Invasive medical devices are commonly used in hospitalized children. They result in more than 250,000 infections each year in the United States.¹ A catheter-related bloodstream infection (CRBSI) is defined as bacteremia or fungemia in a patient with an intravascular catheter that is the presumed source of infection. This definition belies the challenges a clinician faces in accurately diagnosing CRBSI in children, including difficulties in diagnostic testing and variability in epidemiology based on age, intravascular device used, infusate being given, and underlying clinical condition.² Additionally, central venous catheters (CVCs) pose other infectious complications aside from CRBSI, including local infection at exit sites, and they can become secondarily infected through seeding of bacteria due to bacteremia originating from infection at a distant site.

Nationwide approximately 4700 children are hospitalized for initial placement of cerebrospinal fluid (CSF) shunts annually, with more than 2400 children hospitalized for shunt-related infections at a cost of over $250 million.³ The most common CNS shunt placed in children is the ventriculoperitoneal (VP) shunt.⁴ These devices breach natural barriers to infection and provide access for bacteria to enter and proliferate in a site relatively sequestered from the immune system and rich in nutrients. Diagnosis and management of these infections usually involves invasive procedures and prolonged hospitalization, with marked burdens on children and their families, including prolonged hospital stays, potential morbidity, and significant healthcare costs.

Pediatric hospitalists care for a population of children that is increasingly medically complex,⁵ often including children in the neonatal intensive care unit (NICU) and/or pediatric intensive care unit (PICU), both sites of care where intravascular and CNS catheters are commonly used. Knowledge of optimal diagnostic and treatment strategies for device-related infections is a vital skill for hospitalists. This chapter will help provide information clinicians can use to help develop expertise in the management of infectious complications from these devices.

BACKGROUND

Pediatric bloodstream infections account for approximately 13.5% of all nosocomial bloodstream infections,⁶ at per-episode cost of approximately $39,000 per episode in PICU patients.⁷ The majority of nosocomial bacteremias present in the setting of a central venous catheter.⁸ Risk of infection is present both for CVCs inserted directly into a central vein as well as peripherally inserted central venous catheters (PICCs), which are more commonly used outside of the intensive care unit.^9,10

Multiple factors influence the risk of infection, including patient age, medical comorbidities, site of device placement, types of infusate being administered, frequency of times the device is accessed, duration the CVC is in place, and the device type. Table 107-1 lists the most commonly used CVC types in pediatrics. In general, tunneled catheters are a lower long-term risk for infection than non-tunneled catheters, with implanted venous access devices such as portacaths having the lowest rates of infection.¹¹

PATHOPHYSIOLOGY

Central venous catheters provide multiple avenues for microbial colonization and infection (Figure 107-1). First, CVCs have at least one access port/hub which can become colonized with organisms from environmental exposure or from accessing/de-accessing the device. Cutaneous flora may also translocate through the barrier defect provided by the CVC or from contaminated healthcare worker hands during dressing changes or device manipulation, which can result in exit site infections or infection of the subcutaneous length of the catheter in the case of tunneled devices. Fibrin sheaths and thrombi can develop along the intravascular portions of the catheter, an ideal site for colonization by organisms during episodes of transient bacteremia or from bacteremia due to infection at a distant site. Finally, materials being infused through the catheter may be contaminated with organisms, resulting in direct inoculation of both the device and the bloodstream. Thus, infection can be local, systemic, or secondary to another infectious process.

Table 107-2 summarizes the pathogens most commonly associated with catheter-related bloodstream infections. Coagulase-negative staphylococci (CONS) account for approximately 40% of children diagnosed with nosocomial bloodstream infections; this increases to 50% in neonates. Gram-negative aerobic bacilli account for approximately 25% of CRBSI, followed by enterococci (11% to 15%), and Candida species.^12,13 Among gram-negative bacilli isolated, the most common pathogens include Enterobacter species, Pseudomonas aeruginosa, Klebsiella pneumonia, Escherichia coli, Serratia marcescens, Acinetobacter species, and Citrobacter species.¹³ Children with CVCs managed mostly in home healthcare settings may be at higher risk for non-endogenous gram-negative pathogens from water or environmental sources such as Pseudomonas, Acinetobacter, and Agrobacterium species, especially during the summer.^14,15

Available data are biased toward PICUs, NICUs, and oncology units, as surveillance for bloodstream infection is usually performed in these high-risk patient populations. It is likely that a similar spectrum of organisms cause CRBSI in children in non-ICU inpatient settings, but there is limited information to support this. One study of outpatient parenteral therapy in children with osteoarticular infections found that the rate of infectious complications associated with the catheter was 6.3 per 1000 catheter days and occurred in approximately 10% of patients. The average time to development of infectious complications was 24.5 days.¹⁶

CLINICAL PRESENTATION

Children with local site infections and tunnel infections may present with focal erythema, induration, tenderness, warmth, a fluctuant mass, and/or purulent or foul-smelling discharge. For tunnel infections these symptoms can extend along the length of the subcutaneous portion of the catheter. Site infections may extend deep into subcutaneous tissues, including fat, muscle, fascia, or bone. Additionally, these localized infections can still be associated with bloodstream infection.

Catheter-related bloodstream infections are difficult to diagnose clinically. The most common symptom in children with CRBSI is fever, but this has poor specificity as most febrile children with a CVC in place do not have bloodstream infection.¹⁷ For example, among febrile cancer patients with intravascular catheters, only 10% and 24% of all febrile episodes were associated with bacteremia in non-neutropenic and neutropenic patients, respectively.¹⁸ Episodic symptoms of fever and/or hypotension associated with infusion times may be an indicator of introduction of bacteria into the bloodstream through either an infected catheter hub or lumen or from the infusate. Complications of bacteremia may also be the presenting symptoms, which could include sepsis, or disseminated infection to multiple end organs (including the eyes, heart, bones, skin, or visceral organs) which can manifest as embolic phenomena, multifocal abscesses, or immune complex disease such as nephritis. Finally, catheter malfunction, such as inability to infuse fluids or aspirate blood from the catheter, may indicate infection. Such malfunction is often due to fibrin or thrombus deposition which can readily become colonized or infected,¹⁹ but can also arise due to deep venous thrombosis.²⁰

Despite the variety of potential clinical features of CRBSI, in many cases, a positive blood culture alone may be the presenting feature, especially among patients transferred from outside facilities or seen for inpatient consultation.

DIFFERENTIAL DIAGNOSIS

Given that children presenting with fever in the setting of indwelling CVC usually do not have CRBSI, the differential diagnosis can be quite broad. Often children with CVCs have significant chronic medical conditions, including immune-compromised states, previously diagnosed infection at a different site, or systemic illnesses requiring parenteral medications or nutrition. Among children with long-term CVCs in place, catheter malfunction is a common occurrence; secondary infection of a malfunctioning catheter may occur. In all these cases, a thoughtful evaluation that includes blood cultures from the CVC and, when possible, peripheral sites are important for accurate diagnosis.

DIAGNOSIS

Local Infection

Local infections associated with a central venous catheter (including exit site, tunnel, and pocket infections) are readily recognizable by physical examination (Table 107-3). Ultrasonography of the area to assess deep fluid collection/abscess can aid in establishing the extent of infection and identify areas that may require drainage after catheter removal.

Catheter-Related Bloodstream Infections

Accurate diagnosis of CRBSI in children can be challenging. Although pediatric-specific data are lacking, in adults, rates of confirmed catheter infection from removed CVCs in bacteremic patients are only 15%.²¹ Current methods available for increasing accuracy of diagnosing CRBSI in children include quantitative blood cultures obtained through the catheter and a peripheral vein, semiquantitative cultures of a catheter segment, and differential time to positivity of cultures (Table 107-3).^2,17

Diagnosis of CRBSI can be made by comparing quantitative differences in colony counts of a pathogen isolated from blood cultures from a CVC versus a peripheral culture. A fivefold or greater difference in number of colonies from the CVC culture versus the peripheral culture establishes the diagnosis. In practice, this technique is rarely used due to cost and logistic limitations of most microbiology laboratories.

Differential time to positivity of blood cultures drawn from two different sites may also be used to establish the diagnosis. Time to positivity is defined as the duration of time necessary for the blood culture to grow the causative organism. This can only be used reliably when identical blood volumes are drawn from each site, due to the influence of blood volume on time to positivity (see below). This testing method is most commonly performed by obtaining a blood culture from the catheter and a separate culture via peripheral venipuncture. Positive blood culture from the catheter in at least 2 hours less than peripheral culture positivity has a sensitivity of 93% and a specificity of 75% for catheters in place for greater than 30 days, and a sensitivity of 81% and a specificity of 92% for catheters in place for less than 30 days,²² with multiple other studies demonstrating similar results.^23-25 Alternatively, in patients with a dual-lumen catheter, observing a difference in time to positivity between cultures drawn from each hub of ≥180 minutes was associated with a sensitivity of 61% and a specificity of 94%.²⁶ This latter approach may be helpful in cases where obtaining a peripheral culture is not feasible.

In cases where the catheter is removed, the tip may be sent for culture via quantitative or semiquantitative methods. In the first case, a catheter segment including the tip is flushed with broth or vortexed in broth, followed by culture of the broth on agar plates.²⁷ Semiquantitative, or roll-plate methods, involve rolling a sliced segment of the removed catheter across the surface of an agar plate; colony-forming units are counted after overnight incubation.²⁸

Timing of Blood Cultures

Optimal timing of blood cultures has not been studied extensively. In CRBSI, unlike other infections causing bacteremia, patients are only intermittently subjected to bacteremia due to the fact that bacteria residing in the catheter lumen are not continuously exposed to the intravascular space. Prior evidence suggests that bacteremia may precede fever onset,²⁹ which may mean that, in children with CRBSI, by the time fever is noted and blood cultures are drawn, the transient bacteremia has cleared. There is some evidence to suggest that this could be overcome by taking multiple blood cultures in a 24-hour period.^25,30

For critically ill patients with hemodynamic instability, two sets of blood cultures from two separate sites should be drawn promptly before the patient receives initial empiric antibiotic therapy, but this should not delay initiation of antibiotics in a septic patient.³¹ For more stable patients, multiple sets of blood cultures can be obtained in a 24-hour period to improve yield. For patients already on antibiotics, obtaining blood cultures right before the next scheduled dose (when drug levels in the body are lowest) theoretically improve blood culture yield;³² alternatively “isolator” culture bottles—culture bottles with resins or chemical beads designed to remove antibiotics from sampled blood—could be used in these patients, although data are lacking on the effectiveness of these strategies.

Volume of Blood for Culture

The sensitivity of blood cultures for detecting a pathogen can vary by the magnitude of bacteremia, the sensitivity of the blood culture detection system, and the volume of blood aspirated into the bottle. Smaller blood volumes (<0.5 mL) used for culture can prolong time to positivity and reduce the overall likelihood of organism growth, particularly when bacterial concentrations are low in the obtained blood.³³ Volume of blood obtained per culture bottle is more important than number of blood cultures obtained. For example, one study found that positive yield for one large-volume (6 mL) blood culture was 72% versus only 47% for two separate smaller-volume (2 mL) blood cultures.³⁴ Similar results are found in adult data as well, with standard adult-volume cultures (mean 8.7 mL) having a 92% pathogen recovery rate versus only 69% for low-volume (mean 2.7 mL) blood cultures.³⁵ In adults, blood culture yield is increased approximately 3% per mL of blood cultured. In part this may be due to inadequate nutrients present in blood necessary for bacterial growth (as in the case of Haemophilus influenzae) after dilution in culture broth.

Conversely, overfilling blood culture bottles may also reduce yield. Broth dilution of blood, and the phagocytic cells, complement proteins, and antibodies present therein, may enhance blood culture yield.³⁶ Given these constraints, each blood culture system has an optimal blood volume:broth ratio, which is usually in the range of 1:5 to 1:10. Clinicians should become familiar with the blood culture system used at their own institutions and the optimal blood volumes that should be obtained. Finally, despite the lower yield of low-volume blood cultures, it is important to note that modern culture systems supplement the broth with other materials that may increase their yield. Thus, in pediatric blood cultures, even low-volume cultures (e.g. 0.5 mL) may detect significant bacteremias, although 1 to 5 mL of blood is preferred.

Primary Bloodstream Infection versus Contaminated Culture

Given that skin flora—especially coagulase-negative staphylococci (CONS)—are commonly identified in catheter-related bloodstream infections, differentiating contaminated blood cultures from true infections can pose a challenge. Multiple cultures prior to antibiotic administration can help clarify the issue; a single culture contaminated with CONS, micrococcus, or Bacillus species may represent contamination, whereas isolation of the organism from multiple cultures is more concerning for infection. For example, a study using a mathematical model of blood cultures positive for CONS in patients with central venous lines found that the positive predictive value of one positive blood culture was 55% if it was the only culture obtained, 20% if two cultures were performed, and only 5% if three were performed.³⁷ Conversely, if two of two cultures were positive, the positive predictive value was 98% if both samples were obtained from a peripheral vein, 96% if one was obtained through a CVC and the other through a peripheral vein, and 50% if both samples were from a CVC. Differences in antibiotic susceptibility patterns between isolates of CONS obtained from multiple blood cultures may also help differentiate true infection from contamination, as differing susceptibility patterns between isolated organisms could suggest the presence of multiple contaminated blood cultures.³⁸

Age of the child and underlying medical condition also influences the likelihood of a blood culture positive for skin flora being positive for a true pathogen. Neonates more frequently have CRBSI caused by coagulase-negative staphylococci, but it remains the predominant organism identified^39,40 in children in PICUs as well. The identification of CONS in blood culture often results in administration of antibiotics, particularly vancomycin, with such overuse likely being a contributing factor to increasing vancomycin resistance among gram-positive organisms.^41-43 Current recommendations from the American Academy of Pediatrics (AAP) and the Centers for Disease Control and Prevention (CDC) provide a highly sensitive method for detecting primary bloodstream infections, although from a clinician’s perspective these definitions may overestimate the true incidence of bloodstream infection.¹

Future Detection Methods

It is important to note that, although marked strides have been made in blood culture technology, the basic techniques for growing bacteria from blood samples remain relatively unchanged. Future molecular-based methods for detection of bacterial genetic material are being developed and evaluated (citations), though nothing to date has replaced blood cultures as the gold standard for detecting bacteremia.

TREATMENT

Treatment guidelines for catheter-related bloodstream infections in children are based on limited data; most recommendations are derived from data from adult populations. Even adult data are limited, as there are no randomized control trials that address the optimal management of CRBSI.²

Empiric therapy in children with suspected CRBSI should include an antibiotic with gram-positive activity such as nafcillin, oxacillin, or vancomycin, and an anti-pseudomonal gram-negative agent such as cefepime or piperacillin-tazobactam. In cases where infection with multidrug-resistant gram-negative bacteria is suspected, use of an aminoglycoside in addition to an anti-pseudomonal β-lactam may be appropriate. Vancomycin should be used in children with current or recent prolonged hospitalizations or in institutions where methicillin-resistant Staphylococcus aureus is common. Fluoroquinolones are rarely utilized in children as empiric treatment of a suspected CRBSI.

Clinicians should refine antibiotic therapy to target the isolated pathogen(s) based on organism morphology (e.g. coccus or bacillus, gram-positive or -negative, fungal elements) and antimicrobial susceptibility data as they become available.

One of the most important issues to address in treating CRBSI is that of catheter removal (Table 107-4). For adults, most non-tunneled CVCs should be removed in cases of CRBSI.² In children, however, removal may not always be feasible, given lack of alternative vascular access sites or potential complications associated with reinsertion or replacement. Thus, treatment of the infection with retention of the infected catheter is often attempted. Tunneled catheters or catheters with subcutaneous access ports where there is associated tunnel or port site infection present an additional complicating issue; cure of the tunnel or site infection is rarely achieved without removal of the catheter.² Successful treatment of CRBSI with retention of the catheter has been reported, however.^44-46 Regardless, successful treatment of catheter-associated infection should be tailored to the device-type organism identified.

Optimal duration of therapy for catheter-related infection is another issue of debate, and depends in part on pathogen isolated, catheter device involved, whether the device is removed, type of infection (local or systemic), and whether there are complications related to the infection, such as septic thrombosis, endocarditis, or metastatic foci of infection. In the case of complications, duration is tailored to address the most serious complications (e.g. a patient with CRBSI and endocarditis would receive antibiotics for the duration necessary to treat endocarditis). Route of antibiotic administration is also not well-studied, although in cases of resistant organisms parenteral therapy may be the only available option. However, an alternative antibiotic with excellent oral bioavailability may be considered in cases where all of the following conditions are met: (1) the catheter has been removed; (2) the bacteremia has cleared; (3) there are no contraindicating infectious complications; (4) adherence to therapy is assured; (5) there are no significant barriers to medication absorption; and (6) the patient is clinically improved. Following are some pathogen-specific recommendations.

Coagulase-Negative Staphylococci

Coagulase-negative staphylococci (CONS) are considered less virulent than other pathogens causing CRBSI. Infections caused by CONS are often associated with isolated fever or local infection/inflammation at the exit site. Severe systemic illness is rare in patients infected with these organisms, although it is more common in neonates. Removal of the catheter without antimicrobial therapy may be sufficient treatment for CONS bacteremia, although repeat blood cultures should be obtained after catheter removal to document clearance, and most experts recommend at least a short course (e.g. 3 to 5 days) of antibiotic therapy after catheter removal. Treatment of infection when the catheter is retained should last 10 to 14 days after negative blood cultures. Neonates with CONS infection can be treated with retention of the catheter, but if three separate repeat blood cultures remain positive after initiation of adequate antibiotic therapy, the catheter should be removed to minimize the risk of end-organ damage.^46,47 Relapse rates of CONS bacteremia from adult data in cases of retained catheters are 20% versus 3% when the catheter is removed.^48,49