How to Improve Microbacterial Documentation in Febrile Neutropenia: Impact of Reduced Time Between Collection and Incubation, and New Technics of Identification.

Febrile neutropenia (FN) is the most frequent complication of chemotherapy-induced neutropenia. Only 30% of the episodes of FN are microbiologically documented by routine procedures, mainly blood cultures (BC). In order to improve the diagnostic yield of BC in this setting, we prospectively assessed new procedures of BC in FN, by implementing a BC automate in the ward and combining molecular and rapid identification methods directly applied on the blood culture bottle when detected as positive.

This is a prospective study, run in the Hematology department of Henri Mondor Hospital. Adult patients were eligible at their first episode of FN if they were neutropenic (PMN<0.5 ×10⁹/L), febrile (temperature ≥ 38°3 or twice ≥ 38° within 8h), and had a central venous catheter (CVC). For each episode we collected: 2 standard BC sampled through CVC and one through peripheral punction. These BC were incubated in the laboratory and processed according to the routine procedure for identification (morphologic, culturing and biochemical characters including API, Biomérieux) (routine process); 1 study BC (2 aerobic and 2 anaerobic bottles) sampled through the CVC and immediately incubated in the ward (BacT/Alert3D®), then processed directly with Vitek2® (Biomérieux) and Genotype® (Hain Lifescience) for identification (study process). We compared the routine process with the study process on : (1) the rate of BC positivity; (2) the time elapsed between blood sample and BC positivity; (3) the time elapsed between blood sampling and final pathogen identification. Pairwise analyses were performed using the McNemar test to compare documentation rates of both techniques, and the Wilcoxon signed-rank test to compare delays.

120 consecutive episodes of febrile neutropenia were included from Feb 2008 to Mar 2009: 97 (80%) acute leukemia/myelodysplasia, 26 (17%) lymphoid malignancies and 4 (3%) others. The median age was 50y (20-75). Eighteen (15%) episodes were at the initial phase of allogeneic stem cell transplantation (SCT), and 16 (13%) at the initial phase of autologous SCT. 34 episodes (28%) were microbiologically documented by routine BC, 33 (27%) by study BC, without significant difference between the 2 methods (McNemar's chi2=0.20, p=0.65). Fifty bacteria were identified from BC: 24 (48%) streptococci sp., 14(28%) staphylococci, 9 (18%) gram-negative bacteria, and 3 (6%) others. No fungi was identified. In the microbiologically documented episodes, the time elapsed between incubation and positivity in BactAlert was significantly shorter for the study BC (median rank: 740 min [662-947]) than for the routine BC (median rank: 800 min [726-1079]) (p=0.0003). The time needed for final identification was significantly shorter in the Vitek2®+Genotype® study BC (respectively 1.5 days [CI 1.2-2.7] and 1.2 [CI 1.0-2.9) days]) than in the routine BC (median rank: 3.0 days [CI 2.2-4.4]), (respectively p=0.018 and p=0.047).

Immediate incubation combined with the use of rapid method for early identification does not improve the rate of positivity of blood cultures in FN episodes, when compared to BC routine process. However, it reduced the time to positivity about 60 minutes and time to identification about 1.5 days. This could allow an earlier adaptation of antibiotics for the pathogens resistant to the initial antibiotic therapy. Further studies are needed to evaluate the clinical impact of these results.

No relevant conflicts of interest to declare.