Evaluation of Oxoid Denim Blue Agar for detecting Methicillin-Resistant Staphylococcus aureus from Surveillance Specimens

Evaluation of Oxoid Denim Blue Agar for detecting Methicillin-Resistant Staphylococcus aureus from Surveillance Specimens Study report compiled by: Barbara Willey and Nathan Kreiswirth Infection Control and Methods Development Department of Microbiology, Room 1460 Mount Sinai Hospital, 600 University Ave Toronto, Ontario, Canada M5G 1X5 bwilley@mtsinai.on.ca Study conducted by: Barbara Willey, Nathan Kreiswirth, Andrea Tyler, Vanessa Porter Duration of study: November-December, 2006 Report requested by: Brian Kemp, Oxoid, Nepean, Canada Report date: February, 2007

INDEX A. Purpose of Study 4 B. Materials and Methods 4 1. Study setting 4 2. Specimen collection 4 3. Number of specimens 4 4. Type of specimens 5 5. Quality assurance 5 6. Denim Blue numbering and inoculation order 5 7. Specimen and MRSASelect plate labeling 5 8. Plate inoculating and streaking 6 9. Incubation temperature conditions 6 10. Blinded plate examination 6 11. Plate interpretation 6 12. Length of incubation and reading time points 7 13. Denim Blue work-up 7 14. Bio-Rad MRSA Select work-up 8 15. Data documentation and collection 8 16. Resolution of discrepancies 9 17. Turn around time (TAT) 9 18. Data analyses and study results 10 19. Presentations date information and abstracts 11 - CACMID abstract - ASM abstract - Final abstract for posters 2

C. Summary tables Table 1. Patient, specimen and facility demographics 15 Table 2a. Summary of media performance positives 16 Table 2b. Summary of false negatives with PFGE 16 Table 2c. Summary of negative cultures 16 Table 3a. Summary of false positives by specimen type 17 Table 3b. Summary of false positives by TAT 17 Table 4. Summary of discrepancies 18 Table 5. Summary of contaminants 19 Table 6a. Overall sensitivities, specificities, PPV and NPV 20 Table 6b. Sensitivities, specificities, PPV and NPV- nasal swabs 20 Table 6c. Sensitivities, specificities, PPV and NPV- rectal swabs 20 Table 6d. Sensitivities, specificities, PPV and NPV NAGP swabs 21 Table 6e. Sensitivities, specificities, PPV and NPV wound swabs 21 Table 6f. Sensitivities with 95% confidence intervals 21 Table 7a. Results summary by Denim Blue lot number 22 Table 7b. Results summary by MRSASelect lot number 22 Table 8. Results summary by plate inoculation order 22 Table 9a. Comparison of time to positive culture - overall 23 Table 9b. Comparison of time to positive culture both media 23 Table 9c. Summary of time to MRSA notification 23 Table 9d. Summary of time to MRSA final report 23 3

A. Purpose of Study The objective of this blinded prospective study was to compare the Oxoid Denim Blue agar to the Bio-Rad MRSASelect agar to identify the most rapid yet cost effective, sensitive yet specific selective medium for detecting methicillin-resistant S. aureus (MRSA) from surveillance specimens. B. Materials and Methods 1. Study setting The study was conducted in the Mount Sinai Hospital Department of Microbiology, which serves 14 health care facilities in the Greater Toronto Region. All facilities perform screening of patients who are admitted with risk factors for MRSA. Some facilities also perform surveys of patients residing on medium to high risk units, and monitor patients known to have MRSA. As a result of these infection prevention and control programs, the clinical laboratory processes ~100,000 MRSA screens annually. 2. Specimen collection Specimens are obtained either by nurses or by infection control practitioners in accordance with recommended specimen collection guidelines in place in each facility. To reduce specimen numbers, certain facilities submit nasal swabs together with combined axilla/groin/perineum (NAGP) swabs that are intended to be processed as pooled specimens. These consist of a separate nasal swab taken simultaneously to a second swab that first samples the patient s axillae, then the groin surface areas and finally the perineum. These swabs are taped together, accessioned as an NAGP swab and processed together as a pooled specimen by inoculating the two specimens superimposed on the same spot on one plate. 3. Number of specimens The number of samples plated each day was predetermined according to predicted workloads and to accommodate staffing requirements. This resulted in 100 consecutive samples being enrolled each Sunday and 200 each Monday and Tuesday for a total of 500 samples a week. Since the number of MRSA-positive screens processed by the laboratory from all specimen types in the months leading up to the study was ~4%, it was calculated that a minimum of 2,500 specimens would be needed to obtain >100 MRSA. However, at 2,500 specimens only 93 MRSA had been identified, therefore a further 950 swabs were enrolled into the study for a total of 3,450 specimens. 4

4. Type of specimens The 3,450 samples processed consisted of all types of MRSA screening specimens that were received by the laboratory during the study. These included 955 nasals, 925 rectal, 261 wound and 1,309 NAGP swabs. 5. Quality assurance During the study, Bio-Rad supplied 6 lots of MRSASelect and Oxoid supplied 5 lots of Denim Blue. On arrival, the condition of each batch was inspected by the laboratory quality assurance bench technologist and representative samples from each lot were processed as per laboratory protocol. This protocol is as per CLSI standards for sterility, selectivity, specificity and growth rates using appropriate ATCC control organisms. All batches of media used in the study passed these performance challenges before they were utilized in the clinical laboratory. These QC results were documented directly into the clinical laboratory information system (LIS) and are available for review if necessary. The lot numbers for both media were documented in the Access database against each particular specimen number. 6. Denim Blue numbering and rotation of inoculation order To prevent inoculation bias, the media order was rotated every 50 specimens throughout the study so that neither medium was inoculated first more frequently than the other. In practice, plates were organized into their respective orders for the following day in a room with reduced light. Each Denim Blue plate was marked with a consecutive Access auto-number (i.e. 1 to 3,450). The media was then arranged according to the predefined rotation. The numbered/organized plates were replaced into cardboard media boxes to protect them from light. The following morning they were removed from the continuously temperature monitored walk-in media storage refrigerator and allowed to reach room temperature before planting. Plates were taken out of the boxes only when specimen inoculation was imminent. 7. Specimen and MRSASelect plate labeling On receipt in the laboratory, specimens were accessioned into the LIS by technicians as per routine practice. Multiple labels for each specimen were printed during this procedure: one was affixed to the specimen, one to the MRSASelect plate, and one was placed onto a pre-printed worksheet against the Access auto-number that had been assigned to that specimen. The autonumber on the Denim Blue plate was always checked to match that on the worksheet with the LIS label. 5

8. Plate inoculating and streaking Using the frosted marking as a guide, an area the size of a loonie was inoculated by applying the swab to the surface and rotating it so that all sides made contact with the medium. Once inoculated, plates were streaked on a Fisher Scientific Iso-Plater-80. Once streaking was completed, the plates were sorted by media type in preparation for incubation in separate bins. 9. Incubation temperature conditions Inoculated plates were placed into the laboratories monitored 37 o C walk-in incubator continuously throughout the day. Both media types from each specimen were incubated simultaneously. The sorted plates were placed 10 high into separate large plastic bins that were covered with light-weight breathable sheeting. The bins were appropriately labeled to indicate the media type and the incubation time ranges as follows: First bin: 8am 11am Second bin: 11am - 3pm Third bin: 3pm - 6pm Incubation typically occurred within 1h of the laboratory accessioning the specimen. Planting of the required daily number of study specimens was usually completed by 6pm each day (with only a few exceptions). 10. Blinded plate examination The Denim Blue media was examined by a dedicated study technologist while the MRSASelect were worked up by technologists assigned to the MRSA bench as per normal practice. Both routine and study technologists were specifically instructed to refrain from discussing their results in any specific manner. 11. Plate interpretation For the MRSASelect, dusty pink colonies were taken as potential positives in accordance with the manufacturers instructions. However, through experience, hazes and pale pink colonies were typically ignored. For the Denim Blue, denim blue colonies were taken as positives, while hazes and pinpoint dark blue colonies were ignored. 6

12. Length of incubation and reading time points At ~11am after overnight incubation, all bins (both study and routine) were removed from the incubator simultaneously to sort positives from negatives. The examination of plates in the 8-11am bin was treated as their final 24h read. The MRSASelect and Denim Blue plates from these bins that had no growth of the appropriate colour for MRSA were discarded as negative and the corresponding specimens were resulted as No MRSA ; potential positive plates were worked up as per routine/study protocol. The 10:30am read of plates incubated in the 11am-3pm and 3pm- 6pm bins was treated as a preliminary read and recorded in the database under the 18h checkbox. At this initial read, any potentially positive plate was removed from its respective bin and worked up with plates that had completed their 24h incubation. All negative plates from these bins were re-incubated without delay. The 11am-3pm bin was removed for a second and final read at 3pm, and likewise the plates in the 3-6pm bin were read finally at 6pm. The duration of incubation to the first read varied according to when the plates were incubated and as a result the time recorded as 18h and 24h is not precise. A more accurate description of incubation times is as follows: 1. Plates in the 8am-11am bin were read only once at 11am after incubation for 24-27h. 2. Plates in the 11am-3pm bin were read at 11am (20-24h) and again at 3pm after incubation for 24-28h. 3. Plates in the 3-6pm bin were read at 11am (17-20h) and again at 6pm after incubation for 24-27h. 13. Denim Blue work-up After incubation, the presence of denim blue colonies was documented on the worksheet in the 18h or 24h boxes, as previously mentioned. Bio-Rad Pastorex Staph Plus (SS in the database) agglutination was performed directly from blue colonies when possible; results were recorded on the worksheet. If colonies were S. aureus, i.e. Pastorex-positive, the presence of PBP2a was determined using the Denka Seiken kit (Denka in database). If the Pastorex and PBP2a were positive, the specimen was considered an MRSA and this date was recorded as the notification date. If there were insufficient pure colonies on the Denim Blue plate, the Pastorex and/or the PBP2a were performed the next day from a 5% sheep blood agar (BA) sub. Tube coagulase and oxacillin screens were performed directly from the media if sufficient colonies were present. If there were insufficient pure colonies, 7

these tests were performed the next day from the BA sub. If discrepancies between the tube coagulase and Pastorex occurred, the identity of the organism was resolved using the GenProbe Accuprobe S. aureus speciesspecific rdna test. Once all tests were positive the specimen was reported as MRSA, the date was entered onto the worksheet as the result date. 14. MRSASelect work-up The MRSASelect plates were examined at ~18h and 24h for pink colonies as per lab protocol by the technologist assigned to the infection control bench. These results were recorded in the LIS. The presence of pink colonies was also documented on the lab study worksheet. All the work-up related to these specimens with pink colonies (whether MRSA or not) was later extracted from the LIS, as previously described under documentation. A positive PBP2a result in the LIS was used as a proxy for MRSA and this date was documented by the study technologist as the notification date. Similarly, growth on the oxacillin screen was taken as the date of confirmed MRSA and this date was documented as the final result date in the Access database. The number of BA subcultures that were performed from MRSASelect was documented in the database as Y under the BA variable. 15. Data documentation and collection In the study database, the Denim Blue is referred to as DB and the MRSASelect is referred to as BR. The study technologist documented all the daily work results for the Denim Blue medium onto printed worksheets generated from the database that were numbered consecutively according to the Access auto-numbers. On completion of each days work, Denim Blue results were transcribed from these worksheets into a query in the Access database that matched the specific format of the worksheet. For the MRSASelect, in addition to direct documentation into the LIS, any pink colony was documented on the worksheet next to the corresponding LIS label by marking either the 18h or 24h check box as appropriate. A separate research assistant was assigned the daily task of extracting real-time LIS information according to the checks on the MRSASelect worksheet. For each MRSA, the assistant determined from the LIS whether the patient had previously had MRSA: new patients were designated N and those with previous positives were designated P in the Access database. As well, this assistant correlated and collated data extracted from weekly LIS dumps containing demographic and specimen type information. Only the Study coordinator had access to the entire database and review of all results was conducted on a regular basis. 8

16. Resolution of discrepancies Discrepancies were investigated as soon as the results from each medium had been entered into the study database. If MRSA was detected from one medium only, the organisms identity was reconfirmed from the original plate by performing Pastorex and PBP2a agglutinations. Once the isolate had been re-confirmed to be MRSA, the original swab was retrieved from storage and re-planted directly to both Denim Blue and MRSASelect plates. The swab was then placed in BHI broth overnight at 37 o C for enrichment of MRSA, and the following morning it was subbed to both selective media. All replants were worked up as per protocol. If the MRSA that had originally grown from only one medium was recovered from the swab it was considered a true positive, so long as the PFGE DNA fingerprint confirmed this strain to be the same as the fingerprint for the strain known to be associated with that patient. If the fingerprint was different, or if after broth enrichment, the repeat culture was negative for MRSA, the original isolate was considered to be a lab contaminant and was treated as a negative culture (No MRSA) in the study database for evaluation purposes. For all discrepant results, the LIS results from previous and subsequent specimens from the same patient were reviewed by the Study coordinator to determine patient MRSA history. From those with a history of MRSA, the original isolate was pulled from the freezer and SmaI PFGE typing was performed alongside the discrepant isolate from the study. In those cases without patient history where the broth enrichment had produced an MRSA, the original study isolate was also typed by PFGE alongside the isolate retrieved from the broth. In either case, if the PFGE were different, the isolates were determined to be laboratory contaminants. 17. Turn around time (TAT) As described previously, the proxy for TAT to notification for both study arms was taken as the date when the test isolate was determined to be Pastorex and PBP2a-positive. The TAT for notification for each specimen was influenced by the number, size and purity of colonies available for testing directly from the selective media. Similarly, the proxy for TAT to final result was taken as the date when the tube coagulase tests and oxcacillin screens were completed. Because of the staggered incubation times, a comparison of time to positive growth could only be compared for those specimens that grew MRSA from both media. The rationale was that this subset of swabs would have been incubated simultaneously and 18h versus 24h MRSA would be significant. 9

18. Data analyses and study results At present, since no gold standard for detecting MRSA from contaminated specimens exists in the conventional sense, a relative gold standard was applied for the purposes of data analysis. In practice, this meant that if a confirmed MRSA isolate was detected by one or both media that specimen was considered to be a true positive. In other words, the relative gold standard used in this study was the total MRSA identified from both media combined. It should be noted that since there were only two media types in this study, the relative gold standard was not particularly stringent. The use of this type of gold standard results in a slightly falsely elevated sensitivity for both study media. See Table 1 for summary of patient demographics, including the gender balance, the number of patients in the study, and the number of patients with MRSA. Also, find in this table, the mean number and range of specimens per patient, the total number of swabs received from each of the 14 facilities served by the lab, and the number and percent MRSA positive specimens received from each facility. See Table 2a c for a summary of media performance by specimen type. This table also includes data that indicates whether each MRSA was detected from a new or previously identified case. Table 3a summarizes the identities of false positive isolates worked-up from each medium by specimen type, while Table 3b shows this data by TAT. Table 4 contains the patient history and details of DNA fingerprinting for the discrepant isolates, while Table 5 provides evidence for justifying the reclassification of many of the discrepant isolates as laboratory contaminants. The summaries of percent sensitivities, specificities, positive and negative predictive values for the study overall and the per specimen type can be found in Tables 6a-e, while Table 6f summarizes the 95% confidence intervals (CI) for this data overall and by specimen type. Tables 7a and 7b contains summaries of study results by Denim Blue and MRSASelect lot numbers. No difference in performance was noted between batches for either medium. Table 8 demonstrates that no bias was evident when the results were analyzed by inoculated first or inoculated second. Table 9a shows the overall time to positive growth for each medium for the 120 MRSA identified in the study. This data is compromised as the incubation times are not strictly 18h or 24h, and more importantly, because the 24h 10

read includes plates that were read only once (i.e. plates in the 8am-11am bin) and those that were negative at their initial 18h read but positive at their 24h read. The difference between the growth rates of the two media, therefore, is better compared by looking only at those specimens that grew MRSA from both media, as is seen in Table 9b. While these positives still include a combination of 24h plates that were read once and read twice, at least these 113 matched specimens have equivalent incubation times, as we know these plates were placed into the incubator at the same time. So, if one type of media was found to detect more MRSA at 18h than the other, its significance can be measured. The data pertaining to TAT to MRSA notification and TAT to MRSA final reporting is displayed in Tables 9c and 9d, respectively. These data importantly demonstrate in an objective manner that the colonies on Denim Blue MUST be larger earlier than those on MRSASelect since a significantly larger proportion of MRSA could be notified within 24h of planting. Same day notification offers a clear indication that a sufficient volume of colony was available to perform the PBP2a assay directly from the selective medium. If colonies are small, direct PBP2a is not always possible. This difference reaches significance (P <0.0001) suggesting that by using the Denim Blue agar in place of the MRSASelect agar, there could be a resulting fiscal saving for each institution since a significantly larger number of positive results would reach Infection Control/the wards a full day earlier. For newly identified cases of MRSA, this should in turn result in fewer instances of nosocomial transmission, specifically if patients are not suspected to be high-risk. In previously identified cases where the patient would (should) already be in isolation, the impact would be less. So, to calculate the impact of this laboratory improvement in TAT, only the new cases should be taken into consideration. From a laboratory perspective, there could be a moderate fiscal reduction since a significant proportion of the positive isolates would not require subculture to blood agar prior to work-up. Practically, this would save a full day to final reporting, and by effectively reducing the length of time specimens remain on the work-list, labour costs should be reduced. 19. Presentations The preliminary data submitted to AMMI-CACMID 2007 was accepted as a poster for presentation on Friday March 16, 2007 in the Port Royal C and D, World Trade and Convention Centre, Halifax - poster P51. The analyses from the completed study was submitted to ASM Toronto 2007 where it has been accepted as a poster presentation to be held at the Toronto Convention Centre on 22 May 2007 at 9am poster number C-027. The abstract presenting the final cleaned-up analyses will appear as a revised abstract on the poster which will be presented at both conferences. See the three abstracts on the following pages. 11

The AMMI-CACMID abstract is as follows. Preliminary Evaluation of the Oxoid MRSA Denim Blue Agar for the Selective Detection of Methicillin-Resistant Staphylococcus aureus (MRSA): A Prospective and Blinded Comparison to the Bio-Rad MRSASelect Agar N. Kreiswirth, A. Tyler, M. Skulnick, S.M. Poutanen, E. Cudek, N. Nelson, A. McGeer, D.E. Low, B.M. Willey Mount Sinai Hospital, University Health Network, University of Toronto, Toronto, Ontario, Canada Objectives: The rapid, accurate and cost effective detection of MRSA from surveillance specimens is integral to laboratory workflow and to patient management. The availability of selective S. aureus-specific chromogenic agars has vastly improved the sensitivity, specificity and turn-around-time associated with MRSA detection. Previous studies found the MRSASelect (MS) to have a sensitivity of 91% and specificity of 98% using a relative gold standard of four selective media. This study aimed to compare the Denim Blue (DB) agar to the MS that is currently in routine use. Methods: Consecutive specimens received for MRSA screening were set up in parallel on DB and MS agars. The order of plating was rotated every 50 specimens throughout the study to prevent inoculation bias. After incubation in the dark at 37 o C, the DB and MS were examined by different technologists at 18 and 24h for blue or pink colonies, respectively. Standard MRSA confirmatory methods (Bio-Rad Pastorex Staph Plus, Tube coagulase, Denka Seiken PBP2a agglutination, CLSI oxacillin screen) were performed directly from each agar when possible. Discrepancies were resolved by broth enrichment of the original swab and replanting to both media. Results: A total of 2,000 surveillance swabs (525 rectal, 549 nasal, 176 wound, 750 pooled nasal/axilla/groin/perineum swabs) from 1,053 patients in 12 hospitals (range 1-8 swabs/patient) were included in the study. Of these, 71 (3.6%) grew MRSA from at least one medium, with 69 (97.1% of MRSA; 3.4% of specimens) being isolated from DB and 66 (93% of MRSA; 3.3% of specimens) being isolated from MS. 14 (0.7%) specimens planted to DB and 7 (0.4%) planted to MS grew blue and pink colonies, respectively, that proved not to be MRSA. Resulting sensitivities/specificities for DB were 97.2/99.3 and for MS were 93.4/99.6 percent, respectively. The positive/negative predictive values for DB were 83.5/99.9 and for MS were 91/99.7 percent, respectively. Preliminary data regarding the discrepant, found 3/4 did not grow after enrichment suggesting laboratory cross contamination, whilst the other grew MRSA on both agars. Discussion: This preliminary data suggests that the DB and MS have very similar performance characteristics. 12

The abstract as submitted to ASM is as follows: Evaluation of Oxoid s Denim Blue Agar for the Detection of Methicillin-Resistant Staphylococcus aureus (MRSA) from Surveillance Specimens N. Kreiswirth, A. Tyler, V. Porter, M. Skulnick, S.M. Poutanen, N. Nelson, A. McGeer, D.E. Low, B.M. Willey Mount Sinai Hospital, University Health Network, University of Toronto, Toronto, Ontario, Canada Background: The use of chromogens for assisting detection of MRSA has markedly improved the ability of laboratories to manage large numbers of screening specimens in a rapid and labor-efficient manner. First to be successfully launched in Canada, Bio-Rad s MRSASelect (BR), quickly became the standard for selective culture due to its high sensitivity and specificity. This blinded, prospective study compared Oxoid s Denim Blue (DB) to BR. Methods: Consecutive MRSA screens from 12 Toronto hospitals were planted in parallel to DB and BR. The plating order was rotated every 50 swabs to avoid inoculation bias. Incubation was at 37C in the dark for 24h with MRSA growing as blue (DB) or pink (BR) colonies. Hazes/pinpoints were ignored on both media as previous studies proved these were not MRSA. MRSA were confirmed (Bio- Rad Pastorex Staph Plus, tube coagulase (coag), Oxoid PBP2a agglutination, CLSI oxacillin screen) directly from each agar when possible; discrepancies were resolved by replanting original swabs to both media post broth enrichment. Results: Overall, 125 MRSA were identified from 3,450 (3.6%) swabs obtained from 1,952 patients. The DB grew 118 MRSA from 21 of 261 wound, 33 of 925 rectal, 26 of 955 nasal and 38 of 1,309 pooled nasal/axilla/groin/perineum (NAGP) swabs, while the BR grew 120 MRSA from 21 of 261 wound, 34 of 925 rectal, 27 of 955 nasal and 38 of 1,309 NAGP swabs. The 19 DB false positives (13 NAGP, 6 rectal) included 15 coag-neg staphylococci (CNS), 3 enterococci (ENT) and 1 diphtheroid (DIP). The 15 BR false positives (9 NAGP, 4 rectal, 2 wound) included 9 CNS, 3 ENT, 1 DIP, 1 coliform and 1 CNS plus DIP. The overall sensitivities, specificities, positive and negative predictive values for DB were 94.4%, 99.4%, 86.1% and 99.8% versus 96%, 99.5%, 88.9% and 99.8% for BR. Conclusions: There was no significant difference between the sensitivities (P=0.9) or specificities (P=0.6) of DB and BR demonstrating that these media may be used interchangeably. The positive predictive values of both media were <90% which highlights the need to fully confirm all MRSA from potential new cases regardless of the high specificities associated with both agars. 13

The revised abstract that will appear on the poster is as follows: Evaluation of Oxoid s Denim Blue Agar for the Detection of Methicillin-Resistant Staphylococcus aureus (MRSA) from Surveillance Specimens N. Kreiswirth, A. Tyler, V. Porter, M. Skulnick, S.M. Poutanen, E. Cudek, N. Nelson, A. McGeer, D.E. Low, B.M. Willey Mount Sinai Hospital, University Health Network, University of Toronto, Toronto, Ontario, Canada Background: The use of chromogens for assisting detection of MRSA has markedly improved the ability of laboratories to manage large numbers of screening specimens in a rapid and labor-efficient manner. First to be successfully launched in Canada, Bio-Rad s MRSASelect (BR), quickly became the standard for selective culture due to its high sensitivity and specificity. This blinded, prospective study compared Oxoid s Denim Blue (DB) to BR. Methods: Consecutive MRSA screens from 12 Toronto hospitals were planted in parallel to DB and BR. The plating order was rotated every 50 swabs to avoid inoculation bias. Incubation was at 37C in the dark for 24h with MRSA growing as blue (DB) or pink (BR) colonies. Hazes/pinpoints were ignored on both media as previous studies proved these were not MRSA. MRSA were confirmed (Bio- Rad Pastorex Staph Plus, tube coagulase (coag), Oxoid PBP2a agglutination, CLSI oxacillin screen) directly from each agar when possible; discrepancies were resolved by replanting original swabs to both media post broth enrichment. Results: Overall, 120 MRSA were identified from 3,450 (3.5%) swabs obtained from 1,952 patients. The DB grew 116 MRSA from 21 of 261 wound, 32 of 925 rectal, 26 of 955 nasal and 37 of 1,309 pooled nasal/axilla/groin/perineum (NAGP) swabs, while the BR grew 117 MRSA from 20 of 261 wound, 33 of 925 rectal, 27 of 955 nasal and 37 of 1,309 NAGP swabs. The 17 DB false positives (13 NAGP, 4 rectal) included 15 coag-neg staphylococci (CNS), 1 enterococci (ENT) and 1 G+B. The 16 BR false positives (9 NAGP, 4 rectal, 3 wound) included 12 CNS, 2 ENT, 1 G+B, 1 coliform. Overall sensitivities, specificities, positive and negative predictive values for DB were 96.7%, 99.5%, 87.2% and 99.8% and for BR were 97.5%, 99.5%, 87.9% and 99.8%, respectively. Conclusions: There was no significant difference between the sensitivities (P=1.0) or specificities (P=0.86) of DB and BR demonstrating that these media may be used interchangeably. The positive predictive values of both media were <90% which highlights the need to fully confirm all MRSA from potential new cases regardless of the high specificities associated with both agars. 14

C. Summary of results Table 1. Patient, specimen and facility demographics Gender No. of swabs Female 1,670 Male 1,780 Total no. of patients screened No. of patients with MRSA No. of patients without MRSA 1,952 79 1,870 Total no. of swabs enrolled in the study No. of swabs with MRSA No. of swabs without MRSA 3,450 120 3,330 No. of swabs per patient* Mean no. per patient Range *no duplicates from same day/same site 2 swabs 1 to 9 swabs Facilities submitting swabs to MSH No. of swabs per facility No. (%) MRSA per facility A 93 1 (1.1) B 159 13 (8.2) C 296 6 (2) D 17 0 (0) E 76 1 (1.3) F 61 1 (1.6) G 18 0 (0) H 1,056 29 (2.8) I 112 3 (2.7) J 52 6 (11.5) K 93 1 (1.1) L 202 30 (14.9) M 580 12 (2.1) N 635 17 (2.7) 14 facilities 3,450 swabs 120 (3.5) MRSA 15

Table 2a. Summary of media performance by specimen type - positives Swab type No. No. of MRSA detected on: Overall MRSA MRSASelect Denim Blue Both No. % Previous MRSA New MRSA Nasal 955 27 26 25 28 2.9 16 12 Rectal 925 33 32 32 33 3.6 27 6 Wound 261 20 21 20 21 8.1 18 3 NAGP* 1,309 37 37 36 38 2.9 20 18 Totals 3,450 117 116 113 120 3.5 81 40 *Combined nasal and axilla-groin-perineum swabs Table 2b. Summary of false negatives showing SmaI pulsed field types Swab type No. of MRSA Nasal 28 Rectal 33 Wound 21 NAGP 38 All types 120 Medium No. of MRSA No. of MRSA PFGE-types of detected missed missed MRSA MRSASelect 27 1 CMRSA-2 Denim Blue 26 2 CMRSA-1 CMRSA-2 MRSASelect 33 0 / Denim Blue 32 1 CMRSA-9 MRSASelect 20 1 CMRSA-2 Denim Blue 21 0 / MRSASelect 37 1 CMRSA-8 Denim Blue 37 1 CMRSA-2 MRSASelect 117 3 2 CMRSA-2 1 CMRSA-8 1 CMRSA-1 Denim Blue 116 4 2 CMRSA-2 1 CMRSA-9 Table 2c. Summary of negative cultures No MRSA detected Swab type No. BOTH media - No MRSA MRSASelect - No MRSA Denim Blue - No MRSA Nasal 955 927 928 929 Rectal 925 892 892 893 Wound 261 240 241 240 NAGP 1,309 1,271 1,272 1,272 All types 3,450 3,330 3,333 3,334 16

Table 3a. Summary of false positives by specimen and media type Swab Identity of organisms other than MRSA growing as pink or blue Medium No. type colonies based on Gram stain, catalase or Pastorex testing Nasal MRSASelect 0 No false positives Denim Blue 0 Rectal MRSASelect 4 2 enterococci, 1 coagulase-negative staphylococci, 1 coliform Denim Blue 4 2 coagulase-negative staphylococci, 1 enterococci, 1 diphtheroids Wound MRSASelect 3 3 coagulase-negative staphylococci Denim Blue 0 No false positives 7 coagulase-negative staphylococci, 1 diphtheriods, MRSASelect 9 NAGP 1 coagulase-negative staphylococci and diphtheriods Denim Blue 13 13 coagulase-negative staphylococci All 11 coagulase-negative staphylococci, 2 enterococci, 1 coliform MRSASelect 16 swab 1 diphtheriods, 1 coagulase-negative staphylococci and diphtheriods types Denim Blue 17 15 coagulase-negative staphylococci, 1 enterococci, 1 diphtheroids Table 3b. Summary of false positives by TAT Identity of organisms other than Growth on MRSASelect at: Growth on Denim Blue at: MRSA growing as pink or blue 18h 24h Total 18h 24h Total colonies False Positives 7 9 16 1 16 17 Coagulase-negative staphylococci 5 6 11 1 14 15 Enterococcus spp. 1 1 2 0 2 2 Gram negative bacilli - coliform 1 0 1 0 0 0 Diphthroids 0 1 1 0 0 0 Coagulase-negative staphylococci and diphthroids 0 1 1 0 0 0 Nasal+Axilla-Groin-Perineum 1 8 9 1 12 13 Rectal 3 1 4 0 4 4 Wound 3 0 3 0 0 0 Nasal 0 0 0 0 0 0 17

Table 4. Summary of the 7 confirmed discrepancies: 4 failed on Denim Blue and 3 failed on MRSASelect Study no. Swab type MRSASelect Denim Blue Replant result Comments Patient History 357 NAGP CMRSA-2 NO MRSA CMRSA-2 Patient had CMRSA-2 5 days previous 1503 Rectal CMRSA-9 NO MRSA CMRSA-9 Patient had CMRSA-9 on same day + 1 month previous 2069 Nasal CMRSA-1 NO MRSA CMRSA-1 Patient had CMRSA-1 5 days previous 2227 Nasal CMRSA-2 NO MRSA CMRSA-2 Patient had CMRSA-2 on same day + 3 month previous 898 NAGP NO MRSA CMRSA-8 CMRSA-8 Patient had CMRSA-8 before + after specimen 1579 Nasal NO MRSA CMRSA-2 CMRSA-2 Original Denim Blue grew 1 colony only Patient had CMRSA-2 before + after specimen 3407 Wound NO MRSA CMRSA-2 CMRSA-2 Original Denim Blue grew 3 colonies only Patient had CMRSA-2 on same day + 7 days previous 18

Table 5. Summary of the 15 discrepancies deemed laboratory contaminants: 6 on MRSASelect, 9 on Denim Blue Stu no. Swab type 1062 Wound 1108 NAGP 1705 Rectal 2232 Wound 2981 Rectal 3056 Rectal MRSASelect MRSA PF not done CMRSA-2 Mupirocin-S MRSA PF not done CMRSA-2r1 Mupirocin-R CMRSA-2r3 Mupirocin-R MRSA PF not done Denim Blue No MRSA NO MRSA No MRSA NO MRSA NO MRSA No MRSA Replant result No MRSA CMRSA-2 Mupirocin-S No MRSA CMRSA-2r1 Mupirocin-R CMRSA-2r4 Mupirocin-S No MRSA Patient History Patient had CMRSA-2 before and after this specimen (not frozen) Patient had CMRSA-2r2 mupirocin-r strain 20 days previous Patient had MRSA before and after this specimen (not frozen) Patient had CMRSA-2 mupirocin-s strain 5 days previous Patient has NO history of MRSA Patient had CMRSA-11 1 month before study but had multiple negative screens in between Comments Only enterococci grew on replants which were repeated twice Only enterococci grew on replants which were repeated twice MRSA grown from original MRSASelect was unrelated to MRSA from replant Original MRSASelect grew 1 colony only Not frozen as plate had been discarded by the time the discrepancy was identified 512 NAGP No MRSA CMRSA-2 No MRSA Patient has NO MRSA history Original Denim Blue grew 1 colony only 701 Wound No MRSA CMRSA-2 No MRSA Patient had CMRSA-2 3 months before study screens but multiple negative in Original Denim Blue grew 1 colony only between 1529 NAGP NO MRSA CMRSA-2r1 No MRSA Patient had multiple negative screens taken over 3 days 2224 Rectal No MRSA CMRSA-2r1 No MRSA Patient had CMRSA-2 3 months before Only coagulase-negative staphylococci but no screens in between grew on replants that were repeated twice 2297 Rectal NO MRSA CMRSA-2 CMRSA-2 Patient had CMRSA-1 both 1 month prior + 6 days after MRSA from Denim Blue and subsequent broth enrichment different to patient strain 2314 Rectal No MRSA CMRSA-2 No MRSA Patient had CMRSA-2 2 weeks before Only coagulase-negative staphylococci with one negative screen in between grew on replants that were repeated twice 2344 NAGP No MRSA CMRSA-2 No MRSA Patient has NO history of MRSA 2694 NAGP No MRSA CMRSA-2v No MRSA Patient has NO history of MRSA Original Denim Blue grew one colony only 2721 Nasal No MRSA CMRSA-2-II No MRSA Patient has NO history of MRSA Original Denim Blue grew one colony only * Contamination is believed to occur during planting resulting from an increased number of specimens positive for MRSA and an increased pressure to plant greater numbers of specimens in shorter time periods. Glove fingertips and the interior biological safety cabinet environment have been shown to become contaminated when swabs are removed from their transport medium in haste. Measures introduced to reduce contamination include unsheathing of swabs using protective gauze soaked in 70% ethanol, more frequent glove changing, more frequent disinfection of safety cabinet surfaces, and reduced specimen planting speed. These combined interventions have not been entirely effective and intense monitoring is underway as a result of this study. 19

Table 6a. Overall percent sensitivities, specificities, positive and negative predictive values by media type for the 3,450 swabs Medium No. Sensitivities Medium No. Specificities MRSASelect TP 117 MRSASelect TN 3317 97.5%* TP+FN 120 TN+FP 3333 Denim Blue TP 116 Denim Blue TN 3317 96.7%* TP+FN 120 TN+FP 3334 99.5% 99.5% Medium No. PPV Medium No. NPV MRSASelect TP 117 MRSASelect TN 3317 87.9% ** TP+FP 133 TN+FN 3322 99.8% Denim Blue TP 116 Denim Blue TN 3317 87.2% ** 99.8% TP+FP 133 TN+FN 3322 *Sensitivities may be falsely elevated due to the lack of a stringent gold standard **Low positive predictive values (PPV) highlights the ethical responsibility that is incumbent on the manufacturers when marketing these media both products must be correctly presented as selective media to be used in conjunction with confirmatory MRSA testing, especially in newly identified cases. Table 6b. Sensitivities, specificities, positive and negative predictive values by media type for the 955 nasal swabs Medium No. Sensitivities Medium No. Specificities MRSASelect TP 27 96.4% MRSASelect TN 928 TP+FN 28 TN+FP 928 Denim Blue TP 26 Denim Blue TN 929 92.9% TP+FN 28 TN+FP 929 100% 100% Medium No. PPV Medium No. NPV MRSASelect TP 27 MRSASelect TN 928 100% 99.9% TP+FP 27 TN+FN 929 Denim Blue TP 26 Denim Blue TN 929 100% TP+FP 26 TN+FN 931 Table 6c. Sensitivities, specificities, positive and negative predictive values by media type for the 925 rectal swabs 99.8% Medium No. Sensitivities Medium No. Specificities MRSASelect TP 33 MRSASelect TN 888 97.1% TP+FN 34 TN+FP 892 Denim Blue TP 32 94.1% Denim Blue TN 889 TP+FN 34 TN+FP 893 99.6% 99.6% Medium No. PPV Medium No. NPV MRSASelect TP 33 MRSASelect TN 888 89.2% 99.9% TP+FP 37 TN+FN 889 Denim Blue TP 32 Denim Blue TN 889 88.9% TP+FP 36 TN+FN 891 99.8% 20

Table 6d. Sensitivities, specificities, positive and negative predictive values by media type for the 1,309 nasal-axilla-groin-perineum swabs Medium No. Sensitivities Medium No. Specificities MRSASelect TP 37 MRSASelect TN 1263 94.9% TP+FN 39 TN+FP 1272 99.3% Denim Blue TP 37 Denim Blue TN 1259 97.4% TP+FN 38 TN+FP 1272 99% Medium No. PPV Medium No. NPV MRSASelect TP 37 MRSASelect TN 1263 80.4% * 99.8% TP+FP 46 TN+FN 1265 Denim Blue TP 37 Denim Blue TN 1259 74% * TP+FP 50 TN+FN 1260 *Pooling of specimens may create more work, as indicated by low PPV Table 6e. Sensitivities, specificities, positive and negative predictive values by media type for the 261 wound swabs 99.9% Medium No. Sensitivities Medium No. Specificities MRSASelect TP 20 MRSASelect TN 238 95.2% TP+FN 21 TN+FP 241 98.8% Denim Blue TP 21 Denim Blue TN 240 100% TP+FN 21 TN+FP 240 100% Medium No. PPV Medium No. NPV MRSASelect TP 20 MRSASelect TN 238 86.9% 99.6% TP+FP 23 TN+FN 239 Denim Blue TP 21 Denim Blue TN 240 100% TP+FP 21 TN+FN 240 100% Table 6f. Media sensitivities with 95% confidence intervals by swab type MRSASelect Denim Blue Swab type Sensitivities 95% CI Sensitivities 95% CI Overall 97.5% 92.6 99.5 96.7% 91.5 99 Nasal 96.4% 80.8 - >99.9 92.9% 76.3-99.1 Rectal 97.1% 87.6-1 94.1% 83.4 - >99.9 NAGP 94.9% 82.2 99.5 97.4% 85.3 - >99.9 Wound 95.2% 75.6 - >99.9 100% 81.8-1 21

Table 7a. Results summary by Denim Blue lot number Denim Blue 6300 6305 6310 6319 6321 Totals MRSA 16 18 16 13 53 116 False positives 12 1 0 1 3 17 NG or No blue 472 481 484 486 1394 3317 False negatives/discrepancies 1 0 0 1 2 4 Contaminants 0 2 0 1 6 9 No MRSA 484 482 484 487 1397 3334 No. of plates 500 500 500 500 1450 3450 Table 7b. Results summary by MRSASelect lot number MRSASelect 6J2177 6J2179 6K2183 6K2186 6L2191 6L2192 Totals MRSA 17 17 16 37 15 15 117 False positives 2 0 1 8 0 5 16 NG or No pink 481 483 481 955 485 430 3,317 False negatives/discrepancies 0 1 0 1 0 1 3 Contaminants 0 0 2 2 1 1 6 No MRSA 483 483 484 963 485 435 3,334 No. of plates 500 500 500 1,000 500 450 3,450 Table 8. Results summary by plate inoculation order Denim Blue plated first MRSASelect plated second MRSASelect plated first Denim Blue plated second MRSA 68 69 48 48 False positives 9 8 8 8 NG or No Blue/Pink 1,673 1,673 1,644 1,644 False negatives/discrepancies 2 1 2 2 Contaminants 7 3 3 2 No MRSA 1,682 1,682 1652 1652 Totals 1,750 1,750 1,700 1,700 22

Table 9a. Comparison of time to positive culture overall Overall MRSA growth rate MRSASelect Denim Blue P value Positive at 18h 81 84 NS Positive at 24h* 36 32 NS No. of the total 120 MRSA detected 117 116 *Positive at 24h includes plates read at 24h only and those read twice see text for incubation times Table 9b. Comparison of time to positive culture for the 113 swabs where MRSA was detected on both media Comparison of TAT for the 113 MRSA that grew on both Denim Blue and MRSASelect *These include both those read only at 24h and those read twice **This is suggestive of larger colony sizes associated with Denim Blue No. of MRSA P value No. of MRSA that grew on both media at 18h 71 / No. of MRSA that grew on both media at 24h* 22 / No. of MRSA that grew on Denim Blue at 18h but MRSASelect at 24h 12 0.48** No. of MRSA that grew on MRSASelect at 18h but Denim Blue at 24h 8 Table 9c. Summary of time to MRSA notification Days to MRSA notification MRSASelect Denim Blue P value 1 day 50 86 <0.0001* 2 days 67 30 <0.0001* No. of the total 120 MRSA detected 117 116 *Objective demonstration of improved colony sizes associated with Denim Blue Table 9d. Summary of time to MRSA final report Days to MRSA final confirmation MRSASelect Denim Blue P value 2 days 48 86 <0.0001* 3 days 69 30 <0.0001* No. of the total 120 MRSA detected 117 116 *Objective demonstration of improved colony sizes associated with Denim Blue 23