Skin and soft tissue infections (SSTIs) are classified based on the presence or absence of purulence; they are further divided based on severity and complexity.[1] Cellulitis is an infection of deep dermis and subcutaneous tissues characterized by swelling, erythema, pain, and tenderness.[2] A recent analysis of emergency department (ED) visits for SSTIs by the National Hospital Ambulatory Medical Care Survey for 1993–2005 concluded that there is nearly a 3-fold increase in the incidence of SSTIs.[3] Pathogens are typically skin flora with the most common causative pathogens being Group A β-hemolytic Streptococcus and Staphylococcus aureus.[2,4] Community-acquired methicillin-resistant Staphylococcus aureus (MRSA) has become increasingly prevalent such that it is included in the differential for causative pathogens.[5] Nevertheless, the current Infectious Diseases Society of America (IDSA) guidelines recommend that non-purulent cellulitis be treated with antibiotics active against methicillin-sensitive Staphylococcus aureus (MSSA) and β-hemolytic Streptococci only.[1] Broad-spectrum antibiotics are recommended only in immunocompromised patients or patients with systemic signs of illness.[1] The guidelines define systemic signs of illness as the presence of two or more of the systemic inflammatory response syndrome (SIRS) criteria. However, broad-spectrum antibiotics continue to be administered for patients with uncomplicated SSTIs.[6-8] Antimicrobial resistance is considered a growing problem worldwide. The increased use of anti-pseudomonal antibiotics has been linked to increased risk of multidrug-resistant organisms.[9] It is also important to mention that the risk of MRSA infections is increased in IV drug users, prisoners, children and those with penetrating trauma, and purulent drainage.[3]

The recommended duration of the treatment for uncomplicated cellulitis is 5 days of antimicrobial therapy, which has been shown to be as effective as a 10-day course.[1,8,10] Several studies have found that cellulitis treatment failure was not different regardless of the spectrum of activity.[6,11] Data are robust that non-purulent cellulitis can be treated with non-MRSA antibiotics such as penicillin or first-generation cephalosporins.[1,6] The purpose of this study was to determine the treatment failure rate of lower extremity cellulitis when treated with broad-spectrum antibiotics compared to narrow-spectrum antibiotics and to describe the inappropriate use of broad-spectrum antibiotics in lower extremity cellulitis.

A total of 599 patients were identified with a diagnosis of cellulitis between January 2016 and May 2016, of which 479 (79.9%) were excluded from the analysis after applying our inclusion and exclusion criteria [Figure 1]. The final cohort included 120 patients with lower extremity cellulitis [Figure 1]. Among the 120 included patients, 93 (77.5%) have received narrow-spectrum antibiotics and 27 (22.5%) have received broad-spectrum antibiotics. Demographics, laboratory, and microbiological values are shown in Table 1. Only 18 (19.3%) of those who received narrow-spectrum antibiotics were hospitalized and received injectable antibiotics compared to 27 (100%) who received broad-spectrum antibiotics (P = 0.05). The WBC was significantly higher among those who received broad-spectrum antibiotics (10.7 × 10⁹ cell/L ± 5.8) compared those who received narrow-spectrum arm (7.4 × 10⁹ cell/L ± 2) (P = 0.006). 4 patients (14.8%) had elevated lactic acid (i.e., ≥2 mmol/L) levels in the broad-spectrum group, while none of the patients in the narrow-spectrum group had elevated lactic acid (P = 0.001). Of the four patients, two had other reasons for lactic acid elevation in addition to cellulitis (i.e., rhabdomyolysis and missing hemodialysis session). SIRS was present in 6 of 27 (22%) patients in the broad-spectrum group compared to 1 of 18 (1%) in the injectable narrow-spectrum group [Table 1]. The most commonly used antibiotic in the broad-spectrum arm was PTZ 17 of 27 (62.9%), among which only 1 of 17 (5.8%) patient met SIRS criteria [Table 2]. Vancomycin in combination with PTZ was used in 13 of 27 (48.1%) patients, among which only 1 of 27 (7.6%) was positive for MRSA [Table 2]. The most commonly used antibiotics in the narrow-spectrum arm were cephalexin 48 (51.6%), sulfamethoxazole-trimethoprim 23 (24.7%), clindamycin 21 (22.5%), vancomycin 8 (8.6%), and cefazolin 5 (5.3%). MRSA nasal screening was performed in only 4 (4.3%) patients in the narrow-spectrum arm and none tested positive [Table 2]. Only 1 of 27 (3.7%) patient had a positive blood culture in the broad-spectrum arm. Yet, only 15 of 27 (55.5%) had blood samples collected for culture. 13 (48%) patients in the broad-spectrum arm had an all-cause repeat visit compared to 19 (20%) in the narrow-spectrum arm (OR: 3.47, [95% CI: 1.36–8.8]). 1 patient (4%) in the broad-spectrum arm had hospital repeat visit due to cellulitis compared to 3 (3%) in the narrow-spectrum arm (P = 0.89) [Figure 2].

Univariate logistic regression analysis revealed that the length of stay (OR: 1.36, [95% CI: 1.15–1.62]) and body temperature (OR: 1.5, [95% CI: 1–2.38]) were statistically associated with repeat visit [Table 3]. However, after adjusting for confounders in the multivariate logistic regression Table 4, only the length of stay was associated with higher odds for repeat visit (OR: 1.32, [95% CI: 1–1.6]). The mean duration of therapy was not different between the groups: 9.3 ± 2 days in the broad-spectrum arm compared to 8.3 ± 2 days in the narrow-spectrum arm (P = 0.19). 27 patients received more than 10 days of total antibiotic therapy 27 of 120 (22.5%) in both arms. The mean length of stay was similar between the broad-spectrum group and narrow-spectrum group (4.6 ± 2 vs. 4.1 ± 1 days, P = 0.29), respectively.

Cellulitis is a commonly encountered SSTI. The total number of SSTIs related visits to ambulatory care physicians increased from 8.6 million in 1997 to 14.2 million in 2005.[11] Despite the availability of national guideline, uncomplicated cellulitis is commonly treated with broad-spectrum coverage.[7,8,15] Few patients had systemic signs of illness (i.e., SIRS) and the majority of patients in the broad-spectrum arm (77.8%) 21 of 27 did not need such broad coverage.[1] This is contradictory to the most recent IDSA guidelines and consistent with previous studies demonstrating that broad-spectrum antibiotics were overutilized in patients with uncomplicated SSTIs.[6,8,15] Havey et al. found thatthe PTZ use was deemed inappropriate for 86.7% of patients with cellulitis, 66.7% of patients with abscess.[13] In their retrospective study, Jenkins et al. included 533, of which 320 (60%) patients diagnosed with non-purulent cellulitis. Blood cultures were collected from 225 of 320 (70%) patients and only 15 (5%) had positive blood cultures. Not surprisingly, Gram-negative organisms were identified only in 1 (2%) patient. Nevertheless, 144 of 320 (45%) patients received broad-spectrum antibiotics.[6] This study shares some similarities with their studies in that the use of antipseudomonal antibiotics (i.e., PTZ in our study) was commonly administered 17 of 27 (62.9%) with only 6 patients (22%) meeting SIRS criteria [Table 2]. Jenkins et al. conducted another retrospective pre-intervention-post-intervention study.[7] They found that the exposure to antibiotics with antipseudomonal activity decreased from 28% to 18%. Based on their findings, the development of guidelines seems to be an attractive solution to decrease the inappropriate use of broad-spectrum antibiotics. Walsh et al. reported the inappropriate use of antibiotics in SSTIs without complicating factors.[8] They reported that 21 of 163 (17.5%) received antipseudomonal coverage. Only 3 (6.4%) Gram-negative organisms were isolated from wound cultures.

Altogether, we think that our use of broad-spectrum antibiotics in non-purulent cellulitis is not justified given our clinical and laboratory findings. Because we excluded patients with complicated SSTIs, active IV drug users, patients with a history of MRSA infection, and purulent cellulitis, vancomycin use was considered inappropriate in 23 of 30 (76.6%), as only two patients had documented penicillin allergy and only nine had an MRSA nasal screening done of which two had SIRS on admission [Table 2]. In our institution, MRSA nasal screening is frequently done in the intensive care unit (ICU) only and will not be done for patients admitted to the medical floor unless ordered by the provider. MRSA nasal screening was done in 13 (10.8%) patients in both arms [Table 2]. Controversy exists with regard to the utility of MRSA surveillance in hospitalized patients.[13] One study found that MRSA colonization increases the risk of MRSA infection.[16] Safdar et al. conducted a systematic review to determine the relationship between MRSA or MSSA carriage status and subsequent infection.[17] They found that patients colonized with MRSA had 4-fold increased risk for MRSA infection compared to MSSA-colonized patients (OR: 4.08, [95% CI, 2.09–7.94]). In addition, IDSA guidelines recommend the addition of vancomycin or other antibiotics with activity against MRSA in patients colonized with MRSA.[1]

The mean duration of therapy in our study was similar in both arms (9.3 vs. 8.3 days) in the broad-spectrum arm and narrow-spectrum arm, respectively. Our study showed that the majority of our cellulitis patients 96 of 120 (80%) received treatment of 10 days or less, with only 24 of 120 (20%) receiving antibiotics of >10 days. This was surprising as our expectation was that the majority of patients will receive antibiotic therapy for 14 days or more as this was commonly reported in previous studies.[6,8] IDSA guidelines recommend duration with antimicrobial therapy for 5 days, but longer duration of therapy may be needed in patients who showed evidence of delayed response. Hepburn et al. conducted a randomized, double-blind placebo-controlled trial to determine if 5 days of therapy provided equal efficacy to 10 days of therapy.[10] They found that the resolution of signs of infection was similar between both groups. The authors of this study used Levofloxacin as the drug of choice of this study and generalizing results to different antibiotics may not be applicable. Jenkins et al. in his retrospective pre-intervention-post-intervention study also found that patients who received a short course of therapy were not more likely to experience clinical failure than those who received a longer course of therapy.[6]

This study has several limitations. First, this was a retrospective study and we could only control for confounders that we collected. Second, also due to the retrospective nature and the absence of blinding, there could be a chance for reviewer bias; however, we tried to avoid bias during data collection through extensive chart review and excluding patients with questionable non-purulent cellulitis. Third, due to the retrospective nature of this study, we did not evaluate delayed response and the need for antibiotics for more than 5 days, or previous antibiotic use before admission or ED visits. This is important as the most recent IDSA guidelines recommend to broaden the antibiotic coverage in patients who fail oral therapy. Fourth, severity of illness may not have been captured well in patient’s charts. Fifth, using repeat visit as the only criterion for treatment, failure may be insufficient and a closer follow-up and patient contact should have been implemented for better evaluation of the clinical outcomes. Sixth, the relatively small sample size in the broad-spectrum treatment arm could have affected the results of this study; however, it is comparable to the previous studies addressed the same problem. Last, we used SIRS criteria to determine the need for broad-spectrum antibiotics in patients with uncomplicated cellulitis. This is consistent with IDSA guidelines;[1] nevertheless, the task force of the Society of Critical Care Medicine considered SIRS criteria to be unhelpful in identifying sepsis.[17] This was supported by a study conducted in Australia and New Zealand that found one in eight patients admitted to ICUs with infection and organ failure did not fulfill the minimum two or more SIRS requisite to be labeled with sepsis diagnosis.[18]

In conclusion, the use of broad-spectrum antibiotics in patients with uncomplicated SSTIs is common and is not recommended based on the results of our study. We believe that implementation of clinical practice guidelines and providing frequent reminders to providers to adhere to national clinical guidelines will help in reducing the unnecessary use of broad-spectrum antibiotics.