The Effect of West Nile Virus Perceptions and Knowledge on Protective Behavior and Mosquito Breeding in Residential Yards in Upstate New York
- First Online:
- Cite this article as:
- Tuiten, W., Koenraadt, C.J.M., McComas, K. et al. EcoHealth (2009) 6: 42. doi:10.1007/s10393-009-0219-z
- 212 Downloads
A knowledge, attitudes, and practices (KAP) questionnaire combined with entomological surveys of residential mosquito-breeding sites were conducted in two Upstate New York neighborhoods. We tested the hypothesis that “correct” West Nile virus (WNV) knowledge and perceptions correspond with the use of practices that prevent mosquitoes from breeding and biting. Our results demonstrate that perceptions of WNV relate to the number of positive containers in yards and the use of mosquito preventive measures. In contrast, WNV knowledge was not related. Culex pipiens and Cx. restuans were common species found breeding in containers. Aedes japonicus was the most abundant species in 77% of positive containers (buckets, flower pots, and birdbaths). This new, invasive mosquito together with the Culex species identified in this study represent significant potential as vectors of WNV and other arboviruses affecting human and animal health. We conclude that more training and education programs should focus on WNV control strategies and recognizing mosquito breeding in residential yards. This is the first study to directly investigate the relationship between KAP and breeding of WNV vectors in residential yards.
Keywordsvector-borne diseaseWest Nile virusAedes japonicusCulexmosquito control perceptionhuman health behavior
Since West Nile virus (WNV) was first recognized in New York City in 1999, it has spread across North America and emerged as an important public-health problem (CDC, 2008). Many species of mosquitoes are involved in the transmission of the virus, but currently members of the Culex pipiens complex, Culex restuans, and Culex tarsalis are thought to be responsible for the majority of transmission from birds to humans (Kilpatrick et al., 2005; Turell et al., 2005). Because an effective human vaccine is not available, WNV control is limited to reduction of mosquito vector populations and public education (such as promotion of personal protective behaviors).
It is not clear to what extent United States residents are aware of mosquito breeding on their property, or what strategies they use to protect themselves and their families. In addition, reliable information on the contribution of peridomestic breeding sites to mosquito production and WNV risk is lacking. Although a few studies have reported on North Americans’ knowledge, attitudes, and practices about WNV (Wilson et al., 2005; Herrington, 2003; Fox et al., 2006; CDC, 2003; Aquino et al., 2004; Averett et al., 2005; Elliot et al., 2008), none of these studies combine entomological and behavioral determinants into one model.
In this study, we tested the hypothesis that people with greater knowledge and concern about WNV would be more likely to take actions to reduce sources of mosquito breeding on their property and minimize mosquito exposure. In addition, we wanted to identify the most common water-holding container types and key vector mosquitoes in residential yards. By combining behavioral and entomological determinants, the impact of human behavior and attitudes on WNV vector populations could be investigated more fully and options for behavioral change identified.
Materials and Methods
Our study was conducted in two neighborhoods in an upstate New York community where WNV has been isolated from mosquitoes and birds since 2000 but no human infections have occurred to date (New York State Department of Health 2008). One neighborhood was situated in a residential area in downtown Ithaca, NY (42°26′50″N, 76°29′52″W) with houses located close to each other and with small- to medium-sized yards ranging from 186 to 708 (mean, 409) m2. The second neighborhood was situated in a suburban area at the edge of Ithaca, NY (42°25′44″N, 76°25′06″W) with houses farther apart, with large yards (mean, 6,334 (range, 232–18,581) m2) surrounded by agricultural lands and forest. The two study areas were separated by 4.5 km. We selected these geographically different neighborhoods (urban vs. suburban) because of their different-sized yards, which we predicted could have implications for the number of water-holding containers in each yard. All houses in the neighborhoods were included in the study.
A cross-sectional knowledge, attitudes, and practices (KAP) questionnaire was designed to collect data on demographic characteristics, knowledge of WNV, attitudes, and perceptions toward the disease and practices used to avoid or reduce exposure to mosquito bites. A cross-sectional entomological survey was designed to collect data simultaneously on mosquito-breeding places for WNV vectors in residential yards. We compiled KAP survey results and data from entomological surveys to test the hypothesis that “correct” WNV knowledge and perceptions would correspond to the use of practices designed to prevent mosquito breeding and biting. Both surveys were performed between August 1 and September 13, 2006. Approval was obtained from Cornell University Committee on Human Subjects (Protocol # 06-07-017) before the start of the study. The questionnaire was tested for clarity among a pilot group of volunteers before beginning the actual survey. Before our research teams visited the 237 selected households, we notified residents of our study with a letter in which we explained that researchers would come to their houses to ask them to participate in this study by answering a 10-minute questionnaire and to request permission to check their yards. To avoid the possibility that this notification would influence their mosquito control practices, we did not mention mosquitoes. Two to three teams, each consisting of two trained personnel, stopped at houses in the selected areas and, after obtaining consent, conducted interviews and mosquito surveys. Participation for all residents was voluntary. The teams conducted interviews with one consenting adult member of a household during daytime hours (between 10:00 a.m. and 4:00 p.m.). If no one was home at a residence, survey teams returned one more time during the early evening (between 5:00 p.m. and 8:00 p.m.).
Knowledge of WNV was measured by asking questions about the mosquito vector, common infected animals, disease symptoms, risk groups, and preventive measures. Attitudes and perceptions were measured by asking questions about the importance of the WNV problem, probability of getting infected, seriousness of the disease, and responsibility for controlling its vector. Additionally, questions were asked about the use of preventive measures against mosquito bites. Questions were asked using a standard questionnaire with questions derived from existing KAP questionnaires (Koenraadt et al., 2006; Averett et al., 2005).
After each interview, teams secured permission to survey the respondent’s yard for potential mosquito-breeding sites. All water-holding containers and other sources of standing water in the respondent’s yard were inspected for larvae and pupae and described on entomologic survey forms. If larvae were encountered, a subsample was removed and transported to the laboratory for identification to species and staging. Container variables, including size, water temperature, volume of water, and cover status (covered vs. uncovered), were recorded. Larval surveys lasted an average of 16 (range, 3–50) minutes per residence and varied depending on the size of the yard and the number of potential breeding habitats. In the laboratory, the larval samples were identified to species and staged based on published keys (Andreadis et al., 2005; Means, 1987).
Entomologic and KAP data were entered in Microsoft Office Access 2003 (Microsoft Corporation, Redmond, WA). Respondents who answered survey questions correctly scored points for knowledge of WNV. A person was considered knowledgeable about WNV disease risk when he/she mentioned that older people (>55 years) and/or immune-compromised people were most at risk for having severe WNV symptoms (Petersen and Marfin, 2002). We also considered a person knowledgeable of the WNV vector if he or she mentioned that mosquitoes were important in WNV transmission. A person was considered knowledgeable of preventive measures if he or she mentioned at least one of the following preventative practices: eliminate standing water, use insect repellent, wear protective clothing, avoid outdoor areas with mosquitoes, stay indoors at dusk and dawn, use screening, and spray pesticides. The sum of all the points received by the respondent for answers regarding disease risk, vector, and preventative measures determined the overall knowledge score (see also Table 5). Thus, each respondent obtained a score between 0 and 3. A respondent was considered to have low knowledge of WNV when he or she scored 0, 1, or 2 points, and high knowledge when he/she scored 3 points.
Practices to prevent WNV were divided into two strategies: mosquito-breeding prevention practice, and mosquito-bite reduction practice. Successful practices to prevent mosquito breeding included the following: eliminating or changing standing water, covering containers, using larval insecticides, and using biological control. On a household level, mosquito-breeding prevention practice was measured by the presence of positive mosquito-breeding habitats. Mosquito-bite reduction practice was defined as using at least one of the following measures: wearing protective clothing, avoiding outdoor activity, and applying insect repellent containing DEET or Picaridin.
To analyze the relationships between demographic background, knowledge, attitude and perceptions, practice, and presence of larvae/pupae, we used binary logistic regression procedures (SPSS version 14.0; SPSS Inc., Chicago, IL). Factors having a screening significance of P < 0.25 in the bivariate analysis were included in the multivariate analysis (Hosmer and Lemeshow, 1989). This analysis also included all possible two-way interactions in the full model. The final model was selected by first excluding all two-way interactions. If there was no significant loss of fit as evaluated by comparing -2 log likelihoods (χ2 distribution) afterward, the least significant factor was excluded until the model lost significant information compared with the previous model. An association in both bivariate and multivariate analysis was considered to be significant when P < 0.05.
Demographic Characteristics of 97 Respondents in Two Areas of Ithaca, NY, USA
Urban n (%)
Suburban n (%)
Total n (%)
High school or less
Owner or renter
NY state resident (yr)
Knowledge of WNV High-Risk Groups, Vector, and Preventive Measures
Urban n (%)
Suburban n (%)
Total n (%)
Risk groups (N = 97)
Elderly >55 yr
Vector (N = 97)
Breeding places (N = 80)
Commonly infected animals (N = 88)
Dogs and cats
Preventive measures (N = 97)
Eliminate standing water
Avoid outdoor areas with mosquitoes
Stay indoors at dusk and dawn
Associations Among Demographics and WNV Knowledge in Ithaca, NY
Percentage with high knowledgea of WNV
95% confidence limits
High school or less
NY state resident (yr)
Twenty-two percent of the respondents considered WNV to be an “important” to “very important” problem to them personally, whereas 13% regarded WNV as an “important” to “very important” problem to their community. A minority of the respondents (8%) thought it would be “likely” to “very likely” that WNV would harm a member of their family, but if a person gets infected with WNV, 41% thought that it would be “serious” to “very serious.”
Respondents mentioned the newspaper (70%) most often as the main source of information about WNV, followed by television (36%), radio (26%), and the Internet (20%). When we asked respondents to name their most-trusted source for WNV information, the newspaper remained their preferred choice (25%), followed closely by scientific sources (scientific literature and researchers) (24%), their personal doctor (15%), and the Internet (12%). Although popular media (television and radio) were considered significant sources of information about WNV, only 7% and 6% of the respondents, respectively, trusted these sources. Respondents most often mentioned themselves as responsible (54%) for mosquito control, followed by the government (21%) and the health department (10%). The majority of respondents reported that the level of mosquito activity near their own home was low (70%). Thirteen percent described the activity as high, and the remainder reported it as medium (17%).
In our study, respondents claimed that regularly or occasionally eliminating standing water (54%) was the most common practice they used to prevent WNV, followed by wearing protective clothing (34%) and avoiding the outdoors when mosquitoes are active (21%). Of the respondents who reported that they did not regularly eliminate standing water in their yards (55%), the majority explained that they were unaware of any standing water in their yards, or they considered it too much work or too difficult. Respondents blamed warm temperatures (during mosquito peak season) and the lack of mosquito nuisance for not using protective clothing. Of the respondents who used insect repellents occasionally or regularly, 72% used repellent containing DEET, which is recommended as active ingredient by CDC (CDC, 2005). However, most people did have concerns about using repellent (70%) with or without DEET. Concerns included fear of repellent toxicity on skin and bad odor of repellents. All respondents had screening on their windows.
Mosquito Habitat Survey
Entomological Indices of Two Areas in Ithaca, NY
Positive containers per house
Impact of Knowledge and Perceptions on Practices
To link the data of the entomological survey with the KAP survey, we assumed that knowledge and attitudes about WNV of the respondent were representative for the household as a whole. In the survey we interviewed one adult person (>18 years of age) and his/her knowledge and perceptions were considered to be reflective of the container management decisions of the household.
Relationship of Demographics, WNV Knowledge, and WNV Perceptions with the Presence of Mosquito-Positive Containers at Household Level in Ithaca, NY
Percentage with ≥1 positive containers
95% confidence limits
High school or less
Knowledge of risk groups
At least one risk group mentioned
No risk group mentioned
Knowledge preventive measures
≥1 preventive measure mentioned
No preventive measure mentioned
Overall WNV knowledge
WNV as personal problem
Likeliness that WNV will harm family member
Seriousness of WNV infection
Responsible for mosquito control
Level of mosquito activity
With regard to the use of measures to prevent human–mosquito contact, we also did not find a significant relationship with knowledge in the bivariate or multivariate analysis (P > 0.05; Table 5). In the bivariate analysis, the respondent’s age and perceived seriousness of a WNV infection were significantly related to the use of preventive measures. When these factors were included with owner/renter, knowledge of vectors, knowledge of preventive measures, overall knowledge, and level of mosquito activity in a multivariate model (all had screening significance; P < 0.25 (Hosmer and Lemeshow, 1989)), respondents who perceived higher levels of mosquito activity around their house were significantly more likely to use preventive measures than respondents who perceived medium (odds ratio (OR) = 0.12) or low (OR = 0.22) levels of mosquito activity. Respondents aged 55 years or older used at least one preventive measure three times more than all other age groups. If a respondent thought a WNV infection was quite serious, he or she was 2.6 times more likely to use preventive measures than respondents who did not think WNV was a serious infection.
Our study demonstrated that perceptions of WNV significantly relate to prevention practices. Respondents demonstrating concern that WNV could harm a family member had fewer positive containers on their property. Although perhaps coincidental, this finding also could indicate that they were more likely to practice container source reduction. In addition, respondents used preventive measures, such as application of repellants, protective clothing, and avoidance of outdoor activities during peak mosquito hours, when they considered WNV a serious infection or perceived high mosquito activity around their homes.
However, there was no relationship between a resident’s knowledge about WNV risk groups, vectors, and preventive measures and mosquito production on their property. In addition, we found no relationship between these variables and the respondent’s use of mosquito preventive measures. This outcome may be due to our relatively small sample size. Residents were home and allowed us to survey them in only 40% of preselected houses. Whereas telephone surveys can reach more residents and result in larger sample sizes, our study was performed with door-to-door visits of residential areas and benefits from the direct linking of behavioral data with entomological data.
Clearly, knowledge alone is not enough, as is evident from many health campaigns that seek behavior change among targeted populations. A change in perceptions and attitudes is required to establish use of preventive measures against WNV, which can lower WNV disease rates (Gujral et al., 2007). To plan effective WNV prevention and control programs, factors that influence the adoption of protective behaviors should be recognized (Elliott et al., 2008). These factors include attitude toward the behavior, the perceived control individuals have over behavior (i.e., efficacy), and whether they believe other people expect them to comply with the behavior (Ajzen, 1991). For example, although 60% of the respondents in our study knew that eliminating standing water is a good measure to reduce the risk of mosquito bites, only 22% actually eliminated water in their yards. Reasons for not eliminating water included that it was too difficult (no control), too much work, or the respondent was not bothered by mosquitoes.
Compared with other studies in the United States and Canada, knowledge of WNV in Ithaca, NY, was similar (Wilson et al., 2005, Aquino et al., 2004) or slightly lower (Fox et al., 2006; CDC, 2003; Averett et al., 2005). Two-thirds of the respondents knew correctly that individuals older than aged 55 years had a higher risk of developing serious illness from WNV, whereas almost an equal percentage of respondents thought incorrectly that the same is true for children. A similar finding was reported in a study conducted in Connecticut (CDC, 2003). Lack of knowledge about risk groups stresses the need to improve education about WNV risk. One way to accomplish this goal is to improve WNV risk messages in health educational materials disseminated to communities.
Despite reports confirming the safety of DEET (Osimitz and Grothaus, 1995), approximately half of the respondents mentioned concerns about the toxicity of DEET-based repellents. Similar results have been found in other studies (Herrington, 2003, Fox et al., 2006; Aquino et al., 2004; Averett et al., 2005). This perceived toxicity from use of insect repellent may outweigh a person’s perceived risk for infection with WNV, potentially resulting in increased exposure to WNV-infected mosquitoes. Greater emphasis is needed to promote the safety and efficacy of DEET-containing insect repellents. This information could be placed more prominently in health messages about WNV prevention.
Similar to other studies (Fox et al., 2006; Averett et al., 2005), mass media (TV, radio, and newspaper) remain the most common sources of information about WNV. However, when we asked respondents what source of information they trusted most, only a small percentage trusted mass media. Scientific- and health-related sources garnered the highest level of confidence and trust among respondents, as well as physicians and local health departments. This suggests that WNV prevention campaigns conducted through the mass media alone may not be effective. Campaigns also must work to increase communication between doctors and health officials with members of high-risk groups and their support network.
In our study, residents of suburban areas with larger yards tended to have more positive mosquito-breeding sites. This may be related to the fact that it is more work and time involved to minimize standing water over a larger area of land. In addition, it can be difficult to find all mosquito-breeding sites. We also found that residents who said they actively remove mosquito-breeding sites on their property were just as likely to have mosquito-positive containers in their yard as those who said they did nothing to remove mosquito-breeding sites.
More than half of the houses surveyed and almost half of the water-holding containers had immature stages of mosquito species that may transmit WNV as adults, including Cx. pipiens and Cx. restuans, which were responsible for the greatest WNV transmission risk in a previous study of two New York counties (Kilpatrick et al., 2005). The most abundant mosquito found in most containers was Ae. japonicus (77%), an invasive species first detected in the United States in 1998 (Peyton et al., 1999). This species has quickly invaded most regions of New York State (Kaufman et al., 2005) and is an extremely efficient vector of WNV in the laboratory with a vector competence of 93% (Turell et al., 2001). Recently, Kilpatrick et al. (2005) estimated that Ae. japonicus contributed to approximately 5% of WNV risk in regions of New York. This mosquito breeds primarily in artificial water-holding containers, such as buckets, flower pots, and birdbaths, which we found in many residential yards in both urban and suburban settings (Table 4). In addition to WNV, Ae. japonicus is a competent vector of Eastern Equine encephalitis virus, La Crosse virus, and Saint Louis encephalitis and Japanese encephalitis viruses.
The authors thank the residents of the two areas in Ithaca, NY for participating in the survey and for allowing us to collect mosquito larvae and pupae in their yards. We thank graduate and undergraduate students of Medical Entomology, Cornell University, Ithaca, NY, Rachel Hathaway, Edgar Lei, Kalpana Pathak, Rehka Reddy, Sara Woo, Bianca Chang, Rebecca Poulson, Lauren Cator, Jonathan Darbro, Alongkot Ponlawat, and Hongfei Gong for interviewing and collecting mosquito larvae. We thank Morgan C. McKenna, Chris Clarke, and Zheng Yang for their comments on the manuscript. Support for this research was provided by grants from Hatch (NYC-139410) and National Oceanic and Atmospheric Administration (NA04OAAR4310184).