There are presently many non-culture-based methods commercially available to identify organisms and antimicrobial susceptibility from blood culture bottles. Each platform has its benefits and limitations. However, there is a need for an improved system with minimal hands-on requirements and short run times.
In this study, the performance characteristics of the FilmArray® BCID Panel kit were evaluated to assess the efficiency of the kit against an existing system used for identification and antimicrobial susceptibility of organisms from blood cultures.
Positive blood cultures that had initially been received from hospitalised patients of a large quaternary referral hospital in Durban, South Africa were processed as per routine protocol at its Medical Microbiology Laboratory. Positive blood cultures were processed on the FilmArray BCID Panel kit in parallel with the routine sample processing. Inferences were then drawn from results obtained.
Organism detection by the FilmArray BCID panel was accurate at 92.6% when organisms that were on the repertoire of the kit were considered, compared to the combination methods (reference method used in the study laboratory). Detection of the antimicrobial resistance markers provided by the panel and reference method demonstrated 100% consistency. Blood cultures with a single organism were accurately identified at 93.8% by FilmArray, while blood cultures with more than one organism were identified at 85.7%.
The FilmArray BCID Panel kit is valuable for detection of organisms and markers of antibiotic resistance for an extensive range of organisms.
Blood cultures are a crucial element in the diagnostic workup of newly-admitted patients and for monitoring suspected bloodstream infection in inpatients.
Ideally, for organism identification and susceptibility pattern derivation from blood cultures, a system should have a turnaround time of less than two hours and minimal hands-on time.
Hands-on time with samples refers to the actual time spent in preparing the run for the samples to be tested.
The FilmArray® BCID Panel (bioMérieux Clinical Diagnostics, Salt Lake City, Utah, United States) is a PCR-based molecular platform,
This study was undertaken to assess FilmArray, as an option to improve on blood culture reporting. The plethora of bacteria and yeasts covered by the FilmArray panel is frequently recovered in blood cultures. The genetic mutations that were available as antimicrobial resistance genes are also of interest. In a quaternary reference hospital, organisms recovered were from patients exposed to a variety, and possibly suboptimal doses, of antibiotics as they are referred from various clinics across KwaZulu-Natal. This forms a perfect niche for multidrug-resistant organisms. Carbapenems, amongst others, are extensively used in private and public sectors in KwaZulu-Natal, and estimates of
Ethics approval was obtained from the Biomedical Research Ethics Committee of the University of KwaZulu-Natal (BE456/14) and the study was done on blood cultures received by the laboratory. Consent was obtained from the patients whose samples were used as per the existing agreement between the Durban National Health Laboratory Services Medical Microbiology laboratory, the University of KwaZulu-Natal and the hospital where the study was undertaken.
The study was conducted at the quaternary-level facility for the province of KwaZulu-Natal, South Africa, a regional referral hospital with a bed capacity of 846. Becton Dickinson BACTEC (Becton Dickinson, United States) blood cultures were received as per routine work from all wards and intensive care units of the hospital covered by the Medical Microbiology Laboratory of the National Health Laboratory Services, Durban, KwaZulu-Natal.
Per routine processes, the blood cultures are loaded into the automated blood culture continuous monitoring system BACTEC™ FX System (BD Diagnostics, Franklin Lakes, New Jersey, United States). They remain incubated in the instrument pending organism growth and detection. After five days of no signal received from a blood culture bottle, the bottle is retrieved and resulted as ‘no growth after five days’. Blood cultures that become positive are retrieved and worked up by Gram stain and subculture. Plates from subculture are interpreted by microbiologists and further biochemical or automated tests are utilised. Ultimately, positive blood cultures are reported with organism identity and susceptibility.
In this study, blood cultures received between February and April 2015 that became positive were evaluated using a prospective analytical approach. During the study, there was no request made to wards to send more blood cultures or to restrict the number of blood cultures made available for the study. Any blood culture that became positive was included in the study. The number of positive blood cultures from the same patient, patient name, patient clinical diagnosis or antibiotic received by patient with positive blood culture were not used as exclusion criteria.
Data on positive blood culture results were drawn from our laboratory information system (LIS) – TrakCare Lab, version 6.10, InterSystems Corporation (Cambridge, Massachusetts, United States). The BD BACTEC FX system used for blood cultures and the VITEK® 2 microbial ID/AST testing system (bioMérieux Clinical Diagnostics, United States) used by our combination methods were interfaced to the LIS. Data on identification and susceptibility of organisms recovered by FilmArray were recorded into Microsoft Excel 2013 (Microsoft, Redmond, Washington, United States). Identification and susceptibility patterns of organisms recovered from positive blood cultures by combination methods were drawn from the LIS and corresponding blood cultures run on FilmArray were drawn and compared. Reproducibility was achieved by retesting five random positive blood cultures once each on FilmArray. The results from the first run on the instrument were compared to the second run. Four external controls were used as external standards to confirm the good running of the FilmArray panel for this study.
American Type Culture Collection (ATCC) strains were used as external controls and set up by both subculture and FilmArray. Strains used (all obtained from Davies Diagnostics, Randburg, South Africa) included:
The usual protocol of blood culture workup in the study’s laboratory constituted the
For gram-positive cocci in clusters, blood agar, mannitol-salt agar and DNA plates were set up. For gram-positive cocci in chains, blood agar with optochin disc, MacConkey agar and bile esculin agar were set up. For small gram-positive bacilli, blood agar and bile esculin agar were set up. For large gram-positive bacilli, blood agar and egg yolk agar plates were set up. For gram-negative bacilli, chocolate agar and MacConkey agar were set up. For gram-negative cocci, chocolate agar and MacConkey agar were set up. For yeasts, blood agar and Sabouraud’s dextrose agar were set up. Interpretation and further testing of recovered organism(s) were done the next day.
Colonies of organisms were tested by: catalase, indole and oxidase tests; staphylococcal latex agglutination tests; and streptococcal grouping assays. Germ tube tests were done on suspected yeasts colonies. Suspensions were made of the colonies and set up for API 20E, API 20NE, API 20 Strep, API NH, API Coryne, API 20 C AUX (bioMérieux Clinical Diagnostics, United States) and the VITEK 2 microbial ID/AST testing system (bioMérieux Clinical Diagnostics, United States). Antibiotic susceptibility testing was done using the Kirby-Bauer disk-diffusion testing system for oxacillin resistance screening, screening of extended-spectrum beta-lactamases (ESBLs) and Modified-Hodge test screening. ESBLs are enzymes that confer resistance to many beta-lactam antibiotics.
Each positive blood culture was run on a single FilmArray kit within eight hours of becoming positive. The positive blood culture was Gram-stained, subcultured and simultaneously run on FilmArray. One hundred microlitres of the contents of the blood culture bottle were aspirated, mixed with sample buffer (provided) and loaded into the FilmArray pouch. The pouch was loaded onto the FilmArray instrument, which was connected to a computer system. Using the FilmArray software interface, the different steps of DNA extraction, nested and multiplex PCRs and post-amplification analysis may be visualised and timed. At the end of the run, a report was automatically generated which documented it as having any detectable organism(s) as well as any antimicrobial resistance gene(s) –
In this study, identification and antimicrobial susceptibility of organisms obtained by the FilmArray panel were compared to those obtained by combination methods. Identification that was brought down to the nearest precision possible (genus/species/complex/subspecies level achievable) by FilmArray with respect to the results seen with combination methods was labelled as ‘precisely identified’; organisms that were missed or misidentified (genus/species/complex/subspecies level), despite being on the panel, were labelled as either ‘missed’ or ‘misidentified’. The number of organisms, rather than the number of blood cultures, was used to evaluate the performance of FilmArray in the statistical analyses.
Sensitivity, specificity, positive predictive value and negative predictive value as well as agreement were analysed. Calculations of sensitivities, specificities, positive predictive values and negative predictive values were adjusted according to results obtained. ‘True positive’ was defined as an organism on the FilmArray repertoire, identified by combination methods and also accurately identified by FilmArray. ‘True negative’ was defined as any microorganism not on the FilmArray repertoire and not detected by FilmArray. ‘False positive’ describes any microorganism identified by the FilmArray that was not mentioned in the FilmArray kit specifications because of design limitations. ‘False negative’ was any microorganism on the FilmArray repertoire that was identified by combination methods but was reported by FilmArray as any result other than the correct identification (missed/misidentified). Organisms detected by FilmArray while not identified by combination methods were excluded from the number of organisms detected by FilmArray in our calculations. This was due to the assumption that the combination methods was the gold standard method. The combination methods made use of trusted and proven phenotypic methods; and would be (in theory) superior to the FilmArray panel for organism detection. Calculations were done using IBM SPSS Statistics for Windows (Version 22.0; IBM Corp, Armonk, New York, United States).
Over the study period, 2119 blood cultures were received by the laboratory and 22.3% (472/2119) were positive. A total of 113 positive blood cultures were worked up both by combination methods and FilmArray kits.
Three blood culture bottles were reported as invalid by the FilmArray instrument. Four blood culture bottles were used as external controls on FilmArray by inoculation with known ATCC organisms. All four external controls yielded the desired identification. Five blood culture bottles were randomly repeated on FilmArray to evaluate reproducibility, and all five gave the same results each time they were run.
This permitted the actual evaluation of 101 positive blood culture bottles containing clinical isolates, comparing FilmArray to combination methods. Overall, 101 positive blood cultures were tested; 92.1% (93/101) with one organism and 7.9% (8/101) with more than one organism by the combination methods. All positive blood cultures included in the study had organisms detectable on initial Gram stain. In addition, all 101 positive bottles were identified down to the presence of one or more organisms by combination methods. In total, 109 organisms were detected by combination methods from 101 positive cultures. This is because some blood cultures had one organism, and other blood cultures had more than one organism detected. Of the organisms, 86.2% (94/109) were on the repertoire of the FilmArray panel for potential detection.
The performance of FilmArray on positive blood cultures is described (blood culture-wise) in
Performance characteristics of FilmArray compared to combination methods.
Parameter | All positive blood cultures ( |
Positive cultures with one type of organism (93/101) | Positive cultures with more than one type of organism (8/101) |
---|---|---|---|
Sensitivity | 92.6%, |
93.8%, |
85.7%, |
Specificity | 100%, |
100%, |
100%, |
Positive predictive value | 100%, |
100%, |
100%, |
Negative predictive value | 68.2%, |
72.2%, |
50%, |
Cohen’s Kappa | 0.774 | 0.807 | 0.6 |
Positive blood cultures described are organisms that were on the FilmArray BCID repertoire.
Performance of FilmArray on blood cultures with one type of organism – by microorganism and antibiotic resistance marker
Identification by combination methods | Number detected | FilmArray | ||
---|---|---|---|---|
Precisely identified |
Misidentified |
Missed |
||
Methicillin-resistant |
4 | 4 (100.0) | - | - |
Methicillin-resistant |
37 | 37 (100.0) | - | - |
Methicillin-sensitive |
7 | 7 (100.0) | - | - |
Methicillin-sensitive |
3 | 1 (33.3) | 2 (66.7) | - |
|
2 | 1 (50.0) | - | 1 (50.0) |
|
0 | - | - | - |
|
1 | 1 (100.0) | - | - |
|
0 | - | - | - |
|
4 | 3 (75.0) | - | 1 (25.0) |
|
0 | - | - | - |
|
3 | 2 (66.7) | - | 1 (33.3) |
|
1 | 1 (100.0) | - | - |
|
3 | 3 (100.0) | - | - |
|
8 | 8 (100.0) | - | - |
|
0 | 0 | - | - |
|
0 | 0 | - | - |
|
3 | 3 (100.0) | - | - |
|
2 | 2 (100.0) | - | - |
|
0 | 0 | - | - |
|
0 | 0 | - | - |
|
0 | 0 | - | - |
|
2 | 2 (100.0) | - | - |
|
0 | 0 | - | - |
|
0 | 0 | - | - |
|
0 | 0 | - | - |
|
0 | 0 | - | - |
Total yeasts on panel | 2 | 2 (100.0) | - | - |
|
41 | 41 | - | - |
|
1 | 1 | - | - |
KPC | 0 | 0 | - | - |
KPC,
Performance of FilmArray on blood cultures with more than one type of organism – by microorganism and antibiotic resistance marker.
Identification by combination methods | Number detected | FilmArray | ||
---|---|---|---|---|
Precisely identified | Misidentified | Missed | ||
Methicillin-resistant |
0 | - | - | - |
Methicillin-resistant |
3 | 3 (100.0) | - | - |
Methicillin-sensitive |
1 | 1 (100.0) | - | - |
Methicillin-sensitive |
0 | - | - | - |
|
3 | 2 (66.7) | 1 (33.3) | - |
|
1 | 1 (100.0) | - | - |
|
0 | - | - | - |
|
0 | - | - | - |
|
3 | 2 (66.7) | - | 1 (33.3) |
|
0 | - | - | - |
|
0 | - | - | - |
|
0 | - | - | - |
|
1 | 1 (100.0) | - | - |
|
0 | - | - | - |
|
0 | - | - | - |
|
1 | 1 (100.0) | - | - |
|
1 | 1 (100.0) | - | - |
|
0 | - | - | - |
|
0 | - | - | - |
|
0 | - | - | - |
|
0 | - | - | - |
|
0 | - | - | - |
|
0 | - | - | - |
|
0 | - | - | - |
|
0 | - | - | - |
|
0 | - | - | - |
Total yeasts on panel | 0 | - | - | - |
|
3 | 3 | - | - |
|
0 | 0 | - | - |
KPC | 0 | 0 | - | - |
KPC,
Breakdown of conflicting results.
Sample | Blood culture | Combination methods | FilmArray |
---|---|---|---|
1 | One type of organism | Missed | |
2 | One type of organism | Missed | |
3 | One type of organism | Methicillin-sensitive |
Methicillin-sensitive |
4 | One type of organism | Methicillin-sensitive |
Methicillin-sensitive |
5 | One type of organism | Missed | |
1 | More than one type of organism | Missed | |
2 | More than one type of organism |
, Sample was found to have methicillin-resistant
, Sample was found to have methicillin-resistant
Comparison of blood culture results using combination methods against FilmArray.
A minimum of 24 hours was needed by combination methods to obtain results from positive blood cultures. A maximum of 65 minutes was needed when using FilmArray for obtaining identification and potential susceptibility information of an organism from a positive blood culture. Once the kit was manually set up and loaded on the FilmArray instrument, the run, taking 55–60 minutes preceding report generation, was all automated, needing no operator intervention.
Organisms recovered that had resistance genes undetectable by FilmArray included: ESBL-producing
In certain blood cultures run, there was another form of discrepancy – where more organisms were detected by FilmArray than the combination methods. These included 14 isolates comprising these organisms:
Good correlation was seen between the FilmArray and combination methods for identification of organisms and resistance genes. An overall accurate identification rate of 92.6% was achieved for organisms identifiable by FilmArray from all blood cultures. This was slightly superior to the 91% sensitivity achieved overall in a separate study by Blaschke et al., which used a developmental version of the FilmArray,
Higher sensitivity of FilmArray was seen with positive cultures with one type of organism (93.8% [75/80] organisms detected) than with cultures with more than one organism (85.7% [12/14] detected) in our study. This corroborated with a higher proportion of organisms not accurately identified (either missed or misidentified) from cultures with more than one organism tested by FilmArray. This was in line with previous work done on the FilmArray BCID, which also found lower sensitivity for cultures that detected more than one organism.
Of concern were organisms that were on the repertoire of the FilmArray but not detected and still picked up by combination methods. If FilmArray were applied, these results would have been erroneously interpreted as negative. Overall, over 7% of organisms (7/94) were falsely classified as negative in our study, although it did include
During our investigation,
Antibiotic resistance marker detection by FilmArray was not an issue in our study. All
Minimal laboratory personnel training would be needed for incorporation of FilmArray into the routine daily workflow of the laboratory. The stand-alone component of FilmArray during its 55–60 minute run time is beneficial. If implemented, this would permit attention to other laboratory work pending the report on positive blood culture. The FilmArray instrument software interface also enabled quick troubleshooting and can be coupled to a laboratory information system for more efficient updates of results. In addition, when the FilmArray pouch is poorly rehydrated, the system immediately informs laboratory personnel so that an entire hour is not wasted. Finally, the 65 minutes required for running FilmArray on a positive blood culture was substantially shorter than the time required to run the combination methods (a minimum of 24 hours).
FilmArray covered only three types of genetic markers (
In certain blood cultures, detection of organisms by FilmArray was in excess of that of combination methods. These discrepant results were not included in calculations but may have been due to poor performance of combination methods or detection of nucleic acid from non-viable organisms by FilmArray in the blood culture.
Our sample size with regard to assessing performance characteristics of FilmArray was small at 101. Increasing the pool of positive blood cultures tested, with a more diverse range of isolates, would have been beneficial.
One positive blood culture can be run at a time on the FilmArray instrument. Based on the aforementioned rapid reporting times of this instrument, FilmArray would be appropriate to reduce turnaround time in a microbiology laboratory of a hospital. The unavoidable delays in testing more than one positive blood culture could be overcome by using at least two units of the FilmArray instrument. The identity of the organism, along with clinical information, would help clinical microbiologists and clinicians rule out whether a positive blood culture contains a contaminant or an actual pathogen, necessitating start or change of therapy.
Our assessment of the FilmArray panel on positive blood cultures demonstrated reasonable accuracy and practical benefits. Our results were in agreement with the reference method used in this study in the majority of positive blood cultures tested. Yet, for the time being, it should be used coupled to an existing identification and antibiotic susceptibility determination system. This, if utilised for rapid communication of results to the physician in the ward, would ameliorate the choice of initial antimicrobial to patients and also drastically affect patient management.
The authors are grateful to Dr Raveen Parboosing for his contribution to the statistical analysis.
The authors declare that they have no financial or personal relationship(s) that may have inappropriately influenced them in writing this article.
The BioFire FilmArray instrument and kits were provided by bioMérieux. This paper was supported by research project number BREC REF: BE456/14.
M.F. was the principal investigator, and was responsible for the running of tests and write-up of the analysis. F.N. was the statistician in charge of statistical data handling. K.P.M. was the supervisor, providing advice and assisting with the write-up of the analysis.