One objective of this study was to compare the adoption of preventive measures by individuals within and between populations living in regions with different LD epidemiological statuses. Overall, in Neuchâtel, a high proportion of respondents (86% among those who knew of the disease before the survey) reported adoption of at least one of the main preventive measures and three out of four declared having checked for ticks after being in an area at risk for LD. With an incidence reaching 95 cases per 100 000 inhabitants [5], this is good news from a public health perspective given that removing ticks within 24 h after being bitten can reduce the risk of transmission of the bacteria to near zero [38,39]. This high level of adoption of preventive measures was not observed in the Montérégie region where the highest level of adoption for a preventive measure was found to be 50% for the use of protective clothing, among those who knew of LD before they had taken the survey, and under 20% for tick check and 35% for tick repellent. This finding may reflect a lack of knowledge about transmission and distribution of the disease, as described previously in Aenishaenslin et al. [35], given that the region is currently facing emergence of a new disease. Our findings may also suggest that despite their demonstrated efficacy, some preventive measures, such as applying acaricides on one’s property, are not popular in either Quebec or Switzerland, a finding which may also be explained by the low level of social acceptability for this specific measure. Previous studies have shown a low level of adoption for similar preventive measures both in low and high incidence regions for LD in other parts of the world [20,28,30,34,40].
Our findings suggest that the adoption of specific preventive behaviors also vary according to socio-demographic characteristics of respondents, such as gender, age and exposure levels and that the relationship between these characteristics and the adoption of preventive behaviors depends on the region and on the specific preventive measure. For example, being in a younger age group (either 18–34 or 35–54 vs 55+ year old) was positively associated with the adoption of tick repellent use but negatively associated with the use of protective clothing. Also, a higher level of adoption was noted in women for the practice of tick checks in Neuchâtel but not in Montérégie, and a higher proportion was measured in women for risk area avoidance in Montérégie but not in Neuchâtel. In general, gender differences regarding the adoption of health-related behaviors are highly variable and depend on the type of behavior under study [41]. In Phillips and colleagues [28], women were also associated with a greater proportion of preventive behaviors including the practice of risk area avoidance, tick checks and tick repellent in residents of Nantucket Island in Massachusetts, United States. On the opposite, other studies have found no gender differences regarding LD preventive behaviors adoption [30,34].
Another main objective of this study was to test if exposure, knowledge, risk perception, and the perceived efficacy of measures were associated with the adoption of preventive behaviors to a similar degree in both regions. We calculated these associations with four multivariable logistic regression models predicting either a GPB score or an adoption score for three main specific preventive measures (OR was used as an indicator of the strength of association). We could not find significant differences between regional subsets in the strength of association when considering OR confidence intervals. On the other hand, even if we identified different significant factors between regions, we noted a good level of constancy in the strength of association, particularly in the association of risk perception and knowledge with adoption of specific measures when comparing overall models. Knowledge was also a common factor associated with preventive measures in overall models and most of the regional models, as previously reported in other regions [19,33,40]. Several studies have demonstrated that risk perception, expressed by the perceived severity of and the perceived susceptibility to LD, was associated with the adoption of preventive behaviors [19,20,24,33,34,40]. Herrington [20] compared factors associated with preventive behaviors between low and high incidence states in the United States. He observed differences in the strength of association of the perceived severity between these regions: in low incidence states, perceived severity was positively associated with the adoption of preventive behaviors (in general) while it was negatively associated with such behaviors in high-incidence states. As we used a global risk perception score (vs perceived severity) as an independent variable in our models, we are not able to directly compare our findings with these results.
In our study, we decided to restrict multivariable analyses to the subset of respondents who had heard about LD before the survey was administered, given that all other respondents could not have consciously applied preventive measures in order to protect themselves against LD if they did not even know about the existence of the disease. Given that a considerable number of respondents did not know about LD, especially in Montérégie where only 54% had heard about LD prior to the survey, the regional sample sizes were greatly diminished, resulting in large confidence intervals and a reduced statistical power that may partially explain the lack of differences observed in the strength of association between risk perception and the adoption of preventive behaviors between the two regions.
Nevertheless, multivariable models revealed other interesting findings. Perceived efficacy of specific preventive measures was strongly associated with the adoption of three preventive measures in our study. The perceived efficacy of a measure has previously been identified as an important predictor but the relationship was found to be stronger in our study than compared to previous studies [24,26,33].
Another interesting observation is that ‘living in the Montérégie region’ was positively associated with the use of tick repellent in the overall models, while it was negatively associated with the practice of tick checks and the use of protective clothing. One hypothetical explanation is that applying repellent is already well accepted by residents of the region for other reasons, such as to protect themselves against mosquitoes, and the risk of West Nile virus transmission, which is also present in this region [42]. This context is different in Neuchâtel where the use of repellent may be less common. Finally, it may seem reasonable to conclude that living in an emerging and low incidence region such as Montérégie could be negatively associated with specific preventive behaviors as the practice of tick checks and the use of protective clothing when compared to a high incidence region such as Neuchâtel.
This study has several limitations. We obtained data from a web-based survey using panels of respondents. Thus, our study was restricted to Internet users. More aspects of the representativeness of the data are discussed in Aenishaenslin et al. [35]. Also, the cross-sectional design of our study can provide useful data but cannot establish causal relationships, and thus explains our preference for the terms ‘factors associated with preventive behaviors’ in this paper, rather than ‘determinants of preventive behaviors’. A longitudinal design would be of great interest to study temporal changes in preventive behaviors in relation to evolving levels of knowledge and risk perception, particularly in the Montérégie context, where LD is emerging and where such changes will certainly be important in the coming years.
This study was carried out in two regions that were chosen based on their contrasting LD epidemiological situation. Differences observed between the two populations cannot be explained based on their LD epidemiological statuses alone. Other unmeasured contextual factors certainly have an impact on preventive behaviors, such as culture, societal values, and public health communication efforts. Our regional results should therefore be viewed as two case studies, and should be interpreted with respect to their regional specific contexts.
Another limit of the study is that several variables included in our analyses were categorized or dichotomized, resulting in a partial loss of information when compared to the raw survey data which was predominantly ordinal. This was done to allow a useful interpretation of the results in the public health context, to carry out multivariable logistic regression analyses and to maximise statistical power in our analyses.
One explanation for the high OR values found for the perceived efficacy of preventive measures in our study could be the desirability bias of the respondents. Perceived efficacy was based on survey data, and thus the assessment of the adoption of preventive behaviors was self-reported. This may introduce bias such as desirability bias in this measurement, and this bias may be exacerbated when respondents believe in the perceived efficacy of a measure. This may have increased the proportion of reported preventive behaviors and may have moved the estimate of the association between the perceived efficacy and the adoption of the measure in question away from the null. Studies focusing specifically on measuring the observed adoption (vs self-reported) of protective behaviors in relation to the perceived efficacy of a behavior may be of great interest for future research.
Finally, our study measured the association between factors with preventive behaviors, but besides the statistical significance of these factors, quantitative analysis cannot fully explain the relationships between these variables and cannot provide a deep understanding of the motivations and barriers of adoption of preventive behaviors. Qualitative studies may provide essential insights that may help deepen our understanding and ability to interpret behavior related studies, for example to explain observed differences in behavior between age categories.