Methods

Site Description

Monmouth County (40°44′N, 74°17′W) is located in eastern-central New Jersey, a US state on the mid-Atlantic coast. The county is 468.8 mi² in size and had a population of 630,380 (1,344.7 persons/mi²) as of the 2010 census (http://www.census.gov/quickfacts/table/PST045215/34025). The geomorphologic break separating the Inner and Outer Coastal Plain physiographic provinces in New Jersey runs horizontally across the center of the county, and the resulting soil differences are reflected in vegetative differences between these 2 regions.^[20] The Outer Coastal Plain region is characterized by sandier soils that are often dry and acidic, with pine forests and cedar swamps. Previously, the distribution of A. americanum ticks in Monmouth County was restricted to this southern part of the county,^[21] whereas I. scapularis ticks were found throughout. However, recently specimens of A. americanum ticks have been captured in the far north and west of the county.

The county reports several hundred cases of Lyme disease annually, with a 10-year average (during 2005–2014) of 361 cases/year. By contrast, during that same period, there were on average 5.5 cases/year of E. chaffeensis infection and no cases of infection attributed to E. ewingii.^[22]

Risk Model

The relative risk for infection with a tickborne pathogen depends on multiple factors, including risk for exposure to a competent vector, risk for exposure to the pathogen from exposure to the vector, and risk for transmission from exposure to a pathogen-infected vector. To characterize the risk for human exposure to ticks, we define the parameter C_x as the relative proportion of each species (x) of tick submitted to the Monmouth County Mosquito Control Division’s tick identification and testing service. This passive surveillance program, initiated in 2006, allows county residents to submit ticks they have encountered (e.g., found on their skin or clothing) for species identification. This program averages 658.9 submissions/year, although this number has increased markedly in recent years (R.A. Jordan, unpub. data). For C_x , we used the 10-year average of relative submissions data (2006–2015) during peak Lyme disease transmission season in New Jersey (May–August) ( Table ). Although in Monmouth A. americanum ticks are often 3 times as abundant as I. scapularis ticks in field collections,^[9] using the passive surveillance data in our model (where A. americanum ticks are only encountered 1.5 times as often; Table ) provides a more direct measure of human exposure to ticks as well as a more conservative risk estimate.

To characterize risk for exposure to the disease from the vector, we define I_x as the prevalence of the pathogen in ticks (i.e., percentage infected), weighted by life stage. Both I. scapularis and A. americanum ticks have a 3-host life cycle, meaning that adults have had more opportunities to feed on an infected host than nymphs and consequently infection rates differ between life stages. Because transovarial transmission of either B. burgdorferi or E. chaffeensis does not occur,^[23,24] host-seeking larvae are not infected and therefore were not included in the calculations. Relative abundance of nymphs and adults of each species submitted to our passive tick surveillance program during May–August were reported (C_X,N and C_X,D ) and used to weight the infection prevalence of each ( Table ). Infection rates of I. scapularis ticks with B. burgdorferi for both life stages (I_IS,D and I_IS,N ) also were obtained from our passive surveillance program, whereby residents submitting an I. scapularis tick can elect to have it tested for B. burgdorferi through nested PCR assay (following established protocols [8]). Our records show that during a 10-year period of our program, 90.5% of residents submitting I. scapularis ticks during May–August have chosen to have them tested (R.A. Jordan, unpub. data), including 153 adults and 1,146 nymphs. However, the program does not test A. americanum ticks, so infection rates for this species were obtained from other sources. Rates of adult tick infection with E. chaffeensis and E. ewingii (I_AA,D ) in Monmouth County were derived from Schulze et al.^[21] and summed, yielding a total value (accounting for co-infected ticks) of 11.7% infection with human Ehrlichia pathogens (N = 291). Nymphal infection rates with both ehrlichia species were obtained from an unpublished dataset consisting of field-collected nymphs from 4 sites in eastern and western Monmouth County in 2014 (R.A. Jordan unpub. data). Nymphal specimens were disrupted by using a TissueLyser and DNA isolated with QIAgen DNeasy 96 well-plate blood and tissue kits (QIAGEN, Valencia, CA, USA). Specimens were tested for pathogens by using real-time PCR protocols for E. chaffeensis^[25] and E. ewingii.^[26] Both probes were modified slightly (shortened) to allow the use of an MGB quencher as follows (5′–3′): VIC-CGGACAATTGCTTATAACC-MGBNFQ for E. chaffeensis and 6FAM-AACAATTCCTAAATAGTCTCTGAC-MGBNFQ for E. ewingii. Sequence detection primers and reaction conditions were as described previously.^[26] A subset of samples was compared with conventional PCR methods established by this laboratory for detection of these pathogens,^[21] and the 2 methods were found to be in agreement. The resulting infection prevalence of Ehrlichia pathogens in A. americanum nymphs (I_AA,N ) (encompassing E. chaffeensis and E. ewingii, accounting for coinfection) was 9.04% (N = 752). Overall infection prevalence (I_x ) in each species of tick, weighted by life stage, was calculated by multiplying the relative abundance of each life stage times its infection rate and summing across life stages ( Table ).

To characterize transmission risk, we defined T_X as the likelihood of successful pathogen transmission from an infected tick to a human. The general concept of vector competence includes both this direction of transmission from vector to host as well as the probability of the vector becoming infected from feeding on infected hosts, which in our calculations is already reflected in tick infection rates (I_x ). Data on transmission efficiency to animals for both species is scarce because most studies feed multiple infected ticks on 1 host (so the risk imposed by a single feeding tick is unknown) and, because of the difficulty in working with large vertebrates in a laboratory setting, have very small sample sizes.^[27–30] Further, whether probabilities of transmission obtained from animal studies are applicable to humans is unknown. Therefore, although we are including the parameter T_X in the equation so that it can be applied when such data become better known, for purposes of these analyses we are setting it equal to 1 for both diseases, yielding no functional impact on the model.

Calculations

Based on our definitions, we calculate the relative risk for ehrlichiosis compared to Lyme disease as risk = (C_AA × I_AA × T_AA )/(C_IS × I_IS × T_IS ). To then translate this value into expected cases of observed human infection, we move from risk to reported case numbers by using the reported number of cases of Lyme disease in Monmouth County as a benchmark. In other words, if there were X Lyme disease cases, then the expected number of ehrlichiosis cases would be X multiplied by the relative risk estimate calculated as described.