Dennis G. Maki
Leonard A. Mermel
Reliable intravascular access for administration of fluids and electrolytes, blood products, drugs, and nutritional support, and for hemodynamic monitoring is now one of the most essential features of modern medical care (Table 37-1). Each year in the United States, ~150 million intravascular devices are purchased by hospitals and clinics. The vast majority are peripheral venous catheters; however, >5 million central venous devices of various types are sold in the United States annually.
More than one half of all epidemics of nosocomial bacteremia or candidemia reported in the world literature between 1965 and 1991 derived from vascular access in some form [1,2]. One third to one half of episodes of nosocomial endocarditis have been traced to infected intravascular catheters [3,4,5], and healthcare-associated intravascular device-related bloodstream infection (IVDR-BSI) is associated with a 12%-28% attributable mortality [6,7,8,9]. Yet, infusion therapy generally has an underappreciated potential for producing iatrogenic disease. For example, <50% of the intensive care units (ICUs) in the United Kingdom had a written policy concerning the care of central venous catheters (CVCs) after insertion [10].
Infusion-related BSI is too frequently unrecognized, in great measure due to its relative infrequency. The percentage of infusions identified as producing BSI is sufficiently low—<1% on the average—that an average physician or nurse is unlikely to encounter more than an occasional episode. But even a low incidence of infection applied to the estimated 30 million patients who receive infusion therapy in U.S. hospitals annually translates to an estimated 50,000 to 100,000 BSIs nationwide each year [1,2], with 55,000 due to CVCs in U.S. ICUs [11]. Because neither the device nor the infusate is routinely cultured, the source of the BSI in a large proportion of episodes is never recognized.
Intravascular device-related bloodstream infection is largely preventable. This premise forms the thesis for this review: the primary goal must not be simply to identify and treat these iatrogenic infections, but rather to prevent them. By critically scrutinizing existing knowledge of the pathogenesis and epidemiology of device-related infection—the reservoirs of healthcare-associated infection (HAI) pathogens and modes of transmission to patients' infusions—rational and effective guidelines for prevention can be formulated [12].
Sources and Forms of Infusion-Related Inflammation and Bloodstream Infection
There are three major sources of BSI associated with any intravascular device: 1) colonization of the cannula wound, 2) colonization of the cannula hub, or 3) contamination of the fluid (i.e., infusate) administered through the cannula. Cannulas, which cause most endemic IVDR-BSIs, produce BSI far more frequently than contaminated infusate, the source of most epidemics of infusion-associated BSI [1].
It is important to understand the different stages and forms of device-related inflammation or infection, which range from infusion phlebitis—usually unrelated to infection—to asymptomatic colonization of the intravascular device—usually by skin commensals with little intrinsic virulence—to overwhelming septic shock originating from an infected thrombus in a cannulated great central vein or from infusate heavily contaminated by gram-negative bacilli.
P.612
|
TABLE 37-1 |
|||||||||||||||||||||||||||||||
|
Infusion Phlebitis
Infusion phlebitis, defined as inflammation of the cannulated vein—pain, erythema, tenderness, or an inflamed, palpable, thrombosed vein—is a frequent cause of pain and discomfort to the millions of patients who receive infusion therapy through peripheral intravenous (IV) cannulas each year in U.S. hospitals. Most investigators have concluded that infusion phlebitis is primarily a physicochemical phenomenon, and prospective studies have shown that the cannula material, length, and bore size; operator skill on insertion; the anatomic site of cannulation; the duration of cannulation; the frequency of dressing changes; the character of the infusate; and host factors such as patient age, Caucasian race, female gender, and the presence of underlying diseases significantly influence the risk of infusion phlebitis (Table 37-2).
In a prospective clinical study of 1,054 peripheral IV catheters, the Kaplan-Meier risk for phlebitis exceeded 50% by the fourth day after catheterization. IV antibiotics (relative risk (RR), 2.0), female gender (RR, 1.9), catheterization beyond 48 hours (RR, 1.8), and catheter material (polyetherurethane (Vialon), tetrafluoroethylene-hexafluoropropylene (Teflon), RR, 0.7) were strong predictors of phlebitis in a Cox proportional hazards model (each, p <0.003) [13]. The best-fit model for severe phlebitis identified the same predictors plus catheter-related infection (RR, 6.2), phlebitis with the previous catheter (RR, 1.5), and anatomical site (hand: forearm, RR, 0.7; wrist: forearm, RR, 0.6).
Although not all studies have identified an association between phlebitis and catheter-related infection [14,15], this large, prospective study showed a strong statistical association, as have other studies [16,17,18,19,20]. Phlebitis also can be produced by contaminated infusate. Patients with BSI from intrinsically contaminated fluid in a large nationwide epidemic traced to the contaminated products of one U.S. manufacturer in 1970 to 1971 had a much higher incidence of phlebitis than patients receiving IV fluids who did not develop BSI [21].
Only a small proportion of patients with IV cannula-associated peripheral vein phlebitis have infusion-related infection, and <50% of patients with peripheral IVDR-BSI show phlebitis; however, the presence of phlebitis connotes a substantially increased risk of infection and indicates the need for immediate removal of the catheter to reduce the severity of phlebitis, for symptomatic relief, and to prevent catheter colonization from progressing to BSI.
Cannula-Related Infections
Between 5% and 25% of intravascular devices are colonized by skin organisms at the time of removal, as reflected by semiquantitative or quantitative cultures showing large numbers of organisms on the intravascular portion of the removed catheter or its tip. Colonization, which in most instances is asymptomatic, provides the biologic setting for systemic infection to occur and can be considered synonymous with localized infection. However, colonized cannulas are more likely than noncolonized ones to show phlebitis or local inflammation, especially purulence—pus spontaneously draining or expressable from the insertion site—and are far more likely to cause systemic infection (i.e., cannula-related bacteremia or fungemia) [17,22,23].
One of the most serious forms of intravascular device-related infection occurs when intravascular thrombus surrounding the cannula becomes infected. This causes septic (suppurative) thrombophlebitis when it occurs in association with peripheral IV cannulas [24,25], or septic thrombosis of a great central vein when associated with centrally placed catheters [26,27]. With suppurative phlebitis, the vein becomes an intravascular abscess, discharging myriads of microorganisms into the bloodstream, even after the cannula has been removed. The clinical picture is predictable: overwhelming BSI with high-grade and often unremitting bacteremia or fungemia. This syndrome is most likely to be encountered in burned patients or other ICU patients who have heavy cutaneous colonization and develop a cannula-related infection that goes unrecognized,
P.613
permitting microorganisms to proliferate to high levels within the intravascular thrombus. The catheter insertion site is devoid of signs of inflammation >50% of the time, and the clinical picture may not present until several days after the catheter has been removed. In any patient with an IV catheter who develops high-grade BSI that persists after an infected cannula has been removed, it is likely the patient has infected thrombus in the recently cannulated vein, and may even have secondary endocarditis or seeding to other distant sites [28].
|
TABLE 37-2 |
|||||||||||||||||||||||||||||||||||||||||
|
The microorganisms most frequently implicated in suppurative phlebitis are predominantly Staphylococcus aureus, and Candida species [24,25,26,27]. Although coagulase-negative
P.614
staphylococci commonly cause IVDR-BSI, they rarely cause suppurative thrombophlebitis, possible because of their lesser tendency to bind to host-derived protein components of thrombus compared with other pathogens such as S. aureus [29,30].
Suppurative phlebitis of peripheral IV catheters is now rare, and the syndrome of IV suppuration is predominantly a complication of CVCs, characteristically catheters that have been left in place for many days in heavily colonized ICU patients.
Bloodstream Infection From Contaminated Infusate
It also is important to recognize that the infusate—parenteral fluid, blood products, or IV medications—administered through an intravascular device also can become contaminated and produce infusion-related BSI, which is more likely than cannula-related infection to culminate in frank septic shock. Contaminated fluid is a rare cause of endemic infection with short-term peripheral IV devices, but the infusate is more commonly associated with infections of catheters used for hemodynamic monitoring, CVCs, and, possibly, surgically implanted cuffed Hickman or Broviac catheters [31,32,33,34]. Most nosocomial epidemics of infusion-related BSI, however, have been traced to contamination of infusate by gram-negative bacilli, introduced during its manufacture (intrinsic contamination) [21] or during its preparation or administration in the healthcare system (extrinsic contamination) [1,2,35,36].
Diagnosis of Infusion-Related Bloodstream Infection
Clinical Features
Although meticulous aseptic technique during cannula insertion and good follow-up care greatly reduce the risk of IVDR-BSI, sporadic episodes and even epidemics still can be expected occasionally to occur because of human error, intrinsically contaminated products, or the undue susceptibility to infection of many patients. If affected patients are to survive, the causal relationship between an infusion and the BSI must be recognized as early as possible.
The general clinical features of infusion-related bacteremia or fungemia are nonspecific and indiscernible from BSIs arising from any local site of infection, such as urinary tract infection (UTI) or surgical site infection (SSI) (Table 37-3). There also appears to be a poor correlation between clinical judgment and microbiologic confirmation of IVDR-BSI [37]. Infusion-related BSI occurring in ICU patients can be particularly insidious: bacteremia or fungemia usually is identified by positive blood cultures but is attributed to nosocomial pneumonia, UTI or SSI, or is simply accepted as “cryptogenic” and treated empirically.
|
TABLE 37-3 |
||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||
Certain clinical, epidemiologic, and microbiologic findings can be extremely helpful to the clinician evaluating a hospitalized patient with a picture of nosocomial BSI or cryptogenic bacteremia or candidemia, and point toward an IVD as the source (Table 37-3):
P.615
During a large, nationwide outbreak in 1970 to 1971 due to intrinsic contamination of one U. S. manufacturer's products, patients treated with antibiotics to which the epidemic organisms were susceptible remained clinically septic, continued to have positive blood cultures after 24 hours or more of appropriate therapy, and did not improve clinically until their infusions were serendipitously or intentionally removed [21].
Focal retinal lesions—cotton wool spot patches—may be seen in patients with disseminated Candida spp. infection deriving from CVCs, even in those without positive blood cultures [41]. Careful ophthalmologic examination should be routinely performed in the evaluation of patients with CVCs with suspected IVDR-BSI, especially patients receiving total parenteral nutrition (TPN). BSI from arterial catheters may be heralded by embolic lesions that manifest as tender, erythematous papules, 5–10 mm in diameter, appearing in the distal distribution of the involved artery, usually in the palm or sole—Osler's nodes [42,43]. Arterial bleeding from the insertion site often is the harbinger of BSI caused by an infected arterial catheter and may denote an infective pseudoaneurysm [42,44,45]. Endocarditis, particularly right sided, is a rare but well documented complication of flow-directed pulmonary artery catheters [46,47,48].
Blood Cultures
Blood cultures are essential to the diagnosis of IVDR-BSI (see Chapter 9), and in any patient suspected of infusion-related infection, two or three separate 10-ml blood cultures should be drawn [49,50,51], ideally from peripheral veins, by separate venipunctures. If the patient is receiving antimicrobial therapy, blood cultures obtained immediately before a dose is due to be administered and blood antibiotic levels are likely to be low, may provide a higher yield. Use of resin-containing media to adsorb and remove any antibiotic present in the blood specimen [52], adsorb serum factors detrimental to the growth of Enterobacteriaceae [53], and lyse the cell wall of neutrophils, thereby releasing intracellular pathogens [54], also may increase the yield [55].
The use of a biphasic system, such as the Isolator® (E. I. DuPont, Nemours and Co., Wilmington, DE), or systems with selective high blood volume fungal media (BACTEC; Becton Dickinson Diagnostic Instrument Systems, Sparks, MD) appear to significantly enhance the laboratory detection of fungemia [56,57].
The volume of blood cultured is critical to maximize the yield of blood cultures for diagnosis of bacteremia or candidemia: in adults, obtaining at least 20 ml, ideally 30 ml, per drawing—each specimen containing 10 or 15 ml, inoculated into aerobic and anaerobic media—significantly improves the yield, compared with obtaining only 10 ml at each drawing and culturing a smaller total volume [58,59]. It is rarely necessary to obtain > two 15-ml cultures or three 10-ml cultures in a 24-hour period. If at least 30 ml of blood is cultured, 99% of detectable bacteremias should be identified [58].
It is common practice in many ICUs to draw blood cultures through central venous or arterial catheters or, in neonates, through umbilical catheters. Comparative studies of standard blood cultures drawn through central venous or arterial catheters in adults usually have shown good concordance with cultures drawn by percutaneous peripheral venipuncture [60,61,62], but rates of false-positive (contaminated) cultures can be considerably higher with catheter-drawn specimens [63]. The practice of drawing non-qualitative blood cultures through indwelling vascular catheters probably ought not to be encouraged because of the risk of introducing contamination during the manipulation [64]. If, however, to preserve dwindling superficial veins it is considered unavoidable to use a vascular catheter to obtain blood cultures, an attempt should be made to use a newly inserted catheter [60,61,62] and to draw at least every other specimen by percutaneous venipuncture.
If the laboratory is prepared to do pour-plate blood cultures or has available an automated quantitative system for culturing blood, such as the Isolator® system, catheter-drawn blood cultures can permit the diagnosis of IVDR-BSI to be made with reasonable sensitivity and specificity (both in the range of 90%), without removing the catheter [65,66,67,68,69,70,71]. With infected catheters, a quantitative blood culture drawn through the catheter usually shows a marked step-up—often >10-fold—in the concentration of organisms compared with quantitative blood cultures drawn at the same time percutaneously through a peripheral vein.
P.616
Quantitative catheter-drawn blood cultures probably have their greatest utility in diagnosis of device-related infection with surgically implanted cuffed Hickman or Broviac catheters and subcutaneous central venous ports [65,66,67,68].
Finding microbes on Gram stain or acridine orange stain of blood drawn through CVCs has been shown to be highly sensitive and specific for diagnosing IVDR-BSI [73,74]. If confirmed by others using the same and other IVDs, these may be the methods of choice for the rapid diagnosis of serious intravascular catheter-related infections. Intracellular bacteria have been found on Wright-stained peripheral smears in asymptomatic patients with occult CVC-BSI [75].
Microbiology of Intravascular Device-Related Bloodstream Infection
The microbiologic profile of BSI (Table 37-3) can strongly suggest an infusion-related source. Cryptogenic staphylococcal BSI, particularly with coagulase-negative staphylococci, BSI caused by Bacillus or Corynebacterium (especially JK-1) spp. or Enterococcus, or fungemia caused by Candida, Fusarium, Trichophyton, or Malassezia spp., especially in a patient with a CVC, is most likely to reflect catheter-related infection [1,2,38].
BSIs caused by Enterobacter cloacae or, especially, Pantoea (formerly Enterobacter) agglomerans, Burkholderia cepacia, Stenotrophomonas maltophilia, or Citrobacter spp., in the setting of infusion therapy, may signal an epidemic and should prompt studies to rule out contaminated infusate [76]. A BSI cluster should mandate a full-scale investigation, which may include culturing of large numbers of in-use infusions and informing the local, state, and Federal public health authorities. Such actions averted a large, nationwide epidemic in 1973, when, prompted by five unexplained BSIs in three hospitals, intrinsic contamination of one U.S. company's products was identified and a recall put into effect so rapidly that the outbreak was limited to the five initially recognized patients [77]. It must be emphasized, however, that for BSI surveillance to be maximally effective, all blood culture isolates must always be fully identified—that is, identified to the genus and species level. Failure to do so during the 1970 to 1971 nationwide epidemic traced to the contaminated products of one U.S. manufacturer resulted in pre-eminent hospitals experiencing large numbers of infections that were recognized as infusion-related only in retrospect [21].
Cryptogenic nosocomial BSI caused by psychrophilic (cold-growing) organisms, such as non-aeruginosa pseudomonads, Ochrobactrum anthropi (formerly Achromobacter)Flavobacterium, Enterobacter, or Serratia spp. [78,79], or by Salmonella [80] or Yersinia spp. [81], with a picture of overwhelming BSI, may indicate a contaminated blood product.
Cultures of Intravascular Devices
Some laboratories still culture vascular catheters qualitatively, amputating the tip aseptically and immersing it in liquid media. Unfortunately, a positive culture by this technique is diagnostically nonspecific because a single organism picked up from the skin as the catheter is removed can produce a positive—false-positive—culture [82]. Many IVDR-BSIs derive from local infection of the transcutaneous cannula tract (see discussion later). Culture of the external surface of the withdrawn cannula should reflect the microbiologic status of the wound, and quantitative culture should more accurately distinguish infection from contamination. A standardized, semiquantitative method for culturing vascular cannulas in solid media was developed in 1977 [17]. Colony counts on semiquantitative culture are bimodally distributed, as they are in quantitative urine cultures. The method provides excellent discrimination between colonization and insignificant contamination acquired during catheter removal. Fifteen or more colony-forming units (cfu) growing on a semiquantitative plate is regarded as a positive culture, and denotes significant growth or colonization [17]. Based on experience with >10,000 IVDs, positive cultures found using this technique have shown a 15%-40% association with concordant BSI. Cannulas positive on semiquantitative culture also are strongly associated with local inflammation [17].
Good correlation between high colony counts and IVDR-BSI have been demonstrated with cultures of catheter segments semiquantitatively on solid media [83,84,85] or quantitatively in liquid media—removing organisms from the catheter by vortexing or sonication [83,86,87]. The latter techniques appear to have the greatest sensitivity and specificity for the diagnosis of vascular catheter-related infection [72,88]. However, a negative catheter culture may not rule out a catheter-related BSI (CR-BSI) [33,37,72,88,89]. Using >1 catheter culture technique increases the yield [88], as does bedside plating of catheters for semiquantitative culture [90]. Direct Gram stains [91] or acridine orange stains [92] of intravascular segments of removed catheters also show excellent correlation with quantitative techniques for culturing catheters and can permit rapid diagnosis of catheter-related infection.
Given the strong evidence implicating cutaneous microorganisms in the genesis of most BSIs caused by short-term, noncuffed intravascular catheters, a number of studies have shown that a quantitative culture [22,93] or Gram stain [94] of skin at the insertion site also can identify infected catheters with reasonable sensitivity and specificity, greatly exceeding an assessment based solely on clinical signs of infection. The combination of culturing skin both at the insertion site and the catheter hub has been reported to provide even better sensitivity for diagnosis of catheter-related infection, without removing the catheter [33,95,96].
P.617
To reliably diagnose infection caused by contaminated infusate requires a sample of fluid to be aspirated from the line and cultured quantitatively [76]. A variety of techniques are now available for culturing or processing parenteral admixtures and fluid medications in the laboratory for microbial contamination [97,98]. Because there is no evidence that anaerobic bacteria can grow in parenteral crystalloid admixtures, anaerobic culture techniques are not necessary unless blood or another biologic product is involved.
Definitions for Infusion-Related Infection
Using the results of semiquantitative or quantitative culture of the catheter and cultures of the hub of the catheter and of infusate aspirated from the line at the time the catheter is removed, and concomitant blood cultures, it is possible to formulate rigorous definitions for intravascular device-related infection [99] (Table 37-4).
|
TABLE 37-4 |
|
|
The term “catheter sepsis” appears frequently in the literature, but lacks stringent criteria. Although sepsis has been defined by a consensus panel [100], the term as applied to catheter-related infections may be insensitive because many of these infections are due to coagulase-negative staphylococci, and in some studies, only 55%-71% of patients with bacteremia due to this pathogen had leukocytosis [101,102]. Also, the maximal body temperature is <38°C in many patients with coagulase-negative staphylococcal BSI [102,103]. This term should likely not be further promulgated in the literature, especially in those prospective, comparative studies involving IVDs [99].
In addition, the definitions listed previously may be unnecessarily rigorous for use in clinical HAI surveillance because very few clinicians obtain cultures of catheter hubs or infusate, even if the cannula is cultured. Moreover, patients with disseminated candidiasis originating from an infected catheter often have negative blood cultures. It also is important to realize that multiple sites often are colonized when cultures are performed at the time of catheter withdrawal and it may be difficult to distinguish with certainty the source of many catheter-BSIs. Therefore, for routine surveillance, use of the Centers for Disease Control and Prevention (CDC) definitions is recommended [104,105].
P.618
Cannula-Associated Infection
Incidence of Cannula-Related Bloodstream Infection
IVDR-BSI is perhaps the least frequently recognized HAI. The true incidence of IVDR-BSI is underestimated in most centers because a catheter often is not suspected as a source of the patient's clinical picture of nosocomial BSI, and is not cultured. Prospective studies in which every device enrolled is cultured at the time of removal clearly indicate that every type of IVD carries some risk of causing BSI, but the magnitude of risk per device varies greatly, depending on its type [2].
Table 37-5 shows representative rates of infection for various types of intravascular devices. The lowest rates are with small peripheral IV steel needles and Teflon or polyurethane catheters: large prospective studies have shown rates of approximately 0.2 BSIs per 100 peripheral IV catheters [13,14,15,16,19,105,106,107,108,109]; two large, comparative trials have shown that if IV cannulas are inserted under scrupulous aseptic conditions, plastic catheters probably pose no greater risk of device-related bacteremia or candidemia than steel needles [106]. Prospective studies of arterial catheters used for hemodynamic monitoring have found rates of infusion-related bacteremia in the range of 1% [110].
|
TABLE 37-5 |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
The device that poses the greatest risk of iatrogenic BSI is the CVC in its numerous forms [9,111,112,113]. Many prospective studies of short-term, noncuffed, single- or multilumen catheters inserted percutaneously into the subclavian or internal jugular vein have found rates of CR-BSI in the range of 2%-5% [9,83,114,115,116,117,118,119]. Percutaneously inserted, noncuffed CVCs used for hemodialysis have been associated with the highest rates of BSI, >10% [120,121,122]; however, cuffed hemodialysis catheter use appears to be associated with a lower incidence of BSI [123,124]. Peripherally inserted central catheters (PICCs) pose a substantially lower risk of CR-BSI (0.04 to 0.4 per 100 catheter days), comparable to Hickman catheters [125,126]. Swan-Ganz pulmonary artery catheters used for hemodynamic monitoring are associated with a 1% rate of BSI or 0.3 per 100 catheter-days [110]. The lowest rates of infection with CVCs have been with surgically implanted Hickman or Broviac catheters that incorporate a Dacron cuff, which have been associated with rates of infection in the range of 0.1 bacteremias or fungemias per 100 catheter-days [127,128], and surgically implanted subcutaneous central venous ports, associated with a rates of BSI of <0.05 per 100 device-days [127,128]. In prospective studies, the incidence of BSI has been demonstrated to be lower in patients who have subcutaneously implanted ports compared with those with tunneled catheters with a Dacron cuff [129,130,131,132,133,134].
P.619
It has been estimated that 90% of IVDR-BSIs originate from CVCs of various types [2], leading to ~55,000 BSIs in U.S. ICUs each year [11]. Data from the CDC's National Nosocomial Infections Surveillance (NNIS) study have shown that the incidence of secondary BSIs, deriving from identifiable local infections such as UTIs, SSIs, or pneumonias, has remained stable over the past decade; in contrast, the incidence of primary nosocomial BSIs, the largest proportion of which derive from IVDs, has increased more than twofold over this same period [2,135], reflecting the great increase in the use of infusion therapy and, especially, the use of CVCs of all types. It seems clear that the greatest hope for reducing the risk of IVDR-BSI will come from better understanding of infection with CVCs, which will form the basis for more effective strategies for prevention.
Epidemiology
The first and perhaps most important question that must be addressed to develop effective strategies for prevention is to determine the major source or sources of microorganisms that can colonize a percutaneous IVD (Fig. 37-1) and cause invasive infection leading to bacteremia or candidemia. An intravascular catheter can easily become colonized extraluminally by organisms from the patient's cutaneous microflora. Contamination may occur during catheter insertion [136] or shortly thereafter [137]. Microorganisms also can contaminate the catheter hub where the administration set attaches to the catheter, or they may gain access to the fluid column and be infused directly into the patient's bloodstream; the device also can become infected hematogenously from remote sources of local infection; or the device might even be contaminated from its manufacture—which fortunately is very rare.
|
Figure 37-1 Sources of intravascular cannula-related infection. The sources are the skin flora, contamination of the catheter hub, contamination of infusate, and hematogenous colonization of the intravascular device and its fibronectin–fibrin sheath. (HCW, healthcare worker). |
A large body of clinical and microbiologic data indicates that most IVDR-BSIs caused by short-term, percutaneously inserted, noncuffed catheters are caused by extraluminal microorganisms of cutaneous origin that invade the transcutaneous insertion wound at the time the catheter is inserted or in the days after insertion.
Numerous prospective studies of intravascular device-related infection have shown that coagulase-negative staphylococci, the predominant aerobic species on the human skin, are now the most common agents of CR-BSI [1,2,9,48,105,106,107,108,109,111,117,127,135]. The vast majority of vascular CR-BSIs are caused by microorganisms that colonize the skin of hospitalized patients: staphylococci, both coagulase-negative and coagulase-positive (S. aureus); Candida, Corynebacterium, and Bacillus spp.; and, to a lesser degree, aerobic gram-negative bacilli (Table 37-6).
Prospective studies also have shown strong concordance between organisms present on skin surrounding the catheter insertion site and organisms recovered from CVCs producing BSI [22,34,48,93,94,114,115,116,136]. There appears to be a direct parallel between the level and profile of cutaneous colonization at the insertion sites of short-term central venous, arterial, or peripheral IV catheters and the risk of CR-BSI [138,139].
ICU unit and hemodialysis patients with cutaneous colonization by S. aureus experience four- to sixfold higher rates of IVDR-BSI [140,141]. Use of recombinant interleukin-2, with or without lymphokine-activated killer (LAK) cells for cancer immunotherapy, which is associated with frequent dermatotoxicity (desquamation) and heavy cutaneous colonization by S. aureus, has been associated with a prohibitively high incidence of CVC-related S. aureus BSI [142].
Burned patients, who have huge populations of microorganisms on the skin surface, experience very high rates of CR-BSI [143,144] (see Chapter 38).
P.620
|
TABLE 37-6 |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Numerous outbreaks of IVDR-BSI have been traced to contaminated cutaneous antiseptics [145,146,147,148].
High counts of microorganisms on semiquantitative culture of the external surface of a removed catheter are strongly associated with bacteremia caused by the catheter [17,22,83,84,85,95,121,143,149].
Microscopic examination of infected CVCs has shown heavy colonization of the external surface [91,92], especially with short-term catheters [150].
Prospective studies have shown that use of a more effective cutaneous antiseptic, e.g., chlorhexidine, for antisepsis of the insertion site at the time of catheter insertion and in follow-up care of the catheter greatly reduces the risk of infusion-related BSI [32,151,152,153].
Prospective trials have shown that antiseptics or antimicrobials applied topically to the intravascular catheter insertion site can reduce the risk of CR-BSI [122,154].
Surgically implanted Broviac or Hickman catheters, which have a subcutaneous Dacron cuff that becomes ingrown by tissue and poses a mechanical barrier against invasion of the tract by skin organisms, have been associated with considerably lower rates of CR-BSI (~0.20 episodes per 100 catheter-days) [155,156,157] than short-term, noncuffed CVCs (0.6 to 1.0 per 100 catheter days) [22,91,111,114,115,117] (Table 37-5). However, in a clinical trial, nonsurgically implanted, non-tunneled, noncuffed Silastic catheters were inserted for prolonged periods of time with a very low risk of infection [125]. These data suggest that with careful follow-up, insertion of noncuffed CVCs with the Seldinger technique outside of the operating room may be an acceptable alternative to Hickman or Broviac cuffed catheters. With one exception [158], prospective, randomized, clinical trials of a subcutaneous silver-impregnated cuff that can be attached to a short-term (<10 to 14 days) CVC at the time of insertion also can reduce the risk of catheter colonization and CR-BSI [115,116,159]. However, with more prolonged catheterization (>14 days), this device does not appear to be efficacious [160,161,162]. Studies have shown that novel short-term (~7 days) CVCs with an externally antimicrobial [163,164] or antiseptic or heparin [165,166,167,168,169] greatly reduce the incidence of catheter colonization and, in some instances, CR-BSI. Again, efficacy of these novel devices has not been demonstrated with more prolonged catheterization [170,171]. This may reflect the greater importance of the catheter hub as a source of invading pathogens, compared with the skin at the insertion site, with more prolonged catheterization [72,150]. A number of studies have shown that the colonized hubs of IVDs are an important source of pathogens causing CR-BSIs [34], particularly with more prolonged duration of catheterization [33,72,95,150,172,173,174].
Central venous and arterial catheters also can become colonized hematogenously, from remote, unrelated sites of infection, but the evidence suggests that this occurs relatively less frequently than colonization from microbes at the insertion site or catheter hub [32,34,48,115,175,176] (Table 37-7), except in patients with short bowel syndrome [177].
Although infusate not infrequently becomes contaminated by small numbers of organisms, mainly skin commensals such as coagulase-negative staphylococci, with the exception of arterial catheters used for hemodynamic monitoring [31,32], endemic BSIs originating from contaminated infusate also appear to be uncommon in the United States, although common in many facilities with limited resources [32,109,115]. In contrast, contaminated infusate is the single most common
P.621
identified cause of epidemic nosocomial BSI [1,2], caused predictably by microorganisms capable of multiplying in parenteral glucose-containing admixtures, members of the tribe Klebsielleae (Klebsiella, Enterobacter and Serratia), B. cepacia, Burkholderia pickettii, or Citrobacter spp. [76]. Nearly 100 epidemics of infusion-related BSI since 1965 have been traced to contaminated infusate or IV medications, with microorganisms most frequently introduced during preparation or administration in the hospital (extrinsic contamination) or during its manufacture (intrinsic contamination).
|
TABLE 37-7 |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||
Analysis of risk factors predisposing to intravascular catheter-related infection by stepwise logistic regression of data from large, prospective studies of peripheral IV catheters [109], arterial catheters used for hemodynamic monitoring [31], multilumen CVCs used in ICU patients [32], or Swan-Ganz pulmonary artery catheters [175] shows that heavy cutaneous colonization of the insertion site is one of the most powerful predictors of catheter-related infection with all types of short-term, percutaneously inserted catheters (Table 37-8).
Pathogenesis
Examination of an infected IVD by scanning electron microscopy characteristically shows the surface covered by an amorphous film [150,176,189], presumably representing host proteins, with microcolonies of the infecting organism encased in a thick matrix of glycocalyx (slime), all comprising a “biofilm” [190] (Fig. 37-2). Studies of the pathobiology of prosthetic device-related infection have shown considerable differences in the capacity of microorganisms to adhere to various prosthetic materials. In vitro, catheters made of Teflon or polyurethane are more resistant to bacterial adherence, especially by staphylococci, than catheters made of polyethylene, polyvinylchloride, or, especially, silicone [191,192]. These differences are maintained if the experiments are done with previously implanted catheters or catheters precoated with specific plasma procatheters or catheters precoated with specific plasma proteins [193,194,195,196].
Initial attachment of Staphylococcus epidermidis directly to a catheter is mediated, in part, by the hydrophobicity of the strain [197] and by specific adhesins [198,199,200,201,202]. Initial attachment of S. aureus to catheters appears to be more dependent on the presence of preadsorbed plasma or tissue proteins such as fibronectin, thrombospondin, fibrin, vitronectin, and laminin [29,30,196,203]. Because many of these proteins are an integral part of thrombus formation, the presence of thrombus on the catheter surface also appears to promote adherence and catheter-associated infection [88,180,204,205]. Persistence of bacteria and fungi attached to the catheter surface appears to be promoted by surface exoglycocalyx [206,207]. Whereas subtherapeutic levels of antibiotics reduce microbial adherence [195,208], once microorganisms such as coagulase-negative staphylococci colonize a prosthetic surface, host defenses become secondarily impaired and are unable to spontaneously eradicate the infection [209,210]. Moreover, once associated with a foreign surface, microorganisms exhibit increased resistance to antimicrobials [211,212,213,214,215,216]. It should be no surprise that infections of prosthetic implants are difficult to cure with antimicrobial therapy alone, even with prolonged administration of high doses of bactericidal drugs.
P.622
P.623
|
TABLE 37-8 |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Bloodstream Infection from Contaminated Infusate
It took >10 years after the introduction of intravascular plastic catheters before they were ultimately recognized as an important source of serious iatrogenic infection; however, it required >40 years and the occurrence of epidemic gram-negative BSIs in hospitals across the United States in 1970 and 1971 [21] to bring about awareness that fluid given in intravascular infusions—infusate—also was vulnerable to contamination. It has become clear that although the majority of IVDR-BSIs derive from infection of the percutaneous infection wound or contamination of the catheter hub, contamination of infusate is the most common cause of epidemic IVDR-BSI [1]. From 1965 to 1978, 28/30 (85%) reported epidemics of infusion-related BSI were traced to contaminated infusate, with the organisms introduced during its manufacture (intrinsic contamination, which accounted for 7/20 epidemics) or during its preparation and administration in the hospital (extrinsic contamination, which accounted for the remaining 21 outbreaks) [1,2].
|
Figure 37-2 Scanning electron micrograph of an infected central venous catheter (x6,000). The amorphous matrix encasing the microcolonies of Staphylococcus epidermidis is glycocalyx (slime). |
Growth Properties of Microorganisms in Parenteral Fluids
The pathogens implicated in nearly all reported BSIs linked to contaminated infusate have been aerobic gram-negative bacilli capable of rapid growth at room temperature (25°C) in the solution involved [76]: for example, certain members of the family Enterobacteriaceae in 5% dextrose-in-water, and pseudomonads or Serratia spp. in distilled water. It must be emphasized that microbial growth in most parenteral solutions—the exception being lipid emulsion—actually is quite limited.
In 1970, we evaluated the ability of 105 clinical isolates from human HAIs, representing 9 genera and 13 species, to grow at room temperature (25°C) in 5% dextrose-in-water, the most frequently used commercial parenteral solution [217]. Of 51 strains of the tribe Klebsielleae—Klebsiella, Enterobacter, and Serratia spp.—50 attained concentrations of ≥100,000 cfu/ml within 24 hours, beginning with washed organisms at an initial concentration of 1 cfu/ml. In contrast, only 1/54 strains of other bacteria, including staphylococci, Escherichia coli, P aeruginosa, Acinetobacter spp., or Candida spp., showed any growth in 5% dextrose-in-water. With most microorganisms, even with a level of contamination exceeding 106cfu/ml of fluid, evidence of microbial growth was not visible to the unaided eye.
Review of studies of the growth properties of microorganisms in various commercial parenteral products has shown [76] that rapid multiplication in 5% dextrose-in-water appears limited mainly to the tribe Klebsielleae
P.624
and B. cepacia; in distilled water, P. aeruginosa, B. cepacia, Acinetobacter spp., or Serratia spp.; and in lactated Ringer's solution, P. aeruginosa, Enterobacter, or Serratia spp. Normal (0.9%) sodium chloride solution allows growth of most bacteria while supporting the growth of Candida spp. rather poorly. Candida spp. can grow in the synthetic amino acid–25% glucose solutions used for total parenteral nutrition (TPN), but only very slowly; most bacteria are greatly inhibited [218]. Most microorganisms grow rapidly in commercial 10% lipid emulsion for infusion (Intralipid®) [219,220]; in a study of 57 strains, we found that 12/13 bacterial species tested and Candida spp. multiplied in Intralipid almost as rapidly as in bacteriologic media [219]. Infections with Malassezia furfur also have been associated with administration of lipids [221,222,223]. This is not surprising because this dimorphic, lipophilic yeast cannot synthesize medium- and long-chain fatty acids and uses exogenous lipids, such as those found in supplemented TPN, for growth [220]. Use of TPN supplemented with lipids also has been shown to significantly increase the BSI risk by coagulase-negative staphylococci [224,225]. Epidemics also have been reported due to extrinsic contamination of a lipid-based anesthetic, propofol (Diprivan; Stuart Pharmaceuticals, Wilmington, Delaware, U.S.A.) [36]. This anesthetic agent, which initially did not contain a bacteriostatic agent, supported the exuberant growth of several gram-negative, gram-positive bacteria and Candida albicans [226].
The growth properties of most microorganisms in commercial parenteral admixtures and the vast aggregate experience with epidemic or endemic BSIs traced to contaminated infusate have shown that the identity of an organism causing nosocomial BSI can point strongly toward contaminated fluid as the plausible source: P. agglomerans, E. cloacae, Serratia marcescens, B. cepacia, or Citrobacter spp. cultured from the blood of a patient receiving infusion therapy should prompt strong suspicion of contaminated infusate—parenteral fluid or an IV drug (Table 37-6). Conversely, recovery of organisms, such as E. coli, Proteus, Acinetobacter, or staphylococci, all of which grow poorly if at all in parenteral admixtures, suggests strongly that the BSI is unlikely to be due to contaminated infusate.
Mechanisms of Fluid Contamination
As noted, the vast majority of published nosocomial BSIs traced to contaminated infusate occurred in an epidemic setting [1,2]. Parenteral fluids do, however, commonly become contaminated during administration in the hospital. Culture surveys of in-use IV fluids in the hospital have shown contamination rates in the range of 1% to 2% [227,228,229]. However, most of the organisms recovered from positive in-use cultures are common skin commensals that are generally considered of low virulence and grow poorly, if at all, in the parenteral admixture; the level of contamination (<10 cfu/ml) usually is far too low to produce clinical illness, even in the most compromised host. When contamination occurs with gram-negative bacilli capable of proliferation in the product to concentrations >102 to 103 cfu/ml, however, the risk of BSI and even septic shock becomes substantial.
The likelihood of fluid becoming contaminated during use is directly related to the duration of uninterrupted infusion through the same administration set and the frequency with which the set is manipulated. Microorganisms gain access from air entering bottles as they evacuate, from entry points into the administration set—during injections into the line or aspiration of blood specimens from the IVD through the line—or at the junction between the administration set and the catheter hub. Microorganisms capable of growth in fluid, once introduced into a running infusion, may persist in an administration set for many days despite multiple replacements of the bottle or bag and high rates of flow [21]; it appears more likely, however, that the majority of introduced contaminants are rapidly cleared from the running infusion by the continuous flow [227,228,229,230], especially if the organisms grow poorly in the fluid.
A healthcare worker may rarely encounter a filmy cloud in a glass IV bottle. Microscopic examination of the material reveals it is a filamentous fungus, such as Penicillium orAspergillus spp. Molds usually gain access to glass IV bottles through microscopic cracks long before the bottle is hung for use, and over the course of weeks or months grow to produce visible cloudiness or filmy precipitates. Fortunately, “fungus balls” in IV bottles have rarely resulted in systemic infection in patients receiving a mold-contaminated infusion [233].
The incidence of endemic nosocomial BSI caused by extrinsically contaminated IV fluid is not precisely known, but based on studies of the pathogenesis of device-related infection, is 5- to 10-fold lower than the incidence of endemic cannula-related BSI. Moreover, prospective studies of the optimal interval for periodic replacement of administration sets [227,228,229,230,231,232,233] (Table 37-9), which have involved cultures of infusate from large numbers of in-use infusions in an institution, have shown low rates of contamination and a very low risk of related BSI: meta-analysis of five studies in which >9,000 infusions in 5 hospitals were prospectively cultured, with no associated episodes of bacteremia or candidemia identified, yields an incidence of endemic BSI due to contaminated infusate of <1 episode per 2,000 IV infusions. It must be emphasized, however, that IV infusate can be identified as the source of BSI only if it is cultured. Because this rarely occurs in most hospitals, unless there is a cluster of BSIs—an epidemic—occurs, it is likely that most sporadic (endemic) BSIs caused by contaminated fluid go unrecognized or are attributed to the IVD.
Approximately 50% of the BSIs caused by arterial infusions used for hemodynamic monitoring stem from contamination of fluid in the infusion [31], perhaps because
P.625
these infusions consist of a stagnant column of fluid subjected to frequent manipulations, including frequent drawing of blood specimens. However, more recent studies have demonstrated that infusate contamination of hemodynamic pressure monitoring equipment is rare [234]. Over the past 20 years, there have been 28 epidemics of nosocomial BSI traced to contaminated fluid in arterial infusions used for hemodynamic monitoring [235,236,237,238]. Nearly all of these epidemics have involved gram-negative bacilli, particularlyS. marcescens, pseudomonads, or Enterobacter spp. that are able to multiply rapidly in the 0.9% saline commonly used in these infusions.
|
TABLE 37-9 |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Endemic BSIs resulting from transfusion of contaminated blood products have been rare, presumably because most blood products are routinely refrigerated, because contamination is low level, and because of universal awareness that blood products must be used promptly after removal from refrigeration [79,239]. BSI from contaminated whole blood is associated with adverse reactions in 50% of episodes, including fever (80%), rigors (53%), hypotension (37%), and nausea or vomiting (26%), with an associated mortality of 35% [79]. Overwhelming shock usually is due to contamination with massive numbers of psychrophilic (cold-growing) organisms such as Serratia spp., B. cepacia, S. maltophilia, Yersinia spp., or other uncommon, nonfermentative gram-negative bacilli, e.g., Flavobacterium species, in the contaminated unit [79,80,81,241]. Bacteria often have been visible on a direct Gram-stained smear of the product. Blood products should be infused immediately after they are removed from refrigeration. On completion of the transfusion, the entire delivery system should be replaced. If BSI is suspected of being related to a contaminated blood product, the entire infusion should be removed. Aliquots of the remaining product should be cultured aerobically and anaerobically on solid media at both 35°–37°C and 16°–20°C [76]. Platelet units may be stored at room temperature for 5 days before use and may be more prone to contamination with large numbers of microbial pathogens. As many as 10% of platelet pools used for transfusion are contaminated with bacteria [239]. Although most contaminants are skin flora [242], contamination with gram-negative bacilli has been reported [79,80].
The most important measures to prevent rare sporadic BSIs from contaminated in-use infusate are stringent asepsis during the preparation and compounding of admixtures in the hospital central pharmacy or on individual patient-care units, and good aseptic technique when infusions are handled during use, e.g., during injections of medications or changing bags or bottles of fluids. It also appears that replacing the administration set at periodic intervals can prevent the build-up of dangerous introduced contaminants and further reduce the risk of related BSI; during the large, nationwide U.S. epidemic in 1971 due to the contaminated products of one manufacturer, the empiric recommendations that the entire delivery system be routinely changed every 24 hours
P.626
and that at every change of the cannula all equipment be totally replaced resulted in a substantial reduction in epidemic BSIs [21]. Since that time, routinely replacing the delivery system at periodic intervals has been practiced in most North American hospitals as an important measure for reducing the hazard of contaminated infusate. However, in some instances, routine replacement at more prolonged intervals may be associated with epidemics, particularly in vulnerable patient populations and when the fluids infused promote microbial growth [243].
Epidemic Infusion-Related Bloodstream Infections
Outbreaks Due to Intrinsic Contamination
Since 1970, there have been >12 reported epidemics of infusion-related BSI caused by intrinsically contaminated infusate—blood products, IV drugs, or vacutainer tubes (Table 37-10)—illustrating the potential iatrogenic hazards of infusion therapy. The frequency and size of these outbreaks have declined since the late 1980s [2], reflecting appreciation of the importance of stringent quality control during the manufacturing process.
|
TABLE 37-10 |
||||||||||||||||||||||||||||||||||||||||||
|
The first and largest epidemic—and the outbreak that more than any other factor brought about wide-scale appreciation of the iatrogenic hazards of infusion therapy—had its onset several years ago when one U.S. manufacturer of large-volume parenterals began to distribute bottles of fluid with a new elastomer-lined screw cap closure [21]. By early 1970, the first episodes of infusion-related BSI caused by biologically characteristic strains of E. cloacae and P. agglomerans (designated Erwinia at the time) were reported to the CDC, although retrospective review subsequently showed that numerous hospitals had been experiencing epidemic BSIs for a number of months. Although it was established very early, virtually at the outset of the investigation, that epidemic BSIs resulted from contaminated IV fluids, the ultimate source of contamination—intrinsic contamination of the new closures—was not conclusively established until March, 1971. Between July 1970 and April 1971, 25 U.S. hospitals reported nearly ~400 episodes of infusion-related BSI to the CDC (Fig. 37-3). It is likely that there were >10,000 episodes nationwide. More than 20 microbial species, including P. agglomerans, were isolated from the closures of previously unopened bottles. Organisms were readily dislodged from the cap liner and introduced into IV fluid when bottles were handled under conditions duplicating normal in-hospital use. The appearance of epidemic BSIs in individual hospitals paralleled the distribution of the company's product with the new closures, and the epidemic was terminated only by a nationwide product recall in early April 1971.
Since 1975, numerous additional outbreaks have been reported from hospitals in a number of countries, all involving gram-negative bacilli and parenteral products shown to have been contaminated during manufacture [76,77,244,245,246,247,248,249,250,251,252,253,254,255,256]. Most have been of national scope. A large outbreak in Greece in 1981 [245] reaffirmed the findings of the large 1970 to 1971 U.S. outbreak [21] that screw-cap closures are not microbiologically safe for fluids used in medical care that must remain sterile. Outbreaks of pyrogenic reactions [248] and epidemic Pseudomonas spp. BSI [249] have been traced to intrinsically contaminated normal serum albumin, and an epidemic of BSI with
P.627
P.628
Ochrobactrum anthropi has been traced to organisms from contaminated rabbit antithymocyte globulin [250]. Most notably, during the past decade outbreaks of Pseudomonas spp. infection have been traced to intrinsic contamination of 10% povidone–iodine [251], an agent widely used worldwide for cutaneous antiseptic for preparation of the CVC insertion site [252]. Dilute chlorhexidine solution, which is increasingly being used for skin antisepsis, [32,151,152,153], may support the growth of bacteria, leading to epidemic BSI [148].
|
Figure 37-3 Nationwide outbreak of nosocomial bacteremias due to intrinsic contamination of one U.S. manufacturer's large-volume parenteral products. Three hundred ninety-seven cases of IV-associated bloodstream infection in 25 tabulated U.S. hospitals, occurring between July 1, 1970, and April 27, 1971, fulfilled criteria for epidemic cases. The epidemic was curtailed immediately in individual hospitals and nationally by a nationwide recall of the manufacturer's products. (From Maki DG, Rhame FS, Mackel D, Bennett JV. Nationwide epidemic of bloodstream infection caused by contaminated intravenous products. Am J Med 1976;60:47–485, with permission.) |
All of these outbreaks illustrate how subtle and insidious the factors that influence sterility can be. In many instances, there was no documented failure of the sterilization process. Instead, seemingly minor alterations in the manufacturing process resulted in contamination of individual units in the manufacturing plant after the sterilization stage [253].
Although intrinsic contamination is, fortunately, exceedingly rare, its potential for producing harm is great because of the large numbers of patients in multiple hospitals who may be affected. Also, direct contamination of infusate at the manufacturing level gives contaminants an opportunity to proliferate to dangerously high concentrations.
It seems likely that intrinsic contamination is a continuous source of infusion-related BSI, but of such low magnitude that the resulting BSIs are never identified as related to intrinsic contamination. Only when infusion-associated BSIs occur in epidemic numbers is intrinsic contamination likely to be suspected and proven. A substantial increase in the incidence of cryptogenic infusion-associated BSI, particularly with Enterobacter spp., pseudomonads, Burkholderia spp., or Citrobacter spp., should prompt immediate, in-depth studies to exclude intrinsic contamination. There are no clinical clues reliably to differentiate intrinsic from extrinsic contamination. BSI from contaminated fluid has the same manifestations and signs as CR-BSI and other nosocomial BSIs. The few clues to infusion-related BSI—absence of an obvious source of infection, its common occurrence in patients without a predilection to systemic infection, and the dramatic clinical response to discontinuing the infusion (Table 37-3)—do not differentiate between intrinsic and extrinsic sources of contamination. The distinction must be made epidemiologically.
If intrinsic contamination of a commercially distributed product is identified, or even strongly suspected, especially if clinical infections have occurred as a consequence, the local, state, and Federal (CDC and the Food and Drug Administration (FDA)) public health authorities must be immediately contacted. Unopened samples of the suspect lot or lots should be quarantined and saved for analysis.
Outbreaks Due to Extrinsic Contamination
Even when commercially manufactured products are sterile on arrival in the hospital, circumstances of hospital use can compromise that initial sterility. As previously noted, most sporadic infections resulting from infusion therapy, whether due to the cannula or contaminated infusate, are of extrinsic origin. Similarly, most reported epidemics have originated from exposure of multiple patients' infusions to a common source of contamination in the hospital [2,36,235,254,255,256,257,258].
Numerous outbreaks of infusion-related BSI have been caused by use of unreliable chemical antiseptics, or antiseptics such as aqueous benzalkonium and aqueous chlorhexidine used for cutaneous antisepsis [145,146,147,148] or, in more recent years, for decontaminating transducer components used in hemodynamic monitoring [235] (Table 37-10).
Despite the numerous reports of epidemic gram-negative BSIs deriving from contaminated disinfectants used for decontaminating reusable transducer components in hemodynamic monitoring during the 1970s, one third of all nosocomial BSI outbreaks investigated by the CDC between 1977 and 1987 were traced to contamination of infusions used for arterial pressure monitoring [256]. Since 1980, there have been 28 nosocomial BSI outbreaks associated with arterial pressure monitoring reported in the literature, nearly all caused by gram-negative bacilli, most frequently S. marcescens or Burkholderia spp. [235,236,237,238]. Two thirds of these epidemics were linked to failed decontamination of reusable transducer components. Epidemic organisms were most commonly found on metal transducer heads, in the interface between transducers and disposable chamber domes. Eight epidemics were traced to introduction of organisms into closed monitoring systems from external sources of contamination in the hospital, such as contaminated ice used to chill syringes for drawing arterialized blood for blood gas measurements, heparinized saline from multidose vials, and contaminated external devices to calibrate pressure-monitoring systems. The epidemic organisms were found on the hands of healthcare providers in at least nine outbreaks; however, most of the reports do not provide sufficient data to establish the precise mechanism of fluid contamination.
With all forms of infusion therapy, the connection between the administration set and the catheter must be secure. This is especially important with CVCs, where accidental disconnections can result in exsanguination or life-threatening air embolus or blood loss. In TPN, a faulty connection also may increase the risk of iatrogenic infection: one reported outbreak of 23 CR-BSIs caused by different strains of coagulase-negative staphylococci was linked to a manufacturing defect that resulted in hyperalimentation solution leaking from administration set–catheter connections and seeping under dressings, where it resulted in heavy bacterial overgrowth [259]. Another outbreak of coagulase-negative staphylococcal BSIs has been associated with excessive manipulation of a catheter delivery system because of air appearing in the IV pump tubing. This resolved when the IV pump was placed at or below the heart level of the patients and air entry into the tubing ceased [260].
P.629
During the 1970s, numerous outbreaks of gram-negative BSI, particularly with pseudomonads other than P. aeruginosa, were traced to contamination of dialysate in patients' hemodialysis machines [261] (see Chapter 24); however, improved quality control, the decontamination of reused dialyzer coils, and the widespread use of disposable dialyzers have resulted in a marked decline in the incidence of HAI outbreaks traced to contaminated dialysate [2].
Compounding of admixtures is another important means by which contamination can be introduced [262]. The greatest concern about this mode of contamination, especially if it occurs in the central pharmacy, is that a large number of patients may be exposed. Moreover, the delay between compounding and use provides opportunity for proliferation of introduced microorganisms to levels that can cause overwhelming septic shock when administered. Two large outbreaks of candidemia have been traced to contaminated solutions used for IV hyperalimentation [263,264]; in each outbreak, a vacuum system in the hospital's pharmacy used to evacuate fluid from bottles before introducing other admixture components was shown to be heavily contaminated by the epidemic strain of Candida spp. Presumably, organisms refluxed into bottles during compounding of the admixtures. In outbreaks traced to contaminants introducing during compounding, after compounding, bottles were permitted to stand at room temperature for up to 48 hours before use. The necessity for stringent attention to a BSI in central admixture programs cannot be overemphasized. Fluid admixtures should be used within 6 hours or immediately refrigerated.
Investigations of >100 epidemics [1,2] have documented contamination of in-use infusate or contamination of cannula insertion sites, deriving from a myriad of extrinsic sources in the hospital (Table 37-10). In many outbreaks, the hospital reservoir of the epidemic pathogen and even the mode of transmission eluded detection, but the microorganism was found in large numbers on the hands of healthcare providers caring for patients receiving infusion therapy and handling their infusions. Manipulations of the delivery system, especially the administration set, appear to provide a highly effective means for access of microorganisms to in-use infusate, as illustrated by HAI outbreaks across the United States traced to in-use contamination of the IV anesthetic, propofol (Diprivan®). The solution, when initially marketed did not contain a bacteriostatic agent. The anesthetic provided a rich medium for rapid microbial proliferation [226], and outbreaks of primary BSI or SSI with a variety of gram-positive and gram-negative organisms and yeasts were traced to in-use contamination of propofol administered in the operating room, because of poor aseptic technique, storage of opened vials at room temperature, and use of single vials for multiple patients [36,264,265]. Similarly, a veritable explosion of hospital outbreaks of nosocomial candidemia in the past 2–3 decades [265,266,267], primarily in ICUs, has been linked to carriage of the epidemic strain on the hands of healthcare providers handling vulnerable patients' IVDs and infusions.
|
TABLE 37-11 |
||||||||||||||||||||||||||||||
|
Approach to an Epidemic
If an epidemic is suspected, the epidemiologic approach must be methodical and thorough, yet expeditious. It is directed toward establishing the bona fide nature of the putative epidemic infections [269] and existence of an epidemic, defining the reservoirs and modes of transmission of the epidemic pathogens, and, most importantly controlling the epidemic quickly and completely. Control measures obviously are predicated on accurate delineation of the epidemiology of the causative pathogen (see Chapter 6).
The essential steps in dealing with a suspected outbreak of nosocomial BSI can be found in Table 37-11. To illustrate
P.630
the approach to an epidemic of infusion-related BSI, the epidemiologic investigation of an extraordinary outbreak that occurred in the University of Wisconsin Hospital and Clinics is recounted [35].
During a 2-week period in late March 1985, three patients in our university hospital acquired primary nosocomial BSI with a similar nonfermentative gram-negative bacillus. All three patients had had open heart surgery between March 11 and March 25 and became bacteremic 48–148 hours after operation.
The BSI pathogen in each patient was B. pickettii biovariant 1. The organism also was cultured from the IV fluid of two of the patients at the time because, serendipitously, during the outbreak most adult patients in the hospital receiving IV fluids were participating in a study of IV catheter dressings [109]; as part of the study protocol, specimens were routinely obtained from patients' IV fluid when the catheter was removed. Review of nearly 1000 cultures of IV fluid from the infusions of participants in the study since its outset 3 months earlier showed that three additional surgical patients operated on in March had had IV fluid cultures positive for B. pickettii biovariant 1, even though none had shown clinical signs of BSI. Molecular subtyping by restriction enzyme digestion and pulsed-field electrophoresis showed all six isolates to be identical. Three more patients who had been operated on in January had had IV fluid cultured positive for a similar nonfermentative gram-negative bacillus; although the three isolates were no longer available, the results of screening by AP-20E biochemical panel (API Analytab, Inc., Plainville, New York, U.S.A.) at the time were identical to those of the six patients with B. pickettii contamination of IV fluid, with or without associated BSI.
All of the patients had had multiple positive blood cultures and were in septic shock. B. pickettii had not been isolated from any local site of infection, such as the urinary tract, lower respiratory tract, or surgical site, in any of the patients.
Review of nosocomial BSIs over the preceding 7 years showed that B. pickettii had not previously been identified in blood cultures from our institution, indicating that the cluster of three episodes and six instances of contaminated infusate without BSI represented a true epidemic, and with the results of the subtyping, a common-source epidemic.
The CDC and the manufacturer were contacted: none of more than 70 NNIS hospitals had reported B. pickettii BSIs in the past year, and the manufacturer had never identified contamination with B. pickettii in its quality-control microbiologic sampling of fentanyl before distribution, or received any complaints from users about suspected contamination of their fentanyl. Moreover, survey of surrounding hospitals that also used the manufacturer's fentanyl revealed none experiencing nosocomial BSIs with B. pickettii.
A case–control study comparing the 9 infected patients, all of whom had had recent surgery, and 19 operated patients who had had negative IV fluid cultures in the IV dressing study, showed that all 9 cases but only 9/19 operated controls had received fentanyl intravenously in the operating room (p = 0.05); the mean total dose given to the 9 case-patients was far greater than that given to control-patients who received the drug (3,080 vs. 840 µg, p <0.001).
At the time, fentanyl was used at the University of Wisconsin Hospital only in the operating rooms (ORs) as part of balanced anesthesia. The drug was received in 20-ml ampules from the manufacturer and each week, one of three pharmacy technicians, by rotation, pre-drew into sterile syringes all fentanyl likely to be needed the following week in the operating rooms. Each day, one of the technicians delivered enough pre-drawn syringes to the ORs to meet the needs of the patients being operated on that day. Cultures of pre-drawn fentanyl in syringes in the central pharmacy, prompted by the findings of the case–control study, showed that 20/50 (40%) 30-ml syringes sampled were contaminated by B. pickettii in a concentration >104 cfu/ml; none of 35 5-ml or 2-ml syringes showed contamination (p <0.001).
Extensive culturing in the central pharmacy was negative for evidence of environmental contamination by B. pickettii, with one exception: B. pickettii biovariant 1, with an identical antimicrobial susceptibility pattern and restriction enzyme fragment pattern to the epidemic strain recovered from blood cultures or patients' IV infusions, was cultured in a concentration of 28–80 cfu/ml from five specimens of distilled water drawn from a tap in the central pharmacy. The epidemic strain was shown to multiply well in the fentanyl solution, attaining concentrations >104 cfu/ml within 48 hours.
A second case–control study strongly suggested that the epidemic was caused by theft of fentanyl from 30-ml syringes by a pharmacy staff member and replacement with distilled water that the individual thought was sterile, but that, unfortunately, was contaminated by B. pickettii. The pharmacy member resigned early in the investigation. On April 29, the hospital's system for providing fentanyl and other narcotics to the ORs was changed; narcotics were no longer pre-drawn into syringes in the central pharmacy, but were delivered to the ORs in unopened vials or ampules; anesthesiologists' orders for narcotics are filled by a staff pharmacist assigned to the OR. There have been no further B. pickettii BSIs since March 25, 1985, and cultures of >6,000 samples of hospitalized patients' IV fluid in research studies since that time have shown no further contamination by B. pickettii.
This outbreak illustrates the power of epidemiology, e.g., case–control analyses, to identify the probable cause of an epidemic. It further illustrates the potential for contamination of parenteral drugs or admixtures and the extraordinary range of epidemiologic mechanisms of nosocomial BSI deriving from such contamination.
P.631
Strategies for Prevention
Extensive guidelines for prevention of catheter-related infection have been published [12]. Specific interventions are discussed in the following sections.
Aseptic Technique
To accord it due respect, any device for vascular access must be thought of in fundamental terms as a direct conduit between the external world, with its myriad of microorganisms, and the bloodstream of the patient. Vigorous hand hygiene, ideally with an antiseptic-containing preparation, and gloving always must precede the insertion of a peripheral IV cannula and also should precede later handling of the device or the administration set [270]. Furthermore, sterile gloves should be routinely used during the insertion of peripheral IV cannulas in high-risk patients, such as those with severe burns. Sterile gloves are strongly recommended for placement of all other types of IVDs—arterial and all CVCs—which are associated with a higher risk of associated BSI [12].
Although there has been considerable controversy as to the level of barrier precautions necessary during insertion of a CVC, in a study of Swan-Ganz pulmonary artery catheters, the use of maximal barrier precautions—sterile gloves, a long-sleeved sterile surgical gown, a surgical mask, and a large, sterile sheet drape covering the patient, as contrasted to the use of sterile gloves, a surgical mask, and a small, fenestrated drape—was associated with a twofold lower risk of infection [175]. Despite the fact that Swan-Ganz catheters inserted in the ICU using maximal barrier precautions remained in place an average of 22 hours longer than catheters inserted in the OR (inserted with lesser barrier precautions), were more frequently placed in infected patients, and were used more frequently for TPN, catheters inserted in the ICU using maximal barrier precautions were much less likely to be contaminated under the external protective sheath or infected than catheters inserted in the OR by anesthesiologists using lesser barrier precautions. The efficacy of maximal barrier precautions in the prevention of nontunneled CVC-related infection has been demonstrated [271,272]. In this study, the incidence of CR-BSI was 6.3 times higher in those patients who were prospectively randomized to have their catheters inserted with only sterile gloves and small sterile drapes, compared with those patients whose catheters were inserted with maximal barrier precautions (mask, cap, sterile gloves, gown, and large, sterile drape) [272]. In another study, use of maximal barrier precautions with CVC insertion, in addition to a mandatory 5-minute scrub of the insertion site, reduced the incidence catheter colonization from 36% to 17% [273]. Considering that of all IVDs, CVCs are most likely to produce nosocomial BSI, a strong case can be made for mandating maximal barrier precautions during the insertion of such devices, particularly the use of a long-sleeved surgical gown and large, sterile sheet drape, to minimize touch contamination, in addition to sterile gloves, the use of which should be routine [12,272].
Inappropriate catheter care is an independent risk factor for catheter-related infections [186]. Not surprisingly, the use of special IV therapy teams, consisting of trained nurses or technicians to ensure a high level of aseptic technique during catheter insertion and in follow-up care of the catheter, has been associated with substantially lower rates of catheter-related infection [108,273,274,275,276,277,278,279,280,281,282] (Table 37-12). Such teams are highly cost effective, reducing the costs of complications of infusion therapy nearly 10-fold [281,282].
In the absence of a dedicated IV team, some investigators have carried out intensive educational programs in catheter care. In one study, this led to improved care overall and a concomitant reduction of colonization of the catheter insertion site; however, the incidence of catheter hub colonization was unchanged [283]. In some U.S. hospitals, all CVCs, particularly those dedicated to TPN, are cared for by such teams. Other investigators also have shown that institutions can greatly reduce their CR-BSI rate by scrutiny of catheter care protocols and more intensive education and training of nurses and physicians [284]. The importance of adequate staffing of nurses to care for patients with CVCs has recently been demonstrated [286]. After controlling for other risk factors associated with catheter-related infection in a logistic regression model, as the patient-to-nurse ratio doubled due to nurse understaffing in an ICU, the risk of CVC-BSI increased dramatically (OR 62). This seminal observation suggests that in this era of fiscal restraint in healthcare, cost-cutting measures which lead to understaffing of personnel aimed at caring for IVDs will ultimately increase cost and increase the risk of HAIs in today's hospitalized patients.
Cutaneous Antisepsis
Given the evidence for the important role of cutaneous microorganisms in the genesis of many IVDR-infections, measures to reduce cutaneous colonization of the insertion site would seem of the highest priority, particularly the use of chemical antiseptics of the site. Worldwide, an iodophor such as 10% povidone–iodine commonly is used [252]. The lack of published comparative trials of cutaneous antiseptics to prevent catheter-related infection prompted the following prospective investigation: 668 patients' central venous and arterial catheters in a surgical ICU were randomized to 10% povidone–iodine, 70% alcohol, or 2% aqueous chlorhexidine with alcohol for antisepsis of the site before insertion and site care every other day thereafter [32]. Chlorhexidine was associated with the lowest incidence of infection; of the 14 infusion-related BSIs, 1 was in the chlorhexidine group and 13 were in the other two groups (odds ratio, 0.16; p = 0.04) (Table 37-13). This study suggests that the use
P.632
P.633
of 2% chlorhexidine, rather than 10% povidone–iodine or 70% alcohol, for cutaneous antispesis before insertion of an IVD and in post-insertion site care can substantially reduce the incidence of IVD-related infection. Other investigators also have found that use of aqueous chlorhexidine to prepare the catheter insertion site is associated with a lower incidence of catheter-related infection compared with povidone–iodine [151,152,153].
|
TABLE 37-12 |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
TABLE 37-13 |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
In a historical analysis of the impact on the incidence of CR-BSI of using different antiseptics for site care and disinfection of tubing connections in a home TPN program, 0.58 episodes per catheter-year were observed during use of 10% povidone–iodine as contrasted to 0.26 to 0.28 episodes per catheter-year during use of a 0.5% to 2% tincture of iodine or 0.5% tincture of chlorhexidine [287]. Prospective, randomized studies comparing the blood culture contamination rate using povidone–iodine vs. iodine tincture to prepare the puncture site, use of iodine tincture was associated with a contamination rate one-half that of the povidone–iodine group [288].
“Defatting” the skin with acetone is still widely practiced in many centers as an adjunctive measure for disinfecting CVC sites, especially in TPN; however, it was found to be of no benefit whatsoever in a prospective, randomized trial [114]. In contrast, the use of acetone was associated with greatly increased inflammation and discomfort to the patient.
Topical Antimicrobial Ointments
In theory, application of topical antimicrobial agents to the catheter insertion site should confer some protection against microbial invasion. Clinical trials of topical polyantibiotic ointments (polymyxin, neomycin, and bacitracin) on peripheral venous catheters have shown only moderate or no benefit [154,290], and the use of polyantibiotic ointments has been associated with an increased frequency of Candida spp. infections [116,154]. In prospective, randomized trials, application of the topical antibacterial, mupirocin, which is active primarily against gram-positive organisms, to catheter insertion sites has been associated with a significant reduction in CVC colonization, but not arterial or peripheral catheter colonization. Without colonization by Candida spp., the impact on CR-BSI could not be assessed [291]. However, widespread use of mupirocin at catheter insertion sites may lead to resistance [292], and for this reason it is not recommended for routine application to the CVC insertion site.
There have been two prospective studies of topical povidone–iodine ointment applied to CVC sites; one large, randomized trial in a surgical ICU showed no benefit [293], but a more recent comparative trial with subclavian hemodialysis catheters showed a fourfold reduction in the incidence of hemodialysis CR-BSI [122].
Dressings
The importance of the cutaneous microflora in the pathogenesis of IVDR infection might suggest that the dressing applied to the catheter insertion site could have considerable influence on the incidence of catheter-related infection. When used on vascular catheters, transparent dressings permit continuous inspection of the site, secure the device reliably, and are generally more comfortable than gauze and tape. Moreover, transparent dressings permit patients to bathe and shower without saturating the dressing. Clinical trials of these dressings have been prompted by the knowledge that cutaneous occlusion with tape or impervious plastic films results in an explosive increase in cutaneous microflora, with overgrowth of gram-negative bacilli and yeasts [293]. Although polyurethane dressings are semipermeable—impervious to extrinsic microbial contaminants and liquid-phase moisture, and variably permeable to oxygen, carbon dioxide, and water vapor—and studies in healthy volunteers have shown little effect of these dressings on the cutaneous flora [294], a meta-analysis has raised concern that these dressings could increase cutaneous colonization and the risk of catheter-related infection [295].
Transparent polyurethane dressings are more expensive than gauze and tape, and to obviate the issue of greater cost and to increase convenience, many users leave transparent dressings on for prolonged periods, for up to 7 days or even longer. It has been questioned whether transparent dressings left on for prolonged periods might increase the risk of catheter-related infection. There have been a number of trials comparing polyurethane dressings with gauze and tape on peripheral venous catheters [14,15,107,109,296,297,298,299,300]. Three trials [296,297,298] have found significantly higher rates of catheter colonization with transparent dressings left on indefinitely. Other investigators also found a higher rate in catheters dressed with a transparent dressing, but only during the summer months [107]. In additional studies, however, significant differences were found [14,15,109,299,300]. Rates of catheter colonization in all of these trials have been low with all dressings, in the range of 1.6%-8.5%. Only 3 CR-BSIs were identified among the nearly 4,000 catheters studied in all of the reported trials.
In a prospective, randomized trial of various dressings used with 2,088 Teflon peripheral IV catheters [109], the transparent polyurethane dressing studied, left on for the lifetime of the catheter, was not associated with increased cutaneous colonization under the dressing or an increased rate of catheter colonization, compared with the control gauze and tape dressing; however, there were no CR-BSIs. Cutaneous colonization was not heavier under transparent dressings during the spring and summer months, as previously described [107], perhaps because the hospital was air conditioned. Multivariate analysis showed
P.634
cutaneous colonization of the insertion site (relative risk of infection, 3.9) and moisture under the dressing (relative risk, 2.5) to be significant risk factors for catheter-related infection (Table 37-8). These data indicate that it is probably not cost effective to redress peripheral IV catheters at periodic intervals, and that for most patients either sterile gauze or a transparent dressing can be used and left on until the catheter is removed [12].
Studies of transparent dressings on short-term, noncuffed central venous and/or pulmonary artery catheters have yielded conflicting results [34,182,300,301,302,303,304,305,306,307,308], in part reflecting differences in study protocols (e.g., the use of topical antimicrobial ointments under the dressing in the control gauze group, but not in the transparent dressing group) and different dressings studied. One group of investigators [302] reported a threefold increase in infectious complications associated with subclavian catheters using transparent dressings for ≤7 days compared with gauze replaced three times weekly, but the difference did not achieve statistical significance. Others [182] have found a much higher rate of CR-BSI using transparent dressings compared with gauze and tape (16% vs. 0%). In a similar trial performed in an ICU, there were no significant differences in catheter-related infection with transparent dressings compared with gauze dressings when the transparent dressing was changed every 2 days [305]; however, in these high-risk ICU patients, we observed a significant build-up of skin flora, associated with a 50% increase in catheter-related infection, when transparent dressings were left on for up to 7 days between changes, suggesting that if used on CVCs in ICU patients, the dressing studied may need to be replaced more frequently. Other prospective trials, which in aggregate studied hundreds of CVCs in high-risk patients, many of whom were receiving TPN through the catheter, did not find an increased risk of catheter-related infection associated with transparent dressing left on for a prolonged duration, as compared with gauze and tape replaced more frequently [34,300,301,303,304,306,307,308]. In a large, prospective, randomized clinical trial, we found no difference in the incidence of catheter colonization or CR-BSI in patients whose catheters were covered with gauze and tape dressing changed every 2 days, conventional polyurethane dressing, or a polyurethane dressing with a high moisture vapor transmission rate, both replaced every 5 days [34]. However, at the time of catheter withdrawal, insertion site colonization was greater under both polyurethane dressings compared with the gauze and tape dressing group. There was no significant difference in colonization between the polyurethane dressing groups. Similarly, the incidence of CVC-BSI was no different for patients whose catheters were dressed with gauze and tape every 2 days compared with transparent dressing every 5 days in a multi-center trial however, colonization was significantly more common in the polyurethane dressing group [308].
Two randomized studies of the use of transparent dressings on surgically implanted, cuffed Hickman or Broviac catheters have been reported in which microbiologic data were provided [305,309]; in both trials, one in renal transplant patients [305] and the other in bone marrow transplant recipients [309], the transparent dressing studied provided satisfactory cover, and even when left on for prolonged periods, for up to 5–7 days, was not associated with a significantly increased risk of exit site or tunnel infection, or of CR-BSI. These studies of central venous, pulmonary artery, or long-term tunneled catheters suggest that either transparent or gauze and tape dressings can be safely used to cover the insertion site of these devices.
There have been only two reported studies of transparent polyurethane dressings with arterial catheters [15,305]. In one prospective study of dressings for arterial catheters used for hemodynamic monitoring in a surgical ICU, the use of the transparent dressing, even replaced every other day, was associated with a fivefold increased incidence of CR-BSI, compared with gauze and tape [305]. The greatly increased risk of infection associated with transparent dressings used on arterial catheters found in this study may reflect the presence of macroscopic blood in the puncture wound, under arterial pressure, which is common under transparent dressings on arterial catheters; if the blood cannot be cleared, it may provide a rich medium for microbial proliferation, which can result in infection of the catheter. As a consequence, gauze and tape is preferred to transparent dressings for arterial catheters.
Changing Catheters Over a Guidewire
The Seldinger technique, in which the vessel is identified and entered percutaneously with a fine-gauge needle and cannulated with a guidewire passed through the needle, after which the cannula is guided into the vessel over the guidewire, has been a major advance, permitting vessels to be cannulated with large catheters with much less risk of vascular injury and, in the case of subclavian or internal jugular CVCs, pneumothorax, and with less manipulation and potential for contamination. To avoid iatrogenic pneumothorax and other mechanical complications associated with percutaneous insertion of a new cannula, particularly a CVC, new catheters are commonly placed over a guidewire in the site of an old catheter [118]. Prospective, randomized clinical trials have shown that routinely replacing CVCs over guidewires is unnecessary [118,158,310]. In the largest of these studies [118], the incidence of CR-BSI per 1,000 catheter days was nearly twofold higher in patients randomized to the guidewire groups, and 75% of CR-BSIs and fungemias occurred within 72 hours of guidewire exchange or catheter insertion. In a study performed in a pediatric ICU, CVCs were left in place without routine guidewire exchange [311]. Despite prolonged catheterization, the incidence density of catheter-related infection did not increase with the duration of time the catheters were left in situ. Similar results were found in a study of adult oncology patients
P.635
whose noncuffed, non-tunneled CVCs were left in place for a mean of 136 days [125]. These data suggest that routine guidewire exchange of CVCs is unnecessary in patients who are without unexplained fever and without induration or drainage from the catheter insertion site. In two prospective studies of central venous and Swan-Ganz pulmonary artery catheters, the incidence of catheter-related infection was not significantly different among patients whose catheters were routinely changed over guidewires or left in place [118,310]. However, the data derived from all prospective clinical trials of Swan-Ganz pulmonary artery catheters demonstrate that the incidence of BSI rises sharply on the fifth day of catheterization [48]. Because of these conflicting results, firm recommendations regarding the safe duration of Swan-Ganz pulmonary artery catheters cannot be made at this time.
One prospective study of arterial catheters found that there was a greater incidence of CR-BSIs when catheters were routinely exchanged over guidewires compared with catheters removed after 7 days and inserted into a new site [31]. Other studies have shown that the incidence density of arterial catheter-related infection did not increase with the duration of time the catheters were left in place [312,313], and in three prospective studies in which the catheters, transducers, and plasticware were not routinely changed [314,315,316], the incidences of catheter colonization (2.9%) and CR-BSI (0.2%) were quite low and comparable with those in other prospective studies. Therefore, these data suggest that arterial catheters, similar to CVCs, may not need to be replaced at routine intervals as long as signs of localized infection are absent and the patient is without unexplained fever. It is important to remember that all invasive devices, including intravascular catheters of any type, increase the risk of infection, and their need should be assessed on a daily basis. In one study, nearly 20% of patients on medical wards had idle catheters for ≥2 consecutive days, and 20% of all patient-days of IV catheter use were idle and unnecessarily increasing the risk of catheter infection in these patients [317]. In a follow-up study, intensive educational efforts reduced the incidence of idle catheters [318].
If it is considered desirable to replace a central venous or arterial catheter because it has been in place for a prolonged period and there is suspicion of infection (e.g., unexplained fever), it is not unreasonable to replace the catheter in the same site over a guidewire if the patient has limited sites for new access or would be at high risk for the percutaneous puncture required for placement of a new catheter in a new site (e.g., has coagulopathy or is morbidly obese). However, it is imperative that the same meticulous aseptic technique that should be mandatory during insertion of any new catheter must be employed, including the routine use of sterile gloves and a sterile drape, and for CVCs, a sterile gown as well. After vigorously cleansing the site and the old catheter with the antiseptic solution, inserting the guidewire, removing the old catheter, and cleansing the guidewire and site once more with the antiseptic solution, the operator should reglove and redrape the site because the original gloves and drapes are likely to be contaminated from manipulation of the old catheter. After regloving and repreparing the site, the new catheter can be inserted over the guidewire.
It also is important routinely to culture the old catheter after guidewire exchange and, if the patient is febrile or shows other signs of BSI, to obtain blood cultures. If these cultures demonstrate that the old catheter was colonized, the new catheter just placed in the old site should be immediately removed to prevent progression to CR-BSI (or perpetuation of ongoing CR-BSI) because the new catheter has been inserted into an infected tract. Need for continued access would mandate placement of a new catheter in a new site. If, on the other hand, culture of the old catheter is negative, it has been possible to preserve access and to examine the initial catheter microbiologically and exclude it as the cause of fever or BSI, without subjecting the patient to the hazards associated with percutaneous insertion of a new catheter.
If the old insertion site is inflamed at the outset, especially if it is purulent, or the patient shows signs of sepsis that might be originating from the catheter, or the catheter recently has been shown to be infected by quantitative blood cultures drawn through the catheter, it is strongly recommended that a new catheter not be inserted over a guidewire into an old, potentially infected site.
The Effect of Subcutaneous Tunnel Insertion of Central Venous Catheters
Subcutaneous tunnel insertion of CVCs has traditionally been carried out in an effort to reduce the risk of catheter-related infection. In a prospective, randomized study in immunocompromised patients, the incidence of CR-BSI was the same in those patients whose catheters were tunneled, compared to those whose catheters were not [318]. In a randomized trial of catheters used for administering TPN [279], the incidence of catheter sepsis was reduced with tunneling of catheters before, but not after, a trained IV nurse assumed complete responsibility for catheter care. In a more recent, prospective, randomized study, tunneling CVCs inserted into the internal jugular vein dramatically reduced the incidence of CR-BSI (RR 0.19 [320]). However, this may be difficult to extrapolate to the U.S. experience, since most tunneled catheters in the United States are inserted in the subclavian, rather than the internal jugular vein. Also, no blood was drawn through the catheters in this study, which is common practice in the United States. This may have led to a lower incidence of hub-related BSI, further magnifying the difference in tunneled and non-tunneled catheter groups with regards to BSI. In another study [125], non-tunneled CVCs in immunocompromised patients inserted for prolonged periods of time also had a very low incidence
P.636
of BSI when the catheters were cared for by specialized IV nurses. Therefore, CVCs cared for by specialized IV nurses may not need to be tunneled in an attempt to reduce the incidence of catheter-related infection. The utility of tunneled subclavian CVCs in situations without specialized IV nursing teams requires further study.
Measures Aimed at the Delivery System
Numerous studies have shown that most CR-BSIs are caused by infections contaminated infusate [1,2]; however, infusate can occasionally become contaminated and cause endemic bacteremia or fungemia [32,33,34,115,175]. If an infusion runs continuously for an extended period, the cumulative risk of contamination increases, and further, there is increased risk that the contaminants can grow to dangerously high concentrations that will result in BSI in the recipient of the fluid. For nearly 20 years, most U.S. hospitals routinely replaced the entire delivery system of patients' IV infusions at 24- or 48-hour intervals [1,2], to reduce the BSI risk from extrinsically contaminated fluid. Prospective studies (Table 37-9), however, now indicate that IV delivery systems do not need to be replaced more frequently than every 72 hours, including infusions used for TPN or any infusions in ICU patients [227,228,230]; extending the duration of use allows considerable cost savings to hospitals [228]. Other prospective studies have demonstrated that replacement of the IV delivery system every 96–120 hours did not increase the incidence of catheter-related infection [312,321]. It is important to remember that replacement of the IV delivery system at more prolonged intervals may predispose to epidemic BSI, particularly when the fluids infused promote microbial growth [243].
Three clinical settings might be regarded as exceptions to using 72 hours as an interval for routine set change: 1) during administration of blood products or 2) lipid emulsions, or 3) if an epidemic of infusion-related BSI is suspected. In these circumstances, it is most prudent that administration sets be changed routinely at 24- or 48-hour intervals. Minute amounts of blood buffers acidic solutions and provides organic nutrients that greatly enhance the ability of most microorganisms to grow in parenteral fluids [217]. Moreover, most hospital pathogens, including coagulase-negative staphylococci, some gram-negative bacilli, Candida spp., or M. furfur, grow rapidly in commercial lipid emulsion [219,220,225,243], and BSI outbreaks have been associated with administration of lipid emulsion [243,246,323].
Studies suggest that the infusion system, including the administration set and other delivery components, for hemodynamic monitoring may not need routine replacement as long as the catheter insertion site is without induration or discharge and the patient is without an unresolved source of fever [234,314,315,316].
The type of infusion pump and delivery system used may affect the incidence of catheter-related infection. Line breaks associated with some infusion pumps and delivery systems appear to be susceptible to air entry into the tubing, leading to greater manipulation, and at one institution this was associated with an outbreak of coagulase-negative staphylococcal BSI [260]. Some systems require significantly fewer line breaks than others [324], and this may lessen the risk of catheter-related infection because an excessive number of line breaks per day has been demonstrated to be an independent risk factor for catheter-related infection [188].
Needleless IV systems have come into widespread use in an effort to reduce the risk of exposure to bloodborne pathogens. However, the risk of CR-BSI associated with use of these systems has not been systematically addressed in prospective, randomized trials. A number of studies have found that these systems may actually increase the risk of catheter-related infection [258,325,326,327,328,329]. A unifying problem is the inability to clean the inner components of these systems; once they become contaminated, bacteria and fungi can proliferate to large numbers, leading to intraluminal seeding of the blood [325,326]. Prospective studies are needed to determine the safety of needleless systems with regard to needlestick injury, catheter hub colonization, and BSI.
Terminal in-use membrane filters continue to be advocated as a means of reducing the hazard of contaminated infusate. However, filters must be changed at periodic intervals and can become blocked, leading to added manipulations of the system and, paradoxically, greater potential for contamination [330,331]. Some commercial in-line filters may also permit the passage of endotoxin [331,332]. The increased risk for phlebitis associated with the administration of IV antibiotics may be reduced by removing the microparticulates that are associated with compounding these drugs with 0.22 or 0.44 µm in-line filters [334]; however, not all randomized trials have shown a substantial reduction in phlebitis with the use of in-line filters [335]. Moreover, filters are expensive, and their use adds substantially to the costs of phlebitis from microparticulates. Few controlled, prospective clinical trials have been done to assess the effect of in-line filters on the incidence of serious catheter-related infections. In these studies, small numbers of patients were enrolled and the outcomes were variable [336,337]; however, studies carried out in animals suggest that with heavily contaminated infusate, in-line filters reduce mortality [337]. Large, prospective, double-blind clinical studies establishing their efficacy and cost effectiveness are needed before their routine use can be advocated, especially as a control measure for prevention of rare sporadic BSIs resulting from extrinsic contamination of infusate [12].
Innovative Technology
The development and application of novel technology holds the greatest promise for a quantum reduction in
P.637
the risk of infusion-related BSI, especially infection due to the percutaneous device used for intravascular access: innovations in the design or construction of the infusion apparatus that implicitly deny access of microorganisms to the system or that prevent organisms that might gain access from proliferating to high concentrations or colonizing the implanted cannula can obviate poor aseptic technique or undue patient vulnerability.
In a burn model, silver-coated dressings reduced the tissue penetration of P. aeruginosa in animals and may reduce the risk of catheter-related infections [339]. Previous studies of incorporating an antiseptic, namely povidone–iodine into a transparent catheter dressing to suppress cutaneous colonization under the dressing have been disappointing [109]. However, in view of the superiority of chlorhexidine over povidone–iodine for cutaneous antisepsis of vascular catheter sites [32,151,152,153], incorporation of chlorhexidine into a dressing's adhesive has been prove more effective. A chlorhexidine-impregnated urethane sponge composite (Biopatch; Johnson & Johnson, Arlington, Texas, U.S.A.) has been shown significantly to reduce epidural catheter colonization, from 29% to 4% [340]. In a nonrandomized clinical trial using historical controls, use of the composite sponge reduced the incidence of CVC-BSIs from 21 to 13 per 1,000 catheter-days [341]. In two prospective, randomized studies in pediatric patients, this technology has been shown to significantly reduce skin colonization, be at least equivalent to gauze dressings in preventing CR-BSIs and to reduce local infections [342,343]. In a randomized control trial, we found that this technology in adults reduced both CR-BSIs (2.4% vs 6.1%) and local site infections (16.4% vs. 23.9%) compared to gauze dressings. [344]
A tissue–interface barrier (VitaCuff; Vitafore Corporation, San Carlos, California, U.S.A.) has been developed that incorporates aspects of the technology of Hickman and Broviac catheters; the device consists of a detachable cuff made of biodegradable collagen to which silver ion is chelated (Fig. 37-4). The cuff can be attached to CVCs immediately before insertion. After insertion, subcutaneous tissue grows into the collagenous matrix, anchoring the catheter and creating a barrier against invasive organisms from the skin. The silver ion provides an additional chemical barrier against introduced contamination, augmenting the mechanical barrier. However, this device affords protection against extraluminal, not intraluminal, migration of microbial pathogens into the bloodstream. In most prospective, randomized clinical trials of short-term CVCs, the incidence of catheter colonization and CR-BSI was reduced with use of the cuff [116,117,159] (Table 37-14). The cuff did not confer protection, however, against infection with catheters inserted over a guidewire to old sites [116]. A cost-benefit analysis shows that if an institution's rate of short-term CVC-BSI is in the range of 2%-3%, use of the cuff should prove cost effective [116]. In prospective studies of patients in whom the duration of central venous catheterization was >2 weeks [161,162,346], cuff use was not found to reduce the incidence of catheter colonization or CR-BSI. This may be due to the greater importance of hub colonization as a source of catheter-related infections with more prolonged catheterization [72,150]. Therefore, use of the cuff should be considered in patients with an expected duration of catheterization of <2 weeks.
|
TABLE 37-14 |
||||||||||||||||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||||||||||||||
|
Figure 37-4 Schematic depiction of a silver-impregnated, tissue barrier cuff (VitaCuff) and the cuff attached to a central venous catheter in situ. It is important for the cuff to be positioned at least 0.5 to 1.0 cm below the surface of the skin and for the catheter to be well immobilized, preferably with a skin suture, to prevent extrusion. (From Maki DG, Cobb L, Garman JK, Shapiro J, Ringer M, Helgerson RB. An attachable silver-impregnated cuff for prevention of infection with central venous catheters. A prospective randomized multicenter trial. Am J Med 1988;85:307–314.) |
||||||||||||||||||||||||||||||||||||||||||||||||||
Given the multiplicity of potential sources for infection of an IVD and the importance of adherence of microorganisms to the catheter surface in the pathogenesis of infection, it would seem logical that the best strategy
P.638
for prevention might be to develop a catheter material implicitly resistant to colonization. It has been demonstrated that hydrophilic catheters are less likely to become colonized by bacteria in in vitro assays [347,348]. Binding a nontoxic antiseptic or antimicrobial to the catheter surface or incorporating such a substance into the catheter material itself might prove to be the most effective technologic innovation for preventing device-related infection. In a prospective, randomized clinical trial of central venous or arterial catheters in a surgical ICU, catheters coated with cefazolin bonded to the surface with a cationic surfactant were associated with a sevenfold reduction in colonization of the catheter; however, there were no CR-BSIs identified in the study population [163]. More recently, catheters coated with minocycline and rifampin have been shown significantly to reduce the incidence of catheter colonization and CR-BSIs [164]. Widespread use of these devices is also tempered by the risk of the development of resistance to these valuable antibiotics [349]. We have studied a novel CVC in which the catheter material itself, polyurethane, is impregnated with minute quantities of silver sulfadiazine and chlorhexidine (Arrowgard; Arrow International, Reading, Pennsylvania, U.S.A.); in a randomized, comparative trial in 402 patients in a surgical ICU, antiseptic catheters were twofold less likely to be colonized and fourfold less likely to produce BSI [168]. Adverse effects from the test catheter were not seen. In a prospective, randomized study, other investigators also have found that this device reduced the incidence of catheter-related infection [169]; however, with more prolonged catheterization, efficacy appears to wane [170,171] (Table 37-15). Silver-coated catheters also have been shown to reduce the incidence of catheter-related infection [165], as have catheters bound with Benzalkonium or heparin [166,167]. Strategies that hold the greatest promise are those that will change the IVD itself in a novel way that is unlikely to promote resistance to antibiotics or antiseptics. One innovative approach is to apply an electric current to the catheter [350,351]. In vitro studies with these catheters demonstrate that they inhibit the growth of bacteria and fungi [350] and are resistant to colonization, and that application of an electric current sterilizes colonized catheters [350]. In an animal model of S. aureus catheter-related infection, these catheters were more effective in preventing infection than the silver sulfadiazine-chlorhexidine-impregnated catheters [351].
|
TABLE 37-15 |
||||||||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||||||
Strategies aimed at reducing hub-related BSI have been devised, and in a clinical trial, a novel hub incorporating an iodine tincture reservoir reduced the incidence of BSI fourfold [352]. Recently, in an in vitro study in which we contaminated the surface of mechanical valve needleless connectors and then disinfected them with alcohol or used a chlorhexidine-impregnated cap, found that the cap was significantly more likely to eliminate contaminants than was alcohol when used for 10 seconds [353]. Although no clinical trials have been performed on the best antiseptic to use for IVD disinfection, it is clear that sufficient contact time (at least 15–30 secs of alcohol) is necessary to insure that contamants are removed.
The addition of a nontoxic, biodegradable or easily metabolized antiseptic to IV fluid or IV admixtures [354,355,356,357,358] might eliminate the hazard of fluid contamination altogether and, further, reduce the risk of hub contamination, obviating the need for periodic replacement of the delivery system. Use of vancomycin contaiming flushes or catheter/valve dwells have been shown to reduce the risk of CR-BSI and to potentially salvage some contaminated CVCs [359,361]. In hemodialysis patients, the use of EDTA with minocycline reduced CR-BSIs [362]. These solutions have significantly reduced the risk of CR-BSI in high-risk pediatric oncology or neonatal ICU patients.
The Future
We believe that the future is very hopeful for continued progress in the prevention of device-related infection. A great deal of progress has been made over the past decade, with studies showing reduced rates of infection with the use
P.639
of more stringent barrier precautions during CVC insertion, with IV teams, with the use of more effective cutaneous antiseptics, and with the results of the first studies of innovative technologies, such as contamination-resistant hubs, attachable cuffs, and catheters with colonization-resistant surfaces. We are optimistic that future IVDs will be highly resistant to thrombosis and infection, and that it will be possible to allow percutaneously inserted catheters safely to remain in place in high-risk patients nearly indefinitely.
In the past several years, studies have demonstrated that we should have zero tolerance for catheter-related infections. Interventions using a “bundle approach”, that has included many of the interventions discussed previously, have been able to reduce catheter-related infections (including CVC-BSI) rates in ICUs to very low levels (sometimes zero for many months) (Table 37-16) [363,364,365]. Current practice in many healthcare facilities falls short of those used in successful prevention programs [366]. As more and more facilities throughout the world implement the bundle approach for catheter-related infection prevention using the most effective measures [367,368,369,370,371,372], many lives will be saved [363].
|
TABLE 37-16 |
||||||||||||
|
References
P.640
P.641
P.642
P.643
P.644
P.645
P.646
P.647