Peptide Nucleic Acid Fluorescent In Situ Hybridization for Hospital-Acquired Enterococcal Bacteremia: Delivering Earlier Effective Antimicrobial Therapy

Study design.This was a quasiexperimental study (one group, with pretest and posttest design) performed at the University of Maryland Medical Center over 2 years (2005 and 2006). The University of Maryland Institutional Review Board approved the study.

Setting and patients.The University of Maryland Medical Center is an ∼600-bed inner-city tertiary-care teaching hospital that includes 130 ICU beds, the Marlene and Stewart Greenebaum Cancer Center, and a large solid-organ transplant program. All patients with hospital-acquired bacteremia due to E. faecium or E. faecalis during the study period were included as study subjects. Hospital-acquired bacteremia was defined as either type of bacteremia occurring more than 72 h after admission. This included the length of stay after transfer from another hospital setting (18). The AMT did not make interventions for patients with streptococcal infections, as these are often community acquired, present with admitting cultures, occur more often, and are frequently covered by many classes of antibiotics and would consume too much of the AMT time with minimal changes to antibiotic therapy. Only the first episode of enterococcal bacteremia per patient per admission was included. The following patient groups were excluded from the study analysis: pediatric patients and R. Adams Cowley Shock Trauma Center patients (because these patient groups are not evaluated by the hospital's AMT), patients in either group who died before the culture was positive (because the AMT was unable to intervene), and patients with documented endocarditis based on modified Duke criteria (because the developed treatment algorithm did not apply) (17). Surveillance for VRE with perirectal cultures was standard practice for all patients on the day of admission to an ICU or the cancer center and was then performed weekly during both the pre- and postintervention periods. VRE alerts were present in the chart for current and prior VRE colonization, and patients were appropriately isolated. A prior VRE culture in both time periods warranted early linezolid therapy.

Laboratory.Blood cultures were drawn according to standard hospital policy from two separate sites and collected in blood culture bottles (BacTAlert; bioMerieux, Marcy l'Etoile, France). The cultures were placed in a continuous automated detection incubator (BACTEC; Becton Dickinson, MD). When GPCPC were identified by Gram staining signal-positive blood cultures, they were plated onto standard growth media according to the standard laboratory protocol. These results were called in to the treating physician according to normal policy. When there was growth on the plates, enterococci were identified (catalase-negative or alpha- or nonhemolytic colonies) using catalase detection with 3% H₂O₂ and a pyrrolidonyl arylamidase-positive result (Hardy Diagnostics, Santa Maria, CA) (3, 4). Subspecies identification was performed using Vitek 2 (bioMérieux, Marcy l'Etoile, France), and susceptibility tests were performed using a disc diffusion assay following the Clinical and Laboratory Standards Institute (CLSI) protocols (4). PNA FISH testing for Enterococcus species was batched and performed twice a day (8 a.m. and 3 p.m.) following manufacturers' instructions during 2006 (8, 32). After identification of GPCPC from blood culture, a drop of blood was placed on a slide and the probe was added; the slide was then placed in a water bath for 90 min, washed for 30 min, and read under a fluorescence microscope (8, 32). Green fluorescence indicated E. faecalis, while red fluorescence indicated other Enterococcus species, including E. faecium. The PNA FISH results were called in to the physician with follow-up by the AMT. The final cultures were compared to the PNA FISH results.

Intervention.The study subjects were derived from two consecutive time periods. During 2005, E. faecium and E. faecalis were identified using standard microbiological methods, and AMT action on antimicrobial therapy was directed by clinical factors and the final susceptibility results. During 2006, E. faecium and E. faecalis were identified using PNA FISH and standard microbiological methods. The AMT intervened at the time of the PNA FISH results to direct antimicrobial therapy.

Based on our antibiograms from 2004, 100% of our E. faecalis organisms were ampicillin susceptible and 90% were vancomycin susceptible. Of our other Enterococcus species, 99% were E. faecium. None of the E. faecium isolates were ampicillin susceptible based on CLSI standards, and only 15% were vancomycin susceptible. With these data, the AMT developed the treatment algorithm shown in Fig. 1. The AMT discussed the PNA FISH results daily, including weekends, with the treating physician, who then determined whether a change in therapy was necessary. Prior VRE colonization and VRE surveillance cultures were also taken into consideration during the entire study period. The treatment algorithm had preferred therapies with ampicillin for E. faecalis (vancomycin when a patient was allergic to penicillin) and linezolid for E. faecium, and if the organism was not an Enterococcus species, then penicillin or ceftriaxone (vancomycin could be added if meningitis or febrile neutropenia was present) was used. High-dose daptomycin was considered as an alternative to linezolid when necessary with vancomycin-resistant E. faecium (29). Because enterococcal bacteremia is frequently associated with other organisms, especially gram-negative bacteria (28), piperacillin-tazobactam or imipenem/cilastatin could be utilized in place of ampicillin, as both these antibiotics have adequate E. faecalis activity (19, 23, 36). Ampicillin susceptibility was used as a marker for imipenem/cilastatin (36) and piperacillin-tazobactam (19) susceptibility. All central lines were removed after documentation of enterococcal bacteremia.

• Open in new tab
• Download powerpoint

FIG. 1.

Antibiotic treatment algorithm utilized by the AMT with results from the EFOE PNA FISH test. Footnotes: a, vancomycin could be substituted for patients allergic to penicillin; b, high-dose daptomycin could be substituted if a patient is unable to take linezolid; c, vanocmycin could be added with febrile neutropenia or suspected meningitis. Other enterococci were usually E. faecium, with >90% vancomycin resistant.

Data collection.We collected the following information on study subjects via chart review: age, sex, and length of stay; central-line status, APACHE II score at the time of blood culture (21), serum chemistry, complete blood count, and hospital location (all at the onset of bacteremia); and time from report of Gram stain to final enterococcal susceptibilities, other organisms in the blood culture, initial antimicrobial therapy, time to change antimicrobial therapy to appropriate therapy after report of GCPC, and survival status at 30 days. Neutropenia was defined as having an absolute neutrophil count of less than 500 cells at the onset of bacteremia. The prospective intervention group also included the time from the Gram stain to the PNA FISH report and the accuracy of the PNA FISH result compared to the final culture.

Appropriate antimicrobial therapy was defined as the receipt of an antibiotic with activity against the cultured enterococci. The time to appropriate therapy was the period from the time the blood culture was drawn to the time of receipt of appropriate therapy (21). We assessed whether the selected empirical therapy had coverage of all organisms from the blood culture based on their final susceptibilities. Mortality was all-cause mortality and was determined 30 days after the first positive blood culture.

Statistical analysis.The primary objective of this study was to determine if the EFOE PNA FISH test would lead to earlier initiation of appropriate antimicrobial therapy for patients with E. faecium and E. faecalis bacteremia. The secondary objective was to determine the effects of PNA FISH on survival and hospital length of stay. We initially described the study population using proportions for categorical variables and medians for continuous variables. Because the intervention was likely to have a greater effect in patients with E. faecium bacteremia than in those with E. faecalis bacteremia, we stratified our results by enterococcal species. We compared categorical variables using chi-square analysis or the Fisher exact test (if appropriate) and continuous variables using a t test in the pre- and postintervention periods, stratified by enterococcal species. The survival curves of the time to effective antibiotic therapy were compared by a log rank test. Data analysis was performed using SPSS version 15 (SPSS, Chicago, IL). All tests were two tailed, and the level of statistical significance was set at a P value of <0.05.

Description of the study population.Overall, there were 650 patients with blood cultures positive for GPCPC. All blood cultures positive for gram-positive cocci in pairs during 2006 had the PNA FISH test in the laboratory. Of these, 271 patients had hospital-acquired E. faecium or E. faecalis bacteremia during the study period, with 50 patients excluded (23 patients had a repeat episode of enterococcal bacteremia, 10 died before the blood culture turned positive, 3 were from pediatric and 7 were from shock trauma units, and 7 had endocarditis), and 379 were infected with a Streptococcal species and were excluded from analysis. Thus, 224 patients were included, with 129 patients in the preintervention period and 95 patients in the postintervention period. The EFOE PNA FISH test identified all 48 E. faecalis organisms and 44 other enterococcal organisms, all of which were E. faecium by probe and confirmed by culture in the postintervention group, and identified three patients with both E. faecalis and E. faecium in the same blood culture.

The study population had a median age of 56 years (range, 19 to 90 years), with 118 (53%) males. One hundred of 224 patients (45%) were in an ICU at the onset of their enterococcal bacteremia, and 40 (18%) were located in the cancer center; 203 (91%) had central venous access. The median APACHE II score at baseline was 14 (range, 3 to 28), and the white cell count at the index blood culture was 9 × 10³ cells/dl. All E. faecalis isolates were ampicillin susceptible; however, only 100 of 112 (89%) were vancomycin susceptible. In contrast, no E. faecium isolates were ampicillin susceptible, and only 13 (11.6%) isolates were vancomycin susceptible. The overall length of stay prior to developing an enterococcal bacteremia was a median of 9 days (range, 0 to 123 days), and the median length of stay after a positive culture was 11 days (range, 1 to 175 days), with an overall 30-day mortality of 24%. The EFOE PNA FISH test was performed on 95 patients (42%). The time to get the final microbiology report from the blood culture draw was a median of 3.7 days (range, 2.0 to 9.8 days) for all 224 patients, while the time to get the PNA FISH report after the blood culture for those patients was a median of 1.1 days (range, 0.5 to 3.5 days).

Comparison of pre- and post PNA FISH interventions.Obtaining species level identification was 2.6 days quicker with PNA FISH than with conventional culture reporting (P < 0.001; t test). The effects of PNA FISH on antibiotic selection and clinical outcomes with E. faecalis and other enterococci separately offer more information.

There were 112 individual episodes of E. faecalis bacteremia evaluated, with 64 in the preintervention period and 48 in the intervention period (Table 1). The clinical characteristics of the two groups were similar, with a slightly greater severity of illness within the PNA FISH group (APACHE II score, 14 versus 12; P = 0.08). The reporting of PNA FISH for E. faecalis was 3 days earlier than for conventional cultures (1.1 versus 4 days; P < 0.001); however, most patients received effective empirical antimicrobial therapy (99% versus 96%; P = 0.4). There was no difference in mortality between the groups (13% versus 10%; P = 0.73).

There were a total of 112 episodes of non-E. faecalis bacteremia. All were E. faecium except one episode of Enterococcus avium bacteremia, which was in the preintervention period. There were 65 patients in the preintervention period and 47 patients in the intervention period. The demographics were similar, with a trend toward more patients with neutropenia from the cancer center in the intervention group (20% versus 36%; P = 0.06). Over 75% of the patients with E. faecium bacteremia were either from the ICU or neutropenic. The median lengths of stay until onset of bacteremia in the groups were similar (12 versus 14 days; P = 0.7). Reporting of E. faecium with the PNA FISH test was a median 2.3 days earlier than with conventional reporting (P < 0.001; t test). The initial empirical antimicrobial therapy was inadequate in 53 of 65 patients (82%) in the preintervention group and 41 of 47 (87%) in the intervention group (P = 0.4; chi-square test). However, as shown in Fig. 2, there was a significant difference between the lengths of the time it took for the intervention group and the preintervention group to receive appropriate therapy (P < 0.001; log rank test). There was also lower mortality in the intervention group (26% versus 45%; P = 0.04), though there was no difference in the length of stay.

• Open in new tab
• Download powerpoint

FIG. 2.

Graph showing the significant decrease in the time to initiating effective therapy using PNA FISH reporting (solid line) versus standard microbiological reporting (dashed line) with E. faecium (P < 0.001; log rank test).