PEDIATRICS Vol. 107 No. 6 June 2001, pp. 1272-1276

Bacteremia, Central Catheters, and Neonates: When to Pull the Line

Daniel K. Benjamin Jr, MD^*, , William Miller, MD, PhD, MPH, Harmony Garges, MD^*, Daniel K. Benjamin, PhD^§, Ross E. McKinney Jr, MD^*, Michael Cotton, MD^*, Randall G. Fisher, MD, and Kenneth A. Alexander, MD, PhD^*

From the ^* Department of Pediatrics, Duke University Medical Center, Durham, North Carolina;  Department of Epidemiology, Schools of Medicine and Public Heath, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina; ^§ Department of Economics, Clemson University, Clemson, South Carolina; and  Children's Hospital of the King's Daughters, Norfolk, Virginia.

	ABSTRACT

Top Abstract Methods Results Discussion References

Objectives.  Physicians who treat neonates who become bacteremic while dependent on central venous catheters face a serious and common dilemma. We sought 1) to evaluate the relationship between central venous catheter removal and outcome in bacteremic neonates, 2) to determine species of bacteria that are associated with an increased risk of infectious complications if the central catheter is not removed promptly, and 3) to provide evidence-based recommendations for central catheter management.

Method.  A retrospective cohort study of all neonates who had central venous access and developed bacteremia between July 1, 1995, and July 31, 1999, was conducted in the Duke University neonatal intensive care unit.

Results.  The outcome for patients in whom the central catheter was not removed within 24 hours of organism identification was significantly worse (odds ratio = 9.8) than it was for those whose catheters were removed promptly. For patients who were infected with Staphylococcus aureus or with nonenteric Gram-negative rods, delayed removal of the central catheter was associated with complicated bacteremia. Catheter sterilization was attempted in 27 neonates who were infected with enteric Gram-negative rods; only 10 of these infants retained their catheters without infection-related complications. Infants who had 4 consecutive blood cultures that were positive for coagulase-negative staphylococcus (CoNS) were at significantly increased risk for end-organ damage and death, compared with infants who had 3 or fewer positive blood culture for CoNS (odds ratio = 29.58).

Conclusions.  Bacteremic infants experienced fewer infection-related complications when the central catheter was removed promptly. One positive blood culture for S aureus or a Gram-negative rod warrants central line removal in a neonate. Clinicians who are faced with a neonate who has 1 positive culture for CoNS may attempt medical management without central catheter removal, but documentation of subsequent negative blood cultures is crucial. Once a neonate has 3 positive blood cultures for CoNS, the central catheter should be removed.central line, neonate, bacteremia, bacteria, umbilical catheter, Broviac, percutaneous.

Central venous catheters are a cornerstone of neonatal intensive care. Unfortunately, catheter-related sepsis is a frequent life-threatening complication.¹ Although catheter removal is a critical component of medical treatment in neonates who are infected with Candida species,² the benefits of prompt catheter removal in bacteremic neonates are unknown.

The treatment of bacteremic infants who have central catheters is likely to depend on the specific infecting organism. Although some reports have suggested that >70% of neonates who are infected with coagulase-negative staphylococcus (CoNS) can retain their central catheters,³ others have advised that persistent bacteremia with CoNS warrants central catheter removal.⁴ Prompt removal of catheters has been strongly suggested in adult patients who are bacteremic with Staphylococcus aureus¹ and Pseudomonas,⁵ but catheter-management guidelines have not been established for infants who are bacteremic with these organisms.

We sought to answer the practical question of what neonatologists, infectious disease consultants, and primary care nursery physicians should do when faced with a positive blood culture in a neonate who has a central catheter that is vital to the patient's treatment. The goals of this study were 1) to evaluate the consequences of central catheter retention in bacteremic infants; 2) to determine which organisms, when isolated from the blood, warrant prompt catheter removal; and 3) to develop a rational approach to central catheter management in the bacteremic neonate on the basis of the relative risks of infectious complications.

	METHODS

Top Abstract Methods Results Discussion References

The study was conducted at the neonatal intensive care unit (NICU) at Duke University. The nursery, which has 12 level 2 and 24 level 3 beds, admits an average of 700 infants each year, approximately 30% of whom weigh <1500 g on admission. Initial central catheter access usually is obtained via umbilical vessels, but most neonates who require prolonged central access receive percutaneous intravenous central catheters. When an infant is evaluated for possible sepsis, the attending physician's judgment guides the acquisition of cultures and initiation of antimicrobial agents. Virtually all neonates who are evaluated for possible sepsis undergo a blood culture and are started on 2 broad-spectrum intravenous antibiotics. Neonates who have documented bacteremia are evaluated for end-organ damage and have their central catheters removed at the discretion of the attending physician.

We reviewed microbiology laboratory records from July 1, 1995, to July 31, 1999, for positive blood cultures in NICU patients. Inclusion criteria consisted of the presence of a central venous catheter and documented bacteremia. We extracted the following information from the charts of the study population: birth weight, age at the time of infection, end-organ damage, mortality, type of catheter, ventilator status, abnormalities on chest radiograph, abnormalities on abdominal radiograph, and the timing of central catheter removal. We also examined the results of each infant's blood, urine, and cerebrospinal fluid cultures and elapsed time until negative cultures.

The primary outcome was complicated bacteremia, defined as the presence of end-organ damage, multiple positive blood cultures (drawn at separate times) within one episode of sepsis, or death. We defined end-organ damage as evidence of osteomyelitis by plain film, vital organ abscess, a positive echocardiogram showing vegetating lesions, or a positive lumbar puncture. A positive lumbar puncture was documented by a positive culture or the presence of >25 white blood cells/mm³.⁶

We categorized the physician's action as "catheter promptly removed" if the physician removed the central catheter within 24 hours of species identification. We categorized the physician's action as "attempted catheter sterilization" if the physician treated through the central catheter with appropriate antibiotics for >24 hours after species identification. We excluded from analysis those neonates who, before physician notification of bacteremia, died or had their catheter removed.

We followed the status of the central catheter until the infant had completed antimicrobial therapy or was discharged or the medical record noted that the catheter was no longer needed. We defined central catheter retained without infection-related complications as meeting all 3 of the following criteria: 1) negative blood cultures were documented while the catheter was still in the infant, 2) the infant did not experience complicated bacteremia, and 3) the central catheter survived until the end of follow-up.

Statistical analyses were performed with the use of Stata, Version 6.0 Stata Corp, College Station, TX). Comparisons between categorical variables were based on Fisher's exact test, and reported P values are 2-tailed. Odds ratios (OR) and 95% confidence intervals (95% CI) were determined with unconditional logistic regression. Estimates from the logistic regression models were adjusted for nonindependence, or clustering, because some infants had >1 episode of bacteremia. For the logistic modeling, complicated bacteremia was the outcome and central catheter sterilization (dichotomous), birth weight (dichotomized at 1500 g), and age at diagnosis (dichotomized at day of life 4 on the basis of the definition of nosocomial infection as bacteremia in infants >4 days old) were the independent variables.

We also analyzed complicated bacteremia as a function of attempted sterilization by organism. We considered the organisms in 4 groups: 1) nonenteric Gram-negative rods (Alcaligenes, Pseudomonas, and Stenotrophomonas) and S aureus, 2) enteric Gram-negative rods (Citrobacter, Enterobacter, Escherichia coli, Klebsiella, Morganella, Pantoea, and Serratia), 3) Enterococcus, and 4) CoNS.

	RESULTS

Top Abstract Methods Results Discussion References

We evaluated a total of 160 episodes of bacteremia in 122 neonates who had central access (Table 1). Seven neonates had their catheters removed before physician notification of bacteremia and were excluded from the analysis. The mean birth weight was 916 g (standard deviation = 898 g), and the mean age was 33 days (standard deviation = 36 days). Catheter sterilization was attempted in 128 of the episodes of bacteremia. Seven neonates died, none of whom had their catheters removed promptly (Table 1).

Of the neonates who underwent attempted catheter sterilization, 46% (59 of 128) developed complicated bacteremia, compared with 8% (2 of 25) of infants whose catheters were removed promptly (OR = 9.8; 95% CI = 2.2, 43.46). Of the 128 neonates who underwent attempted sterilization, 54 retained their catheters without complication, most of whom (37 of 54) were infected with CoNS (Table 1). No neonate had a positive blood culture after the central catheter was removed. Every patient except for 1 infant who was transferred to another hospital had a documented negative blood culture after central line removal. There was no correlation between bacterial species and attempted central catheter sterilization (P > .99). On species identification, all neonates were placed on at least 1 antibiotic to which the organism was susceptible.

With the use of bivariable analysis, low birth weight was not associated with complicated bacteremia (OR = 2.6; 95% CI = 1.0, 7.0), nor was age at infection (OR = 2.5; 95% CI = 0.8, 8.1). Patients who had >1 positive blood culture were at significantly higher risk of end-organ damage than those who did not (OR = 1.5; 95% CI = 1.2, 1.8).

Using logistic regression modeling, we evaluated the risk of complicated bacteremia as a function of attempted sterilization, birth weight, and age at infection. The results from our logistic regression analysis were similar to the bivariable results provided above. If the physician attempted to sterilize the central catheter, there was a greater likelihood of complicated bacteremia (OR = 17.3; 95% CI = 2.2, 139.4). Again, low birth weight and age at infection were not associated with complicated bacteremia (OR = 2.0; 95% CI = 0.6, 6.9; and OR = 4.2; 95% CI = 1.0, 17.0).

The bedside physician attempted to sterilize the central catheter in 10 neonates who were infected with S aureus. Nine of these had complicated bacteremia, including 4 infants with multiple sites of end-organ damage. One infant who was infected with S aureus experienced reinfection with the same catheter in place 3 weeks later. Catheter sterilization was attempted in 5 neonates who were infected with nonenteric Gram-negative rods. No neonate who was infected with any nonenteric Gram-negative rod (Alcaligenes, Pseudomonas, and Stenotrophomonas) retained a central catheter without complication. Sterilization was attempted in 3 neonates who were infected with Pseudomonas2 died. No neonate in this group whose catheter was removed promptly experienced infection-related complications (Table 1). For patients who were infected with either S aureus or nonenteric Gram-negative rods, complicated bacteremia was associated with attempts to sterilize the central catheter (P = .04, Fisher's exact test).

Sterilization was attempted in 29 infants who were infected with enteric Gram-negative rods, but only 10 (34%) retained their catheters without complication, although this relationship was not statistically significant at traditional levels (P = .07; Table 1). Similarly, catheter sterilization was attempted in 12 neonates who were infected with Enterococcus and 5 (42%) retained their central catheter without complication (P > .99; Table 1).

Central catheter sterilization was attempted in 72 neonates who had CoNS bacteremia. Overall, attempted sterilization was not associated with complicated bacteremia in patients who had positive blood cultures for CoNS (P = .05); nor was it associated with end-organ damage (P = .58). Nevertheless, the number of positive blood cultures was an important determinant of outcome in neonates who had CoNS bacteremia. Neonates who had multiple positive blood cultures, predominantly those with at least 4 consecutive positive blood cultures for CoNS (Table 2), were at greater risk of end-organ damage (P < .001). Once neonates had 4 consecutive positive blood cultures for CoNS, their risk of end-organ damage and death increased dramatically (OR = 29.6; 95% CI = 4.7, 186.1).

We did not have a sufficient number of neonates who had central catheters and who were infected with Listeria (n = 1) or group B Streptococcus (n = 1) to analyze outcomes in those organisms.

Complicated bacteremia occurred in 36 of 82 neonates who had peripherally inserted central cathethers, 13 of 30 neonates who had Broviac catheters, 3 of 8 neonates who had femoral catheters, and 7 of 33 neonates who had umbilical catheters. There was no correlation between type of central catheter and complicated bacteremia (P = .89). There was no correlation between abnormalities on chest radiograph (P = .38), abnormalities on abdominal radiograph (P = .41), or mechanical ventilation status (P = .24) and complicated bacteremia.

	DISCUSSION

Top Abstract Methods Results Discussion References

We found that attempted catheter sterilization in bacteremic neonates was associated with a higher incidence of complicated bacteremia even when we controlled for birth weight and age at infection in a multivariable analysis. We also were able to quantify risks associated with various species of bacterium when catheter sterilization is attempted. We did not attempt to prove which neonates had microbiologic evidence of an infected catheter¹ and which neonates had concomitant bacteremia and central venous access. We simply evaluated and answered the practical question that nursery physicians address when faced with a bacteremic infant who has a central venous catheter: what are the risks of infection-related complications given a particular organism in the bloodstream if catheter sterilization is attempted?

Although the inherent pathogenicity of different bacterial species has considerable variation (S aureus is more virulent than CoNS, and Pseudomonas is more virulent than Enterococcus), we did not find a correlation between species of bacterium and attempted catheter sterilization. We believe that bacterium species should be an essential component of the decision to remove the catheter and have provided evidence to that effect.

We were able to discern variable risks associated with different species of bacteria. In the organisms that traditionally are thought to be more virulent in the neonate (S aureus and nonenteric Gram-negative rods), we found that neonates experienced significantly higher rates of complicated bacteremia when catheter sterilization was attempted.

We evaluated S aureus and nonenteric Gram-negative rod organisms together because these organisms have several similarities. They are seen most commonly in nosocomial infections of immune-compromised hosts with significant antibiotic exposure. They are virulent organisms that usually lead to rapid deterioration in the bacteremic patient. Furthermore, these organisms (because of an adherence advantage) have been shown to be difficult to treat in catheter-associated infections in other patient populations.^1,5 This exploitation, as a result of the organisms' binding to fibrin and fibronectin,^7,8 provides a milieu in which the infecting agent may survive despite antimicrobial therapy.

Our data suggest that central catheters that carry an infection of S aureus or of nonenteric Gram-negative rods should be removed immediately to avoid complicated bacteremia. Not only did most such infants experience end-organ damage when the bedside physician tried to sterilize the catheter, but also there was only 1 instance of catheter salvage in this group.

The association between central catheter sterilization and complicated bacteremia for patients who are infected with enteric Gram-negative rods and Enterococcus did not reach statistical significance, yet with each of these organism groups, the success rate of catheter retention without complications was <50%. We believe that a success rate of <50% is unacceptably low. Bacteremia with an enteric Gram-negative rod or Enterococcus probably warrants central catheter removal in a neonate.

Although we believe that prompt removal of the central catheter in the neonate who is bacteremic with an enteric Gram-negative rod or Enterococcus species is ideal, we realize that the placement of central catheters in neonates (some of whom weigh <500 g) is challenging. Catheter sterilization might be considered in a term infant who has bacteremia that is caused by these organisms, but given the increased risk of end-organ damage associated persistent bacteremia, having multiple positive blood cultures warrants central catheter removal.

The clinician is forewarned that simply because the P value was not <.05, these data do not support the retention of central catheters in neonates who are bacteremic with enteric Gram-negative rods or Enterococcus. This word of caution is particularly applicable to neonates who have persistent bacteremia, a group that had a much higher rate of end-organ damage.

There is considerable disagreement in the clinical literature regarding the ability to manage central catheters in situ in patients who have CoNS. Although some investigators have reported great success in the medical management of CoNS bacteremia, others have advised central line removal after multiple positive consecutive blood cultures.⁴ The microbiologic literature⁹ supports considering 2 or more positive blood cultures in 24 hours as a likely infection and lack of such documentation as likely representing contamination of the blood culture bottle. Given the microbiologic literature and the ability of CoNS to adhere to catheters,^7,8 we anticipated that physicians would be able to use medical management in neonates who have 1 positive culture but that having multiple positive cultures would warrant catheter removal. In fact, most neonates who have 1 positive blood culture for CoNS were able to retain their central catheters without end-organ damage, whereas documentation of 4 consecutive positive blood cultures for CoNS was strongly associated with end-organ damage.

These data support a management model with 3 elements. First, if a patient has 1 positive blood culture for CoNS, the physician may attempt to treat the neonate without catheter removal. Second, documentation of negative cultures for CoNS is crucial, because having 2 positive blood cultures in 24 hours places the neonate in a higher risk stratum.⁹ Third, if the bedside physician obtains 3 consecutive positive cultures, then the central catheter should be removed, because the risk of end-organ damage and death rise dramatically with a fourth positive blood culture for CoNS. For example, if a physician draws on day 1 a blood culture that turns positive for CoNS on day 3, then follow-up blood cultures should be obtained on both day 3 and day 4. If both follow-up cultures are positive for CoNS, then the catheter should be removed.

Our study had several limitations. Because it was retrospective, cause and effect could not be established conclusively. We cannot be certain that removal of the catheter would have altered the outcome in infants who developed complicated bacteremia. For example, it is possible that seeding of distant sites had occurred by the time of the first positive blood culture. Furthermore, the infants who are the most ill from bacteremia frequently require the support provided by central access, and they also often are the infants for whom acquisition of central access is the most difficult. Simultaneously, the neonates who are the most ill from bacteremia are more likely to develop complicated bacteremia. We did not see an association between several common markers for illness in the neonate (abnormal chest radiograph, abnormal abdominal radiograph, and mechanical ventilation requirement) and complicated bacteremia, but the concern for severity of illness as a potential confounder in the catheter sterilization and complicated bacteremia relationship is addressed best by a prospective trial. Because evaluation of end-organ damage was at the discretion of the attending physician, there also may have been children who had undocumented end-organ damage.

We attempted to examine differences across organism species by analyzing complicated bacteremia as a function of catheter sterilization for each organism separately. Unfortunately, our sample size was too small to allow logistic modeling within organism groups.

Physicians who are treating a neonate who has documented bacteremia and central venous access face a difficult decision regarding the removal of the catheter. The catheter often is vital to patient support, yet it may be a persistent source of organisms that cannot be treated adequately in situ. The physician must attempt to balance an acceptable risk of complicated bacteremia with the risk of placing a new central catheter. Clearly, the best answer will come from prospective clinical trials.

Until these are conducted, we believe that these preliminary data indicate that the following model may be appropriate: 1) there are 3 instances in which the bacteremic neonate should have his or her catheter removed immediately: a) S aureus, b) nonenteric Gram-negative rods, and c) enteric Gram-negative rods; 2) a positive blood culture for Enterococcus permits more flexibility in choosing medical management provided that there is documentation of negative blood cultures; after 2 positive blood cultures, the need to remove the catheter becomes more urgent; and 3) a single positive culture for CoNS can be followed by an attempt to sterilize the central catheter, but once a neonate has had 3 consecutive positive blood cultures for CoNS, the catheter should be removed.

	ACKNOWLEDGMENTS

We thank Dr Robert H. Smith, Penny Hodgson, Laura Smith, and the Duke Clinical Microbiology staff for their efforts.

	FOOTNOTES

Received for publication Jun 13, 2000; accepted Sep 28, 2000.

Reprint requests to (D.K.B.) Box 3499 Department of Pediatrics, Duke University Medical Center, Durham, NC 27710.

	ABBREVIATIONS

CoNS, coagulase-negative staphylococcus; NICU, neonatal intensive care unit; OR, odds ratio; CI, confidence interval.

	REFERENCES

Top Abstract Methods Results Discussion References

1. Raad II, Bodey GP Infectious complications of indwelling vascular catheters. Clin Infect Dis 1992; 15:197-210 [Medline]
2. Eppes SC, Troutman JL, Gutman LT Outcome of treatment of candidemia in children whose central catheters were removed or retained. Pediatr Infect Dis J 1989; 8:99-104 [CrossRef][Medline]
3. Cairns PA, Wilson DC, McClure BG, Halliday HL, McReid M Percutaneous central venous catheter use in the very low birth weight neonate. Eur J Pediatr 1995; 154:145-147 [CrossRef][Medline]
4. Benjamin DK Jr, Ross K, McKinney RE Jr, Benjamin DK, Auten R, Fisher RG When to suspect fungal infection in neonates: a clinical comparison of Candida albicans and Candida parapsilosis fungemia with coagulase-negative staphylococcal bacteremia. Pediatrics. 2000; 106:712-718 [Abstract/Free Full Text]
5. Elting LS, Bodey GP Septicemia due to Xanthomonas species and non-aeruginosa pseudomonas species: increasing incidence of catheter-related infections. Medicine 1990; 60:196-206
6. Sarff LD, Platt LH, McCracken GH Jr Cerebrospinal fluid evaluation in neonates: comparison of high-risk infants with and without meningitis. J Pediatr 1976; 88:473-477 [CrossRef][Medline]
7. Vaudaux P, Pittet D, Haeberli A, Host factors selectively increase staphylococcal adherence on inserted catheters: a role for fibronectin and fibrinogen or fibrin. J Infect Dis 1989; 169:865-875
8. Herrmann M, Vaudaux PE, Pittet D, Fibronectin, fibrinogen, and laminin act as mediators of adherence of clinical staphylococcal isolates to foreign material. J Infect Dis 1988; 158:693-701 [Medline]
9. Zaidi AKM, Harrell LJ, Rost JR, Reller LB Assessment of similarity among coagulase-negative staphylococci from sequential blood cultures of neonates and children by pulse-field gel electrophoresis. J Infect Dis 1996; 174:1010-1014 [Medline]