THE GASTROINTESTINAL TRACT SERVES AS THE RESERVOIR FOR... : The Pediatric Infectious Disease Journal (original) (raw)

Infants hospitalized in the Neonatal Intensive Care Unit (NICU) are at great risk for hospital-acquired infections. Gram-positive skin flora is the most common cause of late-onset sepsis in the NICU and strategies to reduce such infections are focused primarily on insertion and maintenance of central venous catheters.1 Although Gram-negative bacilli cause approximately 20–30% of episodes of late-onset sepsis and 30% of healthcare-associated pneumonias in this patient population, strategies to prevent these infections are less well defined than strategies to prevent infections caused by Gram-positive pathogens.2

We report a pilot study to test the hypothesis that Gram-negative bacilli colonizing the gastrointestinal (GI) tracts of very low birth weight (VLBW) infants (birth weight <1500 g) cause subsequent bloodstream infections (BSIs). We sought to determine the frequency of genotypically concordant colonizing flora and invasive strains and evaluate their susceptibility to gentamicin.

MATERIALS AND METHODS

Study Setting.

The study sites were the NICUs at the Morgan Stanley Children's Hospital (58 beds) and the Komansky Center for Children's Health at the Weill Cornell Medical Center (50 beds); both units are part of NewYork-Presbyterian Hospital. These 2 level III NICUs have combined yearly discharges of approximately 1500 infants. The units are geographically separated by 9 miles; there is little shared staffing (other than 1 month per year where a neonatal fellow rotates the site they are not primarily affiliated with), and only rare transfers of patients between the NICUs. The same Department of Hospital Epidemiology serves both NICUs. In this study, VLBW infants were enrolled within the first week of life, including infants transferred from other hospitals. The Institutional Review Boards of Columbia University and Weill Cornell Medical College approved the protocol and waived the requirement for documentation of informed consent, but required an information sheet for parents and allowed them to choose not to have their children in the study.

Surveillance Cultures and Blood Cultures.

From January 2004 to September 2005 surveillance cultures were performed by swabbing the rectal verge of eligible infants with a rayon-tipped swab, obtaining a minute amount of fecal matter. (CultureSwab Collection and Transport System, Becton Dickinson Microbiology Systems, Sparks, MD.) Cultures were done weekly from the first week of life until discharge from the study NICUs. Specimens from both study sites were processed by the Clinical Microbiology Laboratory at Columbia University Medical Center. Selective media for isolating Gram-negative bacteria (MacConkey agar, Becton Dickinson Microbiology Systems, Sparks, MD) were inoculated, incubated at 35°C, and examined for growth at 24 and 48 hours. If a culture grew more than 1 phenotypically distinct colony, the 2 most predominant phenotypes were identified and antimicrobial susceptibility testing was performed using the MicroscanWalkAway SI System (Dade Behring, Deerfield, IL).

Blood cultures (Bactec Peds Plus, Becton Dickinson Microbiology Systems, Sparks, MD) were obtained by the NICU staff as clinically indicated and processed by the Clinical Microbiology Laboratory at each hospital. Prospective surveillance for Gram-negative bacilli causing BSIs as defined by the National Nosocomial Infections Surveillance System, modified for neonates, was performed by the study team.3 Gram-negative bacilli detected on surveillance cultures or causing BSIs were stored at −70°C for later molecular analyses.

Molecular Typing.

Pulsed-field gel electrophoresis (PFGE) was performed to determine the genetic relatedness of antecedent colonizing flora (defined as the most recent surveillance culture with the same species) and the strain causing a BSI. PFGE using the MAPPER system (Bio-Rad, Hercules, CA) was performed using restriction digestion of chromosomal DNA with _Not_I (Escherichia), _Spe_I (Enterobacter, Serratia, Citrobacter), and _Xba_I (Klebsiella). Restriction fragment band patterns were compared and analyzed using Molecular Analyst Fingerprinting Plus software (Bio-Rad) to determine clonal identity. Strain relatedness was interpreted according to established criteria.

RESULTS

Two hundred twenty-one VLBW infants were enrolled during 22 study-months (13 and 9 months of enrollment in NICU 1 and NICU 2, respectively). Fifteen infants developed 20 episodes of BSI caused by Gram-negative bacilli during the study period. All BSI isolates were the same species as the most recent species detected in GI tract surveillance cultures; 9 infants had an additional species detected in surveillance cultures and 3 infants had 1 or more surveillance cultures without detection of colonizing flora. Concordance between colonizing flora and invasive pathogens also occurred in the 2 infants with 2 or more BSIs; the first infant developed BSIs with Escherichia coli followed by S. marcescens. The second infant had 5 BSIs including 3 K. pneumoniae, 1 E. cloacae, and 1 E. coli BSI. This infant had a prolonged hospitalization with chronic GI tract pathology. Between episodes of infection, this infant had no symptoms of infection, negative blood cultures, a normal echocardiogram, and no focal infectious pathology detected by imaging of the GI tract.

Strains associated with 19 of 20 episodes of BSI and the relevant isolates from surveillance cultures (isolated 1–17 days before BSI) were available for PFGE. Seventeen of the 19 strain pairs had indistinguishable PFGE patterns, 1 pair was closely related, and 1 pair was unrelated (Table 1 and Fig. 1). Five (25%) of colonizing-infecting pairs demonstrated resistance to gentamicin. Susceptibility to gentamicin was concordant in the colonizing flora and infecting strains of all episodes.

TABLE 1:

Molecular Epidemiology of Concordant Species Associated With Gastrointestinal Tract Colonization and Subsequent Bloodstream Infection in Very Low Birth Weight Infants

FIGURE 1.:

Selected lanes of pulsed-field gel electrophoresis analysis showing pairs of rectal and blood cultures. Odd lane markers are rectal, even are blood: λ, ladder DNA size marker; lanes 1–2, Infant 2, Escherichia coli Clone C; lanes 3–4, Infant 7, Escherichia coli Clone E; lanes 5–6, Infant 14, Escherichia coli Clones F and G; lanes 7–8, Infant 11, Klebsiella oxytoca Clone I; lanes 9–10, Infant 3, Klebsiella oxytoca, Clone J; lanes 11–12, Infant 8, Klebsiella pneumoniae Clone M; lanes 13–14, Infant 1, Klebsiella pneumoniae Clones O and O2.

Infants were infected with the same strain infrequently. Twin infants, in adjacent isolettes, developed BSIs with S. marcescens on day of life 19 and 26, respectively. Two infants hospitalized in the same room developed BSIs with E. cloacae, on day of life 38 and 178, respectively.

DISCUSSION

In this pilot study, we demonstrated that the majority of BSIs caused by Gram-negative bacilli in the 2 study NICUs were endemic infections caused by endogenous GI tract flora. Similar observations have previously been made in the NICU setting. Almuneef et al4 found that 12 (86%) of 14 infections (including 5 BSIs) were preceded by GI tract colonization with an identical strain, although the interval between colonization and infection was not specified. The role of GI tract colonization with Candida spp. before candidemia has also been described in the NICU; Saiman et al5 demonstrated that colonization preceded candidemia in 15 (43%) of 35 episodes. The importance of the GI tract was further confirmed by Graham et al6 in an analysis of risk factors for BSIs caused by Gram-negative bacilli in VLBW infants. Infants with GI tract pathology, defined as congenital anomalies such as tracheoesophageal fistula, gastroschisis, omphalocele, Hirschprung disease, intestinal atresias, or episodes of necrotizing enterocolitis, were at 5.2-fold increased risk of developing Gram-negative BSIs when compared with controls suggesting that factors that affect the integrity of the bowel wall may allow for translocation of pathogens into the bloodstream causing infection.

For several decades, clinicians and researchers have been attempting to define the best use of surveillance cultures in the NICU. Upon admission to the NICU, only 2% of preterm infants are colonized with Gram-negative flora, but by 15 days of hospitalization the rate increases to 60% and by 30 days increases to 91%.7 In earlier studies, surveillance cultures of the GI tract, pharynx, and/or skin were commonly obtained in efforts to predict which infants would develop infections. Although colonizing flora was common, colonization proved to be a poor predictor of infection8; the sensitivity, specificity, and positive predictive value of surveillance cultures for serious infections has been calculated to be 56%, 82%, and 7.5%, respectively.9 Thus, most NICUs have terminated the practice of performing routine cultures to predict subsequent clinical disease.

Despite the previously noted limitations of surveillance cultures, this study demonstrates that surveillance of the GI tract could potentially be useful in the NICU. Targeted surveillance cultures could improve empiric treatment for late onset sepsis in high risk infants. We demonstrated 100% concordance in susceptibility to gentamicin in colonizing and infecting organisms. Thus, performing surveillance in the NICU has potential to promote antimicrobial stewardship by reducing the use of unnecessarily broad-spectrum antimicrobials and guiding more effective empiric therapy for critically ill infants. In hospitalized adults, Blot et al demonstrated that knowledge of colonization status before infection was associated with higher rates of appropriate therapy for bacteremia caused by antibiotic resistant Gram-negative bacilli.10

This pilot study had limitations. Identification of the 2 predominant phenotypes from a surveillance culture may not be reproducible. Infants may have been colonized with more than 1 strain of each bacterial species. If these differences did not result in different phenotypes, only 1 may have been picked for further testing; this could explain the discordant PFGE results for infant #14, who had different PFGE types of E. Coli isolated from rectal and clinical cultures. Because the number of days between colonization and subsequent BSI varied, the optimal timing for surveillance cultures remains uncertain. The 3 episodes of K. pneumoniae bacteremia in Patient #15 may have been misclassified and not represent unique infections, but rather recrudescence from a persistent infectious focus.

These results were obtained in a research setting, and application in the clinical arena would require significant resource investment and needs further feasibility study. Infants’ dominant colonizing flora and resistance patterns change during their hospitalizations. Understanding the frequency and character of these changes and determining the time period in which infants are most likely to develop infection could allow for an assessment of how targeted surveillance could be adopted into clinical practice. Future studies should address the potential roles of surveillance cultures for guiding empiric therapy for late onset sepsis and determining if outcomes can be improved.

REFERENCES

1. O'Grady NP, Alexander M, Dellinger EP, et al. Guidelines for the prevention of intravascular catheter-related infections. Pediatrics. 2002;110:e51.

2. Sohn AH, Garrett DO, Sinkowitz-Cochran RL, et al; and the Pediatric Prevention Network. Prevalence of nosocomial infections in neonatal intensive care unit patients: Results from the first national point-prevalence survey. J Pediatr. 2001;139:821–827.

3. Garner JS, Jarvis WR, Emori TG, Horan TC, Hughes JM. CDC definitions for nosocomial infections, 1988. Am J Infect Control. 1988;16:128–140.

4. Almuneef MA, Baltimore RS, Farrel PA, et al. Molecular typing demonstrating transmission of gram-negative rods in a neonatal intensive care unit in the absence of a recognized epidemic. Clin Infect Dis. 2001;32:220–227.

5. Saiman L, Ludington E, Phaller M, et al. Risk factors for candidemia in Neonatal Intensive Care Unit patients. The National Epidemiology of Mycosis Survey study group. Pediatr Infect Dis J. 2000;19:319–324.

6. Graham PL, Begg MD, Larson E, Della-Latta P, Allen A, Saiman L. Risk factors for late onset Gram-negative sepsis in low birthweight infants hospitalized in the neonatal intensive care unit. Pediatr Infect Dis J. 2006;25:113–117.

7. Goldmann DA. The bacterial flora of neonates in intensive care–monitoring and manipulation. J Hosp Infect. 1988;11:340–351.

8. Finelli L, Livengood JR, Saiman L. Surveillance of pharyngeal colonization: detection and control of serious bacterial illness in low birth weight infants. Pediatr Infect Dis J. 1994;13:854–859.

9. Evans ME, Schaffner W, Federspiel CF, et al. Sensitivity, specificity, and predictive value of body surface cultures in a neonatal intensive care unit. JAMA. 1988;259:248–252.

10. Blot S, Depuydt P, Vogelaers, et al. Colonization status and appropriate antibiotic therapy for nosocomial bacteremia caused by antibiotic-resistant gram-negative bacteria in an intensive care unit. Infect Control Hosp Epidemiol. 2005;26:575–579.

Keywords:

neonatal infection; gram-negative bloodstream; infection; late-onset sepsis; NICU; gastrointestinal