Fisher, Mark

ABSTRACT Clinical and Laboratory Standards Institute (CLSI) recommends a blood culture contamination (BCC) threshold of <3%, with ≤1% considered optimal. However, there is not a standardized definition of BCC, and the effect of multiple definitions on BCC rates or what definitions laboratories use remain unknown. We surveyed 52 hospitals and analyzed 362,078 blood cultures (BCx) collected 1 September 2019 to 31 August 2021 from 62 intensive care units (ICUs) and 231 wards from 48 of these hospitals. We calculated and compared BCC rates using the College of American Pathologists (CAP) or CLSI criteria (both utilize a limited number of skin commensals to define BCC) and the comprehensive National Healthcare Safety Network (NHSN) commensal list. We characterized factors associated with BCC and related outcomes (central-line associated bloodstream infection [CLABSI] and vancomycin use). BCC, BCx positivity, and single BCx rates were monitored by 100%, 39%, and 21% of hospitals, respectively. Hospitals used CAP (65%), CLSI (17%), and NHSN (17%) criteria to define BCC. Mean BCC rate by CAP (CAP-BCC) was 1.38% for ICUs and 0.96% for wards. BCC rates remained similar by CLSI criteria but increased when using NHSN list. Sharing BCC data outside of the laboratory, measuring additional BCx quality indicators, and limiting central catheter-drawn BCx were associated with lower BCC rates. BCC was associated with higher CLABSI rates in ICUs. This study demonstrated variability in laboratory practices and opportunities to optimize BCx stewardship. IMPORTANCE Blood culture contamination (BCC) is associated with patient harm and unnecessary use of healthcare resources. BCC thresholds have been established; however, multiple BCC definitions exist. There is limited data on how BCC rates differ depending on the BCC definition used, what definitions laboratories most commonly use, or their approach to other blood cultures (BCx) quality indicators such as single rates or BCx positivity. A cross-sectional multicenter survey and analysis of BCx data from intensive care unit and wards revealed that most laboratories did not track single BCx or BCx positivity rates and that there was variability in how BCC was defined. Additionally, BCC rates were influenced by the definition used. BCC was associated with increased central-line associated bloodstream infection rates.

ABSTRACT The purpose of this study was to establish tentative Clinical and Laboratory Standards Institute (CLSI) M45 MIC and disk diffusion (DD) breakpoints (BPs) for Pseudomonas other than Pseudomonas aeruginosa (POPA). Mechanisms of antimicrobial resistance (AMR) and the modified carbapenem inactivation method (mCIM) to detect carbapenemase production were also evaluated. MIC data from P. aeruginosa and POPA from 2013 to 2022 were evaluated to compare the MIC distributions and modal MICs relative to the CLSI M100 P. aeruginosa BPs. A disk-to-MIC correlation study with 83 isolates was completed by testing reference broth microdilution and DD from the same inoculum, and the error-rate bounded method was used to establish DD BPs. For most antimicrobials, the modal MICs between P. aeruginosa and POPA were within 1-doubling dilution and lower than the M100 P. aeruginosa- susceptible BP. For amikacin, the modal MIC for POPA was 2-doubling dilutions lower than P. aeruginosa and was evaluated relative to the Enterobacterales BP. For aztreonam and trimethoprim-sulfamethoxazole, the modal MICs were elevated, and no BPs were set. New DD correlates were established for most antimicrobial agents, except for fluoroquinolones, where the P. aeruginosa correlates were suitable. AMR genes conferring resistance to multiple antimicrobial classes were identified by WGS. Beta-lactamase genes were identified in 30 (36.1%) isolates, with metallo-beta-lactamases (90.6%) predominating. The mCIM had a sensitivity and specificity of 100%. Upon review, the CLSI M45 committee proposed tentative MIC and DD BPs for expanded-spectrum cephalosporins (ceftazidime and cefepime), carbapenems (meropenem and imipenem), fluoroquinolones (ciprofloxacin and levofloxacin), and the aminoglycosides (amikacin and tobramycin). IMPORTANCE Pseudomonas species other than Pseudomonas aeruginosa (POPA) can cause opportunistic infections which may be difficult to treat due to a variety of antimicrobial resistance mechanisms. Antimicrobial susceptibility testing is a critical component of patient management for these infections. Currently, the Clinical and Laboratory Standards Institute (CLSI) M100 non-Enterobacterales breakpoints and methodology are utilized for POPA by US clinical laboratories and likely do not accurately predict susceptibility results. The purpose of this study was to establish tentative CLSI M45 MIC and disk diffusion breakpoints for POPA. Mechanisms of antimicrobial resistance and the modified carbapenem inactivation method to detect carbapenemase production were also evaluated. We present the data used by the volunteers tasked by CLSI to develop POPA breakpoints in the M45 guidelines.

ImportanceBlood culture (BC) use benchmarks in US hospitals have not been defined.ObjectiveTo characterize BC use in adult intensive care units (ICUs) and wards in US hospitals.Design, Setting, and ParticipantsA retrospective cross-sectional study of BC use in adult medical ICUs, medical-surgical ICUs, medical wards, and medical-surgical wards from acute care hospitals from the 4 US geographic regions was conducted. Critical access hospitals, less than 6 months of BC data, and non-US hospitals were excluded. The study included BC use data from September 1, 2019, to August 31, 2021. Data were analyzed from February 23 to July 14, 2024.Main Outcomes and MeasuresThe primary outcome was BC use per 1000 patient-days. Adjusted means with 95% CIs were calculated using mixed-effects negative binomial regression models adjusted for unit type, hospital bed size, geographic region, seasonality, and state COVID-19 case load, with random intercepts accounting for clustering at unit and hospital levels. Secondary outcomes included blood culture positivity, single BCs, BC contamination, and minimum threshold for BC use where blood culture positivity would be optimized.ResultsA total of 362 327 blood cultures were analyzed from 27 medical ICUs, 35 medical-surgical ICUs, 121 medical wards, and 109 medical-surgical wards from 48 hospitals in 19 states and the District of Columbia. The adjusted mean BC use per 1000 patient-days was 273.1 (95% CI, 270.2-275.9) for medical ICUs, 146.0 (95% CI, 144.5-147.5) for medical-surgical ICUs, 80.3 (95% CI, 79.8-80.7) for medical wards, and 65.1 for medical-surgical wards. Blood culture use was significantly higher across all 4 unit types in hospitals with more than 500 beds compared with 500 or less beds and in the West-Midwest compared with other regions. Single blood culture and positive blood culture rates were below 10% across all 4 unit types. Of the 292 units, 97% had a mean BC contamination rate within 3% of the recommended threshold, and 51% were within 1%. The minimum BC use thresholds (ie, BC use below this number may represent undertesting) were 120 BCs per 1000 patient-days for medical ICUs, 80 BCs per 1000 patient-days for medical-surgical ICUs, and 30 BCs per 1000 patient-days for medical-surgical wards.Conclusions and RelevanceThe findings of this study suggest that blood culture positivity may help determine appropriate BC use for individual unit types.