Abbreviations

and Acronyms

TMP/SMX ¼ trimethoprim/sulfamethoxazole

TSN ¼ The Surveillance Network

UTI ¼ urinary tract infection

Accepted for publication October 4, 2013.* Correspondence: Department of Urology,

University of California-San Francisco, 400 Parnas-sus Ave., Suite A-610, San Francisco, California94143.

1608 j www.jurology.com

00

T

©

Urinary Tract Infection/Vesicoureteral Reflux

A Comparison of Inpatient Versus Outpatient ResistancePatterns of Pediatric Urinary Tract Infection

Kara N. Saperston,* Daniel J. Shapiro, Adam L. Hersh and Hillary L. Copp

From the University of California-San Francisco, San Francisco, California, and University of Utah (ALH), Salt Lake City, Utah

Purpose: Prior single center studies showed that antibiotic resistance patternsdiffer between outpatients and inpatients. We compared antibiotic resistancepatterns for urinary tract infection between outpatients and inpatients on anational level.

Materials and Methods: We examined outpatient and inpatient urinary isolatesfrom children younger than 18 years using The Surveillance Network (EurofinsScientific, Luxembourg, Luxembourg), a database of antibiotic susceptibilityresults, as well as patient demographic data from 195 American hospitals. Wedetermined the prevalence and antibiotic resistance patterns of the 6 mostcommon uropathogens, including Escherichia coli, Proteus mirabilis, Klebsiella,Enterobacter, Pseudomonas aeruginosa and Enterococcus. We compared differ-ences in uropathogen prevalence and resistance patterns for outpatient andinpatient isolates using chi-square analysis.

Results: We identified 25,418 outpatient (86% female) and 5,560 inpatient (63%female) urinary isolates. Escherichia coli was the most common uropathogenoverall but its prevalence varied by gender and visit setting, that is 79% ofuropathogens overall for outpatient isolates, including 83% of females and 50%of males, compared to 54% for overall inpatient isolates, including 64% of femalesand 37% of males (p <0.001). Uropathogen resistance to many antibiotics waslower in the outpatient vs inpatient setting, including trimethoprim/sulfameth-oxazole 24% vs 30% and cephalothin 16% vs 22% for E. coli (each p <0.001),cephalothin 7% vs 14% for Klebsiella (p ¼ 0.03), ceftriaxone 12% vs 24% andceftazidime 15% vs 33% for Enterobacter (each p <0.001), and ampicillin 3% vs13% and ciprofloxacin 5% vs 12% for Enterococcus (each p <0.001).

Conclusions: Uropathogen resistance rates of several antibiotics are higher forurinary specimens obtained from inpatients vs outpatients. Separate outpatientvs inpatient based antibiograms can aid in empirical prescribing for pediatricurinary tract infections.

Key Words: urinary tract infections, drug resistance,

anti-bacterial agents, inpatients, outpatients

ANTIBIOTIC resistance in pediatric pa-tients is increasing.1�3 Fewer than50% of all pediatric UTIs are suscep-tible to commonly used antibiotics.4

Because identification and suscepti-bilities are not available at the point

22-5347/14/1915-1608/0

HE JOURNAL OF UROLOGY®

2014 by AMERICAN UROLOGICAL ASSOCIATION EDUCATION AND RESEARC

of care, antibiograms are usefulaids for empirical treatment of UTIwhile cultures are pending. Hospitalbased laboratory data combine out-patient and inpatient sensitivityand resistance patterns to generate

H, INC.

http://dx.doi.org/10.1016/j.juro.2013.10.064

Vol. 191, 1608-1613, May 2014

Printed in U.S.A.

RESISTANCE PATTERNS OF PEDIATRIC URINARY TRACT INFECTION 1609

antibiograms for empirical antibiotic prescribingand yet these data may not accurately reflect uro-pathogen resistance patterns in outpatients.3,5,6

Studies from single centers show that antibioticresistance patterns for pediatric UTI differ bysetting with generally higher resistance ratesamong inpatients than outpatients. Based on thesefindings these studies suggest that antibiogramsshould separate data on outpatients from those oninpatients to maximize the usefulness of antibio-grams for empirical antibiotic selection for UTItreatment. To our knowledge the extent to whichthese differences in resistance patterns betweenoutpatient and inpatient UTIs exist more broadlynationally is unknown.

We compared national patterns of antibioticresistance among common uropathogens betweenantibiograms obtained for outpatients and in-patients. The results of this study show the impor-tance of developing UTI specific antibiogramsstratified by the site where the culture was obtained.

METHODS

Study DesignIn this retrospective study of microbiological results ofpediatric urine cultures we examined urinary isolatesfrom children younger than 18 years that were collectedin the inpatient and outpatient setting from clinical lab-oratories throughout the United States in 2009.

Data SourcesAs previously described,7 we analyzed data from TSN,an electronic surveillance database. TSN collects strainspecific, qualitative and quantitative antimicrobial testresults and patient demographic data from clinical labo-ratories at 195 American hospitals, including academic,nonacademic, pediatric hospitals and governmental hos-pitals, in all 9 United States Census Bureau regions,including Pacific, Mountain, West North Central, EastNorth Central, New England, Mid Atlantic, SouthAtlantic, East South Central and West South Central.Data include the antimicrobial agents tested, organismsidentified, infection site, institution type and test meth-odology. Patient demographic information, including ageand gender, are also available.

Susceptibility testing is performed at all participatinglaboratories using standard United States Food andDrug Administration testing methods. Urine isolate testresults are interpreted according to the NCCLS (NationalCommittee for Clinical Laboratory Standards). The NCCLSsets the standard for the methodology used for suscepti-bility testing, including antibiotic selection, minimuminhibitory concentration parameters, quality control andtest interpretation.8 When multiple isolates were collectedfrom the same patient in a 5-day period, the first isolatewas used to determine susceptibility patterns.

Outpatient urine cultures were defined as urine iso-lates obtained at ambulatory clinics and emergency de-partments. Inpatient cultures were defined as urinary

isolates obtained in the inpatient hospital setting. Weexcluded from study all urinary isolates from skillednursing facilities and rehabilitation facilities. To limitan overestimation of uropathogen resistance we useda strict definition of resistance and included only organ-isms classified as resistant. Bacteria with intermediatesusceptibility were not classified as resistant by our defi-nition since many antibiotics are highly concentrated inurine and may be adequate for treatment. To decreasethe possibility of contamination we limited isolates forthis study to those in which only a single organism grew.

MeasurementsWe compared resistance patterns in the outpatient vsinpatient setting for the 6 most common uropathogens,including Escherichia coli, Enterobacter, Enterococcus,Klebsiella, Pseudomonas aeruginosa and Proteusmirabilis. We report aggregate data on each organismand each of the 15 antibiotics, including TMP/SMX,ampicillin, amoxicillin-clavulanate, nitrofurantoin, ceph-alothin, cefuroxime, ceftriaxone, cefazolin, ceftazidime,gentamicin, ciprofloxacin, piperacillin/tazobactam, imipe-nem, aztreonam and vancomycin. Patients were stratifiedby gender, age and visit setting.

Statistical AnalysisSummary statistics were performed using the frequencyand proportion for categorical data. Chi-square analysiswas done to compare differences in uropathogen preva-lence and resistance patterns between outpatient andinpatient urinary isolates. All analysis was 2-sided withp <0.05 considered statistically significant. Analyseswith fewer than 10 observations were performed withthe Fisher exact test.

RESULTSWe identified a total of 30,978 urinary isolates inthe database, including all inpatient and outpatientsettings. Of the 25,418 outpatient urinary isolates86% were from females. Of the 5,560 inpatient uri-nary isolates 63% were from females.

Organism Prevalence

We stratified urinary isolates by gender and visitsetting (see figure). E. coli was the most commonlycultured organism in males and females in eachsetting, followed by Enterococcus. The E. coli prev-alence in females was 83% (95% CI 83e84) amongoutpatients and 64% (95% CI 63e66) among in-patients (p <0.001). The prevalence of E. coli inmales in the outpatient setting was 50% (95% CI48e52) compared with 37% (95% CI 35e39) in theinpatient setting (p <0.001). The prevalence ofEnterococcus in females was 5% (95% CI 5e6)among outpatients and 13% (95% CI 12e14) amonginpatients (p <0.001). Similarly, in male outpatientsthe Enterococcus prevalence was 17% (95% CI16e18) compared with 27% (95% CI 25e29) in in-patients (p <0.001). The difference in outpatient vs

Uropathogen prevalence by patient setting and gender. spp., species.

1610 RESISTANCE PATTERNS OF PEDIATRIC URINARY TRACT INFECTION

inpatient prevalence in males and females for allother uropathogens was less than 10%.

Given the large number of total E. coli isolates,it was possible to stratify the prevalence of E. coliisolates by age. The E. coli prevalence was higherin outpatient than inpatient isolates for each genderin all age groups

Antibiotic Resistance

The table lists the rates of uropathogen resistanceto various antibiotics stratified by patient setting.We identified numerous differences between theoutpatient and inpatient resistance patterns of theseorganisms. Inpatient resistance rates frequentlyexceeded outpatient resistance rates, especially forthird generation cephalosporins and ciprofloxacin.Notably, E. coli resistance to TMP/SMX was high foroutpatient and inpatient isolates.

DISCUSSIONUsing the TSN database we compared outpatientand inpatient resistance patterns for the 6 mostcommon pediatric uropathogens. This study had2 main findings that may affect the treatment ofchildren with UTIs. 1) Different resistance patternswere noted between outpatient and inpatient set-tings for 5 of the 6 common pediatric uropathogensusing a nationwide database. 2) Prevalence patternsvary based on patient setting, gender and age.These differences should be considered whenempirically treating children who present with uri-nary tract symptoms, and when developing anti-biograms. Our findings extend what was previouslyfound in single center studies.5,9

The different resistance patterns in outpatientand inpatient settings affect empirical antibiotic

choices when treating a child with a UTI. Cliniciansmust choose an antibiotic with a high likelihoodof coverage while considering potential adverse ef-fects and minimizing unnecessary overuse of broad-spectrum antibiotics. Since antibiotics are generallyprescribed before the return of culture results,antibiograms are published to guide empiricalantibiotic choice.

Antibiograms are designed by a set of guidelines8

using antimicrobial resistance and sensitivity data.At most hospitals combined outpatient and inpa-tient laboratory data are currently used to create anantibiogram.10,11

Dahle et al recently examined different suscep-tibility patterns of urinary isolates in a single healthcare system, comparing a community based uro-pathogen antibiogram to a hospital based uropatho-gen antibiogram for children in Utah.5 Similar toour findings, they determined that there was a dif-ference in resistance patterns between outpatientand inpatient uropathogens. We evaluated whetheroutpatient vs inpatient resistance patterns wouldfollow similar trends on a larger, broader scale.

In addition to finding different susceptibilities, wenoted different uropathogen prevalence patterns inmale and female patients. As expected, E. coli wasthe predominant female uropathogen in outpatientcultures. Male outpatient urine cultures had morevariability in organisms. These variations in preva-lence by gender are important observations thatshould be used to guide empirical antibiotic selec-tion, especially since nonE. coli uropathogens areless likely to be susceptible to narrow spectrumantibiotics. The relatively high frequency of Entero-coccus is a particularly important finding in males,which may guide different empirical antibioticchoices based on gender.

Uropathogen resistance rates by setting

% TMP/SMX(95% CI)

% Nitrofurantoin(95% CI)

% Ampicillin(95% CI)

% Cephalothin(95% CI)

% Cefazolin(95% CI)

% Ceftriaxone(95% CI)

% Ceftazidime(95% CI)

% Amoxicillin-Clavulanate(95% CI)

% Ciprofloxacin(95% CI)

% Vancomycin(95% CI)

E. coli: Not done*Outpt 24 (23e24) Less than 1 (less than

1eless than 1)45 (44e46) 16 (15e17) 4 (4e4) Less than 1 (less than

1eless than 1)Less than 1 (less than

1eless than 1)5 (5e5) 5 (4e5)

Inpt 30 (29e32) Less than 1 (less than1eless than 1)

55 (53e57) 22 (19e25) 8 (7e9) 2 (2e3) 2 (1e2) 6 (5e8) 9 (8e10)

p Value (chi-squaretest)

<0.001 0.25 <0.001 <0.001 <0.001 <0.001 <0.001 0.09 <0.001

Enterobacter: Not done* Not done*Outpt 18 (14e21) 23 (19e27) 78 (74e83) 96 (91e100) 91 (88e94) 12 (9e15) 15 (11e19) 1 (0e2)Inpt 13 (10e17) 21 (17e25) 80 (75e85) 98 (94e100) 93 (90e95) 24 (19e29) 33 (28e38) 1 (0e2)p Value (chi-square

test)0.08 0.52 0.66 0.69† 0.44 <0.001 <0.001 1†

Enterococcus: Not done* Not done* Not done* Not done* Not done* Not done*Outpt 1 (1e2) 3 (2e3) 5 (3e7) Less than 1 (less

than 1e1)Inpt 5 (4e7) 13 (11e16) 12 (9e16) 6 (5e8)p Value (chi-square

test)<0.001 <0.001 <0.001 <0.001

Klebsiella: Not done*Outpt 15 (13e17) 17 (15e19) 81 (79e84) 7 (3e10) 7 (6e8) 2 (1e2) 2 (2e3) 4 (2e5) 3 (2e4)Inpt 17 (14e20) 13 (10e16) 82 (79e86) 14 (8e21) 13 (11e16) 4 (2e5) 4 (3e6) 4 (2e7) 4 (3e6)p Value (chi-square

test)0.38 0.06 0.58 0.03 <0.001 0.03 0.03 0.66† 0.2

P. mirabilis: Not done*Outpt 11 (10e13) 94 (92e95) 12 (11e14) 4 (0enot

available)4 (3e5) Less than 1 (less than

1eless than 1)Less than 1 (less than

1eless than 1)1 (0e2) 3 (2e4)

Inpt 13 (8e19) 88 (83e93) 11 (6e16) 0 4 (1e7) 0 (0e0) Less than 1 (lessthan 1e2)

0 (0e0) 6 (3e10)

p Value (chi-squaretest)

0.42 0.01 0.61 0.62† 1† 1† 0.59† 1† 0.02

P. aeruginosa: Not done Not done*Outpt 94 (92e96) 0 (0e0) 31 (24e38) 4 (3e6) 5 (4e7)Inpt 95 (92e98) 0 (0e0) 40 (33e47) 10 (7e13) 11 (8e14)p Value (chi-square

test)0.7 Not applicable 0.08 <0.01 <0.01

*Testing not performed since organism is not susceptible.† Fisher exact test used when percents used to calculate chi-square statistic were based on fewer than 10 observations.

RESISTANCEPATTERNSOFPEDIATRIC

URINARYTRACTINFECTIO

N1611

1612 RESISTANCE PATTERNS OF PEDIATRIC URINARY TRACT INFECTION

As clinicians struggle with the growing rate ofresistance, the need to target empirical antibioticsbecomes more challenging. Prior studies demon-strated that empirical antibiotics are often chosenincorrectly based on longstanding prescribinghabits.9 This concept was studied by McGregor et alwhen they tracked the increasing prevalence ofMRSA (methicillin resistant Staphylococcus aureus)in the outpatient setting.12 They observed thatphysicians changed their choice of empirical treat-ment of cellulitis when they had access to outpatientbased antibiograms with appropriate resistance andsensitivity patterns for community acquired MRSA.However, if they were not provided with antibio-grams, physicians often continued with prior pre-scribing practices. This becomes clinically relevantwhen selecting an empirical antibiotic for a patientwith a UTI at a family practice clinic. Copp et alnoted that Bactrim� is a leading antibiotic pre-scribed by many physicians for pediatric UTIs13 butin our study it had a fairly high outpatient resis-tance rate.

Along these lines, Boggan et al reported thatwhen practitioners are provided with antibiogramsthat target the specific diagnosis of the patient,physicians select more narrow spectrum antibi-otics.9 The investigators asked physicians to choosean empirical antibiotic for theoretical children ofdifferent ages with UTIs. When no antibiogram wasavailable, physicians chose an effective antibiotic32% of the time. When an antibiogram was avail-able that combined adult and pediatric uropathogendata, physicians chose an effective antibiotic 57%of the time. However, when these physicians hadaccess to a pediatric specific antibiogram, 79% chosean effective antibiotic. The mentioned studies sup-port the fact that when clinicians are aware ofresistance and prevalence patterns, more informedand effective antibiotic choices can be made forempirical treatment of UTIs.

This study cannot be used to determine anempirical antibiotic for a specific patient at presen-tation with a UTI. However, the different urinaryisolate resistance patterns of outpatient vs inpatientsettings demonstrate the need to create separateantibiograms at institutions.

Relation to Clinical Practice

A patient who presents to a clinic at age less than2 years with fever and urine studies suggestive ofa UTI most likely requires empirical antibiotics.Usually the patient is on empirical antibiotics forat least 48 hours before culture results are avail-able. A local antibiogram based on urine culturescan be an important tool to aid in the selection ofan antibiotic with a high likelihood of treating theuropathogen.

When urine cultures are stratified to specificallyrepresent the patient being treated, they are mostuseful. These stratifications can reflect inpatientor outpatient status, and patient age and gender.For example, it would be clear that cefazolin is anexcellent empirical antibiotic for females youngerthan 2 years since the most common bacteria in thatage and gender group is E. coli, which is sensitive tocefazolin 96% of the time. In a male patient youngerthan 2 years cefazolin would be less likely to beeffective because it only treats 69% of UTIs (E. coliand P. mirabilis). In this case one may considerusing a different antibiotic, such as amoxicillin-clavulanate, while waiting for cultures to return.

Limitations

This study should be interpreted in the context ofsome database limitations. It was not possible todetermine whether urinary isolates collected in theinpatient setting actually represented infectionsacquired in the inpatient or outpatient setting. Ifa patient was admitted to the hospital, the urineculture obtained on hospital day 0 was counted asan inpatient culture. Clearly, the timing of thatculture should be considered outpatient but in ourdata set it was counted as an inpatient culture.While our data prevented us from determining thiswith certainty, patients from whom urine cultureswere obtained during inpatient admission tendedto have different resistance patterns than thosetreated in the outpatient setting.

There is no information in the TSN data set onassociated clinical signs or symptoms that wouldsupport a UTI diagnosis. Thus, it was not possibleto determine whether some positive cultures wererelated to asymptomatic bacteriuria.

Another limitation of the data set was theinability to identify the specimen source, ie cleancatch vs catheterized specimen. Despite this limi-tation it is unlikely that these cultures representedcontamination. Investigators at participating labo-ratories are instructed to submit only culture datathat they deem clinically positive and we only chosespecimens with single growth bacteria to minimizethe concern for contamination.

Lastly, these aggregate data do not necessarilyreflect resistance patterns in specific communitiessince they were not separated regionally. However,this information is essential to evaluate overalluropathogen resistance patterns and trends in theUnited States.

CONCLUSIONSWe found significant differences in resistance pat-terns between outpatient and inpatient urinaryisolates for several of the most commonly prescribed

RESISTANCE PATTERNS OF PEDIATRIC URINARY TRACT INFECTION 1613

antibiotic agents. We also noted that uropathogenprevalence varies by patient gender and visitsetting. These findings substantiate the need for

separate inpatient and outpatient based antibio-grams to optimize empirical antibiotic selection forpediatric UTI treatment.

REFERENCES

1. Tseng MH, Lo WT, Lin WJ et al: Changing trendin antimicrobial resistance of pediatric uropatho-gens in Taiwan. Pediatr Int 2008; 50: 797.

2. Prabhu A, Taylor P, Konecny P et al: Pyelone-phritis: what are the present day causativeorganisms and antibiotic susceptibilities?Nephrology (Carlton) 2013; 18: 463.

3. Mohammad-Jafari H, Saffar MJ, Nemate I et al:Increasing antibiotic resistance among uropath-ogens isolated during years 2006-2009: impacton the empirical management. Int Braz J Urol2012; 38: 25.

4. Gaspari RJ, Dickson E, Karlowsky J et al:Multidrug resistance in pediatric urinary tractinfections. Microb Drug Resist 2006; 12: 126.

5. Dahle KW, Korgenski EK, Hersh AL et al: Clinicalvalue of an ambulatory-based antibiogram for

uropathogens in children. J Pediatric Infect DisSoc 2012; 1: 333.

6. Lutter SA, Currie ML, Mitz LB et al: Antibioticresistance patterns in children hospitalized forurinary tract infections. Arch Pediatr AdolescMed 2005; 159: 924.

7. Edlin RS, Shapiro DJ, Hersh AL et al: Antibioticresistance patterns in outpatient pediatric uri-nary tract infections. J Urol 2013; 190: 222.

8. Performance Standards for Antimicrobial Suscep-tibility Testing; Twenty-Third Informational Sup-plement, M100S20. Wayne, Pennsylvania: Clinicaland Laboratory Standards Institute� 2013.

9. Boggan JC, Navar-Boggan AM and Jhaveri R:Pediatric-specific antimicrobial susceptibilitydata and empiric antibiotic selection. Pediatrics2012; 130: e615.

10. Halstead DC, Gomez N and McCarter YS: Realityof developing a community-wide antibiogram.J Clin Microbiol 2004; 42: 1.

11. Hindler JF and Stelling J: Analysis and presen-tation of cumulative antibiograms: a newconsensus guideline from the clinical and labo-ratory standards institute. Clin Infect Dis 2007;44: 867.

12. McGregor JC, Dumyati G, Casiano-Col�on AEet al: Usefulness of antibiogram surveillance formethicillin-resistant staphylococcus aureus inoutpatient pediatric populations. Diagn MicrobiolInfect Dis 2009; 64: 70.

13. Copp HL, Shapiro DJ and Hersh AL: Nationalambulatory antibiotic prescribing patterns forpediatric urinary tract infection, 1998-2007.Pediatrics 2011; 127: 1027.