The burden of Lyme disease (LD) and other diseases transmitted by Ixodes spp. ticks is increasing in the United States (Adams et al., 2017; Schwartz et al., 2017). With no vaccines currently available, prevention relies on use of personal protective measures (e.g. tick checks, repellent use) and environmental controls (e.g. yard-based acaricide sprays, host-targeted acaricide treatments; Shen et al., 2011; Wormser, 2005). However, because of a relatively low disease incidence, well-powered studies designed to measure the efficacy of specific prevention approaches on human tickborne disease (TBD) outcomes are exceptionally time- and resource-intensive. Therefore, in studies evaluating environmental controls for the prevention of LD, the primary outcomes have been measures of nymphal Ixodes scapularis abundance and infection prevalence with Borrelia burgdorferi (e.g. nymphal density, nymphal infection prevalence, density of infected nymphs; Curran et al., 1993; Dolan et al., 2009; Dolan et al., 2004; Eisen & Dolan, 2016; Falco & Fish, 1992; Rand et al., 2010; Schulze et al., 1995; Schulze et al., 2001; Stafford, 1991). For decades, these entomological measures have functioned as proxies for human LD risk.

Studies evaluating the relationship between entomological measures and human disease risk have shown inconsistent results (Connally et al., 2006; Pepin et al., 2012; Falco et al., 1999; Mather et al., 1996), indicating that entomological measures may not accurately represent human risk of becoming infected with tickborne pathogens. In addition to the fact that tick abundance can be highly variable even at a fine spatial scale (e.g. within a single yard), entomological indices do not account for human behaviour (Feldman et al., 2015; Pardanani & Mather, 2004). For example, the presence of nymphal I. scapularis ticks indicates some LD risk in an endemic area, and even more so if the vectors are shown to be infected with B. burgdorferi. However, if humans rarely venture into I. scapularis habitat, the risk of acquiring LD is low (Eisen & Eisen, 2018). On the other hand, human-tick encounters, that is finding ticks crawling or attached, provide direct evidence of tick exposure. Even for those who only noticed ticks crawling on themselves, this is a potential marker for having attached ticks that went undetected. This scenario is especially likely with encounters with I. scapularis nymphs, which are very small in size and can be difficult to see. While human-tick encounters do not indicate 100% probability of acquiring LD, the relationship between TBDs and tick encounters, which account for both tick abundance and human behaviour, is potentially stronger than that with entomological measures alone. However, this relationship has never been quantified.

In the first intervention study that prospectively measured human disease outcomes to evaluate the effectiveness of a yard-based acaricide application in LD endemic areas, I. scapularis density did not correlate with the occurrence of human-tick encounters (ticks found crawling on or attached to study household members) or with reported TBDs among household members (Hinckley et al., 2016). Using data from this study, we sought to answer the question of whether human-tick encounters may better capture risk of LD and other TBDs associated with I. scapularis in endemic areas. The specific objectives of this analysis were to (a) evaluate whether human-tick encounters can serve as an accurate measure of LD risk and (b) to identify demographic and property characteristics associated with human-tick encounters.

As described in Hinckley et al., households in counties highly endemic for LD in Connecticut, Maryland and New York were enrolled in a randomized controlled trial to determine the effectiveness of a single, yard-based, spring-time barrier application of acaricide in preventing LD and other TBDs (Feldman et al., 2015; Hinckley et al., 2016). Enrolment criteria included the presence of certain property characteristics known to be favourable to I. scapularis populations and a household size of two or more people. The study was conducted from late April to October in 2011 and 2012; households participated in only one of the two years. Because the randomized controlled trial found no difference in human-tick encounters or human TBD between treated and placebo properties, participants from both groups were treated as a single cohort in our analysis.

Household property characteristics and demographic information were collected via phone-based, introductory surveys at the time of study enrolment during April and May of each study year. Human-tick encounters were self-reported in four web-based surveys conducted approximately monthly during the treatment period, June – September. Designated heads of household answered all questions for the household surveys, including the number of ticks found crawling on or attached to household members. In 2011, human-tick encounters were reported in aggregate for each household; in 2012, these were also reported individually for each household member. Participants in CT and MD were also offered the opportunity in both 2011 and 2012 to send ticks to study investigators for species identification. Self-reported TBD diagnoses were collected in a final, phone-based survey at the end of each study year in October. Self-reported diagnoses were verified by chart review by two physicians and one doctoral-level epidemiologist. Participants who did not complete any monthly surveys were excluded from all analyses, and those who did not complete a final survey at the end of the study year were excluded when assessing relationships between tick encounters and development of a TBD. Participant information that was collected only at the household level was assumed to apply to all members of the household (e.g. race, pet ownership).

Human-tick encounters reported during the study period at either the household or individual level were categorized into the following three groups: experienced any tick encounter (i.e. found at least one tick crawling or attached), found at least one tick crawling and found at least one tick attached. Descriptive statistics were conducted for household (2011, 2012) and individual (2012) level tick encounters, with Pearson’s chi-square tests used to compare tick encounters by household and individual characteristics. To evaluate the association between human-tick encounters and TBD diagnoses at both the household and individual levels, relative risk ratios with 95% confidence intervals and attributable risk per cent were calculated. Multivariable logistic regression models were used to evaluate whether certain demographic and property characteristics were risk factors for human-tick encounters at the household level, while adjusting for covariates. Alpha was set at .05 for all statistical tests of significance. Responses of ‘don’t know’, ‘not sure’ or ‘prefer not to answer’ were classified as missing data for all analyses. All analyses were conducted in R 3.5.1 (R Core Team, 2018). Tidyverse packages (Grolemund & Wicham, 2011; Wickham, 2017, 2018; Wickham, Francois et al., 2018; Wickham, Hester et al., 2018) were used to clean, organize and analyse data. The fmsb package (Nakazawa, 2018) was used to calculate odds ratios and risk ratios.

The study protocol was reviewed and approved by human subjects research committees at the Centers for Disease Control and Prevention, Yale University, Connecticut Department of Public Health, Maryland Department of Health and New York State Department of Health.

The primary objective of this analysis was to evaluate whether human-tick encounters can serve as an accurate proxy for LD risk in endemic areas by measuring the association between human-tick encounters and TBD using a prospective design. We found that, at both the household and individual levels, experiencing any tick encounter and finding a tick crawling on oneself was associated with a three- to four-fold increase in risk of TBD. This relationship is further strengthened by the alignment between demographic and property characteristics associated with human-tick encounters found in this study and those commonly associated with LD in other studies (Jones et al., 2018; Ley et al., 1995; Nadelman & Wormser, 2007; Schwartz et al., 2017). Our findings suggest that tick encounters can serve as a robust proxy for LD and other TBDs associated with I. scapularis in prevention studies in endemic areas. Using any tick encounter (i.e. at least one tick found crawling or attached) as an outcome in future prevention studies would reduce sample size requirements as well as participant follow-up time compared to using TBDs as an outcome. An additional advantage is that using tick encounters as an outcome measure can account for human behaviour, which purely entomological outcomes cannot.

Risk factors for reported tick encounters were concordant with those typically associated with LD. For example, trends in national surveillance data from 2008 to 2015 indicate that confirmed and probable cases of LD tended to be white, male and follow a bimodal age distribution with peaks seen among younger children and older adults [2]. We saw similar trends for tick encounters at the individual level. Being white and male were both associated with experiencing any tick encounter, and those 9 years of age and younger had especially high odds of experiencing any tick encounter, though all age groups had higher odds of experiencing a tick encounter than those 20–29 years of age. Further, studies have shown that previous history of LD is a risk factor for current LD (Nadelman & Wormser, 2007). Similarly, our results suggest that those with a previous history of TBD were at an increased risk for experiencing tick encounters.

We found that higher proportions of households and individuals with properties ≥2 acres experienced tick encounters compared to those with smaller properties. Having a property size >0.5 acre has commonly been considered a risk factor for human exposure to I. scapularis (Maupin et al., 1991). However, this is the first study we are aware of that has quantified the relationship between property size and tick encounters. Interestingly, those with forested areas on their properties did not have increased odds of experiencing tick encounters at the household or individual level, though other studies have documented a positive relationship between forested properties and entomological indices (Carroll et al., 1992; Dister et al., 1997; Duffy et al., 1994; Maupin et al., 1991; Stafford & Magnarelli, 1993) as well as LD diagnoses (Smith et al., 2001). Forested area may be a better indicator of viable tick habitat and entomological risk, rather than human risk.

It is certainly possible that participants in CT and NY had a higher likelihood of experiencing tick encounters than those in MD. Additional risk for LD in CT and NY is supported by public health surveillance data. In 2009, before the intervention study was initiated, all but one of the five counties targeted for recruitment in MD had lower LD incidences (with 4.0, 37.2, 43.3, 52.4 and 95.1 reported cases per 100,000 population) than the one targeted county in CT (79 reported cases per 100,000 population), and all had lower LD incidence rates than the two targeted counties in NY (927.6 and 333.5 reported cases per 100,000 population).

Owning a pet was also associated with increased odds of all forms of tick encounters. Pet ownership has long been considered a risk factor for tick exposure (Fischhoff et al., 2019; Hojgaard et al., 2014; Mead et al., 2018; de Wet et al., 2020) and TBD (Drexler et al., 2014; Jones et al., 2018; Lane & Lavoie, 1988; Ley et al., 1995; Rabinowitz et al., 2007; Steere et al., 1978). Recent evidence suggests increased risk may come from activities pet owners engage in with their pets that expose them to tick habitat rather than pets transporting ticks into the home or through direct contact (e.g. sleeping with pets; de Wet et al., 2020).

D. variabilis represented notable proportions of ticks from human hosts submitted by participants for identification in both CT (64.2%) and MD (45.2%). While these submitted ticks do not represent all tick encounters reported for CT and MD during the study, they do confirm that not all tick encounters reported by participants in these two states were with I. scapularis. Nonetheless, we found that all forms of tick encounters were correlated with both self-reported and verified disease, providing confidence that in regions of the United States where LD is the most commonly reported TBD (e.g. the northeast), and especially in states where I. scapularis is the predominant tick species (e.g. CT), tick encounters may serve as an accurate proxy for LD and other TBDs associated with I. scapularis.

We sought to quantify the relationship between human-tick encounters and TBDs associated with I. scapularis. Human-tick encounters, at both the household and individual levels, are a robust measure of disease risk and inherently account for human behaviour, which entomological measure cannot. We recommend that future LD prevention research consider using tick encounters as an outcome and as a proxy for LD diagnoses, in addition to offering professional identification of tick species encountered to validate results.