Therapyswitchinginpatientsreceivinglong-Actingopioids

Analgesia

TherapySwitchinginPatientsReceivingLong-ActingOpioids

ArielBerger,DeborahLHoffman,SethGoodman,ThomasEDelea,RaafatSeifeldin,andGerryOster

BACKGROUND:Patternsoftherapyswitchinginpatientsreceivinglong-actingopioidshavenotbeenwelldocumented.

OBJECTIVE:Tocomparetherapyswitchingamongpatientsbeginningtreatmentwithcontrolled-release(CR)oxycodone,transdermalfentanyl,orCRmorphinesulfate.

METHODS:Using aUS healthcare claims database, we identified patients beginning treatment with CR oxycodone, transdermalfentanyl,orCRmorphinesulfatebetweenJuly1,1998,andDecember31,1999.Wecompiledclaimsforeachpatientfor6monthsfollowingtherapyinitiationandcomparedtheincidenceoftherapyswitchingamongthe3groups.Wealsoestimatedtotalhealthcarechargesforpatientswhoswitchedtherapyversusthosewhodidnot.

RESULTS:Weidentified1931,668,and449patientsbeginningtherapywithCRoxycodone,transdermalfentanyl,andCRmorphinesulfate, respectively; 16.7%, 25.0%, and 35.9%, respectively,hadcancer.For patients without cancer, rates of therapy switching at6 months were 10.6% (CR oxycodone), 19.0% (transdermal fentanyl), and 26.0% (CR morphine sulfate); for those with cancer,rates were 23.8%, 24.6%, and 29.8%, respectively. Multivariate hazard ratios (vs CR morphine sulfate) for therapy switching inpatients without cancer were 0.36 (95% CI, 0.27 to 0.47) for CR oxycodone and 0.69 (0.51 to 0.94) for transdermal fentanyl; forthose with cancer, corresponding hazard ratios were 0.72 (0.50 to 1.03) and 0.76 (0.50 to 1.16). Total healthcare charges weresignificantly (p 0.01) higher for patients who switched therapy than those whodidnot($23965vs$14299inpts. without cancer;

$58259vs$39618forthosewithcancer).

CONCLUSIONS:Patients without cancer who receive CR oxycodone or transdermal fentanyl are less likely to switch therapy thanthosereceivingCRmorphinesulfate.Totalhealthcarechargesarehigherforpatientswhoswitchtherapy.

KEYWORDS:healthcarecosts,opiodanalgesics,pharmacotherapy..

AnnPharmacother2004;38:389-95.

PublishedOnline,12Jan2004,,DOI10.1345/aph.1D109

ong-actingopioids(eg,controlled-release[CR]oxy-codone,transdermalfentanyl,CRmorphinesulfate)are

acorecomponentoftreatmentforpersistentmoderatetose-verepain.1-5Althoughlong-actingopioidsaregenerallywelltolerated,theincidenceand/orseverityofadverseeffects—particularly,itching,6hallucinations,6constipation,7,8andnausea7—maydifferbetweenagents.Thesedifferencesmaybeclinicallyimportant,astheymaygiverisetodifferencesintolerabilityandpersistencywithprescribedtreatment.

Thetimeneededtotitratetherapyandtoachieveadesiredanalgesiceffectalsomaydifferbetweenlong-actingopioids.2Ithasbeen suggested, for example, thatthe initial dose oftransdermalfentanylmaybetoolow,leadingtogreaternum-bersofpatientsrequiringtitration,9,10difficultiesintitration,11and/ormorefrequentdosingthantherecommended3-day

Authorinformationprovidedattheendofthetext.

Funding for this research was provided by Purdue Pharma, L.P.,Stamford,CT.

intervalbetweenapplications.12,13Consistentwiththeseob-servations,patientsbeginningtreatmentwithtransdermalfentanylmayhavehighlevelsofuseofrescuemedication.14Problemswithdosetitrationmaywellleadtotherapyswitch-ingasaconsequenceofundertreatmentofpain.

These potential differences between long-acting opioidsmayhave important economic implications. Patients whoswitchtherapyduetoadverseeffectsand/orproblemswithdose titration may incur higher costs dueto additional of-fice visits andthe discarding of medication. Patients withinadequately managed painalsomayhave higher costs asa result oftheneedfor additional pharmacotherapy and/orhealthcareservices(eg,hospitalizations).

The purpose ofthisstudywasto examine ratesof thera-py switching in clinical practice among patients beginningtreatmentwithCRoxycodone,transdermalfentanyl,orCR morphine sulfate, using information from alarge UShealthcare claims database. We also compared healthcarecharges for patients who switched long-acting opioid ther-apyversusthosewhodidnot.

Methods

OVERVIEW

Using a large US health insurance claims database, we identified allpatients who began therapy with CR oxycodone, transdermal fentanyl,orCR morphine sulfate between July 1, 1998, and December 31, 1999.Thedateofeachpatient’sinitialreceiptofoneoftheseagentswasdesig-natedtheirindexdate,andpatientswerecategorizedintotreatmentgroups according tothe agent they received on this date. We then com-piledallmedicalandpharmacyclaimsforeachpatientoverthe6-monthperiodfollowingtherapyinitiationandcomparedtheincidenceoftherapyswitchingamongpatientsreceivingtheagents.Wealsocomparedtotalhealthcarechargesforpatientswhoswitchedlong-actingopioidtherapyversusthosewhodidnot.Patientswerestratifiedinallanalysesaccordingtowhetherornottheyhadcancer.

DATABASE

Data for this study were obtained from the Protocare Sciences Man-agedCareDatabase.Thedatabaseiscomprisedofpaidinstitutional,provider,and outpatient pharmacy claims derived from a variety of pri-vate healthcare plans. The database contains healthcare claims and en-rollment data for approximately 3 million persons annually residing inover20 states. Approximately one-quarter of persons inthe database areaged65years.

The database includes patient demographic, eligibility,and vital sta-tus information, inpatient and outpatient diagnoses (inICD-9-CM for-mat15), inpatient and outpatient procedure information (in ICD-9-CM,Physician’sCurrentProceduralTerminology,4thed.[CPT-416],andHealth Care Financing Administration Common Procedure Coding Sys-tem [HCPCS] formats), outpatient drugs (identified by National DrugCode[NDC])andassociatedtherapy-daysdispensed,billedcharges,anddatesofservicefordrugsandmedicalservices.

SAMPLESELECTION

Outpatientpharmacyclaimswerescannedtoidentifyallpersonswhoreceived CR oxycodone, transdermal fentanyl, orCR morphine sulfateduring the study period. The date of each patient’s initial receipt of oneof these agents during this period was designated their index date, andpatients were classified into treatment groups according tothe particularagenttheyreceivedonthisdate.

Patientswereexcludediftheyhad(1)receivedCRoxycodone,trans-dermal fentanyl, orCR morphine sulfate during the6-month period pre-ceding their index date (pretreatment), (2) received multiple long-actingopioids on their index date, (3) received <30 days of long-acting opioidtherapy during the 6-month period following their index date (follow-up), (4) were not continuously eligible for health and drug benefits dur-ing pretreatment and follow-up, or(5)were enrolled ina Medicare sup-plemental or capitated health plan (the database does not include com-pleteclaimshistoriesforthesepersons).

STUDYMEASURES

Attentionwasfocusedontheoccurrenceoftherapyswitchingamongpatients using CR oxycodone, transdermal fentanyl, and CR morphinesulfate. Therapy switching was defined as receipt during follow-up ofalong-actingopioidotherthantheonereceivedontheindexdate.Wealsoexamined total medical care charges for patients who switched long-act-ingopioidtherapyversusthosewhodidnot.

ANALYSES

Since patternsof therapy switching may differ for patients with can-cerversusthosewithout,westratifiedthestudysampleaccordingto whether patients had cancer (except squamous orbasalcellskincarcino- ma,as these rarely cause significant pain) (ICD-9-CMdiagnosis codes 140.XX–172.XX, 174.XX–208.XX). Patients with 2medical claims for cancer during pretreatment or1such claims 14days prior to their indexdateweredeemedtohavecancer.

We assessed the comparability ofCR oxycodone, transdermal fen-tanyl, andCR morphine sulfate patients at baseline. We then comparedthe prevalence of selected pain-related conditions based onthe presenceof2paid medical claims during pretreatment or1such claims withinthe14-dayperiodpriortotheindexdate,withdiagnosesshowninTable

1.Wealsocomparedtheprevalenceofotherchroniccomorbiditiesbased on the presence of 2 paid medical claims during pretreatmentwith diagnoses shown inTable1. Patient characteristics were comparedacross treatment groups using ANOVAfor continuous variables andtestsforcategoricalvariables.

Timetoswitchingwasassessedintermsofthenumberofdaysbetweentheindexdateandthedateoffirstreceiptofanotherlong-actingopioid. Wefocusedattentiononthefirsttherapyswitchonly.Estimatesoftimetoswitchandthecumulativepercentageofpatientsswitchinginthe3groupsweregeneratedusingKaplan–Meiertechniques17;thecorrectedgroup-

Table1.ICD-9-CMDiagnosticCodesUsedinDeterminingPrevalenceofPain-RelatedandComorbidConditions
ConditionsICD-9-CMDiagnosticCodes
Pain-relatedconditions
fibromyalgia729.1X
lowbackpain720.0X–720.2X,721.3X,721.42,722.0X,722.32,722.52,722.73,722.83,722.93,724.02,724.2X, 724.3X,724.6X,724.7X
otherspinalpain(excludinglowback720.XX–724.XX(excludingcodesforlowbackpain;seeabove) painandincludingneck)
osteoarthritis715.XX
othermusculoskeletalpain710.XX–714.XX,716.XX–719.XX,725.XX–729.XX(excluding729.1X) neuropathicpain53.1X,250.6X,350.XX,352.1X,353.XX–357.XX,729.2X
Comorbiddiseases
chronicrenalfailureICD-9-CM585.XX
congestiveheartfailure428.0X
chronicobstructivepulmonarydisease491.XX,492.XX,496.XX coronaryheartdisease410.XX–414.XX
depression296.2X–296.8X,298.0X,300.4X,311.XX
diabetes250.XX
ICD-9-CM=InternationalClassificationofDisease,9thed.

prognosismethod18wasemployedtoadjusttheseestimatesfordifferencesinselectedcovariates,includingage(65yvsyounger),gender,thepres-enceofselectedchronicpain-relatedconditionsandotherchroniccomor-bidities,andpretreatmenthealthcarecharges(medianvs<median).

The probability of therapy switching was assessed using multivariateCoxproportionalhazardsmodels.17Covariatesenteredintothesemodelswereasnotedabove.Patientsweredeemedtobeatriskoftherapyswitching from their index date until either thedateof receipt of anotherlong-actingopioidortheendoffollow-up,whichevercamefirst.

To examine the impact of therapy switching on healthcare charges,we pooled patients across the 3 treatment groups and restratified themaccording to whether they switched therapy during follow-up. Chargesof patients who switched therapy werethen compared with those of pa-tients whohadnot.ANCOVAwasusedto assess the significance of dif-ferences inmean total healthcare charges adjusting for differences in se-lectedcovariatesbetweenthe2groups.

Continuous variables are reported as mean±SD, while categoricalvariables are reported as proportions. All analyses wereperformed usingPC-SASv.8.0.19

Results

PATIENTCHARACTERISTICS

A total of 1931, 668, and 449 patients were identifiedwho began therapy withCR oxycodone, transdermal fen-tanyl,andCRmorphinesulfate,respectively;16.7%,25.0%,and35.9%,respectively,hadcancer.Amongthose withoutcancer,CR oxycodone patients were younger than trans-dermalfentanylorCRmorphinesulfatepatients(p0.01)(Table2). Transdermal fentanyl patients were most likelytobe women. Pretreatment total healthcare charges were

lowestforCR oxycodone patients; the prevalence of mostchroniccomorbidandpain-relatedconditionsalsowaslowest for CR oxycodone patients. Among patients withcancer, there wasnosignificantdifferenceineitherageorgender.Pretreatmenttotalhealthcarecharges,however,weresignificantlyhigheramongtransdermalfentanylpatientsver-susthosereceivingCRoxycodoneorCRmorphinesulfate.

INCIDENCEOFTHERAPYSWITCHING

Among patients without cancer,ratesof therapy switch-ingat6monthswere10.6%forCR oxycodone, 19.0% fortransdermal fentanyl, and26.0%forCR morphine sulfate.Adjusted for covariates usingthe corrected group-progno-sis method, similar results were obtained (CR oxycodone10.5%; transdermal fentanyl 19.6%; CR morphine sulfate26.6%).Mediantimetoswitchingwas63daysforCRoxycodone, 57daysfortransdermal fentanyl, and44 daysforCRmorphinesulfate.

Among patients withcancer, rates of therapy switchingwere23.8%forCRoxycodone,24.6%fortransdermalfentanyl,and29.8%forCRmorphinesulfate.Onanad-justedbasis,similarresultswereobtained(CRoxycodone23.6%; transdermal fentanyl 24.2%; CR morphine sulfate31.1%).Mediantimetoswitchingwas77daysforCRoxycodone, 52daysfortransdermal fentanyl, and26 daysforCRmorphinesulfate.

Table2.CharacteristicsofPatientsReceivingLong-ActingOpioidsbyPresenceofCancer
Variable / CR
Oxycodone(N=1608) / WithoutCancer
CR
Transdermal MorphineFentanylSulfate(N=501)(N=288) / p Value / CR
Oxycodone(N=323) / WithCancer
CR
TransdermalMorphineFentanylSulfate
(N=167)(N=161) / p Value
Age(y),mean±SD / 57.4±16.9 / 64.7±17.7 / 60.3±16.5 / <0.01 / 64.4±12.3 / 64.3±13.7 / 65.3±12.1 / 0.68
Female(%) / 59.0 / 69.3 / 53.8 / <0.01 / 47.4 / 53.9 / 47.8 / 0.36
Chroniccomorbid
condition,nofpts.(%)
chronicrenalfailure / 15(0.9) / 9(1.8) / 7(2.4) / 0.06 / 6(1.9) / 2(1.2) / 2(1.2) / 0.80
congestiveheartfailure / 71(4.4) / 45(9.0) / 28(9.7) / <0.01 / 15(4.6) / 10(6.0) / 7(4.3) / 0.75
COPD / 125(7.8) / 54(10.8) / 34 (11.8) / 0.02 / 46(14.2) / 31(18.6) / 25(15.5) / 0.46
coronaryheartdisease / 132(8.2) / 69(13.8) / 31(10.8) / <0.01 / 33(10.2) / 17(10.2) / 8(5.0) / 0.13
depression / 96(6.0) / 32(6.4) / 15(5.2) / 0.80 / 9(2.8) / 5(3.0) / 4(2.5) / 0.96
diabetes / 169(10.5) / 80(16.0) / 43(14.9) / <0.01 / 55(17.0) / 29(17.4) / 25(15.5) / 0.89
Chronicpain-related
conditions,nofpts.(%)
fibromyalgia / 138(8.6) / 23(4.6) / 19(6.6) / 0.01 / 10(3.1) / 4(2.4) / 1(0.6) / 0.23
lowbackpain / 583(36.3) / 158(31.5) / 95(33.0) / 0.12 / 63(19.5) / 23(13.8) / 41(25.5) / 0.03
neuropathicpain / 185 (11.5) / 53(10.6) / 45(15.6) / 0.09 / 19(5.9) / 7(4.2) / 8(5.0) / 0.72
osteoarthritis / 263(16.4) / 81(16.2) / 49(17.0) / 0.95 / 22(6.8) / 8(4.8) / 15(9.3) / 0.27
othermusculoskeletal / 609(37.9) / 194(38.7) / 114(39.6) / 0.83 / 98(30.3) / 46(27.5) / 45(28.0) / 0.76
pain
otherspinalpain / 622(38.7) / 148(29.5) / 113(39.2) / <0.01 / 79(24.5) / 37(22.2) / 31(19.3) / 0.43
Pretreatmenthealthcare / 11445 / 18426 / 18667 / <0.01 / 33657 / 46089 / 34968 / <0.01
costs($),mean±SD / ±24607 / ±36894 / ±48117 / ±38817 / ±51362 / ±38836
COPD=chronicobstructivepulmonarydisease;CR=controlledrelease.

Hazardratios obtained usingaCox multivariate propor-tionalhazardsmodelareshownin Table 3.

THERAPYSWITCHINGANDHEALTHCARECHARGES

Among the 2397 patients without cancer,341 (14.2%)switchedlong-actingopioidtherapyduringfollow-up.Thosewho switched therapy weremore likely tobe wom-enandtohavelowbackpainand other spinal pain(Table4). Adjusted mean total healthcare charges during follow-upwere $9666 higher among those who switched therapycompared withthosewhodidnot(Table5). Inpatient careaccounted for 59.1% of the total charges among patientswhoswitchedtherapyand54.5%amongthosewhodidnot.Chargesforpain-relatedpharmacotherapyweregreateramong patients who switched long-acting opioid therapycomparedwiththosewhodidnot.

Among651patientswith cancer, 25.5%switchedthera-pyduringfollow-up.Patientswhoswitchedtherapyweremorelikelytohaveotherspinalpaincomparedwiththosewhodidnotswitch.Adjustedmeantotalhealthcarechargesduringfollow-upwere$18641higheramongthosewhoswitchedtherapycomparedwiththosewhodidnot.Inpa-tientcarecomprised48.2%oftotalchargesamongpatientswhoswitchedtherapyand42.2%amongthosewhodidnot.Chargesforpain-relatedpharmacotherapyweregreateramongpatientswhoswitchedlong-actingopioidtherapy.

Discussion

Our findings indicate that patients beginning treatmentwithCR oxycodone or transdermal fentanyl areless likelyto switch therapy thanthose treated initially withCR mor-phine sulfate. In multivariate analyses controlling for dif-ferences in baseline characteristics, patients without cancerwhowerereceivingCRoxycodoneortransdermalfen-tanylwerefoundtobe64%and31%,respectively,lesslikelytoswitchtherapythanCRmorphinesulfatepatients.Amongpatientswithcancer,thosereceivingCRoxy-codone or transdermal fentanyl were approximately 25%less likely to switch therapy, although this result did notachievestatisticalsignificance.

Similar to research on therapy switching in other thera-peuticareas,20,21wealsofoundthatpatientswhoswitchlong-actingopioidtherapyhavesignificantlyhigherhealthcarechargesthanthosewhodonot.Thisfindingpersists after adjustment for covariates suchasage, preex-istingdiagnoses,andpretreatmenthealthcarecharges.Chargesforinpatientandoutpatientcare,aswellasforpain-relatedmedications,weresignificantlyhigherforcan-cer and noncancer patients who switched therapy. Whilewe cannot establish causality with our study design, ourfindings do raise the interesting possibility that better painmanagement mayleadto reduced rates of therapy switch-ingandlowerhealthcarecosts.

Table3.MultivariateCoxProportionalHazardsAnalysisofFactorsAssociatedwithLong-ActingOpioidSwitchingbyPresenceofCancer
Variable / HR / WithoutCancer95%CI / p Value / HR / WithCancer95%CI / p Value
IndexLAOvsCRmorphinesulfate
controlled-releaseoxycodone / 0.36 / 0.27to0.47 / <0.01 / 0.72 / 0.50to1.03 / 0.07
transdermalfentanyl / 0.69 / 0.51to0.94 / 0.02 / 0.76 / 0.50to1.16 / 0.20
Age(65vs<65y) / 0.85 / 0.68to1.08 / 0.18 / 0.85 / 0.62to1.17 / 0.32
Male(vsfemale) / 0.81 / 0.65to1.02 / 0.81 / 0.80 / 0.59to1.09 / 0.16
Chroniccomorbidcondition
chronicrenalfailure / 1.47 / 0.68to3.17 / 0.33 / 0.74 / 0.18to3.09 / 0.68
congestiveheartfailure / 0.87 / 0.54to1.39 / 0.87 / 1.21 / 0.56to2.60 / 0.62
COPD / 1.21 / 0.85to1.74 / 0.29 / 0.83 / 0.52to1.32 / 0.43
coronaryheartdisease / 0.80 / 0.54to1.19 / 0.27 / 0.96 / 0.52to1.76 / 0.89
depression / 1.10 / 0.73to1.65 / 0.66 / 0.60 / 0.19to1.90 / 0.38
diabetes / 0.91 / 0.65to1.27 / 0.56 / 1.29 / 0.86to1.93 / 0.22
Chronicpain-relatedconditions
fibromyalgia / 1.07 / 0.72to1.58 / 0.74 / 1.14 / 0.41to3.14 / 0.81
lowbackpain / 1.33 / 1.03to1.70 / 0.03 / 0.97 / 0.63to1.50 / 0.89
neuropathicpain / 1.04 / 0.76to1.44 / 0.79 / 1.02 / 0.51to2.04 / 0.96
osteoarthritis / 1.00 / 0.74to1.34 / 1.00 / 0.94 / 0.50to1.77 / 0.85
othermusculoskeletalpain / 1.10 / 0.87to1.38 / 0.43 / 0.75 / 0.52to1.09 / 0.13
otherspinalpain / 1.13 / 0.88to1.45 / 0.34 / 1.75 / 1.18to2.60 / 0.01
Pretreatmenthealthcarecharges
(medianvs<median) / 1.24 / 0.97to1.6 / 0.08 / 0.84 / 0.56to1.26 / 0.41
COPD=chronicobstructivepulmonarydisease;CR=controlledrelease;LAO=long-actingopioid.

Wenote several limitations ofourstudy. First, since ourstudywasbasedonaretrospectiveanalysisofhealthinsur-ance claims data,wedonotknowwhy patients were start-edonaparticularlong-actingopioidversusanotherorwhythey switched therapy. While the latter may reflect differ-ences in adverse effects,tolerability,and dose titration ashypothesized,italsomaybeduetodifferencesinthechar-acteristicsofpatientswhoreceivedthesedrugsand/orthose ofthe providers who prescribed them (ie, itmay bethe result of confounding). While we attempted to control

for potential confounders in multivariate analyses, limita-tions inherent in claims data no doubt precluded us fromcontrollingcompletelyforsuchfactors.Forexample,transdermalfentanylandCRmorphinesulfatemaybepreferentiallyprescribedtopatientsinthelater—andhencemore painful—stages of their disease. Additionally, pre-treatment healthcare charges were higher among patientsreceiving transdermal fentanyl and CR morphine sulfatethanthoseonCRoxycodone,suggestingagreaterneedforhealthcare services amongtheformer groups. Transdermal

Table4.CharacteristicsofPatientsReceivingLong-ActingOpioidsSwitchingandNotSwitchingTherapybyPresenceofCancer
Variable / WithoutCancer
NotSwitchingSwitching
(N=2,056)(N=341) / p Value / NotSwitching(N=485) / WithCancer
Switching
(N=166) / p Value
Age(y),mean±SD / 59.4±17.3 / 58.6±17.1 / 0.44 / 64.9±12.7 / 63.6±12.4 / 0.26
Female(%) / 59.7 / 65.7 / 0.04 / 47.6 / 53.6 / 0.18
Chroniccomorbidcondition,nofpts.(%)
chronicrenalfailure / 24(1.2) / 7(2.1) / 0.19 / 8(1.7) / 2(1.2) / 0.69
congestiveheartfailure / 123(6.0) / 21(6.2) / 0.90 / 24(5.0) / 8(4.8) / 0.95
COPD / 176(8.6) / 37(10.9) / 0.17 / 80(16.5) / 22(13.3) / 0.32
coronaryheartdisease / 201(9.8) / 31(9.1) / 0.69 / 45(9.3) / 13(7.8) / 0.57
depression / 117(5.7) / 26(7.6) / 0.16 / 15(3.1) / 3(1.8) / 0.38
diabetes / 251(12.2) / 41(12.0) / 0.92 / 77(15.9) / 32(19.3) / 0.31
Chronicpain-relatedconditions,nof
pts.(%)
fibromyalgia / 150(7.3) / 30(8.8) / 0.33 / 11(2.3) / 4(2.4) / 0.92
lowbackpain / 689(33.5) / 147(43.1) / <0.01 / 90(18.6) / 37(22.3) / 0.29
neuropathicpain / 235 (11.4) / 48(14.1) / 0.16 / 25(5.2) / 9(5.4) / 0.89
osteoarthritis / 334(16.3) / 59(17.3) / 0.63 / 34(7.0) / 11(6.6) / 0.87
othermusculoskeletalpain / 771(37.5) / 146(42.8) / 0.06 / 147(30.3) / 42(25.3) / 0.22
otherspinalpain / 733(35.7) / 150(44.0) / <0.01 / 99(20.4) / 48(28.9) / 0.02
IndexLAO,nofpts.(%)
CRoxycodone1,437(69.9) / 171(50.2) / <0.01 / 246(50.7) / 77(46.4) / 0.35
transdermalfentanyl406(19.8) / 95(27.9) / <0.01 / 126(26.0) / 41(24.7) / 0.35
CRmorphinesulfate213(10.4) / 75(22.0) / <0.01 / 113(23.3) / 48(28.9) / 0.35
Pretreatmenthealthcarecharges($), 13,274±31,256 / 16,771±29,266 / 0.05 / 36,331±38,845 / 39,623±52,271 / 0.46
mean ±SD
COPD=chronicobstructivepulmonarydisease;CR=controlledrelease;LAO=long-actingopioid.
Table5.HealthcareChargesforPatientsReceivingLong-ActingOpioidsSwitchingandNotSwitchingTherapybyPresenceofCancer
Parameter / WithoutCancer($)Mean±SDNotSwitchingSwitching
(N=2056)(N=341) / p Value / WithCancer($)Mean±SD
NotSwitchingSwitching
(N=485)(N=166) / p Value
Inpatientcare / 7,792±25210 / 14167±25558 / <0.01 / 16710±38007 / 28081±38241 / <0.01
Outpatientcare / 4,979±8622 / 7770±8741 / <0.01 / 21393±27658 / 28136±27828 / 0.01
Medications
pain-related / 813±968 / 1170±982 / <0.01 / 675±764 / 1060±769 / <0.01
other / 716±977 / 857±991 / 0.02 / 840±1019 / 982±1025 / 0.13
TOTAL / 14299±28166 / 23965±28553 / <0.01 / 39618±49114 / 58259±49417 / <0.01

fentanylandCRmorphinesulfatepatientsmayhavebeensicker and/or in greater painand therefore mayhave beenmore predisposed to switch therapy.The degree to whichresidualconfoundingaffectsourresultsisthereforeun-known, and caution is accordingly warranted in interpreta-tionofourfindings.

We also note that, while healthcare charges are higheramong patients who switch long-acting opioid therapy incomparison with those whodonot,itdoesnot necessarilyfollow that therapy switching isthe cause of these highercharges. Causation mayruninthe opposite direction. Forexample, higher charges may reflect disease progressionand/orunrelievedpain,whichmaygiverisetotherapyswitching.Whileitisnotpossibleforustoascertainthedi-rection of causation, our results nonetheless arenot incon-sistent withoura priori hypotheses. Additional research isneeded todeterminethe exact relationship between thera-pyswitchingandhealthcarecosts.

Finally,wenotethatthedatabaseweemployedcontainsinformationonlyondrugsthataredispensedthroughretailpharmacies.Thus,totheextentthatpatientsreceivedlong-acting opioids viaother channels (eg, inpatient), we mighthave underestimated therateof therapy switching. We be-lieve, however, that patients typically receive these medi-cations through retail pharmacies and that the magnitudeofanybiasthatthismayhaveintroducedissmall.

Summary

PatientswithoutcancerwhoarebeginningtreatmentwithCRoxycodoneortransdermalfentanylarelesslikelytoswitchtherapythanthosetreatedinitiallywithCRmorphinesulfate.Ourfindingsmaybeofeconomicsignificance,ashealthcarechargesaresignificantlyhigheramongpatientswhoswitchtherapycomparedwiththosewhodonot.

ArielBergerMPH,SeniorAnalyst,PolicyAnalysisInc.,Brookline,MA

DeborahLHoffmanPhD,attimeofwriting,AssistantDirectorof HealthEconomicsandOutcomesResearch,PurduePharma, L.P.,Stamford,CT; now, IndependentConsultant,Hamden,CT

SethGoodman,attimeofwriting,ComputerProgrammer,PolicyAnalysisInc.; now, Programmer,PharMetricsInc.,Watertown,MAThomasEDeleaMBA,SeniorConsultant,PolicyAnalysisInc.

RaafatSeifeldinPhDPharmD,ExecutiveDirectorofWorldWideHealthEconomicsandOutcomesResearch,PurduePharma,L.P.GerryOsterPhD,VicePresident,PolicyAnalysisInc.

Reprints:GerryOsterPhD,PolicyAnalysis,Inc.,FourDavisCt., Brookline,MA02445-7629,fax617/232-1155,

References

1.Parrott T. Using opioid analgesics to manage chronic noncancer pain inprimarycare.JAmBoardFamPract1999;12:293-306.

2.Pappagallo M. Aggressive pharmacologic treatment of pain. Rheum DisClinNorthAm1999;25:193-213.

3.Theuseof opioids forthe treatment of chronic pain: a consensus state-mentfromAmericanAcademyofPainMedicineandAmericanPainSo-ciety.

4.JacoxA,CarrDB,PayneR.Newclinical-practiceguidelinesforthemanagement of pain in patients with cancer.N Engl J Med 1994;330:651-5.

5.AmericanPainSociety.Guidelineforthemanagementofpaininos-teoarthritis,rheumatoidarthritis,andjuvenilechronicarthritis.Glenview,PA:AmericanPainSociety,2002:54-92.

6.Mucci-LoRusso P,Berman BS, Silberstein PT, Citron ML, Bressler L,WeinsteinSM,etal. Controlled-release oxycodone compared with con-trolled-release morphine inthe treatment of cancer pain: a randomized,double-blind,parallel-groupstudy.EurJPain1998;2:239-49.

7.Allan L, Hays H, Jensen NH, Le Polain deWarouxB, Bolt M, DonaldR,etal.Randomisedcrossovertrialoftransdermalfentanylandsus-tained release oral morphine for treating chronic non-cancer pain. BMJ2001;322:1-7.

8.RadbruchL,SabatowskiR,LoickG,KulbeC,KasperM,GrondS,etal.Constipationandtheuseoflaxatives:acomparisonbetweentransdermalfentanylandoralmorphine.PalliatMed 2000;14:111-9.

9.Breitbart W, Chandler S, Eagel B, Ellison N,EnckRE, Lefkowitz M, etal.An alternative algorithm for dosing transdermal fentanyl. Oncology2000;14:695-705.

10.Donner B,ZenzM. Transdermal fentanyl: anewsteponthe therapeuticladder.AnticancerDrugs1995;6:39-43.

11.Johanson GA. New routes of opiate administration. AmJ Hosp PalliatCare1992;9:4-5.

12.DonnerB,ZenzM,StrumpfM,RaberM.Long-termtreatmentofcancerpainwithtransdermalfentanyl.JPainSymptManage1998;15:168-75.

13.KorteW,deStoutzN,MorantR.Day-to-daytitrationtoinitiatetransder-mal fentanyl in patients with cancer pain: short- and long-term experi-encesinaprospectivestudyof39patients.JPainSymptManage1996;11:139-46.

14.RobardsM.Transdermalfentanylinthehospice:asurveyofrescuedos-ingandpaincontrol.AmJHospPalliatCare2001;18:47-50.

15.Professional ICD-9-CM code book. 6th ed. Reston, VA: St. AnthonyPublishing,2000.

16.Current procedural technology.4thed. Chicago: American Medical As-sociation,2000.

17.Allison PD. Survival analysis using the SAS system: a practical guide.Cary,NC:SASInstitute,1995:29-60,111-84.

18.Ghali WA, Quan H, Brant R, van Melle G, Norris CM, Faris PD, et al.Comparison of2 methods for calculating adjusted survival curves fromproportionalhazardsmodels.JAMA2001;286:1494-7.

19.SAS/STATuser’s guide, version 8, 4th ed. Volumes1 2. Cary, NC:SASInstitute,1992.

20.HughesD,McGuireA.ThedirectcoststotheNHSofdiscontinuingandswitchingprescriptionsforhypertension.JHumHypertens1998;12:533-7.

21.Hensley PL, Nurnberg HG. Formulary restriction of selective serotoninreuptakeinhibitorsfordepression:potentialpitfalls.Pharmacoeconomics2001;19:973-82.

EXTRACTO

TRASFONDO:Lospatronesdeconversióndeterapiaenlospacientesusandoopiáceosdelargaduraciónnohansidobiendocumentados.

OBJETIVO:Compararlaconversióndeterapiaentrepacientesiniciandoseenterapiaconoxicodonadeliberacióncontrolada(OLC),fentaniltransdermal(FT)osulfatodemorfinadeliberacióncontrolada(MLC).

METODOS:UsandounbancodedatosdereclamosdeserviciosdesaludenlosEstadosUnidos,identificamospacientesiniciandoseenterapiaconOLC,FT,oMLCentre1/7/98y31/12/99.Coleccionamoslosreclamosparacadapacientedurante6mesesluegodeiniciadalaterapia,ycomparamoslaincidenciadecambioenterapiaentrelostresgrupos.Además,estimamoselcostototalenlosserviciosdesaludentrelosquecambiabandeterapiaylosquenocambiaban.

RESULTADOS:Identificamos1931,668y449pacientesquecomenzaronterapiaconOLC,FT,yMLCrespectivamente;16.7%,25.0%,y35.9%teníancáncer.Enaquellospacientesquenoteniancáncer,latasadecambiodeterapialuegode6mesesfuede10.6%(OLC),19.0%(FT),y26.0%(MLC);larazóndecambioentrepacientesconcáncerfue 23.8%,24.6%,y29.8%.Lastasasderiesgomultivariadas(TR)(vs MLC) paraelcambioenlaterapiaenpacientessincáncerfue0.36 (95%intervalodeconfianza,0.27–0.47)paraOLCy0.69(0.51–0.94)paraFT; paraaquellosconcáncerlasTRcorrespondientesfueron0.72(0.50–1.03),y0.76(0.50–1.16).Elcostototaldelcuidadodesaludfue

significativamentemásalto(p0.01)entrepacientesaquienesselescambiabalaterapiaylosqueno($23965vs$14299enlospacientessincáTherapySwitchinginPatientsReceivingLong-ActingOpioidscáncery$58259vs$39618enlospacienteconcáncer).

CONCLUSIONES:PacientessinenfermedadmalignatratadosconOLCoFTsonmenospropensosarequerircambiosensuterapiacomparadosalosqueseinicianconMLC.Loscostostotalesdelosserviciosdesaludsonmásaltosentrelospacientesaquienesselescambialaterapia.

MitchellNazario

RÉSUMÉ

AVANT-PROPOS:Lesmodalitésdetransfertdespatientssousnarco-analgésiechroniquenesontpasdécritesdanslalittérature.

OBJECTIF:Comparerlestransfertsthérapeutiqueschezlespatientstraitésinitialementavecdel’oxycodoneàlibérationcontrôlée(OLC),dufentanylparvoietransdermique(FT),oudusulfatedemorphineàlibérationcontrôlée(SMLC).

METHODES:Enutilisantunebasededonnéesaméricainederéclamationspourdessoinsdesanté,lesauteursontidentifiélespatientsquiavaientdébutéuntraitementavecdel’OLC,FT,ouSMLCentrele1/7/98etle31/12/98.Ilsontcolligécesréclamationspourchaquepatientdurantles6moissuivantl’initiationdelathérapie,etontcomparél’incidencede

transfertthérapeutiqueentreles3groupes.Ilsontaussiestimélesdépensestotalesdesantépourlespatientsayanttransféréversunautreopiacéparrapportàceuxquisontdemeurésavecletraitementinitial.

RESULTATS:Lesauteursontidentifié1931,668,et449patientsquiontamorcélathérapieavecl’OLC,leFT,etleSMLC,respectivement;16.7%,25%,et35.9%avaientuncancer.Pourlespatientssanscancer,letauxdetransfertdethérapieétaitde10.6%(OLC),19.0%(FT),et26.0%(SMLC)à6mois;pourceuxavecuncancer,lestauxétaientde23.8%,24.6%,et29.8%. L’analyse multivariéepourlerisque(vsSMLC)detransfertchezlespatientssanscancerétaitde0.36(intervalledeconfiance0.27–0.47)pourl’OLCet0.69(0.51–0.94)pourle FT;pourlespatientsaveccancer,lesvaleurscorrespondantesétaientde0.72(0.50–1.03)et0.76(0.50–1.16).Lesdépensestotalesdesantéétaientsignificativementplusélevéespourlespatientsquiontdûtransférerdethérapie(p0.01)parrapportàceuxquiontpoursuivilathérapieinitiale(23965$vs14299$chezlespatientssanscancer,et58259$vs39618$pourceuxavecuncancer).

CONCLUSIONS:Lespatientssanscancerrecevantl’OLCouleFTsontmoinssujetsàuntransfertdethérapiequeceuxrecevantleSMLC.Lesdépensesdesantésontplusélevéespourlespatientsquidoiventchangerdethérapie.

MarcParent