Bloodstream infections cause significant morbidity and mortality worldwide and are among the most common healthcare-associated infections [1–6]. It is estimated that 2 million patients per year in the United States acquire infections while in hospitals, approximately 350,000 (10–20%) of these infections involve the bloodstream, and 90,000 (4.5%) are fatal [1, 6, 7]. Advances in medicine, efforts to control medical costs, and incentives for outpatient care have resulted in an increasingly concentrated population of seriously ill patients in hospitals. The incidence of bloodstream infections in patients treated in United States hospitals has been reported to correlate with increasing use of central venous catheters, patient illness (e.g., oncology, burn/trauma, and high-risk nursery), and other predisposing factors, including microorganism, intensive-care unit (ICU) stay, hand washing practices of medical staff, and adherence to infection control practices [1, 5, 6, 8]. Respiratory, genitourinary tract, and intra-abdominal foci are often identifiable sources of bloodstream infections [9]. Bacteremia due to Enterobacteriaceae other than Escherichia coli are associated with increased mortality compared with bloodstream infections due to Gram-positive species [5]. Gram-negative and polymicrobial bacteremia can result in septic shock and mortality is greater with high-grade bacteremia and polymicrobial infection [4, 5, 10]. Efforts need to be extended to prevent and control serious hospital-acquired infections.

The National Nosocomial Infections Surveillance (NNIS) system reported that from 1986 to 1997, coagulase-negative staphylococci and Staphylococcus aureus were the most common organisms isolated from blood cultures of intensive-care unit (ICU) patients, followed by Enterococcus spp., Candida albicans, and Enterobacter spp. [11]. However, only 50% of all positive blood cultures represent true bloodstream infection [5]. Importantly, although coagulase-negative staphylococci are the most frequently isolated organism from blood cultures, they are clinically significant <15% of the time [5]. Coagulase-negative bacteremia is often the result of long-term use of indwelling central and peripheral catheters as well as other prosthetic devices, the ubiquity of these bacteria as normal skin flora, and the ability of these relatively avirulent organisms to adhere to the surface of biomaterials [5]. Previous studies have reported that S. aureus and E. coli are the two most common, clinically significant causes of bloodstream infections in patients in the United States and Europe [4, 5, 12, 13] and that 6–18% of bloodstream infections are polymicrobial [4, 5]. Bacteremia may be transient or indicative of true systemic infection (i.e., sepsis syndrome) with an initial focal source such as the lungs (e.g., pneumonia) or the urinary tract [10].

The potential for antimicrobial resistance is one consideration for physicians when selecting a regimen with which to treat patients. This is particularly important for the treatment of systemic infections as initial antimicrobial chemotherapy is almost invariably empiric and must be based on knowledge of the most frequently isolated etiological agents and their antimicrobial susceptibility patterns. Early initiation of appropriate antimicrobial treatment is critical in decreasing morbidity and mortality among patients with bloodstream infections [14]. The current study reports the prevalence and antimicrobial susceptibility profiles of blood culture isolates from the United States using The Surveillance Network (TSN) Database-USA (Focus Technologies, Herndon, VA).

In the current study, results from the TSN Database-USA from January 1 to December 31, 2002 were used to estimate the prevalence of specific bacterial species as blood culture isolates in the United States and to determine rates of antimicrobial susceptibility for commonly tested agents among the most prevalent species identified. TSN assimilated antimicrobial susceptibility testing and patient demographic data from a network of 268 hospitals in the United States in 2002 [15]. All blood culture isolates were identified at the participating institutions by routine methods in use at each laboratory. Antimicrobial susceptibility testing of patient isolates was conducted onsite by each participating laboratory as a part of their routine diagnostic testing. An inpatient isolate was defined as such by each laboratory participating in TSN. Data from patients in nursing facilities and hospital outpatients were excluded from the current analysis.

Laboratories contributing to TSN databases are all nationally-accredited and are invited to participate in TSN based on factors such as hospital type (e.g., university teaching hospital, community hospital) and antimicrobial susceptibility testing method used as well as the bed size, patient population, and geographic location of the hospital(s) they serve [15]. Only data generated using nationally approved (Food and Drug Administration-approved) testing methods with MIC results interpreted according to NCCLS [16] recommendations are included in TSN Database-USA. In addition, a series of quality-control filters (proprietary critical rule sets) are used to screen susceptibility test results for patterns indicative of testing error; suspect results are removed from analysis for laboratory confirmation. TSN reflects current testing in United States laboratories and is the antimicrobial susceptibility testing data considered when clinical decisions in participating institutions are made. The TSN database presumes the evidence of infection, but no clinical correlates are applied universally. In TSN, any result from the same patient with the same organism identification and the same susceptibility pattern received within five days is considered a repeat culture and is counted only once in the database. In TSN, all isolates are not tested against all agents and variation can be observed for antimicrobial agents of the same class such as expanded-spectrum cephalosporins (ceftriaxone and cefotaxime) and fluoroquinolones (ciprofloxacin and levofloxacin) for which similar in vitro activities have been previously demonstrated.

The oxacillin-resistance rate was 49.3% among blood culture isolates of S. aureus and 76.7% among blood culture isolates of coagulase-negative staphylococci (data not shown). Among oxacillin-resistant coagulase-negative staphylococci and S. aureus, susceptibilities were 32.1–32.6% (range) and 7.3–7.6% for fluoroquinolones, respectively, 48.1% and 80.9% for gentamicin, 13.7% and 5.9% for erythromycin, 47.2% and 29.8% for clindamycin, 56.0% and 90.9% for trimethoprim-sulfamethoxazole, and 100% and 100% for vancomycin (data not shown).

For E. coli, ≥ 97% of isolates were susceptible to amikacin, cefepime, cefotaxime, ceftriaxone, and nitrofurantoin; ceftazidime non-susceptibility, a commonly used phenotypic marker for estimating extended-spectrum β-lactamase (ESBL) rates, was 3.8%. For K. pneumoniae, ≥ 90% of isolates were reported susceptible to amikacin, cefepime, cefotaxime, ceftriaxone, ciprofloxacin, levofloxacin, and gentamicin; ceftazidime non-susceptibility was 11.5%. Among P. aeruginosa, amikacin, piperacillin-tazobactam, and tobramycin had the highest rates of susceptibility; however, no agent had susceptibilities ≥ 93%.

S. aureus and E. coli were identified in previous studies as the two most common blood culture isolates from hospitalized patients in the United States and Europe [4, 5, 12, 13]. In the current study, coagulase-negative staphylococci were the most common blood culture isolates from laboratories in the United States (42.0% of isolates) (Table 1). However, given that coagulase-negative staphylococci isolated from blood cultures are often contaminants (>85% are clinically insignificant) [5] our results agree generally with those previously published. As TSN collects all laboratory data, year-round, it may present a more accurate description of laboratory testing than do centralized point prevalence studies that often exclude the majority of isolates identified by laboratories in a year [12, 13]. Accepting the over-representation of contaminant coagulase-negative staphylococci in clinical laboratories in the United States as observed in TSN Database-USA, the rank order of other pathogens is similar to previous reports describing centralized surveillance studies [12, 13] and hospital review studies [4, 5]. Six organisms, coagulase-negative staphylococci, S. aureus, E. faecalis, E. coli, K. pneumoniae, and E. faecium accounted for >80% of blood culture isolates. Previously, SENTRY has reported similar results for laboratories in the United States, Canada, Latin America, and Europe [12, 13]. In the current study, overall frequencies of isolation were 78.1% gram-positive bacteria and 21.9% gram-negative bacteria.

Oxacillin-resistant S. aureus are extremely important causes of bloodstream infections and evidence has been presented that oxacillin-resistant S. aureus (Table 3) are increasing globally among bloodstream isolates and among isolates from other anatomical sites [12, 17]. Fluoroquinolone resistance has increased in a consistent step-wise manner in the United States and Europe for Enterobacteriaceae, P. aeruginosa, and S. aureus [12, 13, 18, 19]. It is important for clinicians to be updated with current data concerning the susceptibility of commonly prescribed agents such as the fluoroquinolones and also to be aware of trends in longitudinal data. The rates of change in resistance by pathogen and region can help set priorities for focused intervention efforts.

Early clinical suspicion, rigorous diagnostic measures, aggressive initiation of appropriate antimicrobial therapy, comprehensive supportive care, and measures aimed at reversing predisposing causes (e.g., amelioration of an underlying disease, removal of foreign bodies, drainage of abscesses) are the cornerstone of successful management of patients with sepsis syndrome [5, 10]. The selection of antimicrobials to be used for empiric therapy should be based on the local rates of susceptibility and on the site of infection [10]. Early initiation of appropriate antimicrobial treatment is critical in decreasing morbidity and mortality among patients with bloodstream infections due to gram-negative organisms [14]. The initiation of such therapy is almost always empirical, requiring knowledge of the likely pathogen(s) and their usual antimicrobial susceptibility patterns [10, 20]. Combinations of antimicrobial agents are recommended for empiric therapy for patients with bloodstream infections, particularly for those patients with the most adverse prognostic factors [10]. Combination therapy is recommended to cover the broad range of possible pathogens which may be difficult to distinguish clinically, because of the possibility of polymicrobial infections, because they may prevent the emergence of resistance, and because they may have additive or synergistic antimicrobial activity. For the patient with a nosocomial bloodstream infection, initial treatment should consist of an aminoglycoside initially paired with a broad-spectrum β-lactam. Expanded-spectrum cephalosporins are the β-lactam of choice for the non-neutropenic patient because of the greater likelihood of Klebsiella and Staphylococcus infections in these patients [10]. The regimen of an aminoglycoside paired with a penicillin or cephalosporin having antipseudomonal activity is preferred for neutropenic patients, patients with severe chronic obstructive pulmonary disease or bronchiestatis, patients receiving assisted ventilation, and patients with extensive burns [10].

The in vitro potency of ceftriaxone and cefotaxime against E. coli and Klebsiella suggests that single-agent therapy directed against those bacteria may be successful even in severely compromised hosts [10]. The superior pharmacokinetic and pharmacodynamic properties that exist for ceftriaxone when compared with cefotaxime may be a consideration when choosing between these two agents [18]. In the nosocomial setting, extensive data also confirmed the efficacy of ceftriaxone with or without an aminoglycoside in serious Gram-negative infections, pneumonia, spontaneous bacterial peritonitis and as surgical prophylaxis [21]. Ceftriaxone, cefotaxime, and cefepime all have similar indications for pneumonia, skin and skin structure infections, and urinary tract infections; however only ceftriaxone and cefotaxime have an indication for the sepsis syndrome. In the current study, susceptibilities of isolates to one or both agents in combinations of ceftriaxone, ceftazdime, cefepime, piperacillin-tazobactam or ciprofloxacin plus gentamicin were consistent (range, 74.8–76.3%) but lower than similar β-lactam or ciprofloxacin combinations with vancomycin (range, 93.5–96.6%). Ceftazidime, cefepime, imipenem and meropenem appear most active against P. aeruginosa [10].

In conclusion, susceptibilities to some classes of antimicrobials are decreasing, most notably the fluoroquinolones for Enterobacteriaceae and P. aeruginosa. Many older antimicrobials including ceftriaxone continue to retain high rates of susceptibility against many important bacterial pathogens such as those commonly isolated from blood cultures. Against the most clinically important gram-positive species including pneumococci, and gram-negative bacilli such as E. coli (Table 2) susceptibility to ceftriaxone appears to have changed little, if at all, from 1996 to 2002 [18]. While selective pressure for resistance through antimicrobial use is important, infection control practices are critical to limiting the spread of resistant organisms. The life-threatening nature of bacteremia and sepsis underscores the importance of using timely surveillance data to develop rational antimicrobial therapy recommendations and to design strategies to help control antimicrobial resistance [10, 22].