All patients sought treatment for fever ranging from 37.8°C to 39.8°C and admitted to continuing to inject illicit drugs or methadone. Information about patient characteristics, clinical signs and symptoms, laboratory and micobiologic investigations, and treatment details are summarized in the Table. P. larvae was identified in blood cultures (BacT/ALERT 3D-System; bioMérieux, Marcy l’Etoile, France) of each patient described. The clinical course of P. larvae bacteremia was benign in 3 patients, and complications developed in 2 patients. Patient 1 had relapsing disease and spontaneous bacterial peritonitis; patient 4 had pulmonary embolism without definite evidence of septic embolism. Patients 2 and 3 recovered without specific antimicrobial drug treatment; for patients 4 and 5, defervescence and negative follow-up blood cultures were observed after they received treatment with β-lactam agents (imipenem or cefuroxime). The recurrent P. larvae infection observed in patient 1 was probably the consequence of repeated injection of contaminated methadone rather than an inadequate response to antimicrobial drug therapy.

In 2 cases, culture of the honey used to prepare methadone or of honey-containing ready-to-use methadone also yielded P. larvae. Honey and methadone samples were diluted in sterile phosphate-buffered saline and cultured with or without heat pretreatment (90°C, 10 min) under aerobic and anaerobic conditions at 37°C for 3–4 days by using Columbia blood agar and MYPGD (Mueller-Hinton broth, yeast extract, potassium phosphate, glucose, pyruvate) agar. Colonies from positive blood or honey or methadone cultures with an appropriate macroscopic appearance and gram-stain morphology as well as negative catalase reaction were further identified by PCR amplification and 16S rRNA gene sequencing according to published protocols (8). Obtained sequences were analyzed by using the BLAST algorithm (http://blast.ncbi.nlm.nih.gov/Blast.cgi).

We detected P. larvae in sterile compartments of 5 patients with clinical and laboratory evidence of infection. Given the fact that all patients were injection drug users, the mode of infection was thought to be intravenous administration of contaminated methadone, resulting in P. larvae bacteremia. Our hypothesis is supported by the isolation of P. larvae from honey or honey-containing methadone provided to 2 patients.

Recently, several Paenibacillus species have been reported to cause bacteremic infections in humans. Among these are P. thiaminolyticus (bacteremia in a patient undergoing hemodialysis) (9), P. konsidensis (bacteremia in a febrile patient with hematemesis) (10), P. alvei (prosthetic joint infection with bacteremia) (11), and P. polymyxa (bacteremia in a patient with cerebral infarction) (12). Furthermore, the novel species P. massiliensis, P. sanguinis, and P. timonensis were isolated from blood cultures of patients with carcinoma, interstitial nephropathy, and leukemia, respectively (13). Pseudobacteremia of P. hongkongensis and P. macerans has been reported (14,15).

Several aspects provide strong evidence for a genuine P. larvae bacteremia in the cases described here. First, the present cases were observed over a period of several years, and detection of P. larvae thus occurred in different charges of blood culture bottles, which argues against pseudobacteremia. Second, isolation of P. larvae was reported independently by 2 microbiology laboratories, making contamination highly unlikely. Third, in patient 1 isolation succeeded at different times and in samples of different compartments. Moreover, the detection of P. larvae in honey-prepared methadone and honey strongly suggests genuine bacteremia as a consequence of injection of contaminated material.

Biochemical and molecular identification of P. larvae may be difficult and time-consuming. Misinterpretation of blood culture results because of incomplete differentiation or confusion with other gram-positive spore forming-bacteria (e.g., Bacillus species) has to be taken into consideration. Underestimation of the frequency of true P. larvae bacteremia therefore cannot be excluded. Thus, infectious disease physicians, microbiologists, and pharmacists need to be aware that injection of material contaminated with P. larvae, such as honey-prepared methadone, may cause bacteremic infection.