Multidisciplinary care improves clinical outcome and reduces medical costs for pre-end-stage renal disease in Taiwan

bs_bs_banner

Nephrology19 (2014)699–707

Original

Article

Multidisciplinary

care

improves

clinical

outcome

and

reduces

medical

costs

for

pre-end-stage

renal

disease

in

Taiwan

YUE-REN

CHEN,

1

_YU

_YANG,

2

_SHU-CHUAN

_WANG,

1

_WEN-YU

_CHOU,

1

_PING-FANG

_CHIU,

1

_CHING-YUANG

_LIN,

3,4

WEN-CHEN

TSAI,

5

_JER-MING

_CHANG,

6,7

_TZEN-WEN

_CHEN,

8

_SHYANG-HWA

_FERNG

9

_and

_CHUN-LIANG

_LIN

10

1_{DivisionofNephrology,InternalMedicine,ChanghuaChristianHospital,ChanghuaCity,}2_{SchoolofMedicine,ChungSangMedicalUniversity,}3_Collegeof

Medicine,5_{DepartmentofHealthServicesAdministration,ChinaMedicalUniversity,}4_{DivisionofPediatricNephrologyandClinicalImmunologyCenter,China}

MedicalUniversityHospital,TaichungCity,6_{DepartmentofInternalMedicine,KaohsiungMunicipalHsiao-KangHospital,}7_{DivisionofNephrology,Kaohsiung}

MedicalUniversityHospital,Kaohsiung,8_{DivisionofNephrology,InternalMedicine,TaipeiMedicalUniversityHospital,}9_{DivisionofNephrology,Internal}

Medicine,CathayGeneralHospital,TaipeiCity,and10_{DivisionofNephrology,InternalMedicine,ChiayiChangGungMemorialHospital,Chiayi,Taiwan}

KEYWORDS:

chronickidneydisease,hospitalization,medical

costs,mortality,multidisciplinarycare.

Correspondence:

DrYuYang,DivisionofNephrology,Internal

Medicine,ChanghuaChristianHospital,Chung

SangMedicalUniversity,SchoolofMedicine,

135NanhsiaoStreet,ChanghuaCity,500

Taiwan.Email:2219@cch.org.tw

Acceptedforpublication9July2014.

Acceptedmanuscriptonline27July2014.

doi:10.1111/nep.12316

Disclosure: Theauthors have no conflicts o

f

interesttodisclose.

SUMMARY ATAGLANCE

Thisretrospectivestudyclearly

demonstratesthatmultidisciplinarycareby

nephrologistsindialysispatientsprovided

bettervariousoutcomesandlessmedical

costs.However,thepatientmortalitywas

notdifferent.

ABSTRACT:

Aim: Multidisciplinarycare(MDC)forpatientswithchronickidneydisease (CKD) mayhelp tooptimize disease care and improve clinical outcomes. Ourstudyaimedtoevaluatetheeffectivenessofpre-end-stagerenaldisease (ESRD)patientsunderMDCandusualcareinTaiwan.

Method: Inthis 3-yearretrospectiveobservational study,we recruited822 ESRDsubjects,aged18yearsandolder,initiatingmaintenancedialysismore than3monthsfromfivecooperatinghospitals.TheMDC(n=391)groupwas caredforbyanephrologists-basedteamandtheusualcaregroup(n=431) wascared forby sub-specialists or nephrologists alonemorethan 90 days beforedialysisinitiation.Patientcharacteristics,dialysismodality,hospital utilization,hospitalizationatdialysisinitiation,mortalityandmedicalcost wereevaluated.Medicalcostswerefurtherdividedintoin-hospital, emer-gencyservicesandoutpatientvisits.

Results: TheMDCgrouphadabetterprevalenceinperitonealdialysis(PD) selection,lesstemporarycatheteruse,alowerhospitalizationrateatdialysis initiationand 15%reductionin therisk of hospitalization(P0.05). After adjustingforgender,ageandCharlsonComorbidityIndexscore,therewere lowerin-hospital and higheroutpatient costsin the MDC group during 3 monthsbefore dialysisinitiation(P0.05).Incontrast,medicalcosts(NT$ 146038 vs 79022) and hospitalization days (22.4 vs 15.5 days) at dialysis initiation were higher in the usual care group. Estimated medical costs during 3 months before dialysis till dialysis initiation, the MDC group yieldedareductionofNT$59251foreachpatient(P0.001).Patient mor-talitywasnotsignificantlydifferent.

Conclusion: Multidisciplinarycareinterventionforpre-ESRDpatientscould not only significantly improve the quality of disease care and clinical outcome,butalsoreducemedicalcosts.

The rapid increased incidence and prevalence of chronic kidney diseases (CKD) have been recognized as a global publichealth problemthat consumesa largeproportionof health carebudgets.In Taiwan, thenationalprevalence of CKDishigh,butdiseaseawarenessisinadequate.Only3.5 %

ofCKDpatientsareabletoreporttheirstageofdisease,and awarenessratesarecloselyrelatedtodiseaseseverity.1,2_A s

oneofthe mostrapidlyaging countrieswith anincreasing prevalence of diabetes mellitus, hypertension and

subse-quentlyCKD, Taiwan hasthe highest prevalenceand inci-dence of end-stage renal disease (ESRD) in the world.3 According to the Bureau of National Health Insurance (BNHI)annualreport,thedialysiscostsofESRDpatientsin Taiwan accounted for 5.0–7.52% of the total health-care resourcesinrecentyears.4

Medicalresourceutilizationhasbecomemorefrequentas disease progression and ESRD approach. According to a

recent study, the medical cost of pre-ESRD patients

©2014TheAuthors.NephrologypublishedbyWileyPublishingAsiaPtyLtdonbehalfofAsianPacificSocietyofNephrology.

ThisisanopenaccessarticleunderthetermsoftheCreativeCommonsAttribution-NonCommercial-NoDerivsLicense,whichpermitsuseanddistributioninany medium,

providedtheoriginalworkisproperlycited,theuseisnon-commercialandnomodificationsoradaptationsaremade.

Y-RChenetal.

increased sharplyinthe last6monthsprior todialysis ini-tiation.5 _{Service utilization and hospitalizations were the} major components of cost during the period immediately before and after dialysis initiation.6 _{Co-morbidity, such as} cardiovascular disease, is the major cause of mortalit y

among CKD patients, and CKD patients with increasing co-morbiditymayberesponsible fortherapidescalationof medicalexpenditures.7,8

Optimal management of CKD may improve clinical

outcome and decrease mortality, thus resultingin reduced hospitalization and medicalcost. In 2002,the US National KidneyFoundationlaunchedthepromotionofclinical prac-ticeguidelinesforthediagnosis,evaluationandmonitoring ofCKDwithintheKidneyDiseaseOutcomesQuality Initia-tive(NKFK/DOQI)inanefforttoincreasetheawarenessof optimal CKD care.9 _{Recent trials and studies also have} proven the efficacy of several interventions such as earl y

detection of CKD,10 _{prevention of kidney disease} pro-gression, early referral to nephrologists,11 _{promotion of} pre-dialysiseducations,12_{timelypreparationofrenal}

replace-ment therapy (RRT), and care by a comprehensive,

nephrology-based,multi-disciplinaryteam.13–15

Thepurposeof ourstudyistoevaluatethe effectiveness andassociationofclinicaloutcome,impactofdialysis

modal-ityselection,medicalutilizationandcostsbetweenpre-ESRD patientsreceivingMDCandusualcareinTaiwan.

MATERIALS

AND

METHODS

Study design and subjects

Ourretrospectivestudy wasdirectlylinkedto theNationalHealth

Insurance(NHI)systemwithitsoutpatientandin-hospitaldatabase

intheperiodfromJanuary2005toDecember2009.Patientsaged18

yearsorolderwhowereonmaintenancerenalreplacementtherapy

(includinghaemodialysisandPD)andwhohadcatastrophicillness

cardshadtheirmedicalexpensecoveredbytheNHIandwerecared

forinfivecooperatinghospitalsinthenorthern,centraland

south-ernareas ofTaiwan.Initially,854 subjectswereenrolled, butwe

excluded patients who received temporary dialysis due to acute

deterioration ofrenalfunction,andthosewho diedinthefirst3

months after dialysis initiation. In the end, 822 subjects were

recruitedandfurtheranalyzed.

Atotalof391subjectswereplacedintotheMDCgroupwhohad

receivednephrologists-basedCKDteamcareformorethan90days

beforedialysisinitiation.Theother431subjects,whowereunder

theintervention of sub-specialistsalone suchasendocrinologists,

cardiologistsandnephrologistswithoutreferringtotheMDC,were

definedastheusualcaregroup(Fig.1).

Fig.1 Participantflowchart.

700

©2014TheAuthors.NephrologypublishedbyWileyPublishingAsiaPtyLtdonbehalfofAsianPacificSocietyofNephrolog y

patient management

Themultidisciplinarycare(MDC)consistedofanephrologist,

neph-rologynurseeducator,renaldietician,socialworker,pharmacy

spe-cialists,andsurgeonforvascularaccesscreation,tenchoffcathete

r

implantation.ForthestandardizedinterventionofCKDintheMDC

group,themanagementandeducationwasdependentonthe

dif-ferent stage of CKD and followed the NKF K/DOQI guidelines,

Taiwanpre-ESRD care programand reimbursementpolicy ofthe

NHI.CKDmanagementintheMDCgroupfocusedbothonmedical

management and lifestyle modification. The

case-management

nephrologynursecontactedpatientstoensureregularfollow-ups.

The members of the MDC team met with and followed up on

patients regularly to review and discuss patients’ individualized

therapy,medicalrecommendationsformetabolicabnormality,diet

assessmentandco-morbidity.

Criteria for dialysis initiation

CriteriafordialysisinitiationweremandatedbytheBNHIinTaiwan

including:

1Absolute criteria ofserum creatinine levels above10mg/dL or

creatinineclearanceoflessthan5mL/min.

2Relative criteria of serum creatinine levels above 6mg/dL or

creatinineclearanceoflessthan15mL/min,anduremicconditions

whichthreatenlifeorimpairqualityoflife(notlistedindetail).

Data collection and initial dialysis modality definition

Patients’basiccharacteristicswereprovidedbythefivecooperating

hospitals.WedirectlylinkedtotheNHIsystemandclaimdatato

confirm 1st _{dialysis medicalrecords and admission data to avoi}

d

incorrectdatacollection.IfpatientshadeverexperiencedPDduring

thefirst3monthsafterdialysis initiation,initialdialysismodalit

y

wasdefinedasPD.

Definitionof comorbidity diseases

Based on the definition of Dartmouth-Manitoba’s Charlson

ComorbidityIndex(D-M’sCCI),16_{weusedtheInternational}

Classi-fication ofDiseases (ICD-9)code on thein-hospitaldatabase and

double records in outpatient database to identify co-morbidities

withinoneyearbeforedialysisinitiation.Subsequently,anyinstanc

e

ofICD-9codeoncepresentedintheNHIdatabasewascountedas

co-morbidityandweightedintheCCIscore.

Service utilizations and medical costs,

clinical outcome

Service utilization and patients’ mortality were our clinical

outcomes.Datawerefollowedupuntil31December2009.Medical

serviceutilizationandcosts wereanalyzed from3monthsbefore

dialysisuntil6monthsafterdialysisinitiation(definedasthe

obser-vationperiod).Accidentinjuriesorsurgicalconditionsnotrelatedto

CKDwereexcluded.Medicalcostswerefurtheranalyzedinto

out-patientvisits,in-hospital,andemergencyservices.Serviceutilizatio

n

included frequency ofoutpatient visitsandhospitalization before

MDCimprovesoutcomesandreducescosts

andafterdialysisinitiation,percentageofhospitalizationatdialysis

initiation,andaveragelengthofstay(LOS).Serviceutilizationand

medicalcostduringobservationperiodwerefurtherdividedinto‘3

monthsbeforedialysis’,‘atdialysisinitiation’and‘6monthsafter

dialysisinitiation’.‘Atdialysisinitiation’wasdefinedastheperiodof

first time dialysis prescription (through hospitalization or

outpa-tient). Dialysis costs were excluded in our study (there was no

comparisonofcostofPDvshaemodialysis).

Statistical analysis

Frequency(n) and percentage (%) were used to determine the

distributionof gender,dialysis modality selection, andtemporary

catheteruse,hospitalizationatdialysisinitiation,frequencyof

hos-pitalization,co-morbidityandmortality.Meanandstandard

devia-tion(SD)wereusedtodescribethedistributionofage,hospitaldays,

andmedicalcostson outpatientvisits,in-hospitaland emergency

services.Wealsousedmediantodescribethedistributionofmedical

costs.Thedifferencesincategoricalvariableswereanalyzedby2test

(orFisher’sexacttest).Ageneralizedlinearmodel(GLM)wasused

toevaluatemedicalcostsbetweentheMDCand usualcaregroup

afteradjustingforgender,ageandCCIscore.TheTobitregression

modelwasusedto estimatetheadjusted medicalcosts associated

with the MDC and usual care group after controlling for other

covariates.TheKaplan–Meiersurvivalcurveandlog-rankstatistics

wereused todescribe hospitalization riskandpatientsurvival. A

multivariateCoxregressionmodelwasperformedtopredict

hospi-talizationriskandpatientmortalityafteradjustingforgender,age

andCCIscore.Atwo-tailedP-valuelessthan0.05wasconsidered

statisticallysignificant.Allstatisticalanalyseswereconductedusing

thestatisticalpackageforWindow,SAS9.2(SASInstitute,Cary,NC,

USA)andSPSS16.0(SPSS,Chicago,IL,USA).

RESULTS

Patient characteristics and co-morbidity

In total,822 subjects (usual care, n431; MDC, n391) withmeanageof62.8years(male,51.5%)wereincludedin ourstudy.Patient’scharacteristics,dialysismodality, percent-ageofhospitalizationatdialysisinitiationandco-morbidity areshowninTable1.TheMDCgrouphadahigher percent-ageofPDmodalityselection,lowertemporarycatheteruse, and a lower hospitalization rate at dialysis initiation (P0.05). The CCI scorein the MDCgroup was loweras comparedtotheusualcaregroup(P0.02).

Decline of renal function and eGFR at

dialysis initiation

Table2furtherillustratesthebaselineestimatedglomerular filtrationrate (eGFR) oneyear beforedialysis initiation,at dialysisinitiation,andthedeclineofrenalfunctionbasedon thevalidsubjectsprovidedbyfivecooperatinghospitals.We foundthepatientsoftheMDCgrouphadmorecompliance to outpatients follow-up and more detail laboratory data record(datanotshown).Thedeclineofrenalfunctionwas

©2014TheAuthors.NephrologypublishedbyWileyPublishingAsiaPtyLtdonbehalfofAsianPacificSocietyofNephrology 701

Y-RChenetal.

Table1 Generalcharacteristicofoursubjects DC

Us

ualcare

(n=391) (n=431)

P

P0.001)(Table3).The MDCgrouphada shorterLOSof hospitalization forboth ‘at dialysis initiation’ (15.5days vs 22.4days,P0.001)and‘6monthsafterdialysisinitiation’ (11.2daysvs16.0days,P0.05)(Table3).

Gender,male 210 53.7 213 49.4 0.22

Dialysischoice,PD† 120 30.7 67 15.6 0.001

Temporarycatheteruse 222 56.8 328 76.1 0.001

Initialdialysisadmission 0.05

In-hospital 327 83.6 369 85.6 Outpatient 42 10.7 28 6.5 Emergency 22 5.6 34 7.9 Hospitalizationrate‡ 347 88.7 402 93.3 0.03 CCIscore§ (2.01.6) (2.21.8) 0.02 Age (63.013.9) (62.615.5) 0.72

†Ifpatientshadtheexperienceoftheperitonealdialysis(PD)modelfro

m

dialysisinitiationtoaftera3monthsperiod,itwascalculated.‡Ifpatientsw ere

transferredfromoutpatientservicesoremergencyservicestohospitalizatio

n

atdialysisinitiation,itwascalculated.§Comorbiditydiseaseswerebasedo

n

Dartmouth-Manitoba’sCCI(D-M’sCCI).Valuesareexpressedas(meanSD).

CCI,CharlsonComorbidityindex;GFR,glomerularfiltrationrate;SD,standa

rd deviation.

slowerintheMDCgroup(7.6vs11.1mL/minper1.73m2_, P0.001),alsoeGFR atthedialysis initiationwas lowera s

compared to the usual care group (5.1 vs5.9mL/min per 1.73m2_,P

0.001).

Hospitalization risk and patient mortality

Duringanaveragefollow-uptimeof33.6months,atotalof 104 (24.1%) patients in the usual care and 67 (17.1% )

patients in theMDC groupdied (P0.01).The usualcare grouphadahigherhospitalizationrateduringthe observa-tionperiods(usualcare75.4%vsMDC69.3%,P0.05),an d

even6monthsafterdialysisinitiation(usualcare58.7%v s

MDC50.6%, P0.02).TheMDCgroup hadbetterpatient survival and lower risks of hospitalization in the Kaplan– Meier analysis (log-rank test, P0.05) (Figs2,3). After adjustingforgender,ageandCCI,theMDCgroupwasstill associatedwitha15%reductionintheriskofhospitalization (hazard ration [HR], 0.85; 95% confidence interval [CI] ,

0.72–0.99,P0.05).However,therewasnosignificant

dif-ference in patient mortality between the two groups (HR, 0.79;95%CI,0.58–1.08,P0.14).

Medical costs and frequencyof service utiliza

tion

The average medicalcostfor‘3months beforedialysis ini-tiation’, ‘at dialysis initiation’ and ‘6 months after dialysis initiation’wereNT$43329,NT$114161andNT$224624, respectively(datanotshown).ThemeanLOSof‘atdialysis

initiation’and‘6monthsafter dialysisinitiation’were19.1 daysand13.7days,respectively(datanotshown).

Medicalcostfortheusualcaregroup‘atdialysisinitiatio n’

was significantly higher (NT$ 146038 vs NT$ 7902 2 NT,

3 months before dialysis initiation

There was a higher frequency of outpatient visits (46.6 vs 38.1, P0.001) and higher average medical costs in the

MDC group (NT$ 26629 vs NT$ 20768, P0.001);

however,frequencyofhospitalization(0.8vs1.4,P0.001) andin-hospital(NT$12265vsNT$21519,P0.01), emer-gencycosts(NT$2119vsNT$3143,P0.05)werelowerin theMDCgroup(Table4).

At dialysisinitiation

Medicalcostsatdialysisinitiationweremainlyattributedto in-hospitalcosts.TheMDCgroupincurredlowerin-hospital costs after adjusting for gender, age and CCI score (NT$ 77896vsNT$144825,P0.05)(Table4).However,there

was no difference on outpatient and emergency costs

betweenthetwogroups.

6 months after dialysis initiation

Outpatient visits (21.4 vs 20.0, P0.05) and costs (NT$ 147976dollarsvsNT$126659,P0.01)werehigherinthe MDCgroup,whereasfrequencyofhospitalization(0.9vs1.1, P0.005) and in-hospital (NT$ 66840 vs NT$ 96902, P0.03) and emergency costs (NT$ 3867 vs NT$ 6453,

P0.002) were lower in the MDC group (Table4 and Fig.4).

Medical costs by Tobit regression

Age,CCIscore,temporarycatheteruseandhospitalizationat dialysisinitiationhadsignificantlypositiveeffectsonmedical costs (Table5). Adjusted medical costs were significantly lowerintheMDCgroupandwerereduceduptoNT$59251 foreachpatientduring3monthsbeforedialysisuntildialysis initiation(P0.001).

DISCUSSION

Severalstudieshaveevaluatedtheeffectivenessof compre-hensive, nephrology-based, multi-disciplinary care (MDC) andhave affirmed their substantialbenefits,such as bette r

laboratory parameters and clinical outcomes, slower renal functiondeclines,morefunctionalvascularaccessandshorte r

LOSatdialysisinitiation,andreductioninmedicalcostsand serviceutilization.13–15,17_{However,onlyfewstudieshave} sim-ultaneouslyfocusedonmedicalserviceutilization,costand clinicaloutcomesfortheMDConpre-ESRDpatients.17_From ourlarge-scale population and multi-hospital collaborative

702 ©2014TheAuthors.NephrologypublishedbyWileyPublishingAsiaPtyLtdonbehalfofAsianPacificSocietyofNephrolog y

MDCimprovesoutcomesandreducescosts

Table2 Baselineestimatedglomerularfiltrationrate(eGFR)(mL/minper1.73m2)andeGFRchange(eGFR)betweenmultidisciplinarycare(MDC)andusual care

group

MDC(n=391) Usualcare(n=431) P AdjustedP

# _MeanSD

n

# _MeanSD

BaselineeGFR1yearbeforedialysis 315 12.17.0 218 17.011.7 0.001 0.001

eGFRatdialysisinitiation 315 5.10.7 218 5.90.7 0.001 0.001

eGFR 315 7.06.3 218 11.111.0 0.001 0.001

eGFR/month 315 0.60.6 218 1.11.4 0.001 0.001

eGFR(estimatedglomerularfiltrationrate)asbaselineeGFR1yearbeforedialysis–eGFRatdialysisinitiation.AdjustPforgender,ageandCCIscore;e

GFR

providedbyfivecooperatinghospitalsbasedonvalidsubjects(somesubjectslosttofollow-upandinitiationrenalreplacementtherapy[RRT]outsidethe

study hospitals).

Fig.2 Kaplan–Meiercurveoftimetofirsthospitalizationbetweenmultidisciplinarycare(MDC)andusualcaregroupinthefollow-upperiod.

studythatwasdirectlylinkedtotheNHIsystemanddatabas e

in Taiwan,we confirmed the effectivenessof theMDC for pre-ESRDpatientsnotonlyforimprovingclinicaloutcomes, butalsoforreducingmedicalcosts.

Declines of renalfunction and characteristic

at

dialysis initiation

ConsensusguidelinesforCKDmanagement emphasizethe administrationofnephroprotectiveagents(RAASblockade), reductionofcardiovascularrisk,screeningandintervention forCKD-MBDandanaemia.Unfortunately,somephysicians may be reluctant to prescribe renoprotective medications such as RAAS blockage due to hyperkalaemia or haemo-dynamicmediatedincreaseinserumcreatininelevel.MDC

membranes could help and support additional effort to conductmedicationssideeffectsfollow-up.Dueto combina-tion of lifestyle modifications and more effective medical prescriptionaccordingtoK/DOQIandconsensusguidelines, theneGFRdeclinesintheMDCgroupwasslowerthanthe usual care group. Also in our study, there was no earlier initiationof dialysistherapy andCKDpatientsin theMDC groupinTaiwanweremorelikelytoinitiate dialysisrather thanfacemortality.18,19

Previous researchhas shownthat areduction in tempo-rary catheter useis associated with a lowerinfection rate, avoidanceof emergencydialysis, lower riskof hospitaliza-tion,andadecreaseinmortality,whichresultsinareduction of medical cost.20,21 _{Also the timing of referral and} pre-dialysiseducationhasbeenshowntoinfluencetheselection

©2014TheAuthors.NephrologypublishedbyWileyPublishingAsiaPtyLtdonbehalfofAsianPacificSocietyofNephrology ₇₀₃

Y-RChenetal.

Fig.3 Kaplan–Meiercurveofpatientsurvivalbetweenmultidisciplinarycare(MDC)andusualcaregroupinfollow-upperiod.

Table3 Medicalcosts†andhospitalutilizationinoursubjects

MDC(n=391) Usualcare(n=431) P

Mean(median)‡ SD Mean(median)‡ SD

Medicalcosts,NT

Atdialysisinitiation 79022(53302) 117118 146038(63327) 259431 0.001

After6months 218684(197092) 185792 230014(195463) 243206 0.46

Hospitaldays

Atdialysisinitiation 15.5 12.9 22.4 24.2 0.001

After6months 11.2 21.8 16.0 31.2 0.01

†Excludingdialysiscost;before3monthsandafter6months,costswerebasedontotalsubjects,whileatdialysisinitiation,theywerebasedonthenu

mberof

validsubjects.‡Medianwascalculatedbasedonthenumberofallthesubjects.MDC,multidisciplinarycare;SD,standarddeviation.

of dialysis modality and compliance to therapy prescrip-tion.22,23_{InTaiwan,theproportionofdialysispatients} under-going PD increased from 6.1 to 7.8% between 2001 and 2005, butthe proportionoptingforPD remainslow. MDC may helpCKD patients to receivewell-balanced presenta-tions ofallrenal replacementtherapyoptions.Educational intervention could increase in patient self-care ability and the provision of adequateandgood qualityinformationof dialysis,makingpatientswillingtoundergoPDandself-care dialysis.Ourstudydemonstratedthesameresultsasprevi

-ousstudiesinsofarastheMDCgrouphadlowertemporary catheter use(56.8% vs76.1%, P0.001), a lower rate of

emergencyvisitatdialysisinitiation(5.6%vs7.9%,P0.0 5)

anddoubletheoddsofselectingthePDmodality(30.7%vs 15.6%,P0.001).

Service utilizations and medical costs

Studieshaveindicatedthatresourceutilizationhasbecome morefrequentasESRDapproaches.6,24–26_Asubstantial pro-portionofthesharpincreaseinhospitalizationratesinthe3 monthsbeforeandaftertheinitiationofdialysiswas attrib-utedtovascularaccess andrelatedcomplications.25_Several otherfactorsalsoconfirmedtheincreasedriskof hospitali-zation, such as later referral, temporary catheter use, advancedage,loweralbuminandhematocritlevels,andan

704 ©2014TheAuthors.NephrologypublishedbyWileyPublishingAsiaPtyLtdonbehalfofAsianPacificSocietyofNephrolog y

MDCimprovesoutcomesandreducescosts

Table4 Medicalcosts†ofmedicalserviceadmissionduringtheobservationperiods

MDC(n=391) Usualcare(n=431) Adj.P*

Mean(median)‡ SD Mean(median)‡ SD

Before3months,NT Outpatient Visit 46.6(42.0) 25.5 38.1(34.0) 25.7 0.001 Cost 26629(24429) 17421 20768(17715) 18405 0.001 In-hospital Visit 0.8(0.0) 1.2 1.4(1.0) 1.9 0.001 Cost 12265(0) 35606 21519(0) 45634 0.005 Emergency Visit 1.4(1.0) 2.3 1.9(1.0) 3.0 0.52 Cost 2119(0) 4200 3143(0) 5964 0.03 Atdialysisinitiation,NT Outpatient Cost 764(0) 3906 499(0) 2941 0.39 In-hospital Cost 77896(52860) 117674 144825(62892) 259905 0.0001 Emergency Cost 362(0) 1751 714(0) 3173 0.07 After6months,NT Outpatient Visit 21.4(20.0) 11.5 20.0(18.0) 11.7 0.05 Cost 147976(157056) 112893 126659(74516) 118702 0.01 In-hospital Visit 0.9(1.0) 1.2 1.1(1.0) 1.5 0.005 Cost 66840(9129) 165036 96902(23859) 233118 0.03 Emergency Visit 1.0(0.0) 1.7 1.2(1.0) 1.9 0.06 Cost 3867(0) 8805 6453(311) 14456 0.002

*Adjustingforgender,ageandCharlsonComorbidityIndex(CCI)score.†Excludingdialysiscost;before3monthsandafter6months,costswerebasedontot

al

subjects,whileatdialysisinitiation,theywerebasedonthenumberofvalidsubjects.‡Medianwascalculatedbasedonthenumberofallthesubjects.SD,sta

ndard deviation.

increased number of co-morbidity.5,25,27,28 _{The MDC group} couldfacilitatetheearlycreationoffunctionaldialysisacces s

intheoutpatientsetting,whichobviatedtherequirementof temporarycatheter useanddecreased associated complica-tions, thus decreasing medical service utilization and expenditure.

Inourretrospectivecohortstudyfrom2005to2009based ontheNHIdatabase,wefoundthattheMDCgroupdidhave a significantly lower risk of hospitalization and LOS. The majormedicalcostswerein-hospitalcostsatdialysis initia-tion,and therewasa50% reduction ofmedicalcostsafter theMDCintervention(NT$ 77896 vsNT$144825). Also, CCI score, temporary catheter use and hospitalization at dialysis initiation are major independent risk factors for medicalexpenditure.AftertheTobitmodelanalysis adjust-ment,the estimatedmedicalcostsreduction foreachMDC patientduring3monthsbeforeuntildialysisinitiationcould beuptoNT$59251.Inlight ofthe 8000incidentdialysis patients who allreceived carefromthe MDC in2009, we couldestimatethatsavedmedicalexpenditurecanreach47 5

millionNTdollarseachyear,whichaccountsfor1.4‰ofthe 400billionNT-dollarfiscalbudgetsfortheNHI.29

Clinical outcome

Nephrology care6months before dialysis and the consist-encyofcarearestrongpredictorsofmortality.30,31_MDCcould provide pre-ESRD patients with enhanced disease knowl-edgeandawareness,amorepositiveattitude,betterlifestyle modificationswithcardiovascularriskreduction,and more effectivemedicalprescriptionaccordingtoK/DOQIand con-sensusguidelines;allofthesemaycontributetoadecreasein all-causesmortality.

As expected, our finding confirmed the effectiveness of MDCindecreasingtheriskofhospitalization,butindicated no significantly better survival benefits. We speculatethat ESRDpatientswhodiedwithin90daysafterdialysis initia-tionwereexcludedfromourstudy. Also duetothe acces-sibility of dialysis therapy in Taiwan, pre-ESRD patients underMDCinterventionaremostlikelytoinitiatethe dialy-siselectivelyandexperienceoptimaldialysistherapybefore reachingmortality.

Therearestillsomelimitationsinourstudy.First,dueto our study design, we could not provide detailed cause of hospitalizationandmortality(garbleddata providedbythe

©2014TheAuthors.NephrologypublishedbyWileyPublishingAsiaPtyLtdonbehalfofAsianPacificSocietyofNephrology 705

Y-RChenetal.

Fig.4 Medicalcosts(excludingdialysiscosts)ofmultidisciplinarycare(MDC)andusualcaregroupinobservationperiods.Before3monthsandafter6

months

werebasedonthetotalnumberofsubjects.Atdialysisinitiation,costswerebasedonthenumberofvalidsubjects(adjustingforgender,ageandCCIsco

re).

outpatient;

,in-hospital;■,emergency.

Table5 Tobitmodelanalysisonadjustedmedicalcosts†:before3monthstoinitialdialysisperiods

Total(n=822) MDC(n=391) Usualcare(n=431)

 95%CI P  95%CI P  95%CI P

Low high Low High Low High

Intercept −81501 −177274 −4273 0.10 −53235 −125152 18682 0.15 −177367 −351407 −3327 0.05

MDCversusnon-MDC −59251 −91477 −27025 0.001 — — — — — — — —

Maleversusfemale 1299 −30303 32902 0.94 −2198 −27510 23114 0.86 5457 −49868 60783 0.85

CCIscore 28798 19255 38342

0.001 18799 10892 26706 0.001 36011 19743 52279 0.001

Temporarycatheteruse‡ 64479 28752

100206

0.001 65664 39294 92035 0.001 67096 −1453 135646 0.06

Hospitalizationrate‡ 76068 17858

134278 0.01 45657 4428 86885 0.03 124185 7024

241347 0.04

Scale 229560 218727

240929 125821 117304

134956 291996 273139

312154

†Excludingdialysiscost.‡Eventswereatdialysisinitiation.CCI,CharlsonComorbidityIndex;MDC,multidisciplinarycare.

NHI). Also this was a retrospective observational study of secondary data analysis, so we could not provide detailed biochemistry differences between the two groups. These limited items used as clinical parameters and incomplete informationduetoprimarydataerrorswereneitheravoided noroverlooked.Second,weonlyanalyzedmedicalcosts,and assuch,couldnotevaluatedifferencesinpatients’qualityof life,societyresourceutilizationandexpenditure.Third,the fixedreimbursementpolicyoftheNHIrestrainedlaboratory parameter measurement and medical utilization (such a s

haemoglobin correction by the ESA administration). Furthermore, theNHI systemin Taiwanis auniquehealth insurancecoveragesystem,andourresultsmaynotapplyt o

othercountriesthatconductsimilaranalyses.TheBureauof

Health Promotionin Taiwan has launched pre-ESRD care initiativessince2002.TheMDCprograminTaiwanproposed astandardcareprotocolandannualreportingsystem.Since 2008theTaiwanSocietyofNephrologycommitteealso con-ducted a surveillance program and improvement of the qualityofpre-ESRDcare.ThepropagationoftheMDCmight partially explain the stabilization of increase in incident ESRDpatientsinTaiwan.

CONCLUSION

Pre-ESRDpatients underthe MDCintervention,especially forthoseat3monthspriortoinitiatingdialysis,hadahigher percentage of functional vascular access and better

preva-706 ©2014TheAuthors.NephrologypublishedbyWileyPublishingAsiaPtyLtdonbehalfofAsianPacificSocietyofNephrolog y

lenceinPDselection.Moreover,bothLOSatdialysis initia-tionaswellastheriskofhospitalizationwasreduced.Finally ,

ourstudyconfirmed the effectivenessof MDCnot onlyin improving clinical outcome but also in reducing medical expenditure.

ACKNOWLEDGEMENTS

Thisstudywassupportedbythe Bureau ofHealth Promo-tion, Department of Health, R.O.C. (97-HP-1103, 98-HP-1112,99-HP-1108,100-HP-1104).YuYangwasresponsible for the study conception and execution. Chiou, Chang, Chen,LinandFerngwereresponsibleforplancoordination andcooperation.Wangwasresponsiblefordataanalysisand the drafting of the manuscript. Chou assisted with imple-menting the study and data collection. We thank the cooperationofthehospitalCKDnurses,e.g.Jui-HsinChen, Shu-ChiLu,Jay-JenLin,Ya-HsuehShih,Chih-YingHuang inthecollectionofthesubjects’data.

REFERENCES

1.WenCP,ChengTYD,TsaiMKetal.All-causemortalityattributable

tochronickidneydisease:Aprospectivecohortstudybasedon

462293adultsinTaiwan.Lancet2008;371:2173–82.

2.HsuC-C,HwangS-J,WenC-Petal.Highprevalenceandlow

awarenessofCKDinTaiwan:Astudyontherelationshipbetween

serumcreatinineandawarenessfromanationallyrepresentative

survey.Am.J.KidneyDis.2006;48:727–38.

3.HwangS-J,LinM-Y,ChenH-Cetal.Increasedriskofmortalityin

theelderlypopulationwithlate-stagechronickidneydisease:A

cohortstudyinTaiwan.Nephrol.Dial.Transplant.2008;23:3192–8.

4.DepartmentofHealth.NationalHealthInsuranceAnnualStatistical

Report1996.DepartmentofHealth,2004:Taiwan,2004.[Cited25

August2014.]AvailablefromURL:http://www.nhi.gov.tw

5.AroraP,KauszAT,ObradorGTetal.Hospitalutilizationamong

chronicdialysispatients.J.Am.Soc.Nephrol.2000;11:740–46.

6.StPeterWL,KhanSS,EbbenJP,PereiraBJG,CollinsAJ.Chronic

kidneydisease:Thedistributionofhealthcaredollars.KidneyInt.

2004;66:313–21.

7. KeithDS,NicholsGA,GullionCM,BrownJB,SmithDH.

Longitudinalfollow-upandoutcomesamongapopulationwith

chronickidneydiseaseinalargemanagedcareorganization.Arch.

Intern.Med.2004;164:659–63.

8. GullionCM,KeithDS,NicholsGA,SmithDH.Impactof

comorbiditiesonmortalityinmanagedcarepatientswithCKD.

Am.J.KidneyDis.2006;48:212–20.

9. NationalKidneyF.K/DOQIclinicalpracticeguidelinesforchronic

kidneydisease:Evaluation,classification,andstratification.Am.J.

KidneyDis.2002;39:S1–266.

10.LevinA,StevensPE.EarlydetectionofCKD:Thebenefits,

limitationsandeffectsonprognosis.Nat.Rev.Nephrol.2011;7:

446–57.

11.SmartNA,TitusTT.Outcomesofearlyversuslatenephrology

referralinchronickidneydisease:Asystematicreview.Am.J.Med.

2011;124:1073–1080.e1072.

12.MasonJ,KhuntiK,StoneM,FarooqiA,CarrS.Educational

interventionsinkidneydiseasecare:Asystematicreviewof

MDCimprovesoutcomesandreducescosts

13. HemmelgarnBR,MannsBJ,ZhangJetal.Associationbetween

multidisciplinarycareandsurvivalforelderlypatientswithchronic

kidneydisease.J.Am.Soc.Nephrol.2007;18:993–9.

14. WuIW,WangS-Y,HsuK-Hetal.Multidisciplinarypredialysis

educationdecreasestheincidenceofdialysisandreducesmortality

–acontrolledcohortstudybasedontheNKF/DOQIguidelines.

Nephrol.Dial.Transplant.2009;24:3426–33.

15. CurtisBM,RavaniP,MalbertiFetal.Theshort-andlong-term

impactofmulti-disciplinaryclinicsinadditiontostandard

nephrologycareonpatientoutcomes.Nephrol.Dial.Transplant.

2005;20:147–54.

16. RomanoPS,RoosLL,JollisJG.Presentationadaptingaclinical

comorbidityindexforusewithICD-9-CMadministrativedata:

Differingperspectives.J.Clin.Epidemiol.1993;46:1075–9.

17. WeiS-Y,ChangY-Y,MauL-Wetal.Chronickidneydiseasecare

programimprovesqualityofpre-end-stagerenaldiseasecareand

reducesmedicalcosts.Nephrology2010;15:108–15.

18. ChenYR,YangY,WangSCetal.Effectivenessofmultidisciplinary

careforchronickidneydiseaseinTaiwan:A3-yearprospective

cohortstudy.Nephrol.Dial.Transplant.2013;28:671–82.

19. HwangS-J,YangW-C,LinM-Y,MauL-W,ChenH-C,Taiwan

SocietyofN.Impactoftheclinicalconditionsatdialysisinitiation

onmortalityinincidenthaemodialysispatients:Anationalcohort

studyinTaiwan.Nephrol.Dial.Transplant.2010;25:2616–24.

20. PisoniRL,ArringtonCJ,AlbertJMetal.Facilityhemodialysis

vascularaccessuseandmortalityincountriesparticipatingin

DOPPS:Aninstrumentalvariableanalysis.Am.J.KidneyDis.2009;

53:475–91.

21. PerlJ,WaldR,McFarlanePetal.Hemodialysisvascularaccess

modifiestheassociationbetweendialysismodalityandsurvival.

J.Am.Soc.Nephrol.2011;22:1113–21.

22.GoovaertsT,JadoulM,GoffinE.Influenceofapre-dialysis

educationprogramme(PDEP)onthemodeofrenalreplacement

therapy.Nephrol.Dial.Transplant.2005;20:1842–7.

23.MannsBJ,TaubK,VanderstraetenCetal.Theimpactofeducation

onchronickidneydiseasepatients’planstoinitiatedialysiswith

self-caredialysis:Arandomizedtrial.KidneyInt.2005;68:

1777–83.

24.SmithDH,GullionCM,NicholsG,KeithDS,BrownJB.Costof

medicalcareforchronickidneydiseaseandcomorbidityamong

enrolleesinalargeHMOpopulation.J.Am.Soc.Nephrol.2004;15:

1300–6.

25.MixTCH,StPeterWL,EbbenJetal.Hospitalizationduring

advancingchronickidneydisease.Am.J.KidneyDis.2003;42:

972–81.

26.HunsickerLG.Theconsequencesandcostsofchronickidney

diseasebeforeESRD.J.Am.Soc.Nephrol.2004;15:1363–4.

27.BessetteRW,CarterRL.PredictinghospitalcostinCKDpatients

throughbloodchemistryvalues.BMCNephrol.2011;12:65.

28.LaliberteF,BookhartBK,VekemanFetal.Directall-causehealth

carecostsassociatedwithchronickidneydiseaseinpatientswith

diabetesandhypertension:Amanagedcareperspective.J.Manag.

CarePharm.2009;15:312–22.

29.WuM-S,WuI-W,ShihC-P,HsuK-H.Establishingaplatformfor

battlingend-stagerenaldiseaseandcontinuingquality

improvementindialysistherapyinTaiwan.ActaNephrol.2011;25:

148–53.

30.DoganE,ErkocR,SayarliogluH,DurmusA,TopalC.Effectsof

latereferraltoanephrologistinpatientswithchronicrenalfailure.

Nephrology2005;10:516–19.

31.HuismanRM.Thedeadlyriskoflatereferral.Nephrol.Dial.

Transplant.2004;19:2175–80.