EUCAST has developed a direct antimicrobial susceptibility testing (AST) protocol for automated blood cultures. However, its dependence on mass spectrometry-based microbial identification can be obviated by using a direct inoculum preparation protocol in an automated microbial identification system. This approach can provide AST reports within 24 h of sample collection.
Gram-negative (GN) sepsis is a medical emergency where management in resource-limited settings relies on conventional microbiological culture techniques providing results in 3-4 days. Recognizing this delay in turnaround time (TAT), both EUCAST and CLSI have developed protocols for determining AST results directly from positively flagged automated blood culture bottles (+aBCs). EUCAST rapid AST (RAST) protocol was first introduced in 2018, where zone diameter breakpoints for four common etiological agents of GN sepsis, i.e., Escherichia coli, Klebsiella pneumoniae, Pseudomonas aeruginosa, and Acinetobacter baumannii complex can be reported. However, those clinical laboratories that have implemented this method in their routine workflow rely on mass spectrometry-based microbial identification, which is not easily available, thus precluding its implementation in resource-limited settings. To circumvent it, we evaluated a direct inoculum protocol (DIP) using a commercial automated microbial identification and antimicrobial susceptibility testing system (aMIAST) to enable early microbial identification within 8 h of positive flagging of aBC. We evaluated this protocol from January to October 2023 to identify the four RAST reportable GN (RR-GN) in the positively flagged aBC. The microbial identification results in DIP were compared with the standard inoculum preparation protocol (SIP) in aMIAST. Of 204 +aBCs with monomorphic GN (+naBC), one of the 4 RR-GN was identified in 105 +naBCs by SIP (E. coli: 50, K. pneumoniae: 20, P. aeruginosa: 9 and A. baumannii complex: 26). Of these, 94% (98/105) were correctly identified by DIP whereas major error and very major error rates were 6% (7/105) and 1.7% (4/240), respectively. When DIP for microbial identification is done using the EUCAST RAST method, provisional clinical reports can be provided within 24 h of receiving the sample. This approach has the potential to significantly reduce the TAT, enabling early institution of appropriate antimicrobial therapy.
Sepsis, an important global health problem, is defined as life-threatening organ dysfunction due to a dysregulated host response to infection. The Global Burden of Diseases Study estimated that there were 48.9 million cases of sepsis and 11 million sepsis-related deaths worldwide in 2017, which accounted for almost 20% of all global deaths1. Around 2/3rd of bloodstream infections (BSI) causing mortality are due to gram-negative bacterial pathogens2. The leading causes of mortality amongst gram-negatives (GN) are Escherichia coli, Klebsiella pneumoniae, Pseudomonas aeruginosa, and Acinetobacter baumannii, which account for around 40% of cases amongst 33 bacterial pathogens2.
Blood cultures remain the gold standard for diagnosing BSI, and rapid microbial identification along with antimicrobial susceptibility testing (AST) results is the key to management. It has been estimated that there is a 9% increase in odds of mortality with each-hour delay in instituting appropriate antimicrobials in sepsis3. The turnaround time (TAT) of microbiologically positive blood culture reports with AST results is around 48-72 h with the available microbiological tools in resource-limited settings, even with automated systems. As a result of this subpar TAT, broad-spectrum antimicrobials are used empirically, contributing to the burgeoning problem of antimicrobial resistance (AMR). Recognizing this dire need to reduce TAT for microbiological culture techniques for sepsis, EUCAST and CLSI are moving towards performing AST directly from positively flagged blood culture bottles (+aBC)4,5.
In 2018, EUCAST first introduced the rapid AST (RAST) method for determining AST by Kirby-Bauer disk diffusion method at short incubation times, i.e., 4 h, 6 h and 8 h, directly from +aBC6,7. The method is presently validated for determining AST for +aBCs containing one of the 8 most common causes of BSI namely E. coli, K. pneumoniae, P. aeruginosa, and A. baumannii complex amongst gram-negatives and Staphylococcus aureus, Enterococcus faecalis, E. faecium, and Streptococcus pneumoniae amongst gram-positives8.The breakpoints for AST determination at various time intervals are provided as per microbial species listed above. Hence, before categorical interpretation of AST results, microbial identification is necessary. However, the RAST standard does not specify the method to enable microbial identification within this time frame.
The majority of studies evaluating the EUCAST RAST method in their setting have used mass spectrometry-based microbial identification after short incubation on plated media to identify micro-organisms9,10,11,12,13,14,15,16,17. However, mass spectrometry instruments are not widely available, especially in low to middle-income countries (LMICs), which greatly limits the potential usefulness of this method. Few studies have reported implementation of this method in their centers without using mass spectrometry18,19,20. Tayşi et al.18 reported a broad categorization of GN amongst Enterobacterales, Pseudomonas, and Acinetobacter spp. based on gram stain morphology and oxidase test before interpreting AST results. In other studies from this center, by Gupta et al.19 and Siddiqui et al.20, species-level microbial identification was done by preparing a bacterial pellet from the positively flagged blood-broth mixture and inoculating it on the conventional biochemical tests. While Tayşi et al.18 did not comment upon the accuracy of microbial identification with their approach, Gupta et al.19 reported that with their approach in 165/176 (94%) cases, a RAST reportable gram-negative (RR-GN), i.e., either of E. coli, K. pneumoniae, P. aeruginosa, and A. baumannii complex. However, with the latter approach, the reading of RAST results was done retrospectively using 8 h zone diameter breakpoints only after the full incubation of conventional biochemical results i.e., 18-24 h post-inoculation, and the average time for reporting was around 2 days.
To reduce the TAT of clinical reports further, we propose an alternative methodology to enable early identification of GN present in +aBCs using aMIAST. Before the introduction of mass spectrometry-based microbial identification systems, these automated identification systems were considered the standard of care for microbial identification, where identification was enabled by colorimetric and/or fluorometric changes induced by test bacteria when inoculated in miniaturized biochemical tests harbored in a cassette and matching the results with their isolate database. The average time to identification in these systems is around 4 h to 8 h however, they are limited by the fact that the manufacturers recommend overnight growth of microbes before their respective identification cards can be inoculated. This requirement greatly limits their usefulness in reducing the time to report.
Few studies have evaluated methods to directly identify microbes from +aBCs using these automated systems21,22,23,24,25,26,27. In the case of +aBCs containing monomorphic GN, the majority of studies showed excellent concordance between direct inoculation from bacterial pellet made from positive blood-broth mixture and standard colony incubation. However, in the case of gram-positives, the concordance rates were suboptimal. As the average time to positivity of +aBCs is between 8 h and 16 h and identification of GN takes ~4 h to 8 h in an automated microbial identification system, we hypothesize that by employing direct inoculation protocol in the automated microbial identification, we can complete the clinical reporting of +aBCs with GN having a RR-GN within 24 h of sample receiving.
Setting for the study
The present study was conducted in the clinical bacteriology laboratory of a 950-bed, academic, tertiary care institute of national importance (INI) in Central India from January to October 2023. The laboratory is equipped with a continuous blood culture monitoring system (CBCMS) and aMIAST. The bacteriology laboratory is functional round-the-clock with the availability of technicians and microbiologists for processing and reporting any positively flagged blood culture bottle (+aBCs).
Microbial methods used here
The workflow of the study is shown in Figure 1. The +aBCs showing monomorphic GNs (+naBC) were processed by direct inoculation of corresponding identification cards to enable identification and AST using EUCAST RAST protocol. These results were compared with the standard-of-care (SoC) method for +aBCs i.e., subculturing on conventional plated media through sheep blood agar (SBA), chocolate agar (CA), and MacConkey agar (MA), incubated aerobically for 16 h to 24 h followed by identification and AST cards given by aMIAST when isolated colonies appear. Blood cultures showing gram-positive cocci, gram-positive bacilli, budding yeast cells, and ≥2 different micro-organisms on initial gram staining or plated media were excluded from the study.
The study, funded by the intramural research grant given to Dr. Ayush Gupta by AIIMS Bhopal, was approved by the Institutional Human Ethics Committee (IHEC) vide letter no: IHEC- LOP/2022/IL072.
NOTE: A sample volume of 5 ml was used based on studies done by Quesada et al.25 and Munoz-Davila et al.27.
1. Standard inoculum protocol (SIP) for bacterial identification using aMIAST
2. Direct inoculum protocol (DIP) for bacterial identification using aMIAST
3. AST by EUCAST RAST protocol4
4. Quality control
5. Statistical analysis
General outcomes
During the study period, 240 +naBCs underwent identification by aMIAST using both DIP and SIP. Of these, 15% (36/240) +naBCs were found to be polymicrobial after overnight incubation on the plated media. Of the 204 +naBCs, the proportion of RR-GN identified by SIP was 51.5% (105/204). Amongst them, 47.6% (50/105) were E. coli, 19% (20/105) K. pneumoniae, 8.6% (9/105) P. aeruginosa and 24.8% (26/105) A. baumannii complex. A detailed description of all identification results by SIP is provided in Table 1.
Diagnostic accuracy in microbial identification
Of 105 RR-GN, 98 (93.3%) and 99 (94.2%) were concordant with the DIP, till species-and genus-level identification, respectively. The organism-wise concordance rates were 94% (47/50) for E. coli, 90% (18/20) for K. pneumoniae, 100% (9/9) for P. aeruginosa and 92.3% (24/26) for A. baumannii complex, as shown in Table 1. In 7 +naBCs, results were discordant till species-level identification using DIP, of which aMIAST either gave unidentified (3) or identified a Non-RR-GN (4), as shown in Table 2. As these results will not compel the clinical microbiologist to report, they were considered major errors (ME). The ME rate in our study was 6.7% (7/105) till species-level identification. The proportion of non-RR-GN in aMIAST by SIP was 48.5% (99/204). Amongst them, 60 (60.6%) were concordant by DIP till species-level identification. Amongst 99 non-RR-GNs, an RR-GN was identified using DIP in 4 +naBCs, as shown in Table 2. Such discordance could have led to a reporting error and was considered a Very major error (VME). The overall VME rate using DIP was 1.7% (4/240). A full description of identification results and errors of all gram-negatives is shown in Supplementary Table 1.
Reduction in time to isolate identification (TTI)
The TTI of concordant +naBCs in DIP was significantly less than the TTI in SIP (median (IQR): 507.5 min (685-404) vs 2171 min (2532-1855), P2 vs P1, p<0.00001 (Mann-Whitney test)). The median difference in TTI between both protocols was 1635 min (IQR: 1964-1299).
Figure 1: Workflow of the study: depicting the workflow in the study for positively flagged blood culture bottle with monomorphic gram-negatives being processed by both Standard and Direct inoculum protocol. Abbreviations: +aBC = positively flagged blood culture bottle, +naBC = positively flagged Blood culture bottle with monomorphic gram-negatives, DIP = Direct inoculum protocol, SIP = Standard inoculum protocol, SBA = Sheep blood agar, CA = Chocolate agar, MA = MacConkey agar, RR-GN = RAST reportable gram-negative, TAT = Turn-around time Please click here to view a larger version of this figure.
Figure 2: Direct inoculum protocol for bacterial identification: showing images of the vials during the performance of direct inoculum protocol. Abbreviations: +naBC = positively flagged blood culture bottle with gram-negatives Please click here to view a larger version of this figure.
RAST Reportable Gram-negatives | Isolates tested (n) | Concordant | Misidentification | No identification |
n (%) | n (%) | n (%) | ||
Total | 105 | 98 (93.3%) | 4 (3.8%) | 3 (2.8%) |
Escherichia coli | 50 | 47 (94%) | 1 | 2 |
Klebsiella pneumoniae | 20 | 18 (90%) | 1 | 1 |
Acinetobacter baumannii complex | 26 | 24 (92.3%) | 2# | 0 |
Pseudomonas aeruginosa | 9 | 9 (100%) | 0 | 0 |
# One isolate correctly identified to the genus level, but not to the species level (Acinetobacter baumannii complex identified as A. haemolyticus) |
Table 1: Results of bacterial identification in direct inoculum protocol. The results are only for RAST reportable Gram-negatives. Abbreviations: n = numerator, % = percent.
Gram-negatives | Total isolates tested | Major Error | Very Major Error, n (%) | ||
Number | Misidentification | No ID | (identified as) | ||
n (%) | n | n | |||
Escherichia coli | 50 | 3 (6%) | 1 (A. haemolyticus) | 2 | 0 |
Klebsiella pneumoniae | 20 | 2 (10%) | 1 (Ralstonia pickettii) | 1 | 0 |
Acinetobacter baumannii complex | 26 | 2 (7.7%) | 2 (A. hemolyticus, Cupriaviadus pauculus) | – | 0 |
Salmonella spp. | 10 | NA | 1 (10%) | ||
(E. coli) | |||||
Enterobacter cloacae complex | 8 | NA | 1 (12.5%) | ||
(E. coli) | |||||
Acinetobacter lwoffi | 14 | NA | 1 (7.1%) | ||
(A. baumannii complex) | |||||
Sphingomonas paucimobilis | 9 | NA | 1 (11.1%) | ||
(K. pneumoniae) | |||||
Abbreviations- n: numerator, %: percent, ID: identification, NA: not applicable, | |||||
A: Acinetobacter, E: Escherichia, K: Klebsiella |
Table 2: Details of major and very major errors by direct inoculum protocol. Abbreviations: n = numerator, % = percent, ID = identification, NA = not applicable, A = Acinetobacter, E = Escherichia, K = Klebsiella.
Supplementary Table 1: Detailed results of organism identification in both protocols along with results of errors in direct inoculum protocol. Please click here to download this File.
Using DIP, we successfully identified the RR-GNs with considerable diagnostic accuracy. The mean TTI after positive flagging of aBC was only 507 min (~ 8.5 h). Thus, when done in conjunction with the EUCAST RAST method for AST determination, it can give isolate identification at 8 h AST reading time. This approach has the potential to implement the EUCAST RAST method obviating the need for mass spectrometry-based identification. This is a boon for the low-resource settings who wish to implement the EUCAST RAST method in their routine workflow to reduce time for clinical reporting and circumvent the major obstacles to its implementation.
Before the introduction of the EUCAST RAST method, multiple authors have evaluated the accuracy of direct testing from +aBC for various aMIAST systems21,22,23,24,25,26,27,29,30,31. In these studies, the direct testing protocols differed in either following a single centrifuge step29,30,31,32 or double centrifuge step21,22,24,25,26,27 for making the bacterial pellet. In the single-step method, the blood-broth mixture from a +aBC was centrifuged at high speed in a serum separator tube (SST) to pellet the bacteria above the silicone gel layer. The pellet was used to prepare the inoculum for the inoculation of identification cards. In the double centrifugation method, the blood-broth mixture from a +aBC was first centrifuged at low speed in an SST to pellet out the blood cells. From this tube, the supernatant containing bacteria was removed and transferred to a new tube and underwent high-speed centrifugation. From this tube, the supernatant was discarded, and the pellet was used to inoculate the appropriate identification cards. In these studies, the concordance rate varied from 62%-100% but in general, accuracy was higher with the double centrifugation method.
We found that the method was simple to perform and was undertaken in a routine diagnostic lab with a workforce of >10 lab technicians on rotation, proving the robustness of the method. In ~94% (98/105) +naBCs containing one of the RR-GN, the DIP correctly identified the micro-organism. The concordance rate for the different categories of organisms was also comparable to each other as they were 94%, 90%, 100%, and 92.3%, respectively for E. coli, K. pneumoniae, P. aeruginosa, and A. baumannii complex. We also found that the overall concordance rate for any gram-negative was suboptimal, ~77% (157/204). However, most of these misidentifications were with non-fermentative gram-negatives such as Acinetobacter spp. other than baumannii complex, Pseudomonas spp. other than aeruginosa, Moraxella spp., and Sphingomonas paucimobilis which are usually considered as common contaminants of skin. Misidentifications with non-fermentative gram-negatives were also noted by other authors21,23, which is likely due to the low reactivity of these bacteria in the aMIAST identification cards.
We found a significant reduction in TTI of bacteria within the +naBCs using DIP. The median TTI of DIP was around 4 times lesser than the median TTI of SIP (507 min vs 2171 min) when done in a routine clinical diagnostic laboratory. This TTI of ~8.5 h also included the interval between positive flagging of aBC and performance of aMIAST card inoculation, as the mean time to isolate analysis by aMIAST was only 5.45 h ± 1.6 h. Adding the mean time to positivity of 728 min ± 301 min (~12 h) for concordant +naBCs in our study, this approach of bacterial identification has the potential to give same-day reporting after receiving the aBC in a routine diagnostic laboratory.
There are certain limitations as well, with the DIP. Firstly, as it is an off-label use of inoculum preparation, results should be considered preliminary and so should the AST results by EUCAST RAST. Nevertheless, it serves the major purpose of identifying only the RR-GNs with considerable diagnostic accuracy in a timely manner. Secondly, there is a practical possibility of wastage of testing resources as in around 60% +naBCs; we would not have been able to report due to either polymicrobial infections or identification of a non-RRGN. This wastage of resources applies to any of the newer automated direct AST methods for blood cultures. Thirdly, the rate of polymicrobial infections and identification of common skin contaminants was higher in this study due to poor sample collection practices. Fourthly, we did not confirm the identity of the tested isolates with mass spectrometry, which is the present gold standard for bacterial identification.
This study successfully establishes that even with phenotypic tests, it is possible to do same-day reporting of positive blood cultures, especially in gram-negative bacteremia. This has the potential of considerably reducing the duration of initiating appropriate antimicrobial therapy in the LMICs where the microbiological diagnostics for bacterial identification and AST rely heavily on conventional phenotypic tests. This approach should be validated by conducting a multicentric study, and its potential impact on patient outcomes, and as an antimicrobial stewardship tool should be the focus for future clinical trials in LMICs.
To conclude, DIP for aMIAST complements the EUCAST RAST method to enable early identification of RR-GNs. This obviates the need to rely on advanced microbial identification and AST techniques as the time to report with these approaches is comparable with this approach. In cases of gram-negative bacteria, same-day reporting is achievable through conventional phenotypic methods, if they are optimized maximally. This has the potential to reduce the duration of broad-spectrum antimicrobial treatment and facilitate antimicrobial stewardship in resource-limited settings.
The authors have nothing to disclose.
The study was funded by the intramural research grant given to Dr. Ayush Gupta by AIIMS Bhopal. We acknowledge the contribution of laboratory technicians and resident doctors who performed and read the tests diligently during routine and emergency hours.
ANTIMICROBIAL DISKS | |||
Amikacin disk 30 µg | Himedia, Mumbai, India | SD035-1VL | Antimicrobial susceptibility testing |
Amoxyclav disk (20/10 µg) | Himedia, Mumbai, India | SD063-1VL | Antimicrobial susceptibility testing |
Cefotaxime disk 5 µg | Himedia, Mumbai, India | SD295E-1VL | Antimicrobial susceptibility testing |
Ceftazidime disk 10 µg | Himedia, Mumbai, India | SD062A-1VL | Antimicrobial susceptibility testing |
Ciprofloxacin disk (5 µg) | Himedia, Mumbai, India | SD060-1VL | Antimicrobial susceptibility testing |
Co-Trimoxazole disk (23.75/1.25 µg) | Himedia, Mumbai, India | SD010-1VL | Antimicrobial susceptibility testing |
Gentamicin disk 10 µg | Himedia, Mumbai, India | SD016-1VL | Antimicrobial susceptibility testing |
Imipenem disk 10 µg | Himedia, Mumbai, India | SD073-1VL | Antimicrobial susceptibility testing |
Levofloxacin disk 5 µg | Himedia, Mumbai, India | SD216-1VL | Antimicrobial susceptibility testing |
Meropenem disk 10 µg | Himedia, Mumbai, India | SD727-1VL | Antimicrobial susceptibility testing |
Piperacillin-tazobactam disk (30/6 µg) | Himedia, Mumbai, India | SD292E-1VL | Antimicrobial susceptibility testing |
Tobramycin disk 10 µg | Himedia, Mumbai, India | SD044-1VL | Antimicrobial susceptibility testing |
ATCC Escherichia coli 25922 | Microbiologics, Minnesota USA | 0335A | Recommended Gram negative bacterial strain for quality control in RAST |
BacT-Alert 3D 480 | bioMerieux, Marcy d’ Etoille, France | 412CM8423 | Continuous automated blood culture system |
Biosafety cabinet II Type A2 | Dyna Filters Pvt. Limited, Pune, India | DFP-2/21-22/149 | For protection against hazardous and infectious agents and to maintain quality control |
Blood agar base no. 2 | Himedia, Mumbai, India | M834-500G | Preparation of blood agar and chocolate agar |
Clinical Centrifuge Model SP-8BL | Laby Instruments, Ambala, India | HLL/2021-22/021 | Centrifugation at low and high speed for separation of supernatant |
Dispensette S Analog-adjustable bottle-top dispenser | BrandTech, Essex CT, England | V1200 | Dispensing accurate amount of saline |
MacConkey agar | Himedia, Mumbai, India | M008-500G | Differential media for Lactose fermenters/ non-fermenters Gram negative bacilli |
Micropipette (100-1000 µL) | Axiflow Biotech Private Limited, Delhi, India | NJ478162 | Transferring supernatant after first centrifugation, discarding supernatant after second centrifugation |
Micropipette tips (200-1000 µL) | Tarsons Products Pvt. Ltd., Kolkata, India | 521020 | Transferring supernatant after first centrifugation, discarding supernatant after second centrifugation |
Mueller-Hinton agar | Himedia, Mumbai, India | M173-500G | Antimicrobial susceptibility testing by Kirby-Bauer method of disk diffusion |
Nichrome loop D-4 | Himedia, Mumbai, India | LA019 | For streaking onto culture media |
Nichrome straight wire | Himedia, Mumbai, India | LA022 | For stab inoculation |
Nulife sterile Gloves | MRK healthcare Pvt Limited, Mumbai, India | For safety precautions | |
Plain vial (Vial with red top), Advance BD vacutainer | Becton-Dickinson, Cockeysville, MD, USA | 367815 | Obtaining pellet after second centrifugation |
Sheep blood | Labline Trading Co., Hyderabad, India | 70014 | Preparation of blood agar and chocolate agar |
SST II tube, Advance BD vacutainer | Becton-Dickinson, Cockeysville, MD, USA | 367954 | Supernatant separation in first centrifugation |
Sterile cotton swab (w/Wooden stick) | Himedia, Mumbai, India | PW005-1X500NO | Lawn culture of blood culture broth for antimicrobial susceptibility testing |
Sterile single use hypodermic syringe 5ml/cc | Nihal Healthcare, Solan, India | 2213805NB2 | Preparing aliquots from +aBC |
VITEK DensiCHEK McFarland reference kit | bioMerieux, Marcy d’ Etoille, France | 422219 | Densitometer to check the turbidity of suspension |
VITEK saline solution (0.45% NaCl) | bioMerieux, Marcy d’ Etoille, France | V1204 | Adjustment of McFarland Standard turbidity |
VITEK tube stand | bioMerieux, Marcy d’ Etoille, France | 533306-4 REV | Stand for proper placement of tubes before ID card inoculation |
VITEK tubes | bioMerieux, Marcy d’ Etoille, France | Tubes for inoculum preparation | |
VITEK-2 Compact 60 | bioMerieux, Marcy d’ Etoille, France | VKC15144 | Automated identification and AST system |
VITEK-2 GN card | bioMerieux, Marcy d’ Etoille, France | 21341 | Identification of Gram negative bacilli |