Perceptual Influences on Self-protective Behavior for West Nile Virus, a Survey in Colorado, USA

Craig W. Trumbo; Raquel Harper


BMC Public Health. 2015;15(557) 

In This Article


Study Aims

The first aim of this study was to predict self-protectiuve behvior using the HBM along with an additional set of risk measures. Of the three aditional perspectives examined, the measures based on the cognitive-affective risk percetpion model performed best, with the affective component being clearly superior (dread, anxiety, worry). It is not clear why the cognitive component of this measure presented weaker or directionally opposite results. It seems fairly clear that the affective measures were more effective as a group (certainly more reliable). The three items developed for the cognitive aspect of the measure (knowledge, control, increasing risk) were more difficult to construct, as the cognitive aspects of risk perception that make sense in the context of a technology (their traditional contextual application) did not lend themselves as readily to the risk associated with a vector-borne infectious disease. The inclusion of the item on whether or not WNv is seen to be increasing in prevalence may have also tapped more of an affective response.

The success of the other two risk perception constructs is more provisional. While the ecological measure does have some utility, it was not sufficiently robust to remain in the model alongside the other measures. Further analysis of this measure is certainly warranted to examine its potential use with some more discrete aspects of the data set. Its bivariate relationships with susceptibility and severity suggest it may be useful in other approaches using the HBM. Also, casual examination of other aspects of ecology suggested interesting avenues for future analysis. For example, there was a significant difference in means across ethnicity on the ecology score, with Hispanic/Latinos showing a greater risk response based on the ecological factors.

The second aim of the study was to specifically examine the significance of ethnicity within a trimmed, hybrid model. While this is a single study with a limited sample, the findings do support previous work on WNv showing differential response by Hispanic/Latino persons. Importantly, we see that Hispanic/Latinos were more likely to practice self-protective behavior. This is likely especially motivated by their greater perception of risk/susceptibility and greater exposure to information cues to action.

Implications for Practice

One of the important aspects of these findings for public health practitioners resides in the role of affect in individual behavior. While an understanding of the ecological dimension of WNv and direct experience with risk indicators such as dead birds can play a role in shaping public response, the strongest motivator for protective behavior remains an emotional response to the threat. This may suggest that public health communications on WNv may not benefit as strongly from fact-based messages as they might from more emotional treatment, perhaps especially through the use of personalized narratives.

The three significant variables from the HBM also offer further insight into potential approaches for public health practice. The perception of individual susceptibility points to an opening to gain the public's attention on the importance of avoiding mosquitoes during a WNv outbreak, and the influence of perceived benefits suggests a reinforcing dimension to that line of persuasion: you are susceptible and you will benefit from these actions. Underscoring this most importantly is the final inclusion of information effects, cues to action, that show that messaging on this topic may influence behavior, although the causal ordering cannot be empirically supported here.

Differences on ethnicity are also important. The differences seen in specific information sources reported by Hispanic/Latinos illustrates some possibilities for practice. Greater exposure to television messages may suggest a higher level of attentiveness to the topic. More exposure to information from doctors may suggest either an outreach effort by physicians to inform a more vulnerable population, or more exposure to such information due to a greater likelihood of needing to seek diagnosis or treatment. The stronger role of family members may be based in the cultural domain. Taken together, these findings on ethnicity point toward potentially important avenues for public health message tailoring.

Given the sometimes controversial nature of mosquito control and news media treatment of health risks (and WNv in particular) it is worth mentioning the topic of media effects. As reported above, respondents indicated greater exposure to the three mediated cues as compared to the three personal cues. It is tempting to speculate that some aspects of attitude and behavior might be affected by intense or "hyped" media reporting. However, we did not monitor the media environment prior to or during the study period and do not have specific media exposure measures that would allow us to examine such a direct effect. It's also well established that media don't affect attitudes or behaviors in a direct fashion, but rather interact with and flow through interpersonal sources. Our study was simply not designed to examine this phenomenon.

In a broader sense it is also worthwhile to conclude with a perspective on mosquito-vectored diseases generally, not only WNv. It is becoming clear that changes in global climate will bring increased threat from mosquito vectored diseases. This is anticipated for WNv as well as other diseases, most prominently dengue and Chikungunya virus as the Asian tiger mosquito expands its range into North America.[25,26] Mosquito protection will be an increasingly salient topic for public health communicators in the coming years.


The measures used for the three exploratory scales, while based on extant literature, were assembled ad hoc. More rigorous item development through focus groups and survey pre-testing would have certainly yielded superior measures. This may be especially of concern for the scales measuring cognitive risk perception and proximity risk perception, as the reliabilities were low. The sampling strategy was successful in providing a more representative capture of ethnicity that likely would have been the case for a non-proportional approach. While the survey response rate of 49 % is generally considered acceptable for self-administered mail surveys it does nonetheless represent a limitation, especially as significant differences were found between sample values and Census figures. This may be a notable hindrance in further analyses of these data focusing on ethnicity, as the data only contain 74 cases for the Hispanic/Latino segment. Finally, the study design precludes the basis for causal interpretations.