Abstract
Abstract
Background:
Management of complicated intra-abdominal infections (cIAIs) includes broad-spectrum antimicrobial coverage and commonly includes vancomycin for the empiric coverage of methicillin-resistant Staphylococcus aureus (MRSA). Ideally, culture-guided de-escalation follows to promote robust antimicrobial stewardship. This study assessed the impact and necessity of vancomycin in cIAI treatment regimens.
Patients and Methods:
A post hoc analysis of the Study to Optimize Peritoneal Infection Therapy (STOP-IT) trial was performed. Patients receiving piperacillin-tazobactam (P/T) and/or a carbapenem were included with categorization based on use of vancomycin. Univariate and multivariable analyses evaluated effects of including vancomycin on individual and the composite of undesirable outcomes (recurrent IAI, surgical site infection [SSI], or death).
Results:
The study cohort included 344 patients with 110 (32%) patients receiving vancomycin. Isolation of MRSA occurred in only eight (2.3%) patients. Vancomycin use was associated with a similar composite outcome, 29.1%, vs. no vancomycin, 22.2% (p = 0.17). Patients receiving vancomycin had (mean [standard deviation]) higher Acute Physiology and Chronic Health Evaluation II scores (13.1 [6.6] vs. 9.4 [5.7], p < 0.0001), extended length of stay (12.6 [10.2] vs. 8.6 [8.0] d, p < 0.001), and prolonged antibiotic courses (9.1 [8.0] vs. 7.1 [4.9] d, p = 0.02). After risk adjustment in a multivariate model, no significant difference existed for the measured outcomes.
Conclusions:
This post hoc analysis reveals that addition of vancomycin occurred in nearly one third of patients and more often in sicker patients. Despite this selection bias, no appreciable differences in undesired outcomes were demonstrated, suggesting limited utility for adding vancomycin to cIAI treatment regimens.
P
With regard to cIAIs, MRSA is a relatively uncommon pathogen, especially in community-acquired cases [3,4]. In addition, current guidelines delineate specific patient populations for empiric MRSA coverage; these include patients with healthcare-associated cIAIs with known MRSA colonization or those at risk for MRSA from previous treatment failure or significant antibiotic exposure [1]. Vancomycin may also be included in an empiric regimen to provide for coverage of Enterococcus spp., primarily in healthcare-related infections [1,4].
Utilization of vancomycin has potential consequences. For example, indiscriminate therapy results in promotion of vancomycin resistance in Enterococcus spp. and S. aureus, leading to treatment failure for those in whom it is used as empiric treatment [5–9]. In addition, vancomycin carries the potential for several toxicities, particularly nephrotoxicity, which occurs in as many as 35%–40% of patients [10].
Despite focused recommendations provided by current guidelines and relatively low rates of MRSA in certain cIAIs, vancomycin remains commonly used in the empiric treatment of patients with cIAIs. This study aimed to determine the value of vancomycin use in the management of cIAIs.
Patients and Methods
Subjects and design
Subjects were selected from a database containing patient data collected from the Study to Optimize Peritoneal Infection Therapy (STOP-IT) trial, an open-label, multi-center, randomized controlled trial conducted to determine the effect of antimicrobial duration on treatment outcomes after adequate source control of cIAIs. The original STOP-IT publication contains the detailed study design description [11].
From the original 518 subjects enrolled from August 2008 through August 2013, the present study selected patients receiving therapy with piperacillin-tazobactam (P/T) and or or a carbapenem (ertapenem, imipenem-cilastatin, or meropenem). We reasoned that by focusing on patients receiving these two regimens, we would minimize the impact that resistant pathogens could play in treatment failure (e.g., ampicillin-sulbactam or fluoroquinolone resistant Escherichia coli). Subjects were then allotted into two groups: Those who received vancomycin and those who did not.
Data collection and outcomes
Patient demographics, comorbidities, procedures, antimicrobials, and infection characteristics were collected and stratified by receipt of vancomycin. Receipt of vancomycin was defined as drug administration anytime during the study period. Outcomes assessed included surgical site infection (SSI), recurrent intra-abdominal infection (recIAI), death, and a composite metric of all three outcomes. Original assessment of these complications occurred at 30 days [11].
Statistical analysis
Univariate analyses were performed to determine differences in patient characteristics, treatments, and outcomes based on receipt of vancomycin. Categoric variables were compared using either Pearson chi-square testing or, in cases of small cell sizes, the Fisher exact test. Continuous variables were compared using two-tailed, independent samples t-test. Effects of vancomycin on outcomes were analyzed by univariate and multivariable logistic regression models. Each multivariable model was characterized by outcome metrics as the dependent variable and vancomycin as an independent variable. The coefficient for vancomycin in a model was considered a measure of its effect on the outcome.
Demographics (i.e., age, gender, and race) and severity of illness (i.e., Acute Physiology and Chronic Health Evaluation II [APACHE II]) variables were also included as independent variables to adjust vancomycin means for possible patient differences within the multivariable model of composite outcome. Multivariable models for individual components of the composite outcome (i.e., SSI, recIAI, and death) did not adjust for demographics or severity of illness independently.
A power analysis was conducted to determine the required sample size for producing a significant difference in the composite outcome dependent on receipt of vancomycin. Variables included in the power analysis were β = 0.1, α = 0.05, hypothesized value of 0.30 for non-vancomycin, and a difference in composite outcomes between groups of 0.10. On the basis of the analysis, the study would require 939 patients with an allocation of 629 and 310 to the non-vancomycin and vancomycin groups, respectively. Data were analyzed using SAS 9.3 for Windows (SAS Institute, Cary, NC).
Results
Patient characteristics and treatment modalities
Of 518 patients originally included in the STOP-IT trial, 344 subjects received either P/T and/or a carbapenem. Of these, 110 patients also received vancomycin during the course of treatment for IAI. Patient baseline characteristics and treatment interventions are shown in Table 1. No major differences were observed between groups regarding demographics and co-morbid conditions. Regarding treatment modalities, the procedure for source control was similar between the two groups.
SD = standard deviation.
The majority of patients received P/T in this cohort (>75%); notably, more patients in the vancomycin group received P/T (89.1% vs. 79.1%, p = 0.02). In contrast, more patients without documentation of vancomycin administration received ertapenem, (21.8% vs. 1.8%, p < 0.0001). Patients receiving vancomycin on average received two additional days of total antibiotic therapy (9.1 vs. 7.1 d, p = 0.02).
Infection characteristics
Table 2 represents infection characteristics. Patients who were prescribed vancomycin had higher APACHE II scores at baseline (13.1 vs. 9.4, p < 0.0001) and remained hospitalized an average of four additional days (12.6 vs. 8.6, p < 0.0003). Isolation of MRSA occurred in eight patients (2.3% of the cohort), with only one of these patients receiving vancomycin as part of the treatment regimen. Micro-biologic cultures revealed that twice as many patients in the vancomycin group had Enterococcus spp. isolated (15.5% vs. 6.8%, p = 0.01). Setting of acquisition of the cIAI differed between the groups, with lower rates of community-acquired infections and higher rates of healthcare-associated and hospital-acquired infections in the vancomycin group (Table 2). Sites of the originating infection were similar between groups, except for appendicitis being higher in patients not receiving vancomycin (14.5% vs. 3.6%, p = 0.003).
SD = standard deviation; APACHE = Acute Physiology and Chronic Health Evaluation; MRSA = methicillin-resistant Staphylococcus aureus; IAI = intra-abdominal infection.
Outcomes
Table 3 presents the pre-specified outcomes as set by the original STOP-IT trial: SSI, recIAI, death, and the composite outcome stratified by receipt of vancomycin. Univariate analysis demonstrated no difference in treatment outcomes, both individual and composite, regardless of inclusion of vancomycin in the treatment regimen. The composite outcome occurred in 29.1% versus 22.2% (p = 0.17) of patients who were treated with and without vancomycin, respectively. Numerically, more patients in the vancomycin group had occurrence of recIAI, but this difference was not statistically different (21.8% vs. 13.7%, p = 0.06). Mortality rates were low overall—less than 2%, with an equivalent distribution between the groups.
Multivariable analysis revealed no effect for vancomycin on the individual outcomes: SSI (OR = 1.30, 95% CI 0.57–2.97, p = 0.53), recIAI (OR = 0.96, 95% CI 0.42–1.92, p = 0.78), or death (OR = 0.56, 95% CI 0.08–3.76, p = 0.55). After adjustment for baseline characteristics (i.e., age, race, gender, and APACHE II score), vancomycin did not independently predict the composite outcome (OR = 1.18, 95% CI 0.68–2.04, p = 0.55) in a stepwise regression model. Within our model, APACHE II score remained the only independent predictor of the composite outcome (OR = 1.06, 95% CI 1.02–1.11, p = 0.01).
Discussion
This study demonstrates no reduction in complications by including vancomycin in the treatment of cIAIs in patients receiving concomitant P/T and/or carbapenem therapy after adequate source control. On adjustment for demographics, infection characteristics, disease severity, and treatment strategies, including source control and antimicrobial therapy, vancomycin did not independently predict treatment failure. This suggests a limited utility of vancomycin in the empiric management of cIAIs.
Because of the low incidence of MRSA in cIAIs, it is unlikely that randomized controlled trials designed to test the utility of empiric vancomycin could be conducted. Therefore, post hoc analysis of previous randomized controlled trials conducted for treatment of patients with cIAIs offers a reasonable method to develop a preliminary understanding of important research questions. For example, this methodology may help to discern the role of empiric treatment regimens and thus prevent overexposure to antimicrobials.
This study incorporated statistical analyses including variables collected in the STOP-IT trial with the goal of providing a balanced group of patients on the basis of receipt of vancomycin for the treatment of patients with cIAIs. The only predictor in our model of treatment complications was admission APACHE II score. This finding lends support to the validity of the model because clinical severity often plays a role in outcomes of cIAIs [1,12–14].
Current treatment guidelines for cIAIs recommend vancomycin only in a select subset of patients including those with healthcare-associated infections with previous colonization with MRSA [1]. Nonetheless, vancomycin was included as empiric therapy for approximately one third of the cohort. This therapy included patients across the spectrum of acquisition setting (i.e., community vs. healthcare-associated vs. hospital-acquired). Although patients receiving vancomycin were more likely to have healthcare- or hospital-acquired infections, no association between vancomycin and treatment failure was demonstrated when setting of infection acquisition was included in the model. The limited number of patients with non-community acquired cIAIs hinders the generalizability of omitting vancomycin from empiric therapy for all cIAIs.
The incidence of MRSA in cIAIs varies with reported rates of 5%–19% with influence from several factors including previous antimicrobial exposure, colonization, and acquisition site of the infection [1,3,15–17]. Here we demonstrate a relatively lower rate of MRSA in cIAIs compared with those reported in the literature, which may limit applicability of study results to cohorts containing higher risk patients, especially patients with healthcare- or hospital-acquired cIAIs [3,16,18,19]. Also, nasal colonization predicts subsequent MRSA infection for several types of infections, including cIAIs [16,17]. Testing for MRSA colonization may provide an additional means for promoting antimicrobial stewardship through preventing or limiting exposure to patients whose tests are negative [17].
Vancomycin may add additional coverage of resistant Enterococcus spp. in certain types of cIAIs, especially healthcare-associated infections [1,4]. Our study demonstrated more than twice as many patients in the vancomycin group had Enterococcus spp. isolated. One possible explanation for this disparity may be the disproportionate number of hospital- and healthcare-acquired infections. Previous studies have demonstrated that isolation of Enterococcus does portend a more severe clinical picture [20,21]; of note, the vancomycin group did contain patients with higher APACHE II scores, which functions as a surrogate for clinical severity of illness [12].
Despite the higher percentage of Enterococcus spp. isolates in the vancomycin group, vancomycin did not independently predict treatment failure when isolation of Enterococcus spp. was included in the model. In addition, the inclusion of regimens with in vitro potential activity against Enterococcus spp., barring ertapenem, precludes the ability to rule out the effectiveness of adding vancomycin to other treatment regimens without inherent enterococcal activity (e.g., ciprofloxacin plus metronidazole).
Severity of a given cIAI may elicit a response from a provider to augment the empiric treatment regimen with vancomycin. The discrepancy in vancomycin use with more patients having this agent omitted from the regimen if the appendix constituted the origin of infection supports this postulation. In addition, vancomycin patients had higher APACHE II scores suggesting sicker patient populations prompt the inclusion of vancomycin in the empiric treatment regimen. Despite the underlying higher severity of the cIAIs in those receiving vancomycin, the addition of vancomycin did not independently predict treatment outcomes. Most notably, only one patient in the group deemed sicker and thus having the perceived “need” for vancomycin had MRSA cultured, thus demonstrating that 109 patients unnecessarily received exposure to the potential untoward effects of vancomycin.
Reducing patient exposure to unnecessary antimicrobials constitutes a key component of antimicrobial stewardship [2]. We demonstrate that patients receiving vancomycin on average received an additional 2 d of overall antimicrobial therapy. On the basis of this study, vancomycin plays a limited role in the treatment of patients with cIAIs who receive appropriate and adequate source control. This finding may provide an avenue for future antimicrobial stewardship studies to explore the potential impact of limiting vancomycin as part of empiric therapy.
This study has several limitations. First, the design represents a post hoc analysis of a previously conducted randomized controlled trial, limiting the ability to ensure these two subpopulations represent the overall population and the interpretation of the results. Thus, the analysis relied on statistical methodologies to provide “matching” of subjects based on patient characteristics and disease severity.
Second, the use of a previously populated database for this study may have omitted potential confounders that may have affected the model created. In particular, patients were not randomized to receive select antimicrobials, including vancomycin, for fixed durations, potentially limiting the overall effect of including vancomycin therapy. The analysis presented here operates under the assumption that vancomycin treatment was initiated empirically. Because of documentation of vancomycin as all or none, however, the utilization for culture-guided therapy cannot be ruled out. In addition, the inclusion of patients who only received P/T and/or a carbapenem may have introduced a biased subpopulation (e.g., resistance patterns, treatment site, etc.); however, the inclusion of this subpopulation aimed to limit the role gram-negative resistance would play in affecting treatment outcomes.
Last, the study was not powered sufficiently as evidenced by the power analysis suggesting a requirement of greater than 900 patients to detect a 10% difference for the specified treatment subgroups; therefore, this introduces the possibility for a Type II error and reduces the ability to discern the true effect of adding vancomycin to the treatment of cIAIs.
Conclusion
Despite the low incidence of MRSA in this cohort of patients studied, vancomycin therapy remained a commonly used part of broad-spectrum empiric therapy for cIAIs. On the basis of the present study, the addition of vancomycin appears to have a limited effect on preventing complications from cIAIs. These findings suggest vancomycin should be judiciously used for treatment of patients with cIAIs. Limitation of its use may obviate the potential collateral damage to the patient and the microbiome.
Footnotes
Author Disclosure Statement
No competing financial interests exist.