At the time of the baseline visit, none of the 52 patients with erythema migrans had received antimicrobial drugs; 31 (59.6%) patients had 1 erythema migrans lesion, and 21 (40.4%) patients had multiple erythema migrans lesions (Table 1). Thirty-four (65.4%) patients were male, mean age ± SD was 50.2 ± 15.7 years (range 20–86 years), and 39 (75.0%) patients had concomitant subjective symptoms such as fatigue. All but 4 patients had been exposed to ticks while in areas that included the Lower Hudson Valley (Table 1).

Convalescent-phase blood samples, which were obtained at a mean of 16.7 days (range 7–30 days) after the baseline visit, were screened for antibodies to the C6 peptide of B. burgdorferi and for antibodies to A. phagocytophilum, B. microti, GlpQ protein of B. miyamotoi, and POWV. A total of 46 (88.5%) patients were seropositive by the C6 Lyme ELISA, 32 (69.6%) on both the baseline and convalescent-phase blood samples, and 14 (30.4%) on the convalescent-phase sample only.

None of the 52 patients had evidence of A. phagocytophilum co-infection, although 4 had an IgG titer of 1:64 on the convalescent-phase blood sample, which is regarded as a nonspecific finding (13,17,18). Titers <1:64 were considered to be negative by the performing laboratory and thus were not reported.

Of the 52 patients, 4 (7.7%, 95% CI 3%–18%) had convincing evidence of B. microti co-infection (Table 2), and active babesiosis was clinically suspected for 3 of these patients. For 1 of the 3 patients who had clinical evidence of active babesiosis and a single erythema migrans lesion, fever developed on day 4 of amoxicillin therapy. Another of these patients underwent diagnostic testing for babesiosis because of fever before development of a single erythema migrans lesion. The third patient, who was afebrile, was tested for babesiosis 2 days after beginning antimicrobial drug treatment for a single erythema migrans lesion because of thrombocytopenia and anemia that were documented at the time of study entry. These 3 patients were positive for B. microti DNA by PCR, and 1 of the 3 also had a positive blood smear. All 3 patients received a course of treatment for babesiosis.

A fourth patient without a febrile illness had a convalescent-phase IgG titer of 1:512 and an acute-phase titer of <1:64, consistent with co-infection with B. microti. In addition, 2 patients without clinical evidence of a febrile illness had acute- and convalescent-phase IgG titers of >1,024 (Table 2). Because the exact titer for these serum specimens was not determined, it was not possible to determine if the convalescent-phase sample demonstrated a 4-fold increase in titer. Although none of these 3 patients received anti-Babesia drug therapy, all recovered fully from Lyme disease during the 1-year follow-up period.

Therefore, up to 6 (11.5%; 95% CI 5%–23%) of the 52 patients may have been co-infected with Babesia; 3 of these patients were known to have had fever, hematologic findings consistent with active Babesia infection, or both. All 6 had >1 nonspecific symptoms at study entry; mean ± SD was 3.3 ± 3.4 symptoms (range 1–10 symptoms). In comparison, the mean number of symptoms for the other 46 patients at the baseline visit was 3.1 ± 3.3 (range 0–12 symptoms; p = 0.88). One additional patient had a convalescent-phase IgG titer of 1:128 and an acute-phase IgG titer of 1:256, possibly indicative of a prior Babesia infection antedating the onset of Lyme disease.

None of the 52 patients met criteria for serologic evidence of B. miyamotoi co-infection, although 4 (7.7%) were seropositive for antibodies to GlpQ on acute- and convalescent-phase serum samples but without a discernible increase in values on the convalescent-phase sample. In addition, none of the 52 patients met criteria for serologic evidence for possible or confirmed POWV co-infection; 1 serum sample was positive for IgM to the LB strain of POWV but negative for neutralization antibodies to both the LB strain and a deer tick virus subtype strain of POWV. Therefore, more patients had laboratory evidence of co-infection with Babesia than with A. phagocytophilum, B. miyamotoi, or POWV (possibly as high as 6 [11.5%] for Babesia vs. 0 for the other pathogens tested; p = 0.031).

In this study of 52 adult patients who had erythema migrans but had not been treated for Lyme disease, conducted in the Lower Hudson Valley of New York, the only documented B. burgdorferi co-infection was with B. microti. Several prior studies that used PCR have evaluated I. scapularis ticks found in this region for co-infection with B. burgdorferi; these studies found values of up to 30% for co-infection with A. phagocytophilum (4,5,9,19), up to 24% for B. microti (4,5,9,19), 1% for B. miyamotoi (4), and up to 3.9% for POWV (4,9). In general, lower rates of co-infection were associated with I. scapularis ticks in the nymphal stage than in the adult stage; this finding is relevant to our study because most cases of early Lyme disease in this region result from bites of ticks in the nymphal stage (20).

Extrapolating data on the rate of co-infections by PCR testing of ticks to human co-infection rates should be done cautiously. Confounding factors are the possible existence of nonpathogenic strains of Anaplasma or Babesia in ticks and whether these organisms may have contributed to a portion of the positive PCR results for A. phagocytophilum (21) or B. microti. For example, B. odocoilei is found in I. scapularis ticks but is not regarded as a human pathogen (22). In addition, the potential tick exposure locations of participants in our study were not restricted to the Lower Hudson Valley region of New York. Indeed, tick exposure for 4 of the 52 study participants occurred exclusively in Long Island or Connecticut (Table 1), and 1 of these 4 participants was among the 6 participants with laboratory evidence of B. microti co-infection (Table 2). This participant had no clinical evidence of a febrile illness but had acute- and convalescent-phase B. microti IgG titers >1,024.

We systematically evaluated adult patients with erythema migrans for co-infection with 4 I. scapularis tick–transmitted pathogens. We used well-defined and highly rigorous criteria for defining co-infection and focused on consecutively enrolled patients with the most certain clinical marker of early Lyme disease; namely, an erythema migrans lesion (12). Studies using less stringent case definitions may potentially detect higher numbers of putative co-infections but with less certain validity and less clarity for differentiating sequential from simultaneous infections (13,23). Unlike a previous study of untreated Babesia co-infections in patients considered to have Lyme disease (24), patients in our study with evidence of Babesia co-infection at baseline evaluation were not more symptomatic than those without this co-infection.

Limitations of our study are the relatively small sample size and the assumption that the convalescent-phase serum samples were obtained at the appropriate time to reliably identify the co-infections assessed (mean time from baseline visit to collection of the convalescent- phase blood sample was 16.7 days [range 7–30 days]). Most (75%) of the 52 patients in our study had received doxycycline, raising the question of whether this treatment may have affected the likelihood of seroconversion for antibodies to A. phagocytophilum. However, patients with culture-confirmed human granulocytic anaplasmosis regularly produce high antibody titers within 2 weeks of symptom onset despite receipt of doxycycline (17). Thus, the only theoretical concern about whether doxycycline might have reduced the observed frequency of A. phagocytophilum co-infection would have been for cases of incubating infection that might have been prevented from becoming active.

Another possibility, however, is that we excluded patients whose fever or systemic symptoms were primarily caused by human granulocytic anaplasmosis, rather than Lyme disease, and thus had started antimicrobial drug therapy before study entry. To address this question, we separately looked at acute- and convalescent-phase antibody titers to A. phagocytophilum in 38 patients with erythema migrans who were enrolled into the same study but for whom antimicrobial drug therapy had been initiated before enrollment. For 1 (2.6%) of the 38 patients, we found a 4-fold rise in antibody titers to A. phagocytophilum between the acute- and convalescent-phase serum samples. However, this finding did not differ significantly from what we found for the 52 patients with erythema migrans (1/38 vs. 0/52; p = 0.42).

Another study limitation is our use of serologic testing assays that were not approved by the US Food and Drug Administration; consequently, their performance characteristics are uncertain. Last, our results pertain to a particular geographic area over a discrete time frame and may not pertain to other locations or other periods.

In conclusion, we systematically and rigorously evaluated consecutively enrolled adult patients with erythema migrans for co-infection with the 4 other I. scapularis tick–transmitted pathogens found in parts of New York and in other geographic areas in the northeastern United States. Nearly 90% of the patients evaluated had no serologic evidence of co-infection. B. microti was the only co-infection found, further documenting the clinical relevance of this emerging infection. Similar studies in other geographic areas, in addition to testing acute- and convalescent-phase serum, should include direct diagnostic testing by use of reliable PCR assays to detect potential co-infecting pathogens (particularly for A. phagocytophilum, B. microti, and B. miyamotoi) at the baseline visit.