Abstract
Background. Conflicting data exist on the role of antimicrobial therapy for the treatment of uncomplicated community-onset methicillin-resistant Staphylococcus aureus (MRSA) skin and soft-tissue infections (SSTIs).
Methods. We performed a retrospective cohort study of 492 adult patients with 531 independent episodes of community-onset MRSA SSTIs, which consisted of abscesses, furuncles/carbuncles, and cellulitis, at 2 tertiary care medical centers. The purpose of the study was to determine the impact of active antimicrobial therapy (i.e., the use of an agent to which the organism is susceptible) and other potential risk factors on the outcome for patients with uncomplicated community-onset MRSA SSTIs. Treatment failure was the primary outcome of interest and was defined as worsening signs of infection associated with microbiological and/or therapeutic indicators of an unsuccessful outcome. Bivariate analyses and logistic regression analyses were preformed to determine predictors of treatment failure.
Results. An incision and drainage procedure was performed for the majority of patients. Treatment failure occurred in 45 (8%) of 531 episodes of community-onset MRSA SSTI. Therapy was successful for 296 (95%) of 312 patients who received an active antibiotic, compared with 190 (87%) of 219 of those who did not (P = .001 in bivariate analysis). Use of an inactive antimicrobial agent was an independent predictor of treatment failure on logistic regression analysis (adjusted odds ratio, 2.80; 95% confidence interval, 1.26–6.22; P = .01).
Conclusions. Our findings suggest that certain patients with SSTIs that are likely caused by MRSA would benefit from treatment with an antimicrobial agent with activity against this organism.
Staphylococcus aureus causes skin and soft-tissue infections (SSTIs) that range from folliculitis to life-threatening conditions, such as necrotizing fasciitis [1, 2]. Common clinical manifestations for which patients seek medical care include furuncles, carbuncles, cellulitis, and abscesses. The recent emergence of methicillin-resistant S. aureus (MRSA) outside of the health care setting has complicated the management of these infections [3,4–5]. These isolates, termed “community-associated” or “community-acquired” MRSA, have been described worldwide and are defined by specific epidemiological and molecular criteria [6,7,8–9]. Importantly, community-associated MRSA strains often remain susceptible to non–β-lactam antibiotics, such as clindamycin, sulfonamides, and tetracyclines [10].
Management of uncomplicated community-associated MRSA SSTIs consists primarily of incision and drainage of fluctuant lesions, as recommended by current guidelines [11]. However, the additional role of active antimicrobial therapy (i.e., administration of an agent to which the organism is susceptible) in the treatment of these infections is much less established [8, 12]. This controversy is based on a number of previous reports that detected no differences in patient outcome regardless of whether an active antibiotic was administered [5, 13,14,15,16,17–18]. These studies had several potential shortcomings: methicillin-susceptible or nonstaphylococcal isolates were included in the study, the sample size was small, the primary end points were not clearly defined, information on comorbidities and other potential confounders was not provided, and/or minor infections that often resolve spontaneously (such as folliculitis) were included. In contrast to these studies, other data have suggested a potential association between the use of inactive antimicrobial therapy and treatment failure [19, 20].
Thus, more data are needed to define the importance of active antimicrobial therapy in the management of MRSA SSTIs. We performed a study of a large cohort of patients who presented to a clinic or emergency department with non-life-threatening MRSA SSTI.
Methods
Study population. We conducted a retrospective cohort study of all adult patients (age, ⩾18 years) who presented with community-onset, uncomplicated MRSA SSTI to 2 tertiary care medical centers in Little Rock, Arkansas, during the period from 1 January 2003 to 28 February 2006. Classification as a community-onset infection required the isolation of MRSA from a clinical specimen obtained during an outpatient visit or within 48 h after hospital admission. Isolates were further categorized as community-associated MRSA if none of the following epidemiological risk factors for health care-associated infection were present: hospitalization, surgery, dialysis, or residence in a long-term care facility within the previous year; presence of a permanent indwelling catheter or percutaneous medical device at the time of culture; or previous isolation of MRSA from a clinical or surveillance specimen [21, 22].
The study was conducted at the University of Arkansas for Medical Sciences Hospital (Little Rock), a 341-bed tertiary care referral center that is the only academic medical institution in Arkansas, and at the Central Arkansas Veterans Healthcare System (Little Rock).
Study design and definitions. Adult patients with community-onset MRSA SSTIs were identified by review of the clinical microbiology laboratory records at both institutions. The medical records of all patients from whom a clinical wound specimen yielded MRSA were reviewed to include all potential case patients. SSTIs were classified into 3 categories: cutaneous abscesses, furuncles or carbuncles, and cellulitis.
Minor or superficial skin infections, such as folliculitis or impetigo, were excluded, because these infections commonly resolve spontaneously or with the use of topical antibiotic therapy alone. We also excluded complicated SSTIs (i.e., nonhealing skin ulcer or diabetic foot infection, postsurgical wound infection, or processes involving adjacent deep-tissue structures, including bone, fascia, or tendon sheaths). Zero time was defined as the day of the first incision and drainage procedure. For patients for whom surgical drainage was not performed, zero time was the day of the first positive wound culture result for a sample obtained from a wound exudate swab, needle aspiration, or skin biopsy. Treatment failure was the primary outcome of interest; this was defined as documented worsening of signs of infection at least 2 days after zero time accompanied by ⩾1 of the following events: performance of an additional incision and drainage procedure, subsequent hospital admission, occurrence of new culture-proven MRSA SSTI while the patient was receiving antimicrobial therapy, and/or microbiological failure. The latter was defined as the persistence of MRSA-positive culture of specimens from the original wound site after the completion of antibiotic therapy. Signs and symptoms of infection included erythema, induration, edema, fluctuance, tenderness, and purulent drainage.
If a patient presented with a subsequent episode of culture-proven MRSA SSTI at least 1 month after his or her initial episode, the later episode was included in the analysis if it fulfilled all of the aforementioned criteria and if it occurred at a different body site. Antimicrobial therapy was considered to be active if it included at least 1 agent to which the organism showed in vitro susceptibility. Identification and susceptibility testing of S. aureus isolates were performed in accordance with the Clinical and Laboratory Standards Institute guidelines [23, 24]. Routine testing for inducible clindamycin resistance via the double-disk diffusion assay was performed only during the latter part of the study period: 59 (12%) of 491 isolates were tested, but only 2 (3%) of these 59 isolates displayed inducible clindamycin resistance. On the basis of these results, patients treated with clindamycin were considered to have received active treatment, unless the isolate was found to be clindamycin resistant by automated susceptibility testing [25].
Data obtained from medical records included demographic characteristics, medical and clinical history (including anatomical site of infection), duration of symptoms, peripheral leukocyte count, duration of antibiotic therapy, and duration of follow-up after zero time. The study was approved by the institutional review boards of both hospitals.
Statistical analysis. The Pearson's χ2 test, Fisher's exact test, and Wilcoxon rank sum test were performed to compare categorical and continuous variables, respectively (P < .05). ORs were calculated rather than relative risks to facilitate comparison with the results of logistic regression analysis. Unconditional forward and backward logistic regression analyses were then conducted to determine independent risk factors for treatment failure. Variables with a P value <.2 on bivariate analysis were included in the model. All statistical analyses were conducted using SPSS for Windows software, version 11.0 (SPSS).
Results
During the 38-month observation period, MRSA was isolated from 3680 clinical specimens at both hospitals. We identified a total of 581 eligible episodes of community-onset MRSA SSTI at the 2 hospitals. Of these, 50 episodes were excluded for the following reasons: no follow-up was conducted ⩾2 days after zero time (26 episodes), no susceptibility data were available for the prescribed antibiotic (12 episodes), multiple organisms were isolated (6 episodes), or medical records were incomplete (6 episodes).
The final cohort comprised a total of 531 episodes of community-onset MRSA SSTI in 492 patients. Thirty-one patients (6%) received a diagnosis of a second episode during the study period, and 3 (1%) had >2 independent episodes. Of note, patients who were included in the study and those who were excluded did not differ significantly with regard to a variety of parameters, such as type of SSTI, performance of an incision and drainage procedure, or MRSA susceptibility to the prescribed antibiotic (P > .1 for each).
Patient characteristics and antimicrobial susceptibility profiles. We first examined the baseline demographic and clinical characteristics of the 492 patients at the time of their initial enrollment episode. The median age of this cohort was 47 years; 390 subjects (79%) were male, and 292 (59%) were white. Two hundred fifty-seven subjects (52%) presented to the Central Arkansas Veterans Healthcare System. Twenty-six subjects (5%) used injection drugs, 22 (4%) were HIV positive, and 84 (17%) had diabetes mellitus. Three hundred fifty-one subjects (71%) were categorized as having a community-associated MRSA infection. Antimicrobial susceptibility data are presented in table 1. Three hundred thirty-three isolates (68%) were susceptible to all of the tested antibiotics except β-lactams and usually macrolides.
Antimicrobial susceptibility profiles of community-onset methicillin-resistant Staphylococcus aureus isolates.
Antimicrobial susceptibility profiles of community-onset methicillin-resistant Staphylococcus aureus isolates.
Clinical presentation and antimicrobial therapy. For all additional investigations on the role of active antibiotic therapy and other factors on clinical outcome, all 531 episodes of community-onset MRSA SSTI were evaluated. The proportion of MRSA isolates that caused community-onset SSTI among all MRSA isolates identified per year increased significantly throughout the study period (P < .01 for the trend). Abscesses were the most common presentation (361 episodes), followed by cellulitis (116 episodes) and furuncles or carbuncles (54 episodes). Cellulitis resulted from a focal lesion, such as folliculitis, a nonchronic skin ulcer, or an abscess, in all cases. Active antimicrobial therapy was administered in 312 episodes (59%); in 219 episodes (41%), inactive therapy was administered. The 2 groups were very similar with regard to a variety of demographic and clinical characteristics (tables 2 and 3). A total of 296 (95%) of 312 episodes treated with an active antibiotic within 48 h after zero time were treated successfully, compared with only 190 (87%) of 219 episodes in which the patient did not receive an active agent (P = .001 on bivariate analysis).
Comparison of baseline variables between episodes in patients who received active antimicrobial therapy and episodes in those who received inactive antimicrobial therapy (n = 531).
Comparison of baseline variables between episodes in patients who received active antimicrobial therapy and episodes in those who received inactive antimicrobial therapy (n = 531).
Comparison of clinical characteristics between patients who received active versus patients who received inactive antimicrobial therapy (for 531 episodes)
Comparison of clinical characteristics between patients who received active versus patients who received inactive antimicrobial therapy (for 531 episodes)
Patient outcome. A total of 45 patients (8.5%) experienced treatment failure, which occurred at a median of 3 days after zero time (interquartile range, 2–4 days). According to our definition, treatment failure was determined by worsening signs of infection in association with the presence of at least 1 of the criteria listed in table 4. An additional incision and drainage procedure, subsequent admission to the hospital, or both were necessary in 43 of the 45 episodes. One patient presented with worsening of a skin abscess on follow-up in addition to a second (previously not present) culture-proven MRSA SSTI after receiving cephalexin therapy; both lesions responded quickly after therapy with an active agent (clindamycin) was initiated. Another patient with a cutaneous abscess initially responded to a 14-day course of vancomycin; signs of infection recurred, however, and MRSA with an identical susceptibility pattern was cultured from a worsening wound exudate specimen 2 weeks after the completion of vancomycin therapy. Patients with a successful outcome received antibiotics for a median of 10 days (interquartile range, 7–14 days); β-lactams, tetracyclines, fluoroquinolones, and sulfonamides were administered in 29%, 23%, 18%, and 10% of these episodes, respectively. No patient died as a consequence of their SSTI or within 30 days of follow-up.
Criteria for treatment failure in 45 patients with community-onset methicillin-resistant Staphylococcus aureus skin and soft-tissue infection.
Criteria for treatment failure in 45 patients with community-onset methicillin-resistant Staphylococcus aureus skin and soft-tissue infection.
Predictors of treatment failure. Results of bivariate analyses of clinical characteristics associated with treatment failure are summarized in table 5. Failure to initiate active antimicrobial therapy within 48 h after zero time was the only variable associated with treatment failure (OR, 2.82; 95% CI, 1.49–5.34; P = .001). Twenty-four (83%) of 29 patients for whom inactive therapy had failed had received β-lactams. The site of infection did not impact treatment outcome (data not shown). Information on abscess size, measured as the largest diameter of induration, was available for 351 (97%) of 361 abscess episodes. Abscess diameter of >5 cm was not associated with treatment failure (P > .6), even after stratification by activity of antimicrobial therapy. For 113 (21%) of 531 episodes, patients had received antibiotics for a median duration of 3 days before zero time; receipt of an inactive agent (n = 88) before zero time was not associated with later treatment failure (P = .34, by Fisher's exact test).
Comparison of clinical characteristics between patients with treatment success and patients with treatment failure of community-onset methicillin-resistant Staphylococcus aureus skin and soft-tissue infections (for 531 episodes).
Comparison of clinical characteristics between patients with treatment success and patients with treatment failure of community-onset methicillin-resistant Staphylococcus aureus skin and soft-tissue infections (for 531 episodes).
On logistic regression analysis, failure to initiate active antimicrobial therapy within 48 h after zero time remained the only independent predictor of treatment failure (adjusted OR, 2.80; 95% CI, 1.26–6.22; P = .011). Analysis that was limited to only the first enrolled episode in each patient (n = 492) revealed identical results. A subgroup analysis of all 427 episodes (80%) in which an incision and drainage procedure was performed at zero time also revealed that failure to initiate active antimicrobial therapy within 48 h after zero time was the only independent predictor of treatment failure (P = .013).
Discussion
Similar to published reports from other medical centers, our study describes an increase in the incidence of community-onset MRSA SSTIs at 2 hospitals during a recent 38-month period [4,5–6, 26]. In this large cohort, the overall rate of treatment failure was 8.5% (45 of 531 episodes), as determined by strictly defined criteria. Lack of active antibiotic therapy administered within 48 h after zero time was independently associated with treatment failure (P = .011 on logistic regression analysis). No significant differences in demographic characteristics or comorbidities were observed between the 2 treatment groups. In addition, the anatomical site of infection did not affect clinical outcome, a finding that concurs with the results of a previous study of 219 superficial skin abscesses [15].
To our knowledge, we have performed the largest study to investigate the impact of antibiotic therapy on the outcome of patients with community-onset MRSA SSTI. The large number of enrolled patients may have enabled us to detect a relatively small difference in outcome between patients who received active versus inactive antimicrobial therapy that was not noted in previous studies [13,14,15,16,17–18]. Furthermore, the use of different outcome measures by other investigators may have also influenced study findings. Among patients who presented to an emergency department with community-associated MRSA SSTI in a recent multicenter study, receipt of active antimicrobial therapy was not related to treatment failure [5]. However, outcomes were assessed by telephone follow-up 15–21 days after the initial enrollment. Our definition of treatment failure required the combination of clinical and therapeutic criteria (documented worsening of signs of infection at least 48 h after zero time that necessitated additional surgical drainage procedures or admission to the hospital). These are readily available and clinically relevant criteria, and they demonstrated high interobserver agreement (>95%). Patients with worsening signs of infection whose therapy was switched from inactive to active therapy as the only additional therapeutic measure (without the need for admission or an additional incision and drainage procedure) were not determined to have experienced treatment failure to decrease the potential for misclassification bias. Thus, it is possible that our findings even underestimate the importance of active antimicrobial therapy. The severity of SSTIs may have also varied among different studies; this may bias conclusions regarding the need for antibiotic therapy. We carefully excluded both minor lesions and complicated infections, and we only assessed a relatively homogenous disease spectrum, which consisted of cellulitis, furuncles/carbuncles, or abscesses. Our rate of initial hospitalization at zero time was similar to that reported in previous studies and was not related to patient outcome [4, 13].
The majority of community-associated MRSA infections are caused by a few specific clones, as defined by molecular criteria established by the Centers for Disease Control and Prevention. USA300 is currently the most prevalent clonal type in the United States [4, 5, 10]. Community-associated MRSA strains differ from health care-associated MRSA strains with regard to several characteristics, among them the presence of genetic resistance determinants, such as the mobile staphylococcal cassette chromosome mec type IV, and certain virulence factors, such as the Panton-Valentine leukocidin. Panton-Valentine leukocidin-positive S. aureus strains have been associated with severe SSTIs and pulmonary infections [27,28,29–30]. Because we did not perform PFGE or other typing methods on our patients' isolates, we used community onset of the infection as an approximate indicator. King et al. [4] recently found that 244 (87%) of 279 MRSA strains recovered from patients with community-onset SSTI fulfilled their definition of a community-associated MRSA isolate. Furthermore, susceptibilities to non-β-lactam antibiotics and the presence of risk factors for health care-associated infection (e.g., previous hospitalization or previous MRSA isolation) were very similar in our study, compared with other investigations of community-associated MRSA infections [4, 13, 21, 31]. In addition, a recent surveillance study performed by the Arkansas Department of Health determined that USA300 was the most common causative agent of community-onset staphylococcal SSTIs in central Arkansas (N.S., unpublished data). Thus, we believe that the majority of the strains in our study were community-associated MRSA. Even if this were not the case, our study would more likely underestimate than overestimate the importance of appropriate antimicrobial therapy, given the documented pathogenicity of community-associated strains [32,33–34].
The importance of surgical drainage of purulent collections in patients with cutaneous abscesses and other lesions has been clearly established [11, 14]. On the basis of these findings, we defined the day of any incision and drainage procedure performed during the course of each episode as zero time. This allowed us to examine more specifically whether antimicrobial therapy provided any additional benefit with regard to the outcome of these infections. Furthermore, this designation constituted a more clearly defined time point within the course of disease progression upon which further outcome evaluation could be objectively based. However, it needs to be emphasized that our zero time definition did not allow us to examine the independent impact of surgical drainage on patient outcome. Thus, incision and drainage should continue to be considered the mainstay of treatment of these infections.
There were limitations to our study. MRSA isolates were not available for molecular testing, as was already discussed. Secondly, patients were not observed for a uniform duration; rather, they were observed for an individualized period based on each treating physician's discretion. We cannot exclude the possibility that some patients may have developed clinical deterioration after a period of documented improvement. However, patients in both treatment groups had a median of 2 follow-up visits 2–14 days after zero time. Only 21% of patients were observed for <5 days; these patients' infections had improved significantly such that further observation was not deemed to be necessary on the basis of their physicians' final assessment. In addition, the majority of patients subsequently received care at our hospitals for unrelated reasons; none reported any previously undocumented complications pertaining to an SSTI. Finally, the retrospective design of the study raises the possibility of selection, information, and outcome identification biases. We attempted to limit the influence of selection and information biases by reviewing all available medical, nursing, and pharmacy records of every patient who had MRSA isolated from a nonsterile site during the study period. Strict and reproducible criteria that defined zero time and treatment failure were established to evaluate treatment outcome objectively.
In summary, our study provides detailed information on epidemiological characteristics, underlying comorbidities, and treatment outcomes for a large cohort of patients with community-onset MRSA SSTIs. Incision and drainage of any focal fluid collection should remain the mainstay of therapy. Administration of active antimicrobial therapy was associated with a statistically significant—but only moderate—impact on clinical outcome (success rate, 95% vs. 87%). Additional studies are needed to identify patients who would most benefit from the addition of active antibiotic therapy. Until then, it may be reasonable to closely monitor patients with MRSA SSTI after surgical drainage has been performed and to limit antimicrobial therapy to those patients with a suboptimal response. This selective approach may be most cost-effective, may limit the unnecessary use of antibiotics, and may delay the emergence of additional antimicrobial resistance.
Acknowledgments
We would like to thank Philip Mwangi for his technical assistance.
Potential conflicts of interest. All authors: no conflicts
Comments