomized sampling procedures, and examination of a range of individual PPBs. Researchers captured knowledge, attitudes, and practices both before and after WNV outbreaks in the U.S. and Canada and some analyzed community-wide responses to targeted public education campaigns. Still, others conducted qualitative focus group sessions to understand perceptions of vulnerable populations (adults over 50) and to generate content for future interventions. Some have revealed a sometimes alarming lack of motivation or willingness to engage in key WNV protective behaviors, particularly among the group that needs it most (older adults). The findings suggest that use of a theoretical framework may have value at predicting the 55 perceptions and behaviors of such high-risk individuals. Some weaknesses cited among the existing studies of WNV behavior have included use of self-reported data, recall bias, selection bias, and social desirability bias. 2.6 Conclusions Collectively, the above research studies highlight some primary themes regarding engagement in WNV prevention behaviors. Specifically, although knowledge and awareness of WNV are often high, particularly among persons living in areas that have experienced large WNV epidemics, practice of PPBs remains low. This is often attributable to concern over potential harm or toxicity from insect repellent use (especially for vulnerable groups such as pregnant women and children), which often outweighs perceived susceptibility to the virus. Likewise, significant age disparities have been noted, with individuals over 50 years of age and at highest risk of severe disease often being less likely than their younger counterparts to engage in WNV PPBs. All of these findings warrant further study and none have been studied in Maryland residents, whose WNV knowledge, attitudes and behaviors may differ from those in other geographic areas previously studied. Although previous studies have described the effectiveness of the HBM in explaining behaviors relevant to WNV prevention, such a behavioral analysis has not been previously conducted among Maryland residents over 60 years of age. Accordingly, using survey data gathered via the Maryland WNV Barriers Study the Maryland Department of Health and Mental Hygiene (DHMH), this study assessed the HBM as a theoretical framework to describe the knowledge, attitudes, and behaviors of Maryland residents 60 years of age and older regarding WNV disease. Given the gaps and needs 56 highlighted in the above literature review, the application of secondary DHMH study data to an analysis of the effectiveness of the HBM in predicting WNV protective behavior among Marylanders identified important areas for targeted WNV prevention interventions among Maryland residents at least 60 years of age. The findings from this study will ultimately aid Maryland state health officials in the design and implementation of future educational interventions and could have broad applications in future theory- based research on WNV behavior. 57 CHAPTER 3: METHODOLOGY 3.1 Introduction This dissertation presented a secondary data analysis of a cross-sectional study conducted by the Maryland Department of Health and Mental Hygiene (DHMH) in 2012. The study used telephone interviews to examine the knowledge, attitudes, and behaviors of a sample of Maryland adults 60 years of age and older toward West Nile Virus (WNV) prevention. The Health Belief Model (HBM) served as the theoretical framework for development of the survey instrument used in the study. This dissertation explored the utility of the HBM for predicting older adults’ WNV prevention behavior. This chapter describes the study population, sampling and recruitment methods, survey development, survey implementation, theoretical basis, and statistical analysis plan for the dissertation study. 3.2 Population Description For the cross-sectional study, adults aged 60 years old and older and living independently in Maryland were targeted as the sampling frame. Although national recommendations for West Nile Virus prevention focus on adults above 50 years old, who have been identified by the Centers for Disease Control and Prevention (CDC) as being at highest risk for severe WNV disease (CDC Factsheet, 2008), for purposes of this study, individuals aged 60 years old and older were selected. People aged 60 years and over are more likely to have retired from work and to spend time at home or outdoors engaged in leisure-time activities that might bring them in contact with mosquitoes. They may also have greater financial limitations due to a fixed income that limits their ability to support local mosquito control programs. 58 Accordingly, this group was expected to be more accessible for telephone survey administration and was considered an important target for examination of attitudes and behaviors toward WNV. Prospective subjects were sampled from Maryland counties in which at least two confirmed or probable human WNV cases had been reported during the previous six years. Case record review of human WNV cases reported in Maryland from 2006-2011 identified 11 zip codes from counties in which two or more confirmed or probable human WNV cases were reported to ArboNet, the national CDC database for electronic reporting of arboviral diseases. The CDC National Notifiable Diseases Surveillance System established the following case definitions for human WNV cases (CDC NNDSS, 2011): A confirmed WNV case is one that meets the clinical criteria for neuroinvasive or non-neuroinvasive disease along with specific laboratory criteria including virus isolation or a four-fold change in titer. A probable WNV case is one that meets the clinical criteria for neuroinvasive or non-neuroinvasive disease along with specific laboratory criteria consisting of virus-specific antibodies present in cerebrospinal fluid (CSF) or serum. The study sample consisted of Maryland residents ≥60 years of age residing in zip codes in counties meeting the criteria. This study area comprises the following 11 zip codes: 20902, 20910, 21014, 21060, 21122, 21212, 21214, 21215, 21222, 21224, and 21228, which are found in Anne Arundel County, Baltimore City, Baltimore County, Harford County, and Montgomery County. Populations of residents 60 years old and older by age breakdown are shown in Table 3. 59 Table 3. Population of Maryland Residents ≥ 60 Years Old by Zip Code and Age Category Zip Code/ Age 20902 (n) 20910 (n) 21014 (n) 21060 (n) 21122 (n) 21212 (n) 21214 (n) 21215 (n) 21222 (n) 21224 (n) 21228 (n) 60 to 64 2,325 1,998 2,145 1,627 3,465 1,908 1,142 3,542 3,115 1,966 2,882 65 to 69 1,607 1,216 1,511 1,193 2,531 1,321 730 2,771 2,299 1,451 2,016 70 to 74 1,187 805 1,060 936 1,610 923 458 2,291 1,847 1,079 1,496 75 to 80 932 598 976 770 1,133 715 359 1,961 1,644 980 1,515 80 to 84 784 489 809 566 798 617 319 1,565 1,303 778 1,650 85 & over 831 541 873 480 673 590 347 1,504 1,149 906 2,368 Source: US Census 2010 Eligibility and exclusion criteria for study participation are listed as follows: Eligibility criteria  Age ≥60 years old  Living independently (single family home, apartment, retirement community, etc.) Exclusion criteria  Residing in long-term care facility, nursing home, or institution  Non-English speaking It was important to ensure that only healthy adults over 60 years who were ambulatory and living independently, without assistance, would be eligible to participate in the study. Exclusion criteria were chosen accordingly and also excluded anyone who did not speak English, as translation services were not available to conduct the telephone interviews in other languages. Sample Size Determination The research hypotheses for this study are listed in Table 4. 60 Table 4. Research Question and Hypotheses Research Question Is the Health Belief Model a useful theoretical framework for predicting perceptions and behavior toward West Nile virus prevention among Maryland adults 60 years of age and older? 1. Perceived susceptibility to WNV disease Individuals with high perceived susceptibility to WNV infection are more likely to avoid going outdoors during mosquito feeding hours of dusk and dawn than individuals with low perceived susceptibility. Individuals with high perceived susceptibility to WNV infection are more likely to dress in long-sleeved shirts and long pants to avoid mosquito bites than individuals with low perceived susceptibility. Individuals with high perceived susceptibility to WNV infection are more likely to have used insect repellent in the last 90 days than individuals with low perceived susceptibility. Individuals with high perceived susceptibility to WNV infection are more likely to drain standing water from objects around their property that collect water than individuals with low perceived susceptibility. Individuals with high perceived susceptibility to WNV infection are more likely to accept a human WNV vaccine if one were available than individuals with low perceived susceptibility. Individuals with high perceived susceptibility to WNV infection are more likely to support mosquito control programs in their community than individuals with low perceived susceptibility. 2. Perceived severity of WNV disease Individuals who perceive WNV as a severe disease are more likely to avoid going outdoors during mosquito feeding hours of dusk and dawn. Individuals who perceive WNV as a severe disease are more likely to dress in long- sleeved shirts and long pants to avoid mosquito bites. Individuals who perceive WNV as a severe disease are more likely to have used insect repellent in the last 90 days. Individuals who perceive WNV as a severe disease are more likely to drain standing water from objects around their property that collect water. Individuals who perceive WNV as a severe disease are more likely to accept a human WNV vaccine if one were available. Individuals who perceive WNV as a severe disease are more likely to support mosquito control programs in their community. 3. Perceived benefits of practicing WNV protective behaviors Individuals who perceive benefits in practicing WNV protective behaviors are more likely to avoid going outdoors during mosquito feeding hours of dusk and dawn. Individuals who perceive benefits in practicing WNV protective behaviors are more likely to dress in long-sleeved shirts and long pants to avoid mosquito bites. Individuals who perceive benefits in practicing WNV protective behaviors are more likely to have used insect repellent in the last 90 days. Individuals who perceive benefits in practicing WNV protective behaviors are more likely to drain standing water from objects around their property that collect water. Individuals who perceive benefits in practicing WNV protective behaviors are more likely to accept a human WNV vaccine if one were available. Individuals who perceive benefits in practicing WNV protective behaviors are more likely to support mosquito control programs in their community. 61 4. Perceived barriers to practicing WNV protective behaviors Individuals with high perceived barriers to practicing WNV protective behaviors are less likely to avoid going outdoors during mosquito feeding hours of dusk and dawn than individuals with low perceived barriers. Individuals with high perceived barriers to practicing WNV protective behaviors are less likely to dress in long-sleeved shirts and long pants to avoid mosquito bites than individuals with low perceived barriers. Individuals with high perceived barriers to practicing WNV protective behaviors are less likely to have used insect repellent in the last 90 days than individuals with low perceived barriers. Individuals with high perceived barriers to practicing WNV protective behaviors are less likely to drain standing water from objects around their property that collect water than individuals with low perceived barriers. Individuals with high perceived barriers to practicing WNV protective behaviors are less likely to accept a human WNV vaccine if one were available than individuals with low perceived barriers. Individuals with high perceived barriers to practicing WNV protective behaviors are less likely to support mosquito control programs in their community than individuals with low perceived barriers. 5. Cues to action for preventing WNV infection Individuals who receive cues to action for preventing WNV infection are more likely to avoid going outdoors during mosquito feeding hours of dusk and dawn than individuals who do not receive cues to action. Individuals who receive cues to action for preventing WNV infection are more likely to dress in long-sleeved shirts and long pants to avoid mosquito bites than individuals who do not receive cues to action. Individuals who receive cues to action for preventing WNV infection are more likely to have used insect repellent in the last 90 days than individuals who do not receive cues to action. Individuals who receive cues to action for preventing WNV infection are more likely to drain standing water from objects around their property that collect water who do not receive cues to action. Individuals who receive cues to action for preventing WNV infection are more likely to accept a human WNV vaccine if one were available than those who do not receive cues to action. Individuals who receive cues to action for preventing WNV infection are more likely to support mosquito control programs in their community than those who do not receive cues to action. 6. Self-efficacy for preventing WNV infection Individuals with high self-efficacy for preventing WNV infection are more likely to avoid going outdoors during mosquito feeding hours of dusk and dawn than individuals with low self-efficacy. Individuals with high self-efficacy for preventing WNV infection are more likely to dress in long-sleeved shirts and long pants to avoid mosquito bites than individuals with low self-efficacy. Individuals with high self-efficacy for preventing WNV infection are more likely to have used insect repellent in the last 90 days than individuals with low self-efficacy. Individuals with high self-efficacy for preventing WNV infection are more likely to drain standing water from objects around their property that collect water than 62 individuals with low self-efficacy. Individuals with high self-efficacy for preventing WNV infection are more likely to accept a human WNV vaccine if one were available than individuals with low self- efficacy. Individuals with high self-efficacy for preventing WNV infection are more likely to support mosquito control programs in their community than individuals with low self-efficacy. Conditional probability refers to the probability that an event will occur given the occurrence of another. In this study, the probability of a person being willing to accept a WNV vaccine, given his/her perception that WNV is a serious disease, was used as the basis for determining sample size. In previous published studies, perceived severity has not been found to be a significant predictor of most personal protective behaviors or attitudes toward WNV (Aquino et al., 2004; Herrington, 2004). Therefore, to calculate the sample size in this study, hypothesis 11, the probability that an individual who believes WNV can cause serious disease (an indication of perceived severity) will accept a WNV vaccine was assumed to have the lowest conditional probability. For all other hypotheses, the probabilities of the behavioral outcome occurring were higher and thus required smaller sample sizes. It was important to determine a sample size that would be sufficient to detect statistically significant differences in the relationship between HBM predictors and selected WNV attitudes and behaviors among study participants. Although prior studies of the association between HBM constructs and WNV knowledge, attitudes, and behavior have not considered acceptance of WNV vaccine as an outcome, they have yielded findings of interest concerning other behavioral outcomes, which were used as a guide for this sample size determination. In her 2007 dissertation, Yerby found that individuals with a greater perceived severity score were 24% more likely to use DEET than those with a lower score (2007). This was the only instance in 63 which perceived severity was determined to be a significant predictor of WNV behavior. Yerby also found that participants who perceived WNV as a serious disease had an odds ratio of 1.239 (p= 0.026) of engaging in at least one personal protective behavior (PPB). In their study of WNV knowledge, attitudes, and behaviors, Adams et al. (2003) found that Connecticut residents aged 65 years of age and older had nearly three times higher odds of always using at least one PPB compared to their younger counterparts: OR=2.6 (95%CI: 1.7-3.9). They also determined that individuals who had a high perceived susceptibility to WNV infection (those who were “very worried” about getting WNV) had two times higher odds of always practicing at least one PPB compared to those who were never worried [OR=2.2 (95%CI: 1.2-3.9)]. These odds ratios were believed to approximate the risk ratios (relative risk) of performing specific PPBs given perceived susceptibility and were also used as a basis for sample size calculation. Based on these previously published findings, the sample size for this study was calculated incorporating the following parameters:  The sampling frame (N): the number of Maryland adults aged 60 years old and older meeting eligibility criteria; N=34,000,  Confidence interval: 95%  Significance level: 0.05  Power: 0.80 Null hypothesis: There is no difference in support for WNV vaccine between those individuals with high perceived severity of WNV disease and those with low perceived severity. 64 Alternative hypothesis: Adults with higher perceived severity of disease are more likely to accept a WNV vaccine than those who do not perceive the disease as serious. The Type I error rate α was adjusted to hold for the primary hypotheses of interest and the Bonferroni adjustment was set to α=0.0167 to guard against Type I error associated with multiple comparisons while maintaining the desired alpha level of 5%. Further, the desired power level was assumed to be 80% and that the hypothesis of interest was one-sided (p1/p2>1). Based on the above, the following sample sizes were computed: Relative Risk Ratio Total Sample Size (n1=n2) Type I Error (α) 1.5 484 0.0167 2.0 128 0.0172 Because we expected to detect a relative risk of 2.0 based on the risk ratio identified in the Adams study, the target number of potential enrollees to enroll was 128 for the three primary hypotheses, as indicated in the table above. However, because we also wished to examine the effect of additional variables, the target sample size was increased (doubled) to 256 households. This larger sample size would allow us to better examine the hypotheses within subsets of the study population, i.e. by race, ethnic group, income level, or geographic region. 65 3.3 Sampling Procedures Figure 3 below shows a flowchart of the sampling procedures and data collection methods used for the study. Figure 3. Diagram of Sampling and Data Collection Procedures Sampling Frame Potential enrollees were identified through a list of household telephone numbers obtained from Sales Genie, Inc. (www.salesgenie.com). Sales Genie, Inc. is a marketing Sampling frame: 34,000 households with adults ≥ 60 yo in designated zip codes Stratified random sample (n=1,700 households) Letters with reply postcards mailed to 1,700 potential enrollees 155 postcards returned Phone calls to 106 persons who agreed to be called 75 completed questionnaire 26 Refused 5 Ineligible 49 declined to participate 1,545 did not return postcards Phone calls initiated to 1,545 after 2 weeks 455 people not called due to mid-December study termination 1,090 calls made 136 completed questionnaire 954 Refused or ineligible 66 company that provides marketing and data support services to business clients. The company supplies contact lists for marketing, research, and other projects through the use of data collected from various sources, including census records, voter registration files, deed and assessment property records, and business and consumer Yellow pages. DHMH staff purchased contact information for adults aged 60 years and older living independently in the zip codes identified above. For each zip code, the call list included households that fit the age and housing criteria and provided the following information: (1) Head of household; (2) Address; (3) County; (4) Phone number; and (5) Age. The list provided by Sales Genie, Inc. contained 34,000 records, which served as the sampling frame. Because telephone surveys often yield low response rates due to refusals or unavailability of potential enrollees, it was determined that study recruitment materials inviting prospective participants to enroll in the study should be sent to 1,700 households to increase the probability of achieving the targeted sample size. Recruitment Process Recruitment materials included an introductory letter describing the study and a postage-paid reply postcard with which prospective enrollees could indicate their willingness to be called We assigned a unique identification (ID) number to all prospective enrollees. The ID number was used to track whether individuals returned the reply postcards and was also used to generate a list of non-respondents. The study proposal, survey instrument, invitation letter, and related documents were approved by the DHMH Institutional Review Board in September 2012. Minor modifications to the instrument were submitted to and approved by the DHMH IRB in September 2012 67 following the original approval. The introductory letter and survey instrument are provided in Appendix A and Appendix C respectively. Stratified Random Sample Epidemiologists at DHMH applied stratified random sampling methods (households were stratified by zip code) using a computer-generated random number generator to generate a random sample of 1,700 households. As described in the sample size determination above, we aimed to recruit 256 participants from this sample of 1,700 households. Survey Administration Survey administration took place between October and December 2012. One individual was sampled per randomly selected household. The person to be surveyed was randomly selected by calling and asking for the household member who met the inclusion criteria and who had the most recent birthday. This was done to achieve random selection within each household, as in some instances the same person in a household routinely answers the phone, creating an internal bias. If the person with the most recent birthday was not at home, arrangements were made to call that person during a time that was more convenient to him/her. No other member of the household was interviewed as a proxy, if attempts to recontact the potential enrollee were unsuccessful. This approach aimed to avoid a within-household bias introduced by enrollment of an individual who did not meet the specific criteria. Every effort was made to call each household a total of six times, at varying times of day, depending on whether it was a weekday or weekend. Criteria that constitute valid attempts at phone contact are described in the Data Collection section. 68 Households Called and Enrolled As described above, households that were randomly selected but in which the potential enrollee was unavailable, did not answer, or declined to participate were not replaced. It was anticipated that a sample of 256 adult Maryland residents > 60 years old would be required to achieve a suitable effect size for detecting significant differences in respondents’ reported personal protective behaviors according to their risk perceptions and self-efficacy. However, due to time constraints imposed by the approaching winter holiday season (and the likelihood that prospective enrollees would be unavailable or unable to recall their summer activities) survey administration had to be concluded in mid-December, prior to reaching the enrollment goal. The actual survey administration yielded a sample of 211 enrollees. Introductory Letter Each potential enrollee was mailed an introductory letter on official DHMH letterhead, introducing the study, before being contacted by telephone. Letters were mailed to the 1,700 households generated through the stratified random sampling procedures described above. See Appendix A for the introductory letter. A postage-paid response postcard was enclosed with each letter, allowing potential enrollees the opportunity to refuse to participate prior to any attempt at telephone contact. If a potential participant indicated on the returned postcard a refusal to participate, his/her phone number was removed from the active call list. Potential participants could also return the postcard indicating a desire to participate by providing their phone number and the days and times that were convenient for them to receive calls. 69 If the potential participant did not respond to the postcard within two weeks, study personnel were instructed to accept this lack of response as permission to contact and proceeded to contact the participant for an interview. See Appendix B for the response postcard. 3.4 Theoretical Framework The Health Belief Model (HBM) was created for the purpose of explaining an individual’s engagement in a particular health behavior and provides a theoretical framework for describing individual health behaviors according to perceptions of the risks and benefits associated with the recommended behavior. Developed in the 1950s by social scientists at the U.S. Public Health Service, the model was originally designed to understand the lack of participation in free TB screening programs and lack of compliance with public health recommendations (Hochbaum, 1958; Rosenstock, 1960, 1974; Carpenter, 2010). An outgrowth of stimulus- response theory and cognitive theory, the HBM was ultimately conceived as a value- expectancy theory, combining elements of both reinforcement-reward system as well as subjective personal judgments. As such, the HBM asserts that an individual seeks to avoid illness and that the person will engage in specific behaviors that enable him/her to avoid becoming sick (Carpenter, 2010). The original HBM consisted of four core constructs: 1) perceived susceptibility (a person’s belief that s/he is vulnerable to a certain disease/illness); 2) perceived severity (belief that the illness or disease, is serious and, if contracted, would have serious consequences illness); 3) perceived benefits (one’s belief about the positive outcomes associated with actions taken to reduce susceptibility to or severity of a disease or 70 illness); and,4) perceived barriers (potential negative outcomes associated with engaging in a particular health action to reduce the threat of disease). As research about the HBM evolved, cues to actions (internal or external stimuli that activate an individual’s readiness to prevent disease) and self-efficacy were added. Self-efficacy (confidence in ability to perform the recommended health action), was added to the model in 1988. Since its initial applications by the USPHS, the HBM has since been used as the basis for numerous interventions including mammography screening, AIDS prevention, and medication adherence (Carpenter, 2010; Champion & Skinner, 2008). Figure 4 presents a conceptual model of the HBM as it is applied to this WNV prevention study. Individual Beliefs Action Modifying Factors Diagram adapted from Champion and Skinner, in Glanz 2008. Figure 4. Conceptual Model of Health Belief Model Applied to WNV Prevention Perceived susceptibility to WNV Perceived severity of WNV Perceived benefits of WNV protective behaviors Perceived barriers to WNV protective behaviors Cues to action Info on WNV from media or other sources Knowing someone who got sick or died of WNV WNV Personal Protective Behaviors Drain standing water Dress in long sleeved shirts and long pants Use insect repellent Avoid being outdoors during dusk and dawn Perceived self-efficacy to perform WNV personal protective behaviors Demographic Characteristics Age Gender Race Ethnicity Education Employment status Marital status Household Income WNV Knowledge Transmission Risk groups Attitudes & Perceptions Support for WNV vaccine Support for mosquito control programs 71 3.5 Study Design The study was a cross-sectional design supported by the Emerging Infections Program of DHMH with funding from the Centers for Disease Control and Prevention Arboviral Disease group. It involved design and implementation of a telephone survey on a random sample of Maryland residents. The study protocol involved conceptualization of interview questions, training of interviewers, and development and pilot testing of the survey instrument on a small sample of Office of Infectious Disease Epidemiology and Outbreak Response staff members. Survey Instrument A 36-item questionnaire was developed expressly for use in the DHMH study. It was a new instrument adapted from the content of survey questionnaires used in previous studies of WNV behavior (Yerby, 2007; Aquino, 2004; Tuiten, 2009). Survey items were designed to capture respondents’ knowledge, attitudes, and behavioral practices related to WNV prevention and were framed around similar items in WNV behavior studies by Aquino et al. (2004) and Tuiten et al. (2009). The survey instrument consisted of Likert scale and Likert-type items, yes/no questions, and open-ended questions. Following development of the survey instrument, it was pilot tested with staff members in the Office of Infectious Disease Epidemiology and Outbreak Response, as they were a readily available convenience sample. This group of staff members was used in order to identify any challenges with survey administration. An expert reviewer from the Emerging Infections Program at DHMH with survey development experience also reviewed the pilot instrument and provided feedback, which was used to revise the 72 instrument. Epidemiology staff who expressed interest in participating in survey administration were trained in interviewing techniques. Those trained interviewers administered the questionnaire via telephone interview to consenting enrollees from October through mid-December 2012. The survey instrument included a detailed phone script which preceded the set of survey questions (see Appendix B). Interviewers were instructed to first ascertain that the respondent was at least 60 years of age, to then read a description of the survey, and if the respondent consented, to verify his/her eligibility by asking for the respondent’s zip code and type of residence. Once it was established that the individual met the eligibility criteria and that s/he agreed to participate in the study, the interviewer commenced with survey administration over the phone. This consisted of the interviewer reading each question along with the corresponding response items. The respondent did not have a paper copy available to read, but could ask the interviewer to repeat items as needed. The survey took an average of 19 minutes to complete. Table 5 presents a list of demographic variables and HBM variables, along with corresponding survey questions. The survey instrument captured each respondent’s knowledge, perceptions, and behaviors relating to WNV and their use of arbovirus prevention and control measures, including personal protective behaviors and source reduction behaviors. It also collected the following demographic information: type of dwelling, age, sex, race, ethnicity, geographic area of residence, primary language spoken in household, marital status, employment status, household income level, and education level. In addition, information was collected on the age and sex of participants 60 years of age and over living at the same address in order to indicate the representativeness of 73 those surveyed relative to other household members. Survey responses were recorded on a paper version of the questionnaire and later entered into a secure database. A list of variables measured, their operational definitions, corresponding survey questions, measurement level, and role in analysis are presented in Table 5 below. 74 Table 5. Summary of Variables, Operational Definitions, and Survey Questions Variable Operational Definition Survey Question Measurement Level Role Modifying factors: Demographic Characteristics Type of Dwelling Type of housing in which respondent resides In what type of housing do you live?  Single-family detached home  Townhouse or condominium  Apartment  Active living senior community  Other housing Nominal Independent variable (IV) Age Self-reported age in years (minimum age must be 60 years old) Please tell me the age and gender of yourself and others in the household who are ≥ 60 years old. Ratio Control Gender Self-reported gender (male or female) Please list your gender Male/Female Nominal IV Race Self-reported race or ancestry How would you describe your race?  White or Caucasian  Black or African American  Asian  Native Hawaiian or Other Pacific Islander  Native American/Alaska Native Nominal Control Area of residence Self-reported geographic area of residence Is your home located: In a City/In the Suburbs/In a Rural area? Nominal IV Ethnicity Self-reported as being of Hispanic or non-Hispanic origin Are you Hispanic or Latino or of Spanish origin? Yes/No Nominal IV Language Self-reported primary language spoken at home What is the primary language spoken in your home? English/Spanish/Other Nominal IV Education Self-reported highest year or level of education completed by respondent What is the highest level of education you have completed? -Some high school -HS diploma or GED -Some college, include Associates -Bachelor’s degree -Some graduate school -Graduate or professional degree Ordinal IV 75 Variable Operational Definition Survey Question Measurement Level Role Income Self-reported annual household income Which category best describes your total household annual income?  <$20,000  $20,001-$30,0000  $30,001-$40,0000  $40,001-$50,0000  $50,001-$60,0000  $60,001-$70,0000  >$70,000 Interval IV Marital status Self-reported marital status (married, single, divorced, separated, widowed, domestic partnership) What is your marital status?  Single  Married  Separated  Divorced  Widowed Nominal IV Employment status Self-reported employment status What is your employment status?  Full-time  Part-time  Retired  Unemployed  Other Nominal IV Modifying factors: Knowledge of West Nile Virus Transmission of WNV Knowledge of the vector that transmits WNV to humans How do you think people get WNV?  Eating or drinking contaminated food or water  From bug bites  From birds  Contact with sick people Nominal Control Risk groups for severe disease Knowledge of age groups at greatest risk of severe neurologic disease from WNV What age group or groups do you think are most likely to get seriously ill with WNV?  Young children  Adolescents 11-18  Young adults 19-25  Adults 26-50  Adults > 50 years old Ordinal IV Independent variables: Health Belief Model constructs Perceived susceptibility Individual’s perception of chance that s/he will contract WNV disease How worried are you that you might get sick with WNV? Not at all /Little/Somewhat/Very How likely do you think it is that you will get WNV in next calendar year? Not/Somewhat/Moderate/ Extremely Ordinal IV 76 Variable Operational Definition Survey Question Measurement Level Role Perceived severity Individual’s perception of the seriousness of WNV disease if s/he were to contract it Do you think WNV can cause serious illness? No/Yes Nominal IV Perceived benefits Individual’s belief in the efficacy of the recommended behavior to reduce the risk of contracting WNV? Would you be interested in a community program to help adults over 60 years old repair their damaged window screens and dump standing water in their yard? No/Yes Would you be in favor of a mosquito control program No/Yes How much would you be willing to pay for a WNV vaccine? Nominal IV Perceived barriers Individual’s perceived obstacles to engaging in WNV preventive behaviors Can you tell me why you do not drain water from items on your property that collect water?  Takes too much effort  Do not have anyone to help me  Too dangerous  There is nothing in my yard that collects water  I do not own my home Nominal IV Self-efficacy Individual’s feelings of confidence in his/her ability to protect him/herself from contracting WNV How confident are you that you can protect yourself and your household members from getting WNV? Not at all confident/Somewhat confident/Very confident Ordinal IV Cues to action Self-reported prompts that are likely to stimulate the respondent to engage in behaviors that prevent WNV Do you know anyone who has had WNV? No/Yes Have you ever received information, in any form, about WNV? No/Yes Nominal IV Dependent Variables (DV): Attitudes and Perceptions Support for WNV vaccine Self-reported willingness to accept a human WNV vaccine if one were available If a vaccine were available that was safe and effective, would you be willing to take it? No/Yes Nominal Dependent variable (DV) 77 Variable Operational Definition Survey Question Measurement Level Role Support for mosquito control programs Self-reported willingness to support mosquito control efforts Are or would you be in favor of a mosquito control program like that? No/Yes Nominal DV Dependent Variables: Personal Protective Behaviors Use of insect repellent in last 90 days Self-reported frequency of insect repellent use In the last 90 days, did you always, sometimes, rarely, or never use insect repellent on your skin when you went outdoors? Always/Sometimes/Rarely/ Never Ordinal DV Draining of standing water Self-reported frequency of draining standing water around the home Since the start of this past summer (2012), did someone you asked or hired drain water from items around the outside of your home? No/Yes Nominal DV Dressing in long clothing Self-reported frequency of dressing in long- sleeved and long pants to avoid mosquito bites When you go outdoors in the summer, do you wear long- sleeved shirts and/or long pants? No/Yes Nominal DV Avoid outdoors during mosquito feeding hours Self-reported frequency of avoiding outdoors during mosquito feeding hours (dusk & dawn) When you go outdoors is it usually: At dusk or dawn In the middle of the day Nominal DV As noted above, the survey questionnaire included questions to determine respondents’ general awareness and knowledge of WNV and arboviral disease risk, as well as specific HBM constructs, including perceived susceptibility, perceived severity (perceptions of arboviral disease severity), perceived threat, perceived benefits of practicing personal protective behaviors, perceived barriers, cues to action, and self- efficacy (confidence in their ability to follow public health recommendations, such as eliminating potential mosquito breeding sites, etc.). 78 Other Survey Questions The survey instrument included several additional questions that were not part of the research hypotheses. These included items assessing respondent’s general knowledge about the disease (risk factors, transmission, vector); awareness of mosquito feeding times (dusk and dawn), if they go out during peak mosquito hours, and type of clothing worn outdoors during peak mosquito hours. To assess knowledge of WNV transmission, respondents were asked “How do you think most people get West Nile virus?”; to assess knowledge of WNV risk groups, respondents were asked “What age group do you think is most likely to get seriously ill from West Nile virus?” In addition, they were asked about specific habits/behaviors to protect their home from mosquitoes. Specifically, they were asked “How do you keep your home cool in the summer?” (Fans/Air conditioning/Open windows/Other); “If open windows, do all of the windows in your home have screens? (yes/no)”and “If yes, Are your screens in good condition, with no holes or tears? (yes/no).” In addition to the knowledge questions, survey respondents were also asked about basic demographic information. These questions represented modifying factors other than those included in the hypothesis testing, that may influence individual’s beliefs about WNV, and included respondents’ marital status, employment status, ethnicity, primary language, and geographic location. Participants were also asked about their attitudes and awareness toward specific WNV prevention measures, such as their opinions about tax support of mosquito control programs and about creation of programs to help older adults drain standing water and repair damaged window screens. 79 3.6 Data Collection For those potential enrollees who indicated a willingness to be contacted (either by indicating that on the postcard or by not returning it within two weeks), study personnel attempted to make phone contact. A series of six attempts were made to reach each potential enrollee by phone. These calls took place at varying times of day on multiple days of the week, including evenings and weekends. A valid attempt at phone contact was counted if any of the following results were obtained:  A potential enrollee answers the phone and consents to participate; or  No one is home and either a message is left or no message is left; or  Someone is home and that person does not speak English, s/he refuses to talk or participate, or requests that study personnel call back at another time. These criteria were established because it was recognized that not every household would have voicemail or other means to record a message when no one was home. Likewise, prospective enrollees who did not speak English were excluded because no translators were available to be trained as interviewers. Calls that resulted in an individual’s refusal to participate in the study were also considered valid enrollment attempts because contact had been made with a member of the household who met study criteria. When study personnel contacted a potential enrollee, they verbally provided them with information regarding the potential risks and benefits associated with participation in the study, an explanation of the study rationale and study procedures, and a description of measures that would be taken to ensure security and confidentiality of participant 80 information (see Appendix D for Disclosure Statement). Verbal consent was obtained over the phone in lieu of written informed consent, as the research presented no more than minimal risk of harm to subjects and involved no procedures for which written consent was required outside of the research content. Eight volunteer interviewers from the DHMH Center for Zoonotic and Vector- borne Diseases (CZVBD) and the Emerging Infections Program (EIP) administered the telephone interviews. All interviewers were trained in standardized interviewing techniques by the author, who was a co-Principal Investigator prior to survey administration. Each interviewer was assigned lists of names and phone numbers obtained from the original stratified random sample of 1700 records for survey calls. After obtaining verbal consent, a member of the research study team conducted a telephone interview for all individuals meeting the eligibility criteria. The interview consisted of a standardized telephone-based interview, conducted in English. A script was used for interviewing all subjects and was read at the start of each interview to disclose risks and benefits and to ensure consistency (see phone script in Appendix B). The survey took between 15 and 20 minutes to administer (average=19 minutes). Respondents who completed the survey were offered the option to receive an Information/Disclosure Sheet, detailing the risks and benefits associated with study participation. No incentives were offered for study participation, but general information about WNV was made available by mail or e-mail upon request. These WNV materials were not automatically sent to all study participants because many indicated a preference not to receive such literature. 81 Response Rate A total of 1,191 households were called and of those 1,102 were determined to be eligible to participate in the study (89 were ineligible). A total of 211 individuals were enrolled in the study, yielding a response rate of [19] %. Given that survey administration began in October 2012, time limitations due to the approaching holiday season in December 2012 prevented interviewers from calling the entire sample of 1,700 households. Demographic Data Table 5 provides a comparison of selected demographic characteristics of participants and non-participants. The original sampling frame contained only the following demographic information: Gender, Name, Street Address, City, State, Zip Code, and County/Jurisdiction. More detailed demographic information was only collected for study participants after enrollment. Results of chi-square analysis on associations between the modifying factors and dependent variables are shown below in Table 6. Enrollment status differed significantly by gender (X2=23.99, p<0.001) and also by geographic area of residence or jurisdiction (X2=21.18, p<0.001). 82 Table 6. Comparison of Study Participants and Non-participants Demographic Characteristics* Participants (n=211) Non-participants (n=891) X2 p-value Geographic Area of Residence Anne Arundel Baltimore City Baltimore County Harford County Montgomery County n (%) 53 (16) 42 (15) 39 (34) 28 (19) 49 (22) n (%) 277 (84) 232 (85) 77 (66) 118 (81) 175 (78) 21.18 <0.001 Gender† Male Female n (%) 99 (46.9) 108 (51) n (%) 589 (62) 302 (38) 23.99 <0.001 *Pearson’s Chi-square test was used for nominal variables. †Four respondents declined to specify a gender. The racial breakdown of the study participants was consistent with that of human WNV cases in Maryland and the U.S. Of the 47 human WNV cases in MD in 2012, 81% (n=38) were White and 8.5% (n=4) were Black (DHMH, Final Summary of WNV Results, 2012). This was similar to the racial distribution of the sample: 79% White and 13% Black. Similarly, for human WNV cases reported to the CDC from 1999-2008, the majority of cases for which race was reported were White (Lindsey, Staples, Lehman, & Fischer, 2010). The gender breakdown was also consistent between study participants and Maryland human WNV cases. Among the MD human WNV cases in 2012, 64% were male and 46% female, comparable to those listed for participants in the table above. 3.7 Variables and Measurement This study examined associations between six HBM variables (perceived susceptibility, perceived severity, perceived benefits, perceived barriers, cues to action, and self-efficacy), attitudes toward WNV prevention, and practice of specific WNV preventive measures. It also examined the effects of selected demographic variables (gender, education, employment status, and marital status,), two socioeconomic variables 83 (annual household income and education level) and WNV knowledge. Although the effects of these additional variables were not included in the research hypotheses, their relationship to the dependent variables was of value. Items used in the study instrument included Likert-type and Likert scale items as well as binary response questions and open-ended questions. 3.7.1 Independent Variables: Health Belief Model Constructs Most variables addressing the HBM constructs were measured via Likert-type response items, while others were measured using dichotomous scales (see Table 4). 1. Perceived susceptibility Two questions measured respondents’ perceived susceptibility to WNV (their perception of personal vulnerability to the disease). The first question asked “How worried are you that you might get sick with West Nile virus?” and was measured on a 1 to 4 scale, with 1 = Not at all worried, 2=A little worried, 3=Somewhat worried, 4= Very worried. Higher scores indicated a higher level of perceived susceptibility. For statistical analysis, this variable was dichotomized as Worried/Not worried, with Not worried coded as zero and A little, Somewhat, and Very worried collectively labeled “Worried” and coded as 1. This dichotomized variable was measured on a 0 to 1 scale. The second question asked “On a 1 to 5 scale, with 1 being “not at all likely” and 5 being “extremely likely” how likely do you think it is that you will get West Nile virus in the next calendar year?” Response items were coded as 1=Not at all likely, 2=Not very likely, 3=Moderately likely, 4=Very likely, and 5=Extremely likely. Higher scores 84 indicated a higher level of perceived susceptibility. Because survey administration occurred in late fall/early winter of 2012, that year’s summer WNV season had already concluded. Therefore, the next calendar year was used as the timeframe for this question because it was too late for respondents to comment on their likelihood getting the disease during the current year. 2. Perceived severity The question “Do you think West Nile virus is a serious disease?” was used to measure respondents’ perceived severity of WNV and contained a binary (No/Yes) response option measured on a 0 to 1 scale where 0= No and 1=Yes. A higher score indicated a higher level of perceived severity. 3. Perceived benefits Perceived benefits (positive outcomes that an individual associates with performing the recommended behavior) were measured via two questions that attempted to identify benefits that respondents’ associated with certain WNV protective behaviors. One question was measured using a binary response option, while the other used a Likert- type scale. The first question asked “Would you be interested in a community program to help adults over 60 years old repair their damaged window screens and dump standing water in their yard? (No=0/Yes=1). This question was part of a skip pattern, and was only answered by those individuals who had previously answered “open windows” to a multiple-choice question asking how they keep their home cool (response options were A/C, fans, open windows, or a combination). As a result, only a small subset of the total 85 number of respondents (n=76) provided answers to the question about their support for a program to help them repair damaged window screens. A second question asked “how much would you be willing to pay out of pocket for the (human WNV) vaccine?” Range: 1=$0, 2=<$25, 3=$25-$49, 4=$50-$74, 5=≥$75. For purposes of analysis, this variable was dichotomized into two categories with $50 as the cutpoint, with 0=≥$50 and 1=$0-$49. 4. Perceived barriers The perceived barriers construct was represented by two survey questions that attempted to capture respondents’ perceptions of barriers to two common WNV PPBs: draining of standing water and use of insect repellent, and were based on qualitative responses to survey questions about why respondents did not practice those behaviors. A respondent’s perceived barriers (impediments which they believe prevent them from carrying out recommended WNV prevention measures) were measured in the context of specific WNV personal protective behaviors. Respondents were first asked the following questions about whether or not they performed certain WNV protective behaviors: If they responded “no” to those questions, they were provided with a list of possible reasons for not engaging in the behavior and asked to indicate their reasons for not performing the behavior. These response options included a list of common reasons for not performing the behavior along with an open-ended Other category in which respondents could provide their own answer. Although not a part of the dissertation research study, responses to these open-ended questions were examined in an informal qualitative analysis to identify common themes that might account for Maryland residents’ lack of engagement in certain PPBs. The themes identified were classified according to 86 respondents’ Professed Knowledge, Perceptions of immunity to WNV, Sense of Inevitability (of WNV Disease), and Beliefs about Religion or Fate. If a respondent answered “no” to the question “Since the start of this past summer (2012), did you or someone you asked or hired drain water from items around the outside of your home (such as gutters, buckets, flower pots, kiddie pools, bird baths, or discarded tires)?” s/he was then asked to indicate why or why not using a list of qualitative response items that included “It takes too much effort/I do not have anyone to help me/It is too dangerous/I use products to keep mosquitoes from breeding in my yard/There is nothing that collects water in my yard/I do not own my home/Other.” The qualitative responses to this question were assigned to categories of Personal, Environmental, or Financial reason. As the vast majority of responses fell into the Personal reason category, the response categories were collapsed, with binary response options of Personal/Not personal and these were operationalized as one component of perceived barriers. Responses to this question were included as the independent variable in a series of multivariate logistic regression models to examine hypotheses that test the predictive utility of perceived barriers at explaining respondents’ practice of PPBs. Similarly, in a separate question that asked “In the last 90 days, did you Always, Sometimes, Rarely, or Never use insect repellent on your skin when you went outside,” respondents who answered “rarely” or “never” were asked to indicate why they seldom or never use repellent using the following response items: “I often forget/It feels sticky/It smells bad/It will make me feel sick/It is too expensive/I didn’t know it could help/Other.” The majority of responses fell into the Other category, which consisted of unique qualitative responses not among the response items listed above. These qualitative 87 responses were assigned to categories of Personal, Environmental, or Financial reason. As the vast majority of responses fell into the Personal reason category, the response categories were collapsed, with binary response options of Personal/Not personal, and were measured on a 0 to 1 scale where 0= Not personal and 1=Personal. 5. Cues to Action Prior research on the cues to action construct of the HBM has suggested that cues can sometimes take the form of external events, such as a friend or acquaintance getting sick or dying from a disease or from a health care provider providing information or intervention about the disease (Mattson, 1999; Michel et al., 2010). Accordingly, in this survey, cues to action (triggers that activate a person’s readiness to take action and engage in WNV protective behaviors) were measured using two questions that illustrated such events. The first was “Do you know anyone who has had West Nile virus?” and included a Yes/No response option (No=0, Yes=1). This item was categorized as a cue to action because awareness of other individuals infected with WNV disease could prompt respondents to learn more about WNV and start taking precautions for themselves. The second question was “In the last year, have you ever received information in any form about WNV?” and also used a Yes/No response option (No=0, Yes=1). This question also included a supplemental sub-question for respondents who answered “Yes,” asking them to indicate the source of that information; respondents were asked to indicate whether the information came from Doctor/Radio/TV/Newspaper/Internet/E-mail/Mail/Word of Mouth/Other. This variable was measured on a binary scale with No=0 and Yes=1, and 88 these binary response items were applied to each of the different types of information. Possible scores range from 0 to 1. 6. Self-efficacy A person’s self-efficacy (respondent’s confidence in his/her ability to take the necessary steps to prevent WNV) was measured via a single Likert-type question. The question was “How confident are you that you can protect yourself and your household members from getting WNV?” and was measured on a three-point scale as 1= Not at all confident, 2=Somewhat confident, and 3= Very confident. This variable was dichotomized as Not confident/Somewhat or Very confident for bivariate analysis, and measured as 0=Not confident, and 1=Confident. A higher score indicated a higher self- efficacy for performing specific WNV PPBs. Table 7 below shows how each of the six HBM constructs was operationalized for hypothesis testing. 89 Table 7. Operationalization of Hypotheses Independent variables: Health Belief Model Constructs Construct Hypotheses Survey question(s) Variable construction Range Perceived susceptibility Hypotheses 1-6: Individual’s perception of worry about and likelihood that s/he will contract WNV disease How worried are you that you might get sick with WNV? Worried/Not worried How likely do you think it is that you will get WNV in next calendar year? Not likely /Somewhat likely/ Moderately likely/ Extremely Dichotomized variable (Worry) and 4-point Likert- type scale variable (Likely get WNV) Range 0-1for dichotomized worry scale; 1-4 for likelihood scale Perceived severity Hypotheses 7-12: Perception that WNV serious Do you think WNV can cause serious illness? No/Yes Single item (binary) 0 - 1 Perceived benefits Hypotheses 13-18: Individual’s belief in the benefits of engaging in recommended behaviors to reduce the risk of contracting WNV Would you be interested in a community program to help adults over 60 years old repair their damaged window screens and dump standing water in their yard? No/Yes How much would you pay for a WNV vaccine? $0, <$25, $25- $49, $50-$74, ≥$75 Single items (non- composite) 0 – 1 for each item Perceived barriers Hypotheses 19-24: Individual’s perceived obstacles to engaging in WNV preventive behaviors Why you have rarely or never used insect repellent on your skin in the last 90 days?  I often forget  It feels sticky  It smells bad  It will make me sick  It is too expensive  I did not know it helps  Other Why do you not drain water from items on your property that collect water?  Takes too much effort  No one to help me  Too dangerous  Nothing in yard collects water  I do not own my home  Other Single items (non- composite (barriers to insect repellent use and barriers to draining standing water) 0-1 for each item Cues to action Hypotheses 25-30: Self-reported prompts that stimulate PPB Do you know anyone who has had WNV? No/Yes Have you ever received information, in any form, about WNV? No/Yes Single items (non- composite) 0 – 1 for each item Self-efficacy Hypotheses 31-36: Feelings of confidence in ability to protect self from WNV How confident are you that you can protect yourself and your household members from getting WNV? Not confident/Confident Single (non- composite) 0-1 90 3.7.2 Modifying Factors The research question for this study asked: Is the Health Belief Model a useful theoretical framework for predicting perceptions and behavior toward West Nile virus prevention among Maryland adults 60 years of age and older? The hypotheses are shown in Table 4. Selected variables, namely sociodemographic variables and knowledge variables, were classified as modifying factors solely according to their definition in the conceptual model which, as described by Glanz (2008), suggests that such factors may have an indirect effect on individual’s engagement in a behavior (Glanz et al., 2008). As such, these factors played a key role in answering the research question. These variables were not used, however, as modifiers from a statistical standpoint in analysis. That is, interaction terms were not included in any of the analytic models and models did not examine the effect modification of the above factors on the direct relationship between the HBM constructs and the dependent variables. 3.7.3 Control Variables Three variables were selected for use as controls during multivariate logistic regression analysis. These variables were age, race, and knowledge of WNV transmission. Age: Age, at the time of the interview, was recorded in years for each survey participant. All participants were 60 years of age or older. The survey instrument asked each respondent to indicate his/her age in years. Because the sole age of interest was age ≥ 60 years, as specified by the eligibility criteria, age was initially measured as a continuous variable. However, for purposes of comparison during analyses, age was 91 divided into the following categories: 60-69 years old, 70-79 years old, 80-89 years old, and 90-99 years. These categories were then dichotomized as 60-69 years old and 70 years old and above. Race: This variable was based on self-reported information about racial group affiliation and was measured as a categorical variable. Respondents were asked to identify the racial group with which they most closely identify. The variable response structure was modeled after that of the U.S. Census (ref). Response categories were 1=White or Caucasian, 2=Black or African American, 3=Asian, 4=Native Hawaiian or Other Pacific Islander, 5=American Indian/Alaska Native, 6=Mixed race or Other. For analysis, this variable was dichotomized as White/Non-White. Knowledge of WNV transmission: Subjects’ knowledge of WNV transmission was assessed via the question “How do you think people get West Nile Virus?”—response options for this question included both the correct response (from insect bites, such as mosquito bites) and several incorrect responses (Eating or drinking contaminated food or water/From birds/Contact with sick people/Other). These response items were scored as follows: Bug bites=4, Eating or drinking=3, Birds=2, Sick people=1. Those who scored highest (by selecting value 4) were classified as having greater WNV knowledge than those who chose the other selections. 3.7.4 Other Demographic Factors Gender: Subjects’ gender was measured as a dichotomous nominal variable, where 1=male and 2=female. 92 Education: Educational level was measured according to the number of years of education completed by the respondent. The variable consisted of six categories: (1) some high school, (2) high school diploma or GED, (3) some college, (4) Bachelor’s degree, (5) some graduate school, and (6) graduate or professional degree. The variable was collapsed into two categories: Less than College and College degree or above. The variable was dichotomized in this manner because the majority of the sample had educational levels that exceed those of the US population: 51% of study respondents hold Bachelor’s degrees or higher, compared to only 28.5% of the entire US population, according to the US Census 2008-2012 (Census Bureau). Annual household income level: Participants’ income level was measured as an interval variable with seven categories, each denoting a range of annual household incomes. Response categories were categorized as follows: 1=<$20,000, 2=$20,001- $30,000, 3=$30,001-$40,000, 4=$40,001-$50,000, 5=$50,001-$60,000, 6=$60,001- $70,000, 7=>$70,000. For statistical analysis, the income variable was collapsed into two categories of < $70,000 and ≥ $70,000. This was done because among the study respondents who provided income information, the majority (74%) had income levels below $70,000, while 26% had incomes over $70,000. Area of residence (urban/rural/suburban): this question asked if the respondent’s home was located in the city, suburbs, or a rural area and was measured as a categorical variable. Specifically, it was coded as City=1, Suburbs=2, Rural=3. Ethnicity (Hispanic origin): This variable captured whether the respondent was of Hispanic origin. This was a nominal variable that asked whether the respondent was of 93 Hispanic or Latino or Spanish origin and had a dichotomous response of No=0 and Yes=1. Primary language spoken at home: This nominal variable asked what language is spoken in the respondent’s household. It was measured on a three-point scale with responses English=1, Spanish=2, and Other = 3. Marital status: This item asked respondents to indicate their marital status as Single, Married, Divorced, Separated, Widowed, or in a Domestic Partnership. Scores on this item ranged from 1 to 6, where 1=Single, 2=Married, 3=Divorced, 4=Separated, 5=Widowed, and 6=Domestic partnership. 3.7.5 Knowledge Although knowledge is not a component of the original HBM, the model proposed by Champion and Skinner in Glanz, included it as a modifying factor along with several sociodemographic factors (Glanz, 2008). See Figure 4 for a schematic of the HBM as it is applied to WNV prevention. For this study, survey items that assessed respondents’ knowledge of WNV disease were felt to be useful for providing additional context for respondents’ beliefs and behaviors. These knowledge variables were included in statistical analysis as confounding variables. For the survey question below, interviewers were instructed to read all response items and to indicate to the respondent whether to select one item or more response options, depending on the question. 94 In addition to the knowledge question that was used as a control variable above, the survey also included another knowledge question that sought to capture respondents’ understanding of WNV risk groups, but was not included as a control in statistical analysis. Respondents were asked the following question: “What age groups do you think are most likely to get seriously ill from West Nile Virus?” Response options for this question included Young children (0-10 years old)/Adolescents and teenagers (11-18 years old)/Young adults (18-25 years old)/Adults (26-50 years old)/Adults over 50 years old. Responses were Young children=1, Adolescents and teens=2, Young adults (18-25 yo)=3, Adults (26-50 years old)=4, Adults over 50 years old=5, and Combination/more than one group=6. Only the next-to-last response (adults > 50 years old) was correct, however respondents were instructed that they could provide more than one answer and in some instances they responded with two or more age groups, and sometimes included the correct response along with additional age group selections. A respondent’s knowledge level was determined based on whether s/he could correctly name the type of vector that transmits WNV (insect or bug) in the transmission knowledge question above and could correctly identify the single age group at highest risk of severe WNV disease (adults over 50 years old) in this question. 3.7.6 Dependent Variables Six dependent variables were used as outcomes for testing of the research hypotheses. These variables were: 1. Avoiding the outdoors during mosquito feeding hours (dusk and dawn) Respondents were asked “When you go outdoors is it usually: At dusk or dawn; 95 In the middle of the day? The question had a binary response item of Dusk or dawn=1 or Middle of day=2. It was analyzed using a series of multivariate logistic regression models to test all six HBM constructs to determine if any were significantly associated with avoiding the outdoors at dusk and dawn. 2. Dressing in long clothing (long-sleeved shirts and long pants) to avoid mosquito bites Respondents were asked “When you go outdoors during the summer, do you wear long-sleeved shirts and/or long pants?” The question had a binary response item of No=0 or Yes=1. It was examined in a series of logistic regression models that tested the six HBM constructs against each of the six outcomes to determine if any were significantly associated with dressing in protective clothing when outdoors in the summer. 3. Use of insect repellent in last 90 days Participants were asked “In the last 90 days, did you always, sometimes, rarely, or never use insect repellent on your skin when you went outside?” This was originally a Likert-type question with response categories of Never=1, Rarely=2, Sometimes=3, and Always=4. It was dichotomized as used repellent: 0=No (never or rarely used repellent) and 1=Yes (sometimes or always used repellent). A series of multivariate logistic regression models were constructed to test this outcome against each of the six HBM constructs. 4. Draining of standing water Respondents were asked “In the last year (2012), did you or someone you asked or hired drain standing water from objects around your home (such as gutters, 96 buckets, kiddie pools, bird baths, or discarded tires)?” This question had a binary response item of No=0 or Yes=1. A series of logistic regression models were used to examine the association between this dichotomous variable and the six HBM constructs to determine if any were significantly associated with draining standing water. This outcome was not applicable to all households in the sample, particularly for those respondents who lived in apartment complexes. However, the majority of respondents resided in single family homes. 5. Willingness to accept a WNV human vaccine, if one was available. This was a dichotomous response question with a Yes/No response option where No=0, Yes=1. This variable was included in a series of logistic regression models for testing each of the six HBM constructs determine if they were significantly associated with acceptance of a WNV vaccine. Potential confounding factors such as race and WNV knowledge were added to the model along with control variables. 6. Support for community mosquito control programs Respondents were asked if they would be in favor of a government mosquito control program that used pesticides to reduce mosquitoes in their community This question had a binary response option of No=0, Yes=1. A series of logistic regression models were used to test each of the six HBM constructs to determine if any were significantly associated with support for mosquito control programs. 97 3.8 Data Analysis Plan All statistical tests were performed using STATA v. 12.1 with a level of significance of alpha =.05, unless otherwise noted. All variables were examined for normality to ensure appropriateness of the statistical test selections. Descriptive Analysis Descriptive analyses were performed to describe demographic characteristics of the sample and to examine the frequency of self-reported protective behaviors and attitudes toward WNV. This included calculation of means, range, medians and standard deviations for age, the only continuous variable. For nominal and ordinal variables, frequencies and percentages were calculated to describe the sample. Bivariate Analysis Selected bivariate analyses were used according to the type of variables being analyzed. For categorical variables with two or more levels, chi-square tests were used to examine differences between modifiers, independent variables, and dependent variables. Specifically, Pearson’s chi-square tests examined associations between demographic variables, including gender (male/female), race, education, annual household income, employment status, and marital status, and the following dependent variables: avoiding the outdoors during mosquito feeding hours, dressing in long-sleeved shirts and long pants, having used insect repellent in the last 90 days, draining items containing standing water, willingness to accept a WNV vaccine, and support for mosquito control programs. Chi-square tests were also used to examine associations between the above six outcomes and the six HBM constructs. Respondent age, race, and WNV transmission knowledge 98 were selected as control variables in multivariate analysis, according to their significance in bivariate analysis. Spearman’s rank correlations were performed via correlation matrix to identify any issues of multicollinearity and to determine the strength of relationships between the independent variables prior to multiple logistic regression analysis. This helped ensure that none of the predictors were too highly correlated with one another, which may have skewed regression results in the multivariate analysis. Regression Analysis Multivariate logistic regression analysis was used to examine associations between the HBM constructs, demographic modifying factors, and the six dependent variables. Specifically, logistic regression analysis was performed to test associations between each of the six Health Belief Model constructs and the six outcomes: (1) avoidance of outdoors at dusk and dawn, (2) dressing in long-sleeved shirts and long pants, (3) having used insect repellent in the last 90 days, (4) draining standing water from around the property, (5) willingness to accept a WNV vaccine, and (6) support for mosquito control programs. Stepwise regression model entry techniques were applied and specified a threshold of p<0.20 to determine which variables from the bivariate analysis were entered into each multivariate logistic regression model. This model building process involved application of a forward selection procedure followed by backward elimination to ensure consistency among results. Control variables, demographic and knowledge variables, and HBM variables meeting that minimum p- value criterion were entered into logistic regression models to determine their association 99 with dichotomous outcomes pertaining to individual attitudes toward WNV prevention and protective behaviors. For all six dependent variables, the utility of the HBM constructs of perceived severity, perceived susceptibility, benefits, barriers, and self-efficacy in predicting the designated outcomes of practicing or supporting at least one personal protective behavior were examined. A minimal amount of missing data was identified in the survey records, and a complete case method was used to address missing data. This method, also known as listwise deletion, involves discarding cases for which there is not complete data. 3.9 Testing of Hypotheses Research Question: Is the Health Belief Model a useful theoretical framework for predicting perceptions and behavior toward West Nile virus prevention among Maryland adults 60 years of age and older? Hypotheses The 36 research hypotheses tested in this dissertation study were framed according to four individual personal protective behaviors (PPBs) and two attitude outcomes: willingness to accept a WNV vaccine and support for community mosquito control programs. The rationale for examining each PPB individually rather than as a composite variable was based on the previous WNV research literature, which suggested a benefit to considering such outcomes on an individual basis. While a couple of studies have examined PPBs as a single composite variable (McCarthy et al., 2001; Adams et al., 2003), several others have demonstrated the value of testing each PPB on its own (Herrington, 2003; Wilson et al., 2005; Bitto et al., 2005; Yerby, 2007), which allows 100 researchers to pinpoint specific activities (e.g. draining standing water and insect repellent use) for development of targeted interventions in the future. Multivariate logistic regression models were used to test the 36 study hypotheses and followed the guidance of Hosmer & Lemeshow (2000). The model building process involved use of forward selection procedures that were repeated via backward elimination to ensure consistency of results for each model. Per Hosmer & Lemeshow, a threshold was set for entry and removal of variables from each model; in this instance the selected threshold was p<0.20. Hypotheses 1-6 tested the perceived susceptibility construct as the primary predictor for each of the six outcomes. 1. Perceived susceptibility to WNV infection is positively associated with all six outcomes (avoidance of the outdoors, dressing in long shirts and pants, use of insect repellent in last 90 days, draining of standing water, acceptance of a WNV vaccine, and support for mosquito control programs). Hypotheses 1-6 were tested using a series of logistic regression models with perceived susceptibility (in the form of two categorical variables) as the independent variable and each of the six binary outcomes as the dependent variables. Three control variables (age, race, and WNV transmission knowledge) and confounders were entered into the regression models according to significance in bivariate analysis and model entry and removal criteria (p<0.20). 2. Perceived severity of WNV infection is positively associated with all six outcomes (avoidance of the outdoors, dressing in long shirts and pants, use of insect repellent in last 90 days, draining of standing water, acceptance of a WNV vaccine, and support for mosquito control programs). 101 Hypotheses 7-12 were tested using a series of six multiple logistic regression models with perceived severity (binary variable) as the independent variable and each of the six binary outcomes as the dependent variables. Both forward and backward stepwise selection procedures were applied. The three control variables and selected confounding variables were entered into the model according to their significance in bivariate analysis and according to model entry and removal criteria (p<0.20), to examine their effect on the association between perceived severity and the six outcomes. 3. Perceived barriers to WNV prevention are negatively associated with all six outcomes (avoidance of the outdoors, dressing in long shirts and pants, use of insect repellent in last 90 days, draining of standing water, acceptance of a WNV vaccine, and support for mosquito control programs). Hypotheses 13-18 were tested using a series of logistic regression models with the two perceived barrier variables (both binary) as the independent variables and the six binary outcomes as the dependent variables. An inverse relationship was predicted such that the higher perceived barriers, the lower the frequency of engaging in WNV PPBs. Both forward and backward selection methods were used. The three control variables and selected confounding variables were entered into the model according to their significance in bivariate analysis and model entry and removal criteria (p<0.20). 4. Perceived benefits of practicing PPBs are positively associated with all six outcomes (avoidance of the outdoors, dressing in long shirts and pants, use of insect repellent in last 90 days, draining of standing water, acceptance of a WNV 102 vaccine, and support for mosquito control programs). Hypotheses 19-24 were tested using a series of logistic regression models with forward and backward selection using the two perceived benefits questions as the independent variables (binary variables) and the six binary outcomes as the dependent variables. The three control variables and selected confounders were entered into the model according to their significance in bivariate analysis and model entry and removal criteria (p<0.20). 5. Cues to action are positively associated with all six outcomes (avoidance of the outdoors, dressing in long shirts and pants, use of insect repellent in last 90 days, draining of standing water, acceptance of a WNV vaccine, and support for mosquito control programs). Hypotheses 25-30 were tested using a series of logistic regression models, with the two binary cues to action variables as the independent variables and the six binary outcomes as the dependent variables. The three control variables and selected confounding variables were entered into the models according to their significance in bivariate analysis and model entry and removal criteria (p<0.20). 6. Self-efficacy is positively associated with all six outcomes (avoidance of the outdoors, dressing in long shirts and pants, use of insect repellent in last 90 days, draining of standing water, acceptance of a WNV vaccine, and support for mosquito control programs). Respondents whose responses indicated that they had a high level of confidence in their ability to protect themselves and/or their families from WNV infection were more likely to have used insect repellent in the last 90 days than those with a 103 low level of confidence. Therefore, a higher self-efficacy score was expected to be associated with more frequent insect repellent use. Hypotheses 31-36 were tested using a series of logistic regression models with the dichotomized self- efficacy variable as the independent variable and the six binary outcomes as the dependent variables. The three control variables and selected confounding variables were entered into the model according to their significance in bivariate analysis and model entry and removal criteria (p<0.20). 3.10 Data Management Frequencies were calculated for each variable to identify missing data and any data errors. A codebook was developed that assigned numeric values to categories of all categorical variables. This process minimized the likelihood of errors in data entry. A quality control review was performed in which an epidemiologist who had not been involved in instrument development or survey administration reviewed every fifth record, comparing responses on the paper survey form to those entered in the database. Security and Confidentiality All documents containing personal information were kept confidential and secured in a locked file cabinet at the Maryland Department of Health and Mental Hygiene (DHMH). Data from paper survey forms were initially entered into an Excel spreadsheet and subsequently imported into STATA for analysis. STATA v. 12.1 was used for both data storage and data analysis. Data files were password protected. 104 Human Subject Research Approval The parent study was approved by the Institutional Review Board of the Maryland Department of Health and Mental Hygiene in September 2012. This dissertation study was approved by the University of Maryland Institutional Review Board on February 28, 2014. The approval letter is shown in Appendix E. 105 CHAPTER 4: RESULTS 4.1 Introduction The purpose of this study was to examine the utility of the HBM for predicting WNV personal protective behaviors (PPBs) among a sample of adults 60 years of age and older in Maryland. This chapter presents results of statistical analyses that described the 211 enrollees and tested hypotheses about associations between the HBM constructs and WNV personal protective behaviors. Univariate analyses were performed to describe the sample and to summarize respondents’ outcome behaviors (PPBs), HBM constructs, and WNV knowledge. Bivariate analysis included Spearman’s rank correlations and chi-square tests, and were used to examine associations between HBM constructs and demographic variables and between modifying factors and dependent variables. Finally, regression models were used to test the research hypotheses, examining the predictive capacity of each HBM construct to predict the designated outcomes. 4.2 Description of Sample The study consisted of 211 Maryland residents aged 60 years old and older who completed a survey that captured their knowledge, attitudes, and practices regarding WNV. All participants lived independently and were sampled from zip codes with two or more probable or confirmed WNV cases in the previous five years. 4.2.1 Demographic Characteristics Table 8 below lists the frequencies and percentages for various demographic characteristics of the study sample. The sample consisted of nearly equivalent numbers of males (49%) and females (51%) and respondents ranged in age from 60 to 99 years old 106 (mean age 70 years; SD=8.22), with a little more than half (55%) falling between 60 and 69 years old. The majority of participants were White (79%) and almost exclusively of non-Hispanic White origin (98%). Respondents primarily speak English at home (98%). Nearly 20% of respondents hold a Bachelor’s degree and 32% have a graduate or professional degree. The majority of participants were either married or in a domestic partnership (57%). Most respondents were also retired (72%) and just under half reported an annual household income less than $70,000 (48%), while 26% reported incomes over $70,000. Table 8. Demographic Characteristics Demographic Variable Frequency Percentage Gender Male Female Refused 99 108 4 46.9 51.2 1.9 Age (Mean 70 years; SD=8.22; Median 69; Range: 60-99) 60 – 69 years 70 – 79 years 80 – 89 years 90 – 99 years Refused 112 60 25 6 8 53.1 28.4 11.8 2.8 3.8 Race White Black Asian Pacific Islander Mixed race Refused 167 28 3 1 4 8 79.1 13.3 1.4 0.5 1.9 3.8 Hispanic ethnicity Yes No Refused 1 205 4 0.5 97.6 1.9 Primary language English Other: Filipino Refused 207 3 1 98.1 1.4 0.5 Education Some high school High school/GED Some college Bachelor’s Some graduate school Graduate or Professional Refused 8 41 46 41 1 67 7 3.8 19.4 21.8 19.4 0.5 31.7 3.3 Marital status Single Married Separated Divorced Widowed 20 119 1 17 44 9.5 56.4 0.5 8.1 20.9 107 Demographic Variable Frequency Percentage Domestic partnership Refused 1 9 0.4 4.2 Employment status Full-time Part-time Retired Unemployed Other: Disabled Refused 28 19 151 6 2 5 13.3 9 71.5 2.8 1 2.4 Annual household income ≤$20,000 $20,001-$30,000 $30,001-$40,000 $40,001-$50,000 $50,001-$60,000 $60,001-$70,000 >$70,000 13 21 18 17 18 16 55 6.2 9.9 8.5 8.1 8.5 7.6 26.1 Home Ownership Yes No Refused 196 12 3 92.9 5.7 1.4 Time in residence Mean: 28.8 SD: 15.06 Range: 2 - 75 Home Region City Suburbs Rural 55 145 11 26.1 68.7 5.2 Housing Type Single family home Townhouse or condo Senior community Other 174 30 2 5 82.5 14.2 1.0 2.3 Zip codes 20902 20910 21014 21060 21122 21212 21214 21215 21224 21222 21228 27 22 28 21 35 22 8 17 4 12 15 12.8 10.4 13.3 9.9 16.6 10.4 3.8 8.1 1.9 5.7 7.1 Most respondents own their home (93%) and have lived at their current residence an average of 28 years (SD=15.06; range 2 - 75 years). The majority reside in a suburban (69%) or urban area (26%) and live in single family homes (83%) that they own (93%). Respondents are distributed across the 11 study zip codes, with a range of residents per zip code of 4 to 35. The highest percentage of study participants (17%) live in the 21122 108 zip code (Anne Arundel County). The lowest percentage of participants hailed from the 21224 zip code (Baltimore City). 4.2.2 Behavioral Outcomes Frequencies and percentages of engaging in WNV PPBs are shown in Table 9. Just over half of all respondents (51%) reported avoiding the outdoors during prime mosquito feeding hours of dusk and dawn, and 56% dress in long-sleeved shirts and long pants when outdoors. The majority of respondents (52%) indicated that, in the previous 90 days, they never used insect repellent on their skin when outdoors, while only 28% always or sometimes used it. Sixty-two % of participants routinely drain standing water from objects around their homes. Table 9. Personal Protective Behavior Outcomes Personal Protective Behavior (Frequency) % of respondents Avoid going outdoors at dusk and dawn Yes No (105) 51.2 (100) 48.8 Dress in long-sleeved shirts & long pants Yes No (115) 55.6 (92) 44.4 Use insect repellent in last 90 days Always Sometimes Rarely Never (9) 4.3 (51) 24.3 (40) 19 (110) 52.4 Drain standing water from items around home Yes No (129) 62.3 (77) 37.2 Study participants’ attitudes toward WNV prevention efforts are presented in Table 10. These attitudes include willingness to accept a WNV vaccine, acceptance of a community mosquito control program, and being in favor of tax support for mosquito control activities. Most respondents (~70%) indicated willingness to accept a WNV vaccine if one was available, and a majority (56%) would not pay more than $25 for it. In addition, 109 84% of participants expressed support for community mosquito control programs. Among those who stated they would be in favor of mosquito control programs, 86% favored using tax revenues to fund these programs. Table 10. Attitudes Toward WNV Prevention Programs and Services Attitude (Frequency) % of respondents Willingness to accept WNV vaccine Yes No Maybe DK (146) 69.5 (26) 12.4 (26) 12.4 (12) 5.7 If yes, amount willing to pay for vaccine $0 <$25 $25-49 $50-74 $75-99 >=$100 DK (14) 9.2 (72) 47.1 (34) 22.2 (12) 7.8 (6) 3.9 (11) 7.2 (4) 2.6 In favor of mosquito control program Yes No DK (175) 83.7 (14) 6.7 (20) 9.6 Favor tax support of mosquito control programs Yes No DK Refused (180) 86.1 (17) 8.1 (11) 5.3 (1) 0.5 In favor of a community program to help adults over 60 years old repair damaged window screens and dump standing water in their yard? Yes No DK Refused (41) 19.4 (37) 17.5 (3) 1.4 (130) 61.7 4.2.3 Health Belief Model Constructs This section presents frequencies and percentages for HBM constructs. Perceived Susceptibility Most participants did not consider themselves susceptible to WNV disease, with 34% indicating they were not at all worried , 38% only a little worried about getting sick with WNV, and 84% stating it was not at all likely (53%) or not very likely (31%) they would get WNV disease in the next calendar year. 110 Perceived Severity Nearly all study participants (96%) believed WNV can cause serious disease; however, 4% indicated they did not know if it causes serious illness or not. Perceived benefits Just over half of all respondents (56%) expressed willingness to pay for a WNV vaccine if it cost $25 or less. The remainder of participants were willing to pay varying higher amounts for a vaccine, with approximately 19% of respondents willing to pay $50 or more. The question about participants’ interest in a community program to assist persons over 60 with window repair and draining of standing water was part of a skip pattern: it was asked as a follow-up question after establishing how respondents kept their homes cool in the summer months. As a result, more than half the respondents (61%) did not provide answers to that question; 19% expressed interest in such a program and 18% indicated they would not be interested in such a program. Perceived barriers Survey questions designed to operationalize the perceived barriers construct were open-ended and asked respondents why they did not engage in certain WNV PPBs, namely using insect repellent on exposed skin when outdoors and draining standing water from objects around their homes. Qualitative responses to those questions were subsequently divided into the following categories: Personal, Environmental, and Financial. Responses in the Personal reason category included such statements as “I don’t go outside much,” “I always wear long clothing when outside,” and “I don’t consider mosquitoes a problem (never bitten).” Responses in the Environmental category included “I am not aware of any mosquitoes in the area” and “There is nothing in my 111 yard that collects water.” In general, perceived barriers to practicing recommended PPBs included fear of toxic chemicals from repellents, cost, and a perception that taking steps to reduce mosquitoes, such as using insect repellent and draining standing water, are ineffective. Many respondents (81%) indicated they did not drain standing water on their property because they believed no items on their property collected water (81%). Cues to Action The majority of participants (58%) reported having received some form of information about WNV in the past year. Conversely, 89% indicated they did not know anyone who had ever gotten WNV. Self-efficacy Respondents expressed confidence in their ability to protect themselves and their household members from WNV, with most stating they were somewhat (52%) or very (25%) confident. 112 Table 11. Health Belief Model Constructs Health Belief Model Construct (Frequency) % Perceived Susceptibility How worried are you that you will get sick with WNV? Not at all worried A little worried Somewhat worried Very worried Don’t know (70) 33.8 (79) 38.2 (39) 18.8 (16) 7.7 (3) 1.4 How likely is it that you will get WNV in the next calendar year? Not at all likely Not very likely Moderately likely Very likely Extremely likely (110) 53.4 (63) 30.6 (25) 12.1 (4) 1.9 (2) 1.0 Perceived severity Do you believe WNV is serious disease? Yes Don’t know (199) 96.1 (8) 3.9 Perceived benefits Would you be interested in a community program to help adults over 60 years old repair their damaged window screens and dump standing water in their yard? Yes No Don’t know Refused or N/A (41) 19.4 (37) 17.5 (3) 1.4 (130) 61.7 How much would you be willing to pay for a WNV vaccine? $0 <$25 $25-49 $50-74 $75-99 >=$100 Don’t know (14) 9.2 (72) 47.1 (34) 22.2 (12) 7.8 (6) 3.9 (11) 7.2 (4) 2.6 Perceived barriers If never or rarely used repellent in last 90 days, why not? Often forget Feels sticky on skin Smells bad Did not know it could help Other Sticky and smells Forget, sticky, and smells Don’t know (5) 3.4 (8) 5.4 (1) 0.7 (1) 0.7 (124) 84.3 (1) 0.7 (1) 0.7 (6) 4.1 If do not drain standing water around home, why not? Takes too much effort Nothing collects water in yard Other (1) 1.4 (60) 81.1 (13) 17.5 Cues to Action Received information on WNV in the past year Yes No Don’t know (123) 58.3 (85) 40.3 (3) 1.4 Know anyone who had WNV disease Yes No Don’t know (22) 10.7 (182) 88.8 (1) 0.5 Self-efficacy How confident are you that you can protect yourself and your family from WNV? Not confident Somewhat confident Very confident Don’t know (30) 14.5 (107) 51.7 (52) 25.1 (18) 8.7 113 4.2.4 Knowledge Table 12 describes participants’ responses to the two knowledge questions included in the survey instrument. Table 12. Knowledge of West Nile Virus Risk and Transmission Knowledge of West Nile Virus Frequency (%) How do you think most people get WNV? Eating or drinking Insect bites Birds Sick people Other Don’t know 3 (1.5) 155 (74.9) 8 (3.9) 6 (2.9) 23 (11.1) 12 (5.8) What age group or groups do you think is/are most likely to get seriously ill from WNV? Young children (0-10 yo) Adolescents and teens (11-18 yo) Young adults (19-25 yo) Adults (26-50 yo) Adults >50 yo Young children and adults >50 yo All of the above Don’t know 15 (7.4) 5 (2.5) 3 (1.5) 1 (0.5) 50 (24.6) 74 (36.5) 25 (12.3) 29 (14.3) Correct responses are italicized. As shown in the table, most respondents (75%) correctly answered the question on WNV transmission, responding that people get it from insect (mosquito) bites. Among the 11% of respondents who provided other responses not among those listed in the response categories, their responses included germs, poor ventilation, and being in a crowd. For the question asking which age group or groups are most likely to get seriously ill from WNV, only 25% provided the single correct answer of adults over 50 years of age. An additional 36% indicated both young children and adults over 50 years old are most at risk, while 12 % indicated all of the listed age groups were likely to become seriously ill with WNV infection. 114 4.2.5 Summary This section provided a description of the sample and a summary of responses to survey questions. The results indicate the sample was predominantly White and well- educated with middle to high income levels. Participants were aware of WNV and knowledgeable about how it is transmitted and who is at risk. Likewise, perceptions of confidence in ability to avoid WNV infection were also very high in this sample. The next section presents results of bivariate analyses to test associations between the HBM constructs, demographic characteristics, and study outcomes. 4.3 Bivariate Analysis Bivariate analyses were conducted to examine associations between respondent demographic characteristics, West Nile Virus (WNV) knowledge, and outcome variables as well as between Health Belief Model (HBM) constructs and outcome variables. Correlation analysis was used to examine associations between independent variables to check for collinearity prior to regression analysis. Spearman’s Rank correlation tests were run for the HBM questions that were measured on an ordinal scale (note that many of these variables were collapsed for use in the chi-square tests). The correlation analysis is presented in Table 13 below. None of the variables were highly correlated with one another, indicating that multicollinearity would not be an issue. 115 Table 13. Spearman Rank Correlations How worried about WNV Likely get WNV Think WNV serious Confident protect How much willing pay Barriers to repellent How worried about WNV 1.0000 Likely get WNV 0.4201 (0.001) 1.0000 Think WNV serious -0.0714 (0.52) 0.0210 (0.85) 1.0000 Confident protect against WNV -0.1003 (0.37) -0.2156 (0.05) -0.0191 (0.86) 1.0000 How much willing pay for vaccine -0.0142 (0.89) 0.0247 (0.82) 0.0105 (0.92) -0.1266 (0.25) 1.0000 Barriers to repellent use 0.0699 (0.53) -0.1952 (0.08) -0.0910 (0.41) 0.0363 (0.74) -0.1963 (0.075) 1.0000 For each of the six behavioral outcomes: (1) avoiding the outdoors at dusk and dawn, (2) dressing in long-sleeved shirts and long pants, (3) use of insect repellent in the last 90 days, (4) draining of standing water, (5) willingness to accept a West Nile Virus (WNV) vaccine, and (6) support for community mosquito control programs, two tables of bivariate results are presented. The first table shows Pearson chi-square tests of independence between demographic characteristics, a knowledge variable, and one of the six outcomes. The second table shows Pearson chi-square tests of independence between the HBM constructs and the same outcome. For instances in which a cell count was less than five, Fisher’s Exact test results are reported in lieu of a chi-square test statistic. As per the definition of race defined by the US Census Bureau, individuals’ responses to the questions asking about race and ethnicity on the survey were based solely on self- identification and have no scientific or anthropological basis. 116 Avoiding Outdoors at Dusk and Dawn Table 14 shows chi-square tests of independence for respondent characteristics, knowledge, and avoiding the outdoors during mosquito feeding hours. No statistically significant associations were identified. Table 14. Demographic Characteristics, Knowledge, and Avoiding the Outdoors at Dusk and Dawn Characteristic Yes – % (n) Chi-square p Gender Male Female 51.6 (49) 52.0 (52) 0.004 0.95 Age 60 – 69 years old ≥ 70 years old 56.5 (61) 48.2 (40) 1.29 0.26 Education Less than Bachelor’s Bachelor’s or above 51.1 (45) 51.4 (54) 0.002 0.97 Marital status Single Married or domestic partnership 44.0 (33) 56.9 (66) 3.03 0.08 Race White Non-white 54.0 (87) 41.9 (13) 1.53 0.22 Annual household income ≤ $70,000 > $70,000 47.4 (46) 55.6 (30) 0.92 0.34 Knowledge: How do people get WNV? Correct response: Mosquito bites Incorrect responses 50.7 (76) 56.3 (27) 0.45 0.50 Table 15 presents the results of chi-square tests of independence on associations between HBM constructs and avoidance of the outdoors during mosquito feeding hours (dusk and dawn). A single component of perceived barriers (barriers to draining standing water) did not achieve significance in its association with avoidance of the outdoors at prime mosquito feeding times (p=0.06). That is, individuals who reported barriers to draining standing water around their home (such as physical inability to perform the task or the belief that no objects on the property collected water) were more likely to avoid the outdoors at dusk and dawn (p=0.06). 117 Table 15. HBM Constructs and Avoiding the Outdoors at Dusk and Dawn Construct Yes: % (n) Chi-sq p Perceived susceptibility How worried are you that you might get WNV? Worried Not worried 52.3 (67) 51.4 (36) 0.015 0.90 How likely is it that you will get WNV in the next year? Likely Not likely 44.8 (13) 53.2 (90) 0.70 0.40 Perceived severity: Do you think WNV is a serious illness?* Yes Don’t know 51.3 (98) 71.4 (5) -- 0.45 Perceived benefits Would you be interested in a community program to help repair damaged window screens? Yes No 52.6 (20) 55.8 (19) 0.08 0.78 How much would you be willing to pay for WNV vaccine? ≤ $49 $50 or above 52.2 (59) 63.3 (19) 1.18 0.28 Perceived barriers Why have you not used insect repellent in last 90 days? Personal Non-personal 53.1 (51) 51.9 (14) 0.014 0.91 Why have you not drained standing water from objects around your home? Personal Non-personal 50.9 (29) 83.3 (10) -- 0.06 Cues to Action: Have you received any WNV information in the last year? Yes No 53.3 (64) 50.0 (39) 0.21 0.65 Do you know anyone with WNV? Yes No 68.2 (15) 49.7 (86) 2.67 0.10 Self-efficacy: How confident you can protect yourself from WNV? Confident Not confident 54.0 (81) 51.9 (14) 0.043 0.84 *None of the participants reported a “No” response to the perceived severity question; thus, the only operational categories were “Yes” and “Don’t know”. Dress in Long-Sleeved Shirts and Long Pants Table 16 presents chi-square test results for demographic characteristics, WNV knowledge, and dressing in long-sleeved shirts and long pants. Race /ethnicity was significantly associated with dressing in protective attire while outdoors to avoid mosquito bites (p<0.05). Out of all participating racial groups, whites accounted for the highest proportion of individuals who dress in protective clothing to avoid mosquito bites. Age also showed borderline significance: individuals 60 to 69 years old showed a borderline significant association with dressing in long clothing when outdoors (p=0.05). 118 Table 16. Demographic Characteristics, Knowledge, and Dressing in Long Shirts and Pants Characteristic Yes –% (n) Chi-sq p Gender Male Female 60.4 (58) 52.0 (52) 1.41 0.24 Age 60 – 69 years old ≥ 70 years old 49.5 (53) 63.5 (54) 3.76 0.05 Education Less than Bachelor’s Bachelor’s or above 47.8 (43) 62.5 (65) 4.24 0.04 Marital status Single Married or domestic partnership 61.8 (47) 52.6 (61) 1.60 0.21 Race White Non-white 52.2 (84) 75.0 (24) 5.64 0.02 Annual household income ≤ $70,000 > $70,000 59.2 (58) 51.8 (28) 0.77 0.38 Knowledge: How do people get WNV? Correct response: mosquito bites Incorrect responses 55.6 (84) 57.1 (28) 0.03 0.85 Table 17 presents chi-square independence test results for HBM constructs and the practice of dressing in long-sleeved shirts and long pants to avoid mosquito bites. No statistically significant associations were noted. Even so, a higher proportion of respondents who believed they were not likely to get WNV indicated they routinely dress in long-sleeved shirts and long pants when outdoors. Table 17. HBM Constructs and Dressing in Long Sleeved Shirts and Pants Construct Yes% (n) Chi-sq p Perceived susceptibility How worried are you that you might get WNV? Worried Not worried 54.2 (71) 62.1 (41) 0.50 0.48 How likely is it that you will get WNV in the next year? Likely Not likely 43.3 (13) 58.2 (99) 2.30 0.13 Perceived severity: Do you think WNV is a serious illness?* Yes Don’t know 55.4 (107) 71.4 (5) -- 0.47 Perceived benefits: Would you be interested in program to repair damaged window screens? Yes No 52.5 (21) 67.6 (23) 1.75 0.19 How much would you be willing to pay for WNV vaccine? ≤ $49 >$50 53.5 (61) 53.3 (16) 0.001 0.99 Perceived barriers Why have you not used insect repellent in last 90 days? Personal Non-personal 52.0 (51) 55.6 (15) 0.11 0.75 Why have you not drained standing water from objects around your home? Personal Non-personal 51.8 (29) 71.4 (10) -- 0.24 Cues to Action Have you received any WNV information in the last year? Yes No 55.8 (67) 56.3 (45) 0.003 0.95 Do you know anyone with WNV? Yes No 68.2 (15) 54.9 (96) 1.41 0.24 Self-efficacy: How confident are you that you can protect yourself from WNV? Confident Not confident 55.6 (84) 53.6 (15) 0.41 0.84 119 *None of the participants reported a “No” response to the perceived severity question; thus, the only operational categories were “Yes” and “Don’t know”. Use of Insect Repellent in the Last 90 Days Table 18 presents chi-square tests of independence for respondent demographic characteristics, WNV knowledge, and use of insect repellent in the previous 90 days. Both age of respondent and WNV transmission knowledge showed significant associations with use of insect repellent during that time period (p<0.05). As with support for mosquito control, individuals 60 to 69 years old were more likely to engage in insect repellent use than individuals 70 years old and older (p<0.01). In addition, respondents who correctly identified insect bites as the means by which WNV is spread were more likely to have used insect repellent during the previous 90 days. No other statistically significant associations were noted between respondent characteristics and insect repellent use. The above findings were significant, suggesting a major role for respondent knowledge and age. Table 18. Demographic Characteristics, Knowledge, and Use of Insect Repellent in the Last 90 Days Characteristic Yes – % (n) Chi- square p Gender Male Female 51.0 (49) 42.7 (44) 1.38 0.24 Age 60 – 69 years old ≥ 70 years old 56.4 (61) 35.6 (31) 8.40 0.004 Education Less than Bachelor’s Bachelor’s or above 45.1 (41) 50.0 (53) 0.48 0.49 Marital status Single Married or domestic partnership 41.8 (33) 51.7 (60) 1.86 0.17 Race White Non-white 47.6 (78) 43.8 (14) 0.15 0.69 Annual household income ≤ $70,000 > $70,000 52.0 (51) 48.2 (26) 0.21 0.65 Knowledge: How do people get WNV? Correct response: mosquito bites Incorrect responses 53.3 (81) 31.4 (16) 7.35 0.007 120 Table 19 shows results of chi-square tests of independence for HBM constructs and use of insect repellent in the previous 90 days. Individuals who expressed worry about getting sick with WNV demonstrated a statistically significant association with insect repellent use during that time period. Specifically, persons who reported being a little worried about WNV infection had the highest proportion of those using repellent. This was a significant association and no other significant associations were found between the remaining HBM constructs and repellent use. Table 19. HBM Constructs and Use of Insect Repellent in the Last 90 Days Construct Yes: % (n) Chi-sq p Perceived susceptibility How worried are you that you might get WNV? Worried Not worried 57.3 (75) 31.9 (22) 13.27 <0.001 How likely is it that you will get WNV in the next year? Likely Not likely 54.8 (17) 46.5 (80) 0.73 0.39 Perceived severity: Do you think WNV is a serious illness?* Yes Don’t know 49.0 (96) 14.3 (1) -- 0.12 Perceived benefits: Would you be interested in a program to repair damaged screens? Yes No 51.2 (21) 51.4 (18) 0.0003 0.99 How much would you be willing to pay for WNV vaccine? ≤ $49 $50 or above 51.7 (60) 65.4 (17) 0.23 0.63 Perceived barriers: Why have you not used insect repellent in last 90 days? Personal Non-personal 25.7 (26) 18.5 (5) 0.61 0.44 Why have you not drained standing water around your home? Personal Non-personal 40.7 (24) 14.3 (2) -- 0.06 Cues to Action: Have you received any information about WNV in the last year? Yes No 46.7 (57) 49.4 (40) 0.14 0.71 Do you know anyone with WNV? Yes No 36.4 (8) 49.4 (88) 1.34 0.25 Self-efficacy: How confident are you that you can protect yourself from WNV? Confident Not confident 51.6 (79) 48.3 (14) 0.11 0.74 *None of the participants reported a “No” response to the perceived severity question; thus, the only operational categories were “Yes” and “Don’t know”. Draining Standing Water Table 20 shows results of chi-square tests of independence for respondent characteristics and draining standing water on property. Age and transmission knowledge were significantly linked with the practice of draining standing water from items around one’s property. Specifically, persons aged 60 to 69 years old were significantly more 121 likely to drain standing water from objects around their home than persons aged 70 and over (p<0.001). Likewise, individuals who correctly identified mosquito bites as the mode of WNV transmission were also significantly more likely to drain standing water around their homes (p<0.001) than those who incorrectly identified the mode of WNV transmission. Table 20. Demographic Characteristics, Knowledge, and Drain Standing Water Characteristic Yes – % (n) Chi- square p Gender Male Female 60.6 (57) 62.8 (64) 0.09 0.76 Age 60 – 69 years old ≥ 70 years old 74.8 (80) 45.8 (39) 16.77 <0.001 Education Less than Bachelor’s Bachelor’s or above 55.4 (51) 66.7 (68) 2.57 0.11 Marital status Single Married or domestic partnership 56.9 (45) 63.7 (72) 0.89 0.35 Race White Non-white 61.1 (99) 58.1 (18) 0.10 0.75 Annual household income ≤ $70,000 > $70,000 61.6 (61) 56.6 (30) 0.36 0.55 Knowledge: How do people get WNV? Correct response: mosquito bites Incorrect responses 68.9 (104) 40.8 (20) 12.36 <0.001 Table 21 shows chi-square test results for HBM constructs and draining standing water around one’s property. Perceived susceptibility and perceived severity both yielded statistically significant associations with the practice of draining standing water (p<0.05). Persons who indicated they were worried about getting WNV were significantly more likely to routinely drain water from objects around their homes (p<0.001); those who consider WNV a severe disease were also significantly more likely to drain standing water around their homes than those who did not feel it was a serious illness (p<0.01). No other statistically significant associations were noted between the other demographic characteristics and this outcome. 122 Table 21. HBM Constructs and Drain Standing Water Construct Yes: % (n) Chi-sq p Perceived susceptibility How worried are you that you might get WNV? Worried Not worried 72.1 (93) 44.9 (31) 15.71 <0.001 How likely is it that you will get WNV in the next year? Likely Not likely 72.4 (21) 60.2 (103) 1.56 0.21 Perceived severity: Do you think WNV is a serious illness?* Yes Don’t know 63.7 (123) 14.3 (1) -- 0.008 Perceived benefits Would you be interested in a community program to help repair damaged window screens? Yes No 82.5 (33) 71.4 (25) 1.31 0.25 How much would you be willing to pay for WNV vaccine? ≤ $49 $50 or above 59.0 (69) 72.4 (21) 1.77 0.18 Perceived barriers: Why have you not used insect repellent in last 90 days? Personal Non-personal 58.6 (58) 48.2 (13) 0.94 0.33 Why have you not drained standing water from objects around your home? Personal Non-personal 3.5 (2) 7.1 (1) -- 0.55 Cues to Action: Have you received any information about WNV in the last year? Yes No 66.7 (80) 55.0 (44) 2.77 0.096 Do you know anyone with WNV? Yes No 68.2 (15) 61.7 (108) 0.35 0.56 Self-efficacy: How confident you can protect yourself WNV? Confident Not confident 68.8 (104) 58.6 (17) 1.16 0.28 *None of the participants reported a “No” response to the perceived severity question; thus, the only operational categories were “Yes” and “Don’t know”. Willingness to Accept A WNV Vaccine Table 22 shows associations between selected demographic characteristics, WNV knowledge, and willingness to accept a WNV vaccine. Only gender and race were significantly associated with respondents’ willingness to accept a WNV vaccine, with both having p-values <0.05. Males showed greater vaccine acceptance than females, and whites demonstrated higher vaccine acceptance than non-whites. 123 Table 22. Demographic Characteristics, Knowledge, and Willingness to Accept a WNV Vaccine Characteristic Yes % (n) Chi-sq p Gender Male Female 90.5 (76) 77.5 (62) 5.17 0.02 Age 60 – 69 years old ≥ 70 years old 87.1 (81) 79.4 (54) 1.71 0.19 Education Less than Bachelor’s Bachelor’s or above 87.0 (67) 81.8 (72) 0.83 0.36 Marital status Single Married or domestic partnership 78.5 (51) 88.8 (87) 3.20 0.07 Race White Non-White 88.0 (125) 63.6 (14) 8.77 0.003 Annual Household Income ≤ $70,000 > $70,000 84.9 (73) 87.5 (42) 0.17 0.68 Knowledge: How do people get WNV? Correct response: mosquito bites Incorrect responses 85.4 (105) 81.4 (35) 0.38 0.54 Table 23 below presents associations between HBM constructs and willingness to accept a WNV vaccine. Perceived susceptibility was significantly associated with this outcome. Participants who expressed worry about getting WNV were more likely to accept a vaccine than those not worried about getting it. As indicated, none of the other HBM constructs were significantly associated with willingness to accept a WNV vaccine. Table 23. HBM Constructs and Willingness to Accept a WNV Vaccine Construct Yes: % (n) Chi-sq p Perceived susceptibility How worried are you that you might get WNV? Worried Not worried 90.0 (99) 73.2 (41) 7.91 0.005 How likely is it that you will get WNV in the next year? Likely Not likely 83.3 (20) 84.5 (120) 0.02 0.54 Perceived severity: Do you think WNV is a serious illness?* Yes Don’t know 85.2 (138) 50.0 (2) 3.66 0.11 Perceived benefits: Would you be interested in a program to repair damaged window screens? Yes No 81.8 (27) 76.7 (23) 0.25 0.61 How much would you be willing to pay for WNV vaccine? ≤ $49 $50 or above 99.1 (111) 100.0 (29) -- 0.61 Perceived barriers Why have you not used repellent in last 90 days? Personal Non-personal 83.5 (66) 91.3 (21) -- 0.51 Why have you not drained standing water from objects around your home? Personal Non-personal 81.3 (39) 76.9 (10) -- 0.73 Cues to Action: Have you received any information about WNV in the last year? Yes No 81.2 (82) 89.2 (58) 0.52 0.16 Do you know anyone with WNV? Yes No 82.4 (14) 84.3 (123) 0.04 0.84 Self-efficacy: How confident you can protect yourself WNV? Confident Not confident 86.2 (106) 85.2 (23) 0.018 0.90 124 *None of the participants reported a “No” response to the perceived severity question; thus, the only operational categories were “Yes” and “Don’t know”. Support for Community Mosquito Control Programs Table 24 lists the associations between respondent demographic characteristics, WNV knowledge, and being in favor of a community mosquito control program. None of these associations were statistically significant. Individuals between 60 and 69 years old accounted for the majority of respondents who favored mosquito control programs. A greater proportion of individuals with annual household income levels below $70,000 expressed support for community mosquito control programs than those with higher incomes. Table 24. Demographic Characteristics, WNV Knowledge, and Support for Mosquito Control Program Characteristic Yes – % (n) Chi- square p Gender Male Female 93.2 (82) 92.2 (83) 0.06 0.81 Age 60 – 69 years old ≥ 70 years old and above 91.8 (89) 92.3 (72) 0.02 0.89 Education Less than Bachelor’s Bachelor’s or above 95.2 (80) 90.3 (84) -- 0.25 Marital status Single Married or domestic partnership 90.3 (65) 95.1 (97) 1.53 0.22 Race White Non-white 91.8 (135) 93.3 (28) 0.08 0.78 Annual household income ≤ $70,000 > $70,000 93.4 (85) 93.2 (41) 0.0024 0.96 Knowledge: How do people get WNV? Correct response: mosquito bites Incorrect responses 92.7 (127) 91.1 (41) 0.12 0.73 Table 25 shows chi-square tests of independence for associations between HBM constructs and support for a mosquito control program. Both perceived severity and a single component of perceived benefits (amount respondent was willing to pay for WNV vaccine), demonstrated statistically significant associations with favoring mosquito 125 control (p<0.05). Individuals who preferred to pay less for a WNV vaccine (<$50) were more likely to support community mosquito control efforts. Table 25. HBM Constructs and Support for Mosquito Control Program Construct Yes: % (n) Chi- square p Perceived susceptibility How worried are you that you might get WNV? Worried Not worried 94.8 (110) 87.3 (55) 3.21 0.07 How likely is it that you will get WNV in the next year? Likely Not likely 93.3 (28) 92.1 (140) -- 0.82 Perceived severity: Do you think WNV is a serious illness?* Yes Don’t know 93.1 (163) 71.4 (5) 4.46 0.035 Perceived benefits Would you be interested in a community program to help repair damaged window screens? Yes No 88.9 (32) 96.9 (31) -- 0.21 How much would you be willing to pay for WNV vaccine? ≤ $49 $50 or above 98.1 (104) 84.6 (22) -- 0.003 Perceived barriers: Why have you not used insect repellent in last 90 days? Personal Non-personal 88.6 (78) 96.0 (24) -- 0.27 Why have you not drained standing water from objects around your home? Personal Non-personal 89.3 (50) 91.7 (11) -- 0.81 Cues to Action: Have you received any information about WNV in the last year? Yes No 94.5 (104) 88.9 (64) 1.96 0.16 Do you know anyone with WNV? Yes No 86.4 (19) 93.6 (147) -- 0.22 Self-efficacy: How confident are you that you can protect yourself from WNV? Confident Not confident 94.9 (130) 88.0 (22) -- 0.19 *None of the participants reported a “No” response to the perceived severity question; thus, the only operational categories were “Yes” and “Don’t know”. Summary The bivariate analyses revealed some significant associations between demographic characteristics, knowledge, and HBM constructs as well as outcome variables. Worry about getting sick with WNV infection was significantly related to multiple behavioral outcomes, including acceptance of WNV vaccine, insect repellent use, and draining of standing water around the home. Only those variables found to be significantly associated with the outcomes in the bivariate analysis will be included as covariates in the multivariate logistic regression models. 126 4.4 Multivariate Analysis Multivariate analysis in the form of logistic regression was conducted to test the study hypotheses and to determine which HBM constructs predicted WNV prevention attitudes and behaviors among adults ≥60 years old, after adjusting for confounders. Operationalization and measurement of the independent variables that are used in these analyses are described in the Variables and Measurement section of Chapter 3. Prior to initiating the multivariate logistic regression analyses, bivariate tests were used to determine which control variables should be included in the multivariate logistic regression models. Specifically, Pearson’s chi-square tests were used to identify significant associations between respondent characteristics and each of the six dependent variables: (1) avoidance of the outdoors at dusk and dawn, (2) dressing in long-sleeved shirts and long pants when outdoors, (3) having used insect repellent in the last 90 days, (4) draining standing water from around the property, (5) willingness to accept a WNV vaccine, and (6) support for community mosquito control programs. 4.4.1 Summary of Bivariate Analysis Results Bivariate analyses revealed significant associations between selected HBM constructs and study outcomes. Perceived susceptibility to WNV infection was a significant predictor of insect repellent use (OR=3.04, 95% CI: 1.65-5.59), draining standing water (OR=3.33, 95% CI: 1.82-6.11), and acceptance of a WNV vaccine (OR=3.29, 95% CI: 1.39-7.77). Perceived severity was a significant predictor of draining standing water (OR=10.54, 95% CI: 1.24-89.36). Perceived benefits (OR=9.45, 95% CI: 1.63-54.88) was significantly associated with support for community mosquito control programs. 127 Although no single demographic characteristic was found to be significantly associated with all six outcomes, three demographic variables demonstrated significant relationships with two or more outcomes in bivariate analysis. Age was found to have a significant association with use of insect repellent (χ2(1)= 8.40, p<0.01; OR=2.34, 95%CI: 1.31-4.19) and draining of standing water (χ2(1)= 16.77, p<0.001; OR=3.49, 95%CI: 1.89-6.43) and a borderline significant association with dressing in long-sleeved shirts and pants when outdoors (χ2(1)= 3.76, p=0.05; OR=0.56; 95%CI: 0.31-1.01). Race was also found to be significantly associated with dressing in long-sleeved shirts and pants when outdoors (χ2 (1)= 5.64, p<0.05; OR=0.36, 95% CI: 0.15-0.85) and with acceptance of a WNV vaccine (χ2(1)= 8.77, p<0.01; OR=4.20, 95% CI: 1.54-11.49). Similarly, knowledge of WNV transmission was found to be significantly associated with use of insect repellent (χ2(1)= 7.35, p<0.01; OR=2.49, 95% CI: 1.27-4.89) and draining of standing water (χ2(1)= 12.36, p<0.001; OR=3.21, 95% CI: 1.65-6.24). 4.4.2 Selection of Control Variables for Multivariate Analysis Based on the results of the above chi-square tests, age, race, and WNV knowledge were selected as control variables to be included in all multiple logistic regression analyses. The results of logistic regression analysis used to test the 36 research hypotheses are presented below and the same three controls were included in all models along with selected independent variables that met specific criteria for model entry, as described below. 4.4.3 Building Logistic Regression Models For each of the six dichotomous outcomes -- (1) avoidance of outdoors at dusk and dawn, (2) dressing in long-sleeved shirts and long pants, (3) having used insect 128 repellent in the last 90 days, (4) draining standing water from around the property, (5) willingness to accept a WNV vaccine, and (6) support for mosquito control programs – each HBM construct was entered into a logistic regression model, regardless of its significance in bivariate analysis, along with the three control variables and other independent variables that met specific criteria designating a maximum p-value. Specifically, a logistic regression model was built and variables entered into it using model building techniques described by Hosmer & Lemeshow (2000), which involved identifying statistically significant variables in bivariate analysis, specifying a threshold for entry and removal of variables from the model, and applying stepwise procedures using forward selection with a test for backward elimination (Hosmer & Lemeshow, 2000) to generate a final model. For this analysis, variables with a p-value < 0.20 were used as the threshold for variable removal and entry (Hosmer and Lemeshow specify a cutoff of p<0.25, but 0.20 was selected for this analysis instead as a more conservative approach). Accordingly, to test each hypothesis, each of the six outcomes was regressed on each of the six HBM constructs and additional variables that met the p<0.20 criterion were also entered into the model. Results are presented below according to the six dependent variables. The tables are grouped accordingly and include all independent variables that were entered into the initial model (regardless of whether they remained in the final model). Only those variables that maintained a significance level of p<0.20 were retained in the final model (shaded in each table). Statistical significance was denoted using bold red font. 129 4.4.4 Results of Multivariate Logistic Regression Analysis Outcome #1: Avoiding outdoors at mosquito feeding hours of dusk and dawn Hypotheses 1, 7, 13, 19, 25, and 31 tested the above outcome by regressing the dependent variable on each of the six HBM constructs to determine their effect on respondents’ avoidance of the outdoors during mosquito feeding hours of dusk and dawn. The single covariate that met the p<0.20 criteria for entry into the models for this outcome was marital status, which was included in the initial models. Table 26 shows results of the six adjusted logistic regression models to test those six hypotheses. 130 Table 26. Logistic Regression Models for HBM Constructs and Avoidance of the Outdoors (Hypotheses 1, 7, 13, 19, 25, and 31)* Independent Variable Adj OR p- value 95% CI Lower Upper Hypothesis 1: Individuals with high perceived susceptibility to WNV are more likely to avoid the outdoors at dusk and dawn than those with low perceived susceptibility to WNV. Worried about getting WNV Yes No (reference) 1.33 1 0.42 0.66 2.68 Likelihood of getting sick with WNV Not at all likely (ref) Not very likely Moderately likely Very likely Extremely likely 1 1.59 1.44 1.17 0 0.21 0.48 0.88 0.77 0.52 0.15 3.27 4.00 9.44 Marital status Single/Separated/Div/Widowed Married (reference) 0.68 1 0.23 0.37 1.27 Hypothesis 7: Individuals who believe WNV causes serious illness are more likely to avoid the outdoors at dusk and dawn than those who do not believe it is serious. Perceive WNV as a serious illness† Yes Don’t know 0.21 1 0.10 0.03 1.32 Marital status Single/Separated/Div/Widowed Married (reference) 0.60 0.11 0.32 1.12 Hypothesis 13: Individuals who perceive benefits in practicing WNV protective behaviors are more likely to avoid the outdoors at dusk and dawn than those who do not perceive benefits. Amount willing pay for WNV vaccine $0-$49 ≥ $50 (reference) 0.65 1 0.63 0.11 3.76 Perceived benefits of community program to repair windows Yes No (reference) 0.78 1 0.72 0.19 3.05 Marital status Single/Separated/Div/Widowed Married (reference) 1.28 1 0.73 0.31 5.26 Hypothesis 19: Individuals with high perceived barriers to practicing WNV protective behaviors are less likely to avoid the outdoors at dusk and dawn than those with low perceived barriers. Perceived barrier to draining standing water Personal Not Personal (ref) 0.13 1 0.06 0.02 1.08 Perceived barrier to insect repellent use Personal Not personal (ref) 3.29 1 0.15 0.65 16.74 Marital status Single/Sep/Div/Widowed Married (reference) 0.13 1 0.01 0.02 0.66 Hypothesis 25: Individuals who receive cues to action are more likely to avoid the outdoors at dusk and dawn than those who did not receive cues to action. Received information about WNV in last year Yes No (reference) 1.12 1 0.73 0.59 2.12 Know anyone with WNV Yes No (reference) 1.75 1 0.27 0.65 4.71 Marital status Single/Separated/Div/Widowed Married (reference) 0.64 1 0.17 0.35 1.20 Hypothesis 31: Individuals with high self-efficacy for preventing WNV infection are more likely to avoid the outdoors at dusk and dawn than those with low self-efficacy for preventing WNV infection. Confident can protect self from WNV Not confident (ref) Somewhat confident Very confident 1 0.80 1.32 0.63 0.59 0.32 0.48 1.98 3.67 Marital status Single/Separated/Div/Widowed Married (reference) 0.68 1 0.26 0.35 1.33 *Age, Race, and WNV knowledge were entered as controls for each of the six hypothesis tests. 131 † None of the survey participants reported a “No” response to the perceived severity question; thus, the only response categories were “Yes” and “Don’t know”. None of the hypothesized associations between HBM constructs and avoidance of the outdoors at dusk and dawn were supported, but a demographic characteristic was significantly associated with this outcome. Marital status was significantly associated with avoidance of the outdoors in hypothesis 19 with single/unmarried respondents less likely to avoid the outdoors than married respondents (OR=0.11; 95% CI: 0.02-0.63). Outcome #2: Dressing in long-sleeved shirts and long pants when outdoors Hypotheses 2, 8, 14, 20, 26, and 32 tested the above outcome by regressing the dependent variable on each of the six HBM constructs to determine their effect on respondents’ choice of attire when outdoors in the summertime. A single covariate, education level, met the p<0.20 criteria for entry into the models for this outcome and was included along with the target HBM construct. Table 27 displays results of the adjusted logistic regression models to test those six hypotheses for the above outcome. 132 Table 27. Logistic Regression Models for HBM Constructs and Dressing in Long- Sleeved Shirts and Long Pants (Hypotheses 2, 8, 14, 20, 26, and 32)* Independent Variable Adjusted OR p- value 95% CI Lower Upper Hypothesis 2: Individuals with high perceived susceptibility to WNV are more likely to dress in long- sleeved shirts and long pants than those with low perceived susceptibility to WNV infection. Worried about getting WNV Yes No (ref) 0.99 1 0.98 0.48 2.03 Likelihood of getting sick with WNV Not likely (ref) Not very likely Moderately likely Very likely Extremely likely 1 1.01 2.45 1.43 0 0.99 0.10 0.74 0.48 0.85 0.17 2.10 7.04 11.88 Education 1 person] Among the [#] individuals you mentioned, whose birthday will occur next?  May I please speak with that individual? o If person is potential enrollee, go to Survey description o If person coming to the phone, go to Q2 o If person not home at this time, go to Q1.1 o If not a good time, go to Q1.1 o If voicemail go to Q1.2 o If person does not speak English, s/he is not eligible. =STOP= 1.1 IF POTENTIAL ENROLLEE NOT HOME OR IF NOT A GOOD TIME: His/her participation in this research study is very important. When would be a good time to reach him/her? RECORD PERSON’S NAME TO ASK FOR AND DAY/TIME TO CALL. Thank you for your time. I will call again at that time. Good-bye. =STOP= 1.2 If voicemail: “Hello, my name is _________ and I am calling from the Maryland Department of Health for Mr./Mrs. ____________. We are interested in talking with adults 60 184 years of age and older about West Nile virus. If you or a member of your household is at least 60 years old, please call us at 410-767-5649 as we would very much like to talk with you. Thank you.” 2. Hello my name is _________. I’m calling from the MARYLAND DEPARTMENT OF HEALTH AND MENTAL HYGIENE. We are doing a survey about West Nile Virus. Specifically, we are interested in talking with adults 60 years of age and older. I would like to confirm that you are in that age group. Are you at least 60 years of age? ___ If Yes, go to Survey Description ___ No; go to Q2.1. 2.1 IF NO, May I speak with him/her? ___ NOT AT HOME; GO TO Q2.2. ___ Yes; COMING TO THE PHONE; GO BACK TO Q2. ___ No; Thank you for your time. =STOP= 2.2 Is there another phone number where I could reach him/her? ___ Yes; RECORD ALTERNATE PHONE NUMBER. Thank you very much for your time. =STOP= ___ No; When would be a good time to call back to reach him/her? [RECORD DAY/ TIME]. Thank you very much for your time. =STOP= Survey Description [Once age/randomization criteria have been met] Now, I’d like to tell you a little bit more about this survey. We are talking with adults 60 years of age and older to learn about things you may already be doing around your home that would protect you from a disease caused by the West Nile virus, what you know about this disease, and your opinions about ways in which we might help protect your community from this disease. We will use this information to develop educational materials and other programs to help protect people from getting sick with this disease. It will take about 10 to 15 minutes to answer these questions. We will not ask you any questions about your health or that of any other members of your household. Your answers will not be linked to your name, address or other information that may identify you. Likewise, we will not use your name, address or any other identifying information in any reports or materials that we may publish. Your participation in this survey is completely voluntary. At any time, you may decide that you do not want to answer a specific question. That is OK. If you decide that you do not want to finish the interview that is OK too. However, your answers are very important and will help us develop better programs to serve your community. 185 We used a publicly available source of information to identify households located in the communities of interest to us. However, we know that this information is sometimes not correct. 3. Would you be willing to answer two questions so that we can make sure that you are eligible to participate in this survey? ___ Yes ___ No If NO: Your participation in this research study is very important. We are trying to determine what adults 60 years old and over know about how to prevent West Nile Virus. May I schedule a time to talk that would be better for you? ___ Yes; RECORD DAY/TIME. Thank you very much for your time. =STOP= ___No, sorry to have disturbed you. Good-bye. =STOP= If YES: Thank you. In what zip code do you live? [PLEASE CHECK BOX NEXT TO APPROPRIATE ZIP CODE] Zip codes Zip codes  20902  21214  20910  21215  21014  21222  21060  21224  21122  21228  21212 If Potential Enrollee does not live in one of the listed zip codes then: READ: I’m sorry; the zip code in which you live is not one of our study zip codes. Thank you for your interest in participating. READ: In what type of housing do you live? [LIST ITEMS BELOW] 1. Single-family detached home 2. Townhouse or condominium 3. Apartment 4. Active living senior community 5. Other housing: _____________________________ If Potential Enrollee lives in a long-term care facility, nursing home, or institution, then ineligible to participate: READ: I’m sorry, you are not eligible to participate based on your housing. Verbal Consent [IF ALL ELIGIBILITY CRITERIA ARE MET] READ: It seems that the information we had about your household is correct. I would now like to ask you our survey questions. But before I do so, do you have any questions for me? If yes, answer questions. If no, continue 186 READ: Now, or at any time during or after the survey, you may also contact Kim Mitchell, MPH, WNV Prevention Project Coordinator at 410-767-5649 with any questions about this project or Gay Hutchen Administrator, DHMH Institutional Review Board, at 410-767-8448 with any questions you may have about your rights as a survey participant. At the end of the survey, I will also provide you with an opportunity to receive a written description of this project including who to contact with any further questions. Are you willing to take part in this survey? ___ Yes ___ No Interviewer’s initials ______ If yes, go to first question (Q4). If no: Your participation in this research study is very important. We are trying to determine what adults 60 years old and above know about how to prevent West Nile Virus. May I schedule a time to talk that would be better for you? ___ Yes; RECORD DAY/TIME. Thank you very much for your time. =STOP= ___ No, thank for their time. Good-bye. =STOP= BEGIN INTERVIEW Knowledge of West Nile Virus and Severity of Disease READ: We are going to start the survey now. These first questions ask about what you know and think about West Nile virus disease. 4. Have you ever heard of West Nile virus? a. Yes IF YES, go to Q5. b. No IF NO, go to Q4a. c. Don’t know d. Refused 4a. Would you be willing to answer some questions about ways to avoid mosquitoes in the summer? Circle Yes / No. If yes, skip to Q11. 5. How do you think most people get West Nile virus? a. Eating or drinking contaminated food or water b. From bug bites, such as mosquito bites c. From birds d. Contact with sick people e. Other (list):_______________________________________________________ f. Don’t know g. Refused 6. How worried are you that you might get sick with West Nile virus? [READ CATEGORIES] 187 a. Very worried b. Somewhat worried c. A little worried d. Not at all worried e. Don’t know f. Refused 6a. If Very worried or somewhat worried: can you tell me why you are worried about getting sick with WNV: a. I hear about it on the news b. I know someone who had it c. I get sick easily d. Other: __________________________ e. Don’t know 6b. If A Little worried or Not at all worried: can you tell me why you are not worried about getting sick with WNV: a. I cover up when I go outdoors b. I use bug spray/repellent c. I almost never go outside in the summer d. Mosquitoes are not a problem where I live e. I don’t think WNV is in my area f. Other: ______________________________ 7. Do you think West Nile virus can cause serious illness? a. Yes b. No c. Don’t know d. Refused 8. What age group do you think is most likely to get seriously ill from West Nile virus? [READ ALL ITEMS] a. Young children (0-10 years old) b. Adolescents and teenagers (11-18 years old) c. Young adults (19-25 years old) d. Adults (26-50 years old) e. Adults >50 years old f. Don’t know g. Refused 9. On a 1 to 5 scale, with 1 being “not at all likely” and 5 being “extremely likely” how likely do you think it is that you will get West Nile virus in the next calendar year? Not at all likely 1 Not very likely 2 Moderately likely 3 Very likely 4 Extremely likely 5 188 9a. IF NOT AT ALL LIKELY TO MODERATELY LIKELY: why do you think it is unlikely, not very likely, or moderately likely that you will get WNV in the next calendar year? a. I don’t go outside b. I always use repellent c. I am afraid of mosquitoes d. No one I know has gotten it e. Other: _________________________________ f. Don’t know g. Refused 9b. IF VERY LIKELY OR EXTREMELY LIKELY: why do you think it is likely that you will get WNV in the next calendar year? h. I often spend time outside i. I never use repellent j. I do not know how to avoid it k. Other: ________________________________________ l. Don’t know m. Refused 10. Do you know anyone who has had West Nile virus? a. Yes b. No c. Don’t know d. Refused Prevention READ: Next I would like to ask you some questions about preventing West Nile virus. 11. How confident are you that you can protect yourself and your household members from getting West Nile virus? Very confident 1 Somewhat confident 2 Not at all confident 3 Don’t Know 4 Refused 5 11a. IF VERY CONFIDENT OR SOMEWHAT CONFIDENT: Can you tell me what makes you feel confident that you can protect yourself and your household members from WNV (circle all that apply)? a. I always use insect repellent and encourage others in the house to use it b. I do not go outside when mosquitoes are active c. I empty containers holding standing water on my property d. I educate others in my household about how to avoid mosquito bites e. Other [ASK RESPONDENT TO LIST]: _____________________ f. Don’t know g. Refused 189 11b. IF NOT AT ALL CONFIDENT: Can you tell me why you do not feel confident that you can protect yourself or your family members from WNV (circle all that apply)? a. I do not know how to prevent WNV b. I do not know how to use repellent c. It is too inconvenient to take steps to protect myself and others who live with me d. Don’t know e. Refused 12. Since the start of this past summer (2012), did you or someone you asked or hired drain water from items around the outside of your home, such as gutters, buckets, flower pots, kiddie pools, bird baths, or discarded tires…? a. Yes IF YES, GO TO 12a b. No IF NO, GO TO Q13 c. Don’t know d. Refused 12a. If yes, how many times each summer do you have the water drained from those objects around your home? ____ [INDICATE APPROXIMATE NUMBER OF TIMES] 13. [IF NO], can you tell me why you do not drain water from items on your property that may collect water [Circle all that apply]? a. It takes too much effort (e.g., too time consuming, not enough energy, or too heavy) b. I do not have anyone to help me c. It is too dangerous d. I use products (e.g., mosquito-dunks) in the containers to keep mosquitoes from breeding in the water e. There is nothing that collects water in my yard f. I do not own my home g. Other (list)______________________________________________________ h. Don’t know i. Refused 14. How do you keep your home cool in the summer (circle all that apply)? a. Fans– GO TO Q16 b. Air-conditioning– GO TO Q16 c. Open windows – GO TO Q14a d. Other (list): ________________________________________ e. Don’t know f. Refused 190 14a. [If open windows, ASK] do all of the windows in your home have screens? a. Yes b. No c. Don’t know d. Refused 14b. [IF YES (ALL WINDOWS HAVE SCREENS)] are your screens in good condition (i.e., no holes or tears)? a. Yes b. No IF NO, GO TO 14c c. Don’t know d. Refused 14c. [IF NO, ASK], can you tell me why you have not repaired the damaged screens? a. It takes too much effort (e.g., too time-consuming, not enough energy) b. They are too difficult to check or repair c. I do not have anyone to help me d. It is too expensive e. I did not think it would help f. Not applicable—I do not own my home g. Other: [ASK RESPONDENT TO LIST:]____________________________ 15. Would you be interested in a community program to help adults over 60 years old repair their damaged window screens and dump standing water in yard? e. Yes f. No g. Not applicable—I do not own my home h. Don’t know i. Refused 15a. [IF YES, ASK] how do you think the program should be supported? _________________________________________________________ READ: These next few questions are about programs in your community. 16. Are you aware of any government programs that use pesticides (i.e., larvicides or insecticides) to reduce mosquitoes in your community? a. Yes b. No c. Refused 17. Are you or would you be in favor of a mosquito control program like that? a. Yes b. No 191 c. Don’t know d. Refused 18. Are you in favor of tax money being used to support mosquito control programs in your community? a. Yes b. No c. Don’t know d. Refused 18a. If no, how do you think a mosquito control program in the community should be supported? _________________________________________________________________ 19. Which, if any, of the following insect repellents or other products have you heard of [circle all that apply]? a. DEET (N,N-diethyl-meta-toluamide) b. Citronella c. Picaridin d. IR3535 e. Oil of lemon eucalyptus f. Permethrin g. Skin So Soft h. None of them i. Don’t know/not sure j. Refused 20. In the last 90 days (since the beginning of September), did you always, sometimes, rarely or never use insect repellent on your skin when you went outside? a. Always b. Sometimes c. Rarely d. Never 20a. IF RARELY or NEVER, ASK: can you tell me who you rarely or never use insect repellent on your skin (can choose more than one) a. I often forget b. It feels sticky on my skin c. It smells bad d. It will make me feel sick or get a rash e. It is too expensive f. I didn’t know it could help g. Other: [ASK RESPONDENT TO LIST] ______________ h. Don’t know i. Refused 192 20b. IF RARELY or NEVER, ASK: would you be more likely to use insect repellent if it were [can choose more than one]: a. Given away free at community centers or other places b. Sent to you as a free sample in the mail c. Odorless d. Did not leave a residue on skin e. Other [ASK RESPONDENT TO LIST]: ____________________________ f. Don’t know g. Refused 21. In the summer months, about how many hours do you spend outside during the day: _______ [please list a number or range] a. Don’t know b. Refused 21b. When you do go outdoors, is it usually: c. At dusk or dawn d. In the middle of the day 22. When you go outdoors during the summer, do you wear long-sleeved shirts and/or long pants? a. Yes – GO TO Q22a b. No – GO TO Q23 c. Don’t know d. Refused 22a. If yes, at what time of day do you dress in long-sleeved shirts and/or long pants? e. At dusk or dawn f. In the middle of the day 23. If there were a West Nile virus vaccine available that is both safe and effective, would you be willing to take it? a. Yes – GO TO Q23b b. No – GO TO Q23a c. Maybe d. Don’t know e. Refused 23a. IF NO, please indicate why not: a. I don’t like vaccines b. I don’t believe it would work c. I am afraid of needles 193 d. I am afraid it would make me sick e. I am not worried about getting West Nile virus f. Other: ___________________________ 23b. If YES and the vaccine was not covered under your health insurance plan, how much would you be willing to pay out of pocket for the vaccine? a. $0 b. <$25 c. $25-49 d. $50-74 e. $75-99 f. >$100 Educational material READ: These next questions ask about how to distribute information about WNV. 24. In the last year, have you received information in any form about West Nile virus (this can be written material, electronic media, word-of-mouth, or some other method)? a. Yes b. No c. Don’t know d. Refused 24a. [IF YES] please tell me who provided it or in what form that information was provided [can choose more than one]: a. Doctor or other health care provider b. Radio [ASK RESPONDENT TO LIST STATION __________] c. Television [ASK RESPONDENT TO LIST CHANNEL________] d. Newspaper [If yes, ASK RESPONDENT TO NAME:] ___________________ e. Internet or website [If yes, ASK WHICH ONE:] ___________________ f. Magazine [If yes, ASK RESPONDENT TO NAME:] ___________________ g. Other written material (brochure, newsletter, flyer, fact sheet, door hanger) h. E-mail i. Mail j. Word of mouth k. Other [PLEASE ASK RESPONDENT TO LIST:] ___________________ l. Don’t know m. Refused 25. How do you prefer to get information about health issues and health care services [can choose more than one]? 194 a. Doctor or healthcare provider b. Radio c. Television news d. Internet/e-mail e. Mail f. Newspaper g. Public gatherings (e.g., health fairs, senior center events) h. Other [ASK RESPONDENT TO LIST:]: ______________________ i. Don’t know j. Refused Demographic READ: These last few question are about you. 26. Is your home located? a. In a city b. In the suburbs c. In a rural area d. Refused 27. Do you own your home? a. Yes b. No c. Don’t know d. Refused 28. How long have you lived at your current residence: ___ [approximate months or years] 29. Would you please tell me the age and gender of yourself and the other people in your household who are 60 years old or older? [List respondent age and sex on line #1] Age Gender Age Gender 1. 1. 1111 5. 2. 6. 3. 7. 4. 8. 30. Are you Hispanic or Latino/a or of Spanish origin? Yes ___ No ___ 31. How would you describe your race? You can list more than one category. CHECK ALL THAT APPLY. 195  White or Caucasian  Black or African American  Asian  Native Hawaiian or other Pacific Islander  American Indian/Alaska Native/Chicano/Mestizo DO NOT READ:  Other [specify] __________________  Don’t know or not sure  Refused 32. What is the primary language spoken in your household? a. English b. Spanish c. Other: _____________________________ d. Refused 33. What is the highest level of education you have completed (circle only one)? a. Some high school b. High school Diploma or GED c. Some college, including Associate degree d. Bachelor’s degree e. Some graduate school f. Graduate degree g. Refused 34. What is your marital status? a. Married b. Single c. Separated d. Divorced e. Widowed f. Refused 35. What is your employment status? a. Full time b. Part time or “semi-retired” c. Retired d. Unemployed e. Other: [ASK RESPONDENT TO LIST]: ____________________________ f. Refused 36. Which category best describes your total household annual income? [please read categories to respondent and circle answer] 196 a. ≤ $20,000 b. $20,001 to $30,000 c. $30,001 to $40,000 d. $40,001 to $50,000 e. $50,001 to $60,000 f. $60,001 to $70,000 g. Over $70,000 h. Don’t know i. Refused That concludes the survey. Thank you so much for your time. Do you have any questions? Questions: If you would like, I can send you a copy of the information that I provided about the study for your records. I can send it by mail, e-mail, or fax. Yes ___ No ___ [RECORD PREFERENCE; IF APPROPRIATE GET CONTACT INFORMATION] For Office Use: Date: __________________ Time: _____________________________ Administered by [name of interviewer]: ____________________________________ 197 I will send you a copy of the INFORMATION THAT I PROVIDED ABOUT THE STUDY (Project Information Sheet) for your records. Would you prefer that I mail, email or fax the form to you? RECORD PREFERENCE AND GET MAILING ADDRESS, EMAIL, OR FAX NUMBER AS APPROPRIATE. DETACH PAGE HERE -------------------------------------------------------------------- Name: ___________________________________________________________ Mailing address: ___________________________________________________ E-mail address: ____________________________________________________ Fax number: _______________________________________________________ 198 APPENDIX D: PROJECT INFORMATION SHEET Maryland Department of Health and Mental Hygiene Assessing Barriers to Prevention of West Nile Virus in Persons at Least 60 years old in Maryland Principal Investigators: Kimberly C. Mitchell, MPH (Maryland DHMH), Katherine Feldman, DVM, MPH (Maryland DHMH) Introduction Thank you for taking part in this study by the Maryland Department of Health and Mental Hygiene and the Centers for Disease Control and Prevention. The information you have given us will enable us to plan programs that help people avoid getting sick with West Nile virus and other health conditions that might be of interest to your community. Your household contact information was picked at random by computer from a publicly available database of households in Maryland. Potential Risks There were minimal risks involved in this study. Some of the survey questions asked about your income, education, and other information that may usually be kept private. Benefits Benefits you may get from this study include the option to be given contact information for local agencies that you may call to get information and resources about West Nile virus. You may contact the WNV Prevention Project Coordinator below to request this information. Persons to Contact If you have questions about this study, please contact Kim Mitchell, MPH, WNV Prevention Project Coordinator, at 410-767-5649. If you have questions about your rights as a participant in the study, please contact Ms. Gay Hutchen, Administrator, Maryland Department of Health and Mental Hygiene, Institutional Review Board, 201 West Preston Street, Baltimore, MD 21201, 410-767-8448. Confidentiality Your answers will be kept private and will be used only for research. We will not collect your name and your household information will not appear on any project reports or published materials. Costs There is no cost to you for completing the survey. Right to Refuse or Withdraw 199 This study is completely voluntary. There is no penalty for not participating or not answering all of the survey questions. You did not give up any legal rights by being part of the survey. 200 APPENDIX E: UMCP IRB APPROVAL LETTER 201 202 REFERENCES AbuSabha, R., Achterberg, C. (1997). Review of Self-Efficacy and Locus of Control for Nutrition and Health-Related Behavior. Journal of American Dietary Association; 97, 1122-1132. Adams, M.L., LoBianco, L., Wilcox, L., Hadler, J.L., Griffith, K.S. (2003). Knowledge, Attitudes and Behaviors about West Nile Virus--Connecticut, 2002. CDC Morbidity and Mortality Weekly Report, 52(37), 886-888. http://www.cdc.gov/mmwr/preview/mmwrhtml/mm5237a4.htm . Akompab, D. A., Bi, P., Williams, S., Grant, J., Walker, I. A., Augoustinos, M. (2013). Heat waves and climate change: applying the Health Belief Model to Identify Predictors of Risk Perception and Adaptive Behaviours in Adelaide, Australia. Int J Env Res Pub Health; 10, 2164-2184. Aquino, M., Fyfe, M., MacDougall, L., Remple, V. (2004). West Nile Virus in British Columbia. Emerging Infectious Diseases, 10(8), 1499-1500. Arcury, T. A., Estrada, J. M., Quandt, S. A. (2010). Overcoming language and literacy barriers in safety and health training of agricultural workers. J Agromedicine; 15(3), 236-248. Artsob, H., Gubler, D. J., Enria, D. A., Morales, M. A., Pupo, M., Bunning, M., Dudley, J. P. (2009). West Nile Virus in the New World: Trends in the Spread and Proliferation of West Nile Virus in the Western Hemisphere. Zoonoses Public Health, 56, 357-369. Atkinson, N., & Permuth-Levine, R. (2009). Benefits, Barriers, and Cues to Action of Yoga Practice: A Focus Group Approach. Am J Health Behav, 33(1), 3-14. Averett, E., Neuberger, J. S., Hansen, G., Fox, M. H. (2005). Evaluation of West Nile Virus Education Campaign. Emerging Infectious Diseases, 11(11), 1751-1753. Beaujean, D. J., Bults, M., van Steenbergen, J., Voeten, H. A. (2013). Study on public perceptions and protective behaviors regarding Lyme disease among the general public in the Netherlands: implications for prevention programs. BMC Public Health, 13:225. Berner, Y. N., Lang, R., Chowers, M. Y. (2002). Outcome of West Nile Fever in Older Adults. JAGS, 50(11), 1844-1846. Bish, A., Yardley, L., Nicoll, A., Michie, S. (2011). Factors Associated with Uptake of Vaccination Against Pandemic Influenza: A Systematic Review. Vaccine, 29, 6472-6484. 203 Bitto, A., Hakim, J. A., Pula, M. F. (2005). What Seniors Say About West Nile Virus— The Threat is Not Over. California J Health Promotion, 3(3), 127-136. Bonar, E. E., Rosenberg, H. (2011). Using the Health Belief Model to predict injecting drug users’ intentions to employ harm reduction strategies. Addictive Behaviors, 36, 1038-1044. Bosch, S. A., Leong, K., Musgrave, K., Powers, J., Wong, D. (2010). Zoonotic Disease Risk Perception and Use of Personal Protective Measures among Wildlife Biologists: An Application of the Health Belief Model. Human Dimensions of Wildlife 15, 221-228. Brewer, N. T., Rimer, B. K. (2008). Perspectives on Health Behavior Theories that Focus on Individuals. In K. Glanz, B. K. Rimer, K. Viswanath (Eds.), Health Education and Health Behavior: Theory, Research, and Practice (4th Ed., pp. 45-65). San Francisco, CA: Jossey-Bass. Brewer NT, Gilkey MB. (2012). Comparing theories of health behavior using data from longitudinal studies: a comment on Gerend and Shepherd. Ann Behav Med, 44(2), 147-148. Bults, M., Beaujean, D. J., de Zwart, O., Kok, G., van Empelen P., van Steenbergen J., . . . Voeten, H. (2011). Perceived risk, anxiety, and behavioural responses of the general public during the early phase of the influenza A (H1N1) pandemic in the Netherlands: results of three consecutive online surveys. BMC Public Health, 11:2. Bursac, Z., Gauss C. H., Williams, D. K. , Hosmer, D. W. (2008). Purposeful Selection of Variables for Logistic Regression. Source Code for Biology and Medicine, 3(17). Butterworth, M. (2009). Knowledge, Perceptions, and Practices: Mosquito-borne Disease Transmission in Southwest Virginia. (Master’s Thesis) Retrieved from ProQuest Dissertations and Theses. Calvocoressi L., Kasl S. V., Lee, C. H., Stolar, M., Claus, E. B., Jones, B. A. (2004). A Prospective Study of Perceived Susceptibility to Breast Cancer and Nonadherence to Mammography Screening Guidelines in African American and White Women Ages 40 to 79 Years. Cancer Epidemiol Biomarkers Prev, 13, 2096-2105. Carpenter, C. J. (2010). A Meta-Analysis of the Effectiveness of Health Belief Model Variables in Predicting Behavior. Health Communication, 25(8), 661-669. Case, D. O., Andrews, J. E., Johnson, J. D., Allard, S. L. (2005). Avoiding versus seeking: the relationship of information seeking to avoidance, blunting, coping, dissonance, and related concepts. J Med Lib Assoc, 93(3), 353-362. 204 Centers for Disease Control and Prevention. (2011). Nationally Notifiable Disease Surveillance System (NNDSS). Arboviral diseases: neuroinvasive and non- neuroinvasive case definitions. Centers for Disease Control and Prevention. (2008). West Nile Virus (WNV) Factsheet. Centers for Disease Control and Prevention. West Nile Virus: Information and Guidance for Clinicians. Clinician Outreach and Communication Activity (COCA) Conference Call. August 30, 2012. Centers for Disease Control and Prevention. “West Nile Virus: Questions and Answers”. N.P. Accessed October 10, 2012. Centers for Disease Control and Prevention. West Nile Virus: A Risk You Can Do Something About, 2004. Centers for Disease Control and Prevention. West Nile Virus in the United States: Guidelines for Surveillance, Prevention, and Control, 4th revision, June 2013. Champion, V. L., & Skinner, C. S. (2008). The Health Belief Model. In K. Glanz, B. K. Rimer, K. Viswanath (Eds.), Health Education and Health Behavior: Theory, research, and practice (4th Ed., pp. 45-65). San Francisco, CA. Jossey-Bass. Chin, Y., Huang, T., Hsu, B. R. (2012). Impact of action cues, self-efficacy and perceived barriers on daily foot exam practice in type 2 diabetes mellitus patients with peripheral neuropathy. J Clin Nursing, 22, 61-68. Coe, A. B., Gatewood, S., Moczygemba, L. R., Goode, J., Beckner, J. O. (2012). The Use of the Health Belief Model to Assess Predictors of Intent to Receive the Novel (2009) H1N1 Influenza Vaccine. Inov Pharma, 3(2), 1-11. Conner, M. (2010). Cognitive Determinants of Health Behavior. In A. Steptoe (Ed.), Handbook of Behavioral Medicine: Methods and Applications (pp. 19-30). Springer Science + Business Media, LLC. Curry, V. J., Cole, M. (2001). Applying social and behavioral theory as a template in containing and confining VRE. Crit Care Nurs Q, 24(2): 13-19. Deavenport, A., Modeste, N., Marshak, H. H., Neish, C. (2011). Closing the Gap in Mammogram Screening: An Experimental Intervention Among Low-Income Hispanic Women in Community Health Clinics. Health Education & Behavior 38(5), 452-461. DeBiasi, R. L. (2011). West Nile Virus Neuroinvasive Disease. Curr Infect Dis Rep, 13, 350-359. 205 Debate, R. D., Plichta, S. B., Tedesco, L. A., Kerschbaum, W. E. (2006). Integration of Oral Health Care and Mental Health Services: Dental Hygienists’ Readiness and Capacity for Secondary Prevention of Eating Disorders. J Behav Health Serv and Research, 113-125. Durham, D. P., Casman, E. A., Albert, S. M. (2012). Deriving behavior model parameters from survey data: self-protective behavior adoption during the 2009-2010 influenza A (H1N1) pandemic. Risk Anal, 32(12), 2020-2031. Eichler, E. A. (2011) Public Attitudes, Knowledge, and Practices on West Nile Virus. (Master’s Thesis). Retrieved from ProQuest Dissertations and Theses (Accession #). Elder, J. P., Ayala, G. X., Harris, S. (1999). Theories and Intervention Approaches to Health-Behavior Change in Primary Care. Am J Prev Med, 17(4), 275-284. Elliott, S. J., Loeb, M., Harrington, D., Eyles, J. (2008). Heeding the Message? Determinants of Risk Behaviours for West Nile Virus. Can J Pub Health, 99(2), 137-141. Elmieh, N. (2009). Public Health Responses to West Nile Virus: The Role of Risk Perceptions and Behavioral Uncertainty in Risk Communication and Policy. Retrieved from ProQuest Dissertations and Theses. Farley, S. D., Stasson, M. F. (2003). Relative Influences of Affect and Cognition on Behavior: Are Feelings or Beliefs More Related to Blood Donation Intentions? Exper Psych, 50(1), 55-62. Finfgeld, D. L., Wongvatunyu, S., Conn, V. S., Grando, V. T., Russell, C. L. (2003). Health Belief Model and Reversal Theory: A Comparative Analysis. J of Advanced Nurs; 43(3), 288-297. Flood E., Rousculp M. D., Ryan, K. J., Beusterien, K. M., Divino, V. M., Toback, S., . . . Mahadevia, P. J. (2010). Parents’ Decision-Making Regarding Vaccinating their Children Against Influenza: A Web-based Survey. Clin Therapeutics, 32(8), 1448-1467. Fox, M. H., Averett, E., Hansen, G., Neuberger, J. S. (2006). The Effect of Health Communications on a Statewide West Nile Virus Public Health Education Campaign. Am J Health Behav, 30(5), 483-494. Freimuth, V. S., Hovick, S. R. (2012) Cognitive and Emotional Health Risk Perceptions Among People Living in Poverty. J Health Comm, 17, 303-318. Fulton, J. P., Buechner, J. S., Scott, H. D., DeBuono, B. A., Feldman, J. P., Smith, R. A., Kovenock, D. A. (1991). Study Guided by the Health Belief Model of the 206 Predictors of Breast Cancer Screening of Women Ages 40 and Older. Pub Health Reports, 106(4), 410-420. Galloway, R. (2003). Health Promotion: Causes, Beliefs, and Measurement. Clin Med and Res, 1(3), 249-253. Gargano, L. M., Painter, J. E., Sales, J. M., Morfaw, C., Jones, L. M., Murray, D., . . . Hughes, J. M. (2011). Seasonal and 2009 H1N1 influenza vaccine uptake, predictors of vaccination and self-reported barriers to vaccination among secondary school teachers and staff. Human Vaccines, 7, 89-95. Glanz K, Rimer B, Viswanath K. (2008). Health Behavior and Health Education: Theory, Research and Practice. (4th ed.). San Francisco, CA: Jossey-Bass. Gray, L., MacDonald, C., Mackie, B., Payton, D., Johnston, D., Baker, M. J. (2012) Community Response to Communication Campaigns for Influenza A(H1N1): A Focus Group Study. BMC Public Health, 12(205), 1-12. Greene, K., & Brinn, L. S. (2003). Messages Influencing College Women’s Tanning Bed Use: Statistical versus Narrative Evidence Format and a Self-Assessment to Increase Perceived Susceptibility. J Health Comm, 8: 443-461. Gubler, D. J., Reiter, P., Ebi, K. L., Yap, W., Nasci, R., Patz, J. A. (2001). Climate Variability and Change in the United States: Potential Impacts on Vector- and Rodent-Borne Diseases. Environmental Health Perspectives, 109, 223-233. Gujral, I., Zielinski-Guttierez, E., LeBailly A, Nasci R. (2007). Behavioral Risks for West Nile Virus Disease. Emerging Infectious Diseases, 13(3), 419-425. Hall, K. S. (2012). The Health Belief Model Can Guide Modern Contraceptive Behavior Research and Practice. J Midwifery and Women’s Health, 57, 74-81. Harrison, J. A., Mullen, P. D., Green, L. (1992). A meta-analysis of the health belief model with adults. Health Education Research, 7, 107-116. Doi:10.1093/her/7.1.107 Hayes, E. B., Piesman, J. (2003). How can we prevent Lyme disease? New England Journal of Medicine, 348, 2424-30. Hayes, E. B., Komar, N., Nasci, R. S., Montgomery, S. P., O’Leary, D. R., Campbell, G. L. (2005). Epidemiology and Transmission of West Nile Virus Disease. Emerging Infectious Diseases, 11(8), 1167-73. Hayes, E. B., Sejvar, J. J., Zaki, S. R., Lanciotti R. S., Bode, A. V., Campbell, G. L. (2005). Virology, Pathology, and Clinical Manifestations of West Nile Virus Disease. Emerging Infectious Diseases, 11(8), 1174-79. 207 Henry, L. (2004). West Nile Virus: Knowledge, Behaviors, Attitudes Among Seniors, Denton County, Texas. Herrington, J. E., Campbell, G. L., Bailey, R. E., Cartter, M. L., Adams, M., Frazier, E.L., . . . Gensheimer, K. F. (1997). Predisposing Factors for Individuals’ Lyme Disease Prevention Practices: Connecticut, Maine, and Montana. Public Health Briefs, 87, 2035-2038. Herrington, J. (2003). Pre-West Nile Virus Outbreak: Perceptions and Practices to Prevent Mosquito Bites and Viral Encephalitis in the United States. Vector-borne and Zoonotic Diseases 3(4), 157-173. Herrington, J. E. (2004). Risk Perceptions Regarding Ticks and Lyme Disease: A National Survey; Am J Prev Med, 26(2), 135-140. Hochbaum, G. M. (1958). Public Participation in Medical Screening Programs: A Socio- Psychological Study. Washington, D.C.: U.S. Dept. of Health, Education, and Welfare. Hosmer, D. W. & Lemeshow, S. (2000). Applied Logistic Regression, 2nd Ed. John Wiley and Sons, Inc., Hoboken, NJ. Janssen, E., Waters, E. A, van Osch, L., Lechner, L., de Vries, H. (2014). The Importance of Affectively-laden Beliefs About Health Risks: the Case of Tobacco Use and Sun Protection. J Behav Med, 37, 11-21. Janz, N. K., Becker, M. H. (1984). The Health Belief Model: A Decade Later. Health Education Quarterly, 1(1), 1-47. Jones, C. L., Jensen, J. D., Scherr, C. L., Brown, N. R., Christy, K., Weaver, J. (2014). The Health Belief Model as an Explanatory Framework in Communication Research: Exploring Parallel, Serial, and Moderated Mediation. Health Communication. DOI: 10.1080/10410236.2013.873363 Keer, M., van den Putte, B., Neijens, P. (2010). The Role of Affect and Cognition in Health Decision Making. Br J Soc Psych, 49, 143-153. Kiehn, L, Murphy, K. E., Yudin, M. H., Loeb, M. (2008). Self-reported Protective Behavior Against West Nile Virus Among Pregnant Women in Toronto. J Obstet Gynaecol Can, 30(12), 1103-1109. Krawczyk, A. L., Perez, S., Lau, E, Holcroft, C. A., Amsel, R., Knauper, B., Rosberger, Z. (2012). Human papillomavirus vaccination intention and uptake among college women. Health Psychology, 31(5), 685-693. 208 LaBeaud, A. D, Kile, J. R., Kippes, C., King, C. (2007). Exposure to West Nile Virus during the 2002 Epidemic in Cuyahoga County, Ohio: A Comparison of Pediatric and Adult Behaviors. Public Health Reports, 122, 356-61. Lawton, R., Conner, M., McEachan, R. (2009). Desire or Reason: Predicting Health Behaviors from Affective and Cognitive Attitudes. Health Psych, 28(1), 56-65. LeDang, H., Nuberg, L. E., Bruwer, J. (2014). Farmers’ perceived risks of climate change and influencing factors: a study in the Mekong Delta, Vietnam. Environ Manage, 54, 331-345. Leguen P. (2003). Current Knowledge, Practices, and Attitudes of Miami-Dade Residents about West Nile Virus Disease and its Prevention. LePrevost, C. E., Blanchard, M. R., Cope, W. G. (2011). The Pesticide Risk Beliefs Inventory: A Quantitative Instrument for the Assessment of Beliefs about Pesticide Risks. Int J Environ Res Public Health, 8, 1923-1935. Liao, Q., Cowling. B. J., Lam, W. W., Ng, D. M., Fielding, R. (2014). Anxiety, Worry, and Cognitive Risk Estimate in Relation to Protective Behaviors during the 2009 Influenza A/H1N1 Pandemic in Hong Kong: Ten Cross-sectional Surveys. BMC Infectious Diseases, 14(169), 1-11. Lin, P., Simoni, J. M., Zemon, V. (2005). The Health Belief Model, Sexual Behaviors, and HIV Risk among Taiwanese Immigrants. AIDS Educ and Prev, 17(5), 469- 483. Lindsey, N. P., Staples, J. E., Lehman, J. A., Fischer, M. (2010). Surveillance for Human West Nile Virus Disease—United States, 1999-2008. CDC Morbidity and Mortality Weekly Report, 59(SS-2), 1-18. Lindsey, N. P., Sejvar, J. J., Bode, A. V., Pape, J., Campbell, G. (2012). Delayed Mortality in a Cohort of Persons Hospitalized with West Nile Virus Disease in Colorado in 2003. Vector-borne and Zoonotic Diseases, 12(3), 230-235. Lindsey, N. P., Lehman, J. A., Staples, J. E., Fischer, M. West Nile Virus and Other Arboviral Diseases—United States, 2013. (2014). CDC Morbidity and Mortality Weekly Report, 63(24), 521-526. Loeb, M., Elliott, S. J., Gibson, B., Fearon, M., Nosal, R., Drebot, M., . . . Eyles J. (2005). Protective Behavior and West Nile Virus Risk. Emerging Infectious Diseases, 11(9), 1143-46. 209 Lollis, C. M., Johnson, E. H., Antoni, M. H. (1997). The efficacy of the health belief model for predicting condom usage and risky sexual practices in university students. AIDS Educ Prev, 9(6), 551-63. Mahoney, C. A., Thombs, D. L., Ford, O. J. (1995). Health Belief and Self-efficacy Models: Their Utility in Explaining College Student Condom Use. AIDS Educ and Prev 7(1), 32-49. “Maryland Arbovirus Surveillance Results, 2012” Maryland Department of Health and Mental Hygiene. Accessed December 16, 2012. Maryland Department of Agriculture. (2006). Maryland Mosquito Control Survey. Mattson, M. (1999). Toward a Reconceptualization of Communication Cues to Action in the Health Belief Model: HIV Test Counseling. Communication Monographs. 66, 240-265. McCarthy, T. A., Hadler, J. L., Julian, K., Walsh, S. J., Biggerstaff, B. J., Hinten, S. R., . . . Petersen LR. (2001). West Nile Virus Serosurvey and Assessment of Personal Prevention Efforts in Area with Intense Epizootic Activity. Ann NY Acad Sciences, 951, 307-316. McCauley, M., Minsky, S., Viswanath, K. (2013). The H1N1 pandemic: media frames, stigmatization and coping. BMC Public Health, 13:1116, doi: 10.1186/1471- 2458-13-1116. McQueen, A., Vernon, S. W., Rothman, A. J., Norman, G. J., Myers, R. E., Tilley, B. C. (2010). Examining The Role of Perceived Susceptibility on Colorectal Cancer Screening Intention and Behavior. Ann Behav Med, 40(2), 205-217. Michel, G., Kuehni, C. E., Rebholz, C. E., Zimmerman, K., Eiser, C., Rueegg, C. S., von der Weid, N. (2011). Can health beliefs help in explaining attendance to follow- up care? The Swiss Childhood Cancer Survivor Study. Psycho-Oncology, 20, 1034-1043. Molaei, G., Andreadis, T. G., Armstrong, P. M., Anderson, J. F., Vossbrinck, C. R. (2006). Host Feeding Patterns of Culex Mosquitoes and West Nile Virus Transmission, Northeastern United States. EID 12(3), 468-474. Morin, A., LeMaitre, T., Farrands, A., Carrier, N., Gagneur, A. (2012). Maternal knowledge, attitudes, and beliefs regarding gastroenteritis and rotavirus vaccine before implementing a vaccination program: Which key messages in light of a new immunization program? Vaccine, 30, 5921-5927. 210 Mosher, C. E., Lipkus, I. M., Sloane, R. Kraus, W. E., Snyder, D. C., Peterson, B. . . . Demark-Wahnefried. (2008). Cancer Survivors’ Health Worries and Associations with Lifestyle Practices. J Health Psychol, 13(8), 1105-1112. doi: 10.1177/1359105308095964. Murphy, C. C., Vernon, S. W., Diamond, P. M., Tiro, J. A. (2014). Competitive Testing of Health Behavior Theories: How Do Benefits, Barriers, Subjective Norm, and Intention Influence Mammography Behavior? Ann Behav Med, 47, 120-129. Murray, K. O., Mertens, E., Despres, P. (2012). West Nile Virus and Its Emergence in the United States of America. Veterinary Research, 41(67), 1-14. Murray, K. O., Walker, C., Gould, E. (2011). The virology, epidemiology, and clinical impact of West Nile virus: a decade of advancements in research since its introduction into the Western Hemisphere. Epidemiol Infect, 139, 807-817. Nan, X. (2011). Communicating to Young Adults about HPV Vaccination: Consideration of Message Framing, Motivation, and Gender. Health Comm, 27(1), 10-18. National Cancer Institute (2005a). Theory At A Glance: A Guide for Health Promotion Practice (2nd Ed.) NIH Publication No. 05-3896: US Department of Health and Human Services. Norman, P., Brain, K. (2005). An application of an extended health belief model to prediction of breast self-examination among women with a family history of breast cancer. Br J Health Psych, 10, 1-16. Painter, J. E., Borba, C., Hynes, M., Mays, D., Glanz, K. (2008). Use of Theory in Health Behavior Research from 2000 to 2005: A Systematic Review. Ann Behav Med, 35: 358-362. Pasick, R. J., Burke, N. J. (2008). A Critical Review of Theory in Breast Cancer Screening Promotion Across Cultures. Ann Rev Pub Health, 29, 351-368. Petersen, L. R., Fischer, M. (2012). Unpredictable and Difficult to Control—The Adolescence of West Nile Virus. N Engl J Med, 367, 1281-1284. Phuanukoonan, S., Brough, M., Bryan, J. H. (2006). Folk Knowledge about dengue mosquitoes and contributions of the Health Belief Model in Dengue control Promotion Northeast Thailand. Acta Tropica, 99, 6-14. Pinto, B. M., Floyd, A. (2008). Theories Underlying Health Promotion Interventions Among Cancer Survivors. Seminars in Oncol Nursing, 24(3), 153-163. 211 Poss, J. E. (2001). Developing a New Model for Cross-Cultural Research: Synthesizing the Health Belief Model and the Theory of Reasoned Action. Advanced Nursing Science, 23, 1-15. Prati, G., Pietrantoni, L., Zani, B. (2011). Compliance with recommendations for pandemic influenza H1N1 2009: the role of trust and personal beliefs. Health Education Research, 26, 761-769. Quine, C. P., Barnett, J., Dobson, A. D., Marcu, A., Marzano, M., Moseley, D., . . . Uzzell, D. (2011). Frameworks for risk communication and disease management: the case of Lyme disease and countryside users. Philosophical Transactions of the Royal Society B, 366, 2010-2022. Reid, A. E., & Aiken, L. S. (2011). Integration of Five Health Behaviour Models: Common Strengths and Unique Contributions to Understanding Condom Use. Psychology and Health, 26(11), 1499-1520. Reiter, P. L., Brewer, N. T., Gottlieb, S. L., McRee, A., Smith, J.S. (2009). Parental health beliefs and HPV vaccination of their adolescent daughters. Social Science and Medicine, 1-6, doi: 10.1016/j.socscimed.2009.05.024. Renner, B., Reuter, T. (2012). Predicting vaccination using numerical and affective risk perceptions: The case of A/H1N1 influenza. Vaccine, 30, 7019-7026. Reuter, T., Renner, B. (2011). Who Takes Precautionary Action in the Face of the New H1N1 Influenza? Prediction of Who Collects a Free Hand Sanitizer Using a Health Behavior Model. PLOS One; 6(7): e22130. doi:10.1371/journal.pone.0022130. Rios-Gonzalez, A., Jansen, K., Javier, H. (2013). Pesticide risk perceptions and the differences between farmers and extensionists: towards a knowledge-in-context model. Environmental Research, 124, 43-53. Roden, J. (2004). Revisiting the Health Belief Model: Nurses applying it to young families and their health promotion needs. Nurs and Health Sciences, 6, 1-10. Rosenstock, I. M. (1960). What Research in Motivation Suggests for Public Health. American Journal of Public Health, 50, 295–302. Rosenstock, I. M. (1974). The Health Belief Model and Preventive Health Behavior. Health Education Monographs, 2(4), 354–386. Rosenstock, I. M. (1974) Historical Origins of the Health Belief Model. In M. H. Becker (Ed.) The Health Belief Model and Personal Health Behavior. (pp. 1-8). Thorofare, NJ: Charles B. Slack. 212 Rosenstock, I. M., Strecher, V. J., Becker, M. H. (1988). Social Learning Theory and the Health Belief Model. Health Educ Q, 15(2), 175-83. Schellenberg, T. L., Anderson, M. F., Drebot, M. A., Vooght, M., Findlater, A. R., Curry, P. S., . . . Osei, W. D. (2006). Seroprevalence of West Nile Virus in Saskatchewan’s Five Hills Health Region, 2003. Can J Pub Health, 97(5), 369- 373. Schofield, I., Kerr, S., Tolson D. (2007). An exploration of the smoking-related health beliefs of older people with chronic obstructive pulmonary disease. J Clin Nursing, 16, 1726-1735. Sejvar, J. J., Lindsey, N. P., Campbell, G. L. (2011). Primary Causes of Death in Reported Cases of Fatal West Nile Fever, United States, 2002-2006. Vector-borne and Zoonotic Diseases,11(2), 161-164. Senay, I., Alford, S. H., Kaphingst, K. A. (2013). Judging Risk for Multiple Diseases: The Role of Disease Worry. J Health Psychol, 18(4), 554-560. Smithburn, J. S., Hughes, T. P., Burke, A. W., Paul, J. H. (1940). A neurotropic virus isolated from the blood of a native of Uganda. Am J Trop Med Hyg, 20, 471–92. Swaim, R. A., Barner, J. C., Brown, C. M. (2008). The relationship of calcium intake and exercise to osteoporosis health beliefs in postmenopausal women. Res in Social and Admin Pharm, 4, 153-163. Tavafian, S. S. (2012). Predictors of Cervical Cancer Screening: An Application of Health Belief Model, Topics on Cervical Cancer with an Advocacy for Prevention. Dr. R. Rajamanickam (Ed.), ISBN: 978-953-51-0183-3. Doi: 10.5772/27886. InTech, Available from: http://www.intechopen.com/books/topics-on-cervical-cancer-with-anadvocacy- for-prevention/predictors-of-cervical-cancer-screening-an-application-of-health-belief- model Tuiten, W., Koenraadt, C., McComas, K., Harrington, L. C. (2009). The Effect of West Nile Virus Perceptions and Knowledge on Protective Behavior and Mosquito Breeding in Residential Yards in Upstate New York. EcoHealth, 6, 42-51. University of Twente. Health Belief Model. http://www.utwente.nl/cw/theorieenoverzicht/Theory%20Clusters/Health%20Co mmunication/Health_Belief_Model.doc/ . Accessed March 2, 2013. Van der Bijl, J. J. (2001). The Theory and Measurement of the Self-Efficacy Construct. Schol Inquiry for Nursing Practice, 15(3), 189-207. 213 Walker, R. (1999). Models in Health Psychology: An Introduction. J Diab Nursing 13(6), 188-191. Wang, S., Charron-Prochownik, D., Sereika, S. M., Siminerio, L., Kim, Y. (2006). Comparing three theories in predicting reproductive health behavioral intention in adolescent women with diabetes. Pediatric Diabetes, 7, 108-115. Wei, J., Hansen, A., Zhang, Y., Li, H., Liu, Q., Sun, Y., Bi, P. (2014). Perception attitude, and behavior in relation to climate change: A survey among CDC health professionals in Shanxi province, China. Environmental Research, 134C, 301- 308. Whittingham, M. J., Stephens, P. A., Bradbury, R. B., Freckleton, R. P. (2006). Why do We Still Use Stepwise Modeling in Ecology and Behaviour? J Anim Ecol, 75(5), 1182-89. Wieland, M. L., Weis, J. A., Yawn, B. P., Sullivan, S. M., Millington, K. L., Smith, C.M., . . . Sia, I. G. (2012). Perceptions of Tuberculosis Among Immigrants and Refugees at an Adult Education Center: A Community-Based Participatory Research Approach. Journal of Immigrant and Minority Health, 14, 14-22. Wolf, R., Lewis, D., Cochran, R., Richards, C. (2008). Nursing Staff Perceptions of Methicillin-Resistant Staphylococcus aureus and Infection Control in a Long- Term Care Facility. Am Med Dir Assoc, 9, 342-346. Wong, L. P., AbuBakar, S. (2013). Health Beliefs and Practices Related to Dengue Fever: A Focus Group Study. PLOS Neglected Tropical Diseases, 7, e2310. doi:10.1371/journal.pntd.0002310. Yarbrough, S., Braden, C. J. (2001). Utility of health belief model as a guide for explaining or predicting breast cancer screening behaviours. J Adv Nursing, 33(5), 677-688. Yerby, L. G. (2007). The Validation of a West Nile Virus Survey Based on the Health Belief Model. (Doctoral Dissertation). Retrieved from ProQuest Dissertations and Theses (Accession #). Zielinski-Gutierrez E. (2003). West Nile Virus and Other Mosquito-borne Infections. The Health Care of Homeless Persons—Part I: Mosquito-borne Infections. pp. 181- 186. Zimmerman, R. S., & Vernberg, D. (1994). Models of preventative health behavior: Comparison, critique, and meta-analysi. In G. Albrecht (Ed.), Advances in medical sociology, Health behavior models: A reformulation (pp. 45-67). Greenwich, CT: JAI Press.