Dr. Stith's case descriptions do not allow us to determine whether these infections were examples of human granulocytic ehrlichiosis caused by an Ehrlichia species related to E. phagocytophila or E. equi or examples of human ehrlichiosis caused by E. chaffeensis. Although physicians do not need to distinguish between these infections for case management, assignment of a specific identity to the etiologic agent is useful for epidemiologic purposes.

Human ehrlichiosis is primarily a presumptive clinical diagnosis, and treatment should be promptly initiated when infection is suspected. Definitive diagnosis is difficult in a physician's office or hospital laboratory. Examination of a Giemsa-stained thin blood smear for characteristic morula inclusions in the leukocytes may help establish the diagnosis. The presence of thrombocytopenia and leukopenia, along with elevated liver enzyme levels, provides an incentive for prolonged scrutiny of multiple blood smears, perhaps of smears obtained at different times of the day. A negative blood smear result, however, does not exclude the diagnosis of ehrlichiosis.

The strategy for confirming the presumptive diagnosis of ehrlichiosis rests on seroconversion. In some cases, the acute serum sample might contain specific IgM reactivity but generally is nonreactive. Thus, an acute sample should be stored and analyzed in parallel with a convalescent sample obtained 1 month later. Our experience with human granulocytic ehrlichiosis suggests that multiple convalescent samples might be necessary to show seroconversion [1]. Serologic testing for ehrlichiosis is not yet standardized, however, and the lack of information on the antigen used for Dr. Stith's samples is disturbing but not unusual. Although some commercial laboratories may use E. canis or E. equi antigen (which cross-reacts with E. chaffeensis and the agent of human granulocytic ehrlichiosis, respectively), experience with rickettsial immunofluorescence techniques vary widely.

Additional diagnostic support may be obtained from the Centers for Disease Control and Prevention or some academic centers. Anticoagulated blood samples should be obtained before treatment is initiated. Such material may be used in DNA amplification assays or in attempts to directly isolate the agent. We have recently shown the utility of mouse inoculation for the diagnosis of human granulocytic ehrlichiosis (unpublished data). This procedure is similar to the hamster inoculation, which serves as the confirmatory gold standard for Babesia microti infection [2]. Some research laboratories might offer (as a professional courtesy or as a collaboration) polymerase chain reaction, mouse inoculation, or in vitro cultivation, as well as serologic testing.

We cannot overemphasize the importance of reporting all cases of ehrlichiosis to state health departments. By doing so, we may more accurately estimate the public health burden of these emergent tick-borne zoonoses.