Conflicts of Interest of Conflicts article todisclose. Lake City, UT; General Hospital, Columbia, MD; School of Nursing, UniversitySchool ofNursing, of

Denver, CO; Michigan School of Medicine, AnnArbor,MI; Michigan School ofMedicine, of andUniversity Center Medical Michigan of University Pediatrics, and Medicine Emergency This article is protected Allrights bycopyright. reserved. “Ac cite thisarticle asan Medicine Emergency Academic the in forpublication approved and hasbeenpeer-reviewed that Article anAccepted is This Financial disclosure statement MCHB/HRSA. EMSC program ofthe U03MC00006, U03MC00007,U03MC00008, (PECARN) issupportedby cooperativ Injury 1R49CE00100201.ThePediatric Prevention Grant Support MedicineNational Meeting,Dallas, May2014. TX, Meeting Presentation 916-456-2235,Ema 4545, Fax: Davis, 2516Stoc University California, of to: correspondence Address 12 8 1 Article Contribution type :Original Date:14-Dec-2015 Accepted Revised Date:02-Dec-2015 :07-Sep-2015 Date Received Cincinnati, OH; Medicine, New York University School YorkUniversity Medicine, New Department of Pediatrics, Universi Department ofPediatrics, University Department ofPediatrics, Michael Tunik, M.D. Tunik, Michael M.P.H. Department of Emergency Medicine, Universi Department ofEmergency Medicine,

Accepted ArticleM.D. Relationship of Physician-Identified Patient Race and Ethnicity to Use of Computed Computed toUseof Ethnicity Raceand Patient Physician-Identified of Relationship JoAnne E. Natale, M.D, Ph.D. E.Natale,M.D, JoAnne M.D, M.P.H. 7, 8 10 , Lawrence J. Cook,M.Stat.,Ph.D. , Lawrence J. , Kimberly Quayle, M.D. 11 St. LouisChildren’s Hospital, Washington Louis, University, St. MO; 10 : This work was supported by a grant fro a : Thisworkwassupportedby 7 Department ofPediatrics, Children’s Nationwide Hospital, Columbus,OH; 12 Department ofPediatrics Department for the Pediatric Emergency Care Applied Research Network (PECARN) (PECARN) Network Research Applied Care Emergency Pediatric forthe : This work was presented in part at the Society for Academic Emergency : The authors noconflicts have : The ofinterest relevant tothisarticle todisclose. 4 , DavidMonroe,M.D. Tomography inPediatri cepted Article”; doi: 10.1111/acem.12943 doi:10.1111/acem.12943 cepted Article”; JoAnne E. Natale, MD, PhD,F E.Natale, JoAnne il: [email protected] il: [email protected] , but has yet to undergo copy-editing and proof correction. Please yet correction. and proof , buthas toundergo copy-editing : The authors have no financial re nofinancial authorshave : The California, Davis, Sacramento,CA; Davis, California, 1 6 , Jill G. Joseph, M.D,, Ph.D. G.Joseph, Jill Department of Pediatrics,Ci Department of ty of California, SanDiego,ty CA; ofCalifornia, 11 , Nathan Kuppermann, M.D,M.P.H. , NathanKuppermann, of California, Davis, Sacramento, CA; CA; Sacramento, Davis, of California, e agreements U03MC00001,U03MC00003, agreements e of Medicine, New York City, NY; NewYork ofMedicine, kton Blvd, Sacramento, CA 95817, Phone:916-734- kton Blvd, Sacramento,CA95817, and Emergency Medicine, University of Colorado, of University Medicine, andEmergency 9 5 , KentPage, M.Stat. , Benjamin Kerrey, M.D., M.S. Kerrey,, Benjamin M.D., U03MC22684, andU03MC22685fromthe 4 ty of California, Davis, Sacramento, CA CA ofCalifornia,Davis,Sacramento, ty Departments of Pediatrics and Emergency andEmergency ofPediatrics Departments

c Blunt Torso Trauma Emergency Care Applied Research Network Research Applied Care Emergency m the Centers for Disease Controland for m theCenters AAP, Department ofPediatrics, AAP, Department 2 , Alexander J. Rogers, M.D. J. , Alexander ncinnati Children’s Hospital, lationships relevant tothis 9 , Kathleen Adelgais,M.D., , Kathleen 9 UniversityUtah, Salt of 3 Departments of Departments 1,12 5 Howard County , F. Holmes, James 2 6 Betty Irene Moore Irene Betty Moore , BemaBonsu, K. 3 ,

from 2 large urban pediatric EDs revealed less fre less EDsrevealed pediatric urban 2large from For (ED). including department intheemergency are differ management ofclinical aspects that This article is protected Allrights bycopyright. reserved. services medical There iscontinuing evidenceand ethnic ofracial disparities inchildren’s access toand use of Background Introduction affect provider choicesregarding emerge studies Further use ofabdominalCTinchildren. Hispanic. Thissuggeststhatnon-clinicalfactor abdominal CT toreceive lesslikely Hispanic were pediat After blunttorsotrauma, Conclusions (oddsratio in whitenon-Hispanicchildren an ofreceiving Hispanic children,thelikelihood forblacknon-Hi interval0.7,0.9) confidence an ofreceivinghospital clustering, thelikelihood abdominal ultrasounduse,riskforintra-abdomin abdominal children underwent 44.2% ofHispanic 51.8%ofwhitenon- Overall, (n=1,291, 11.9%). blacknon-Hispanic Hispanic (n=5,847,54.0%), physicia patients, treating Among 12,044enrolled Results clusteringwithin hospitals. children of estimatin generalized using multivariable analyses abdominal injury intervention acute undergoing we rule, prediction clinical previously-derived documented aswh race/ethnicity departments participa yearswithin 24hoursofblunt 18 oldpresenting We performed a planned secondary analysis of Methods trauma. abdominal withreceiving associated independently To Objective Abstract

Accepted Article determine whether a child’s race or ethnicity as determined by the treating physician is is physician treating the by determined as ethnicity raceor child’s a whether determine

1,2 , as well as in selected health outcomes health selected asin , aswell ting in a pediatric research network, 2007-2010. Treating physicians physicians 2007-2010.Treating network, ting inapediatricresearch ric patients by identified treating the as physicians black non- ite non-Hispanic, black non-Hispanic, or Hispanic. Using a Using Hispanic. black non-Hispanic,or ite non-Hispanic, ncy department radiographic imaging. radiographicimaging. ncy department 0.9, 95% confidence interval 0.8,1.1). interval confidence 0.9, 95% spanic than for white non-Hispanic children. For children. than forwhitespanic non-Hispanic entially provided to children of different races, races, different of entially providedtochildren classified each child’s risk for having anintra- having for risk child’s each classified s influence clinician decision-making regarding clinician decision-making s influence a prospective observationa a prospective Hispanic, 32.7% of black non-Hispanic, and 32.7%ofblacknon-Hispanic, Hispanic, to define injury severity. We performed We defineinjury severity. to abdominal CT did not differ from that observed CTdidnotdifferfromthatobserved abdominal (n=3,687, 34.1%), or Hispanic of any race race Hispanic ofany 34.1%),or (n=3,687, should focus on explaining how patient race can can race howpatient shouldfocusonexplaining torso trauma to 20 North American emergency emergency American torso traumato20North al injury undergoing acute intervention,and injuryacute al undergoing abdominal CT was lower (oddsratio0.8,95% abdominal CTwaslower quent use of laboratory and radiologic and testing quent useoflaboratory ns documented race/ethnicity aswhitenon- ns documentedrace/ethnicity example, a recent report based onclaimsdata report arecent example, g equations to control for confounding and for andfor confounding g tocontrolfor equations imaging than those identified aswhitenon- imaging thanthoseidentified CT imaging.age, adjusting sex, After for computed tomography (CT) afterblunttorso computedtomography(CT) 3 . Perhaps most concerning, it appears appears it concerning, most . Perhaps l cohort of children < cohort ofchildren l whom cliniciansidentifiedasbe minorbluntheadtr with evaluation ofchildren Memb behavior inthiscontext. whet onunderstanding group hasfocused trauma. Our Clinical managementintheemerge presenting withacute painabdominal fluids, administrationofnarc intheuseofCTs onrace based differences EDrevealed pediatric asingle from data record medical ofelectronic analysis Similarly, Injury isthe leading cause ofbothpediatric and morbidity mortality trauma-associated mortality isinfrequentbutove mortality trauma-associated to receive computed tomography (CT) as part of their ED evaluation for abdominal pain forabdominal EDevaluation oftheir as part (CT) tomography computed receive to suggesAmbulatory Survey Care (NHAMCS) Medical ethnicity as identifiedtreating by physicians. evaluation ofblunttorsotraumainchildrenis study todeterm was current The objectiveofthe This article is protected Allrights bycopyright. reserved. Accepted Articlegroup observed inabiracial towhitenon-Hi compared in blackchildrenwhen 4 . Another study analyzing data HospitalNational fromthe otic analgesia, hospital admission, and surgery forchildren admission, andsurgery hospital otic analgesia, ers of our group previously reported that cranial CTscansforthe thatcranial reported group previously ers ofour ing other than white non-Hispanic otherthanwhite ing ncy department isparticularly ncy department 6 . canning, ultrasonography, ultrasonography, canning, independently associated auma were less frequentl wereless auma ine if the use of abdominal CTscansforthe abdominal iftheuseof ine rall higher Hispanicpatients rall and inbothblack spanic children, an effect thatwasalso an effect spanic children, her race and ethnicity influence clinician influence andethnicity her race ted that blackted that as children were likely half crucial in response to pediatric inresponsetopediatric crucial 9 . 7 . Among pediatric patients, pediatric . Among provision ofintravenous provision y orderedamong patients with patient race and with patientrace 5 .

8 . previous abdominalCT or di tobe wereknown examination, ifthey apreexisting excluded had patients were than24hoursbefore greater the injury occurred injury,chest or and intra-abdominal to screenfor assessmentwithsonog ultrasound including focused thefollowi abdominal traumaresulting of in any physical assaulti more, crushinjury totorso, or 20feet fall crash, vehicle motor orrollover ejection high-speed, mechanisms: injury following fractures bone ormultiplenonadjacentlong paralysis withblunttorsotr consciousness inassociation of level decreased 1) following: the one of any included criteria inclusion study, parent For the Study Setting and Population fully elsewhere atallbut1site. waived informed consentwas primary studywas reviewed and Institutional approved by the ReviewBoard at each site and This article is protected Allrights bycopyright. reserved. Accepted Article 2010 Mayand January 2007 between sites (PECARN) Network Research Applied Care Emergency 20Pediatric EDsat the in yearsage withbl younger of 18 children than We performed a planned secondary analysis of Study Design Methods 10 , while specific methods relevant to this , whilespecific methodsrelevant 10 agnostic peritoneal lavage. agnostic lavage. peritoneal . Participating hospitals are listed inthe Acknowledgements. The pregnant, or were transferre were pregnant, or nvolving the abdomen; or 4) physician concern for concern or4)physician nvolving theabdomen; unt torso trauma. The primary study was conducted study was unt torsotrauma.Theprimary The methods of the primary study are described described are The methodsoftheprimary study auma; 2)blunttraumatic event witheither a large prospective observational cohortstudy of observational prospective a large neurologic disorder re disorder neurologic presentation or was pene orwas presentation ng diagnostic tests: abdominal CT,abdominal diagnosticng tests:abdominal pelvic radiography. Pa radiography. pelvic raphy for trauma (FAST), laboratory testing (FAST), fortrauma raphy ; 3) blunt torso trauma due to any ofthe duetoany ; 3)blunttorsotrauma secondary analysis are described below. below. aredescribed analysis secondary d from another hospitalwith another d from stricting dependable stricting dependable tients were excluded if wereexcluded tients trating. Additionally, (n=439). size sample insufficient dueto Islander) Pacific (such as categories main three these than different This article is protected Allrights bycopyright. reserved. publications w from theED outcomes ofchildrendischarged Additional informationonthestudy population, injury and2)anintra-abdominal identifiableonCT, theirclinicalsuspic recorded physician the treating scanatthe discretion was treating of physician. (ifperfo aCTscan beforeobtaining examination The treating physiciancompleted and documented Measures AcceptedInaddition, fromtheanalysis. excluded were andwith unknownrace as identified by physician blacknon-H main categories: whitenon-Hispanic, asnon-Hispanic treated Hispanic orunknownwere ethnicity were and race here, analysis reported Island Pacific American, Asian, black/African orunknown) (Hispanic/Latino, non-Hispanic/Latino Inaddition, the treating recorded physician theandethnicity patient’s determined Article 10,11 . we excluded patientswith we excluded er, white, unknown, or other). In the secondary In thesecondary white, unknown,orother). er, categorized asfollows.Ethnicitieslistednon- categorized Additionally, whether a CT wasobtainedornot, a CT whether Additionally, methods, interobserver agreement, and agreement, methods,interobserver ithout CT scans are reported inotherstudy reported ithout CTscansare rmed). The decision toobtainanabdominalCT Thedecision rmed). ethnicity as non-Hispanic or unknown (n=780) orunknown(n=780) ethnicity as non-Hispanic ispanic, and Hispanic of any race. Children Children race. any of andHispanic ispanic, ion for: 1) the presence of an intra-abdominal of anintra-abdominal ion for:1)thepresence a structured patient history and physical historyand physical a structuredpatient . Race and ethnicity were dividedintothree were and ethnicity . Race injury that would require acute intervention. acute injury thatwouldrequire and race (American Indian/Alaskan Native, (American andrace race and ethnicity andethnicity race results of the model. Applying results ofthemodel.Applying potential outliers, and all analyses revealed noobservations potentially negatively influencing the from the PECARN prediction rule as well as important demographic and clinical variables and clinical demographic asimportant aswell rule prediction PECARN the from basedonthePECARN undergoing acute intervention abdominalultrasound usein age,controlling sex, for andethnicity, andrace rates CT abdominal between associations the estimate to modeling groups. ethnicity and We comparison oftherace used We injuryacute intervention. undergoing acuteinterven intra-abdominal injury undergoing should be noted that all covariates are categori are should benotedthatallcovariates range (IQR) (25 (IQR) range conti ,while appropriate where dataus describedcategorical We Data Analysis high-risk we levels, these combined riskcategories forthisanalysis. intervention. Because preliminary analyses demonstrated similarin moderate- outcomesand ch each classification toclassify weused here, reported analysis the secondary of the three race and ethnicity groups se ethnicity age, by and race of thethree rate calculated We wasobtainedintheED. CT This article is protected Allrights bycopyright. reserved. Accepted Articleacute identified intervention moderate, fourrisklevels:low (Table high, low,and 1) very PECARNclinicalpredic The previously-derived th – 75 th percentile). We defined the outcomeof percentile). definedthe We standard covariate rules, our rules, covariate standard nuous data were described using using described nuous datawere ild’s risk for an intra-abdominal injury undergoing an acute an acute injury undergoing ild’s riskforanintra-abdominal ing counts, percentage this clinicalpredicti cal. We testedforinfl We cal. s of children undergoing abdominal CT foreach CT abdominal childrenundergoing s of a chi-square test of independence for each each for testofindependence a chi-square x, Glasgow ComaScale(G Glasgow x, employed standard multivariable logistic employed multivariablelogistic standard tion rules for intra-abdominal injury undergoing rulesforintra-abdominal tion tion, and clinical suspicion for intra-abdominal andclinicalsuspicionforintra-abdominal tion, ED, and risk for an intra-abdominal injury ED,andriskfor prediction rule. We selected covariates covariates selected predictionrule.We s and 95 percent confidence intervals (CI) intervals confidence s and95percent model had 670 CT scansobtained model had670 interest as whether an abdominal whether an as interest the median and the median on rule 4-level risk on rule4-level uential data pointsand uential CS) score, risk for an CS) score,riskfor interquartile 10 10 . In . It (Table 2). (Table 2). whitenon-Hisp for 32.7% and44.2%respectively 4,803(44.4%)patientsreceive (Table 2).Overall, 11 age of hadamedian children 11.9%). Evaluated non-Hispanic (n=3,687, (n=5,847, 54.0%),black ( 10,825 physician, designation thetreating by 12,044(80.9%)of enrolled The primary study of Characteristics Results ofthestudy. reporting Institute,9.3 (SAS North Carolina).The Cary, abdominal injury intervention. acute undergoing forintra- forrisk model the of part is becauseit model the in included wasnot GCS intervention. withrisk associated the modelasitwasstrongly injury clinical suspicionforintra-abdominal Physician models. other for all QICu the than lower QICu was its because intervention acute useinED,andri abdominalultrasound age, sex, and ethnicity, race with model the from results the present We interactions. order first including GEEmodels all of QICu the Then wecompared This article is protected Allrights bycopyright. reserved. explored per variable Accepted Article Independence Quasilikelihood underthe clusteringwithin hospitals. children Toselect of the best fitmodel, we calculated the

study subjects 12 . We used generalized estimating equations (GEE) to account for the to accountfor the . estimating (GEE) usedgeneralized equations We model Criterion(Q undergoing acute intervention was notincludedin intervention undergoing acute 89.9%) had race/ethnicity of white non-Hispanic of hadrace/ethnicity 89.9%) 14,882 eligible patients. Among these, per eligible these, patients.Among 14,882 funding agencies had no role intheconductor role hadno agencies funding for an intra-abdominal injury undergoing acute acute injury anintra-abdominal undergoing for sk for an intra-abdominal injury undergoing foranintra-abdominal sk containing all combinations of these variables ofthesevariables containing allcombinations Data analysis was performed withSASversion performed was Data analysis 34.1%), or Hispanic of any race (n=1,291, any race 34.1%), orHispanicof d an abdominal CT scan with ratesof51.8%, scan d anabdominalCT anic, black non-Hispanic, and Hispanic groups andHispanic anic, blacknon-Hispanic, .3 years (IQR 6.0, 15.1) and 61.5% were male years and61.5%were 6.0,15.1) .3 (IQR ICu) statistic 13,14 for each model. foreachmodel. abdominal CTinchildrenwithblunttorsotrauma of analysis secondary conductedaplanned We Discussion that observedinwhitenon-Hispanic an abdominal of receiving children. Thelikelihood 0.8, (odds ratio[OR] physicians lesslikely toreceive were black non-Hispanic Multivariable analyses (Table 4)revealed that 5%. than greater intervention withclinicalsuspic GCS scoresof14andthose children to be evaluated with abdominal CTscan children tobeevaluated thatwhitenon- revealed and adjustedanalyses This article is protected Allrights bycopyright. reserved. Accepted Articlechildren tore black non-Hispanic gr white,non-Hispanic categories. Similartothe st twogroupswere these comparisons between abdominal toreceive likely generally more were children(Tab CT scansthanblacknon-Hispanic ph by thetreating identified aswhitenon-Hispanic in suspicion forintra-abdominal clinical with scoresof14andthose GCS with presenting ofchildren exception the With results Main 95% confidence interval [CI] 0. [CI] 95%confidenceinterval jury undergoing acute intervention greater than10%,children greater juryacute intervention undergoing ceive abdominal CTscans,excep ceive children (OR 0.9, 95% CI 0.8, 1.1). 0.9, 95%CI 0.8,1.1). (OR children Hispanic children were more likely more thanblack Hispanic childrenwere children identifiedchildren by the treating physicianas abdominal CT scans ordered by those same by thosesame CTscansordered abdominal atistically significant in fewer comparison atistically infewer significant data from 20 hospitals describing ED use of from20hospitalsdescribing ED useof data ion for intra-abdominal injury undergoing acute injury acute ion forintra-abdominal undergoing le 3). Similarly, white le CT scans than Hispanic children, although CTscansthanHispanicchildren, oup, Hispanic children were more likely than morelikely Hispanicchildrenwere oup, s. This work extends and complements research andcomplementsresearch s. Thisworkextends 10 ysician were more likely to receive abdominal to receive morelikely ysician were CT in Hispanic children didnotdifferfrom CTinHispanicchildren . With a large sample size, bothbivariable . alargeWith samplesize, 7, 0.9) than whitenon-Hispanic 7, 0.9)than t for children presenting with presenting t forchildren non-Hispanic children non-Hispanic in both black and Hispanic patients in bothblackandHispanic infrequent, butmultivariable analyses the TraumaBank National of Data reveal that itishigher remains animportantpriority care ofpediatric provision in equity onmortality, care oftrauma effects the of care. Regardless for coverage toprovideinsurance state efforts differences inmanagement differences of minority trauma patients at trauma centers with worse than expected mortality expected worsethan with centers trauma at patients trauma of minority related to race/ethnicity in an analysis of administrative data from California from data ofadministrative ananalysis in race/ethnicity to related This article is protected Allrights bycopyright. reserved. patients ofrace/ethnicity toth relationship focused onthe whetherrace/ethnicity, patient as treating determined by the linked physician, toclinical was me collection protocol-driven data id clearly information, collected prospectively likelihood ofmortality in the influence to appear ethnicity and Race betroubling. to continue diagnostic interventions managementrequires meticulous inthe ED and objective evaluation, racial/ethnic disparities in Because injury isthe leading cause ofpediatric morbidityand mortality itsearly and a foridentifying responsible study alsothose were c Importantly, CTuse. regarding decision-making Accepted patientsw CT evaluationinpediatric Article 3,5,6,15 by focusing on the more urgent problem of abdominal injury, and useof by ofabdominalinjury, onthemoreurgent problem focusing adulttraumapatients 20-22 24 . Among pediatric patients, trauma . Among pediatric , and data reported here raise co hereraise reported , anddata thods. The purpose oftheanalyses thods. Thepurpose 8 . In. contrast tonational data, trauma mortality was not ho have experienced torso trauma. torsotrauma. experienced ho have entified inclusion and exclusion criteria,and entified inclusionandexclusion children and an emphasis on culturally competent competent culturally on emphasis andan children 16-18 e ED evaluation of abdomin EDevaluationof e linicians responsible for linicians responsiblefor nd recording the race/ethnicity of each patient. patient. ofeach race/ethnicity the nd recording , perhaps partially explained by the clustering clustering by the explained , perhapspartially ncerns regarding equitable use of equitableuse regarding ncerns -associated mortality-associated overall is reported here was toassess here reported ordering CTsinthis ordering 23 al paininpediatric , perhapsdueto 19 and/or history examination andphysical anacuteinterventi toundergo serious enough modalities. In reporte the analyses useofimaging differential warranting thereby and ethnicity, race by varies severity injury CT inpediatric trauma Each patients. requires careful consideration.it ispossible First, that despite similar demographic characteristics and illness severity testing receive to patients minority than likely more were patients pediatric white where pain consistent with several ofracial disparities reports inthe evaluation complaints ofabdominal There are a variety of possible ex possible of a variety are There management regarding decisions imaging. to corresponded that thesepatientdescriptors and r were determined study part ofthecurrent toracialandethnic methods forassigning patients Whether positive ornegative, investigations disparities ofpotential careful require attention to This article is protected Allrights bycopyright. reserved. te radiological andlaboratory Our studyresults aregenerally consistent withtheobservation ofracial disparities inordering Accepted practices diagnostic in differences true are there findings fromsuch unlikely arise thatthereported estimating equati generalized using differences and ethnicityalsobetween varies used institutions.We multivariable analyses adjusted forsite inpatientma differences in differences forobserved unlikely toaccount Article nagement, creating apparent racialdi apparent nagement, creating sts for pediatric ED patients ED pediatric sts for 10 planations fortheobserveddisp d here, we validly assessed risk assessed validly we d here, . Therefore racial and ethnic differences in injury severity are severity injury in differences andethnic racial . Therefore on as documented in specific elements ofthe elements documentedinspecific on as the impressions ofthosemaking clinical the impressions ecorded by treating physicians, therebyassuring ons to account for such clustering, makingit ons toaccountforsuch CT use. Second, there may beinter-institutional may use.Second,there CT for pediatric patients presenting withblunttorso pediatricpatients for confounding. Finally, it is conceivable that Finally, itisconceivablethat confounding. categories. Racial and ethnic data analyzed as ethnic dataanalyzed and categories. Racial 4 , including those presenting with chest chest with presenting those , including sparities ifthedistributionofrace sparities 25 . Our findings are also are . Ourfindings arity in the use ofabdominal arity inthe of an intra-abdominalinjury of 5,15 . cranial trauma CTforminorhead minor blunt head trauma, parental minor bluntheadtrauma, inou overuse isnotclear, Although for thereason torsotrauma. commonly childrenwithblunt inwhitenon-Hispanic abdominal CTscanning,thisfinding supportsth atsuch Aschildren non-Hispanic whitechildren. c PECARN predictionruleorby intervention intra-abdominal injury undergoing request for CT scanning after blunttorsotrauma request forCTscanning ED selected that demonstrate others gr racial/ethnic inother rather thanunderuse theresultsreportedhere that It isconceivable injuries. greater im assume factors as clinical overcome apparently are disparity detected the to are leading influences Thus,whatever longer observed. is provocative tonote thatat the highest ofclinical levels riskthe racial disparity inCTuse isno children exposed to ionizing radiation fromCTimaging toionizing radiation children exposed children. The latter isparticularly as intriguing there are documented risksoffuture cancer in This article is protected Allrights bycopyright. reserved. such as implicit bias theinfluenceof Iftrauma. thisisthecase, Accepted commonlyrisk patientsandwasmore Article 26 , clearlyIn appraisal. requires careful further evaluating thispossibility, it linician suspicion,abdominalCT 9 request was an important indicat an important was request , may actually reflect the overuse in white non-Hispanic inwhite theoveruse reflect , mayactually interventions, notably hospitaladmissions interventions, cited inwhitenon- non-clinical factors on physician decision-making, decision-making, onphysician factors non-clinical oups. Previous work by Previous workby oups. reflect overuse in white non-Hispanic children inwhitenon-Hispanic overuse reflect portance in those patients with moreserious inthosepatientswith portance was defined using risk stratification by the riskstratification definedusing was was the documented reason for CT scanning in forCTscanning reason was the documented e concept that abdominal CT isoverusedmore concept thatabdominal e a low risk are highly unlikely to benefit from tobenefit highly unlikely a lowriskare r prior study on cranial CTuseinchildrenwith cranial r priorstudy on 28-31 Hispanic children . Regardless of whether risk for . whetherriskfor Regardless of rates were highest inthewhite highest rates were or for CT use in the lowest or forCT members of our group and ofour members 9 . Here, parental . Here,parental 27 anduseof This article is protected Allrights bycopyright. reserved. Accepted thisstudychoosing in toparticipate may notbe evaluation of pediatric emergency traumaareNonetheless, care.awarethat we institutions yearsover20 participating nearly 3 from ins including CTs 10, and abdominal race/ethnicity of Our analyses patients. trauma ofpediatric evaluation emergency the in patterns practice this study.Further, also itcan be asked disparities observedare widely whether reflective here of this relatively of small proportion However, analyzed. were whosedata children from different systematically ways in been treated ethnicity notsufficiently complete.Thiscr was excl were children eligible 6.5% ofotherwise betw differences andreducing conservative bias a creating thereby misclassification, some in resulted may have asnon-Hispanic ethnicity patients.Thefa trauma for caring physicians Article andinsuran andethnicity, impression ofrace fo the However, covariates. independent ofthese relations describethe unableto are we therefore importantly, statusnorpare neitherinsurance Most beconsidered. needto that limitations hascertain study current the that recognize We Limitations definitely influencingfactors identify the neitherthis However, abdominal CTscanning. only twopatients, suggesting that parental played request little role inthe decision to obtain patientswouldbeunlikely to decision for CT evaluation in each patient. for CTevaluationineach decision titutions provides a and broad detailed sampling ct that we classified children with unknown children with weclassified ct that ce status isnotcommonly emergency byce status known ntal socioeconomic status was documented; socioeconomicstatuswas ntal uded because information regarding race and/or and/or race regarding uded becauseinformation eates a potential bias should these children have apotentialbiasshouldthesechildrenhave eates nor the head trauma study was designed to designed was study theheadtrauma nor generally representative of all EDs evaluating all EDsevaluating of representative generally 825 pediatric patients obtained prospectively prospectively 825 pediatricpatientsobtained hip of race and ethnicity to abdominal CTuse ethnicity to and ofrace hip een Hispanics and non-Hispanics. In addition, Hispanicsandnon-Hispanics. een cus of our analysis was based on the physician on thephysician wasbased ofouranalysis cus change the majorconclusionson the change

(https://implicit.harvard.edu/implicit) biases that conscious orunconscious This article is protected Allrights bycopyright. reserved. Accepted al. et Chang Individualequitable all children. care clinic for effective toovercoming approaches such concerning disparities highest-quality toensure the and regardin providerdecision-making can affect shouldfocusonexplaining studies children. Further clinician d factors may influence child abdominal CTimaging thanwhitenon-Hispanic In blunttorsotrauma,non-Hispan after summary, Conclusions Article PECARN predictionrulemodel,however,pr report.basis Substituting and forouranalyses clinical derived suspicion forthe fromthe risk asthe risk perceived than rather risk calculated the used andtherefore algorithms stratification collinearity have We variables. inthese interest a primary inthe empirically value of derived risk PECARN clinical rule prediction rather onclinician based than suspicion, due tothe substantial intra-abdominal inju riskfor evaluated patient the assumed quality emphasis on and protocolized care inparticipating institutions.Finally, we estimated the inflate than rather diminish would self-selection such by introduced any bias that likely is it However, patients. trauma pediatric 32 and completeassessments free on-line ofimplicit social cognition leading to ecision-making regarding use of useof regarding ecision-making may influence their practice patterns 33 . g ED radiographic imaging and ondeveloping and radiographicimaging ED g ovides similar results (data not shown). ovides similarresults(datanot ry undergoing acute intervention basedonthe acuteintervention ry undergoing ians may consider the recent recommendations of may therecent consider ians association between race and CTusebasedon and betweenrace association ic black children are less likely to receive toreceive lesslikely are ic blackchildren how factors such as patient race or ethnicity orethnicity race aspatient factors such how ren. Thisfinding thatnon-clinical suggests abdominal CT after trauma in abdominal CTaftertrauma This article is protected Allrights bycopyright. reserved. Accepted Hoyle, Suhajda, N.Schamban J. J. R.Ruddy, Burr, K.Call, Atabaki, J. Safetyand Regulatory Subcommittee Affairs Schalick, (SRAS):W. J. Callahan, Co-Chairs; S. Lillis,K. R.Lichenstein, B. R.Holubkov,E.Jacobs, Millar, R.Ruddy,Gerardi, M. Shults M. Enriquez, R. Ehrlich, P. Chair; Stanley, R. (QAS): Subcommittee Assurance Quality A.Walker K. Shaw, M.Dean, P.Dayan, J. Chamberlain, A.Cooper, S Protocol ConceptReviewandDevelopment Teach Moler, S. L.Donaldson, G.Foltin,F. Brown, Cimpello,A. Grants and Publications Subcommittee (GAPS):M.Gorelick, D.Borgialli, Chair; E.Alpern, K. S.Zuspan H.Rincon,M.Tunik, E. Kim,D.Monroe,G.O’Gara, Subcommittee(FABS):Feasibility T. and Budget M.Fjelstad A.Jones, Wade, M.Miskin, J. A. Donaldson,S.Zuspan, Central DataManagement Walker. MCHB/EMSC liaisons: D.Kavanaugh, H. Park. Lichenstein,Lillis, K. P.Mahajan, R.Ruddy, R.Maio, D.Monroe, R.Stanley, M.Tunik,A. M. Dean,M.Gerardi, Chamberlain, P.Dayan, J. Chair; PECARN Steering Committee:N. Kuppermann, this study initiated. was ofthefollowing indi efforts the acknowledge We Article (K. Quayle); and Women Children’sLillis). Hospital ofBuffalo (K. Children’s Medical Washington Center (K.Adelgais); Louis University/St. Children’s Hospital Utah/Primary of University Garcia); (M. ofRochester University (A. Rogers); Michigan Holmes);Universi J. Center (N.Kuppermann, (K. Yen); University Nationwide Children'sBonsu); Hospital California (B. Davis of Medical Medical Center(S.Blumberg); Medical College Wisconsin/Children’s of Hospital ofWisconsin Hu Center(D.Monroe); Medical Howard County Children’s Hospital Kooistra); Medical (J. Children’s Center Hospital DeVos Kerrey); (B. Hospital ofPhiladelphia (F. Nadel); Children’s National Medical Atabaki); Center(S. Cincinnati Michigan(P. Mahajan);Children’s Hospital of Hospital Center Children’s Hospital Boston(L. (M.Tunik); Lee); Hospital Children’s of Participating centerssiteand investigators ar children inthisstudy. possible;andall study wouldnothavebeen inPECARN coordinators thanktheresearch We Acknowledgements

andCoordinating Cent the clinicians around the PECARN whoenrolled the cliniciansaroundPECARN e listedbelow inalphabe ubcommittee (PCRADS): D. Jaffe, Chair; J. J. Chair; D.Jaffe, (PCRADS): ubcommittee ty of Maryland (J. Menaker); University of University of Menaker); Marylandty (J. of New York – Presbyterian (M. Kwok);Children’s (M. York–Presbyterian New R. Holubkov,P.Mahajan, , without whose dedication and hard work this andhardwork , withoutwhosededication M. Gorelick, J. Hoyle, E. Jacobs, D. Jaffe, R. D.Jaffe, Jacobs, Hoyle, E. M. Gorelick,J. viduals participating in PECARN atthetime viduals participating inPECARN rley Medical Center (D. Borgialli); Jacobi Borgialli); Jacobi (D. Center rley Medical er (DCC): M. Dean, R. Holubkov, S.Knight, R.Holubkov, er (DCC):M.Dean, Singh, Chair; A. Drongowski, L. Singh,Fukushima, Chair;A.Drongowski, E. Alpern, D. Borgialli, J. Callahan, J. Callahan,J. Borgialli, J. E.Alpern,D. tical order:Bellevue tical

R. Maio,N.C.Mann, This article is protected Allrights bycopyright. reserved. R, CaperellK,Pitetti 6. 75. A,Gonzales A, Harries Herring FahimiJ, departments toemergency children presenting 5. PayneNR, Puumala SE.Racial disparities in depa pediatric patientsintheemergency 4. 2004;292:1977-82. in children.JAMA Ponsky K,EichelbergerZJ, Kittle TA,Huang Brody MR,GilbertF, JC, et al.Hospital- appendectomy andnegative rupture appendiceal of risk andthe characteristics and patient-level 3. Localio AR,Fiks PrasadPA, JS, Gerber primar by prescribing inantibiotic differences 2. PineW, J, Flores G,Rabke-Verani of care linguistic issuesintheemergency 1. References 17. Crompton JG, Crompton JG, Pollack KM,Oyetunji T,Chan disparities inmotorcycle-related mortality:an analysis ofthe NationalBank. Data Trauma 17. Newgard RacialMullins RJ. disparities JR, BS, ArthurM,Hedges in CD,Diggs mortality adultshospitalized among 16. department. Pediatri S,Davis Borrero TJ, MD, Weaver Johnson withmanage ethnicity and Association ofrace 15. Hilbe Generalized HardinJM. JW, Estimating New & York:Chapman Equations. Hall/CRC; 2003. 14. criterionin Akaike's PanW. information 2001;57:120-5. 13. KemperE,Holfor P, Concato Peduzzi J, inlogistic re pervariable number ofevents 12. 2013;20:426-32. Lillis N, K,MonroeD, YenK,Kuppermann assessment ofchildre intheclinical agreement 11. 2013;62:107-16 . Lillis Holmes JF, Borgialli K,Monroe D, D,KerreyBT,Identifying Mahajan P,et al. of clinically impor lowrisk children atvery 10. Pediat Arch race/ethnicity. Roge JG, Joseph NataleJE, with children among use tomography computed 9. Lipsitz SR, MearaJG, RosenH,SalehF, USchildren. in traumamortality affects 8. 2009;40:71-4. research ofsafety Journal DellingerAM, Rudd NN, GilchristJ, Borse childhood injuriesintheUnitedStates:keyfindings theCDCchildhoodinjury from report. 7. department. Pediatrics2013;131:1098-106. Accepted Maybury OB, C,Ha RS,Bolorunduro Villegas Pedestriansand race- vehicles insurance-based worsen struck by motor further disparities in 18. 2010;200:191-6. ofsurgery American journal Article cs 2013;132:e851-8. cs 2013;132:e851-8. r Adolesc Med 2012;166:732-7. r AdolescMed Cross KP. Race and acute abdominal Raceandacute CrossKP. rs AJ, MahajanP,CooperA, rs AJ, after injury. Med Care 2008;46:192-9. after injury.Med Care2008;46:192-9. rtment. Pediatr Emerg Care 2013;29:598-606. Emerg Care rtment. Pediatr Journal of pediatric surgery 2009;44:1952-7. 2009;44:1952-7. ofpediatricsurgery Journal children. Pediatr Emerg Care 2002;18:271-84. 2002;18:271-84. PediatrEmerg Care children. gression analysis. J ClinEpidemiol analysis. 1996;49:1373-9. J gression Sabharwal A. The importance of cultural and cultural of The importance A. Sabharwal tant blunt abdominalinjuries.AnnEmerg Med tant blunt y care pediatricians. Pediatrics 2013;131:677-84. Pediatrics y pediatricians. care ment of abdominal pain in the emergency ofabdominal painintheemergency ment n with blunt abdominal trauma. Acad Emerg Med blunt abdominaltrauma.AcadEmerg with n with abdominalpain.Pediatrics2012;130:e1069- d TR, Feinstein AR. A simulation Feinstein studyd TR, ofthe AG, Grundmeier RW, Bell LM, et al. Racial Racial al. LM,et Bell RW, AG, Grundmeier minor blunt head trauma: association with bluntheadtrauma:association minor generalized estimating equations. Biometrics Rogers SO, Jr. Lack of insurance negatively negatively LackRogersSO, Jr. ofinsurance R, Alter H. Computed tomography use among among use tomography H.Computed Alter R, EM, Myaskovsky L, Zuckerbraun EM, Myaskovsky L,NS, Zuckerbraun et al. RA, Ballesteros MF, Sleet DA. Unintentional RA, Ballesteros SleetDA. MF, Borgialli Borgialli D, Kerrey BT, et al. Interobserver orderingradiology tests for laboratory and g DC, Efron DT, Haut ER, et al.Racial EfronDT,HautER,et g DC, ut ER, Stevens K,CornwellEE,3rd,etal. ut Wisner DH, et al. Cranial et al. DH, Wisner pain in a pediatric emergency emergency paininapediatric 2015, at Implicit. 14, November Project 2011.(Accessed https://implicit.harvard.edu/implicit/.) 33. ChengLevy RR. Jenkins TL, MA, Emmanuel Child Health DJ, Disparities: Can What a Pediatrics2015;136:961-8. Clinician Do? 32. Lancetcohort study. 2012;380:499-505. This article is protected Allrights bycopyright. reserved. Accepted Pearce Little MS,Salotti JA, MP,McHughLee K, C,KimKP,et al. Radiation exposure andsubsequentriskof from CTscansinchildhood 31. risk.JAMA estimated cancer S,SolbergRT, Weinmann LI, A,Greenlee et al. Miglioretti E,Williams DL, Johnson pediatri in tomography The useofcomputed 30. linkage 2013;346:f2360. of11millionAustralians.BMJ study etal.Cancer GB, Byrnes GoergenZ, SK, MW, Butler Brady Forsythe AV, MathewsJD, tocomputedtomogr risk in680,000peopleexposed 29. exposure. radiation of source increasing tomography--an Computed EJ. Hall DJ, Brenner Med2007;357:2277-84. N Engl J 28. 24. PatelKM,Poll JG, Joseph ChamberlainJM, Hospitalization Pediatric AreAssociated Rates 27. Nosek Sabin BA, J, GreenwaldFP. A,Rivara Physicians' implicit and explicit attitudes gender and ethnicity, MDrace, by about race 26. KimballTR,Khoury HambrookJT, P,Cnot Article pe of department evaluation 25. CouncilonCommunityAmerican ChildH.PolicyP, Committee onNativestatement-- and children's Pediatrics2010;125:838-49. health equity rights. 24. Ramirez M,ChangBickler DC, Pediatric SW. injury inracial/ethnic outcomes minorities reduce in California:diversity may 23. Shafi S,GentilelloLM. Ethnic disparities ininitial management patients oftrauma ina department nationwide sampleofemergency 22. Millham F, NB. Jain Arethere Surg racial disparities World traumaJ care? in 2009;33:23-33. 21. j American pelvic fracture? ER,etal. M,Haut AH, Oyetunji TA,KhouryA, Cubangbang Bolorunduro Haider OB, diagnostic fewer are Disparities intraumacare: 20. discussion 9-81. racialdisparitiesin phenomenon helpexplain Hui X, ZG,Zafar SN, Hashmi HaiderAH, trauma patients tend worse-than-expected at tocluster centers traumamortality:with can this 19. 2010;148:202-8. pe forinner-city case the outcomes: trauma ournal of surgery 2013;205:365-70. 2013;205:365-70. ournal ofsurgery diatric chest pain.Congenita diatric chest pediatrics2013;167:700-7. disparity. JAMA surgery 2013;148:76-80. surgery 2013;148:76-80. JAMA disparity. destrian injury Surgery programs. destrian prevention cs and the associated radiation exposure and exposure radiation associated the and cs . J Health Care Poor Underserved 2009;20:896-913. Underserved Poor HealthCare . J trauma outcomes? AnnSurg 2013;258:572-9; outcomes? trauma visits. ArchSurg 2008;143:1057-61;discussion61. With Race/Ethnicity. Pediatrics 2007;119:e1319- Race/Ethnicity. Pediatrics With tests conductedforuni tests Schneider EB, Efron DT, et al. Minority etal. Efron DT, EB, Schneider leukaemia and brain tu leukaemiaandbrain a J. Disparities exist intheemergency Disparitiesexist a J. ack MM.Differences inSeverity-Adjusted aphy scans in childhood or adolescence: data data childhood oradolescence: scansin aphy l heart disease 2010;5:285-91. 2010;5:285-91. disease l heart nsured patientswith nsured mours: a retrospective aretrospective mours: Rate of IAI on CT, % IAI% Rate of onCT, Abdominal CTinED,% ED dispositiontohome, % % Moderate/High, Mild,% % Low, intervention IAIRisk for undergoing ED, %* in Ultrasound performed GCS 15,% Male, % Age (IQR) age Median ≥ Table 2. IAI: injury GCS: Glasgow ComaScale; intra-abdominal This article is protected Allrights bycopyright. reserved. al. et JF Holmes from Modified Table 1Riskstratification criteria risk level for IAI risk levelfor 2 years, 2 % PECARN rule

Accepted Articleintervention undergoing Very Low 0.1% Without criteriaforlo Without 0.1% Low Very Variable Variable Moderate Moderate High Low Low

Patient characteristics byPatient characteristics provider-det intervention undergoing IAIRisk of 0.7% 1.4% 5.4% 11.3 (6.0, 15.1) 11.3 (6.0,15.1) Entire Group Entire Group (n=10,825) (n=10,825) 12.4 44.4 50.6 43.7 14.2 42.1 86.8 61.5 90.8 9.1 10

criteria for moderate or high risk level risklevel orhigh moderate criteria for orvomiting without sounds breath pain, decreased Thoracic wall trauma, comp level risk criteriaforhigh Abdominal tendernesswithout abdominaltrauma GCS score<14withblunt signor wall trauma/seatbelt abdominal Evidence of

among withblunttorsotrauma patients 11.7 (6.4, 15.3) 11.7 (6.4,15.3) Non-Hispanic Non-Hispanic ermined patient race and ethnicity andethnicity patientrace ermined (n=5847) (n=5847) White 13.6 51.8 45.6 48.1 15.3 36.5 86.1 62.5 91.2 Clinical Criteria 9.8 w, moderate, or high risk level risklevel moderate, orhigh w, laints ofabdominal 10.9 (6.1, 14.9) 10.9 (6.1,14.9) Non-Hispanic Non-Hispanic Race/Ethnicity (n=3687) (n=3687) Black Black 10.3 32.7 58.6 36.3 12.4 51.3 88.7 60.0 90.9 6.5

9.9 (4.4, 14.7) 9.9 (4.4,14.7) (n=1291) (n=1291) Hispanic 10.4 44.2 50.7 44.5 14.3 41.2 13.1 84.0 61.3 88.2

This article is protected Allrights bycopyright. reserved. IAI intervention for undergoing suspicion andprovider severity, Table 3. Hispanic n=1265 Hispanic n=3632,and n= inED,overall *For ultrasoundperformed abdominal injury;IQR: CT: computed tomography; ED:emergency department; GCS:Glasgow Coma Scale;IAI: intra- Accepted >10% 6-10%* 1-5%*,# <1%*,^,# intervention acute undergoing ClinicalIAI suspicion for *,^,# Moderate/High Mild*,# Low *,^,# intervention IAIRisk for undergoing 15*,^,# 14 score GCS Female *,^,# Male*,^,# Sex <2*,# ArticleAge group,y Variable ≥ 2*,^,# Percentage of children receiving abdominal receiving children Percentage of

interquartile range range interquartile 31.9 (28.9, 35.1) 31.9 (28.9,35.1) 29.5 (27.6,31.5) 48.8 (47.4,50.2) 58.1 (52.2,63.7) 51.0 (48.9,53.1) 52.2 (50.6,53.9) 53.3 (52.0,54.7) 35.6 (31.5,39.9) 92.3 (87.8, 95.5) 92.3 (87.8,95.5) 95.3 (92.1,97.5) 86.0 (83.7,88.2) 39.0 (37.5,40.4) 75.0 (73.3,76.6) non-Hispanic non-Hispanic (n=5847) (n=5847) White

10,642, white non-Hispanic n=5745, blacknon- whitenon-Hispanicn=5745, 10,642, Children, % (95% CI) Children, %(95%CI) CT by race/ethnicity, age, sex, injury age, sex, byCT race/ethnicity, 95.5 (88.9, 98.8) 95.5 (88.9,98.8) 89.5 (82.3,94.4) 75.2 (70.8,79.3) 23.0 (21.5,24.5) 58.4 (55.8,61.1) 21.9 (18.2,26.0) 17.1 (15.5,18.9) 29.4 (27.9,31.0) 51.1 (43.7,58.5) 32.1 (29.7,34.5) 33.1 (31.2,35.1) 33.3 (31.7,34.9) 27.2 (22.5,32.4) non-Hispanic non-Hispanic (n=3687) (n=3687) Black Black

91.2 (76.3, 98.1) 91.2 (76.3,98.1) 91.4 (81.0,97.1) 85.9 (80.0,90.6) 32.4 (29.5,35.4) 65.0 (60.9,68.9) 34.6 (27.8,41.9) 25.0 (21.4,28.9) 40.6 (37.6,43.5) 51.1 (40.5,61.5) 42.5 (38.1,47.0) 45.2 (41.7,48.7) 44.7 (41.8,47.6) 40.1 (32.3,48.4) (n=1291) (n=1291) Hispanic

Ml 12(.,14 This article is protected Allrights bycopyright. reserved. Accepted 1.2(1.1,1.4) 6.1(4.7, 7.9) 1[reference] generalized IAI:estimating equations; CI: confidenceCT: computed interval; tomography; ED:Emergency department; GEE: 1.1(1.02,1.2) Moderate/High Mild 0.9(0.8,1.1) 1.4(1.1,1.7) Low IAI-intervention for Risk ED in Ultrasound performed 0.8(0.7,0.9) (male) Sex 1 [reference] AdjustedOR(95%CI) Age group ( Hispanic Black non-Hispanic Article non-Hispanic White Race/Ethnicity Variable IAI interventionrisk (PECARNrule calculated) undergoing for blunt torsotrauma:ResultsofGEE,adjustedfo Table 4. abdominal injury CI: confidenceCT: computed interval; tomography; GCS:Glasgow ComaScale;IAI: intra- vs.Hispaniccomparisons # P<0.05forblacknon-Hispanic vs.Hispaniccomparisons ^ P<0.05forwhitenon-Hispanic comparisons vs.blacknon-Hispanic * P<0.05forwhitenon-Hispanic Relationship of physician-identified patient race and ethnicity to use of CT in pediatric pediatric in useofCT to and ethnicity race patient of physician-identified Relationship ≥ 2yas 10(.,12 1.0(0.9,1.2) years) 2 intra-abdominal injury; OR:oddsratio intra-abdominal r age group, sex, ultrasound performed in ED and ultrasoundperformedinED group, sex, age r