Outcomes of Catheter-associated Infections in Pediatric Patients with Short Bowel Syndrome

Abstract

Children with short bowel syndrome requiring long-term total parenteral nutrition are at high risk for catheter-associated infections. The optimal management of catheter infections in this patient population is unknown. We conducted a retrospective observational study in children with short bowel syndrome to compare outcomes of catheter-associated infections treated with catheter removal plus antibiotic therapy versus antibiotic therapy alone.

INTRODUCTION

In children, short bowel syndrome (SBS) is a chronic condition with significant morbidity and mortality.¹ Advances in medical care have greatly improved the prognosis for children with this condition due in part to the long-term administration of total parenteral nutrition (TPN) through central venous catheters. Although TPN provides the necessary nutrients for children to thrive, its use is associated with well-described complications including catheter-associated infection.² The incidence of catheter-associated infection in children receiving home TPN varies between 0.41 and 1.5 episodes of infection per patient year;^2–6 the incidence of infection is six-fold higher in infants with SBS than TPN-dependent infants with other diseases, which may be due to translocation of enteric organisms.⁷ Children with SBS require constant and prolonged venous central access and catheter removal may not be an option. Consequently, the management of catheter-associated infections in these patients may differ from that of other patient populations. The purpose of this study was to compare the outcomes of children with SBS and catheter-associated infections treated with central catheter removal plus antibiotic therapy versus antibiotic therapy alone.

MATERIALS AND METHODS

We identified all children treated at Duke University Medical Center who met the following eligibility criteria: 1) age ≤18 years; 2) SBS, defined as congenital intestinal atresia or surgical bowel resection resulting in decreased intestinal length requiring TPN for at least 4 weeks; 3) documented positive catheter blood culture at Duke inpatient or outpatient facilities between January 1, 1998 and July 1, 2007; 4) presence of a central venous catheter through which TPN was being administered; 5) no other likely source of infection; and 6) adequate antimicrobial treatment for the infection administered through the central venous catheter.

The beginning of an infection episode was defined as the time of first positive blood culture. Blood cultures positive for the same organism and antimicrobial susceptibilities were considered the same episode of infection if they occurred within 30 days of the previous positive culture and the central line was not removed. If a blood culture was positive for multiple organisms, each pathogenic organism was considered to represent a separate infection. For example, if a subject had an initial blood culture that was positive for organism A, a second culture that was positive for organisms A and B within 30 days of the initial culture, and a third culture that was positive for organism B only within 30 days of the second culture, then the first and second cultures would be considered to be one infection associated with organism A, while the second and third cultures would be considered a separate infection associated with organism B. All single positive cultures for likely contaminants, including Gram-positive rods (n=1), diphtheroids (n=1), Streptococcus species (n=1), and Leuconostoc species (n=1), were excluded. Single positive cultures for coagulase-negative Staphylococcus (CoNS) were included if the infection was treated for ≥ 5 days; all episodes with multiple positive cultures for CoNS and Streptococcus species were included.

Prompt catheter removal was defined as removal of the central line within 24 hours of species identification. Time required to clear infection was defined as the number of hours between the first and last positive blood cultures for each organism. Because time required to clear infection could be confounded by the time at which the next blood culture was drawn, this parameter was only calculated for episodes in which the last positive culture was followed by a negative culture or catheter removal within 72 hours. For organisms with reported antimicrobial susceptibilities, adequate therapy was considered if the organism was susceptible to the therapy provided. For organisms without reported susceptibilities, adequate antimicrobial therapy was based on published data. Mortality attributed to infection was defined as death occurring while the patient remained on treatment for the infection. Renal insufficiency was defined as a serum creatinine level three times the baseline level or a serum creatinine ≥4 mg/dL. Disseminated infection included endocarditis, infected thrombus, osteomyelitis, renal abscess, endophthalmitis, and meningitis occurring within 12 weeks of the initial positive blood culture. Significant hypotension was defined as an episode requiring treatment with vasopressors. Mechanical ventilation during the treatment period was also documented.

Episodes of infection were classified into 4 groups according to the causative organisms: 1) CoNS and Streptococcus species; 2) Other Gram-positive cocci (GPC); 3) Gram-negative rods (GNR); and 4) Fungi (including Candida and mold species). The primary outcome variable, time required to clear infection, was compared within these groups between patients with and without prompt catheter removal.

We conducted the analysis with STATA 9 (College Station, TX) and used the Wilcoxon rank sum test and Fisher's exact test where appropriate to compare values between groups. Reported p-values are two-tailed, and p-values < 0.05 were considered significant. The Duke University Institutional Review Board provided permission to conduct this analysis.

RESULTS

We identified 52 patients with SBS. The etiologies for SBS were necrotizing enterocolitis (56%, 29/52), congenital atresia (15%, 8/52), gastroschisis (8%, 4/52), volvulus (6%, 3/52), trauma (4%, 2/52), and other (12%, 6/52). The total number of line infections was 195, and the median number of line infections per patient was 1.5 (range: 1–27). The median age of patients was 0.33 years (range: 4 days-15 years); patients older than 1 year of age accounted for 85/195 (43%) infections; 33/52 (63%) of the patients were male; the majority of patients had SBS for more than 1 month (75%, 38/52); most of the catheters were centrally inserted (89%, 174/195); and about half of patients were in the hospital when their bacteremia developed (52%, 101/195). There was no significant difference in age (0.71 years vs. 0.85 years, p=0.24) or sex (64% male vs. 68% male, p=0.56) between patients with and without prompt catheter removal.

The most commonly identified organisms were GNR [70/195 (36%); Klebsiella species 26/195 (13%), Escherichia coli 13/195 (7%)], CoNS (46/195, 24%), and Candida species (36/195, 18%). Of a total of 379 positive blood cultures, 52 (14%) were positive for more than one pathogenic organism. The majority of patients (154/195, 79%) were not exposed to effective antimicrobial therapy in the 72 hours prior to blood culture evaluation.

In the episodes in which the catheter was removed, a median of 2.5 positive cultures [range: 1–8] were drawn before removal. Time required to clear infection could be calculated for 159/195 (82%) episodes of infection. Catheters were promptly removed in 14/159 (9%) of these episodes and eventually removed in 70/145 (48%) of the remaining cases. The median time to removal was 5 days. There was no difference in time to clear infection between episodes in which the catheter was promptly removed and not promptly removed for infections caused by CoNS and streptococci or GNR. The results were unchanged when single positive cultures for CoNS were excluded. No catheters were promptly removed in patients with Staphylococcus aureus infection. However, the time required to clear fungal infections was significantly less when the catheter was removed promptly (median [IQR]; 0 [0–42.5] hours vs. 84.3 [62.6–197.0] hours, p=0.007; Figure 1). For treatment of the 38 episodes of fungal infection, amphotericin B deoxycholate was used in 27/38 (71%), fluconazole in 22/38 (58%), and an echinocandin (micafungin and/or caspofungin) in 12/38 (32%).

Figure 1.

Time required to clear infection, stratified by organism group and timeliness of catheter removal. The asterisk (*) indicates a significant difference between groups (p<0.05). CoNS: Coagulase-negative Staphylococci; GPC: Gram-positive cocci; GNR: Gram-negative rods.

Of the 181 episodes in which the catheters were not promptly removed, renal insufficiency occurred in 12 (7%) cases, disseminated infection in 7 (n=4 renal abscess, n=2 meningitis, n=1 infected thrombus) (4%), hypotension in 13 (7%), and mechanical ventilation in 10 (6%). None of these complications occurred in patients with Staphylococcus aureus infection. Of the 14 episodes in which the catheter was promptly removed, no morbidities were observed, except for 1 patient who required mechanical ventilation due to sepsis and respiratory compromise. Frequency of secondary outcomes between the 2 groups, however, was not different. Three of the 14 (21%) infection episodes with prompt catheter removal resulted in surgical complications consisting of 1 pneumothorax and 2 episodes of bleeding requiring treatment. Twelve patients died prior to hospital discharge, 5 (56%) from complications of their infections (n=2 CoNS, n=1 Candida albicans, n=1 Enterococcus faecalis, n=1 Enterococcus faecalis and Eschericia coli). In all 5 of these patients, the catheter was not promptly removed, death occurred within 5 days of the last positive culture, and the patient was receiving appropriate antimicrobial therapy prior to withdrawal of care.

DISCUSSION

Central catheters are often essential for TPN administration, replacing catheters can be difficult in patients with limited access, and surgical complications can arise. As such, the risks of catheter retention in the setting of infection must be weighed against those of surgery and general anesthesia, as well as the consumption of the limited anatomical sites that are well-suited for catheter placement. Although there was no catheter management-dependent difference in time required to clear infection for Gram-positive and Gram-negative organisms, the time required to clear infection was significantly longer in episodes of infection caused by fungal organisms when the catheter was not removed promptly. Our data are supported by a previous study of patients with SBS in which catheter salvage with antimicrobial therapy was attempted in all patients.⁷ In this study, only 2/12 (17%) of catheters infected with yeast were salvaged without removal of the catheter, compared to a 50% salvage rate in 40 episodes of infection with other organisms. Taken together with the current observations, these data support the prompt removal of catheters infected with fungal species in children with SBS.

Although the sample size in our cohort was too small to examine differences in morbidities associated with catheter retention and mortality attributed to infection, none of the patients for whom the catheter was promptly removed died, while 5 patients with delayed removal died. Because of the retrospective nature of this study, we cannot interpret this observation as a direct effect of catheter management. It is possible that surgery was not a viable option for those patients who died because of their unstable condition or that the terminal nature of the patient's condition necessitated supportive care only. Previous investigators have found that delayed removal of central venous catheters in children with candidemia was associated with increased fatality.⁸

The strengths of this study include its focus on a well-defined group of patients for whom the appropriate management is uncertain; documentation of every blood culture at the institution's inpatient and outpatient facilities in the catheter life of each patient; information on catheter placement and time of removal; and documentation of antimicrobial therapy. The primary limitations of this study are its retrospective design and single institution experience. Because our study included only the subset of patients with catheter-associated infection, we are unable to report an accurate incidence of catheter-associated infection in all patients with SBS on long-term TPN. A larger, prospective multi-center study is the next logical step in advancing our understanding of the catheter management-related outcomes in this population.