Sentinel surveillance outperforms risk-based for tracking tick-borne disease

Abstract

Background: Lyme disease (LD) emerged in southern Québec at the start of the century, with many municipalities now endemic. A coordinated active surveillance programme has been in place in the province of Québec since 2014, including a limited number of sentinel field sites resampled each year and a larger set of accessory field sites that change yearly according to the LD surveillance signal. We aimed to evaluate whether a sentinel approach to active surveillance was more representative of LD risk to human populations, compared to risk-based surveillance.

Methods: We compared enzootic hazard measures (average nymph densities) from sentinel and accessory sites with LD risk (number of human LD cases) across the study area between 2015 and 2019 using local bivariate Moran's I analysis.

Results: Hazard measures from sentinel sites captured spatial risk significantly better than data from accessory sites (χ²=20.473, p<0.001). In addition, sentinel sites successfully tracked the interannual trend in LD case numbers, whereas accessory sites showed no association despite the larger sample size.

Conclusion: Where surveillance aims to document changes in tick-borne disease risk over time and space, we suggest that repeated sampling of carefully selected field sites may be most effective, while risk-based surveillance may be more usefully applied to confirm the presence of emerging disease risk in a specific region of interest or to identify suitable sites for long-term monitoring as LD and other tick-borne diseases continue to emerge.

Introduction

Lyme disease (LD) is a tick-borne disease that has been emerging in southern Canada over the past three decades. Ixodes scapularis is the vector of Borrelia burgdorferi, the primary agent of LD in Canada, east of the Rocky Mountains ^{Footnote 1 Footnote 2}. Populations of I. scapularis ticks, first established in the north-eastern and midwestern United States, have expanded their geographic distribution northward via migratory birds to invade southern Canada ^{Footnote 3} with the first established populations appearing in Manitoba, Ontario, Québec and Nova Scotia ^{Footnote 4}. In response to this emerging health threat, public health authorities require effective surveillance systems to monitor the emerging risk of LD.

Active acarological surveillance, whereby forested field sites are sampled to collect questing ticks in the environment, is commonly used to assess enzootic hazard for LD ^{Footnote 5}. Active surveillance usually consists of drag sampling, where a piece of white flannel cloth is dragged across the forest floor such that questing ticks cling to the passing fabric, allowing them to be collected and analyzed. From such field studies, enzootic hazard is calculated as the density of nymphal ticks (DON) or density of infected nymphs ^{Footnote 6 Footnote 7}. Density of nymphal ticks and density of infected nymphs have both been associated with LD risk in different studies in North America ^{Footnote 8 Footnote 9 Footnote 10 Footnote 11}. Although some studies have evaluated the association between enzootic hazard and LD risk in southern Canada where LD is emerging, it is worth re-evaluating this link as the epidemiological portrait continues to evolve ^{Footnote 10 Footnote 12}. In addition, because increasingly large regions of southern Canada need to be surveyed as Ixodes spp. continue to increase their geographic range, there is a growing need to adapt active surveillance approaches to ensure their sustainability and relevance within the evolving epidemiological context ^{Footnote 13}.

Due to complex ecological requirements, tick populations tend to expand their geographical range heterogeneously in space ^{Footnote 14 Footnote 15}. To reflect this, active surveillance systems must be able to capture the spatially heterogenous LD risk pattern across a region. Some provincial public health authorities have developed risk-based criteria to decide which sites to target for surveillance whilst others visit the same sites repeatedly over time ^{Footnote 16 Footnote 17 Footnote 18}. Currently, it is not known which of these approaches best represents LD risk in space and time.

Among the ten provinces in Canada, Québec has the third-highest number of reported LD cases ^{Footnote 19 Footnote 20}. Québec is the largest and second most populated province in Canada, with a total population of nearly 8.5 million ^{Footnote 21}. Most of the population resides in the south of the province, where the highest I. scapularis tick densities occur. In the past five years, the number of human LD cases has more than tripled, an increase that is consistent with the expanding geographic distribution of I. scapularis in southern Canada ^{Footnote 22}.

Active tick surveillance has been carried out in southern Québec since 2007, with the first coordinated provincial surveillance system established in 2014 by the Institut national de santé publique du Québec in collaboration with the Université de Montréal ^{Footnote 18 Footnote 23}. From 2015 to 2019, active surveillance was carried out at two types of field sites: sentinel sites, which are kept constant through time and are visited every year, and accessory sites, which change every field season and are selected through a risk-based algorithm ^{Footnote 24}. These two types of sites were intended to serve different objectives within the surveillance program, with sentinel sites designed to provide a geographically representative surveillance signal allowing risk to be compared between regions and over time, and accessory sites selected each year to confirm the risk status of areas where LD was thought to be emerging. Sentinel surveillance was initiated in 2015 based on the hypothesis that repeated sampling of a small number of carefully selected sites could provide a more representative portrait of evolving LD risk at the provincial scale than annual risk-based surveillance. In addition, sentinel surveillance has several important logistical advantages, including reduced annual effort for site selection and lower overall sampling effort. However, the spatial and temporal representativeness of sentinel vs. risk-based surveillance have yet to be formally compared.

In this study, we analyzed LD surveillance data collected over a five-year period (2015–2019) to test the hypothesis that sentinel surveillance provides a more representative signal of LD risk in space and time than a risk-based approach, in the epidemiological context of pre-emerging/emerging LD risk in southern Québec.

Methods

Study site

The province of Québec is in eastern Canada, located between the provinces of Ontario and New Brunswick. The ten most southern administrative regions of the province encompass the emergence zone for I. scapularis and are targeted annually by the Institut national de santé publique du Québec for active surveillance (Figure 1). This study analyzed data collected from 2015 to 2019 by active surveillance within this study zone.

Figure 1 - Text description

This figure shows the ten southern-most administrative regions in the province of Québec, Canada: Capitale-Nationale, Mauricie and Centre-du-Québec, Estrie, Montréal, Outaouais, Chaudière-Appalaches, Lanaudière, Laurentides, Laval, and Montérégie. This region constitutes the active surveillance zone targeted by the provincial Lyme disease surveillance program. The program has sentinel sites (n=21) which are visited yearly and accessory sites which change on a yearly basis according of Lyme disease risk signal across the study zone. There are between 60 and 80 accessory sites sampled each year.

Active surveillance in Québec

A network of 21 sentinel sites for the active surveillance of LD risk in the environment was designed by the Québec Tick-Borne Disease Expert Panel (Groupe d'expert sur les maladies transmises par les tiques), a panel bringing together public health authorities, laboratory experts, scientific and medical advisors and epidemiologists. Two sentinel sites were chosen per administrative region, except in Montréal where three sites were selected (Figure 1) ^{Footnote 25}. The sites were placed in provincial or regional parks, which are readily accessible to the public, contain suitable habitat for the establishment of tick populations and are located in geographically distinct areas of the administrative region. These sentinel sites have remained the same throughout the study period (2015–2019) and were usually visited twice during the field season (May–August), the first time in early June, followed by a second visit at least two weeks later. Sites considered endemic for LD (one site in Montérégie and another in Estrie) were only visited once. Occasionally, other sites were only visited once per season due to logistical constraints (e.g. park closure).

In addition to the network of sentinel sites, 60–80 accessory sites were sampled once per field season. Accessory sites were selected according to the LD risk signal, generated from past passive and active acarological surveillance data and reported human cases. A standardized drag sampling protocol was carried out during each site visit, in both sentinel and accessory sites. Two field technicians dragged a 1 m² piece of white flannel cloth along two parallel transects: the first in the vegetation along the edge of a public footpath; and the second in the forest 25 m from the path. Each team member sampled 1,000 m² for total area sampled of 2,000 m² per site. Presence of ticks on the cloth was checked every 25 m and collected ticks were stored in tubes containing 70% ethanol. Subsequently, ticks were classified by species at the Québec Public Health Laboratory (Laboratoire de santé publique du Québec).

Human Lyme disease surveillance

Risk was calculated by the number of human LD cases reported at the municipal scale over the five-year study period, divided by the logarithm of human population size. A potential source of error is the misclassification of the municipality in which LD was acquired. As only half of LD cases recall have been bitten by a tick ^{Footnote 26}, it was sometimes difficult for individuals to identify the precise location where they acquired LD. However, in Québec, each LD case is subject to a public health investigation, which includes a review of the clinical and personal history of the patient to determine the most likely location of acquisition, limiting this source of error.

Statistical analyses

Enzootic hazard from active surveillance: The DON was calculated as a measure of enzootic hazard ^{Footnote 24}. Due to their small size, nymphs represent a greater hazard to humans as they are likely to be missed during self-examination ^{Footnote 27}. As tick densities are relatively low in southern Québec, we decided not to use density of infected nymphs, which may not be representative due to the low numbers of collected ticks.

Using seasonality models of I. scapularis phenology in southern Québec, we estimated standardized nymph densities for a reference date of June 15 ^{Footnote 28}. This allowed us to correct for temporal variability in nymph densities due to site visits occurring at different periods of the tick life cycle. We used these estimated nymph densities to compute the mean density per site across the study period. The data were georeferenced using the start location of the surveillance transect.

Densities of nymphs measured annually at both sentinel sites and accessory sites were interpolated across the study zone to generate a hazard map based on each type of surveillance. Interpolation was done using a Kernel density estimation (QGIS version 3.18; Zurich, Switzerland). A distance of 80 kilometres was used as the radius of interpolation, as correlogram-revealed spatial dependency of active surveillance data up to this distance ^{Footnote 24 Footnote 29}. The resulting hazard maps were used to assign an estimated value of DON based on sentinel surveillance and risk-based surveillance to each municipality across the study zone.

Temporal association between enzootic hazard and LD risk

To assess the association between enzootic hazard (average nymph density) and LD risk (number of human LD cases) across the study period. Pearson correlations between these two variables were tested using R version 4.0.4 ^{Footnote 30}. The estimated DON was calculated at sentinel and accessory sites as described in the previous section. The resulting average nymph density derived from all sentinel or accessory sites in the same year was then correlated with total human cases reported that year.

Spatial association between enzootic hazard and LD risk

Bivariate local Moran's I analyses were performed using GeoDa 1.18.0 to determine the spatial association between enzootic hazard and LD risk. Bivariate local Moran's I can capture the relationship between a value of one variable in space, and the average neighbouring values for another variable ^{Footnote 31}. GeoDa creates cluster maps that determine if the spatial association between the variables is significant or not across municipalities. If significant, the maps indicate if 1) both variables represent high values, 2) both represent low values or 3) one variable is a high value and the second is a low value. Furthermore, some municipalities may remain "undefined" if they do not have an attributed value of either one of the variables or be "neighbourless" if adjacent polygons are missing data.

The results from the Moran's I analyses were transcribed into a contingency table. From the contingency table, we were able to calculate if hazard measures were positively associated with risk as predicted (i.e. both risk and hazard are low or high), or if they diverged (i.e. risk is high whilst that hazard is low, or vice versa), for sentinel and accessory site hazard measures. Chi-square tests were performed to evaluate significant statistical differences in hazard-risk associations between site types.

Results

Active surveillance

A total of 197 site visits were conducted at sentinel sites between 2015 and 2019: 28 in 2015; 45 in 2016; 43 in 2017; 39 in 2018; and 42 in 2019. A total of 346 accessory site visits were carried out over the same period: 47 in 2015; 104 in 2016; 55 in 2017, 65 in 2018; and 75 in 2019. Average nymph density across the study period was 0.13 (0.10–0.16) nymphs/100 m² at sentinel sites and 0.08 (0.07–0.11) nymphs/100 m² at accessory sites. Sentinel sites identified the regions of Montérégie, Estrie and Outaouais as having the highest DON (Table 1), whereas for accessory sites, the highest DON were found in Outaouais followed by Montérégie. It is worth noting that for accessory sites in Mauricie-et-Centre-du-Québec, high average nymph density in 2016 was due to a single site where 2.17 nymphs/100 m² was recorded.

Table 1: Average nymph densities^{Footnote a} for sentinel and accessory sites in the ten administrative regions included in the active surveillance system for Lyme disease in Québec, Canada, from 2015 to 2019

Admin region	Density of nymphs (nymphs/100 m2)Footnote b
	Year					Average
	2015	2016	2017	2018	2019
Sentinel sites
CN	0	0	0	0	0	0
MC	0.02 (0–0.05)	0.01 (0–0.03)	0	0	0	0.01 (0–0.01)
ES	0.74 (0–1.47)Footnote c	0.22 (0–0.43)Footnote c	0.27 (0–0.53)Footnote c	0.05 (0–0.1)	0.05 (0–0.1)	0.23 (0.11–0.35)Footnote c
MT	0.02 (0–0.05)	0.05 (0–0.1)	0.25 (0.09–0.41)Footnote c	0	0.1 (0.02–0.18)	0.09 (0.05–0.14)
OU	0.13 (0.09–0.17)	0.06 (0.02–0.1)	0.20 (0.15–0.25)Footnote c	0.03 (0.01–0.04)	0.85 (0.3–1.4)Footnote c	0.20 (0.10–0.30)Footnote c
CA	0	0	0	0	0	0
LV	0.02 (0–0.05)	0.01 (0–0.03)	0.01 (0–0.03)	0.03 (0–0.05)	0.05 (0.01–0.09)	0.02 (0.01–0.04)
LN	0	0	0.02 (0–0.03)	0	0	0.005 (0–0.01)
LA	0.04 (0–0.08)	0.04 (0.01–0.06)	0.11 (0.05–0.18)	0	0.02 (0–0.03)	0.05 (0.03–0.06)
MR	0.45 (0.02–0.88)Footnote c	0.38 (0.14–0.62)Footnote c	1.23 (0.55–1.90)Footnote c	0.47(0–0.93)Footnote c	0.05 (0–0.1)	0.57 (0.36–0.77)Footnote c
Accessory sites
CN	0	0	0	0.05 (0–0.1)	0.01 (0–0.02)	0.01 (0–0.02)
MC	0	0.54 (0–1.09)Footnote c	0	0.01 (0–0.01)	0	0.06 (0–0.13)
ES	0.01 (0–0.02)	0	0.2 (0.05–0.35)	0	0.02 (0.01–0.04)	0.02 (0.01–0.04)
MT	0	0	0.06 (0–0.13)	0	0	0.02 (0–0.03)
OU	0	0.35 (0.13–0.57)Footnote c	0.81 (0.28–1.36)Footnote c	0	0	0.31 (0.31–0.49)Footnote c
CA	0	0	0	0	0.01 (0–0.01)	0.002 (0–0.004)
LV	0.01 (0–0.02)	0	0.06 (0–0.12)	0	0	0.02 (0–0.04)
LN	0	0	0.12 (0.01–0.23)	0	0.12 (0.001–0.25)	0.05 (0.02–0.09)
LA	0	0.002 (0–0.003)	0	0	0.03 (0.01–0.05)	0.005 (0–0.008)
MR	0.02 (0–0.04)	0.37 (0.28–0.46)Footnote c	0.05 (0.01–0.08)	0.01 (0–0.01)	0.01 (0.01–0.02)	0.19 (0.14–0.24)
Figure 1 - abbreviation Abbreviations: CA, Chaudière-Appalaches; CN, Capitale-Nationale; ES, Estrie; LA, Laurentides; LN, Lanaudière; LV, Laval; MC, Mauricie-et-Centre-du-Québec; MR, Montérégie; MT, Montréal; OU, Outaouais Table 1 Footnote a 95% confidence intervals across the study period Table 1 Return to footnote a referrer Table 1 Footnote b No colour: <0.20 nymphs/100 m2 Table 1 Return to footnote b referrer Table 1 Footnote c Red: >0.20 nymphs/100 m2 Table 1 Return to footnote c referrer

These densities were subsequently adjusted using the seasonality model to account for tick phenology prior to using the data for further analysis (Figure 2).

Figure 2 - Text description

The line graph shows the number of human Lyme disease cases between 2015 and 2019 in southern Québec, Canada. During this period, a total of 1,063 cases were acquired within the study zone and reported to public health authorities: 108 in 2015, 125 in 2016, 247 in 2017, 210 in 2018, and 372 in 2019. The bar charts shown average nymph densities (nymphs /100 m²) collected in sentinel sites (light grey) and accessory sites (dark grey) between 2015 and 2019. Average nymph density across the full study period was 0.13 (0.10–0.16) nymphs/100 m² at sentinel sites and 0.08 (0.07–0.11) nymphs/100 m² at accessory sites.

Within significant associations, some showed positive associations (both variables either "high" or "low") whilst others showed negative associations (one variable was "high" whilst the other was "low"). In the context of surveillance, positive associations between active surveillance and LD risk suggest reliability of active surveillance sites. In this analysis, sentinel sites showed positive association with LD risk for 388 municipalities (36.8%), whereas accessory sites showed positive association with LD risk for 302 (28.6%). The proportion of positive vs. negative association was significantly higher for sentinel vs. accessory sites (χ²=20.473, p<0.001).

Statistical analyses

Correlation between enzootic hazard (average DON) and LD risk (number of human cases) showed a positive association (r=0.88; 95% CI, -0.02–0.99) for data obtained from sentinel sites. This association was weakly significant by Pearson's correlation test (p=0.05). In contrast, for data collected from accessory sites, the correlation between enzootic hazard and LD risk was negative (r=-0.32; 95% CI, -0.937–0.784) and not significant (p=0.60).

Interpolated data at the municipal level across Québec were used in local bivariate Moran's I to see if nymph densities collected during active surveillance methods were associated with the degree of LD risk (number of human cases/logarithm of the population (Figure 3). The cluster maps show whether there is significant spatial association between these two variables. Accessory site data had a greater proportion of non-significant classifications (n=490, 46.4%) compared to sentinel site data (n=348, 33.0%) (Table 2). Limited number of sampling sites during active surveillance meant that some of the study zone was undefined or neighbourless in the analyses, as no data was collected in these areas to incorporated within the analysis.

Figure 3 - Text description

This figure illustrates cluster maps derived from local bivariated Moran's I analyses, comparing human Lyme disease risk (number of Lyme disease cases divided by the logarithm of the population) with nymph densities obtained from active surveillance data. The first panel a) use active surveillance data from sentinel sites. The second panel b) uses active surveillance data from accessory sites. The cluster maps show whether there is significant spatial association between these two variables. Within significant associations, some showed positive associations (both variables either «"high" or "low") whilst others showed negative associations (one variable was "high" whilst the other was "low"). In the context of surveillance, positive associations between active surveillance and Lyme disease risk suggest reliability of active surveillance sites. The finite number of sites sampled during active surveillance activities meant that some of the study zone was undefined or neighbourless in the analyses, as no data was collected in these areas to be incorporated within the analysis.

Table 2: Outcome of local bivariate Moran's I for human cases data compared with active surveillance data from sentinel and accessory sites

Local bivariate Moran's I outcome	Sentinel	Accessory
Not significant	348	490
High-high	44	53
Low-low	344	249
Low-high	124	181
High-low	13	15
Neighbourless	17	22
Undefined	165	45
Totals
Positive association	388	302
Negative association	137	196

Discussion

This paper demonstrates the ability of active surveillance at a limited number of high-quality sentinel sites to capture spatiotemporal LD risk trends in a context of emerging disease over a five-year period. In contrast, roughly twice the number of site visits carried out at accessory sites during this same period using a risk-based approach provided a less accurate geographic portrait of emerging risk and failed to capture the steep increase in human cases over time, even suggesting that risk had decreased rather than increased over the study period. Sentinel and risk-based surveillance provide complementary information and serve different purposes within a surveillance system; this study demonstrates that the analysis and interpretation of the resulting surveillance data should take these differences into account. Specifically, where surveillance aims to document changes in tick-borne disease risk over time and space, we suggest that repeated sampling of carefully selected field sites may be most effective, while risk-based surveillance may be more usefully applied to confirm the presence of emerging disease risk in a specific region of interest or to identify suitable sites for long-term monitoring as LD and other tick-borne diseases continue to emerge.

In our analyses, we used enzootic hazard measures, in the form of nymph density collected at sentinel and accessory sites, to track the temporal trend in LD risk between 2015 and 2019. A Pearson correlation test at the provincial level demonstrated that average nymph density calculated from sentinel sites was positivity correlated with LD risk (number of human LD cases) (r=0.88), compared with average nymph density calculated from accessory sites where no significant correlation was found. As accessory sites change yearly, average nymph densities will not only account for interannual variation, but also for the heterogeneous spatial distribution of tick populations which make the yearly variation more difficult to interpret. However, as our study period was limited to five years, this result should be interpreted with caution. Furthermore, in previous research we noted that a positive association between nymph densities from sentinel sites and annual human cases was not always evident at the regional scale (e.g. in the Estrie region) ^{Footnote 24}. It would be interesting to explore the reasons for regional variation in this relationship; for instance, it is possible that the sentinel sites chosen in Estrie were not optimal to represent the epidemiological portrait at this scale. In the meantime, we suggest that average nymph density calculated from sentinel sites at a broader scale may be more robust and informative for evaluating interannual variation in LD risk.

The spatial relationship between enzootic hazard and LD risk was more reliably represented by sentinel sites compared to accessory sites. For both analyses, interpolation was used to permit representation of enzootic hazard across the full extent of the study zone; hence, the interpolation will not capture fine-scale heterogeneity in tick population establishment across space, which could subsequently lead to information bias. It is thus important to consider that sentinel surveillance provides a general risk indicator to follow spatiotemporal trends, whereas the targeted information derived from risk-based surveillance strategies may be more appropriate for confirming the establishment of endemic LD risk at the municipal scale ^{Footnote 13}. Areas where risk was not well captured by sentinel surveillance (e.g. the North Shore of Montréal) (see Figure 3) could subsequently be surveyed using an exploratory approach; risk-based surveillance (accessory sites) could be added to the surveillance strategy and the most informative sites retained as part of the sentinel system. While we show that two sentinel sites per administrative region were able to capture broad-scale trends in LD risk, increasing the number of sentinel sites per region would be useful in allowing better geographic representativeness and higher-resolution risk estimates. Another source of information bias could come from underreporting of LD cases, for instance in lower-risk areas due to reduced awareness of the public and clinicians. Previous studies suggest that this may not be a significant issue in Canada ^{Footnote 32}; however, this reiterates the advantage of having a longer time series or repeating a similar study to determine if the relationship between enzootic hazard and LD risk holds, especially as awareness of LD may change with time.

Sentinel surveillance for tick-borne disease in not a novel concept. Many studies, including some in southern Canada, have sampled sites repeatedly to determine geographic or ecological risk of LD associated with presence of ticks ^{Footnote 33 Footnote 34 Footnote 35}. In the United States, data from repeated field sampling have shown a positive correlation between density of infected nymphs and human cases ^{Footnote 11}; however, these sites were not part of a coordinated surveillance system. In Canada, field data from repeated site visits have been used to develop and evaluate indicators to determine the likelihood of establishment of I. scapularis, thus contributing to knowledge of hazard distribution ^{Footnote 34 Footnote 36}. A national sentinel surveillance system for tick-borne disease was launched in 2019 by the Canadian Lyme Disease Research Network (CLyDRN); however, the data generated from this new surveillance initiative remain to be analyzed ^{Footnote 37}. The surveillance system put in place in Québec, which uses both sentinel and risk-based surveillance, permitted the first comparative evaluation of these surveillance approaches in the southern Canada—an area where LD is emerging. As the epidemiological portrait of LD is fast evolving, the relationship between enzootic hazard measured at sentinel sites and LD risk may have to be re-evaluated regularly to determine if this relationship holds. Clow et al. ^{Footnote 13} proposed a framework for surveillance of tick-borne diseases where surveillance is described as an adaptive process, with surveillance goals modified over time as the epidemiological context continues to evolve.

Strength and limitation

According to this framework, active surveillance at sentinel field sites is considered suitable for both the emergence and endemic phases of the disease process. Although we have shown the ability of sentinel sites to track spatiotemporal risk more reliably than accessory sites, this remains to be demonstrated for endemic regions. Furthermore, an important limitation of sentinel surveillance is its inefficiency in pre-emergence context; as sentinel sites are a small subset of possible surveillance locations, they have limited sensitivity for capturing early emergence signals. This highlights the complementary role of sentinel surveillance within larger surveillance network that includes other surveillance methods such as passive acarological surveillance (e.g. eTick) ^{Footnote 10 Footnote 38}.

Conclusion

Our study demonstrated the capacity of sentinel surveillance to track spatiotemporal risk of LD in a region where the risk is spreading. In Canada, where tick-borne diseases continue to emerge, this study can support the planning of active surveillance strategies. Active surveillance at sentinel sites allows for comparable hazard measures through space and time, whilst limiting sampling effort to a restricted number of sites. A careful decision-making process must support site selection, to ensure that these are representative of the underlying epidemiological context and that the resulting data provide a robust portrait of emerging disease trends in space and time.

Authors' statement

CG — Conceptualization, methodology, analysis, interpretation, writing original draft, review and editing

PAL — Supervision, conceptualization, methodology, interpretation, review and editing

CB — Supervision, methodology, interpretation, review and editing

FM — Supervision, review and editing

KB — Review and editing

RP — Review and editing

Competing interests

None.

Acknowledgements

We thank the many field and laboratory technicians who were involved in tick collection and testing. We gratefully acknowledge the local, regional, and national parks for providing access to field sites.

Funding

The Institut national de santé publique du Québec (INSPQ) and Ministère de la santé et des services sociaux (MSSS) funded data collection in sampling sites across Québec as part of annual surveillance activities. The Laboratoire de Santé Publique du Québec (LSPQ) identified the species of tick specimens collected. The National Microbiology Laboratory confirmed the species and undertook the testing of ticks for tick-associated pathogens.