Diagnostic Microbiology and Infectious Disease 79 (2014) 273–279
Contents lists available at ScienceDirect
Diagnostic Microbiology and Infectious Disease journal homepage: www.elsevier.com/locate/diagmicrobio
Inappropriate initial antibiotic treatment for complicated skin and soft tissue infections in hospitalized patients: incidence and associated factors B.A. Lipsky a, b,⁎, L.M. Napolitano c, G.J. Moran d, L. Vo e, S. Nicholson e, M. Kim e a
University of Oxford, Oxford, UK University of Washington, Seattle, WA, USA c University of Michigan, Ann Arbor, MI, USA d Olive View-UCLA Medical Center, Sylmar, CA, USA e Janssen Scientiﬁc Affairs, LLC, Raritan, NJ, USA b
a r t i c l e
i n f o
Article history: Received 12 April 2013 Received in revised form 11 February 2014 Accepted 13 February 2014 Available online 24 February 2014 Keywords: Complicated skin and soft tissue infections Inappropriate antibiotics treatment Incidence
a b s t r a c t We analyzed 525 hospitalized adults treated with intravenous antibiotic(s) for complicated skin and soft tissue infections (cSSTIs) to assess incidence of, and risk factors associated with, inappropriate initial antibiotic treatment (IIAT). IIAT was given to 22.5% of enrolled patients. The rate of IIAT did not vary by type of facility (academic versus community) but was signiﬁcantly higher in rural than urban hospitals (38.9% versus 21.3%, P = 0.02). Pathogens were exclusively gram-positive in 68% of patients, exclusively gram-negative in 13%, and mixed in 19%. Staphylococcus aureus was the most frequently isolated pathogen (in 65%), 54% of which were methicillin-resistant. Signiﬁcant independent risk factors for IIAT were: admission to a rural hospital (odds ratio = 2.34; 95% conﬁdence interval: 1.06–5.19), dialysis treatment (3.86; 1.15–12.93), cancer other than non-melanoma skin cancer (5.23; 1.78–15.36), and infection with gram-negative (3.43; 1.79–6.60) or mixed (4.52; 2.62–7.78) pathogens. IIAT for cSSTIs was relatively frequent in these hospitalized patients, especially those with selected risk factors. © 2014 Elsevier Inc. All rights reserved.
1. Introduction Complicated skin and soft tissue infections (cSSTIs) are a frequent clinical problem that often result in hospitalization for antibiotic therapy and sometimes require surgical procedures (Nichols, 1999). Initial antibiotic treatment for cSSTIs is almost always empirical, since culture and sensitivity results are not yet available when treatment is initiated (Fung et al., 2003). Selecting an antibiotic regimen is therefore mostly based on the expected pathogen(s), as well as the severity of the infection, various patient factors (e.g., history of allergic reactions, renal, or hepatic insufﬁciency), the presumed source of infection, and local antibiogram data. This selection process has become more challenging in the past decade, as both the types of causative pathogens and the likelihood that they will demonstrate antibiotic-resistance have increased (Chambers, 2001; Klein et al., 2007). These circumstances heighten the risk that the initial antibiotic regimen will be ineffective against the causative pathogen(s). Several studies have reported on the incidence of inappropriate initial antibiotic therapy (IIAT) for various types of bacterial infections (Eagye et al., 2009; Ibrahim et al., 2000; Kollef et al.,
⁎ Corresponding author. Tel.: +44-1865-559078. E-mail address: [email protected]
(B.A. Lipsky). http://dx.doi.org/10.1016/j.diagmicrobio.2014.02.011 0732-8893/© 2014 Elsevier Inc. All rights reserved.
1999, 2008; Kumar et al., 2006; Luna et al., 2006; Micek et al., 2007; Schramm et al., 2006; Shorr et al., 2010; Shorr et al., 2008; Zilberberg et al., 2008, 2009, 2010). These investigations have generally found that patients who receive IIAT, especially those with pneumonia or bacteremia, have worse outcomes (Eagye et al., 2009; Ibrahim et al., 2000; Kollef et al., 1999, 2008; Kumar et al., 2006; Micek et al., 2007; Schramm et al., 2006; Shorr et al., 2008; Zilberberg et al., 2010). Some studies have also shown that, in patients who received IIAT, the risk of worse outcomes is not attenuated by subsequent treatment escalation (i.e., broadening the spectrum) of antibiotic therapy (Zilberberg et al., 2008). Despite the frequency of cSSTIs, very few studies have examined the incidence of, risk factors for, or consequences of IIAT in these patients (Eagye et al., 2009; Itani et al., 2011; Lipsky et al., 2010; Zilberberg et al., 2009, 2010). In routine clinical practice, prescribing patterns for many treatments can vary widely, based on the kinds of infections seen, the specialty of the practitioner, the types of healthcare facilities, and the facility’s location. To increase the generalizability of the ﬁndings of a study of IIAT, it is important to assess a relatively heterogeneous population. Therefore, we designed this study to examine the incidence of IIAT among patients hospitalized for cSSTI in a variety of settings in the US. We also sought to determine what factors might be associated with an increased likelihood of IIAT in our patient population by examining
B.A. Lipsky et al. / Diagnostic Microbiology and Infectious Disease 79 (2014) 273–279
their demographic data, clinical characteristics, and the patterns and appropriateness of the initial IV antibiotic therapy they received. 2. Methods 2.1. Study design and patients We conducted this analysis as part of a prospective, multicenter, observational study designed to investigate the clinical characteristics of cSSTIs in hospitalized patients and the antibiotic treatments they received (Lipsky et al., 2012). We enrolled patients admitted to any of the 62 selected hospitals from June 1, 2008, to December 31, 2009, with 1 of 4 types of infection: diabetic foot infection (DFI), surgical site infection (SSI), cellulitis, or deep soft tissue abscess (DSTA). We speciﬁcally selected hospitals of varying size, ownership status (e.g., private, city/county, university), academic afﬁliation (any type), and geographic region. Each of the treating physicians who agreed to participate in this study independently made all decisions about patient management and prospectively collected the requisite data on standardized forms. Adult patients with any of the designated types of cSSTI were eligible for inclusion in the study if they: 1) were to receive intravenous (IV) antibiotics as the primary treatment regimen during their hospitalization, 2) had an expected inpatient hospital stay of at least 48 hours, and 3) had at least 1 pathogen identiﬁed on culture from the site of the cSSTI. We considered wound isolates to be contaminants, rather than pathogens, if they were coagulase-negative staphylococci isolated as one among other probable pathogen(s) or from only a swab specimen or they were Corynebacterium species. We also excluded patients whose hospitalization data were incompletely recorded and those for whom we could not determine the pathogen type or appropriateness of their antibiotic therapy.
and cancer other than non-melanoma skin cancer). Characteristics assessed for the participating hospitals included their afﬁliation type (academic versus community), geographic region (i.e., Midwest, Northeast, South, or West), and location (urban or rural). We obtained the following microbiological data: the speciﬁc pathogen(s) isolated, wound pathogen type (i.e., whether the culture isolates were only gram-positive, only gram-negative, or mixed), and the antibiotic susceptibility of each isolate. We documented the initial IV antibiotic regimen used for the study infection, including the proportion of patients who received 2 or more classes of antibiotics. Furthermore, we documented all surgical and non-surgical procedures each patient underwent related to the study infection, as well as whether or not we considered the procedure as “source control” of the infection (i.e., incision and drainage, surgical debridement, excision, or amputation of the infected wound). 2.4. Data analysis We calculated the rates of IIAT by infection type, wound pathogen type, hospital characteristics, HCA risk factors, and initial antibiotic regimen. After stratifying study patients based on whether they received appropriate or IIAT, we conducted descriptive analyses on the 2 cohorts. For continuous variables, we calculated the mean and SDs, using the student’s t test to determine statistical signiﬁcance. We summarized categorical variables using frequencies and percentages and used the chi-square test for statistical testing. To assess factors associated with receiving IIAT, we used a logistic regression model, including those independent variables we considered clinically important a priori. The following independent variables were considered: age, gender, infection type, HCA risk factors, pathogen type, and hospital characteristics. We estimated odds ratios (ORs) and 95% conﬁdence interval (CI) for the factors analyzed.
2.2. Deﬁnitions 3. Results We classiﬁed initial IV antibiotic treatment as inappropriate if: 1) it was not given within 24 hours of hospital admission, or 2) the selected agents were not active against the identiﬁed pathogens, based on in-vitro susceptibility testing done at the local site (when available), or usual spectrum of coverage for a speciﬁc isolate (based on susceptibilities in standard reference texts, a proxy measure used for susceptibility in those cases in which test results were unavailable) (Zilberberg et al., 2010, 2012). For patients who lacked adequate data on pathogen susceptibility results or spectrum of coverage of selected agents, we classiﬁed the appropriateness of antibiotic therapy by achieving consensus among 3 physician authors (BAL, GM, and LN), each of whom have extensive experience and expertise in treating cSSTI patients. We classiﬁed infections as healthcare-associated (HCA) if the affected patient met at least 1 of the following criteria: had been hospitalized within the 6 months immediately preceding admission for the study-related infection; was a resident of, or was admitted from, a nursing home; was in an immunosuppressed state (organ transplantation, neutropenia, primary or secondary immunodeﬁciency disorder, therapy with high-dose corticosteroids, chemotherapy, or radiation with the past 90 days); had received a course of systemic antibiotic therapy during the 30 days prior to hospital admission for the study infection; or, was undergoing renal dialysis treatment (Zilberberg et al., 2010).
A total of 525 patients met our criteria for inclusion in this analysis; 36% had SSI, 28% had DFI, 20% had DSTA, and 16% had cellulitis (Table 1). The overall mean age was 52.6 years, 56% of the study population was male, and 74.3% were white. Of those enrolled, 118 (22.5%) received IIAT (Fig. 1). These patients, compared to those who received appropriate initial therapy, were signiﬁcantly older (56.6 versus 51.5 years, P = 0.004) but had no other signiﬁcantly different demographic characteristics. Of the patients receiving IIAT, 36 (30.5%) did not receive antibiotics within 24 hours of admission, and 82 (69.5%) did not receive antibiotics that were active against their infecting pathogen(s). The most frequent co-morbidity in the study population was diabetes (in 57.9%), but cancer other than non-melanoma skin cancer was the only co-morbidity for which there was a signiﬁcant difference between the 2 cohorts (2.2% in the IIAT patients and 7.9% in those who received appropriate initial antibiotic treatment, P = 0.020). Compared with patients receiving appropriate treatment, those receiving IIAT had signiﬁcantly higher rates of several HCA risk factors, including hospitalizations within the past 6 months (58.5% versus 46.0%, P = 0.017) and residence in or admission from a nursing home or assisted living facility (5.9% versus 1.5%, P = 0.006).
2.3. Data elements
3.2. Rate of IIAT
We obtained baseline characteristics on each enrolled patient, including their age, gender, ethnicity, race, infection type, and any active co-morbidities (i.e., chronic lung disease, diabetes mellitus, hepatic dysfunction, peripheral vascular disease, renal insufﬁciency,
As shown in Fig. 1, patients with DFI and SSI had the highest rates of IIAT (24.5% and 24.6%, respectively), while patients with DSTA had the lowest rate (17.1%); none of these differences in rates were statistically signiﬁcant. The rates of IIAT were higher among patients
3.1. Baseline characteristics
B.A. Lipsky et al. / Diagnostic Microbiology and Infectious Disease 79 (2014) 273–279
Table 1 Baseline characteristics by appropriateness of initial antibiotic treatment.
Age, mean (SD) Male Hispanic/Latino Race White Black/African American Asian American Indian/Alaska native Native Hawaiian/Paciﬁc Islander Admission from emergency room Infection type SSI DFI DSTA Cellulitis Active co-morbidities Chronic lung disease Diabetes mellitus Hepatic dysfunction Peripheral/vascular disease Renal insufﬁciency Cancer other than non-melanoma skin HCA risk factors Hospitalization within past 6 months Resident of or admission from nursing home or assisted living facility Immunosuppressed state Previous antibiotic use (past 30 days) Receiving dialysis Hospital characteristics Community location Urban Rural Hospital type Academic Community Region of US Midwest Northeast South West Number of beds 100–200 N200 Unknown
Overall (N = 525)
Appropriate Treatment (n = 407)
Inappropriate Treatment (n = 118)
P-value (appropriate versus inappropriate)
52.6 (15.8) 294 (56.0) 46 (8.8)
51.5 (15.2) 235 (57.7) 38 (9.3)
56.6 (17.1) 59 (50.0) 8 (6.8)
0.004 0.136 0.463 0.222
390 (74.3) 103 (19.6) 10 (1.9) 3 (0.6) 19 (3.6) 371 (70.7)
301 (74.0) 76 (18.7) 10 (2.5) 3 (0.7) 17 (4.2) 294 (72.2)
89 (75.4) 27 (22.9) 0 (0.0) 0 (0.0) 2 (1.7) 77 (65.3)
191 (36.4) 147 (28.0) 105 (20.0) 82 (15.6)
144 (35.4) 111 (27.3) 87 (21.4) 65 (16.0)
47 36 18 17
57 (10.8) 304 (57.9) 21 (4.0) 92 (17.5) 79 (15.0) 19 (3.6) 382 (72.8) 256 (48.8) 13 (2.5)
39 (9.7) 241 (59.1) 15 (3.6) 65 (15.9) 53 (13.1) 9 (2.2) 292 (71.7) 187 (46.0) 6 (1.5)
17 (14.0) 64 (54.4) 6 (5.3) 27 (22.8) 25 (21.1) 9 (7.9) 90 (76.3) 69 (58.5) 7 (5.9)
0.342 0.163 0.597 0.221 0.105 0.020 0.331 0.017 0.006
58 (11.1) 276 (52.6) 14 (2.7)
47 (11.6) 213 (52.3) 7 (1.7)
11 (9.3) 63 (53.4) 7 (5.9)
0.497 0.840 0.012
489 (93.1) 36 (6.9)
385 (94.6) 22 (5.4)
104 (88.1) 14 (11.9)
167 (31.8) 358 (68.2)
130 (31.9) 277 (68.1)
37 (31.4) 81 (68.6)
158 (30.1) 102 (19.4) 51 (9.7) 214 (40.8)
118 (29.0) 72 (17.7) 44 (10.8) 173 (42.5)
40 (33.9) 30 (25.4) 7 (5.9) 41 (34.8)
57 (10.9) 401 (76.4) 67 (12.8)
46 (11.3) 315 (77.4) 46 (11.3)
11 (9.3) 86 (72.9) 21 (17.8)
(39.8) (30.5) (15.3) (14.4)
Note: Data are number (%) of patients, unless otherwise indicated.
infected with mixed gram-positive and gram-negative organisms (42.2%) and with only gram-negative organisms (34.3%) when compared with patients infected with only gram-positive pathogens
(14.6%; for both comparisons P b 0.01). Rates of IIAT did not vary by type of facility (academic versus community) or region but were signiﬁcantly higher in rural than urban hospitals (38.9% versus 21.3%,
100% 80% 60%