Abstract
Background:
Early initiation of broad-spectrum antibiotic agents is a cornerstone of the care of necrotizing skin and soft tissue infections (NSTI). However, the optimal duration of antibiotic agents is unclear. We sought to characterize antibiotic prescribing patterns for patients with NSTI, as well as associated complications.
Patients and Methods:
Using an NSTI registry, we characterized antibiotic use at a quaternary referral center. Kaplan-Meier analyses were used to describe overall antibiotic duration and relative to operative source control, stratified by presence of other infections that independently influenced antibiotic duration. Factors associated with successful antibiotic discontinuation were identified using logistic regression.
Results:
Between 2015 and 2018, 441 patients received antibiotic agents for NSTI with 18% experiencing a complicating secondary infection. Among those without a complicating infection, the median duration of antibiotic administration was 9.8 days (95% confidence interval [CI], 9.2–10.5) overall, and 7.0 days after the final debridement. Perineal NSTI received fewer days of antibiotic agents (8.3 vs. 10.6) compared with NSTI without perineal involvement. White blood cell (WBC) count and fever were not associated with failure of antibiotic discontinuation, however, a chronic wound as the underlying infection etiology was associated with greater odds of antibiotic discontinuation failure (odds ratio [OR], 4.33; 95% CI, 1.24–15.1).
Conclusions:
A seven-day course of antibiotic agents after final operative debridement may be sufficient for NSTI without any secondary complicating infections, because clinical characteristics do not appear to be associated with differences in successful antibiotic discontinuation.
Necrotizing skin and soft tissue infections (NSTI) are serious, potentially life- and limb-threatening infections requiring rapid diagnosis and intervention to limit morbidity and mortality. 1 Beyond aggressive surgical debridement, early initiation of antibiotic agents remains one of the most important therapeutic interventions. 2 However, the etiologic organism(s) are unpredictable. A complex, polymicrobial milieu often exists, presenting a challenge to ensuring adequate coverage and necessitating empiric broad-spectrum antibiotic agents that are not without risk.
Although empiric antibiotic agents are intended to cover the most likely and the most virulent pathogens, data to guide ongoing antibiotic management in NSTI are lacking. Determining how and when to de-escalate and discontinue broad-spectrum coverage remains elusive, with individual practitioners utilizing different combinations of clinical characteristics, laboratory findings, and source control to guide treatment duration. A recent multicenter study revealed substantial variation in care pattern and antibiotic duration even at centers with expertise in the management of NSTI, illustrating how little is known about best practices in complex patients with NSTI. 3 Existing guidelines—based largely on expert opinion—recommend continued antimicrobial therapy until the last debridement in a clinically improved and afebrile patient.4,5 Although some limited observational data exist, there remains a knowledge gap in defining the appropriate duration of antibiotic agents.6,7 A growing body of evidence supporting shorter courses for most bacterial infections and the increasing focus on patient safety have renewed interest in addressing this knowledge gap and reducing the overutilization of unnecessary antibiotic agents. 8
We sought to characterize the antibiotic prescribing patterns in a large NSTI cohort and evaluate patient characteristics that may influence antibiotic management decisions to help determine the optimal duration of therapy. We hypothesized that the success of antibiotic discontinuation would be related to normalization of white blood cell (WBC) count and fever, and time after final debridement.
Patients and Methods
Data collection
All adult patients presenting with NSTI to Harborview Medical Center (Seattle, WA) were included in a prospective, quality improvement registry beginning in January 2015; the details of this registry are described elsewhere. 1 Harborview is a 413-bed acute care hospital that serves as a public safety-net hospital for King County, as well as the level 1 trauma and burn center for Washington, Wyoming, Alaska, Montana, and Idaho. This study was approved by the Institutional Review Board at the University of Washington.
Our institution used standardized antibiotic regimens for patients with suspected NSTI based on body site involvement and underlying patient conditions as part of a clinical care pathway. 5 For perineal NSTI, piperacillin-tazobactam or meropenem, plus vancomycin, and clindamycin were administered; for non-perineal NSTI, penicillin, levofloxacin, vancomycin, and clindamycin were administered. Specific regimens were tailored to target the most likely organisms based on anatomic site, while also taking into consideration pre-existing patient conditions (e.g., neutropenia or penicillin allergy). Of note, patients at our institution typically receive high-dose clindamycin (1,200 mg intravenously every six hours) as part of empiric therapy given the theoretical antitoxin effects in severe group A Streptococcus (GAS) and clostridial infections. 9 Operative cultures—either swabs or tissue from deep surgical sites—are routinely obtained, and these results guide subsequent antibiotic de-escalation.
Antibiotic administration, laboratory values, and physiologic data were obtained from the electronic medical record using Microsoft Amalga (Microsoft Corporation, Redmond, WA). Fever was defined as a temperature of 38.5°C or greater. Medical charts and microbiologic data were reviewed to identify patients who experienced a secondary infection (e.g., bacteremia, pneumonia, osteomyelitis) that could impact the duration of antibiotic agents independent of the NSTI. Patients were deemed to have experienced a complicated NSTI if the administration of antibiotic agents for a secondary or concurrent infection overlapped with the administration of antibiotic agents for NSTI. Conversely, an uncomplicated NSTI was defined as a clinical course without a secondary or concurrent infection requiring antibiotic agents that overlapped with those administered for NSTI.
Patient charts were reviewed to determine when antibiotic agents were definitively discontinued for either NSTI or their complicating infection. Antibiotic duration did not include antibiotic agents administered at another institution prior to transfer. Debridement was defined as an operative procedure occurring at our institution during which necrotic tissue was excised, or residual infection was drained. Operations that were intended for wound preparation for reconstruction were not considered debridement for the purposes of source control.
Our primary aims were to characterize the antibiotic prescribing patterns in NSTI and patient characteristics at the time of antibiotic discontinuation. Discontinuation success was defined as discontinuation of all antibiotic agents with no subsequent need for antibiotic agents for NSTI. Finally, we assessed the risk of Clostridioides difficile infection (CDI) with antibiotic administration.
Statistical analysis
Data were analyzed using R (Version 3.6.1, R Foundation for Statistical Computing, Vienna, Austria). Categorical variables were reported as absolute values and percentages, and continuous variables as medians and interquartile ranges. The Mann-Whitney U test was used to compare continuous variables.
Kaplan-Meier analyses were used to estimate probability of continuing antibiotic agents. Patients were censored if they died, were placed in comfort care, or were discharged while taking antibiotic agents. The overall duration of antibiotic was defined as the time between hospital admission and the last dose of antibiotic administered for either NSTI or a complicating infection.
Logistic regression was used to identify patient and clinical characteristics associated with outcomes of interest: antibiotic discontinuation success and CDI. Patient characteristics selected a priori were based on expert opinion and published recommendations, and included overall duration of antibiotic agents, time since last debridement, WBC count, and fever.
Results
Patient and infection characteristics
Between 2015 and 2018, 445 adult patients with NSTI were identified. Four patients died or were placed on comfort care prior to receiving antibiotic agents or were transferred after having completed a course of antibiotic agents elsewhere and were excluded.
Patients included in this study were predominantly middle-aged, (median age, 55 years; interquartile range [IQR], 43–62), male, and white (Table 1). Three hundred ninety-six patients (90%) were transferred from another institution. Data regarding timing of presentation relative to an outside hospital were available for 286 patients, with 75% of patients transferred to our facility within one day of presentation. Seventy-eight patients (18%) endorsed a penicillin allergy. Diabetes mellitus was the most common pre-existing condition, and the perineum was the most frequently involved site of the NSTI (37%), followed by the leg (34%). Nearly one-third of infections had no clear etiology. Infectious complications occurred in 81 patients (18%; Table 2), with bacteremia being the most common complicating infection (56%), followed by osteomyelitis (22%).
Demographic and Clinical Characteristics of Patients Who Received Any Antibiotic for NSTI Stratified by Presence of a Complicating Infection
NSTI = necrotizing skin and soft tissue infections; CHF = congestive heart failure; CRF = chronic renal failure; HTN = hypertension; IQR = interquartile range; qSOFA = Quick Sequential Organ Failure Assessment; WBC = white blood cell count.
Some patients had more than one body site involved.
Clinical Outcomes Among Patient With NSTI Who received Any Antibiotic Agents, Stratified by Infectious Complication
NSTI = necrotizing skin and soft tissue infections; ICU = intensive care unit; IQR = interquartile range.
Of 28 days.
Some patients had more than one complicating infection.
Among patients with limb involvement.
Treatment and clinical outcomes
Four hundred thirty-two patients (98%) underwent at least one surgical debridement at our institution, with a median of three debridements (IQR, 2–4) per patient. Surgical debridements were completed after a median of three and two days for patients with and without an infectious complication, respectively. Overall, patients required few days of mechanical ventilation or intensive care; however, hospital stays were relatively long (Table 2). Approximately 20% of patients with limb involvement required an amputation, and mortality was 15% overall. Among the 64 (14.5%) patients who died, the median time to death was three days from admission.
Microbiology data
Three hundred seventy-six patients (85%) who underwent a debridement had intra-operative specimens obtained for wound culture within 48 hours of admission. There were no differences in demographic or infection characteristics between those who did and did not have culture data. Most infections were polymicrobial (73%), with streptococcal and staphylococcal species predominating (Table 3). Clostridia were rarely identified in wound cultures (3%).
Microbiologic Data Among Patients With Surgical Wound Culture Obtained Within Forty-Eight Hours of Admission
MRSA = methicillin-resistant Staphylococcus aureus; MSSA = methicillin-sensitive Staphylococcus aureus; GAS = group A Streptococcus; GBS = group B Streptococcus.
Initial antibiotic selection
Among patients who received antibiotic agents for NSTI, all but one patient received antibiotic agents within 24 hours of admission. The majority (87%) of patients received one of the institutionally recommended antibiotic regimens, however only 75% received an appropriate regimen for both their body site involvement and allergy profile. Some patients received penicillin despite a reported allergy (16/78). There was no difference in receipt of an appropriate antibiotic regimen in the first 24 hours after admission based on transfer status.
Antibiotic duration
The median duration of antibiotic agents for uncomplicated NSTI was 9.8 (95% CI, 9.2–10.5) days (Fig. 1A). Patients who experienced a complicating infection received a median of 23.6 (95% CI, 17.6–35.9) days of antibiotic agents. Patients with uncomplicated NSTI with perineal involvement received fewer days of antibiotic agents overall compared with patients without perineal involvement (median, 8.3 [95% CI, 9.2–10.5] vs. 10.6 days [95% CI, 9.7–11.6]; p = 0.01). Among the 416 patients who remained on antibiotic agents at the time of their final debridement, antibiotic agents were continued for a median of 7.0 (95% CI, 6.5–7.8) days thereafter for patients with uncomplicated NSTI, and 19.7 (95% CI, 14.6–31.2) days thereafter for patients with a complicating infection (Fig. 1B). No differences in treatment duration were observed by transfer status, and there were no differences in antibiotic duration among patients with extremity involvement based on whether they received amputation. Among the 327 patients whose antibiotic discontinuation was observed, WBC count at the time of discontinuation did not differ between those who did and did not experience a complicating infection (median 10.0 vs. 10.5; p > 0.05).
Kaplan-Meier curve plotting probability of continuing on antibiotic agents by (
Most patients received empiric vancomycin and clindamycin for NSTI (96% and 98%, respectively) consistent with our institutional recommendations. Among patients with uncomplicated NSTI, the median duration of vancomycin administration was 3.8 days, and clindamycin administration was 5.1 days. The subset of uncomplicated NSTI involving confirmed methicillin-resistant Staphylococcus aureus (MRSA) or coagulase-negative staphylococcus received a longer median duration of vancomycin (5.0 vs. 3.6 days; p = 0.001) compared with uncomplicated NSTI involving other pathogens. The median duration of clindamycin was similar between patients with and without GAS or clostridium (5.6 vs. 4.8 days; p > 0.05) NSTI, respectively.
Clostridoides difficile infections
Twenty (4.5%) patients developed CDI. Every one-week difference in antibiotic duration was associated with 20% greater odds of CDI (OR, 1.20; 95% CI, 1.07–1.32). Among patients who received clindamycin, every one-day difference in clindamycin exposure was associated with 6% greater odds of CDI (OR, 1.06; 95% CI, 1.02–1.09).
Factors associated with successful antibiotic discontinuation among uncomplicated NSTI
Among the 274 patients with uncomplicated NSTI, antibiotic agents were discontinued successfully on the first attempt in 260 (95%) patients after a median total duration of 9.7 days (IQR, 8.9–10.3) and 6.9 days (IQR, 6.4–7.6) from the last day of surgical debridement, respectively (Table 4). Among the 14 patients for whom the first attempt of antibiotic discontinuation after a median total duration of 6.7 days (IQR, 4.8-9.5) failed, antibiotic agents were subsequently re-started with a longer overall duration compared with those whose antibiotic agents were successfully discontinued (median 16.8 vs. 9.7 days; p < 0.01). The WBC counts were not significantly different at the time of antibiotic agent discontinuation, and there was no difference in incidence of fever in the preceding 48 hours. Among the subgroup of patients who had antibiotic agents stopped after less than seven days from time of final debridement, there was no association between either WBC count or temperature in the preceding 48 hours and antibiotic discontinuation success or failure. Secondary analyses evaluating pre-existing conditions, body site, and infection etiology were performed; a chronic wound as the underlying infection etiology was associated with greater odds of antibiotic discontinuation failure (OR, 4.33; 95% CI, 1.24–15.1).
Patient and Clinical Characteristics at the Time of First Antibiotic Discontinuation Among Patients With Uncomplicated NSTI Whose Discontinuation Was Observed
NSTI = necrotizing skin and soft tissue infections; WBC = white blood cell count; IQR = interquartile range;
p < 0.01.
Discussion
Current practice guidelines do not adequately address the optimal duration of antibiotic agents in NSTI because of the lack of high-quality evidence to inform best practices. We described the prescribing practices at a quaternary referral center with experience caring for NSTI. Among patients with uncomplicated NSTI, the median duration of antibiotic agents was approximately 10 days overall, and seven days after final debridement. Commonly cited indicators of the adequacy of source control, including fever and WBC count, were not associated with antibiotic discontinuation success or failure, suggesting that a seven-day course of therapy after definitive surgical debridement may be sufficient in uncomplicated NSTI. This is consistent with results from other studies, including one with 142 patients with NSTI, which concluded no differences in mortality, length of stay (LOS), or re-admission among patients treated with seven days or less of antibiotic compared with longer courses, provided adequate surgical debridement was achieved 7 ; an observational study of 322 patients also demonstrated no difference in recurrence, intensive care unit LOS, or mortality among patients treated with less than or greater than 48 hours of antibiotic after final debridement 10 ; and a small prospective study comparing short duration of antibiotic (48 hours after source control) to a historical control population which demonstrated similar treatment failure and mortality. 11
Despite the high morbidity and mortality associated with NSTI, these findings suggest that a short course of antibiotic agents after source control is safe and effective without negatively affecting patient outcome. It is possible that a shorter course of less than seven days may be appropriate for some patients. Further randomized controlled trials are needed to elucidate the shortest effective duration of therapy.
In the absence of consensus guidelines, clinicians often utilize clinical characteristics and laboratory findings to guide antibiotic management. Commonly cited indicators of the adequacy of infection control—including fever and WBC count—were not associated with the success or failure of complete antibiotic discontinuation in this study. However, failure of antibiotic discontinuation was an uncommon event, and this study may have been underpowered to detect small differences in these factors. The available data do, however, suggest that both overall antibiotic duration, and duration after source control, are important when considering antibiotic discontinuation in uncomplicated NSTI. Shorter durations raise concern for unchecked progression of infection, and premature antibiotic discontinuation may contribute to the need to restart antibiotic agents, and a longer duration of therapy overall. An underlying chronic wound associated with acute NSTI was also associated with failure of antibiotic discontinuation and may represent poor penetration of antibiotic agents because of pre-existing vascular disease, difficult-to-clear organisms, or unappreciated infectious complications such as osteomyelitis.
In this study, complicating infections were common, and antibiotic duration among patients with a complicating infection was longer. In particular, patients with osteomyelitis and septic arthritis receive long antibiotic courses—up to six weeks. Longer antibiotic duration in the subset of patients with concomitant infectious complications is likely driven by the infectious complication rather than the NSTI itself.
Perineal NSTI is a unique clinical entity from non-perineal NSTI, with lower mortality and a potentially less aggressive natural history. 1 Notably, we found that overall antibiotic duration was shorter for patients with perineal involvement compared with those without perineal involvement. This is also supported by a retrospective study of patients with Fournier gangrene that demonstrated no difference in mortality, primary closure, surgical site infection, or CDI among patients treated with less than seven days of antibiotic agents and those treated with longer course. 6 These finding supports prior assertions that perineal NSTI is a unique clinical entity from non-perineal NSTI, with lower mortality and a potentially less aggressive natural history. 1
We also sought to characterize the empiric use of specific antibiotic agents. As expected, uncomplicated NSTI involving MRSA or coagulase-negative staphylococcus received a significantly longer duration of vancomycin compared to uncomplicated NSTI involving other pathogens. However, this highlights a stewardship opportunity as the median duration of vancomycin was 3.6 days among patients without MRSA or coagulase-negative staphylococcus. The clinical significance of coagulase-negative staphylococcus is unclear in the context of NSTI, although it was recovered frequently (24%) from surgical specimens in our cohort. As such, treatment decisions should be made on a case-by-case basis.
Given the high prevalence of GAS in NSTI, and the potential production of exotoxins in Clostridium, empiric clindamycin is thought to limit virulent toxin production in these species. 12 Furthermore, clindamycin retains activity during the stationary phase of bacterial growth, and is associated with improved survival when used in combination with a β-lactam for invasive GAS infection. 13 The global emergence of clindamycin resistance in streptococcal infections,14–17 has led us to postulate whether oxazolidinones antibiotic such as linezolid may reduce virulence factor and toxin production in clindamycin resistant isolates. Experimental studies in mice have demonstrated in vitro activity and efficacy of oxazolidinones against fulminant GAS myonecrosis, including isolates harboring clindamycin resistance. 18 Furthermore, oxazolidinones exhibit activity against MRSA, and may facilitate the consolidation of vancomycin and clindamycin into a single regimen. Lehman et al. 19 sought to compare linezolid to the combination of vancomycin and clindamycin as empiric therapy for NSTI in a real-world setting. Linezolid was associated with one less day of MRSA-active therapy and lower incidence of acute kidney injury, suggesting that linezolid may be a safe alternative.
Ultimately, optimal antibiotic duration is a difficult balance between achieving source control and minimizing the risk of antibiotic-associated complications. Clostridioides difficile infection is a potentially preventable complication associated with antibiotic administration, and associations between dose and duration of antibiotic agents and risk of CDI have been well described. 20 In this study, we demonstrated that both a longer antibiotic course and a longer clindamycin course were associated with greater odds of CDI. The increased risk of CDI may be in part attributed to the combination of levofloxacin and clindamycin, both with high propensity for causing CDI, and the high dose of clindamycin used at our institution. Antimicrobial stewardship interventions such as targeted education and prospective audit of clindamycin are warranted to facilitate rapid discontinuation among patients without GAS or clostridium.
This study has several limitations. Only antibiotic agents administered during hospitalization at our institution were included in the antibiotic duration, yet 90% of patients were transferred from another institution with most patients transferred to our facility within one day of presentation. Despite this fact, we did not find any differences in initial antibiotic selection or antibiotic duration based on transfer status. This suggests that most patients were transferred early in their clinical course and were treated similar to patients presenting primarily to our institution. These findings may limit the generalizability to smaller facilities where patients may initially present; however, we believe most transfers are because of lack of surgical or intensive care capabilities to manage these patients long-term, and that patients are likely to be transferred to larger referral centers where these data are generalizable.
Contrary to other previously published studies, our study was not designed to compare a fixed short versus long duration of therapy for NSTI. Specifically, we aimed to derive the optimal duration of therapy from our study cohort who did not experience antibiotic discontinuation failure. The exact clinical decision-making process guiding antibiotic discontinuation was often difficult to assess retrospectively but our definition of antibiotic discontinuation as a proxy for clinical success was the most pragmatic strategy to help define the optimal duration of therapy. As a surrogate, we utilized temperature and WBC count to define the clinical context at the time of antibiotic changes. Similarly, time since final debridement was selected as a surrogate indicator of wound appearance and adequacy of source control. Finally, the patients in this registry were cared for at a single institution and may not be generalizable to other facilities. However, our institution has significant NSTI and this registry represents one of the largest prospectively collected datasets of an otherwise rare disease.
Conclusions
In summary, a seven-day course of antibiotic agents after final operative debridement may be sufficient for uncomplicated NSTI, because clinical characteristics such as WBC count and fever do not appear to be associated with antibiotic discontinuation success or failure. Additionally, patients with perineal NSTI may require shorter durations of therapy compared with patients without perineal involvement. To our knowledge, our study is the largest investigation to date evaluating antibiotic prescribing pattern and contributes to the available evidence to guide optimal antibiotic duration in uncomplicated NSTI and help target antimicrobial stewardship efforts. Further prospective data are needed to inform best practice to define the shortest effective duration of therapy in the management of invasive necrotizing soft tissue infections.
Footnotes
Funding Information
This work was supported by the National Institutes of Health [T32 5T32GM121290 to D.L.H].
Author Disclosure Statement
No competing financial interests exist.