Are bacterial culture quantifications reliable? Comparative performance of the WASP automated inoculation instrument in the era of ISO 15189 accreditation.

Purpose. Isolating colonies and obtaining accurate colony counts from bacterial cultures are critical steps for the optimal management of infected patients. The uncertainties in the colony count results from the bacterial cultures were evaluated by verifying the performance of the WASP inoculation system according to the International Organization for Standardization (ISO) 15189 standard. Methodology. We first (i) evaluated the cross-contamination and precision of the WASP instrument (Copan Diagnostics, Italy) and (ii) established enumeration reading grids for urine, swab, bronchopulmonary specimens (BPSs) and catheter tip cultures. Subsequently, 72 clinical samples were tested to compare the results of the WASP, PREVI Isola (bioM erieux, France) and manual inoculation methods. Results. The WASP method did not show cross-contamination. The coefficient of variation for the colony counts in the repeatability experiment was evaluated for 10 μl and 30 μl loop protocols and determined to be 29 and 14%, respectively. The agreement between the automated and manual methods and between the automated methods for the colony counts was high (94.4 and 100%, respectively). The WASP method yielded better isolation quality compared to the manual method (P=0.020) and to the PREVI Isola only when polymicrobial specimens were considered (P=0.014). For quantification evaluation, the measurement uncertainty was evaluated to 1.8 10 1 for a suspension of Escherichia coli at 10 . Conclusion. We report the verification of the performance of the WASP instrument and describe a rapid procedure for achieving semi-quantitative cultures from BPSs and catheter tips. Quantitative interpretation of the bacterial cultures should be performed with caution.

Purpose. Isolating colonies and obtaining accurate colony counts from bacterial cultures are critical steps for the optimal management of infected patients. The uncertainties in the colony count results from the bacterial cultures were evaluated by verifying the performance of the WASP inoculation system according to the International Organization for Standardization (ISO) 15189 standard. Methodology. We first (i) evaluated the cross-contamination and precision of the WASP instrument (Copan Diagnostics, Italy) and (ii) established enumeration reading grids for urine, swab, bronchopulmonary specimens (BPSs) and catheter tip cultures. Subsequently, 72 clinical samples were tested to compare the results of the WASP, PREVI Isola (bioM erieux, France) and manual inoculation methods. Results. The WASP method did not show cross-contamination. The coefficient of variation for the colony counts in the repeatability experiment was evaluated for 10 µl and 30 µl loop protocols and determined to be 29 and 14%, respectively. The agreement between the automated and manual methods and between the automated methods for the colony counts was high (94.4 and 100%, respectively). The WASP method yielded better isolation quality compared to the manual method (P=0.020) and to the PREVI Isola only when polymicrobial specimens were considered (P=0.014). For quantification evaluation, the measurement uncertainty was evaluated to 1.8 103 1 for a suspension of Escherichia coli at 104 1. Conclusion. We report the verification of the performance of the WASP instrument and describe a rapid procedure for achieving semi-quantitative cultures from BPSs and catheter tips. Quantitative interpretation of the bacterial cultures should be performed with caution.

During the last decade, clinical microbiology laboratories (CMLs) have had to adapt to new quality requirements in the context of increasing economic constraints, leading to automation and concentration processes [1, 2]. Furthermore, the central role of CMLs in the management of patients suffering from infections in an era of multidrugresistant organisms and antibiotic shortages has been highlighted [3, 4]. The interpretation of cultures in clinical bacteriology is generally considered to be a qualitative result that requires the ability to recognize pathogenic bacteria in the commensal flora. However, the interpretation of urine, catheter tips and respiratory sample cultures needs semi-quantitative assessment to establish a diagnosis of infection [5, 6]. These quantifications depend on factors that are independent of the bacteriologist, such as bacterial strain metabolism or the presence of antibiotics in the sample. Technical management of the specimen is also Received 8 May 2018; Accepted 7 September 2018 Author affiliations: 1Unit e de Bact eriologie-Hygi ene, D epartement de Virologie, Bact eriologie-Hygi ene, Parasitologie-Mycologie, Unit e Transversale de Traitement des Infections, Hôpitaux Universitaires Henri Mondor, DHU ’Virus, Immunit e et Cancers’, Assistance Publique – Hôpitaux de Paris, F94000 Cr eteil, France; 2EA Dynamyc, Universit e Paris Est Cr eteil– Ecole V et erinaire de Maison Alfort, F-94000 Cr eteil, France; 3Laboratoire de Microbiologie-Hygi ene, Centre Hospitalier Intercommunal de Villeneuve-Saint-Georges, F-94195 Villeneuve-Saint-Georges, France; 4INSERM, UMR 1137, IAME, Facult e De M edecine, Site Xavier Bichat, Universit e Paris Diderot, Sorbonne Paris Cit e, Paris, France. *Correspondence: V. Fihman, Keywords: automation; growth & development; accreditation; ISO 15189; quantification; PREVI® Isola; WASP. Abbreviations: BHI, brain heart infusion; BPS, bronchopulmonary specimen; c.f.u., colony-forming unit; CML, clinical microbiology laboratory; CNA, colistin/nalidixic acid/sheep blood agar; COH, Columbia horse blood; CV%, coefficient of variation; ISO, International Organization for Standardization; SD, standard deviation; TS, trypticase soy. 000847 ã 2018 The Authors involved. For this reason, clinical diagnostic laboratories should focus on reducing the uncertainties of measurements using accurate, precise and reproducible methods. The International Organization for Standardization (ISO) 15189 : 2012 standard ( std:iso:15189:ed-3:v2:en) specifies particular quality and competence requirements to reach this goal [7]. Several studies have shown better isolation of bacteria from automated seeding machines than from manual inoculation [8–10]. Several automated systems are currently available: the InoqulA (Becton, Dickinson and Company, Kiestra Lab, Drachten, The Netherlands), the PreLUD (I2A, Montpellier, France), the PREVI Isola (formerly MicroStreak, bioM erieux, Marcy l’Étoile, France) and the WASP (Copan Diagnostics, Brescia, Italy) [11]. Direct comparisons of these instruments are scarce [12], and the existing studies have focused primarily on samples with readily automated seeding, such as urine and swabs, leaving out more complex clinical specimens of high clinical value, such as respiratory tract aspiration [8, 10, 13]. In this study, we measured the uncertainties in the colony count results from the bacterial cultures through the performance evaluation of the WASP inoculation system according to the ISO 15189 standard. We described rapid procedures to achieve semi-quantitative cultures of the catheter tips and bronchopulmonary specimen (BPS) cultures. Finally, we examined the agreement of the results between the WASP and PREVI Isola automated systems and the manual method used in case of failure.


Strains and bacterial suspensions
Five species were used for this study: four reference strains (Escherichia coli ATCC 25922, Pseudomonas aeruginosa ATCC 27853, Staphylococcus aureus ATCC 25923 and Streptococcus pneumoniae CIP 104485) and one clinical strain of Klebsiella pneumoniae that was isolated from a liver abscess in our laboratory and exhibited a hypermucoviscous phenotype [14]. Colonies of each bacterial species were obtained on Columbia agar with 5% horse blood (COH agar; bioM erieux) at 35 C in 5% CO2 incubators. Monomicrobial suspensions were then prepared in saline solution and adjusted to a McFarland turbidity of 0.5 (DensiCheck, bioM erieux). Serial 10-fold dilutions in saline solution were performed to obtain different concentrations ranging from 108 to 103 c.f.u. ml 1 and each concentration was controlled by plating 100 µl on COH agar as measured by a pipette. Polymicrobial suspensions were obtained by mixing three different monomicrobial suspensions at three different concentrations (Table 1).

Sample collection
A total of 72 consecutive specimens were obtained from patients hospitalized in the Henri-Mondor University hospital, Cr eteil, France, in April 2015. Forty-six urine samples were collected on V-Monovette with boric acid (Sarstedt AG & Co., Nümbrecht, Germany), 13 wound samples were collected with E-swab (Copan, Brescia, Italy) and 13 BPSs were included. The samples were received in the clinical bacteriology department within 2 h after sampling, as recommended [5]. All samples were inoculated using manual, WASP and PREVI Isola methods. Dedicated enumeration reading grids were used for the automated methods (see [15] and this study).

Media, inoculation and incubation
All bacterial suspensions and clinical samples were inoculated on media recommended by the European Society of Clinical Microbiology and Infectious Diseases and the Soci et e Française de Microbiologie [5]. (i) Urine samples were inoculated onto UriSelect 4 plates (Biorad, Marnes-laCoquette, France) and colistin/nalidixic acid/sheep blood agar (CNA; bioM erieux) when Gram-positive bacteria were suspected on direct Gram staining. Media were incubated for 24–36 h at 35 C in aerobic incubators. (ii) E-swab samples were inoculated onto trypticase soy (TS; Oxoid, Dardilly, France), Drigalski and CNA agar plates (bioM erieux) incubated in aerobic incubators and COH agar incubated in anaerobic conditions. All media were incubated at 35 C for 48 h. (iii) BPSs were manually liquefied using a volume-to-volume dilution with mucolytic SL solution (Copan) and then inoculated on the same media as used for the E-swab samples with the addition of PolyViteX-supplemented chocolate agar (bioM erieux). These media were incubated in 5% CO2 incubators. In the preliminary experiments, we identified the best of the four different inoculation protocols available from WASP to isolate all the species in the polymicrobial suspensions (data not shown). (i) For the urine samples, we used the single streak type 6 pattern with a 10 µl loop, which made a single central 2.5 cm long streak throughout the plate, followed by a zigzag pattern. (ii) For the E-swab samples, a 10 µl loop protocol inoculated the media by depletion with a fourquadrant type 4 pattern. (iii) For the BPS samples, a 30 µl loop protocol inoculated the media by depletion with the five-quadrant type 1 pattern. (iv) For the catheter tip culture, we adapted the semi-quantitative method routinely performed for many years in our laboratory, as follows [16]: the catheter tip was placed in 3ml saline solution and vortexed for 1min, and then the container was inoculated by the WASP instrument using the sensitivity streak 3 pattern with threefold 10 µl loop streaking on the same COH agar. The PREVI Isola automated system used a single 18 µl protocol as described previously [15]. Manual inoculations were performed by two skilled technicians using the same protocols as WASP for the urine and E-swab samples. For the BPS samples, quantitative cultures were made by a serial-dilution method [6]. A visual representation of the seeding patterns is presented in Fig. 1.

Performances analysis
In paragraphs and, ISO 15189 : 2012 defines the performance of an analytical procedure that must be verified to the extent possible. However, the detection and quantification limits, as well as the analytical sensitivity and specificity of the bacterial culture, depend mainly on the bacterium tested, the culture media used and the incubation conditions. These parameters, which may affect the performance of an automated inoculation system, fall outside the scope of this study. We also focused our efforts on measuring the precision (including repeatability and reproducibility) and technical accuracy of the isolations obtained. The precision is the closeness of the agreement between the measured quantity values obtained by repeated measurements on the same or similar objects under specified conditions. The precision only reflects the random error distribution and has no relationship to the true value. It is classically evaluated by the combination of repeatability and reproducibility studies. The repeatability study allows the intra-series variability to be determined under well-defined conditions (same operator, same equipment, same laboratory, same series, etc.). The intra-laboratory reproducibility consists in evaluating inter-series variability. The sources of variations are numerous (operator, reagent batches and environmental conditions). In addition, it was necessary to establish the comparability of the results obtained by the different seeding methods available in our laboratory, as required by paragraph 5.6.4 of ISO 15189 : 2012. We conducted the verification process as follows.

The analytical specificity was evaluated by inter-sample contamination, alternating rich samples and sterile samples. Two sets of four brain heart infusion (BHI) broths were inoculated with E. coli and K. pneumoniae strains, respectively. After overnight incubation at 35 C, we performed the following sequence of inoculation: two BHI broths of E. coli, a sterile broth, two BHI broths of K. pneumoniae and a sterile broth; this sequence was performed twice consecutively. A ‘urine protocol’ was performed to inoculate UriSelect 4 agar plates with a 10 µl loop. The number of bacterial cells in the broths was over 107 c.f.u. ml 1 for both species as determined by colony counting via the serial dilution method. Measurement precision of the WASP instrument The repeatability (intra-series experiment) was assessed by a 20-fold repeat of the same experiment within 1 h: a 103 c.f.u. ml 1 E. coli suspension was plated on UriSelect 4 with a 10 µl loop and colony counts were made after incubation at 35 C. The reproducibility (inter-series experiment) was evaluated with the following experiment: a 104 c.f.u. ml 1 E. coli suspension was prepared in Schaedler broth with 20% (v/v) glycerol. This concentration was chosen because of its clinical relevance as this threshold is used for the interpretation of cultures from clean catch voided urine and bronchoalveolar lavage samples. This suspension was frozen at 80 C. For 15 consecutive days, one sample of this suspension was thawed at 35 C for 30min and plated (i) on UriSelect 4 agar with the ‘urine protocol’ and (ii) on COH and Drigalski agar with the ‘catheter protocol’. Colonies were counted after incubation at 35 C for 24 h. A preliminary study on the repeatability of the bacterial colony count after manual seeding of the calibrated suspension at 103 c.f.u. ml 1E. coli suspension showed a coefficient of variation (CV%)=50% (data not shown). We defined a CV% <50% as an acceptability criterion for the automated method. The measurement uncertainty is the margin of doubt that exists for the result of any measurement. The width of the margin was considered to be at least equal to the square root of the sum of the reproducibility variances obtained with different inoculation patterns on different non-selective media (UriSelect 4 and COH agar). Development of enumeration reading grids for the WASP automated system Monomicrobial suspensions were used to establish a specific reading grid for each inoculation protocol: urine, E-swab, BPS and catheter tip samples, respectively. Agreement between inoculation methods We analysed the ability of different inoculation methods (i.e. the WASP, manual and PREVI Isola methods) to separate the different types of bacterial species from the clinical specimens. The results of the culture were considered to be concordant if the change in c.f.u. ml 1 between the grown micro-organisms was not different by more than one scale on the corresponding reading grid.

Quality of isolation
We evaluated the quality of the isolation using a standardized scoring system as described previously [8]. Briefly, the isolation quality was divided into four categories: very poor (QIS1), poor (QIS2), fair to good (QIS3) and very good to excellent (QIS4). This scale was based on the following characteristics: the number of isolated colonies, the need for subculture to continue the identification and the susceptibility testing, the ability to distinguish different types of colonies in polymicrobial samples and the distribution of colonies over the entire agar surface. Three independent investigators have been trained to use this scoring system. They agreed over a 3-day period by comparing their classification results on routine cultures (data not shown). Then, this four-category QIS was applied through the blinded reading of all the clinical specimen positive cultures included in the study.

Risk analysis
We identified the risks using a 5M model. Hazard classification and control were carried out using the generic method developed by the United States Department of Defense [17].

Statistical analysis
Statistical analyses were performed using GraphPad Prism 5.02 software (GraphPad Software, Inc., La Jolla, CA, USA). The 2 test or Fisher’s exact test were used for comparisons of the number of QIS scores obtained from the different inoculation methods. A P value <0.05 was considered statistically significant.


No contaminant was observed on either the sterile plates or those inoculated with E. coli or K.pneumoniae.

Measurement precision
The WASP instrument showed repeatable results. Using the 10 µl urine protocol, the expected number of colonies was 10. It varied between 7 and 28 (mean=15; SD=4, CV% =29%). The WASP instrument also showed reproducible results. With the 10 µl urine protocol, the expected number of colonies was 100. It varied between 46 and 97 (mean=74; SD=13, CV%=18%). With the 30 µl catheter protocol, the expected number of colonies was 300, and it varied between 146 and 300 (mean=220; SD=39; CV% =18%) on COH agar plates. On Drigalski plates, the number of colonies was between 90 and 235 (mean=168; SD=50; CV%=30%). During the reproducibility study, we did not show a regular decrease in the bacteria count that could have been attributed to preservation by freezing. Thus, we estimated the measurement uncertainty to be at least equal to 1.8 103 c.f.u. ml 1 for a suspension of E. coli at 104 1.

Enumeration reading grids
Fig. 1 shows the automated reading grids that we developed for the WASP inoculation system according to the inoculation protocols. There was no difference between the different monomicrobial suspensions for the quantification obtained (data not shown). The photographs selected were of E. coli and P. aeruginosa for urine samples and BPS, respectively, and two of S. aureus for swabs and catheter tips. The colony counts were the same for non-selective media, such as COH, TS, or chocolate agar; moreover, Drigalski agar showed 10- to 100-fold lower concentrations of E. coli and P. aeruginosa than TS agar. No differences were observed between the CNA and COH agar for the S. aureus and S. pneumoniae isolate colony counts. Agreement between inoculation methods Among the 72 clinical specimens included, 18 (25%) were sterile with the three inoculation methods. Four specimens showed discordant results: (i) three urine samples were found to have 102 to 103 c.f.u. ml 1 of P. aeruginosa, E. faecalis and mixed vaginal flora with automated methods and were sterile with manual inoculation. (ii) One protected distal sampling of BPS showed 3 103Haemophilus influenzae on manually streaked chocolate agar, although those automated methods were sterile. These four discordant results were not clinically significant because the positive cultures were interpreted as colonization and not infection by the patients’ physicians. We did not find significant differences in the colony counts between the three methods tested in either the analysis of the clinical specimens or the artificial bacterial suspensions. Finally, the agreement between the automated and manual methods was high [68 samples of the 72 tested, 94.4% (CI 95%: 89.2–99.7%)]. The agreement between the WASP and the PREVI Isola was 100%.

Quality of isolation
The agreement between the QIS assigned by the three investigators was 100%. Compared to the manual method, the WASP instrument resulted in a better QIS4 score (P=0.020) when the 54 clinical specimens with positive cultures were analysed (Table 2). No significant difference was shown between the WASP and PREVI Isola instruments. However, when only polymicrobial specimens were considered, the WASP instrument produced a significantly higher QIS4 score than the PREVI Isola (P=0.014) and manual (P=0.037) methods. The PREVI Isola and manual methods did not result in significantly different results (P=0.77). The WASP instrument produced QIS1 or QIS2 scores for 8 out of the 53 non-sterile clinical samples (15%). This occurred in five cases for samples with more than 106 c.f.u. ml 1 of Proteus mirabilis (four urine samples and a wound swab).

Risk analysis
A risk level of high, serious, medium or low was assigned as a combination of the severity and the probability of the occurrence of each risk (Table 3). These risk levels have defined priorities for risk decision-making when the WASP system is routinely implemented.

Automation constitutes one of the most promising enhancement solutions for clinical bacteriology laboratories; this solution should be implemented within the framework of the current mandatory certification procedure [1, 18]. We report here the first evaluation of the WASP automated plating instrument according to ISO 15189 quality standards. In this study, the WASP instrument did not produce contamination between successive samples. Unlike the previously reported studies, we deliberately included a hypermucoviscous K. pneumoniae strain that could have clung strongly to the loop [12, 14, 15, 19]. Despite this challenging strain, the loop sterilization system between consecutive samples seemed to be effective. Nevertheless, the inoculation system is not changed or sterilized between successive agar plates streaked for the same specimen and if one of the first agar plates had been contaminated prior to inoculation, the contamination would have been transmitted to subsequent agars. As previously reported [8–10, 13], we showed that the automated WASP system was more efficient than the manual method for optimal isolation quality, as estimated by the quality score from Froment et al. Compared to their results [8], we identified a lower percentage of QIS2 and QIS3 classes from the manual method (48 vs 81%). This could be at least partially explained by the lower proportion of polymicrobial specimens in our study (52 vs 77%). In addition, we analysed all the plates of a sample as a whole. This way of conducting the analysis of the cultures improves the QIS score calculated per sample. An alternative solution to improve the colony isolation from polymicrobial specimens is the use of selective media, such as Drigalski or CNA agar. Interestingly, we observed a two-log decrease in the number of colonies of targeted bacterial species when using the corresponding selective plates in comparison to non-selective agar plates. Thus, this last category of agar plates should be used to count the different types of colonies after their identification on selective media. We confirmed that the differences in the performance of different inoculation systems, especially the manual vs automated methods, were particularly noticeable for polymicrobial samples [20]. Thus, the WASP system appeared to be more efficient than the PREVI Isola system in obtaining enough colonies to perform identification and susceptibility testing as necessary. Otherwise, we noted that a significant load of P. mirabilis was detrimental to the quality of the isolation procedure. These results were consistent with recently published data supporting the view that the performance of the WASP is inferior to that of the comparable automated system (InoquIA) when the bacterial load reaches or exceeds 108 1, with the performance differing according to the bacterial species tested [12]. Nevertheless, one advantage of the WASP system is the ability to change the streaking pattern by adjusting the spacing or length of the striations. Thus, another recent study confirmed that two different streaking patterns available on the WASP system yielded significantly different results when the number of single colonies from the mixed cultures was considered [20]. Such modifications of the streaking pattern could have improved our results. A key recommendation of the ISO 15189 standard is to verify that the different methods implemented in a clinical laboratory for the same analysis produce similar results for the same sample. Our study was the first study to show an acceptable correlation between the culture results produced by two automated methods used routinely in our laboratory and a manual backup method. Further, our work has shown that the quantitative interpretation of bacterial cultures should be used with caution. Quantitative bacteriuria is a major criterion for the diagnosis of urinary tract infections [21–23]. Similarly, quantitative interpretation of BPS cultures is recommended for the diagnosis of ventilator-associated pneumonia [6, 24]. The evaluation of measurement uncertainty in colony counts has been discussed in other specialities, such as food microbiology (ISO/TS 19036 : 2006) or water quality (ISO 29201 : 2012), but, to the best of our knowledge, it has never been reported in clinical microbiology. We showed that the measurement uncertainty associated with an automated plating instrument was at least 20% for a bacterial suspension estimated at 104 1. This uncertainty in microbiological methods is usually structured into two components: (i) distribution uncertainty (also called intrinsic variability) resulting from the random distribution of micro-organisms in the sample and (ii) technical uncertainty (also called operational variability) resulting from uncertainties in the different technical steps of the analysis. Based on the within-laboratory reproducibility of the data, our experiences have evaluated the uncertainty related to the culture media, incubation and inoculation patterns. Thus, we showed that the precision of the bacterial colony counts did not depend on the loop used (CV%=18% for both 10 and 30 µl loops), while Drigalski selective medium should not be used (CV%=30%). However, in clinical bacteriology, the intrinsic variability is also affected by the presence of an antibiotic in the sample or the viscosity differences that may exist between a sample that is purulent and one that is not. In addition, the measurement uncertainty is probably greater for lower bacterial concentrations, as seemed to be suggested by our repeatability results with a bacterial suspension of 103 c.f.u. ml 1 and a 10 µl loop protocol (CV%=39%). Indeed, the CV% of reproducibility experiments is generally higher than that of repeatability experiments that do not consider the variability related to the environment and the reagent batches. We believe that our study noted that the evaluation of the number of bacteria in a sample must be considered as an estimated and not a quantitative variable along a continuous scale of values. A grey zone should be defined around the thresholds used for diagnosis of infections that would consider these measurement uncertainties. In daily clinical practice, susceptibility testing could be performed even if the bacterial count is found in this grey zone below the pathological threshold. During this study, we established new inoculation methods and corresponding reading grids that allow rapid sample management and interpretation of culture results from several common and clinically relevant types of specimens, such as BPSs, catheter tips, or wound samples (Fig. 1). Our methods for BPS will deliver a semi-quantitative evaluation with a single calibrated loop without using serial dilution, which constitutes a significant simplification [6]. Similarly, modification of the method described by Brun-Buisson et al. [16] will automate the culture of catheters, which is time-consuming using the manual method. The major findings of this study can be summarized as follows: (i) the implementation of the WASP inoculation system improved the quality of bacterial isolation compared to an older automated system, such as PREVI Isola, and a manual method; (ii) the performance verification was in accordance with the ISO 15189 rules and demonstrated that the measurement uncertainty of a calibrated suspension of E. coli at 104 c.f.u. ml 1 was at least 20%; and (iii) we described new inoculation methods and corresponding reading grids for BPSs and catheter tip cultures to simplify the reporting of results. Nevertheless, our work has several limitations. First, the number of clinical samples included in the present work was relatively small; as such, our findings should be confirmed by large-scale studies that include a higher number of clinical specimens from different origins. However, the diversity of the samples tested was representative of the daily activity of our laboratory. Second, we did not evaluate the time saved by using an automated inoculation system. Indeed, the manual inoculation of agar plates and broths represents 24% of the technician’s time [25]. Nevertheless, regular interventions from a technician were necessary for maintenance, reloading the culture medium in the system and discharging inoculated media. An accurate assessment of all the steps involved in the use of the device has yet to be Table 3. Risks analysis using the 5M model for inoculation by WASP system 5Mcategories Critical point Probability of occurrence* Severity† Risk category Elements to master Control measures Materials Type of samples accepted A IV Medium Various containers accepted Transform samples in a liquid format using flocked swabs with transport medium; use of plastic containers and fitted custom-built supports Sample pre-treatment A III Serious Fluidity and homogeneity of sample Manual liquefaction and crushing before loading on WASP system; automatic vortex of the sample by WASP system before inoculation Media plates – patient mismatch E I Medium Concordant identification of sample and inoculated media Connection of the WASP system to the laboratory information system; unambiguous labelling of agar plates Machine Precision of inoculation system D I Serious Defect and aging of calibrated loops Loop camera control attesting the correct volume of sample inoculated Inter-sample contamination E II Medium Cleanliness of instrument and loops Daily maintenance; incinerator to sterilize the loop after streaking each sample Suitable media for use D I Serious Traceability of analytical processing steps Lot number and expiration date of the media recorded on WASP instrument; network backup of data Methods Sample volume, streaking protocols and media inoculated A III Serious Perennial settings Protection by password Break-down of automatic system D II Medium Maintain the quality of the inoculation results Remote maintenance; local support engineers available at day+1; Laboratory technicians trained for manual inoculation Manpower Misuse of the WASP system (mistake when loading agar plate input silos or incubation atmosphere after inoculation) B III Serious Skill to operate the WASP system Formation and in-house training; implementation of smart incubators (WASPLab) Milieu Biosecurity E I Medium Contamination of the environment by infectious agents contained in sample Automatic decapping/recapping and HEPA filter to protect the atmosphere Environmental requirement D IV Low Soil strengthening; operating temperature between 5 and 40 C Pre-installation visit; air conditioning and temperature monitoring *Frequent, A; probable, B; occasional, C; remote, D; improbable, E. †Catastrophic (diagnostic false-negative, contamination within laboratory), I; critical (diagnostic false-positive, limited testing restriction), II; marginal (delay in release of urgent sample report, clinically relevant errors), III, negligible (delay in release of routine sample report, non-clinically relevant errors), IV. carried out. Third, we could have looked for the presence of antibiotics in the clinical samples examined by performing a culture inhibition test. This test was difficult to automate. Furthermore, it is not currently recommended to perform it systematically on all samples received by CMLs. Our results should be interpreted as reflecting those obtained in daily practice. Finally, our study supports the use of automated inoculators, such as the WASP system, and the cautious interpretation of quantitative cultures. Seeding systems coupled with smart incubators can reduce the time required to obtain organism identification and antibiotic susceptibility testing reports [26]. Prospective clinical trials are needed to evaluate the costs and benefits of these innovative technologies for patient care. Funding information This research did not receive any specific grant from funding agencies in the public, commercial or not-for-profit sectors Conflicts of interest V. F. and J.W.D. have consulted for Beckman Coulter, France. Ethical statement Because this study assessed a new laboratory method without returning results for clinical management, research ethics committee approval was not required in France. References 1. Ledeboer NA, Dallas SD. The automated clinical microbiology lab- oratory: fact or fantasy? J Clin Microbiol 2014;52:3140–3146. 2. Sautter RL, Thomson RB. Consolidated clinical microbiology laboratories. J Clin Microbiol 2015;53:1467–1472. 3. Buehler SS, Madison B, Snyder SR, Derzon JH, Cornish NE et al. Effectiveness of practices to increase timeliness of providing targeted therapy for inpatients with bloodstream infections: a laboratory medicine best practices systematic review and meta-analysis. Clin Microbiol Rev 2016;29:59–103. 4. Galar A, Leiva J, Espinosa M, Guill en-Grima F, Hern aez S et al. Clinical and economic evaluation of the impact of rapid microbiological diagnostic testing. J Infect 2012;65:302–309. 5. Cornaglia G, Courcol R, Herrmann JL, Kahlmeter G. European Manual of Clinical Microbiology Soci et e Française de Microbiologie (SFM) and European Society of Clinical Microbiology and Infectious Diseases (ESCMID); 2012. 6. Sharp SE, Robinson A, Saubolle M, Santa Cruz M, Carroll K et al. Lower Respiratory Tract Infections. Washington, DC: ASM Press; 2004. 7. Carey RB, Bhattacharyya S, Kehl SC, Matukas LM, Pentella MA et al. Implementing a Quality Management System in the Medical Microbiology Laboratory. Clin Microbiol Rev 2018;31. 8. Froment P, Marchandin H, vande Perre P, Lamy B. Automated versus manual sample inoculations in routine clinical microbiology: a performance evaluation of the fully automated InoqulA instrument. J Clin Microbiol 2014;52:796–802. 9. Glasson JH, Guthrie LH, Nielsen DJ, Bethell FA. Evaluation of an automated instrument for inoculating and spreading samples onto agar plates. J Clin Microbiol 2008;46:1281–1284. 10. Mischnik A, Mieth M, Busch CJ, Hofer S, Zimmermann S. First evaluation of automated specimen inoculation for wound swab samples by use of the Previ Isola system compared to manual inoculation in a routine laboratory: finding a cost-effective and accurate approach. J Clin Microbiol 2012;50:2732–2736. 11. Croxatto A, Prod’hom G, Faverjon F, Rochais Y, Greub G. Laboratory automation in clinical bacteriology: what system to choose? Clin Microbiol Infect 2016;22:217–235. 12. Croxatto A, Dijkstra K, Prod’hom G, Greub G. Comparison of inoculation with the InoqulA and WASP automated systems with manual inoculation. J Clin Microbiol 2015;53:2298–2307. 13. Bourbeau PP, Swartz BL. First evaluation of the WASP, a new automated microbiology plating instrument. J Clin Microbiol 2009; 47:1101–1106. 14. Decr e D, Verdet C, Emirian A, Le Gourrierec T, Petit JC et al. Emerging severe and fatal infections due to Klebsiella pneumoniae in two university hospitals in France. J Clin Microbiol 2011;49: 3012–3014. 15. Nebbad-Lechani B, Emirian A, Maillebuau F, Mahjoub N, Fihman V et al. New procedure to reduce the time and cost of broncho-pulmonary specimen management using the Previ Isola® automated inoculation system. J Microbiol Methods 2013;95: 384–388. 16. Brun-Buisson C, Abrouk F, Legrand P, Huet Y, Larabi S et al. Diagnosis of central venous catheter-related sepsis. Critical level of quantitative tip cultures. Arch Intern Med 1987;147:873–877. 17. Defense DoD. (editor). System Safety Standard Practice, MIL-STD882E ed; 2012. 18. Dauwalder O, Landrieve L, Laurent F, de Montclos M, Vandenesch F et al. Does bacteriology laboratory automation reduce time to results and increase quality management? Clin Microbiol Infect 2016;22:236–243. 19. Iversen J, Stendal G, Gerdes CM, Meyer CH, Andersen CØ et al. Comparative evaluation of inoculation of urine samples with the copan WASP and BD kiestra inoqula instruments. J Clin Microbiol 2016;54:328–332. 20. Quiblier C, Jetter M, Rominski M, Mouttet F, Böttger EC et al. Performance of copan WASP for routine urine microbiology. J Clin Microbiol 2016;54:585–592. 21. Grabe M, Bartoletti R, Bjerklund Johansen TE, Cai T, Çek M et al. Guidelines on Urological Infections. European Association of Urology; 2015. 22. McCarter YS, Burd EC, Hall GS, Zervos M, Sharp SE et al. Laboratory Diagnosis of Urinary Tract Infections. Washington, DC: ASM Press; 2009. 23. Stein R, Dogan HS, Hoebeke P, Ko cvara R, Nijman RJ et al. Urinary tract infections in children: EAU/ESPU guidelines. Eur Urol 2015;67:546–558. 24. Chastre J, Fagon J-Y. Ventilator-associated Pneumonia. Am J Respir Crit Care Med 2002;165:867–903. 25. Greub G, Prod’hom G. Automation in clinical bacteriology: what system to choose? Clin Microbiol Infect 2011;17:655–660. 26. Theparee T, Das S, Thomson RB. Total laboratory automation and MALDI-TOF improve turnaround times in the clinical microbiology laboratory: a retrospective analysis. J Clin Microbiol (in Press).

V Fihman, S C Bleunven, J M Le Glaunec, F Maillebuau, I De Rochebouet, B Nebbad-Lechani, M Desroches, J W Decousser
Journal of medical microbiology