A systematic review of the association between delayed appropriate therapy and mortality among patients hospitalized with infections due to Klebsiella pneumoniae or Escherichia coli: how long is too long?

Recent data from the Study for Monitoring Antimicrobial Resistance Trends program [1] found that during 2011 the global rates of extended-spectrum beta lactamase (ESBL)-containing Escherichia coli from intra-abdominal infections ranged from 6% (North America) to 38% (Asia). During the preceding decade, rates for urinary tract infections (UTIs) caused by ESBL E. coli and Klebsiella pneumoniae increased from 3.3 to 8.0% and from 9.1 to 18.6%, respectively, across 217 hospitals in the United States [2]. Results from the Healthcare Cost and Utilization Project Nationwide Inpatient Sample database [2] show that the number of complicated (ie, hospitalized) UTIs caused by carbapenem-resistant Enterobacteriaceae (CRE) in the United States increased from 0% in 2000 to 2.3% in 2009. The clinical burden of these resistant infections is substantial: ESBL-producing Enterobacteriaceae and CRE account for an estimated 1700 and 610 deaths annually, respectively, in the United States [3].

Reasons for concern regarding the rising rates of resistance are many. In more than 25% of patients with resistant Gram-negative infections, initial empirical antibiotic therapy has no microbiological activity against the pathogen causing the infection [4, 5]. Patients who receive delayed appropriate therapy (DAT) may have a 1.5- to 2-fold increase in morbidity and mortality relative to those who receive early appropriate therapy [6].

A second reason for concern is that resistance limits treatment options. For ESBLs, clinicians are often left using last-line agents, such as carbapenems, empirically. Clinicians have had to rely on older, often toxic agents (eg, colistin) with limited efficacy data to treat infections caused by CRE [7, 8].

This review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.

There were several notable findings from this systematic literature review and meta-analysis that predominately (95%) included studies of patients with bloodstream infections due to K. pneumoniae or E. coli. First, the results suggest that overall, delays in appropriate therapy for patients with K. pneumoniae or E. coli infections result in higher mortality rates than when patients receive early appropriate therapy. On average, patients who receive any DAT (any delay exceeding 24 h after culture collection or C& SR) have up to 2-fold higher odds of death. This is consistent with the observed association between time to appropriate therapy and mortality for serious infections due to other pathogens [31]. Likewise, the results of this meta-analysis indicate that a delay in appropriate therapy that exceeds 48 h or lasts until after the time of C& SR is associated with an up to 2-fold higher risk of mortality. Taken together, the results for overall DAT suggest that even a delay of 24 to 48 h may negatively impact patient outcomes. Although not statistically significant, the ORs associated with DAT > C& SR only, which approximate a 48- to 72-h delay for reporting of culture and susceptibility data, were highly consistent with the ORs for a > 48- to > 72-h DAT. Finally, the results also suggest that the odds of mortality increased in a stepwise pattern as delays increased from > 24 to > 72 h. Although there were too few studies to draw definitive conclusions, it may be that the 24-h time point represents an inflection point beyond which a higher mortality risk is observed with increasing delays in initiating appropriate therapy. In general, the findings from this meta-analysis underscore the importance of timely initiation of appropriate therapy especially among patients with bloodstream infections due to K. pneumoniae or E. coli.

These findings have important implications for clinical practice, particularly highlighting the critical importance of using appropriate antibiotic therapy at initiation of treatment or “getting it right the first time.” Delivery of early, appropriate therapy is a fundamental pillar of antibiotic stewardship, and delivery of an appropriate drug during the course of an infection is one of the most important measures clinicians can take for their patients to ensure the highest probability of a successful outcome. In particular, the findings suggest that clinicians should be more aggressive with their empiric antibiotic choices, especially among patients with bloodstream infections due to K. pneumoniae or E. coli. As part of this proposed practice shift, it is of the upmost importance that therapy is “de-escalated” once culture and susceptibility data are available, when possible. Otherwise, prolonged overuse of broad spectrum antibiotic will exacerbate our growing resistance problems and will limit the future utility of current antibiotics. Because the increasing prevalence of antibiotic resistance increases the likelihood of a patient receiving inappropriate empirical therapy, our findings also highlight the need to develop tools to both identify patients at high risk for antibiotic resistance and facilitate rapid diagnosis of the underlying causative pathogen to ensure early appropriate therapy.

Several things should be noted when interpreting the findings. Nineteen of the 20 studies included in this meta-analysis were of patients with bloodstream infections due to K. pneumoniae or E. coli. It is unclear whether the results are applicable to other infection types. Study design limitations included those inherent to any meta-analysis, such as heterogeneity across studies that may originate from differences in study populations, including those that combined both hospital- and community- onset infections, different definitions of response, different definitions of in vitro susceptibility and appropriateness, and the lack of a multivariate control. Heterogeneity was noted in several of the meta-analyses. Interestingly, the studies driving the heterogeneity were those that assessed the association between DAT and infections due to ESBLs. The definition of appropriate therapy varied across these studies and likely contributed to the observed heterogeneity. For example, the study by Tuon and colleagues [26] only considered carbapenems as appropriate for the treatment of infections due to ESBLs and classified all other beta lactam therapies as inappropriate; and resulted in a lower rate for appropriate therapy. However, cefepime and piperacillin/tazobactam are active against most ESBLs and therefore may have been considered appropriate therapy in other studies.

The use of a key objective endpoint (mortality) across all studies helped to minimize any bias. Nonetheless, the presence of publication bias cannot be excluded. In addition, a targeted analysis taking into account all confounders was not possible, limiting the ability of this meta-analysis to definitively establish the presence of a causal relationship. Because most of the included articles were retrospective observational cohort studies that were not randomized with regard to DAT, possible treatment selection bias and bias in management decisions with respect to dosage adjustments could not be analyzed or accounted for in the present analysis. It was also difficult to determine a definitive cutpoint for time delay given the limited number of studies included in this analysis. It is possible that there are different cutpoints for different infection types or disease severities. Another potential limitation is that the published literature included here focused predominantly on adults with bloodstream infections; pediatric and other infection types might differ in terms of the impact of DAT. Additional studies in these other infection types are necessary to explore this question. For example, Kumar and colleagues [32] reported that appropriate antibiotic therapy within the first hour is associated with improved survival in patients with septic shock. A final limitation is that this analysis did not evaluate other outcomes, such as length of hospital stay and costs. Data are needed to evaluate how delays in therapy affect these outcomes and whether there are different time delay cut-points.

In summary, results of this meta-analysis, which predominately (95%) included studies of patients with bloodstream infections due to K. pneumoniae or E. coli, indicate that delayed appropriate therapy increases the risk of mortality, especially when appropriate treatment is delayed by > 48 h. Given the results of this study documenting the adverse outcomes associated with delayed appropriate therapy, our findings suggest that there is a need to shift current treatment practices away from antibiotic escalation strategies that contribute to delayed appropriate therapy and toward early, relatively aggressive and comprehensive, antibiotic therapy, especially among patients with bloodstream infections due to K. pneumoniae or E. coli. As part of this proposed practice shift, it is critical that clinicians “streamline” therapy once culture and susceptibility data are available, when possible, to shorten the duration of time that patients unnecessarily receive broad spectrum antibiotics. Otherwise, prolonged overuse of broad spectrum antibiotics will exacerbate the growing resistance problem and will limit the future utility of current antibiotics. These findings also highlight the need for developing additional treatment options and strategies for infections in which therapeutic resistance is most likely to arise.