We often tell our colleagues not to give antibiotics before obtaining blood cultures, but exactly how much culture yield are we potentially losing? In this prospective study of patients presenting to a single German ICU with sepsis between 2010 and 2017, the authors compared the yields of blood cultures drawn before and after the initiation of empiric antimicrobial therapy. Examining over 1300 blood cultures, they found that cultures were positive in 50.6% of cases when drawn before antibiotics versus only 27.7% of cases when drawn after antibiotics (p<0.0001). So, you can quote this study to tell docs that when they give the drugs first they're cutting their chances of finding a bug in half. PMID: 29879482
Can you trust your micro lab's antimicrobial susceptibility reports? That depends in part on whose antibiotic discs they're using. Here's a paper from the EUCAST lab in Sweden reporting variation in quality of the antibiotic discs for used for susceptibility testing in hospitals across the globe. The European Committee on Antimicrobial Susceptibility Testing, by the way, is essentially Europe’s version of our CLSI (the Clinical Laboratory Standards Institute): the very smart people who decide when bacteria should be considered resistance to antibiotics. Anyway, in 2014 the authors took discs for 16 different antibiotics from 9 major manufacturers and tested them against reference bacteria with known susceptibilities. They found that 8/9 manufacturers produced multiple discs that generated zones of inhibition significantly greater or lesser than the target values, and 5/9 manufactures produced multiple discs with major variations from the target values.
The EUCAST lab took these findings to the manufacturers and then repeated the study in 2017; the good news is that the antibiotic disc quality improved overall, but 7/9 manufacturers still produced multiple discs with discrepancies and 3/9 produced multiple discs with major discrepancies. The manufacturers producing the most reliable discs were Oxoid, SirScan, Bio-analyse, and Bio-rad. The antibiotics with the most unreliable discs were amoxicillin-clavulanate, piperacillin-tazobactam, and cefoxitin. I wonder if that has to do with the vagaries of detecting ESBL and/or inducible ampC beta-lactamases at different drug concentrations? PMID: 29886174
The BioFire blood culture ID PCR panel reliably rules out gram-negative VAP when used on respiratory samples. This study took tracheal aspirate and BAL fluid samples from 165 patients with suspected ventilator-associated pneumonia and ran the BioFire blood culture PCR panel on them. A pathogen was identified in the majority of cases, and 55% of samples contained Pseudomonas aeruginosa, Acinetobacter baumannii, or Klebsiella pneumoniae. Compared to the results of MALDI-TOF, the sensitivity of the BioFire assay was 79% with a specificity of 98% and a negative predictive value of 97%. Results of the BioFire (which could have been obtained within one hour) could have caused the empiric antibiotics to be changed to more appropriate empiric therapy in more than a third of cases. This data seems encouraging: add this on to the nasal MRSA screen and we could dramatically reduce our usage of vancomycin and carbapenems for VAP, so long as we ID specialists do a good job of convincing the ICU teams that this approach is reliable and safe. PIMD: 29906599
What's the strongest predictor of MRSA bacteremia? The nasal screening swab is a better predictor than any clinical variable. This restrospective study compared the clinical characteristics of 100 patients with MRSA bacteremia to 276 patients with MSSA bacteremia. They found that the MRSA screening nasal swab outperformed all other clinical parameters in predicting whether a patient's Staphylococcus aureus bacteremia was due to MRSA, and that this was true using either a 30-day cutoff for a positive swab or any prior positive. In the univariate analysis, no medical comorbidity was associated with MRSA bacteremia: not diabetes, ESRD, cirrhosis, HIV, cancer, or total Charlson Comorbidity Index. Hospitalization in the past 6mo or 12mo were the only other factors associated with MRSA bacteremia. When the authors created multivariate models of MRSA bacteremia risk, including other variables in the model only modestly improved the diagnostic accuracy of MRSA swab alone. PMID: 29890954