|Home | About | Journals | Submit | Contact Us | Français|
These updated guidelines replace the previous management guidelines published in 2001. The guidelines are intended for use by health care providers who care for patients who either have these infections or may be at risk for them.
Recommendations related to the unique aspects of the following subjects may also be found in the text: treating short-term peripheral venous catheters, nontunneled and long-term CVCs, implanted catheter–related infections (other than infections related to hemodialysis catheters), treatment of pediatric patients with catheter-related infections, and treatment of infections related to hemodialysis catheters. Recommendations are also made regarding antibiotic lock therapy, pathogen-specific treatment, management of suppurative thrombophlebitis, management of persistent bloodstream infection, and detection and management of an outbreak of CRBSI. A full listing of all recommendations may be found in table 1.
In 2001, the IDSA published a clinical practice guideline on the management of intravascular catheter-related infection . IDSA updates its guidelines when new data or publications might change a prior recommendation or when the Expert Panel feels clarifications or additional guidance are warranted. For the 2009 Update, the indications for treatment and agents of choice from the 2001 guideline were reviewed . The previous document is a source for a more detailed review of earlier studies.
The Expert Panel addressed the following clinical questions in the 2009 Update:
“Practice guidelines are systematically developed statements to assist practitioners and patients in making decisions about appropriate health care for specific clinical circumstances” . Attributes of good guidelines include validity, reliability, reproducibility, clinical applicability, clinical flexibility, clarity, multidisciplinary process, review of evidence, and documentation .
The IDSA Standards and Practice Guidelines Committee convened a multidisciplinary panel of experts in the management of intravascular catheter-related infections. Expert Panel participants included representatives from the following collaborating organizations: European Society of Clinical Microbiology and Infectious Diseases, Pediatric Infectious Diseases Society, American Society of Nephrology, Society for Critical Care Medicine, and the Society for Health-care Epidemiology of America.
For the 2009 update, the Expert Panel completed the review and analysis of data published from 2001 through June 2008. Data published after June 2008 were also considered in the final preparation of the guideline. Computerized literature searches of the PubMed database were performed with combinations of the following search terms: “catheter-related,” “infections,” “cultures,” “management,” “treatment,” “peripheral,” “non-tunneled,” “central venous catheter,” “arterial catheter,” “implanted catheter,” “pediatric,” ”hemodialysis,” “antibiotic lock,” “bacteremia” “suppurative thrombophlebitis,” “endocarditis,” and “outbreak.”
In evaluating the evidence regarding the management of intravascular catheter–related infections, the Expert Panel followed a process used in the development of other IDSA guidelines. The process included a systematic weighting of the quality of the evidence and the grade of recommendation (table 2) .
The Expert Panel met face-to-face on 1 occasion and via teleconference on 8 occasions to complete the work of the guideline. The purpose of the meetings was to discuss the questions to be addressed, make writing assignments, and discuss recommendations. All members of the Expert Panel participated in the preparation and review of the draft guideline. Feedback from external peer reviewers was obtained. All collaborating organizations were also asked to provide feedback and endorse the guidelines. The following organizations endorsed the guidelines: American Society of Nephrology (pending), European Society of Clinical Microbiology and Infectious Diseases, Pediatric Infectious Diseases Society, Society for Critical Care Medicine (pending) and the Society for Healthcare Epidemiology of America (pending). The guideline was reviewed and approved by the IDSA Standards and Practice Guidelines Committee and the Board of Directors prior to dissemination.
All members of the Expert Panel complied with the IDSA policy on potential conflicts of interest, which requires disclosure of any financial or other interest that might be construed as constituting an actual, potential, or apparent conflict. Members of the Expert Panel were provided with the IDSA’s conflict of interest disclosure statement and were asked to identify ties to companies developing products that might be affected by promulgation of the guideline. Information was requested regarding employment, consultancies, stock ownership, honoraria, research funding, expert testimony, and membership on company advisory committees. The Expert Panel made decisions on a case-by-case basis as to whether an individual’s role should be limited as a result of a conflict. Potential conflicts of interest are listed in the Acknowledgements section.
At annual intervals, the Expert Panel Chair, the Standards and Practice Guidelines Committee liaison advisor, and the Chair of the Standards and Practice Guidelines Committee will determine the need for revisions to the guideline on the basis of an examination of current literature. If necessary, the entire Expert Panel will be reconvened to discuss potential changes. When appropriate, the Expert Panel will recommend revision of the guideline to the Standards and Practice Guidelines Committee and the IDSA Board for review and approval.
Each year in the United States, hospitals and clinics purchase >150 million intravascular devices to administer intravenous fluids, medications, blood products, and parenteral nutrition fluids, to monitor hemodynamic status, and to provide hemodialysis . Different types of intravascular catheters are currently being used (table 3), leading to a myriad of infectious complications (table 4). The focus of these guidelines is on the management of such complications, particularly CRBSI. In the United States, ~80,000 CVC-related bloodstream infections occur in intensive care units each year . In addition, the risk of bloodstream infection varies according to the intravascular device , the type of and intended use for the catheter, the insertion site, the experience and education of the individual who installs the catheter, the frequency with which the catheter is accessed, the duration of catheter placement, the characteristics of the catheterized patient, and the use of proven preventative strategies [7, 8]. For the purpose of this guideline, short-term catheters are defined as those devices that are in situ for <14 days.
Most CRBSIs emanate from the insertion site, hub, or both . For long-term catheters—particularly tunneled catheters—the catheter hub is a prominent source of microbes causing bloodstream infection . In order of prevalence, the 4 groups of microbes that most commonly cause CRBSI associated with percutaneously inserted, noncuffed catheters are as follows: co-agulase-negative staphylococci, S. aureus, Candida species, and enteric gram-negative bacilli. For surgically implanted catheters and peripherally inserted CVCs, they are coagulase-negative staphylococci, enteric gram-negative bacilli, S. aureus, and P. aeruginosa .
Clinical findings are unreliable for establishing the diagnosis of intravascular device–related infection because of their poor sensitivity and specificity. The most sensitive clinical finding, fever, has poor specificity. Inflammation or purulence around the insertion site has greater specificity but poor sensitivity [4, 15]. Blood cultures that are positive for S. aureus, coagulase-negative staphylococci, or Candida species, in the absence of other identifiable sources of infection, should increase the suspicion for CRBSI [16–18]. Improved symptomatology within 24 h after catheter removal suggests but does not prove that the catheter is the source of infection .
Laboratory criteria for diagnosing intravascular catheter-related infections are precise, but differences in definitions and methodologies among various studies have made data difficult to compare [4, 18]. When a catheter segment is submitted for culture, it is adequate to culture only the catheter tip and not the subcutaneous portion of the catheter . If a pulmonary artery catheter is removed because of suspected infection, the highest yield is to culture the introducer, rather than the catheter itself . Semiquantitative (roll plate) or quantitative catheter culture techniques (luminal flushing or sonication methods) are the most reliable diagnostic methodologies and have much greater specificity than broth cultures [22–25]. A recently inserted catheter (i.e., one that had been indwelling for <14 days) is most commonly colonized from a skin microorganism along the external surface of the catheter. Thus, the roll-plate method has high sensitivity. Intraluminal spread of microbes from the catheter hub into the bloodstream is increasingly important for long-term catheters (i.e., those that have been indwelling ≥14 days). In some studies, the roll-plate method was less sensitive than other methods that also sampled the internal surface of such catheters [10, 26], but other studies have not found this to be the case . For subcutaneous ports, culture of the material inside the port reservoir is more sensitive than catheter tip culture for the diagnosis of CRBSI [28–30].
Antimicrobial coatings may lead to false-negative culture results [31, 32]. For silver sulfadiazine– or chlorhexidine-coated catheters, specific inhibitors can abrogate this effect, but this is not the case for minocycline- or rifampin-coated catheters [31, 32]. The specific components of the inhibitor solution to be used when culturing silver sulfadiazine or chlorhexidine can be found elsewhere .
Various methods have been used to diagnose a catheter-related infection without catheter removal. In one method, a moist cotton swab can be used to do a semiquantitative culture of a 3-cm radius around the catheter insertion site, and alginate swabs can be used to sample the inner surface of each catheter hub (1 swab per hub). Swab samples are streaked on blood agar plates. Growth of >15 cfu/plate of the same microbe from the insertion site swab sample and hub swab sample cultures and from a peripheral blood culture suggests CRBSI . This approach also has good negative predictive value for CRBSI when <15 cfu/plate are detected on insertion site and hub swab sample cultures.
Although catheter colonization with accompanying systemic signs of infection suggests catheter-related infection, a definitive diagnosis of CRBSI requires positive percutaneous blood culture results with concordant microbial growth from the catheter tip or catheter-drawn cultures that meet the above-described quantitative culture or DTP criteria. The accuracy of all diagnostic microbiologic methods greatly increases with increasing pretest probability. Thus, diagnostic tests for vascular catheter-related infection should not be done unless there is a high index of suspicion. Overall, quantitative blood cultures are the most accurate method by which to diagnose CRBSI [34, 35]. No single test is clearly superior for short-term CRBSI diagnosis. For diagnosis of CRBSI in patients with long-term catheters, quantitative blood cultures are the most accurate test, but DTP also has a high degree of accuracy. Neither method requires catheter removal. If a blood sample for culture cannot be obtained from a peripheral vein, ≥2 catheter blood samples for culture should be drawn through different catheter lumens ; however, it is unclear whether blood samples should be drawn through all catheter lumens in such circumstances.
It is important to remember that the definition of CRBSI used in the current document, which deals with management of infections related to intravascular devices, differs from surveillance definitions used to define central line–associated bloodstream infection .
Contamination rates are lower if a dedicated phlebotomy team collects the blood samples for culture . Skin preparation with either alcohol, alcoholic chlorhexidine (>0.5%), or tincture of iodine (10%) leads to lower blood culture contamination rates than does the use of povidone-iodine [39, 40]. Contamination rates among blood samples obtained through newly inserted intravenous catheters are higher than contamination rates among blood samples obtained from peripheral veins [41, 42]. Blood samples obtained through catheters that are in use are associated with a higher rate of false-positive results, compared with cultures of percutaneous blood samples . Thus, there is higher specificity and a greater positive predictive value when blood samples are obtained from a peripheral vein for culture, compared with when blood samples are obtained through catheters for culture [44, 45]. Negative predictive values are excellent for cultures of blood samples obtained from either a peripheral vein or a catheter.
DTP uses continuous blood culture monitoring for growth (e.g., radiometric methods) and compares the DTP for qualitative blood culture samples obtained from the catheter and from a peripheral vein. The greater the inoculum of microbes inoculated into blood culture bottles, the shorter the DTP .
When studied among patients with cancer and patients hospitalized in intensive care units who had both long-term and short-term catheters, this method has been shown to have accuracy comparable to that of quantitative blood cultures, as well as greater cost-effectiveness [35, 47–49]. Most microbiology laboratories do not perform quantitative blood cultures, but many laboratories will be able to determine DTP. DTP may not discriminate between CRBSI and non-CRBSI for patients who are already receiving antibiotics .
PCR to target bacterial 16S ribosomal DNA is sensitive and specific for diagnosing catheter-related infection but is not routinely used in clinical microbiology laboratories .
Antibiotic therapy for catheter-related infection is often initiated empirically. The initial choice of antibiotics will depend on the severity of the patient’s clinical disease, the risk factors for infection, and the likely pathogens associated with the specific intravascular device (figure 3 and table 4). In the largest published comparative trial of CRBSI treatment involving antimicrobial therapy and catheter removal, 88% of 169 patients had a successful microbiologic outcome when evaluated 1–2 weeks after the end of treatment, and there was an 83% microbiologic success rate among 98 cases of CRBSI due to S. aureus . Coagulase-negative staphylococci are the most common cause of catheter-related infection. Most of these pathogens exhibit methicillin resistance, and this should be considered when choosing empirical therapy for catheter-related infection [53, 54]. Vancomycin is associated with a lower clinical success rate in treating MRSA bacteremia if the MIC is ≥2 μg/mL [55, 56]. Standardized treatment advice can be formulated for each CRBSI on the basis of these guidelines. When such standardized treatment advice is automatically delivered to treating physicians, compliance with the guidelines increases significantly .
There are no compelling data to support specific recommendations for the duration of therapy for device-related infection. However, the Expert Panel’s recommendations are presented in figures 1–4. Management of CRBSI should be distinguished on the basis of the removal or retention of the catheter, and a distinction should be made between complicated CRBSI, in which there is suppurative thrombophlebitis, endocarditis, osteomyelitis, or possible metastatic seeding, and uncomplicated CRBSI (figures 1–4). Intravenous administration of thrombolytic agents, such as urokinase, should not be used as adjunctive treatment for CRBSI [58, 59].
Diagnosis and management of illness among patients with a nontunneled CVC or arterial catheter and unexplained fever are summarized in table 1 and figures 1 and and3.3. CVCs in patients with fever and mild-to-moderate disease should not routinely be removed, because the majority of the catheters from patients with suspected catheter-related infection are sterile . Recent studies suggest that the risk of arterial CRBSI approaches that associated with short-term CVCs [63–65].
One study found that 1 in 4 patients with S. aureus colonization of an intravascular catheter subsequently developed S. aureus bacteremia if they did not receive immediate anti-staphylococcal antibiotics . Similarly, other studies have found that S. aureus and Candida catheter colonization, compared with catheter colonization due to enterococci or gram-negative bacilli, was more likely to be associated with CRBSI and that CRBSIs due to S. aureus and Candida species were more likely to be associated with complications than CRBSIs due to enterococci or gram-negative bacilli [26, 67].
The diagnostic evaluation for new onset of fever in patients hospitalized in the intensive care unit is a daily problem for intensive care physicians . New onset of fever often leads to the removal of intravascular catheters and the reinsertion of new catheters over a guidewire or into another site. However, few of these patients have a catheter-related infection [33, 50, 69]. For hemodynamically stable patients without documented bloodstream infection and without prosthetic valves, pacemakers, or recently placed vascular grafts, systematic catheter removal may not be necessary for new onset of fever. Catheter removal only when bloodstream infection is documented or when there is hemodynamic instability reduces unnecessary catheter removal . Nevertheless, if a catheter is to be removed for suspected catheter-related infection and the patient is at high risk for mechanical complications during catheter reinsertion, a guidewire exchange of the catheter can decrease the risk of mechanical complications . The tip of the removed catheter should be sent for culture. If the tip has positive culture results, this newly inserted catheter should be replaced a second time, because bacterial contamination of the newly inserted catheter can be expected.
Surgically implantable intravascular devices consist of either a tunneled silicone catheter (e.g., Hickman, Broviac, or Groshong catheters) or a subcutaneously implanted port reservoir (e.g., Port-A-Cath). Because the removal of such devices is often a management challenge, it is important to be sure that one is dealing with true CRBSI rather than with contaminated blood cultures (e.g., contaminated due to coagulase-negative staphylococci), catheter colonization without concomitant bloodstream infection, or fever from another source (figures 3 and and4).4). Microbiologic data suggestive of true CRBSI caused by potential skin flora rather than contamination include the following: multiple blood samples with positive culture results obtained from different sites; quantitative blood cultures performed on samples drawn from a catheter with growth of >15 cfu/mL of blood or isolation of the same organism from a catheter culture and a percutaneous blood culture, especially if a culture performed on blood drawn from the catheter shows growth at least 2 h earlier than a culture performed on blood drawn from a peripheral vein . Although several studies suggest that catheter exchange over a guidewire can be used successfully to manage CRBSI associated with long-term catheters , most of these were small, uncontrolled studies with poor definitions, and none of these studies used antimicrobial catheters as a replacement for the infected catheter [73–77]. Management of CRBSI for patients with a long-term CVC or implantable device is summarized in tables 5 and 6 and in figure 2.
The pediatric population is diverse, and the probability of infection varies with patient risk factors, the type and location of the device, and the nature of the infusate [78, 79]. Among premature infants, birth weight is inversely proportional to the risk of infection, with infants who have extremely low birth weight (1000–1500 g) having an increased risk, compared with infants who have very low birth weight . Most nosocomial bloodstream infections among pediatric patients are related to the use of an intravascular device , and in critically ill neonates, the incidence of CRBSI can be as high as 18 cases per 1000 catheter-days . Most CRBSIs among children are caused by coagulase-negative staphylococci (which account for 34% of cases), followed by S. aureus (25%) . Among neonates, coagulase-negative staphylococci account for 51% of CRBSIs, followed by Candida species, enterococci, and gram-negative bacilli [78, 84]. Infants with short-gut syndrome, a disorder that is clinically defined by malabsorption, diarrhea, steatorrhea, fluid and electrolyte disturbances, and malnutrition, often resulting from anatomic removal of bowel during the newborn period due to necrotizing enterocolitis, are more likely to have CRBSI due to gram-negative bacilli .
Several problems arise when the clinical and laboratory definitions of infection established for adults are applied to children [18, 86]. Although specific pediatric blood culture devices are commercially available, difficulty in obtaining blood samples and concerns about drawing a large volume of blood may result in lower volumes of blood being submitted for culture, which would reduce the negative predictive value of the culture. Often, only results from blood samples obtained via the catheter are available to guide patient management. Peripheral cultures are not often performed when catheter cultures are obtained, because venipuncture can be difficult for infants and young children. A recent study suggests that, among pediatric oncology patients with a double lumen CVC, catheter-related infection can be diagnosed by a ≥5-fold difference in colony count between the 2 lumens; this method has 62% sensitivity, 93% specificity, and 92% positive predictive value, compared with a comparison between the colony count for 1 lumen and for a peripheral blood sample . However, validation in a prospective study is needed to confirm these findings. In addition, placement of catheters or changing a catheter over a guidewire is difficult for young children, and catheter removal for diagnostic purposes is often not done out of concern over losing access (figure 3). Because of these limitations, definitive CRBSI can often not be diagnosed in children. In these circumstances, many physicians treat their patients as if they had presumed CRBSI.
Although indications for catheter removal among children should follow the recommendations for adults, because of greater vascular access difficulties in children, it is often necessary to attempt CRBSI treatment without catheter removal. Several studies have reported successful CRBSI management among children without catheter removal [88–90]. Such children should be closely monitored, and the device should be removed in the event of clinical deterioration or recurrence of CRBSI. In contrast, treatment of catheter-associated fungemia without removal of the catheter has a low success rate and is associated with higher mortality [91, 92]. Recent reports involving children with Candida CRBSI found that the addition of antifungal lock therapy led to a high cure rate without catheter removal, but there are insufficient data to recommend routine catheter salvage using this approach for this infection unless there are unusual extenuating circumstances (e.g., no alternative catheter insertion site) [93–95].
Antimicrobial agents that are appropriate for infants and children and the recommended dosages of specific agents, by patient age and weight, are summarized in table 6. Antibiotics should be administered through the catheter. In contrast to the recommendation for adults, empirical antifungal coverage in critically ill patients with femoral catheters is not recommended for children. Antifungal therapy should be initiated when yeast is isolated from a blood culture or when the suspicion of fungemia is high [90, 96–98]. The selection of an appropriate antifungal agent depends on the organism that is isolated and the drug characteristics, including available pediatric dosing information, toxicities, route of administration, and formulations.
Conventional treatment for CRBSI has not been established to be different from that previously described for adults (table 6 and figures 1–4), but certain procedures may not apply to infants and young children. For example, as noted in figures 1 and and2,2, echocardiographic examination is not used commonly for small infants and children with CRBSI who do not have other indicators of endocarditis. The optimal duration of therapy has not been established for treating catheter-related infection in children with or without catheter removal [89, 90]. Therefore, recommendations regarding the duration of therapy for pediatric patients with CRBSI should mirror adult recommendations. Lastly, antibiotic lock therapy should also be used, with the recognition that dwell times may be variable, based on limited venous access and the necessity to use the catheter.
It is frequently not feasible to obtain a peripheral blood sample for culture from patients who are receiving dialysis . In some patients, the peripheral veins have been exhausted as a result of multiple failed dialysis fistulas or grafts. Moreover, it is important to avoid drawing blood from peripheral veins that will be used for future creation of a fistula or graft, because venipuncture may injure the vein.
A substantial proportion of patients who receive dialysis who have CRBSI are treated successfully in the outpatient setting. Hospitalization is only indicated for patients with severe sepsis or metastatic infection. CRBSI in patients who are undergoing hemodialysis has several unique features that may dictate differences in their management, compared with that of other patients (tables 1 and and77 and figure 2).
CRBSI in patients who are undergoing hemodialysis may be caused by several different pathogens, but such cases are most often due to coagulase-negative staphylococci or S. aureus [99–101]. If possible, antibiotic selection should be made on the basis of pharmacokinetic characteristics that permit dosing after each dialysis session (e.g., vancomycin, ceftazidime, or cefazolin) , or antibiotics that are unaffected by dialysis (e.g., ceftriaxone) should be used. The majority of gram-negative organisms causing CRBSI in patients who are undergoing hemodialysis are susceptible to both aminoglycosides and third-or fourth-generation cephalosporins [99, 100], but cephalosporins are preferred, because aminoglycosides carry a substantial risk of inducing irreversible ototoxicity . Validated dosing schedules for cefazolin and vancomycin to ensure therapeutic concentrations have been published (table 8) [104, 105].
Among patients who are undergoing hemodialysis who have CRBSI involving long-term catheters, not only is the catheter the source of the infection, but it is also the vascular access for providing ongoing dialysis. The 4 potential treatment options for such patients are (1) intravenous antibiotics alone, (2) prompt catheter removal with delayed placement of a new long-term catheter, (3) exchange of the infected catheter with a new one over a guidewire, or (4) use of systemic antibiotics and an antibiotic lock in the existing catheter (figure 5) . Administration of intravenous antibiotics alone is not a satisfactory approach, because bloodstream infection recurs in the majority of patients once the course of antibiotics has been completed [101, 106–109]. Moreover, the risk of treatment failure among patients who are undergoing hemodialysis whose CRBSI is treated with antibiotics alone is a 5-fold higher, compared with the risk among patients who undergo catheter removal . For patients whose symptoms resolve after 2–3 days of intravenous antibiotic therapy and who do not have evidence of metastatic infection, guidewire exchange of the catheter is associated with cure rates that are comparable to those associated with immediate removal and delayed placement of a new catheter [74–76, 111, 112]. Patients with hemodialysis-associated CRBSI due to gram-negative pathogens or CRBSI due to coagulase-negative staphylococci may have the catheter retained and be treated with adjunctive anitibiotic lock therapy for 3 weeks, or they may have the catheter exchanged over a guidewire and then receive the same antibiotic course (figure 2).
For antibiotic lock therapy, the antibiotic is combined with heparin and instilled into each catheter lumen at the end of each dialysis session (table 9) [99, 113, 114]. The success rate is 87%–100% for infection due to gram-negative pathogens and 75%– 84% for infection due to Staphylococcus epidermidis, but it is only 40%–55% for hemodialysis-associated CRBSI due to S. aureus [100, 114, 115].
Antibiotic lock therapy for CRBSI is used in conjunction with systemic antibiotic therapy and involves installing a high concentration of an antibiotic to which the causative microbe is susceptible in the catheter lumen. In 14 open trials of CRBSI involving long-term catheters with catheter retention and administration of standard parenteral therapy without the adjunctive use of antibiotic lock therapy, the average success rate was 67%. The likelihood of success varies with the site of infection (e.g., tunnel or pocket infection are unresponsive to salvage) and with the microbe causing the infection (e.g., co-agulase-negative staphylococci are likely to respond; S. aureus and P. aeruginosa are not). Recurrent bacteremia after parenteral therapy is more likely to occur if that therapy is administered through a retained catheter than if the catheter is removed . This likely reflects the inability of most antibiotics to achieve therapeutic concentrations needed to kill microbes growing in a biofilm [117–122]. Antibiotic concentrations must be 100 to 1000 times greater to kill sessile bacteria within a biofilm than to kill planktonic bacteria [117–122]. Because the majority of infections involving long-term catheters or totally implanted catheters are intraluminal, eradication of such infections is attempted by filling the catheter lumen with supratherapeutic concentrations of antibiotics and leaving them indwelling for hours or days, thereby creating an antibiotic lock. In 21 open trials of antibiotic lock therapy for CRBSI involving long-term catheters, with or without concomitant parenteral therapy, catheter salvage without relapse occurred in 77% of episodes. Two controlled clinical trials of the use of antibiotic lock therapy together included only 92 patients, and treatment was successful in 58% of the control subjects and 75% of the patients treated with an antibiotic lock [123, 124]. Compared with bacterial infection, Candida CRBSI is more difficult to eradicate with antibiotic lock therapy [93, 125–127]. In the largest published series on the use of antibiotic lock for CRBSI due to S. aureus, treatment failure was observed in one-half of the cases . Antibiotic lock solutions contain the desired antimicrobial concentration (table 9), and they are usually mixed with 50–100 units of heparin or normal saline in sufficient volume to fill the catheter lumen (usually 2–5 mL). Rapidly decreasing antibiotic concentrations may occur over time in the distal lumen of hemodialysis catheters instilled with an antibiotic lock, especially among ambulatory patients with femoral catheters . Thus, to maintain a concentration of vancomycin that is >1000 times the MIC90 of staphylococci during the entire dwell time, a concentration of 5 mg/mL is preferred, and the antibiotic lock fluid should be changed at least every 48 h.
Although the duration of antibiotic lock therapy has varied substantially among different studies (3–30 days), most studies have used a 2-week duration. Vancomycin, cefazolin, and ceftazidime remain stable in heparin solutions at 25°C and 37°C for several days . Not all antibiotic-heparin combinations can be used, because precipitation occurs when some antibiotics are mixed with heparin, especially with increasing antibiotic concentrations . Table 9 lists antibiotic lock solutions that can be used without the risk of precipitation.
The use of an antibiotic lock does not obviate the need for systemic antimicrobial therapy. However, when blood cultures have become negative and signs of sepsis have resolved, systemic antimicrobial therapy can be given orally in some patients. The combination of an orally administered, well-absorbed antibiotic (clindamycin, doxycycline, a fluoroquinonolone, or linezolid) and an antibiotic lock that can be left in place for 24–48 h may be more practical in some cases of outpatient management of CRBSI with coagulase-negative staphylococci .
Catheters that are in place for <2 weeks are most often infected extraluminally, and patients with catheters in place for longer periods may also have evidence of extraluminal infection . Antibiotic lock therapy is unlikely to have any impact on extraluminal infection.
Other antimicrobial locks are being evaluated for the treatment of CRBSI. One pediatric CRBSI study had a high success rate using a 70% ethanol antimicrobial lock .
On occasion, symptomatic patients with catheters have multiple catheter-drawn blood cultures that are positive for co-agulase-negative staphylococci or, more rarely, gram-negative bacilli, but also have concurrent percutaneous blood cultures with negative results. Such patients can be considered to have an intraluminally colonized catheter. If these colonized catheters are left in place, patients may go on to develop a true CRBSI. Therefore, if such catheters cannot be removed, antibiotic lock therapy without systemic therapy can be given through the retained catheter.
Coagulase-negative staphylococci are the most common cause of catheter-related infection. Most patients have a benign clinical course; rarely, patients develop sepsis with a poor outcome. For example, S. lugdunensis is an uncommon cause of catheter-related infection; however, it can cause endocarditis and metastatic infections similar to those caused by S. aureus .
The interpretation of blood cultures positive for coagulase-negative staphylococci remains problematic, because they are the most common contaminant and, at the same time, they are the most common cause of CRBSI. A high proportion of positive blood cultures performed on samples drawn from multiple sites remains the best indication for true CRBSI due to coagulase-negative staphylococci [17, 133].
No randomized trials have evaluated the treatment of co-agulase-negative staphylococcal CRBSI. Such infections may resolve with removal of the catheter without antibiotic therapy, and some experts recommend that no antibiotic therapy be administered to patients without endovascular hardware unless fever and/or bacteremia persist after catheter withdrawal. Other experts recommend that such infections be treated with antibiotics (table 4). Specific management strategies for coagulase-negative staphylococcal infection associated with different catheters and devices are summarized in table and figures 1–4.
There are no data from randomized trials with adequate sample size to determine the optimal duration for the treatment of S. aureus CRBSI. Traditionally, S. aureus bacteremia has been treated with a 4-week course of therapy because of concern about the risk of infective endocarditis [134, 135]. However, several studies have suggested that the risk of infective endocarditis or other deep tissue infection related to S. aureus bacteremia may be sufficiently low among selected patients with uncomplicated CRBSI to recommend a shorter course of therapy (i.e., ≥14 days) [136–140]. Identifying patients without risk factors for hematogenous complications and pursing an aggressive evaluation which may include TEE is important before proceeding to short-course therapy .
Many patients (25%–30%) with S. aureus bacteremia will have hematogenous complications, including cardiac or musculoskeletal involvement [142–146]. Clinical identifiers can be helpful in determining which patients with S. aureus bacteremia have a complicated infection [143, 144, 146]. One of the most consistent predictors of hematogenous complications is positive blood culture results 72 h after initiation of appropriate antimicrobial therapy and catheter removal [143–146]. Additional predictors of hematogenous complications include community-acquired infection and skin changes consistent with septic emboli [143, 144]. Failure or delay in removing the catheter increases the risk for hematogenous complications . In addition, removal of vascular catheters infected with S. aureus has been associated with a more rapid response to therapy and/or a higher cure rate, compared with catheter retention [139, 144, 147, 148].
Patients with S. aureus CRBSI have a significantly higher risk of hematogenous complications if they have a retained foreign body, if they are hemodialysis-dependent, if they have AIDS, or if they are diabetic or receiving immunosuppressive medications . For this reason, a longer course of therapy is prudent for immunosuppressed patients with S. aureus CRBSI.
Many cases of infective endocarditis are not suspected clinically and are therefore not detected . Studies using TEE to identify infective endocarditis among patients with S. aureus bacteremia have shown high rates of valvular vegetations (25%–32%) [142, 150, 151]. TEE is superior to transthoracic echocardiography in detecting valvular vegetations . Additionally, TEE is most sensitive when performed 5–7 days after the onset of bacteremia .
A combination of antibiotic lock therapy and systemic therapy has been used to salvage infected ports and long-term (e.g., hemodialysis) catheters for some patients with S. aureus CRBSI [99, 107, 153]. Although some catheters without evidence of exit site infection or tunnel infection may be salvaged, most patients with S. aureus CRBSIs eventually experience relapse and require removal of the catheter [99, 107].
Patients with catheters that are colonized with S. aureus who are not bacteremic are at risk for subsequent S. aureus bacteremia [66, 154], and administering antistaphylococcal therapy within 24 h after removal of the catheter may reduce the likelihood that the patient will develop bacteremia.
In the largest randomized, published study assessing treatment of CRBSI in adults, a group of patients who received linezolid was compared with a control group that received non–weight-based vancomycin (for MRSA infection) or oxacillin (2 g every 6 h) or dicloxacillin (500 mg orally every 6 h); for suspected gram-negative bacteremia, aztreonam or amikacin was recommended . The rate of successful microbiologic outcome at test of cure for patients with methicillin-susceptible S. aureus CRBSI was 82% for the linezolid group and 83% for the control group (95% confidence interval [CI], −16 to 14) ; for patients with MRSA CRBSI, it was 81% for the linezolid group and 86% for the control group (95% CI, −26 to 16). The rate of successful clinical outcome at test of cure for patients with methicillin-susceptible S. aureus CRBSI was 67% in the linezolid group and 67% in the control group (95% CI, −19 to 19); for patients with MRSA CRBSI, it was 79% in the linezolid group and 76% in the control group (95% CI, −21 to 27). Kaplan-Meier survival curves for intention-to-treat populations found that there was no statistically significant difference between the 2 treatment groups among patients with S. aureus bacteremia (hazard ratio [HR], 0.70; 95% CI, 0.34–1.44) or among patients who had gram-negative bacteremia (HR, 1.94; 95% CI, 0.78–4.81). However, patients without bacteremia at baseline were less likely to survive in the linezolid group than in the control group (HR, 2.20; 95% CI, 1.07–4.50). Nevertheless, linezolid has not been recommended for empirical therapy in this guideline (i.e., for patients in whom bacteremia is suspected but not confirmed).
Enterococci account for 10% of all nosocomial bloodstream infections [155, 156], many of which are caused by intravascular catheters. Sixty percent of Enterococcus faecium and 2% of Enterococcus faecalis nosocomial bloodstream infections are resistant to vancomycin . Antimicrobial resistance to newer agents, such as linezolid, has been reported [157, 158].
The risk of endocarditis as a complication of enterococcal CRBSI is relatively low. In a multicenter study involving >205 cases of CRBSI due to vancomycin-resistant enterococci, only 1.5% had definitive evidence of endocarditis . However, signs and symptoms of endocarditis, persistent bacteremia, and enterococcal bacteremia in the presence of a prosthetic valve warrant further evaluation with TEE [160, 161]. Enterococcal bacteremia that persists for >4 days has been independently associated with mortality [162, 163].
There are no data from randomized trials with adequate statistical power to determine the role of combination antimicrobial therapy or the optimal treatment duration for enterococcal CRBSI. Several retrospective cohort studies found no statistically significant differences in outcomes among patients with uncomplicated enterococcal bloodstream infection treated with combination therapy versus monotherapy [164, 165]. However, one large series found that combination therapy with gentamicin and ampicillin was more effective than monotherapy when the catheter was retained in cases of enterococcal CRBSI . The combination of ampicillin and high-dose ceftriaxone was used successfully in a nonrandomized study of enterococcal endocarditis for patients in which use of an aminoglycoside was precluded because of either antimicrobial resistance or nephrotoxicity .
An open-label clinical trial among solid-organ transplant recipients reported a 63% success rate in treating vancomycin-resistant enterococci bloodstream infections with linezolid . Quinupristin-dalfopristin has been reported for use in treating bloodstream infections due to E. faecium, with an overall clinical response rate of 69% in the small subset of patients with CRBSI . An open-label study of neutropenic patients found a 44% cure rate in an intention-to-treat analysis of enterococcal bacteremia treated with daptomycin . In a retrospective cohort study, chloramphenicol treatment of vancomycin-resistant enterococci bloodstream infections had a clinical response rate of 61% .
During the past 2 decades, rates of gram-negative bacillary intravascular device infection and secondary bacteremia among adults have decreased, supplanted by infections due to coagulase-negative staphylococci, S. aureus (often MRSA), and Candida species . The incidence of infections due to antibiotic-resistant gram-negative pathogens has increased over the past decade [86, 173], and patients with CRBSI due to MDR gram-negative pathogens are at greater risk for inappropriate initial antibiotic therapy, which results in increased morbidity and mortality (table 4) [172–177]. Risk factors for infection due to MDR gram-negative bacilli include being critically ill, being neutropenic, having received prior antibiotic therapy, and having a femoral catheter [172, 178–180].
Over the past decade, the incidence of gram-negative bacilli resistant to third- and fourth-generation cephalosporins has increased [15, 86, 173]. MDR Klebsiella pneumoniae and Escherichia coli expressing extended-spectrum β-lactamases have been associated with poor clinical outcomes when treated with cephalosporins or piperacillin-tazobactam versus carbapenems, even when the organisms appear to be susceptible in vitro [173, 177]. In addition, there is increasing concern over the evolution of MDR gram-negative bacilli having carbapenemases that confer resistance to carbapenems, and some of these enzymes may also be active against cephalosporins . No randomized, controlled trials have evaluated various treatments for gram-negative bacilli that produce the new β-lactamases or carbapenemases and require therapy with polymyxin (colistin) or an aminoglycoside . Treatment failure among patients with Enterobacter bacteremia who are administered cephalosporins has also been observed .
Suggested management recommendations are summarized in figures 1–4 and tables 4–6. Most of the recommendations for the management of CRBSI due to MDR gram-negative bacilli have been limited by small numbers of cases derived from outbreaks or small clusters of infections, concerns over the accuracy and interpretation of in vitro susceptibility data, and confounding by concurrent use of combinations of antibiotics. When culture and susceptibility data are available, the initial antibiotic regimen can be adjusted to a single agent for the remainder of the therapeutic course, usually for 7–14 days . Recommendations and guidelines for the management of sepsis have been recently published . Recommendations for antimicrobial therapy for specific gram-negative pathogens are shown in table 4.
Short-term catheters infected with gram-negative bacilli should be removed (figure 1). Patients with CRBSI associated with long-term CVCs or implanted ports should have the device removed, and appropriate antibiotic therapy should be administered (figure 4). Several studies have advocated the removal of an infected catheter for patients with CRBSI due to MDR gram-negative bacilli that have a propensity for biofilm production, such as Acinetobacter baumannii, Pseudomonas species, and Stenotrophomonas maltophilia [172, 179, 180, 183]. However, these studies are limited by small numbers of patients and lack data on the efficacy of combination therapy with an antibiotic lock and systemic antibiotics. Recent studies in which antibiotic lock and systemic antibiotics were used to treat gram-negative rod CRBSI have found high success rates [99, 114].
Fluconazole administered at a dosage of 400 mg daily for 14 days after the first negative blood culture result is obtained is equivalent to amphotericin B in the treatment of candidemia caused by Candida albicans and azole-susceptible strains . For Candida species with decreased susceptibility to azoles (e.g., C. glabrata and C. krusei), echinocandins (caspofungin administered with a 70-mg intravenous loading dose, followed by 50 mg daily administered intravenously; micafungin at a dosage of 100 mg daily administered intravenously or anidulafungin with a 200-mg intravenous loading dose followed by 100 mg daily administered intravenously) or lipid formulations of amphotericin B (ambisome or amphotericin B lipid complex) administered intravenously at a dosage of 3–5 mg/kg daily are highly effective [185–187]. Conventional amphotericin B therapy is also effective but is associated with more adverse effects.
There are limited clinical data to suggest that antifungal lock therapy with amphotericin B may result in catheter salvage for patients with candidemia [93, 127]. Echinocandins , lipid formulations of amphotericin B [194, 195], or ethanol-based lock solutions [196, 197] eradicate biofilm-containing Candida in vitro, but catheter retention in combination with antifungal lock therapy is still investigational at the present time.
If Candida is grown from blood samples obtained from a patient with a long-term catheter or implantable port, the decision regarding catheter removal should be based on predictors of a catheter-related candidemia versus another source of infection (e.g., the gastrointestinal tract). Predictors of CRBSI involving long-term catheters include the following: a >3:1 quantity of Candida growing from the catheter-drawn blood cultures, compared with percutaneous blood cultures; catheter-drawn blood cultures growing >2 h before percutaneous blood cultures [36, 48, 49, 198]; candidemia in a patient who has not received chemotherapy or steroid therapy within 1 month before the onset of infection and who has no dissemination or other apparent source for the bloodstream infection except the intravascular catheter; candidemia in a patient receiving hyperalimentation through the catheter; and persistent candidemia unresponsive to systemic antifungal therapy [199, 200]. Any of these conditions should raise suspicion for Candida-related CRBSI and the need to remove the catheter. Management of candidemia and other fungal infections is summarized in tables 4–6, figures 1 and and2,2, and in the recent IDSA guidelines for the management of candidiasis .
Isolation of these microorganisms from a single blood culture set does not prove true bloodstream infection. Confirmation by multiple percutaneous blood culture results positive for the same organism is required before meaningful conclusions can be drawn as to the significance of the culture results. CRBSIs due to Micrococcus and Bacillus species are difficult to treat successfully unless the infected catheter is removed [202, 203]. A high incidence of CRBSI due to Micrococcus species has been reported among patients treated for pulmonary arterial hypertension with continuous epoprostenol .
Suppurative thrombophlebitis may involve central or peripheral veins or arteries and result in high-grade and persistent bacteremia or fungemia [205–210]. Patients who undergo chemotherapy for malignancy and patients with solid tumors who develop S. aureus CRBSI may be at increased risk for suppurative thrombophlebitis, because S. aureus is the most common offending organism [207, 211–214]. Septic pulmonary emboli and other metastatic infections may complicate this condition [207, 215]. Patients may remain febrile and bacteremic or fungemic for prolonged periods of time despite the initiation of appropriate antimicrobial therapy; however, few patients have physical examination findings that suggest the diagnosis of suppurative thrombophlebitis . Only a minority of patients require surgery for the definite resolution of suppurative thrombophlebitis.
Because infected intravascular thrombus and intraluminal abscess may remain intact after catheter removal, this infection may become manifest after catheter removal . When peripheral veins are involved, many older children and adult patients have localized pain, erythema, and edema, and a smaller subset of patients demonstrate an abscess, palpable cord, or purulent drainage [206, 217, 218]. A patient with suppurative thrombophlebitis caused by a peripheral arterial catheter may present with a pseudoaneurysm or embolic lesions of the involved hand [205, 210]. Patients with suppurative thrombophlebitis of the great central veins may have ipsilateral neck, chest, or upper extremity swelling [208, 209, 219]. There are no randomized studies to guide the optimal choice or duration of antibiotics, use of anticoagulants, thrombolytic agents, or excision of the involved vessel, but anticoagulation with heparin should be considered .
Colonized intravascular catheters are the most commonly identified source of nosocomial endocarditis and account for one-to two-thirds of reported cases [24, 25, 34, 221–224]. Staphylococci are the main etiologic agents, followed by Enterococcus and Candida species [24, 25]. The risk of nosocomial endocarditis is greatest among patients with S. aureus bacteremia who have prosthetic heart valves, pacemakers, malignancy, or who are receiving dialysis through a catheter [24, 25, 34, 44, 225, 226].
There are no data from randomized clinical trials to establish the indications for TEE, but clinical examination has a low sensitivity for diagnosing infective endocarditis. A TEE should be offered to all patients with S. aureus bacteremia, with the possible exception of patients whose fever and bacteremia resolve within 3 days after catheter removal who have no underlying cardiac predisposing conditions for endocarditis and no clinical signs of endocarditis .
Repeatedly positive blood culture results and/or unchanged clinical status for 3 days after catheter removal usually reflects serious sequelae of CRBSI, such as suppurative thrombophlebitis, endocarditis, or metastatic foci of infection.
Outbreaks of CRBSI occur infrequently and are most commonly caused by contaminated infusate . These infections can be difficult to recognize and are sufficiently uncommon that they may go unnoticed by clinicians. Any fluids administered through an intravenous catheter can become contaminated, either during the manufacturing process or while being prepared or administered in the health care setting. Numerous outbreaks of bloodstream infection related to contaminated, intravenously administered products have been reported [227–237]. In addition, medical equipment can become contaminated because of inadequate infection-control practices [238–254]. In some instances, health care workers have adulterated intravenous narcotics for illicit use and have contaminated the narcotics in the process .
Bacteria that are most often implicated in contamination of infusate include gram-negative bacilli capable of reproducing at room temperature, such as Klebsiella species, Enterobacter species, Serratia species, Burkholdaria cepacia, Ralstonia pickettii, and Citrobacter freundii . Gram-negative bacilli that are unusual human pathogens or that are frequently found in the environment should alert the clinician to the possibility of contaminated infusate.
Because the clinical picture of contaminated infusate is the same as that of bloodstream infection due to other causes, contamination of infusate often will not be detected unless there is a cluster of unusual bloodstream infections or several patients develop a bloodstream infection due to the same organism. Contaminated infusate should be suspected when no other infection is present that would account for a bloodstream infection or when the abrupt onset of shock occurs in association with infusion of parenteral medication or fluid.
Contamination in the hospital pharmacy should be suspected if an increase in bloodstream infection due to the same microorganism occurs among patients on different hospital units. Suspected contamination should prompt an immediate and thorough investigation. Assistance from public health authorities may be required, especially if related outbreaks occur in multiple health care settings.
We thank Drs. Stijn Blot, Vance G. Fowler, Mark E. Rupp, Richard Watkins, and Andreas F. Widmer, for their thoughtful review of earlier drafts of the manuscript, and Dr. Jennifer Hanrahan, for assistance in drafting the outbreak management section.
Financial support. The Infectious Diseases Society of America.
aIt is important to realize that guidelines cannot always account for individual variation among patients. They are not intended to supplant physician judgment with respect to particular patients or special clinical situations. The IDSA considers adherence to these guidelines to be voluntary, with the ultimate determination regarding their application to be made by the physician in the light of each patient’s individual circumstances.
Potential conflicts of interest. L.A.M. has received research funding from Angiotech and Theravance and has served as a consultant to Cadence, Ash Access Technology, and CorMedix. M.A. is a consultant for Angiotech and Covidien. E.B. has served on advisory boards for or received research or conference funds from Pfizer, Merck Sharp and Dohme, Cerexa, Cardinal-Health, Sanofi-Aventis, GlaxoSmithKline, Astellas and Astra-Zeneca. D.E.C. has served on the speaker’s bureaus of Pfizer, Wyeth, Sanofi Pasteur, and Merck and has received research funding from Bard, Nomir Medical Technologies, Data and Safety Monitoring Board, and Johnson & Johnson. B.J.A.R. has received research grants from Schering-Plough and Gilead Sciences; has served as a speaker for Bayer, Schering-Plough, Pfizer, and Tibotec; and has served as a consultant to Pfizer and Schering-Plough. P.F. has clinical research contracts with MedImmune and Tibotec. I.I.R. has received research grants from Cubist, Schering-Plough, Versicor, Enzon, Cook Medical, Schering-Plough, and Wyeth; has served on the speaker bureaus of Merck, Pfizer, Cook, and Schering-Plough; has served as a consultant to Clorox, Cubist, and Cook; and has received royalties related to patent licensed to American Medical Systems, Horizon Medical Products, and TyRx on which he is a coinventor. D.K.W. has served on the Pfizer speaker’s bureau; has received research funding from GeneOhm Sciences, Verimetrix, and Astellas Pharma; and has served as a consultant to 3M Healthcare. N.P.O. and R.J.S.: no conflicts.