Introduction

Acute bacterial skin and skin structure infection (ABSSSI) is defined as a bacterial infection of the skin with a lesion area size of at least 75 cm2 (lesion size measured by the area of redness, edema, or induration). ABSSSI include cellulitis/erysipelas (A diffuse skin infection characterized by spreading areas of redness, edema, and induration), wound infection (an infection characterized by purulent drainage from a wound with surrounding redness, edema, and induration), and major cutaneous abscess (an infection characterized by a collection of pus within the dermis or deeper that is accompanied by redness, edema, and induration). Common bacterial pathogens causing ABSSSI are Streptococcus pyogenes and Staphylococcus aureus including methicillin-resistant S. aureus. Less common causes include other Streptococcus species, Enterococcus faecalis, or Gram-negative bacteria [1]. Typically, this infection has a relatively benign course [2]. Treating infectious diseases often emphasize the need to determine microbiology to plan the most effective treatment. However, in cases of cellulitis and acute bacterial skin and skin structure infections (ABSSSI), diagnosis and management mostly depend on the morphological features of the lesion and the clinical setting; the causal pathogen is relatively less important [3]. The role of blood culture in the management of ABSSSI remains controversial. This relates to the incidence of bacteremia in ABSSSI being relatively low, and contamination rates relatively high, which devalues the clinical significance of blood culture [4,5,6,7,8]. Infectious Diseases Society of America (IDSA) guidelines for the management of skin and soft-tissue infections state that blood cultures are recommended only for patients with deleterious conditions, such as malignancy, sepsis, immersion injury, animal bites, neutropenia, and severe cell-mediated immunodeficiency [3]. However, evidence to support these recommendations is limited. Several reported factors for positive blood cultures in cellulitis cases include: age > 65 years, nonlower extremity involvement, liver cirrhosis, and systemic inflammatory response syndrome [9]. Factors associated with a positive blood culture in patients with ABSSSI remain controversial, and findings vary among studies because of sample sizes or study designs. This study investigates factors associated with positive blood cultures, and correlation between bacteremia and mortality in patients with ABSSSI on the basis of clinical conditions including comorbidities and initial laboratory evaluations. This study has a large sample size, which provides a high statistical power of the results. On the basis of our findings, when presented with a case of ABSSSI, clinicians should be able to effectively differentiate between patients in whom blood culturing should be performed, and patients who should be treated more carefully because of the presence of high-risk factors for mortality.

Materials and methods

Patient selection

The Institutional Review Board of Chang Gung Memorial Hospital, Chiayi, Taiwan, approved this retrospective study. We collected retrospective data on patients who had a discharge diagnosis of ABSSSI, identified according to the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes 528.3 (cellulitis and abscess of oral soft tissues), 681.00–681.9 (cellulitis and abscess of the finger and toe), and 682.0–682.9 (other cellulitis and abscess), and who were admitted via our emergency department (ED) from March 2015 to December 2016. Each patient’s medical record was reviewed for documentation of ABSSSI or a related diagnosis, clinical characteristics, final diagnosis, hospitalization duration, comorbidities, and mortality. Some of the results have been published (age > 65 years, liver cirrhosis, lymphedema, systemic inflammatory response syndrome, chronic kidney disease, and leukocytosis of > 13.5 × 103/µL are shown to be associated with higher risk of bacteremia) [9, 13]. In the current study, we use data collected from these studies for further analysis to evaluate the association between laboratory data and clinical condition recorded in the ED, comorbidities, and bacteremia in ABSSSI patients. All blood cultures were drawn in ED at the same time. Patients aged ≤ 18 years, those found not to have ABSSSI on chart review, and those having a discharge diagnosis of another infectious disease were excluded from the study.

Data analysis

All patients who visited our ED between March 2015 to December 2016, who had a discharge diagnosis of ABSSSI, and who did not meet the exclusion criteria were enrolled in our study. We reviewed the patients’ charts and recorded the following variables: age, gender, clinical condition, site of infection, laboratory data, comorbidities, hospital stay, blood culture result, and survival status on discharge. We defined these variables as follows: anemia, hemoglobin < 10 mg/dL in ED; thrombocytopenia, platelet counts < 100 × 103 µL in ED; hyperglycemia, serum glucose level of > 200 mg/dL in ED; hypoalbuminemia, albumin < 3.0 g/dL in ED; hyperthermia, body temperature > 38 °C in ED; hypothermia, and body temperature < 35 °C in ED. The WBC counts, CRP, hemoglobin, platelet count, serum glucose, albumin, and creatinine were assessed in the first laboratory analyses in the ED.

Differences between the bacteremia and nonbacteremia groups were assessed using the Chi-square test for all the categorical variables (Table 1). To identify possible effects on bacteremia, a multivariate analysis was applied for variables with a p value of < 0.05 for differences between the bacteremia and nonbacteremia groups (Table 2). Differences between mortality and survival were also assessed using the Chi-square test (Table 3). Factors with a p value of < 0.05 for differences between mortality and survival were also included in multivariate analysis (Table 4). Because bacteremia is the primary factor evaluated in this study, it was included in the logistic regression model.

Table 1 Clinical characteristics of the bacteremia and nonbacteremia groups
Full size table
Table 2 Multivariable analysis to identify independent risk factors for bacteremia in patients with ABSSSI
Full size table
Table 3 Clinical characteristics of the mortality and survival groups
Full size table
Table 4 Multivariable analysis to identify independent risk factors for mortality in patients with ABSSSI
Full size table

Results

Of 1322 patients, 122 patients had positive blood cultures, and 28 patients died. Therefore, the positivity rate of blood cultures is 9.2%, and the mortality rate is 2.1%. The analysis of variables between bacteremia and nonbacteremia groups is summarized in Table 1. Compared with the nonbacteremia group, the bacteremia group has a higher incidence of diabetes mellitus (58.2% vs. 37.1%; p = 0.047) and chronic kidney disease (39.3% vs. 22.8%; p = 0.008). Significant differences are observed in the WBC count > 10 × 103/µL, CRP > 20 mg/dL, anemia, hyperglycemia, hypoalbuminemia, creatinine > 1.5 mg/dL, and hyperthermia. In addition, the bacteremia groups have longer hospital stays and a higher mortality rate.

Multivariate analysis reveals major contributing factors to bacteremia including: diabetes mellitus (p = 0.001, OR 4.18, 95% CI 2.03–14.85), chronic kidney disease (p = 0.003, OR 3.69, 95% CI 0.64–8.73), CRP > 20 mg/dL (p < 0.001, OR 2.55, 95% CI 1.68–9.26), hyperglycemia (p = 0.016, OR 1.95, 95% CI 1.09–4.15), and hypoalbuminemia (p = 0.001, OR 3.43, 95% CI 2.25–9.58) (Table 2).

The analysis of variables between the patients who survived and those who did not is summarized in Table 3. Significant differences are observed in the age subgroup of > 65 years; and percentage of patients with chronic kidney disease, liver cirrhosis, WBC count > 10 × 103/µL, CRP > 20 mg/dL, thrombocytopenia, hypoalbuminemia, and hyperthermia (Table 3). After controlling for factors that are significantly related to mortality, we find that age ≥ 65 years (odds ratio [OR]: 4.37, Table 4), liver cirrhosis (OR: 1.06), hypoalbuminemia (OR: 7.12), and bacteremia (OR: 1.97) significantly affect mortality. A power analysis was conducted for a logistic regression. It is determined that, for a sample size of 1322 participants, with an alpha of 0.05, OR 1.97 (bacteremia), the statistic power is nearly 1.0.

An extremity area was the most common (70.7%) site of cellulitis. No significant difference in bacteremia or mortality is found according to the site of infection (Tables 5, 6).

Table 5 Mortality according to the site of infection
Full size table
Table 6 Bacteremia according to the site of infection
Full size table

Staphylococcus and Streptococcus species are the most frequent causative organisms. Polymicrobial infections and anaerobic organisms account for 13.1% and 6.6% of infections, respectively. However, no difference in mortality is found according to the isolated organisms (p = 0.412, Table 7).

Table 7 Micro-organisms of patients with ABSSSI with positive blood culture
Full size table

Discussion

Our study examines the factors of clinical conditions, comorbidities, and laboratory data associated with bacteremia, which include diabetes mellitus and chronic kidney disease, high CRP level, hypoalbuminemia, and hyperglycemia. High CRP levels may indicate severe infection or inflammation caused by ABSSSI. Low albumin and high serum glucose levels are correlated with chronic kidney disease, diabetes mellitus, which are often associated with chronic illness and poor nutrition status. An association between comorbidity and positivity rates of blood culture has been described before, especially for malignancy, immunodeficiency, or diabetes mellitus [3, 10]. Diabetes mellitus is a comorbidity with a high prevalence, and the relationship between diabetes mellitus and positive blood cultures has been described [11]. In our study, the positivity rate for patients with diabetes is also higher than in patients without diabetes. A study addressing hypoalbuminemia (< 3 g/dL) and bacteremia as factors independently associated with prolonged hospitalization and mortality gives results similar with our findings [12]. In the logistic regression model for factors associated with mortality, bacteremia is identified as an independent risk factor for mortality (OR: 1.97) in ABSSSI. Therefore, a blood culture should be considered for patients with ABSSSI with diabetes mellitus or chronic kidney disease or those exhibiting abnormal CRP, glucose, or albumin levels because of the positive correlation between bacteremia and mortality. The causative organism has no significant association with mortality. The site of infection also demonstrates no clinical correlation with mortality.

A Swedish study shows that blood cultures are more frequently performed in patients who fulfill the severe inflammatory response syndrome (SIRS) criteria [13]. However, they did not compare the positivity rates of blood cultures between patients who did or did not fulfill the SIRS criteria. An American retrospective chart review compares the positivity rates of blood cultures in patients with cellulitis with or without fever [8]. Unexpectedly, patients without fever have positive blood cultures significantly more frequently than patients with fever, which supports our findings that hyperthermia is not associated with a higher positivity rate of blood cultures in ABSSSI. The results of this study may have clinical implication for the clinical choice to send blood cultures or not in patients with ABSSSI.

The most common cause of cellulitis is beta-hemolytic Streptococci (groups A, B, C, G, and F); most commonly, group A Streptococci or Streptococcus pyogenes are observed. S. aureus (including methicillin-resistant strains) is a notable but less common cause [14,15,16]. We contend that the increasing incidence rate of cellulitis caused by S. aureus (including methicillin-resistant strains) should be a global health concern. Gram-negative aerobic bacilli are identified in a minority of cases. In immunocompromised patients, the spectrum of potential pathogens is much broader, and infectious disease consultation is warranted, because a weakened immune system is a risk factor for mortality [14].

Our study has several limitations. Its study design (we used the initial laboratory data in the ED) may result in small populations of some factors (e.g., hyperthermia, thrombocytopenia, and leukocytosis), which might represent a significant impact on mortality in ABSSSI patients. This may limit the evaluation of these factors and other possible significant predictors. Furthermore, because of this retrospective setup, we selected patients based on the diagnosis documented in the patient files, and we were dependent on the documentation of health care providers. Moreover, we used all-cause mortality as an end-point, which may have overestimated the number of deaths due strictly to ABSSSI admissions.

There are several strengths to this study. Several characteristics and variables that have not previously been associated with bacteremia in cellulitis, which have positive correlations with mortality, are examined. The large number of patients that we studied provides a high power for estimates in the model that we used.

Conclusion

These findings can be used to stratify patients with ABSSSI according to the degree of risk for bacteremia, and may be helpful for deciding the most appropriate means of care—outpatient treatment or hospitalization.