A total of 1,669 residents of Block Island, most of the island population, enrolled in our study from 1991 to 2000. We excluded those participants who did not report spending at least 1 month on the island during May through October, which resulted in a sample of 1,490. The mean age of the sample was 43 years (95% CI 42.2–44.1) and approximately half (51%) were female.

Figure 1

Figure 1. . Reported tick bite and itch by serosurvey year on Block Island, Rhode Island, 1991–2000.

We determined how frequently the 1,490 persons in our sample recalled a deer tick bite. Each year an average of 27% (95% CI 23%–31%) of the study participants reported >1 tick bite during the prior year (range 20%–37%). An average of 17% (95% CI 13%–21%) of study participants reported itch associated with tick bite (Figure 1).

We then determined the prevalence of B. burgdorferi infection in nymphal deer ticks that contained B. burgdorferi DNA and that were swept from vegetation on Block Island throughout the course of this study. Of 135 such ticks tested, 23% (95% CI 20%–26%) contained amplifiable B. burgdorferi DNA. We therefore predicted that the maximum yearly incidence of Lyme disease would be 6.2% (95% CI 4.6%–8.1%) (27% with tick bite x 23% of ticks infected).

We next calculated yearly incidence rate of new Lyme disease cases among study participants from 1991 to 2000. The average incidence was 1.74% (95% CI 1.1%–2.4%) (Table). Thus, the predicted incidence of Lyme disease using the basic model (95% CI 4.6%–8.1%) was 3%–6% greater than the observed incidence. This difference would not be expected due to chance.

We next determined whether increasing exposure to ticks increases the probability of tick-associated itch. This analysis was limited to participants whose sera displayed no evidence of tickborne illness before study entry, who had >1 serosurvey visit, and who were >2 years of age. Among these 610 participants, 52% reported at least 1 tick bite (mean 2.2, 95% CI 2.0–2.4), and 32% reported itch with any tick bite. The probability of itch doubled as the number of reported tick bites increased from 1 to 2 (21% to 46%, respectively) and doubled again from 2 to 4 reported bites (46% to 97%, respectively; linear trend p < 0.001).

Figure 2

Figure 2. . Risk of acquiring Lyme disease according to reports of tick bite–associated itch.

Finally, we tested the hypothesis that tick-associated itching is associated with decreased risk of Lyme disease. The acquisition of Lyme disease increased from 15% to 25% to 31% among participants who reported no itch, 1 episode of tick-associated itch, and 2 reports of itch, respectively. In contrast, the frequency of Lyme disease decreased to 13% in participants who reported 3 episodes of tick-associated itch and 10% in those with >4 such reports. We used a multiple logistic regression model to estimate the likelihood of acquiring Lyme disease for participants reporting none to 1, 2, and >3 reports of itch, controlling for number of study visits and reports of tick bite. Consistent with the bivariate analysis, the risk of acquiring Lyme disease was higher for 1 report of itch (OR 2.7, 95% CI 0.4–2.3), and decreased among those who reported itch at >3 study visits (range 3–12, OR 0.18, 95% CI 0.05–0.5) (Figure 2). Confining the analysis to only those who had Lyme disease illness by omitting participants whose evidence of Lyme disease was by seroconversion alone did not alter the inverse relationship between itch and developing Lyme disease. Persons who consistently report itching in association with tick bites are less likely to experience an episode of Lyme disease than do those who fail to react against tick bite.

These observations suggest that residents of disease-endemic sites who experience persistent tick-associated itch are less likely to develop Lyme disease than are those who do not experience this reaction. Itch was reported at the site of tick bite, and the frequency of itch increases as the number of reported tick bites increase, which strongly suggests that tick-associated itch is associated with an acquired cutaneous hypersensitivity response. Such a relationship has definitively been established in the case of I. ricinis, the European analog of the North American deer tick. After having been bitten by these ticks repeatedly, persons experience both immediate and delayed cutaneous hypersensitivity reactions. They express tick-specific IgE antibodies, as well as dermal and perivascular infiltrates of CD8+ T lymphocytes and Langerhans' cells (13). Additional studies of other tick species also suggest that tick-associated itch is mediated by tick-specific antibody or cellular infiltration (12,18,19). People in our study who reported a single episode of tick-associated itch were more likely to acquire Lyme disease than those who did not report itch, probably because tick-associated itch is a marker for tick exposure. In contrast, persons who had repeated tick-associated itch were protected from developing Lyme disease. We are confident, therefore, that the people in our study who described repeated episodes of tick-associated itching were experiencing cutaneous hypersensitivity and that this immune reaction protected them from acquiring Lyme disease.

The antitick immune response that protects people from acquiring Lyme disease might act through any of several effector mechanisms. A heightened awareness by a person of an attached tick would result in the removal of the potentially infecting tick before pathogen transmission could occur. Because vector ticks must remain attached for at least 2 days before spirochetes are transmitted, prompt tick removal should prevent spirochetal infection (8,20). Vector ticks generally are removed by persons within such a time period (21–23). Alternatively, immunity against tick salivary antigens might interfere with pathogen transmission independent of early host recognition and removal of ticks. Indeed, preexposure of mice and guinea pigs to uninfected ticks prevents Lyme disease after challenge by spirochete-infected ticks using infestation conditions that prevent removal of feeding ticks by host grooming (24,25). Ticks feeding on such immune hosts detach sooner and retain less host material than do ticks feeding on nonimmune hosts. Although natural hosts, such as Peromyscus leucopus, do not develop robust antitick immunity, inbred strains of mice and guinea pigs become immune to the salivary antigens secreted by feeding ticks (10,26,27). This immunity results in reduced volume of engorgement, abnormal blood meal composition, prolonged feeding, and frequently death of the tick (10,27). Antitick immunity may specifically neutralize some components of tick saliva that ensure successful feeding and facilitate pathogen transmission, such as vasodilators, anticoagulants, and immunosuppressors (10,11,28). Prevention of B. burgdorferi transmission in guinea pigs is associated with antivector antibody (25). Persons who experience frequent deer tick bites produce an array of specific antitick antibodies (29–31). Cell-mediated immune factors against tick-derived antigen might similarly play a role in developing cutaneous hypersensitivity and protect against infection (13,19). A strong delayed cutaneous hypersensitivity response to sandfly bites in persons has been correlated with reduced transmission of leishmanial parasites (32,33). An array of antitick immune reactions may prevent Lyme disease and other tickborne diseases.

Our analysis incorporates several potentially confounding assumptions. Although our estimate of the prevalence of spirochetal infection in vector ticks is based on a small sample of ticks, this rate is consistent with that derived in other parts of southern New England at approximately the same time (5,6). Although our study participants live on Block Island where deer ticks predominate, some of our participants may have been bitten by other kinds of ticks or been bitten by ticks off the island. While approximately a quarter of our participants reported a tick bite, previous studies have shown that persons are often unaware that they have been bitten. From a third to three fourths of patients with Lyme disease cannot recall having been bitten by a tick (6,34–36). Assuming that a similar percentage of our study participants were unaware of a previous tick bite, the tick bite prevalence on Block Island would have varied from half to all of our participants. A tick bite prevalence of 71% was reported at another site where Lyme disease is highly endemic (37). Any underestimate of the frequency of tick attachment would increase the likelihood that other factors, including antitick immunity, help limit the incidence of Lyme disease. Finally, our predicted Lyme disease incidence assumes that people are bitten only once in any year. An increase in yearly tick bite frequency among persons would increase the likelihood that the tick bite is from an infected tick, thus similarly increasing the difference between predicted and observed incidence rates. We believe that none of these assumptions is notably confounding.

Several factors help prevent Lyme disease in persons who live in disease-endemic regions. These factors include the paucity of ticks that are infected with the Lyme spirochete, the limited number of persons who are bitten by ticks, acquired immunity against the spirochetal pathogen, and the immune reaction to ticks that develops in the course of tick attachment. Persons who express an immune reaction against the vector tick appear to acquire Lyme disease less frequently than do those who experience no such immune reaction. The protective effect of the immune response to tick salivary protein against the agent of Lyme disease in persons suggests that a tick protein-based vaccine might be developed that would protect against infection by the agent of Lyme disease and possibly other tickborne infections.