pediatrics 2014 tsai e322 9

DOI: 10.1542/peds.2013-1248; originally published online January 13, 2014; 2014;133;e322Pediatrics

Ming-Chou Chiang, Ren-Huei Fu, Chiang-Wen Lee and Yhu-Chering HuangMing-Horng Tsai, Shih-Ming Chu, Jen-Fu Hsu, Reyin Lien, Hsuan-Rong Huang,

in the NICURisk Factors and Outcomes for Multidrug-Resistant Gram-Negative Bacteremia

http://pediatrics.aappublications.org/content/133/2/e322.full.htmllocated on the World Wide Web at:

The online version of this article, along with updated information and services, is

of Pediatrics. All rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275.Boulevard, Elk Grove Village, Illinois, 60007. Copyright 2014 by the American Academy published, and trademarked by the American Academy of Pediatrics, 141 Northwest Pointpublication, it has been published continuously since 1948. PEDIATRICS is owned, PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly

at Chulalongkorn University on August 1, 2014pediatrics.aappublications.orgDownloaded from at Chulalongkorn University on August 1, 2014pediatrics.aappublications.orgDownloaded from

Risk Factors and Outcomes for Multidrug-ResistantGram-Negative Bacteremia in the NICU

WHATS KNOWN ON THIS SUBJECT: There is a perception thatGram-negative bacilli (GNB) bloodstream infection is increasing inthe NICU, and those infections caused by a multidrug-resistant(MDR) strain are a growing threat to hospitalized patients.

WHAT THIS STUDY ADDS: Exposure to broad-spectrum antibioticsis the most important risk factor for MDR GNB bacteremia, whichis associated with higher mortality. Neonates with risk factors forbacteremia caused by a MDR GNB strain may benet fromempirical antimicrobial therapy with carbapenem.

abstractOBJECTIVES: To assess the risk factors antibiotic therapy and out-comes of multidrug-resistant (MDR) Gram-negative bacilli (GNB)bacteremia in NICU patients.

METHODS: Episodes of MDR GNB bacteremia were compared witha non-MDR GNB bacteremia group in an 8-year cohort study.

RESULTS: Of 1106 bacteremias, 393 (35.5%) were caused by GNB. Se-venty (18.6%) were caused by an MDR strain. The most frequent mech-anism of resistance was extended-spectrum b-lactamase production(67.1%), mainly by Klebsiella pneumoniae (59.6%). Previous antibioticexposure to third-generation cephalosporin (odds ratio [OR]: 5.97;95% condence interval [CI]: 2.3715.08; P , .001) and carbapenem(OR: 3.60; 95% CI: 1.2610.29; P = .017) and underlying renal disease(OR: 7.08; 95% CI: 1.7428.83; P = .006) were identied as independentrisk factors for MDR GNB acquisition. Patients with MDR GNBbacteremia more likely received inadequate initial antibiotic therapy(72.9% vs 7.8%; P , .001) had higher rates of infectious complication(21.4% vs 10.5%; P = .011) and overall case fatality 1rate (28.6% vs10.5%; P , .001). Independent risk factors for overall mortality werepresence of infectious complications after bacteremia (OR: 3.16; 95%CI: 1.417.08; P = .005) and underlying secondary pulmonaryhypertension with or without cor pulmonale (OR: 6.19; 95% CI: 1.8820.31; P = .003).

CONCLUSIONS: MDR GNB accounted for 18.6% of all neonatal GNB bac-teremia in the NICU, especially in those with previous broad-spectrumantibiotic therapy and underlying renal disease. The most frequentmechanism of resistance was extended-spectrum b-lactamase (ESBL)production. Neonates with MDR GNB were more likely to developinfectious complications, which were independently associated witha higher overall case-fatality rate. Pediatrics 2014;133:e322e329

AUTHORS: Ming-Horng Tsai, MD,a,b,c Shih-Ming Chu, MD,b,d

Jen-Fu Hsu, MD,b,d Reyin Lien, MD,b,d Hsuan-Rong Huang,MD,b,d Ming-Chou Chiang, MD,b,d Ren-Huei Fu, MD, PhD,b,d

Chiang-Wen Lee, MD,c and Yhu-Chering Huang, MD, PhDb,e

aDivision of Neonatology and Pediatric Hematology/Oncology,Department of Pediatrics, Chang Gung Memorial Hospital, Yunlin,Taiwan; bCollege of Medicine, Chang Gung University, Taoyuan,Taiwan; cChang Gung University of Science and Technology,Chiayi, Taiwan; and Divisions of dPediatric Neonatology andePediatric Infectious Disease, Department of Pediatrics, ChangGung Memorial Hospital, Taoyuan, Taiwan

KEY WORDSGram-negative bacilli, nosocomial infection, bacteremia,extended-spectrum b-lactamaseproducing bacteria, mortality

ABBREVIATIONSCIcondence intervalCLSIClinical Laboratory Standards InstituteESBLextended-spectrum b-lactamaseGNBGram-negative bacilliMDRmultidrug-resistantORodds ratio

Dr Tsai conceptualized and designed the study and drafted theinitial manuscript; Drs Chu and Hsu cared for the patients andcollected and veried the data; Dr Lien cared for the patientsand approved the agreement of interstitial review board; DrHuang cared for the patients and carried out the initialanalyses; Drs Chiang and Fu cared for the patients and helpedwith data verication; Dr Lee designed the data collectioninstruments and coordinated and supervised data collection; DrHuang critically reviewed and revised the manuscript; and allauthors approved the nal manuscript as submitted.

www.pediatrics.org/cgi/doi/10.1542/peds.2013-1248

doi:10.1542/peds.2013-1248

Accepted for publication Aug 26, 2013

Address correspondence to Yhu-Chering Huang, MD, PhD, Divisionof Pediatric Infection Disease, Department of Pediatrics, ChangGung Memorial Hospital, 5, Fu-Shing St, Kwei-Shan, Taoyuan 333,Taiwan. E-mail: [email protected]

PEDIATRICS (ISSN Numbers: Print, 0031-4005; Online, 1098-4275).

Copyright 2014 by the American Academy of Pediatrics

FINANCIAL DISCLOSURE: The authors have indicated they haveno nancial relationships relevant to this article to disclose.

FUNDING: No external funding.

POTENTIAL CONFLICT OF INTEREST: The authors have indicatedthey have no potential conicts of interest to disclose.

e322 TSAI et al at Chulalongkorn University on August 1, 2014pediatrics.aappublications.orgDownloaded from

Bacteremia is a common complication ofneonates with a long duration of hospi-talizationintheNICUandisassociatedwithadditional costs, morbidity, and long-termadverse outcomes.1,2 Although the mostcommon pathogen in the NICU is Gram-positive coccus, the incidence of bacter-emia caused by Gram-negative bacilli(GNB) has increased in the past decades,3

and GNB bacteremia is often associatedwith a higher mortality rate.35 The emer-gence of multidrug resistance (MDR)among these organisms deservesparticular concern, because treatmentoptions of antimicrobial agents for anMDR strain are often limited and in-appropriate initial antibiotics will pre-dispose these neonates to an especiallyhigh risk of severe sepsis and a pooroutcome.6,7

When a preliminary blood culturereveals growth of a GNB species, clini-cians require additional guidance indeciding whether the patient can betreated safely with rst-line antimi-crobial agents. Most current studieshave focused on the molecular epide-miology and possible transmissionroute of MDR GNB in NICU patients,811

and clinical data with regard to MDRGNB have only been available in smallcase series or on a single outbreak.12,13

Understanding the characteristics thatdifferentiate critically ill neonates atrisk of infection due to MDR GNB fromthose caused by a non-MDR strain willassist clinicians with an early treat-ment decision for GNB bacteremia. Wetherefore conducted this study to as-sess the incidence of, risk factors foracquisition, antibiotic therapy for, andoutcomes of MDR GNB bacteremia inNICU patients.

METHODS

Setting, Patients, and Study Design

This study was carried out in the NICUof Chang Gung Memorial Hospital, auniversity-afliated teachinghospital innorthern Taiwan. The NICU contained

a total capacity of 49 beds equippedwithventilator and 58-beds of special carenurseries. All infants ,34 to 35 weekscompleted gestation, with a birthweight ,2 kg or .5 kg, or with anyclinical signs of respiratory distressor cardiovascular, gastrointestinal, orneurologic problems requiring surgicalor intensive treatment were eligible toadmission in our NICU. From January2004 to December 2011, all hospitalizedneonates with at least 1 episode ofbacteremia caused by GNB were in-cluded in the study. A neonatology spe-cialist recorded all basic informationon inpatient admission, including de-mographic characteristics, brief hospi-tal course, all nosocomial infections,and discharge diagnosis every weekdaybeginning before January 2004. Thisprospectively collected neonatal data-base contained microbiologic data-bases, and all neonates correspondingto the study denition were retrievedfrom this database.

To identify risk factors for MDR GNB in-fection, all enrolled patients were di-vided into 2 groups: patients withbacteremiaduetoanMDRGNBandthosewithbacteremiaduetoanon-MDRGNB. Ifmultiple episodes of GNB bacteremiaoccurred in an individual during thestudy period, the subsequent episodethat occurred ,1 month after the pre-vious episode of GNB bacteremia wasexcluded from the analysis. We alsocompared patients who died with thosewho survived to determine the in-dependent risk factors for mortality.

More detailed information, including thepresence of a central venous catheter,use of mechanical ventilation andtotal parenteral nutrition, antimicrobialtherapy in the 30 days preceding in-fection, and treatment courses for bac-teremia, were retrospectively reviewedfrom the medical records. Severity ofillness was evaluated at the onset ofeach episode of bacteremia by usingthe Neonatal Therapeutic Intervention

Scoring System.14 Outborn infants whohad been hospitalized in another hos-pital for 2 weeks and those whose de-tailed hospital courses were missing orunavailable were excluded from analy-sis. All recorded data describing thebacteremia episodes were reviewed by2 investigators (S.-M.C. and J.-F.H.) forface validity. This study was approved bythe institutional review board of ChangGung Memorial Hospital, with a waiverof informed consent.

Denitions

Criteria from the Centers for DiseaseControl and Prevention were applied todene neonatal bacteremia.15 Patientswere considered to have an MDR in-fection in the following situations: (1)extended-spectrum b-lactamase (ESBL)producing Enterobacteriaceae, (2) micro-organisms with intrinsic resistancemechanisms such as Stenotrophomonasmaltophilia, and (3) MDR strains in-cluding Pseudomonas aeruginosa andAcinetobacter baumannii. MDR strainswere dened as those resistant to atleast 1 agent in $3 of the following an-timicrobial categories: carbapenems(imipenem andmeropenem), penicillins(piperacillin, ticarcillin, and piperacillin/tazobactam), broad-spectrum cephalospo-rins (ceftazidime and cefepime), mono-bactams (aztreonam), aminoglycosides,and uorquinolones.16 In cases of poly-microbial bacteremia, which was denedas .1 microorganism identied froma single set of blood culture,17 the episodewas dened as an MDR GNB case if 1 ofthe isolates was an MDR GNB strain.

All comorbidities of prematurity, in-cluding respiratory distress syndrome,intraventricular hemorrhage, broncho-pulmonary dysplasia, necrotizing entero-colitis, and periventricular leukomalacia,were dened on the basis of the latestupdated diagnostic criteria in the stan-dard textbook of neonatology.18 Shockwas dened as a mean blood pressure

ARTICLE

PEDIATRICS Volume 133, Number 2, February 2014 e323 at Chulalongkorn University on August 1, 2014pediatrics.aappublications.orgDownloaded from

less than the lower limit according togestational age that was unresponsiveto uid treatment or required vasoac-tive agents.19 Congenital infection andearly-onset sepsis were dened asa denite infectious focus and septice-mia that occurred within the rst 3 daysof life.3 Previous antibiotic therapy wasdened as systemic antibiotic use for.72 hours in the preceding 30 daysbefore bacteremia onset. Empiricalantibiotic therapy was considered tobe inappropriate if the treatmentregimen did not include at least 1 anti-biotic that was active in vitro against theinfecting microorganisms within 24hours of blood culture collection. In-fectious complications were dened asa newly infectious focus or persistentorgan dysfunction occurring within 1week and directly related to bacteremiabut not concurrent with the onset ofbacteremia. Early case-fatality rate wasdened as death within 7 days of thebacteremic episode, and overall case-fatality rate was dened as death byany cause within 30 days of the onset ofbacteremia.

Microbiologic Studies andAntibiotic Policy of Our NICU

In our NICU, empirical antibiotics wereprescribed for the coverage of bothGram-positive and Gram-negative or-ganisms, usually oxacillin or vanco-mycin plus cefotaxime or gentamicin,once late-onset sepsis was suspected.Antimicrobial regimens were modiedat the attending physicians discretion,mostly according to the results andantibiotic susceptibility patterns ofblood cultures. The identication of allcausative microorganisms was per-formed by using standard microbio-logic methods. Antibiotic susceptibilitypatterns were determined accordingto methods recommended by the Na-tional Committee for Clinical Labora-tory Standards Institute (CLSI) for diskdiffusion method, and categorical as-signment was carried out by using CLSI

breakpoints.20 The following agents weretested: ertapenem, imipenem,meropenem,cefepime, ceftazidime, cefotaxime,aztreonam, piperacillin/tazobactam,amoxicillin/clavulanate, ciprooxacin,levooxacin, gentamicin, amikacin, andomoxef. ESBL production was screenedand conrmed in all isolates with a pro-le suggestive of resistance by per-forming a double-disc synergy testaccording to CLSI guidelines.21 Thepresence of blaSHV, blaDHA, blaCMY, andblaCTX-M genes was investigated by poly-merase chain reaction amplication, aspreviously described.22,23 The molecularcharacterization of ESBL GNB was typedby infrequent-restriction-site polymerasechain reaction, and restriction patternswere analyzed by applying previouslyestablished criteria.23

Statistical Analysis

Categorical variableswere comparedbyusing the x2 test or Fishers exact test;odds ratios (ORs) and 95% condenceintervals (CIs) were calculated. Contin-uous variables were compared by theMann-Whitney U test and the t test,depending on the distributions. Multi-variate logistic regression analysis offactors potentially associated with MDRGNB acquisition and mortality includedall statistically signicant variables withP , .1 in univariate analysis, gender

and gestational age, and all clinicallyimportant variables, whether or notthey were statistically signicant. Mea-sures of goodness-of-t were obtainedto assess the performance of the mod-els. The analysis was performed by us-ing the stepwise logistic regressionmodel of SPSS, version 15.0 (IBM SPSSStatistics, IBM Corporation, Armonk, NY).

RESULTS

During the study period, a total of 1106episodes of bacteremia were recorded.Three hundred ninety-three (35.5%) ofthemwerecausedbyGNB ina totalof333neonates. Seventeen episodes of GNBbacteremia were excluded from theanalysis because they occurred ,1month after the previous episode of GNBbacteremia. Of the 376 episodes of GNBbacteremia enrolled into analyses, 70(18.6%) were caused by an MDR strain.In the case and control groups, therewere 70 episodes in 61 patients and 306episodes in 278 patients (9 and 22 ofthese patients, respectively, experi-enced .1 episode of GNB bacteremia).There were 6 patients with multipleepisodes of GNB bacteremia caused byboth an MDR and a non-MDR strain.

Of the 70 episodes of MDR GNB bac-teremia (Table 1), 47 (67.1%) wereESBL-producing bacteria, includingKlebsiella pneumoniae (n = 28; 59.6%),

TABLE 1 MDR GNB and non-MDR GNB Pathogens That Cause Bacteremia in the NICU

Organism MDR GNB (n = 70), n (%) Non-MDR GNB (n = 306), n (%)

Klebsiella pneumoniae 22 (31.4) 68 (22.2)Klebsiella oxytoca 5 (7.1) 31 (10.1)Escherichia coli 12 (17.1) 79 (25.8)Enterobacter cloacae 5 (7.1) 21 (6.9)Enterobacter aerogenes 0 (0) 16 (5.2)Pseudomonas aeruginosa 9 (12.9) 7 (2.3)Acinetobacter baumannii 3 (4.3) 39 (12.7)Serratia marcescens 0 (0) 10 (3.3)Citrobacter freundii 0 (0) 3 (1.0)Stenotrophomonas maltophilia 3 (4.3) 0 (0)Hafnia alvei 0 (0) 2 (0.7)Neisseria meningitidis 0 (0) 2 (0.7)Chryseobacterium meningoseptium 2 (2.6) 0 (0)Flavobacterium 0 (0) 1 (0.3)Morganella morganii 0 (0) 1 (0.3)Polymicrobial microorganisms 9 (12.9) 26 (8.5)


Escherichia coli (n = 9; 19.1%), Klebsi-ella oxytoca (n = 6; 12.8%), and Entero-bacter cloacae (n = 4; 10.6%). Rates ofresistance to nonb-lactam antibioticsamong ESBL-Enterobacteriaceae wereas follows: amikacin, 59.6% (28 of 47);gentamicin, 72.3% (34 of 47); cipro-oxacin, 12.8% (6 of 47); levooxacin,35.6% (16 of 45); and cotrimoxazole,66.7% (14 of 21). All 47 isolates weresusceptible to imipenem, but 4 (8.5%)of them were resistant to ertapenem.The ESBLs were characterized in 38available isolates as follows: 20 fromthe SHV family, 13 from the CTX-M family,and 5 from the combined CTX-M family (2CTX-M3+SHV1, 1 CTX-M14+DHA-1, and 2CTX-M27+CMY-2). The molecular typingof all ESBL-Enterobacteriaceae strainsidentied 29 different pulsed-eld gelelectrophoresis (PFGE) patterns, withthemost common PFGE prole E (total of4 cases) isolated during an epidemic in2009. Four nonESBL-producing isolates,including 3 E coli and 1 E cloacae caseswere found to be resistant to mono-bactams, aminoglycoside, and broad-spectrum cephalosporin. The remainingMDR GNB non-Enterobacteriaceaewere, in order of frequency, as follows:Paeruginosa (n= 9), A baumannii (n= 5),Smaltophilia (n=3), andChryseobacteriummeningoseptium (n = 2). Two cases of Abaumannii were of a pandrug-resistantstrain.16 All other A baumannii and Paeruginosa were resistant to mostb-lactam antibiotics but were susceptibleto carbapenems. S maltophilia strainswere susceptible only to cotrimoxazoleand ciprooxacin. C meningoseptiumisolates were susceptible only tociprooxacin and piperacillin/tazobactam.Among the above isolates, 9 (12.9%)were the polymicrobial bacteremiaepisodes. The GNB isolates in the non-MDR GNB group are summarized inTable 1.

Baseline and demographic character-istics of the neonateswithMDRGNBandnon-MDR GNB bacteremia are shown in

Table 2. Variables including birthweight, gestational age, gender, peri-natal history, and most underlyingchronic conditions were similar be-tween these 2 groups. The presence ofunderlying neurologic sequelae, renaldisease, previous episode of bacter-emia, use of total parenteral nutritionand/or intralipid, central venous cathe-ter, and antibiotic therapy with several

classes of antibiotics within 1 monthbefore bacteremia were more fre-quently found among the MDR GNBgroup. In addition, infants in the MDRGNB group had a signicantly higherrate of being outborn than those in thenon-MDR GNB group (P = .009). Afterapplying a logistic regression model(Table 3), the only independent riskfactors for MDR GNB acquisition were

TABLE 2 Baseline and Demographic Characteristics of all Episodes of GNB Bacteremia in theNeonatal Intensive Care Unit Patients

Characteristic MDR GNB (n = 70) Non-MDR GNB (n = 306) P

Birth body weight, median (IQR), g 1340.0 (8872050.0) 1345.0 (9002047.5) .809Gestational age, median (IQR), wk 29.0 (26.036.0) 30.5 (27.035.0) .544Gender (male/female), n (%) 35/35 (50.0/50.0) 156/150 (51.0/49.0) .969Inborn/outborn, n (%) 37/33 (52.9/47.1) 210/96 (68.6/31.4) .009Age at onset of bacteremia, median (IQR), d 31.5 (14.066.3) 24.0 (13.050.5) .070Perinatal history, n (%)

Cesarean/natural vaginal delivery 40/30 (57.1/42.9) 183/123 (59.8/40.2) .610Premature rupture of membrane 13 (18.6) 57 (18.6) .991Maternal fever and/or chorioamnionitis 1 (1.4) 9 (2.9) .697Low Apgar score at 5 min (#7) 38 (54.3) 130 (42.5) .073Congenital infection and/or early-onset sepsis 2 (2.9) 20 (6.5) .395

Underlying chronic conditionsa, n (%)Congenital anomaliesb 8 (11.4) 16 (5.2) .057Neurologic sequelae, congenital or acquired 16 (22.9) 41 (13.4) .048Cardiovascular diseasec 4 (5.7) 19 (6.2) .871Bronchopulmonary dysplasia 36 (51.4) 134 (43.8) .247Pulmonary hypertension and/or cor pulmonale 5 (7.1) 10 (3.3) .168Congenital gastrointestinal tract pathology 4 (5.7) 18 (4.9) .607Gastrointestinal sequelaed 3 (4.3) 14 (4.6) .605Renal diseasee 9 (12.9) 4 (1.3) ,.001Hematologic disease 0 (0) 3 (1.0) .538

Previous surgery (within 1 month), n (%) 10 (14.3) 45 (14.7) .940Use of corticosteroids (within 1 week), n (%) 11 (15.7) 29 (9.5) .127Previous antibiotic exposure

(within 1 month), n (%)Third-generation cephalosporin 53 (75.7) 108 (35.3) ,.001Vancomycin or teicoplanin 40 (57.1) 103 (33.7) ,.001Carbapenem 12 (17.1) 10 (3.3) ,.001Monobactam 5 (7.1) 10 (3.3) .135Aminoglycoside 41 (58.6) 159 (52.0) .317Antifungal drugs 5 (7.1) 7 (2.3) .037Antianaerobic antibiotics (metronidazole) 11 (15.7) 18 (5.9) .005

On high-frequency oscillatory ventilator, n (%) 11 (15.7) 26 (8.5) .067Invasive mechanical ventilation

(within 1 week), n (%)36 (51.4) 146 (47.7) .575

Use of total parenteral nutrition and/orintrafat, n (%)

56 (80.0) 207 (67.6) .042

Use of central venous catheter, n (%) 61 (87.1) 224 (73.2) .014Previous episode of bacteremia, n (%) 25 (35.7) 72 (23.5) .036

IQR, interquartile range.a Indicating the presence at onset of late-onset sepsis; some patients had .1 underlying chronic complex condition.b Including all neonates with documented and undocumented syndrome, chromosomal anomalies, and genetic and meta-bolic diseases; simple cleft palate and polydactyly were not included.c Including congenital complicated heart disease, cyanotic heart disease, and acyanotic heart disease with heart failuresigns.d Including short bowel syndrome, gastrointestinal pseudo-obstruction, adhesion ileus, and chronic malnutrition.e Including congenital nephrotic syndrome, immunoglobulin A nephropathy, and renal failure requiring hemodialysis.

ARTICLE


previous antibiotic exposure to third-generation cephalosporin (OR: 5.97;95% CI: 2.3715.08; P , .001) and car-bapenem (OR: 3.60; 95% CI: 1.2610.29;P = .017) and presence of underlyingrenal disease (OR: 7.08; 95% CI: 1.7428.83; P = .006). The goodness-of-t testof Hosmer and Lemeshow revealedgood agreement between observedand predicted values of the model (P =.73).

Clinical manifestations, antibiotic treat-ment, and patient outcomes are sum-marized in Table 4. Infants with MDRGNB bacteremia had a signicantlyhigher rate of septic shock than thosewith non-MDR GNB bacteremia (40% vs22.2%; P = .002). Otherwise, no signi-cant differences in terms of clinicalmanifestations were found. All of thepatients received empirical antibioticswithin a few hours after blood culturecollection, and there were no signi-cant differences between groups re-garding the most frequently usedantibiotic type, except that the most-used broad-spectrum antibiotic com-bination, namely carbapenem plusvancomycin or teicoplanin, were morefrequently prescribed in the MDR GNBgroup (18.6% vs 9.2%; P = .023). Infants

with MDR GNB bacteremia more fre-quently received inadequate initialempirical antibiotic treatment whencompared with the susceptible controlgroup, and the time to adequate anti-biotic therapy was also longer (42.7616.8 versus 4.56 10.8 hours; P, .001).Although the duration of mechanicalventilation and hospitalization wascomparable between the 2 groups,bacteremia due to MDR GNB was as-sociated with a poorer outcome andwith signicantly higher rates of in-fectious complication, early case fa-tality, and overall case fatality.

The results of univariate and multi-variate analyses of factors potentiallyassociatedwith overall case fatality aresummarized in Table 5. The mean ges-tational age of neonates with GNBbacteremia who died within 30 dayswas relatively lower than those whosurvived (29.76 5.5 vs 31.16 4.7 days;P = .098). After adjustment, indepen-dent risk factors for mortality werepresence of infectious complicationsafter bacteremia (OR: 3.16; 95% CI:1.417.08; P = .005) and underlyingsecondary pulmonary hypertensionwith/without cor pulmonale (OR: 6.19;95% CI: 1.8820.31; P = .003). The

goodness-of-t Hosmer and Lemeshowtest revealed good agreement betweenobserved and predicted values of themodel (P = .43).

Because our study extended overa relatively long period of time, thestudy subjects were split into thosebefore 2007 and those after 2008 todetermine if the practices within ourNICU may have changed and con-founded the results. However, theresults shown in Tables 25 were stillconsistent over time (data not shown).

DISCUSSION

In this study, we found that MDR GNBbacteremia was not uncommon in theNICU and accounted for nearly one-fthof all episodes of GNB bacteremia. Thisstudy reveals that acquisition of MDRGNB bacteremia is not associated withextremely low birth weight, being bornextremely preterm, perinatal compli-cations, or most underlying chronicconditions, but is associated with un-derlying renal disease and previousantibiotic exposure to third-generationcephalosporin and carbapenem. Themost common mechanism of antimi-crobial resistance in the NICU wasESBL production. In addition, neo-nates with MDR GNB were more likelyto receive inappropriate antibioticsand develop infectious complications,which were independently associatedwith a higher overall case-fatalityrate.

Previous studies have identied verylow birth weight (,1000 g), being bornextremely preterm, and prolonged ex-posure to antimicrobial agents as theindependent risk factors associatedwith resistant Enterobacteriaceae in-fection in critically ill neonates.2427

However, some of these associationswere not observed in this study, and wefurther identied antibiotic exposureof third-generation cephalosporin andcarbapenem within 1 month beforeGNB bacteremia as the independent

TABLE 3 Risk Factors for MDR GNB Acquisition by Univariate and Multivariate Analysis

Risk Factors Univariate Analysis Multivariate Analysis

OR (95% CI) P OR (95% CI) P

Birth weight (every 100-g decrease) 0.99 (0.851.17) .974Gestational age (every 2-week decrease) 1.07 (0.941.22) .285Outborn 2.12 (1.253.59) .005 2.07 (0.884.75) .098Underlying chronic conditions

Congenital anomalies 2.33 (0.965.89) .063 1.01 (0.313.27) .982Neurologic sequelae 1.91 (0.993.65) .051 1.39 (0.643.01) .408Renal disease 11.10 (3.3137.21) ,.001 7.08 (1.7428.83) .006

On high-frequency oscillatory ventilator 2.33 (1.114.89) .026 2.33 (0.965.66) .062Use of total parenteral nutrition and/or intrafat 1.73 (0.933.21) .084 0.64 (0.211.98) .442Use of central venous catheter 2.16 (1.064.42) .035 2.63 (0.739.46) .140Previous episode of bacteremia 1.81 (1.043.15) .037 0.65 (0.311.38) .262Previous antibiotic exposure

Third-generation cephalosporin 5.72 (3.1510.35) ,.001 5.97 (2.3715.08) ,.001Vancomycin or teicoplanin 3.78 (2.196.53) ,.001 0.75 (0.311.82) .523Carbapenem 6.12 (2.5314.84) ,.001 3.60 (1.2610.29) .017Antifungal drugs 3.29 (1.0110.67) .048 2.03 (0.498.43) .327Antianaerobic antibioticsa 2.98 (1.346.64) .007 1.11 (0.413.03) .832

a Metronidazole.


risk factor. These diverse results canbe explained by the different de-nitions of a resistant GNB isolate,2427

different study designs,24,25 differentempirical antibiotic policies, and dif-ferent inclusion criteria (nosocomialinfection instead of bacteremia in theother studies).24,25,27 Previous antibi-otic therapy has been recognized to besignicantly related to bacterial re-sistance development,2428 but allissues regarding appropriate controlgroup selection, case denition, de-scription of previous antibiotic expo-sure, and adjustment for confoundingfactors should be rened. Therefore,

a prospective study should be con-ducted to investigate the impact ofspecic antibiotic type, treatment du-ration or total dosage in grams, andinterval on the development of neo-natal MDR GNB and the colonizationand risk of subsequent bacteremia.29

In the current study, we found signi-cantly higher rates of infectious com-plications and early and overall casefatality for the infants with MDR GNBbacteremia, which is consistent withprevious reports.3032 The durations ofmechanical ventilation and hospitali-zation were comparable between theMDR GNB group and the controls, but

these results may be masked by thehigher proportions of neonates in theMDR GNB group who died within 30days after the onset of bacteremia. Al-though inappropriate empirical anti-biotic therapy has been related tohighermortality in GNB bacteremia,3133

we were unable to establish a directassociation between increased mor-tality in neonates with MDR GNB bac-teremia and a delay in appropriateantibiotics. We found initial inap-propriate antibiotic therapy to be sig-nicantly related to overall mortality(OR: 2.35; 95% CI: 1.234.51; P = .010)and infectious complications (OR: 3.27;95% CI: 1.716.23; P , .001) but notearly case-fatality rate. On the basis ofthe current study, we suspected thatalthough most neonates with MDR GNBbacteremia can survive the rst weekeven after receiving inappropriate an-timicrobial therapy initially, some ofthem would have a signicantly higherrisk of progressively clinical deteri-oration, infectious complications, or anadditional episode of nosocomial in-fection, which may lead to an adverseoutcome.

According to the clinical presentationsand laboratory ndings (data notshown) in the current study, infantswith GNB bacteremia caused by anMDRstrain could not be differentiated fromthose with GNB bacteremia caused bya non-MDR strain until the resultsof antimicrobial susceptibility wereavailable. Although a signicantly higherrate of septic shock was noted ininfants with MDR GNB bacteremia, notall of these cases were due to initialinappropriate antibiotics. Clinical out-comes of these neonates were not onlyaffected by the initial empirical antibi-otic therapy but also by the pathogensand host factors34 such as underlyingconditions or immunity of the patients,concomitant infectious focus, and re-taining or removal of infected cathe-ters or endotracheal tubes.

TABLE 4 Clinical Features, Antibiotic Treatment, and Outcome of Patients With MDR GNBBacteremia Compared With the Susceptible Control Group

Characteristic MDR GNB (n = 70) Non-MDR GNB(n = 306)

P

Clinical features, n (%)Fever 32 (45.7) 134 (43.8) .713Apnea, bradycardia, and/or cyanosis 53 (75.7) 213 (69.6) .207Feeding intolerance and/or abdominal distension 54 (77.1) 209 (68.3) .186Hyperglycemia or hypoglycemia 21 (30.0) 76 (24.8) .312Tachycardia 25 (35.7) 92 (30.1) .275Coagulopathy and/or DIC 33 (47.1) 110 (35.9) .076Concomitant infection 10 (14.3) 42 (13.7) .903Shock within 48 hours after bacteremia onset 28 (40.0) 68 (22.2) .002Polymicrobial bacteremia 9 (12.9) 26 (8.5) .197NTISS score at onset of bacteremia, mean 6 SD 17.3 6 5.4 17.2 6 4.8 .962

Empirical antibiotic treatment, n (%) .349Combination therapy 64 (91.4) 284 (92.8)b-lactam + aminoglycoside 5 (7.1) 25 (8.2) NSb-lactam + third-generation cephalosporin 14 (20.0) 66 (21.6) NSGlycopeptide + aminoglycoside 1 (1.4) 11 (3.4) NSGlycopeptide + third-generation cephalosporin 29 (41.4) 133 (43.5) NSGlycopeptide + carbapenem 13 (18.6) 28 (9.2) .023Above combination + antianaerobes (metronidazole) 2 (2.9) 21 (6.9) NS

Monotherapy 6 (8.6) 22 (7.2) NSThird-generation cephalosporin 2 (2.9) 12 (3.9) NSCarbapenem 2 (2.9) 3 (1.0) NSGlycopeptide 2 (2.9) 7 (2.3) NS

Inadequate initial empirical antibiotic therapya, n (%) 51 (72.9) 24 (7.8) ,.001Time to adequate antibiotic therapy .48 h, n (%) 24 (34.3) 18 (5.9) ,.001Outcome, n (%)

Infectious complicationsb 15 (21.4) 32 (10.5) .011Recurrent bacteremia within one monthc 6 (8.6) 32 (10.4) .650Early case-fatality rate (7 d) 12 (17.1) 22 (7.2) .009Overall case-fatality rate (30 d) 20 (28.6) 32 (10.5) ,.001

Duration of mechanical ventilation, median (IQR), d 44.0 (10.065.8) 34.0 (5.070.5) .423Duration of hospitalization, median (IQR), d 78.5 (42.0122.8) 75.0 (41.0114.0) .769

DIC, disseminated intravascular coagulopathy; IQR, interquartile range; NS, nonsignicant; NTISS, Neonatal TherapeuticIntervention Scoring System.a Indicates patients who did not receive any antimicrobial agent to which the causative microorganisms were susceptiblewithin 24 h of blood culture collection.b Dened as a newly infectious focus or persistent organ dysfunction occurring within 1 week and directly related tobacteremia but not concurrent with the onset of bacteremia.c Includes both Gram-positive and Gram-negative bacteremia.

ARTICLE


Molecular analysis of ESBL-producingstrains in the current study showeduncommon,horizontalpatient-to-patienttransmission in our NICU, becausemostof these ESBL-producing strains weredifferent strains. These results werecompatible with our nding that broad-spectrum antibiotic selection washighly associated with MDR GNB bac-teremia in our NICU. However, severalclonal strains of ESBL-producing Kpneumoniae and E coli were identiedover a period of 2 to 8 months, sug-gesting their prolonged existence inthe NICU environment or long-termcolonization. Therefore, both infectioncontrol measures and applying new

empirical antibiotic policies, especiallydecreased usage of broad-spectrumcephalosporin, may help to reducethe incidence of MDR GNB colonizationand outbreak.35,36 Our results sug-gested that initial antimicrobial regi-men could be based on previousantibiotic use, and local surveillancedata and local NICU antibiograms arerecommended for the optimization ofinitial antibiotic choice.37,38

There were some limitations to thisstudy. Because our data source wasonly a single institution, there might bedifferent epidemics in other centers,although a large cohort of bacteremiaswere studied in this research. Because

our cases of MDR GNB included manyvarieties of both Enterobacteriaceaeand non-Enterobacteriaceae, the inu-ences of different types of micro-organisms on clinical manifestationsand outcomes were not studied.

ACKNOWLEDGMENTSThe authors thank Mrs Chiao-ChingChiang for maintaining the databaseofourNICUandallnursingstaffworkingin our NICUs for keeping extremely de-tailed patient records, which contrib-uted greatly to the completion of thisresearch. Miss Hsiao-Jung Tseng andMrYu-Jr Linhelpedwithstatistical anal-ysis.

REFERENCES

1. Atif ML, Sadaoui F, Bezzaoucha A, et al.Prolongation of hospital stay and addi-tional costs due to nosocomial blood-stream infection in an Algerian neonatalcare unit. Infect Control Hosp Epidemiol.2008;29(11):10661070

2. Schlapbach LJ, Aebischer M, Adams M,et al; Swiss Neonatal Network and Follow-Up Group. Impact of sepsis on neuro-developmental outcome in a Swiss NationalCohort of extremely premature infants.Pediatrics. 2011;128(2). Available at: www.pediatrics.org/cgi/content/full/128/2/e348

3. Cohen-Wolkowiez M, Moran C, Benjamin DK,et al. Early and late onset sepsis in latepreterm infants. Pediatr Infect Dis J. 2009;28(12):10521056

4. Makhoul IR, Sujov P, Smolkin T, Lusky A, ReichmanB. Epidemiological, clinical, and microbio-logical characteristics of late-onset sepsisamong very low birth weight infants in Israel:a national survey. Pediatrics. 2002;109(1):3439

5. Gordon A, Isaacs D. Late onset neonatalGram-negative bacillary infection in Australiaand New Zealand: 1992-2002. Pediatr InfectDis J. 2006;25(1):2529

6. Apisarnthanarak A, Holzmann-Pazgal G,Hamvas A, Olsen MA, Fraser VJ. Antimicro-bial use and the inuence of inadequateempiric antimicrobial therapy on the out-comes of nosocomial bloodstream infec-tions in a neonatal intensive care unit. InfectControl Hosp Epidemiol. 2004;25(9):735741

7. Foglia EE, Fraser VJ, Elward AM. Effect ofnosocomial infections due to antibiotic-resistant organisms on length of stay andmortality in the pediatric intensive careunit. Infect Control Hosp Epidemiol. 2007;28(3):299306

TABLE 5 Risk Factors for Overall Mortality by Univariate and Multivariate Analysis

Risk Factor Survived (n = 324) Died (n = 52), n (%) P Adjusted OR (95% CI) P

Gestational age, median (IQR), wk 31.0 (27.035.0) 28.0 (24.335.0) .031 0.95 (0.881.03) .238Birth weight, median (IQR), g 1387.5 (924.82043.8) 1090 (756.32057.5) .084Male gender, n (%) 167 (51.5) 23 (44.2) .307Outborn, n (%) 112 (34.6) 16 (30.8) .581Low Apgar score at 5 min (#7), n (%) 135 (41.7) 33 (63.5) .003 1.46 (0.663.22) .348Concomitant infection, n (%) 42 (13.0) 10 (19.2) .224Polymicrobial bacteremia, n (%) 34 (10.5) 4 (7.7) .804Initial inadequate antibiotics, n (%) 58 (17.9) 17 (32.7) .008 1.37 (0.513.69) .534Underlying chronic conditions, n (%)

Congenital anomalies 17 (5.2) 7 (13.5) .025 2.79 (0.829.45) .099Neurologic sequelae 43 (13.3) 14 (26.9) .011 1.74 (0.783.86) .175Cardiovascular disease 20 (6.2) 3 (5.8) .906Bronchopulmonary dysplasia 142 (43.8) 28 (53.8) .178Pulmonary hypertension 3 (0.9) 12 (23.1) ,.001 6.19 (1.8820.31) .003Gastrointestinal sequelae 14 (4.3) 3 (5.8) .716Renal disease 7 (2.2) 6 (11.5) .004 3.16 (0.8411.88) .088Hematologic disease 1 (0.3) 2 (3.8) .052

MDR GNB, n (%) 50 (15.4) 20 (38.5) ,.001 1.67 (0.624.52) .311Infectious complications, n (%) 32 (9.9) 15 (28.8) ,.001 3.16 (1.417.08) .005

IQR, interquartile range.


8. Anderson B, Nicholas S, Sprague B, CamposJ, Short B, Singh N. Molecular and de-scriptive epidemiology of multidrug-resistantEnterobacteriaceae in hospitalized infants.Infect Control Hosp Epidemiol. 2008;29(3):250255

9. Toltzis P, Dul MJ, Hoyen C, et al. Molecularepidemiology of antibiotic-resistant gram-negative bacilli in a neonatal intensive careunit during a nonoutbreak period. Pediat-rics. 2001;108(5):11431148

10. Mammina C, Di Carlo P, Cipolla D, et al.Surveillance of multidrug-resistant gram-negative bacilli in a neonatal intensivecare unit: prominent role of cross trans-mission. Am J Infect Control. 2007;35(4):222230

11. Toltzis P, Dul MJ, Hoyen C, et al. The effect ofantibiotic rotation on colonization withantibiotic-resistant bacilli in a neonatal in-tensive care unit. Pediatrics. 2002;110(4):707711

12. Maragakis LL, Winkler A, Tucker MG, et al.Outbreak of multidrug-resistant Serratiamarcescens infection in a neonatal in-tensive care unit. Infect Control Hosp Epi-demiol. 2008;29(5):418423

13. Gngr S, Ozen M, Akinci A, Durmaz R. AChryseobacterium meningosepticum out-break in a neonatal ward. Infect ControlHosp Epidemiol. 2003;24(8):613617

14. Gray JE, Richardson DK, McCormick MC,Workman-Daniels K, Goldmann DA. Neonataltherapeutic intervention scoring system:a therapy-based severity-of-illness index.Pediatrics. 1992;90(4):561567

15. Horan TC, Andrus M, Dudeck MA. CDC/NHSNsurveillance denition of health care-associatedinfection and criteria for specic types ofinfections in the acute care setting. Am JInfect Control. 2008;36(5):309332

16. Magiorakos AP, Srinivasan A, Carey RB,et al. Multidrug-resistant, extensively drug-resistant and pandrug-resistant bacteria:an international expert proposal for in-terim standard denitions for acquiredresistance. Clin Microbiol Infect. 2012;18(3):268281

17. Sutter D, Stagliano D, Braun L, et al. Poly-microbial bloodstream infection in pediat-ric patients: risk factors, microbiology, andantimicrobial management. Pediatr InfectDis J. 2008;27(5):400405

18. Taeusch HW, Ballard RA, Gleason CA. AverysDiseases of the Newborn. 8th ed. Philadelphia,PA: Elsevier Saunders; 2006

19. Kermorvant-Duchemin E, Laborie S, RabilloudM, Lapillonne A, Claris O. Outcome and prog-

nostic factors in neonates with septic shock.Pediatr Crit Care Med. 2008;9(2):186191

20. Clinical and Laboratory Standards Institute.Performance Standards for AntimicrobialSusceptibility Testing: Twenty-second In-formational Supplement M100-S22. Wayne,PA: Clinical and Laboratory Standards In-stitute; 2012

21. Clinical and Laboratory Standards Institute.Performance Standards for AntimicrobialDisk Susceptibility Tests: Approved Stan-dardTenth Informational SupplementM02-A10. Wayne, PA: Clinical and LaboratoryStandards Institute; 2009

22. Chia JH, Chu C, Su LH, et al. Development ofa multiplex PCR and SHV melting-curvemutation detection system for detectionof some SHV and CTX-M beta-lactamases ofEscherichia coli, Klebsiella pneumoniae,and Enterobacter cloacae in Taiwan. J ClinMicrobiol. 2005;43(9):44864491

23. Shu JC, Chia JH, Kuo AJ, Su LH, Wu TL. A 7-year surveillance for ESBL-producingEscherichia coli and Klebsiella pneumo-niae at a university hospital in Taiwan: theincrease of CTX-M-15 in the ICU. EpidemiolInfect. 2010;138(2):253263

24. Singh N, Patel KM, Lger MM, et al. Risk ofresistant infections with Enterobacteriaceaein hospitalized neonates. Pediatr Infect Dis J.2002;21(11):10291033

25. Huang Y, Zhuang S, Du M. Risk factors ofnosocomial infection with extended-spectrumbeta-lactamase-producing bacteria in a neo-natal intensive care unit in China. Infection.2007;35(5):339345

26. Litzow JM, Gill CJ, Mantaring JB, et al. Highfrequency of multidrug-resistant gram-negative rods in 2 neonatal intensive careunits in the Philippines. Infect Control HospEpidemiol. 2009;30(6):543549

27. Linkin DR, Fishman NO, Patel JB, Merrill JD,Lautenbach E. Risk factors for extended-spectrum beta-lactamase-producing Enter-obacteriaceae in a neonatal intensive careunit. Infect Control Hosp Epidemiol. 2004;25(9):781783

28. Le J, Nguyen T, Okamoto M, McKamy S,Lieberman JM. Impact of empiric antibioticuse on development of infections caused byextended-spectrum beta-lactamase bacte-ria in a neonatal intensive care unit.Pediatr Infect Dis J. 2008;27(4):314318

29. Kritsotakis EI, Tsioutis C, Roumbelaki M,Christidou A, Gikas A. Antibiotic use and therisk of carbapenem-resistant extended-spectrum-b-lactamase-producing Klebsiellapneumoniae infection in hospitalized patients:

results of a double case-control study. JAntimicrob Chemother. 2011;66(6):13831391

30. Gudiol C, Tubau F, Calatayud L, et al. Bac-teraemia due to multidrug-resistant Gram-negative bacilli in cancer patients: riskfactors, antibiotic therapy and out-comes. J Antimicrob Chemother. 2011;66(3):657663

31. Peralta G, Snchez MB, Garrido JC, et al.Impact of antibiotic resistance and of ad-equate empirical antibiotic treatment inthe prognosis of patients with Escherichiacoli bacteraemia. J Antimicrob Chemother.2007;60(4):855863

32. Schwaber MJ, Carmeli Y. Mortality anddelay in effective therapy associated withextended-spectrum b-lactamase produc-tion in Enterobacteriaceae bacteraemia:a systematic review and meta-analysis.J Antimicrob Chemother. 2007;60(5):913920

33. Kang CI, Kim SH, Park WB, et al. Bloodstreaminfections caused by antibiotic-resistantgram-negative bacilli: risk factors for mor-tality and impact of inappropriate initialantimicrobial therapy on outcome. Anti-microb Agents Chemother. 2005;49(2):760766

34. Makhoul IR, Sujov P, Smolkin T, Lusky A,Reichman B; Israel Neonatal Network.Pathogen-specic early mortality in verylow birth weight infants with late-onsetsepsis: a national survey. Clin Infect Dis.2005;40(2):218224

35. Lee J, Pai H, Kim YK, et al. Control ofextended-spectrum beta-lactamase-producingEscherichia coli and Klebsiella pneumoniae ina childrens hospital by changing antimicro-bial agent usage policy. J Antimicrob Che-mother. 2007;60(3):629637

36. Velasco C, Rodrguez-Bao J, Garca L, et al.Eradication of an extensive outbreak ina neonatal unit caused by two sequentialKlebsiella pneumoniae clones harbouringrelated plasmids encoding an extended-spectrum beta-lactamase. J Hosp Infect.2009;73(2):157163

37. Paterson DL. Impact of antibiotic resistancein gram-negative bacilli on empirical anddenitive antibiotic therapy. Clin Infect Dis.2008;47(suppl 1):S14S20

38. Bagattini M, Crivaro V, Di Popolo A, et al.Molecular epidemiology of extended-spectrumbeta-lactamase-producing Klebsiella pneu-moniae in a neonatal intensive care unit. JAntimicrob Chemother. 2006;57(5):979982

ARTICLE


DOI: 10.1542/peds.2013-1248; originally published online January 13, 2014; 2014;133;e322Pediatrics

Ming-Chou Chiang, Ren-Huei Fu, Chiang-Wen Lee and Yhu-Chering HuangMing-Horng Tsai, Shih-Ming Chu, Jen-Fu Hsu, Reyin Lien, Hsuan-Rong Huang,

in the NICURisk Factors and Outcomes for Multidrug-Resistant Gram-Negative Bacteremia

ServicesUpdated Information &

tmlhttp://pediatrics.aappublications.org/content/133/2/e322.full.hincluding high resolution figures, can be found at:

References

tml#ref-list-1http://pediatrics.aappublications.org/content/133/2/e322.full.hat:This article cites 35 articles, 15 of which can be accessed free

Subspecialty Collections

_of_care_transition_-_discharge_planning_subhttp://pediatrics.aappublications.org/cgi/collection/continuityContinuity of Care Transition & Discharge Planning

medicine_subhttp://pediatrics.aappublications.org/cgi/collection/hospital_Hospital Medicine

gy_subhttp://pediatrics.aappublications.org/cgi/collection/neonatoloNeonatology

orn_infant_subhttp://pediatrics.aappublications.org/cgi/collection/fetus:newbFetus/Newborn Infantthe following collection(s):This article, along with others on similar topics, appears in

Permissions & Licensing

tmlhttp://pediatrics.aappublications.org/site/misc/Permissions.xhtables) or in its entirety can be found online at: Information about reproducing this article in parts (figures,

Reprints http://pediatrics.aappublications.org/site/misc/reprints.xhtml

Information about ordering reprints can be found online:

rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275.Grove Village, Illinois, 60007. Copyright 2014 by the American Academy of Pediatrics. All and trademarked by the American Academy of Pediatrics, 141 Northwest Point Boulevard, Elkpublication, it has been published continuously since 1948. PEDIATRICS is owned, published, PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly

at Chulalongkorn University on August 1, 2014pediatrics.aappublications.orgDownloaded from

pediatrics 2014 tsai e322 9

Documents