Original Paper Neonatology 2011;99:23–31 DOI: 10.1159/000292567
Received: September 3, 2009 Accepted after revision: February 15, 2010 Published online: June 30, 2010
Risk Factors Associated with Development of Persistent Coagulase-Negative Staphylococci Bacteremia in the Neonate and Associated Short-Term and Discharge Morbidities Ann Anderson-Berry Britt Brinton Elizabeth Lyden Roger G. Faix Department of Pediatrics, Divisions of Neonatology and Infectious Diseases, Infection Control Office, University of Utah Health Sciences Center, Salt Lake City, Utah, and Department of Pediatrics, Division of Neonatology, and College of Public Health, University of Nebraska Medical Center, Omaha, Nebr., USA
Key Words Coagulase-negative staphylococci ⴢ Late-onset bacteremia ⴢ Nosocomial infection ⴢ Neonatal intensive care unit
Abstract Background: Coagulase-negative staphylococci (CoNS) are the most common cause of late-onset sepsis in neonatal intensive care unit settings. Historically, authors have questioned the clinical significance of these bacteremia episodes. Recently, clusters of CoNS bacteremia associated with significant morbidity have been reported. The frequency and importance of these clusters of bacteremia and their associated morbidity remains unclear. Objective: We studied a prolonged cluster of 52 cases of persistent CoNS bacteremia in a level III neonatal intensive care unit to clarify risk factors, morbidity and outcomes associated with persistent CoNS bacteremia. Study Design: A retrospective case-control study of infants with CoNS bacteremia 1 48 h after initiation of appropriate antibiotics and gestational age-matched control infants was performed. We reviewed patient characteristics prior to and during bacteremia, and outcomes at discharge or death. Results: Persistently infected infants were significantly more likely to have greater duration of exposure to parenteral nutrition, hydrocortisone, antibiotics, and me-
© 2010 S. Karger AG, Basel 1661–7800/11/0991–0023$38.00/0 Fax +41 61 306 12 34 E-Mail
[email protected] www.karger.com
Accessible online at: www.karger.com/neo
chanical ventilation prior to infection. Persistently infected infants were significantly more likely than controls to experience feeding intolerance and to require inotropic support, increased respiratory support, and blood product transfusion during bacteremia. Infants with persistent CoNS bacteremia took longer to achieve full enteral feeds, had higher rates of chronic lung disease and increased length of stay compared to controls. No significant difference in mortality was noted. Conclusions: Persistent CoNS bacteremia is associated with clinically significant morbidity including feeding intolerance, respiratory failure, blood transfusion and chronic lung disease in this patient population. This impacts hospital course, increases length of stay and impacts medical needs after discharge. During this study, CoNS bacteremia was difficult to clinically eradicate despite in vitro antibiotic susceptibility. Copyright © 2010 S. Karger AG, Basel
Coagulase-negative staphylococci (CoNS) are the bacteria most frequently recovered from blood culture in infants 63 days of age at most neonatal intensive care units (NICU) in North America [1]. CoNS have generally been considered to be a rare cause of fulminant, fatal disease, and an infrequent cause of significant clinical deterioration [2–5], although death and/or severe morbidity in Ann Anderson-Berry, MD Joint Division of Neonatology, University of Nebraska School of Medicine and Creighton University School of Medicine Omaha, NE 68198 (USA) Tel. +1 402 559 9280, Fax +1 402 559 9333, E-Mail alanders @ unmc.edu
NICU infants have been attributed to this organism [6, 7]. It is not always clear whether CoNS recovered in culture from a normally sterile site represents a contaminant (CoNS pseudo-bacteremia) or a true infection [8]. True bloodstream infection can be either non-persistent CoNS bacteremia (true bacteremia resolving !48 h after initiation of appropriate antibiotics) or persistent CoNS bacteremia (lasting 1 48 h after initiation of treatment with antibiotics to which the organisms were susceptible in vitro). Most clinicians accept that persistent recovery of this organism over several days or from several normally sterile sites is consistent with a truly infectious episode [7, 9–12]. We report a recent cluster of episodes of persistent CoNS bacteremia that occurred over more than 2 years and involved 52 infants. To define risk factors and clinical outcomes associated with persistent CoNS bacteremia, we conducted a case-control study comparing the affected infants to a cohort matched for estimated gestational age and dates of hospitalization in our NICU. We hypothesized that infants with persistent CoNS bacteremia would sustain increased morbidity, including respiratory failure, transfusion requirement, and feeding intolerance, during the persistent episode compared to control infants. We also speculated that length of stay (LOS) and discharge outcomes such as chronic lung disease (CLD) will increase.
Methods Study Population The institutional review board at the University of Utah Health Sciences Center (Salt Lake City, Utah, USA) approved this study, waiving the need for parental consent. A retrospective cohort analysis was first performed to assess temporal trends among infants admitted to the University of Utah NICU from January 1, 1999, through December 31, 2004, for episodes of blood culture positive for CoNS and for events of persistent CoNS bacteremia (blood culture remaining positive after 648 h of antibiotic therapy to which the isolate was deemed susceptible on in vitro testing). The computerized microbiology database at the University of Utah was queried to identify infants admitted to the NICU, including both live births at University Hospital and infants born at and transferred from another hospital, who had a blood culture positive for CoNS at 63 days of age between January 1999 and December 2004. Any infant not continuously hospitalized from birth was excluded. For infants with more than one episode of CoNS bacteremia during their hospitalization, only the first episode was included in the case-control study analysis. Infants who had focal infection at the time bacteremia was first detected were also excluded from analysis, although the presence of occult foci could not be absolutely precluded in those who were included. No attempt was made to determine whether episodes consisting of a single positive blood culture for CoNS represented contamination.
24
Neonatology 2011;99:23–31
Study Design We performed a retrospective case-control study to examine risk factors associated with persistent CoNS bacteremia in the NICU and to assess the morbidity and mortality associated with persistent CoNS bacteremia during the persistent episode and subsequent hospitalization. Infants were studied from 2002 and 2003, the years with the highest rates of persistent CoNS bacteremia. For each case with persistent CoNS bacteremia, a pool of potential controls matched for estimated gestational age, year of hospitalization and continued hospitalization at the same postnatal age as the corresponding case infant when first diagnosed with CoNS was identified from a database of all NICU admissions during 2002 and 2003. Four controls were randomly selected from this pool for the corresponding case infant. Complications and outcomes associated with persistent CoNS were compared to controls of the same postnatal age. Cases were not matched for intrauterine growth restriction, multiple gestation or gender. Clinical Management Blood cultures were processed with a commercial computerassisted, continuously reading automated system. Standard methodology was used for antimicrobial susceptibility testing. All infants underwent assessment of peak and trough serum vancomycin concentrations and adjustment of the dosage regimen to attain peak 120 and trough ⬃10 g/ml during the first 6 months of the study period, with increases to peak ⬃40 and trough 15–20 g/ml if bacteremia persisted, and increase of target concentrations routinely for all infants during the remainder of the study period. Gentamicin was continued if susceptibility was confirmed. Rifampin was typically added if bacteremia persisted for 4 days despite therapy with agents effective on in vitro testing, and linezolid was added if bacteremia persisted for an additional 10 days [13]. Antimicrobial therapy evolved clinically during this time period as persistent CoNS bacteremia with clinical deterioration was observed and determined to require more aggressive intervention to resolve. Indwelling vascular catheters were removed if treatment through the catheter failed to yield sterilization by 2 days. However, removal of catheters was not associated with resolution of the persistent CoNS infection. Blood cultures were drawn from any in situ indwelling vascular catheters as well as peripheral sites, and relative time to positive culture employed to determine if the catheter was the likely source of infection [14]. Echocardiography was routinely performed after 4 days, along with sonographic evaluation for thrombi or vegetations in the great vessels. Systematic physical examination was performed at least daily to assess for soft tissue, bone, joint or other metastatic infection, and additional imaging studies were obtained as determined by the responsible clinician or consultant [15]. Abscesses or sequestered infection were rare (n = 3), but were surgically drained. Cultures of blood were obtained daily (every other day in a few cases) until sterilization was confirmed (3 consecutive negative cultures with at least 24 h separating each, with the first of the 3 defining the date of sterilization) [16]. Cerebrospinal fluid and catheterized or percutaneously aspirated urine were obtained in most persistently infected infants who could tolerate the sampling procedures. Study Definitions For the purposes of this study, respiratory failure was defined as a requirement to increase respiratory support by mean airway
Anderson-Berry/Brinton/Lyden/Faix
Data Analysis The incidence proportion of CoNS bacteremia versus other organisms isolated from blood was calculated per 1,000 NICU admissions in the 6-year cohort, and incidence ratios and 95% confidence intervals (CIs) were used to detect differences [17, 18]. To determine independent associations between potential risk factors and persistent CoNS bacteremia in the case-control study, odds ratios (ORs) with estimated 95% CIs were calculated using univariate, conditional logistic regression. Continuous risk factors were entered into the model as continuous variables. Risk factors prior to CoNS onset which were significant in univariate analysis (p ! 0.05) were included in a backward selection, multiple logistic regression model to determine the best subset of risk factors to predict persistent CoNS.
Results
Both CoNS bacteremia and persistent CoNS bacteremia were observed to increase significantly during the first 5 years of the retrospective cohort analysis (fig. 1), even with the exclusion of second or subsequent episodes of such infections in individual subjects. Further analysis using the case-control format was conducted for 2002 and 2003, the years with the highest frequencies of these types of infection. Data from these 2 years were analyzed as one cluster of increased CoNS bacteremia. It is notable that roughly 50% of first episodes of CoNS bacteremia during these 2 years were also persistent episodes. Persistent CoNS bacteremia was identified in 52 infants during 2002 to 2003. These 52 infants were then compared to 206 controls matched for estimated gestational age and year of hospitalization and continued hospitalization at the postnatal age at which the corresponding study patient underwent onset of the persistent bacteremia. For 2 of the 52 infants, only 3 matched controls were available. Duration of persistence ranged from 3 to 44 days. In vitro susceptibility testing revealed all of the CoNS isolates to be susceptible to vancomycin [minimum inhibitory concentration (MIC) !4.0 g/ml], while susceptibility to gentamicin, clindamycin, oxacillin and Morbidity of Coagulase-Negative Staphylococci
CoNS bacteremia
Color version available online
100
Persistent CoNS bacteremia
90 80 70
Episodes/1,000 admissions
pressure 62 cm H2O, increase in O2 by 620% or to institute mechanical ventilation. Feeding intolerance was represented by the need to place the infant NPO or to decrease enteral feedings by 620 ml/kg/day. Transfusion needs were determined by the responsible attending physician (although practice guidelines in the NICU suggested packed red blood cells for hematocrit !35% if requiring new or increasing ventilator support or supplemental oxygen; platelets for platelet count ^50,000/mm3; fresh frozen plasma if prothrombin time 117 s, and cryoprecipitate if fibrinogen !100 mg/dl). LOS was defined as the time of birth through discharge home or death. CLD was defined as requirement for supplemental oxygen at 36 weeks postmenstrual age.
60 50 40 30 20 10 0 1999
2000
2001
2002
2003
2004
Year
Fig. 1. Frequency of CoNS bacteremia and persistent CoNS bac-
teremia (bacteremia 1 48 h after initiation of antibiotic therapy to which isolate was susceptible per in vitro testing) between January 1, 1999, and December 31, 2004, among University of Utah NICU patients. Frequency of both infections increased significantly from 1999 to 2003 (p ! 0.001). (If an infant had more than one bacteremia episode, only the first episode is reported.)
trimethoprim-sulfamethoxazole varied from 17 to 69%. It is notable that the median MIC of vancomycin for CoNS isolates rose from 0.5 g/ml in 1997 to 1.5 g/ml by 2003. First-line antibiotic therapy was impacted by this cluster of CoNS, with vancomycin dosing adjusted to achieve higher peak and trough levels. Prior to the first positive culture, infants with persistent CoNS bacteremia were more likely to have prior exposure to total parenteral nutrition (PN), hydrocortisone, indwelling central catheter within 72 h, a greater number of total days of antibiotic exposure, a greater number of ventilator days, and to have an absolute neutrophil count (ANC) !1,000 at any time (table 1). Few were still receiving hydrocortisone (n = 3) or still had ANC !1,000 (n = 0) at the time persistent CoNS bacteremia was detected. None received dexamethasone, granulocyte colony-stimulating Neonatology 2011;99:23–31
25
Table 1. Patient characteristics prior to CoNS bacteremia
Characteristic
Persistent CoNS (n = 52)1
Gestational age (range), weeks Birth weight, g Male sex TPN days Central line associated (indwelling within 72 h) ANC