In vivo induction of AmpC beta-lactamases produces high-level resistance to many beta-lactam antibiotics in Enterobacteriaceae, often resulting in the need to use carbapenems or cefepime (FEP). The clinical effectiveness of piperacillintazobactam (TZP), a weak inducer of AmpC beta-lactamases, is poorly understood. Here, we conducted a case-control study of adult inpatients with bloodstream infections (BSIs) due to Enterobacter, Serratia, or Citrobacter species from 2009 to 2015 to assess outcomes following treatment with TZP compared to FEP or meropenem (MEM). We collected clinical data and screened all isolates for the presence of ampC alleles by PCR. Primary study outcomes were 30-day mortality and persistent bacteremia at Ն72 h from the time of treatment initiation. Of 493 patients with bacteremia, 165 patients met the inclusion criteria, of which 88 were treated with TZP and 77 with FEP or MEM. To minimize differences between covariates, we carried out propensity score matching, which yielded 41 matched pairs. Groups only differed by age, with patients in the TZP group significantly older (P ϭ 0.012). There were no significant differences in 30-day mortality, persistent bacteremia, 7-day mortality, or treatment escalation between the two treatment groups, including in the propensity score-matched cohort. PCR amplification and sequencing of ampC genes revealed the presence of ampC in isolates with cefoxitin MICs below 16 g/ml, in particular in Serratia spp., and demonstrated that these alleles were highly genetically diverse. Taken together, TZP may be a valuable treatment option for BSIs due to AmpC betalactamase-producing Enterobacteriaceae, diminishing the need for broader-spectrum agents. Future studies are needed to validate these findings.

Interventions, such as education and guideline development, can aid in reducing ART medication errors, but a committed stewardship program is necessary to elicit the greatest impact.

Klebsiella pneumoniae liver abscess (KPLA) has been described as an invasive syndrome with extrahepatic complications. The majority of KPLA is caused by capsular serotype K1 and K2 isolates. We report a case of carbapenem resistant Klebsiella pneumoniae liver abscess. The patient initially presented with infected right above-the-knee amputation and was later found with a large liver abscess. Initial antimicrobial susceptibility showed carbapenem resistant K. pneumoniae (CRKP). Further molecular workup revealed that the isolate was a less virulent non-K1/K2 serotype, and both rmpA and kfu genes were negative. The lack of outer membrane porins likely contributed to the carbapenem resistance. To our knowledge, this is a first reported case of carbapenem resistant, non-K1/K2 serotype K. pneumoniae liver abscess in the United States.

No abstract

We appreciate Aitken et al.'s careful review (1) of our recent paper assessing the use of piperacillin-tazobactam for the treatment of bloodstream infections caused by Enterobacteriaceae harboring AmpC ␤-lactamases (2). We respectfully disagree with their assertion that the numerical increase in mortality seen with the use of piperacillintazobactam compared to the mortalities of patients on carbapenems and/or cefepime in our and other studies suggests that inferior outcomes are to be expected with piperacillin-tazobactam.Aitken et al. are correct to point out that the odds ratio (OR) for the 30-day survival, not mortality, of patients receiving piperacillin-tazobactam versus that of patients receiving cefepime or meropenem was 0.5 in the propensity-matched data set. Therefore, the OR for 30-day mortality was 2.0 (95% confidence interval, 0.5 to 7.7), representing 3 additional deaths in the piperacillin-tazobactam group. This result was not statistically significant (P ϭ 0.33), indicating that we were unable to detect an association between the treatment regimen and the occurrence of mortality in this study. It is worth noting that the large confidence intervals reported here suggest a low level of the precision of the OR. Thus, it is inappropriate to draw a firm conclusion about the relative odds of mortality based on its numerical value alone. As we pointed out in our publication, propensity score matching lowered our ability to detect differences in outcomes. In the overall cohort, the OR for 30-day survival was 1.16, as reported, and the OR for 30-day mortality was 0.86 (95% confidence interval, 0.3 to 2.3; P ϭ 0.8). It is worth noting that in creating the propensity score-matched cohort, fewer deaths were excluded from the piperacillin-tazobactam group (3/9) than from the carbapenem/ cefepime group (6/9).Aitken et al. also suggested that using multivariable logistic regression is preferred to propensity score matching for some data sets. We opted for propensity score matching to minimize differences between baseline characteristics of treatment groups. In addition, a low number of outcome events in our data set made it difficult to perform multivariate models. Multivariable analyses of the overall cohort were attempted by incorporating variables that were significantly different between treatment groups, but these models demonstrated poor fit. However, they also showed no association between treatment groups and mortality, supporting our findings from the propensity score-matched analysis.The commenters further mention a paper by Harris and colleagues from 2017 with similar results (i.e., a numerically higher rate of recurrence of Enterobacter spp. bacteremia without a statistically significant difference) and a small patient population (3).