Special Articles Journal of General and Family Medicine 2015, vol. 16, no. 3, p. 138 142. Clinical Usefulness of Multi-facility Microbiology Laboratory Database Analysis by WHONET Sachiko Satake, PhD, 1 Singaku Kaneko, 2 Ikuyo Yokozawa, 2 Mayumi Shibasaki, 3 Osamichi Mita, 4 Mayumi Takase, 5 Sakiko Miyashita, 6 Yoshiro Fujita, MD, 7 John Stelling, MD, MPH, 8 and Yasuharu Tokuda, MD, MPH 9 1 School of Health Sciences, Gunma University, Japan 2 Maebashi Red Cross Hospital, Japan 3 JCHO Gunma Central Hospital, Japan 4 Kiryu Kosei General Hospital, Japan 5 Gunmaken Saiseikai Maebashi Hospital, Japan 6 Fujioka General Hospital, Japan 7 Chubu Rosai Hospital, Japan 8 WHO Collaborating Centre for Surveillance of Antimicrobial Resistance, Brigham and Women s Hospital, USA 9 Japan Community Healthcare Organization, Tokyo, Japan Clinically important issues might be addressed more comprehensively by analyzing microbiological data assembled from several institutions instead of from a single one. Microbiology laboratory databases collected from six collaborative hospitals (A F) between 2003 and 2013 were converted into WHONET files and analyzed. Three important areas of investigation are presented in this article including early warning of the emergence of resistant bacteria, evaluation of a new susceptibility testing method, and establishment of strategies for identifying patients harboring resistant organisms. Combined databases from microbiology laboratories of a collaborative hospital network might help in developing integrated responses to prevent infection, decrease transmission, and ensure that appropriate antibiotics are selected for patients with infectious diseases. Keywords: antimicrobial susceptibility, antibiogram, multidrug-resistant Pseudomonas aeruginosa, methicillinresistant Staphylococcus aureus, minocycline, infection control Corresponding author: Sachiko Satake, PhD Gunma University, 3-39-22 Showa-machi, Maebashi 371-8511, Japan E-Mail: satake@gunma-u.ac.jp Received for publication 30 October 2014 and accepted in revised form 18 February 2015 2015 Japan Primary Care Association 138
Clinical Usefulness of Multi-facility Microbiology Laboratory Database Analysis by WHONET Table 1. Antimicrobial resistance rates of Pseudomonas aeruginosa isolated during 2005 Antibiotics Total Piperacillin 5.7 5.8 5.6 11.4 3.5 6.1 ND Ceftazidime 7.1 7.6 6.3 12.5 4.2 3.0 17.0 Aztreonam 10.5 8.2 17.5 13.6 7.7 3.6 18.0 Imipenem 12.6 12.8 5.1 20.5 13.7 10.8 22.0 Gentamicin 7.4 5.2 3.6 8.0 5.8 11.5 22.0 Amikacin 4.9 2.9 1.6 4.6 4.2 7.9 17.0 Levofloxacin 11.1 11.7 11.6 9.1 8.9 9.0 20.0 Total patients (n) 1301 343 289 89 313 167 100 ND, not done. Background The antimicrobial susceptibility rate and the distribution of resistant bacteria are important for infectious disease clinics and infection control in supporting and evaluating therapy guidelines and containment interventions. 1,2 Useful strategies for the analysis and presentation of antimicrobial susceptibility test results are available in CLSI documents, 3 for example addressing the issue of repeat isolations of a microbial strain from a given patient through the analysis of first isolates of a given species from a patient. For monitoring incidence of infection, then analyzing one phenotype strain per patient may be more appropriate. Analyses of multicenter data might permit the data analysis to explore issues that cannot be addressed through separate analyses of data from individual institutions. Analyses of multicenter data might also clarify systematic problems in laboratory testing methods or in database analysis. The software WHONET 5.6 is a database for microbiology laboratory management that was developed and provided by the World Health Organization (WHO) Collaborating Centre for Surveillance of Antimicrobial Resistance, Boston. Six laboratory databases with information collected between 2003 and 2013 were converted to the WHONET file structure, and analyzed at Gunma University. Antimicrobial susceptibility was tested using the Vitek system (SYSMEX biomérieux) at the laboratory of Hospital A, and the MicroScan WalkAway (Siemens) at those of hospitals B, C, D, E and F. The laboratory of Hospital A is located in Aichi Prefecture (400 km west of Tokyo) and the other five are located in Gunma Prefecture (100 km north of Tokyo). The analytical findings of the individual databases are described with comments on each laboratory. In this study, we explore a number of ways to analyze data which would have benefits to clinicians, policy-makers, and infection control staff in the management of patients with resistant infections and in the tracking and containing of resistant microbes. Spread of Multidrug-Resistant Pseudomonas Aeruginosa The first isolate of Pseudomonas aeruginosa per patient in 2005 was analyzed. The rates of antibiotic resistance to ceftazidime (CAZ), aztreonam (AZT), imipenem (IPM) gentamicin (GM), amikacin (AMK), levofloxacin (LVFX) were statistically significantly different in hospital F compared with the other laboratories (Table 1). Resistance profiles were analyzed using WHONET to detect multidrug-resistant P. aeruginosa (MDRP). The rate of MDRP resistance to IPM, AMK, LVFX, was 1.8% in five laboratories and 13.0% in hospital F. However, the rate of MDRP in hospital F decreased to that of the other hospitals in 2007 (Table 2). Deviation of Antimicrobial Susceptibility Test Results between Two Machines The annual resistance rates to minocycline (MINO) of methicillin-resistant Staphylococcus aureus (MRSA) 139
Journal of General and Family Medicine 2015, vol. 16, no. 3 Table 2. Multidrug-resistant Pseudomonas aeruginosa (MDRP) isolated between 2003 and 2007 Ratio (%) of MDRP Number of patients Year Hospital F Other hospitals Hospital F Other hospitals MDRP Non-MDRP MDRP Non-MDRP 2003 11.2 2.3 12 95 32 1,351 2004 13.4 1.9 11 71 16 820 2005 13.0 1.8 13 87 20 1,101 2006 3.5 0.7 3 82 7 1,038 2007 1.0 0.8 1 103 7 886 Table 3. Minocycline resistance rates of methicillin-resistant Staphylococcus aureus isolated between 2003 and 2013 Years Total 2003 0.1 ND 0.0 0.3 0.0 0.0 0.0 2004 0.4 ND 0.0 1.4 0.5 0.0 0.0 2005 7.0 26.8 0.6 0.0 0.0 0.0 0.0 2006 8.1 30.5 0.0 1.0 0.6 0.0 0.0 2007 8.8 41.4 0.0 0.0 0.8 0.0 0.7 2008 11.6 41.9 1.5 1.4 7.2 0.0 3.6 2009 42.4 47.6 41.5 33.8 49.4 ND 30.7 2010 53.1 50.0 52.9 46.4 55.7 64.1 52.7 2011 49.4 43.3 47.8 41.9 53.8 65.3 50.0 2012 42.6 38.3 44.7 31.5 45.0 56.9 44.8 2013 30.9 27.3 39.5 27.1 27.7 42.1 34.7 ND, not done. isolated from all six hospitals between 2003 and 2013 were analyzed using WHONET. Analyses were based on the first isolate with antibiotic findings per patient per year. The resistance rates of MINO between 2005 and 2007 at hospitals B, C, D, E and F ranged from 0.0% to 1.0%, whereas the rates at hospital A between 2005 and 2007 ranged from 26.8% to 41.4%. The resistance rates of MINO did not significantly differ among the six hospitals by 2010 (Table 3). Analytical Method That Does Not Miss Emergence of Resistant Bacteria Clinical microbiology laboratory data from six hospitals in 2009 were analyzed. When several strains with the same species were repeatedly isolated from the same patient, we compared results obtained when analyzing 1) the first isolate per patient per 365 days versus 2) the most resistant finding for each antibiotic per patient. The rates of P. aeruginosa resistance to imipenem in the per-patient analysis using the first isolate did not differ significantly among the six hospitals. However, when the imipenem resistance rates of P. aeruginosa were analyzed using the most resistant test results, the rates at hospitals D and E were higher than those in the others. The rates of S. aureus resistance to oxacillin were higher for isolates from hospital F than at the others when the data were analyzed per patient with the first isolate or per patient with the most resistant result. The oxacillin resistance rates of isolates from hospitals B, D, E were higher than others when analyzed per strain (Table 4). The average numbers of MRSA isolates per patient at 140
Clinical Usefulness of Multi-facility Microbiology Laboratory Database Analysis by WHONET Table 4. Analysis of resistance rates during 2009 Analytical method Imipenem resistance rate of Pseudomonas aeruginosa By patient, first isolate with antibiotic results 9 8 9 11 17 13 By patient, most resistant result against each antibiotic 10 12 12 20 25 14 By isolate 14 16 14 25 27 21 Oxacillin resistance rate of Staphylococcus aureus By patient, first isolate with antibiotic results 36 35 30 35 37 51 By patient, most resistant result against each antibiotic 39 38 33 40 39 55 By isolate 46 53 37 52 58 53 Table 5. Average numbers of MRSA isolates per patient Number of MRSA isolates (A) 365 564 235 985 180 224 MRSA-positive patients number (B) 189 234 145 326 64 144 Ratio (A/B) 1.9 2.4 1.6 2.9 2.8 1.5 hospitals B, D and E were 2.4, 2.9 and 2.8, respectively (Table 5). Discussion The higher rate of MDRP in hospital F compared in early 2006 with the other hospitals in 2005 was communicated to the hospital infection control and link nurse committees. This information improved staff awareness, and the infection control team evaluated the management of urethral catheter, and UTI surveillance was initiated in June 2006. By 2007, these aggressive interventions decreased the rate of MDRP. The resistance rates to MINO of MRSA considerably differed among the hospitals in 2005, but the cause could not initially be defined. Three years later, Siemens informed their clients that the results of MINO resistance were misreported by the MicroScan Pos series because the minimum inhibitory concentrations (MIC) were two- to four-fold lower than that of the standardized method of Clinical and Laboratory Standards Institute (CLSI). 4 Subsequently, Siemens reported that good MIC agreement could be obtained using the standard method and modification by an improved panel. The improved panel was adopted at all five hospitals between November 2008 and March 2009 and this immediately changed the rate of resistance to MINO. This finding led to an error being identified by a recently introduced method of testing minocycline susceptibility. Thus, the cause of the significantly different MINO resistance rate of MRSA among hospitals was established and the MINO database of S. aureus accumulated using the Micro- Scan Pos series before 2008 should be used as epidemiological information with care. To guide clinical decisions regarding empirical therapy for initial infections, antibiograms that comprise only the first isolate of a given species per patient should be analyzed according to the CLSI. 3 However, basing summaries only on the first isolate per patient might result in changes related to the emergence of new resistance patterns being overlooked. Analysis of the complete database is recommended to detect such resistant strains. When strains of the same species have been isolated repeatedly from the same patient, WHONET can be analyzed using the highest resistance value per antibiotic per patient. For example, an IPM- 141
Journal of General and Family Medicine 2015, vol. 16, no. 3 resistant strain emerged during IPM therapy for P. aeruginosa infection, and a first isolate approach omits this isolate from consideration. Analysis using the highest resistance result per antibiotic per patient can facilitate monitoring the spread of resistant bacteria among hospitalized patients. Therefore, this is considered to be a useful analytical method from the viewpoint of infection control. The analysis of combined data showed that knowing the number of patients harboring resistant organisms was more valuable than simply counting the number of isolates of resistant organisms to understand the epidemiological spread of resistant bacteria in patients at a single hospital. Acknowledgments We thank the medical technologists at each hospital laboratory who were responsible for sending the microbiological data to us for 11 years. Research reported in this publication was supported by the National Institute of General Medical Sciences of the National Institutes of Health under award number R01GM103525. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Conflict of interest None. Conclusion Providing feedback to a hospital that the rate of multiple-drug-resistant bacteria was statistically higher than that of other hospitals triggered an appropriate intervention to control infection. An analysis of multicenter data confirmed that the MINO resistance rate of MRSA before 2009 measured using MicroScan should be used carefully as epidemiological information. Analysis using the highest resistance value per antibiotic per patient is a useful appropriate method for infection control purposes as it is not affected by the number of isolates per patient, and it can improve detection of rare resistance phenotypes and identification of novel phenotypes among hospitalized patients. References 1 Trevino S: Antibiotic resistance monitoring: a laboratory perspective. Mil Med. 2000;165:40 42. 2 Murray PR, Baron EJ, Landry ML, Jorgensen JH, Pfaller MA: Manual of clinical microbiology, 9th ed. ASM press, 2007, 41 42. 3 Clinical and Laboratory Standards Institute: Analysis and presentation of cumulative antimicrobial susceptibility test data; Approved Guideline, 3rd ed. CLSI document M39-A3. Wayne: Clinical and Laboratory Standards Institute, 2009, 1 49. 4 Clinical and Laboratory Standards Institute: Methods for dilution antimicrobial susceptibility tests for bacteria that grow aerobically; Approved standard, 7th ed. CLSI document M7-A7. Wayne: Clinical and Laboratory Standards Institute, 2006, 1 49. 142