a Brazilian hospital in patients with hematological diseaseat admitted to an intensive care unitspecialized Predictors of mortality among patients with acuteleukemias Hematology, Transfusion and Cell Therapy

w w w . h t c t . c o m . b r

Hematology, Transfusion and Cell Therapy

Original article

Predictors of mortality among patients with acute leukemias admitted to an intensive care unit

specialized in patients with hematological disease at a Brazilian hospital

Lorena Costa Corrêa

, Dahra Teles

, Odin Barbosa da Silva

, Gustavo Henriques Trindade-Filho

, Paula Loureiro

,

Maria do Socorro Mendonc¸a Cavalcati

aUniversidadedePernambuco(UPE),Recife,PE,Brazil

bFundac¸ãodeHematologiaeHemoterapiadePernambuco(Hemope),Recife,PE,Brazil

a r t i c l e i n f o

Articlehistory:

Received19February2018 Accepted7January2019 Availableonline26April2019

Keywords:

Acuteleukemia Intensivecare Bloodmalignancies Mortality

Sepsis

a bs t r a c t

Introduction:Hematologists deal every daywith high mortality rates ofacute leukemia patients.ManytimesthesepatientsneedIntensiveCareUnit(ICU)supportandsomegen- eralICUteamsbelievethatthesepatientshaveamuchgreaterchanceofdyingthanpatients withotherpathologies.InBrazil,datarelatedtomortalityratesandICUsforacuteleukemia patientsarescarce.

Methods:Therefore,toassessmortalitypredictorsinpatientswithacuteleukemiaadmitted toa specializedhematologicalICU,weevaluated demographics,supportivecare,hospi- talizationtime,diseasestatus,admittingdiagnosis,neutropenia,numberoftransfusions andAcutePhysiologyandChronicHealthEvaluation(APACHE)/SepsisRelatedOrganFail- ure Assessment (SOFA)scoresaspossible factorsassociatedwithmortality. Datawere extractedfromthefirstadmissionrecordsof110patientswithacuteleukemiaadmitted totheHemocentrodePernambuco(Hemope)ICUbetween2006and2009.

Results:Inthisretrospectivecohortstudy,72/110ofthepatientsweremen,and64/110were fromthemetropolitanareaofRecife.Thepatients’agemedianwas43.5years(±17.9);67.3%

hadacutemyeloid leukemia(AML)and32.7%hadacutelymphoidleukemia.Themain admittingdiagnosisintheICUwassepsis (66.7%).Themean APACHEIIscorewas18.3.

Ofthetotal,65(59%)died,andthemortalityratewasindependentlyrelatedtolongerhos- pitalization(p<0.001),theincreaseintheAPACHEIIscore(p<0.038)andhavingreceived hemodialysis(p<0.006).Neutropenia,receivingmultipletransfusionsandusinganykind ofmechanicalventilationorvasoactivedrugonadmissionwerenotrelevanttomortality.

Factorsassociatedwithhighermortalityrateswere:longerhospitalization,increaseinthe APACHEIIscore,anduseofhemodialysis.

∗ Correspondingauthorat:RuaViscondedeMamanguapeS/N–Encruzilhada,Recife,PE,Brazil.

E-mailaddress:llorenacosta@gmail.com(L.C.Corrêa).

https://doi.org/10.1016/j.htct.2019.01.004

2531-1379/©2019Associac¸ ˜aoBrasileiradeHematologia,HemoterapiaeTerapiaCelular.PublishedbyElsevierEditoraLtda.Thisisan openaccessarticleundertheCCBY-NC-NDlicense(http://creativecommons.org/licenses/by-nc-nd/4.0/).

Conclusion: Withthesedata,topreventorganlesionsbeforeadmissiontotheICU,abetter strategymightbetoreducemortalityforleukemiapatients.

©2019Associac¸ ˜aoBrasileiradeHematologia,HemoterapiaeTerapiaCelular.Published byElsevierEditoraLtda.ThisisanopenaccessarticleundertheCCBY-NC-NDlicense (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Introduction

Acuteleukemias aredisorders ofhematopoieticstem cells, characterizedbyfastanduncontrolledproliferationofimma- turecells(blasts)inthebonemarrow.¹Theseveryaggressive disorders suppress the productionand normal function of bonemarrowcells,causingimmunosuppression.Itrequires urgenttreatment,andthe complete remissionrateishigh, approximately60–75%foracutemyeloidleukemia(AML).²

Apart from the bone marrow suppression promoted by leukemia,theuseofchemotherapy(CT)drugsraisesimmuno- suppression,resulting infrequentinfectiouscomplications, whichinturnleadtoorgandysfunction.Patientsareoftensep- tic,withacutelunginjury/acuterespiratorydistresssyndrome (ALI/ARDS)andacutekidneyfailure(AKF).Theseconditions leadpatientstorequirehospitalizationinintensivecareunits (ICUs).^2,3

The effervescence of successful anticancer therapies is linked to the better management of critically ill patients, whichclamorsfortherethinkingoftheroleofintensivecare, regardingindications,benefitsandprognosticfactorsrelated toneoplasmpatientsincriticalconditions.Intheearly1990s, cancersurvivorsmaintainedagood qualityoflifeandhad nolimitationstodailyactivitiesafteradmissiontointensive care.⁴Despitehighratesofmortality(45–87%),studieswithin this population ofcriticallyill patients showed that inten- sivecarewasbeneficial.^5–13Thereafter,predictorsofmortality amongpatientswithcancerintheICUstartedtoberepeatedly evaluated.^3,6,13–17

Currently,althoughtherearemanyanalysesonpredictors ofmortalityamongcancerpatientsintheICU,reportsofunits specializedinexclusivelytreatingpatientswithblooddisease areinsufficient.¹⁸

Thus,theaimofthisstudyistoassessmortalityriskfactors amongpatientswithacuteleukemiaadmittedtoaspecialized hematologicalICU.

Materials and methods

Thisisaretrospectivecohortstudy.Allpatientswithacute leukemiaover16yearsofageontheirfirstadmissiontothe ICUoftheHemocentroofPernambuco(Hemope)Foundation HospitalbetweenJanuary2006andDecember2009hadtheir medicalrecordsretrospectivelyevaluatedinthisstudy.Dur- ing thisperiod, therewere 426admissions, 137withacute leukemia,and 110patientsmet theinclusion criteria. This

ICU isspecialized inpatientswith blood disease.Itispart oftheHemope FoundationHospital,wherethereisaward with 36 beds for treating acute leukemia (32), hemophilia (2) andsicklecell disease(2).Thisisacenterofexcellence inPernambucoandneighboringstates.TheResearchEthics Committee approved this project under the CAEE number 11810213.5.0000.51952013.

Thediagnosisofacuteleukemiawasestablishedaccord- ing to criteria of the World Health Organization (WHO), revised in 2008. The disease status was categorized as:

aplasia, remission, progression or relapse. Aplasia was defined as the moment after CT when the hemoglobin (Hb) is lower than 10g/dl, associated leukocyte count is lower than 1000cells/mm³ and platelets are lower than 100,000cells/mm³.Remissionwasdefinedasthestatusafter CTwhenthepatienthasnoactivedisease(<5%blastsinthe bonemarrow).Progressionwasdefinedasfailureoftherapy anytimebeforeadmissionorwithrefractorydisease.Relapse wasclassifiedasareturnofdiseaseactivityafteraperiodof improvement(remission).Reasonsforadmissionwerethose mentioned inthe medicalrecordsand were categorizedas respiratoryfailureofsepticandnonsepticorigins,acuterenal failure(ARF),sepsis,sensoryalterations,post-operativebleed- ingandshock.

The Acute Physiology and Chronic Health Evaluation (APACHEII)scorewasappliedasapredictivemortalityindex within the first 24h after admission. It is a severity clas- sification system based upon 12 initialroutine physiologic measurements(temperature,heartrate,meanarterialpres- sure, serum sodium, serum potassium, serum creatinine, oxygenation, hematocrit, blood cell count, Glasgow coma score),ageandprevioushealthstatus.Anincreasingscoreis closelyrelatedtosubsequentin-hospitaldeath.¹⁷ Toassess progressionoforgan failure, theSepsis RelatedOrgan Fail- ureAssessment(SOFA)scorewasevaluatedondays1and3.

Thisscorehasadescriptivefunctionandindividualizesthe dysfunctionofeachorgandaily.¹⁹Thetwoscoresarecomple- mentaryandbothofthemwereretrospectivelycalculatedfor this paper athttp://www.globlarph.com/APACHEII.html and http://www.sfar.org/scores2/sofa2.html,respectively.^17,19

Supportivecarewascategorizedastheuseofinvasiveor non-invasivemechanicalventilation(IMV/NIMV),vasoactive drugs(VAD)onadmission,renalreplacementtherapy(RRT) andtransfusionsupport.TheIMVwasdefinedastheuseof anorotrachealtubeatanytimeduringtheICUstay,andthe NIMVwasconsideredeverytimethatmechanicalventilatory supportwasnecessarybutanorotrachealtubewasnot;VAD wasconsidered whenthe useofnoradrenalin ordopamine wasnecessary,and;theuseofanykindofhemodialysis,at leastonceandforanyduration,definedRRT.

Statisticalanalysis

Coredemographics,comorbidities,diseasecharacteristicsand factorsassociatedwithmortalityintheICUwereevaluated usingtheSTATAstatisticalsoftwareversion 13.0(StataCorp LP,CollegeStation,TX,USA).

Resultswerepresentedinthefrequencytablesforqualita- tivevariables.Descriptivestatistics(mean,standarddeviation, median and range) were used for continuous quantitative variables,accordingtotheoutcomeintheICU(deathorsur- vival).Theoutcomeresultswerecomparedbynonparametric Mann–WhitneyUtests.

Forquantitativevariables,measurementsofcentralten- dencyanddispersion weremade.Inthisstudy,considering thehighmortalityrateinthatICU(59%)onbivariateanaly- sis,theprevalencerate(PR)andtheir95%CIswereestimated toevaluatethecorrelationbetweenthevariablesanddeathin theunit.^20,21Variablesindependentlyassociatedwiththemor- talityratewereanalyzedwithamultipleCoxregression(with robustvariance)usingvariablesthathadp-valueslowerthan 0.20inunivariateanalysis.Achi-squaretestforlineartrend wasusedforcompletebloodcount(CBC)variables.Regarding non-quantitativevariables,anonparametricMann–Whitney Utestwasusedforcomparisonbetweenpatientswhodied andthosewhodidnot.

Themodelingprocessstartedwiththevariablethathadthe lowestp-valueduringbivariateanalysis,followedbytheaddi- tionofalltheotherswithp-valueslowerthan0.20,withjust thevariableswithdescriptivelevelslowerthan0.05(p<0.05) remainingforthefinalmodel.Finally,PRswereestimatedfor eachvariableinthefinalmodelwiththeirrespectiveconfi- denceintervals.

Results

Sixty-fivepatientsdiedduringICUhospitalization,totalling a59.1%mortalityrate.Table1showsdemographicandclin- icalcharacteristics ofthe studiedpopulation,whileTable2 hashemogramdata.Seventy-fouroutof110patients(67.3%) hadAML,withFrench-American-British(FAB)subtypesclas- sifiedasAML-M3(n=17,23.0%), AML-M1(n=16,21.6%)and AML-M2(n=15,20.3%). Theothers (36/110) hadacute lym- phoidleukemia(ALL),58.3%casesofB-cell(21/115)and30.6%

oftheT-celllineage(11/115).

Themainadmissiondiagnosiswasacuterespiratoryfail- ure (ARF) secondary to sepsis (39/110, 37.1%), followed by sepsis(37/110,33.6%).Theothercausesofadmissionappearin Table1.Atthatpoint,themaininfectionwaspulmonaryand 30%ofthebloodcultureswerepositive,Pseudomonasaerugi- nosaandKlebsiellapneumoniaebeingthemostcommonagents detected.

The APACHE II scores were calculated for 85 admitted patients(77.3%),duetothelackofdataintheother25medical records.ThemeanAPACHEIIscoreforthese85patientswas 18.3(±6.4),rangingfrom8to43,asshowninFigure1.TheSOFA scorewascalculatedretrospectively,however,onthefirstday ofadmission,63patientscouldnothavetheSOFAcalculated, anumberthatincreasedto81onthethirdday.Thus,themean variationinSOFAin27patientswas2.7(±3.5).

Table1–Characterizationofpatientswithacute leukemiahospitalizedintheICU,accordingtosocial anddemographicfeatures,basepathology,reasonsfor admission,severityprofileandorgandysfunctionatthe HemopeFoundationfrom2006to2009.

Number %

Gender

Male 72 65.5

Female 38 34.5

Age(years)

Median(dp) 43.5(17.9)

Cityofresidence

MetropolitanareaofRecife 46 41.8

Othermunicipalities 64 58.2

Bloodcondition

AML 74 67.3

ALL 36 32.7

Leukemiastatus^a

Aplasia 40 40.4

Diagnosis 26 26.3

Progression 21 21.2

Relapse 10 10.1

Remission 2 2.0

Reasonsforadmission^b

Acuterespiratoryfailure,septic 39 37.1 Acuterespiratoryfailure,nonseptic 18 17.1

AKF 1 0.9

Sepsis 37 33.6

Sensoryalteration 4 3.8

Bleeding 1 1.0

Oncologicemergencies 1 1.0

Post-operative 1 1.0

Rupturedspleen 3 2.9

Variables Mean Standard

deviation APACHEII^c

Mean(dp) 18.3 (6.4)

Timeofhospitalization(days)^d

Median(dp) 19.1 (20.5)

SOFAD1^e

Mean(dp) 8.7 (3.6)

SOFAD3^f

Mean(dp) 11.1 (5.0)

SOFA^g

Mean(dp) 2.7 (3.5)

Ignoreddata:(a)11;(b)5;(c)25;(d)1;(e)63;(f)81;(g)83.

Inregardtosupportivecare,99/110patientsusedmechan- ical ventilation (MV) ofany kind, and only 21/110 needed RRT. The use of vasopressors on admission was not rele- vant.Amongtransfusedpatients,52.7%received1–5unitsof leukocyte-poorpackedredcells(LPPRC),and51.8%received 1–5platelettransfusions(Table3).

Afterbivariateanalysis,mortalityamongindividualswho underwenthemodialysiswashigherthanthosewhodidnot (p<0.001),andtheuseofMVmightalsobeafactorassoci- atedwithhighermortalityrates(p=0.001)inthisscenario.We additionallyfoundthatpatientswhohadahigherhemoglobin

Table2–Distributionofpatientswithacuteleukemia admittedtotheICUandbloodcellcountdataat admissionanddischargeattheHemopeFoundation from2006to2009.

Variables No. %

Hemoglobinvaluesatadmission(g/dL)^a

<5 5 4.7

5–6.9 21 19.6

7–10 64 59.8

>10 17 15.9

Neutrophilsvaluesatadmission(cells/mm³)^b

<500 50 64.9

500–999 3 3.9

1000–4999 13 16.9

5000–9999 4 5.2

10,000–30,000 6 7.8

30,001–99,999 1 1.3

≥100,000

Hemoglobinvaluesondischarge(g/dL)^c

<5 0 0.0

5–6.9 14 14.3

7–10 63 64.3

>10 21 21.4

Neutrophilsvaluesatadmission(cells/mm³)^d

<500 38 57.6

500–999 2 3.0

1000–4999 16 24.2

5000–9999 3 4.5

10,000–30,000 7 10.6

30,001–99,999 0 0.0

≥100,000 0 0.0

Ignoreddata:(a)3;(b)33;(c)12;(d)44.

value(p=0.039)and morethan 500neutrophils/mm³ inthe blood count (p=0.014) atadmission were less likely to die (Table4).

The summary and comparison of the central ten- dencymeasurementsofthesevariablesamongHemope-ICU patientsappearinTable5.Variablesindependentlyassociated withthemortalityrateinthispopulationwereevaluatedwith amultipleCoxregressionwithrobustvariance.Inthismodel,

Table3–Distributionofpatientswithacuteleukemia hospitalizedintheICUaccordingtosupportive therapiesapplied.HemopeFoundation,2006to2009.

Variables Number %

Mechanicalventilation

None 11 10.0

IMV 77 70.0

IMVorNIMV 22 20.0

Kidneysupport

None 89 80.9

Yes 21 19.1

Leukocyte-poorpackedredcells

Didnotuse 25 22.7

1–5 58 52.7

6–10 11 10.0

>10 16 14.5

Plateletconcentrate(transfusions)

Didnotuse 19 17.3

1–5 57 51.8

6–10 11 10.0

>10 23 20.9

IMV:invasivemechanicalventilation;NIMV:non-invasivemechan- icalventilation.

thefollowingvariableswereincluded:havingundergoneRRT, usedIMV,stayedlongerintheICU,presentedhigherAPACHE IIscores,presentedhigherHb valuesonadmission,hadan increaseinneutrophilsondischarge,andhadahighernumber ofplatelettransfusions.TheSOFAvalueswerenotanalyzed duetothesignificantlossofinformation.

Amongthesampleofpatientswithacuteleukemiainthis study,themortalityintheHemopeICU wasindependently associatedwithRRT[PR_adj1.47(1.12–1.94)],longer hospital- ization [PRadj 1.01 (1.00–1.01)]and higher APACHEII scores [PR_adj1.02 (1.00–1.04)].Mortality inthissetting forpatients whounderwentRRTwas47%higherthanamongthosewho didnot(p=0.006),regardlessoftheICUhospitalizationtime and the APACHE II score. Death risk alsoincreased by1%

(p=0.001)foreachadditionaldayattheunit,independently oftheAPACHEIIscoreandofhavingRRTornot.Inaddition,

10 20 30 40 50

APACHE2

Figure1–APACHEIIscoresinpatientswithacuteleukemiahospitalizedintheICUattheHemopeFoundationfrom2006to 2009.

Table4–DistributionofpatientswithacuteleukemiahospitalizedintheICU,accordingtotheircharacteristicsand occurrenceofdeath,attheHemopeFoundationfrom2006to2009.

Variables Total DeceaseintraICUnumber(%) PR(CI95%) p

Gender 0.564

Male 72 44(61.1) 1

Female 38 21(55.3) 0.90(0.64–1.27)

Bloodcondition 0.911

AML 74 44(59.5) 1

ALL 36 21(58.3) 0.98(0.70–1.37)

Reasonforadmission 0.135

Acuterespiratoryfailure,septic 39 25(64.1) 1

Acuterespiratoryfailure,nonseptic 18 8(44.4) 0.69(0.39–1.23)

AKF 7 6(85.7) 1.34(0.91–1.96)

Sepsis 31 16(51.6) 0.81(0.53–1.22)

Others 10 6(60.0) 0.94(0.53–1.64)

Hemoglobinlevelsonadmission(g/dL) 0.039^a

>10 17 8(47.1) 1

7–10 64 36(56.3) 1.19(0.69–2.07)

<7 26 20(76.9) 1.63(0.94–2.83)

Neutrophilcountonadmission(cells/mm³) 0.091

≥500 27 13(48.2) 1

<500 50 35(70.0) 1.45(0.94–2.24)

Hemoglobinlevelsondischarge(g/dL) 0.292^a

>10 21 11(52.4) 1

7–10 63 36(57.1) 1.09(0.69–1.73)

<7 14 10(71.4) 1.36(0.80–2.31)

Neutrophilcountondischarge(cells/mm³) 0.014

≥500 28 11(39.3) 1

<500 38 28(73.7) 1.88(1.14–3.10)

IMV 0.001

None 33 1(3.0) 1

Yes 77 64(83.1) 27.4(3.9–191.2)

Kidneysupport <0.001

None 89 47(52.8) 1

Yes 21 18(85.7) 1.62(1.25–2.11)

Leukocyte-poorpackedredcells(IU) 0.634

Didnotuse 25 14(56.0) 1

≤10 69 40(58.0) 1.04(0.69–1.55)

>10 16 11(68.8) 1.28(0.76–1.99)

Plateletconcentrate(transfusions) 0.152

Didnotuse 19 9(47.4) 1

≤10 68 39(57.4) 1.21(0.72–2.03)

>10 23 17(73.9) 1.56(0.91–2.66)

PR:prevalencerate;IMV:invasivemechanicalventilation;LPPRC:leukocyte-poorpackedredcells;PC:plateletconcentrate.Boldvalues-the Valueswithp<0,001ortendingto0,001

a Chi-squaretestforlineartrend.

theincreaseofoneunitintheAPACHEIIscorerelatestoa 2%increaseinthemortalityrate(p=0.038),independentlyof havingundergoneRRTornot(Table6).

Discussion

Weanalyzed110patientswithacuteleukemiasduringcriti- calillness.ThisisararelyfoundpatientprofileingeneralICUs andtherearefewICUswhicharespecializedinhematological diseasesaroundtheworld.Therefore,describingthispopula- tionandevaluatingpredictorsofoutcomeareveryimportant

toimproveknowledge,notonlyabouttheircriticalillness,but alsoabouttheirprognosisduringthestayintheICU.

Wefoundahighmortalityrateinthestudiedpopulation (59.1%),beingRRTthemostimportantriskfactor.Thismor- talityseemsnottobedifferentfromthosepreviouslyreported byAzoulay et al.and Abraham et al.incancer patientsin theICU.^5,6 Twenty-ninepercentofthepatientswereadmit- tedwithAKFandthemortalityratewas85%forthosewho neededhemodialysis.Thesedataaresimilartothoseformerly reportedbySantosetal.Twelvepercentto49%oftheAKFrate amongcancerpatientsandahighermortalitywererelatedto RRT(80%).Althoughtheseresultsare poor,theyareinline

Table5–Descriptivestatisticsofquantitativevariablesofpatientswithacuteleukemia,accordingtooccurrenceofdeath, attheHemopeFoundationfrom2006to2009.

Variables Death p-Value

No(n=45) Yes(n=65)

Age 45 65 0.292

APACHE2^a 31 54 0.025

Durationofhospitalization(days)^b 44 65 0.002

SOFAD1^c 10 37 0.002

SOFAD3^d 6 23 0.003

DeltaSOFA^e 5 22 0.004

Ignoreddata(a)25;(b)11(c)63;(d)81;(e)83.

Table6–Estimationofthemortalityprevalenceratioamongpatientswithacuteleukemia,withthemultipleCox regressionmodel.

Variables PRcr PRadj(CI95%) p-Value

Withkidneysupport 1.62 1.47(1.12–1.94) 0.006

Timeofhospitalization(days) 1.01 1.01(1.00–1.01) 0.001

APACHEII 1.03 1.02(1.00–1.04) 0.038

PRcr:crudeprevalenceratio;PR_adj:adjustedprevalenceratio.

withtheoutcomesfoundinnon-cancerpatientsadmittedto theICUwithAKF.^22,23

Otherimportantpredictorsofmortalitywerelongerhos- pitalizationtimeattheunit andincreaseinthe APACHEII score.Wedidnotfindanydescriptionrelatedtothelengthof hospitalizationinotherstudies.Andalthoughtheincreasein theAPACHEIIscorehasbeenassociatedwithahigherchance ofdeathintheICU,thisindexwasnotabletoproperlyesti- matemortalityinthis studybecauseanaverageAPACHEII scoreof18predictsamortalityrateof20–30%,whichisvery farfromthemortalityratesinourstudy.Despitethisdispar- ity,similardatahavealreadybeendescribedinpublications byHampshireandJonge.^24,25

Sepsiswasthereasonforadmissionforalmost70%inthis study.Sepsisisafrequentconditionamongpatientsadmitted toICUsandalsoamongpatientswithcancer.Itsmortalityrate isaround30–50%,notonlyforcancerpatients,butalsoforthe generalpopulationduringcriticalillness.Therefore,despite havingahighmortalityrateinourstudy,itwasnotsodiffer- entfromtheratefoundamongsepticpatientswithoutcancer.

Consequently,wemustthinkaboutleukemicpatientsinthis setting,aspatientswithaseverelife-threateningdisease,but withanalmost50–70%chanceofsurviving,whenadmittedto theICU.^26–29

Therewasalsoatendencyforlowermortalitywhenthere wasahigherhemoglobinlevelonadmission(p=0.039)and whenneutrophil levels were higherthan 500cells/mm³ on discharge(p=0.014).Onecanthinkthatrecoveringfromneu- tropeniamaybeaprotectingfactoragainstmortalityinthis scenario,butatthispoint,asalsomentionedinformerpubli- cations,studieswithagreaternumberofpatientsareneeded toconfirmorrefutethistrend.²⁶

Finally,thepredictorsofmortalityfoundinourstudyare alsoinaccordancewithICUstudiesrelatedandnotrelatedto cancerpatients.Inthismanner,itisimportanttonotethatto offerICUsupporttoleukemicpatientsistogivethemachance

tosurviveaftercriticalillness.Ofcourse,multicenterprospec- tivestudies withacontrolpopulation maybenecessaryto elucidatemanyotherquestions.

Conclusion

Ourmortalityratewashigh,butnotdifferentfrompatients admittedtoICUsforsepsiswithoutleukemiasorotheronco- logic diseases. Predictors of higher mortality were: longer hospitalization,day-by-dayincreasingAPACHEIIscoresand undergoinghemodialysis,whichisalsosimilartothegeneral populationwithcriticalillness.

Allthesedataareimportantforpublichealthbecausethey callattentiontothefactthatfactorsassociatedwithdeathin theleukemicpopulationarenotdifferentfromthereasonsfor deathinthegeneralpopulationduringcriticalillness.There- fore,itmightbeimportanttorevisitadmissionpoliciesinthe ICU, andperhaps,tomodifythe rankingofpriorityrelated toleukemiapatients,whichmayimprovetheoutcomesafter criticalillnessesinthispopulation.

Conflicts of interest

Theauthorsdeclarenoconflictsofinterest.

Acknowledgments

We would like to thank Prof. Dr. Cesar de Almeida Neto oftheFundac¸ãoPró-Sangue,HemocentrodeSãoPauloand DepartmentofMedicalSciencesattheCollegeofMedicine, UniversityofSãoPaulo,forhisinsightfulcomments,which improvedourmanuscript.

references

1. SchellongowskiP,StaudingerT,KundiM,LaczikaK,LockerGJ, BojicA,etal.Prognosticfactorsforintensivecareunit admission,intensivecareoutcome,andpost-intensivecare survivalinpatientswithdenovoacutemyeloidleukemia:a singlecenterexperience.Haematologica.2011;96(2):7–231.

2. FreedmanAS,FriedbergJW,AsterJC.Classificationofthe hematopoieticneoplasms.[cited2013October].Available from:

http://www.uptodate.com/contents/classification-of-the- hematopoietic-neoplasms.

3. TacconeFS,ArtigasAA,SprungCL,MorenoR,SakrY,Vincent J.Characteristicsandoutcomesofcancerpatientsin EuropeanICUs.CritCare.2009;13(1):1–10.

4. YaulE,RohatinerAZ,ListerTA,HindsCJ.Longtermprognosis andqualityoflifefollowingintensivecareforlife-threatening complicationsofhaematologicalmalignancy.BrJCancer.

1991;(64):42–938.

5. AzoulayE,SoaresM,DarmonM,BenoitD,PastoresS,Afessa B.Intensivecareofthecancerpatient:recentachievements andremainingchallenges.AnnIntensiveCare.2011;1:1–13.

6. Ben-AbrahamR,SegalE,HardanI,ShpilbergD,StemerS, ShitritA,etal.Hemato-oncologypatientsinacuterespiratory failureintheICU.Harefuah.1997;133(3–4):4–91.

7. KressJP,ChristensonJ,PohlmanAS,LinkinDR,HallJB.

Outcomesofcriticallyillcancerpatientsinauniversity hospitalsetting.AmJRespirCritCareMed.

1999;160(6):61–1957.

8. SapolnikR.Intensivecaretherapyforcancerpatients.J Pediatr(RioJ).2003;Suppl.2:42–231.

9. AzoulayE,AlbertiC,BornstainC,LeleuG,MoreauD,Recher C,etal.Improvedsurvivalincancerpatientsrequiring mechanicalventilatorysupport:impactofnoninvasive mechanicalventilatorysupport.CritCare.2001;29(3):25–519.

10.ThiéryG,AzoulayE,DarmonM,CiroldiM,DeMirandaS,Lévy V,etal.Outcomeofcancerpatientsconsideredforintensive careunitadmission:ahospital-wideprospectivestudy.JClin Oncol.2005;23(19):13–4406.

11.LecuyerL,ChevretS,ThieryG,DarmonM,SchlemmerB, AzoulayE.TheICUtrial:anewadmissionpolicyforcancer patientsrequiringmechanicalventilation.CritCareMed.

2007;35(3):14–808.

12.LeucemiaAguda[Internet].RiodeJaneiro:INCA;2008[cited 2012Nov6].Availablefrom:

http://www.inca.gov.br/conteudoview.asp?id=344 13.CapraME.EstimativaDoNúmerodeCasos,Distribuic¸ão

RegionaleSobrevidadePacientescomDiagnósticode LeucemiaMielóideAgudanoEstadodoRioGrandedoSul- Brasil[dissertac¸ão].PortoAlegre:UniversidadeFederaldoRio GrandedoSul,FaculdadedeMedicina,Pós-Graduac¸ãoem Medicina:CiênciasMédicas;2004.

14.UTIHistóricoDaCriac¸ãodasUTIS[Internet];2011.Available in:www.medicinaintensiva.com.br.

15.HorsterS,StemmlerHJ,MandelPC,MückA,TischerJ, HausmannA,etal.Mortalityofpatientswithhematological malignancyafteradmissiontotheintensivecareunit.

Onkologie.2012;35(10):61–556.

16.KnausWA,DraperEA,WagnerDPZJ.APACHEII:aseverityof diseaseclassificationsystem.CritCareMed.

1985;13(10):29–818.

17.SilvaOB,CorreaL,LoureiroP,AraujoE,TelesD,Vasconcelos LA,etal.Predictorsofmortalityinpatientsfroma

hematologicICUinBrazil.CritCareMed.2012;16:S146.

18.DaviesHT,CrombieIK,TavakoliM.Whencanoddsratios mislead?BMJ.1998;316:92–89.

19.VincentJL,MorenoR,TakalaJ,WillattsS,DeMendonc¸aA, BruiningH,etal.TheSOFA(Sepsis-relatedOrganFailure Assessment)scoretodescribeorgandysfunction/failure.

IntensiveCareMed.1996;22:10–707.

20.BarrosAJ,HirakataVN.Alternativesforlogisticregressionin cross-sectionalstudies:anempiricalcomparisonofmodels thatdirectlyestimatetheprevalenceratio.BMCMedRes Methodol.2003;3(1):21.

21.CoutinhoLM,ScazufcaM,MenezesPR.Métodosparaestimar razãodeprevalênciaemestudosdecortetransversal.Rev SaúdePública.2008;42(6):8–992.

22.HampshirePA,WelchCA,McCrossanLA,FrancisK,Harrison DA.Admissionfactorsassociatedwithhospitalmortalityin patientswithhaematologicalmalignancyadmittedtoUK adult,generalcriticalcareunits:asecondaryanalysisofthe ICNARCCaseMixProgrammeDatabase.CritCare.

2009;13(4):137.

23.JongeE,BosMM.PatientswithcancerontheICU:thetimes theyarechanging.CritCare.2009;13(9):122.

24.SantosP,MaximinoJ,PaivaA,BaldaiaAJ,LoureiroA,FariaF.

Outcomeofcriticallyillpatientswithhaematological malignanciestreatedwithrenalreplacementtherapy.PortJ NephrolHypert.2011;25(4):50–145.

25.LianoF,PascualJ.Epidemiologyofrenalfailure:aprospective, multicenter,community-basedstudy.MadridAcuteRenal FailureGroup.KidneyInt.1996;50(3):8–811.

26.HanSS,KimS,AhnSY,LeeJ,KimDK,ChinHJ,etal.Duration ofacutekidneyinjuryandmortalityincriticallyillpatients:a retrospectiveobservationalstudy.BMCNephrol.2013;14:133.

27.LegrandM,MaxA,PeigneV,MariotteE,CanetE,Debrumetz A,etal.Survivalinneutropenicpatientswithseveresepsisor septicshock.CritCareMed.2012;40(1):9–43.

28.SalluhJI,SoaresM.PolíticasdeAdmissãodePacientes OncológicosnaUTI:HoradeReverosConceitos.RevBrasTer Intensiva.2006;18:8–217.

29.WilliamsMD,BraunLA,CooperLM,JohnstonJ,WeissRV, QualyR,etal.Hospitalizedcancerpatientswithsevere sepsis:analysisofincidence,mortality,andassociatedcosts ofcare.CritCare.2004;8(5):81–129.