 |
| About Bioline | All Journals | Testimonials | Membership | News |
|
||||||
|
||||||
Indian Journal of Medical Microbiology, Vol. 29, No. 3, July-September, 2011, pp. 283-287 Original Article Direct inoculation on Phoenix panels for identification and antimicrobial susceptibility from positive BACTEC cultures: First study from India S Duggal1, SK Jesaiwal2, N Tandon2, TD Chugh2 1 Department of Microbiology, Jr Specialist, Baba Saheb Ambedkar Hospital, Rohini, Delhi-110085, India Date of Submission: 10-May-2011 Code Number: mb11068 PMID: 21860111 DOI: 10.4103/0255-0857.83914 Abstract Purpose: This was a prospective study planned in a super-specialty hospital in Delhi to reduce turnaround times of identification-susceptibility results of positive blood cultures. Materials and Methods: One hundred consecutive single morphology non-duplicate cultures were inoculated on Becton Dickinson Phoenix™ panels by growth recovered directly from liquid BACTEC™ media and after pure growth on solid media. Results: Complete concordance was observed in 72.4% of gram-negative and 45.8% of gram-positive isolates. For gram-negative isolates, categorical agreement (CA) was >83% and essential agreement (EA) was >96% among all antibiotics tested, very major errors (VME) were 0.13%, major errors (ME) 0.54%, and minor errors (MiE) were 3.01%. For gram-positive isolates, VME was 0.73%, 1.10% MiE and no ME. It was observed that average time from receipt of specimen to release of reports was 30:34 h and 32 h for gram-negative and gram-positive isolates if reports of "Direct" panels were to be released. Conclusions: By direct panel inoculation, a decrease of at least 18-20 h in turnaround time was observed compared with the standard method. This helps early change to effective antibiotic therapy and also reduces the expenditure incurred for a patient's hospital stay by average Rs 20,000 ($443) per day.Keywords: BACTEC, blood cultures, direct inoculation, Phoenix, panels Introduction With the advent of automation in Microbiology, fast and reliable reports are expected. Blood culture system BACTEC 9120 and Phoenix 100 for identification/antibiotic susceptibility testing are faster methods to conventional but involve subculture of positive flagged bottles on solid media from which the panels are inoculated next day (growth panel). [1],[2] This was a prospective study planned for diagnosis of invasive infections faster by inoculating directly from the blood culture bottles onto the Phoenix panels (direct panels). It was conducted over 7 months (February-September 2010) and results of 100 direct and growth panels were compared, including gram-positive and gram-negative organisms. Materials and Methods All blood cultures were done on BACTEC 9120. The protocol described by the manufacturers [1],[2] was compared with a protocol developed for direct inoculation of panels. After a bottle signalled positive, a direct smear was gram stained and for single morphology cultures, direct panels; gram-positive or gram-negative, were set up in Phoenix 100 within 1 h of the signal. Simultaneously, a subculture was done from the Bactec bottle on blood and MacConkey agars for growth of colonies and the next day, a panel from pure growth was set up. Two millilitres from positive BACTEC bottle was taken in serum separator tube (SST) and spun at 3000 rpm (1157 g) for 15 min (Eppendorf centrifuge 5804). After discarding the supernatant, Phoenix identification broth was added to the pellet in SST and mixed by inverting 2-3 times followed by vortexing. [3] Standard inoculum was thus prepared; adjusting its turbidity to 0.5-0.6 McFarland. Phoenix susceptibility broth was prepared by adding 25 μL of standard inoculum and one drop of indicator into this broth, mixed by inverting and inoculated on susceptibility side of the panel. The panels were then sealed and loaded into Phoenix system within 30 min of inoculation. [1],[2] From the standard inoculum, a Blood agar plate was always inoculated to check for its purity. During the study period, 2 cultures were found polymicrobial based on gram stain and therefore, excluded from the study. A total of 100 BACTEC culture positive samples were analysed and the results from direct as well as growth panels compared, considering growth panel as reference method. Reports were released based on findings of the growth panel. The results were compared in terms of differences in identification, confidence values in identification, time to completion of result and categorical and essential agreement in antibiotic sensitivity. Complete concordance was defined as agreement between the two isolates to species level; species discordance, if the species were different; genus discordance if the genus was also different. In terms of antimicrobial susceptibility testing, it was said to have ′essential agreement′ (EA) when MIC was within ± one 2-fold dilution compared with the reference method; categorical agreement (CA) when interpretive results (susceptible, intermediate, or resistant) with the reference method was the same. The errors were labelled as very major error (VME) if the direct panel reported an antibiotic as sensitive as compared to growth panel (reference method) reporting it as ′resistant′; major error (ME) if interpretation was ′resistant′ compared to ′sensitive′ and minor error (MiE) if interpretation was ′sensitive/intermediate′ or vice versa and ′resistant/intermediate′ or vice versa compared with the reference method. [4] Results A total of 100 isolates were compared, 76 gram-negative and 24 gram-positive isolates. Majority of gram negatives were Salmonella spp. (31), followed by Escherichia coli (20), Klebsiella spp. (10) Enterobacter cloacae (1), Proteus mirabilis (1), Acinetobacter spp. (8), Pseudomonas spp. (3), Stenotrophomonas maltophilia (1) and Burkholderia gladioli (1). Amongst gram-positive isolates, majority were staphylococci - [5 Staphylococcus aureus, 14 coagulase-negative staphylococci (CoNS)], Enterococcus faecium (3) and 2 streptococci (S. pneumoniae, S. vestibularis). None of the cultures during the study yielded mixed growth on subculture. Concordance in identification Phoenix gave no identification if the confidence value was <90%. This was seen in 7 gram-negative (3 Salmonella, 2 Acinetobacter, Proteus and Burkholderia one each) and one positive (CoNS) isolate. Concordance of other isolates was matched. There was no genus discordance among gram-negative isolates. Amongst Salmonella spp., complete concordance was observed in 19 (67.8%), species discordance in 9 (32.1%) cases. After serological confirmation, it was found that direct panels correlated better than growth panels in 4 cases. Corrected concordance therefore was 82.1%. Amongst other members of Enterobacteriaceae tested (31), there was only one species discordance (3.2%), others (96.8%) were completely concordant. Klebsiella pneumoniae isolates were comparable to subspecies level except in one. Among 10 non-fermenters, complete concordance was seen in 6 cases with 4 species discordances. Eight out of 18 staphylococcal isolates showed complete concordance (44.44%), 5 species and 5 genus discordances. Amongst enterococci, concordance in identification was seen in 2 out of 3 cases, and one showed genus discordance. Streptococcus pneumoniae showed complete concordance while the other streptococcus showed species discordance. Confidence values of direct and growth panel identification results were compared. Discordant identification results are shown in [Table - 1]. It was observed that amongst completely concordant panels, better or same confidence values in direct panels was observed in 8 (72.7%) out of 11 gram-positive and 32 (59.25%) out of 54 gram-negative panels. Antimicrobial susceptibility results were comparable in 76 out of 78 gram-negative bacteria. The other two panels where these results could not be read were ′CDC group Vb-3′ in growth panel (susceptibility not included in Phoenix taxonomy) and ′Klebsiella pneumoniae ssp. pneumoniae′ in direct panel (no growth on susceptibility side of panel). The CA was >93% amongst all antibiotics tested. For Salmonella isolates, ampicillin, chloramphenicol, ciprofloxacin, levofloxacin, ceftriaxone and trimethoprim-sulfamethoxazole were compared. Antimicrobial susceptibility was comparable in all 31 isolates. For ciprofloxacin, CA was 83.9% and EA was 96.8%; all others had 100% categorical and essential agreement. All these isolates were nalidixic acid resistant (tested by disc diffusion). [5] For other gram-negative isolates, ampicillin, amoxycillin-clavulanic acid, cefuroxime, ceftazidime, ceftriaxone, cefepime, aztreonam, piperacillin/tazobactam, imipenem, meropenem, gentamicin, amikacin, tobramycin, ciprofloxacin, levofloxacin, tetracycline and trimethoprim-sulfamethoxazole were compared for 43 isolates where susceptibility results were available. The CA varied between 88.4% for piperacillin/tazobactam, 90.7% for meropenem, 93% for tobramycin and levofloxacin, 95.3% for imipenem and amikacin; 97.7% for gentamicin, ceftazidime, ceftriaxone, cefepime and trimethoprim-sulfamethoxazole to 100% for others. The EA was 95.3% for piperacillin/tazobactam, imipenem, meropenem and trimethoprim-sulfamethoxazole, 97.7% for tobramycin, ceftriaxone to 100% for others. For other antibiotics it was 100%. Overall, for gram-negative isolates, VME was only one (0.13%), ME were 4 (0.54%) and MiE were 22 (3%). Similarly, susceptibility results were comparable in 15 of 24 gram-positive isolates. This was attributed to ′susceptibility not included in Phoenix taxonomy′ in 8 cases and ′insufficient growth in susceptibility side of panel′ in one case, all in direct panels. For staphylococci (13), penicillin- G, ampicillin, oxacillin, cefazolin, cefalothin, ampicillin-sulbactum, trimethoprim-sulphamethoxazole, vancomycin, clindamycin, erythromycin, quinupristin-dalfopristin, levofloxacin, gatifloxacin, moxifloxacin, rifampin, gentamicin, tetracycline, chloramphenicol and linezolid were compared. CA varied between 84.6% for chloramphenicol, 92.3% for gentamicin, tetracycline and gatifloxacin to 100% for others. EA was 92.3% for chloramphenicol and tetracycline but 100% for others. Two enterococcal isolates were compared for ampicillin, trimethoprim-sulphamethoxazole, vancomycin, erythromycin, linezolid, quinupristin-dalfopristin, levofloxacin, gatifloxacin, moxifloxacin, gentamicin-synergy, streptomycin-synergy, and tetracycline and both CA and EA were 100%. For gram-positive isolates, there were no ME, 2 VME (0.73%) and 3 MiE (1.1%). Of 1188 antibiotic-organism combinations tested, CA was present in 1156 (97.3%) combinations [Table - 2]. The time to complete reporting (time from panel loaded in Phoenix to final result) was similar in both the methods. It ranged between 05:51 and 16:22 h (mean, 10:86 h) for gram-negative direct panels and 06:02 and 16:16 h (mean, 11:09 h) for growth panels. For gram positive panels, it was 10:18-15:59 h (mean, 12:88 h) in direct and 10:07-15.55 h (12:81 h) in growth panels. In 11 out of 24 tests, results were faster in direct panels. Time for detection of positive growth in BACTEC varied from 12:68 to 34:9 h (average 18.98 h) and 3:04 to 37:24 h (average 37:24 h) for gram-positive and gram-negative isolates, respectively. It was found that the average time from receipt of specimen to release of reports was 29:31 h (14:55-52:83 h) and 32:62 h (24:13-49:3 h) for gram-negative and gram-positive isolates; if reports of direct panels were to be released. To improve the direct results for positive panels, especially for identification, the technique was slightly modified and 10 new culture-positive gram-positive isolates were tested in Phase II. In this phase, the time and speed of centrifugation were modified. Instead of 3000 rpm for 15 min, these cultures were centrifuged at 5000 rpm (3214 g) for 10 min before panel inoculation. This method was compared with results of manual identification and susceptibility testing. [5] There were no major discordances. Complete concordance was seen in 7 cases (in other three, S. aureus was identified as S. haemolyticus, coagulase-negative Staphylococcus as S. aureus and one isolate was novobiocin-sensitive but identified as S. saprophyticus). It did not give susceptibility results for Streptococcus mitis and Streptococcus pneumoniae (Streptococcus panels are recommended but not used). However, when Identification was manually entered as Enterococcus spp., susceptibility results could be retrieved and they matched with each other and were comparable with disc diffusion results. Discussion Blood cultures are gold standard for diagnosis of bacteremia, and rapid results can significantly change the treatment outcomes and cost of stay in the hospital. [6] Phoenix uses an optimized colorimetric oxidation-reduction indicator for susceptibility testing and a variety of fluorometric and colorimetric indicators for bacterial identification. The time to completion of result till final susceptibility ranges from 7.5 h to 16 h, with a mean of 10.5 h [1],[2] In the present study, the mean was 10.86 h for gram-negative and 12.88 h for gram-positive direct panels. According to the US Food and Drug Administration, [4] the percentage categorical and essential agreement must be greater than 89.9%. For all antibiotics, both gram-positive and gram-negative, the results clearly had very good agreement. CA was 83.87% for ciprofloxacin and 88.37% for piperacillin/tazobactam but EA was 96.77% and 95.34%, respectively (ie, >90%), therefore, were considered to be in agreement. In addition, the major error rate must be less than 3% of all the susceptible organisms tested and the very major error rate of 1.5% or less [4] and our results were much below the acceptable limits. There was no genus discordance in gram-negative isolates. Species discordances were seen in Enterobacteriaceae (Klebsiella pneumoniae identified as Klebsiella oxytoca and others were among Salmonella spp.). Overall concordance for gram-negative isolates was 80.26% and for gram-positive isolates it was 53.8%, which increased in phase II. However, further refinements are needed in the procedure for better results. The results are similar to a study of gram-negative isolates on Vitek 2 system [7] and Phoenix [8] where 82.2% and 95.7% of these strains were correctly identified to species level, 80.7% gram-positive isolates were correctly identified by direct inoculation. [8] VMEs in the present study were 0.13%, 0.73% for gram-negative and gram-positive isolates, respectively. These values are comparable to 0% and 1.1% in a study. [8] Another study comparing direct versus growth panels for gram-negative isolates has shown 0.1% very major errors, 0.1% major errors, and 0.8% minor errors. [3] Gram-positive organisms, such as Bacillus cereus, Pediococcus pentosaceus, Alloiococcus otitidis and Pasteurella pneumotropica were identified without susceptibility results in 5 direct panels; as they were not in Phoenix taxonomy. In India, till the time this study was conducted, only combined Identification/susceptibility panels were being supplied. To counter these problems, technique was modified which gave better results, it could reliably differentiate between various gram-positive genera but coagulase testing still needs to be done for staphylococcal isolates. A modified technique has been tried by other researchers using saponin where again some isolates were not identified (3%) and some mis-identified (15%). [9] The mean time for direct and growth panel results for gram-negative and gram-positive panels were comparable; rapidity observed in some panels could be attributed to growth phase of these organisms at the time of panel inoculation. Our laboratory is hospital-based and offers services 24 h a day. From the time gram stain results are communicated to a clinician to the final report, the expectation of patients and clinicians is very high. With this method of direct panel inoculation, we were able to deliver results at least 18-20 h earlier. The results were well within acceptable limits in terms of all parameters and small revisions in protocol were made to ensure better results. However, few rapid tests, such as coagulase test for staphylococcal isolates and serotyping for Salmonella isolates must be done from growth on solid media before issuing final reports. This is the first study for direct panel inoculation from positive blood cultures in India. Also no single study has compared results of Salmonella isolates from blood cultures on similar lines. Magnitude of Salmonella infections is very high in Southeast Asia and the results for these isolates are encouraging. For the endemic and other regions, direct panel inoculation provides a solution for early detection of salmonellosis; and to prevent its complications. To conclude, the results of direct Phoenix panels both for gram-negative and gram-positive pure isolates from BACTEC are reliable to be used for clinical specimens. Direct inoculation saves time (almost one day) and therefore proves cost-effective. This is important for managing invasive bacterial infections, where culture is the gold standard of diagnosis. Therefore, we recommend setting up Direct Phoenix panels for all gram-negative and gram-positive isolates from positive BACTEC bottles to have reliable and faster results for the benefit of patients. Acknowledgement We would like to acknowledge the help from Becton Dickinson, Gurgaon, India, in conducting this study. References
Copyright 2011 - Indian Journal of Medical Microbiology The following images related to this document are available:Photo images[mb11068t2.jpg] [mb11068t1.jpg] |
| |||||||||
