See discussions, stats, and author profiles for this publication at: llttPSV/l/WWWVIESEHI :hgatevnet/publication/8417 188
`
`Partial Compliance and Risk of Rehospitalization Among California Medicaid
`
`ResearchGate
`
`Patients With Schizophrenia
`
`Article In Psychiatric SerVIces - September 2004
`DOI
`lD llTSIappi p5 55 3 2:35- SoLII’ce' IanELI
`
`ClTATlONS
`508
`
`4authors:
`
`0 Peter Weiden
`.
`University Cl lll'nO's at Chlcago
`238 PUBLIC/\TlONS 11,563 CIT/\TIDNS
`SEE PROFILE
`
`Amy L Gregg
`AmerisourceBergen
`41 PUBLICATIONS 1,379 CITATIONS
`SEE PROFILE
`
`Some of the authors ofthis publication are also working on these related projects:
`
`mm
`
`treatment resistance View project
`
`mu
`
`MM Lines View project
`
`REAOS
`530
`
`Chris Kozma
`260 PUBLICATIONS 2,056 CITATIDNS
`SEE PROFILE
`
`Julie Lockleai’
`EMD Serono
`43 PUBLICATIONS 1,205 CITATIONS
`SEE PROFILE
`
`Alluomemiullnwingmis pagewas uploaded W Amy LOTugg On 29 May 2014
`The user has requested cnha ncemcnt of the downloaded file.
`
`1
`
`LATU DAO4357213
`
`Exhibit 2137
`
`Slayback v. Sumitomo
`IPR2020-01053
`
`1
`
`Exhibit 2137
`Slayback v. Sumitomo
`IPR2020-01053
`
`

`

`
`
`Pater 3' Wtidm, 313E.
`Chris Kazan, P31E
`
`31133.1: 1.9111211, P3121111.E, MEA.
`
`Giyfeetire: The objective (1? this study was in cvainate the inflationsiaip
`between compliance with an antipsychotic medication regimen and risk
`of hospitalization in a, cohort Of California f‘éedicaiéi patients with schib
`()phrenia. Afetizods: Compliance behavior was estimated by using a netw
`rospecfive review 0f Caiifm‘nia Medicaid Pharmacy refiii anti meiiicai
`ciaii‘ns i'nr 4,325 outpatients fur whom antipsychotics were prescfiiieci
`for treatment of schizophrenia from 3,999 to 203%. Compliance iieiiaifw
`ior was estimated by using four different Liei'initiuns: gaps in medica—
`tion the any5 medication consistency" and persistence, and a medication
`possession ratiu. Patients were failnweti fur one year anti 31.21111 an aver~
`age of 3.9.3 dispensing events. Logistic regressinn mocieis using each
`cnrnpiiance estimate were used to determine the odds ui' haspitaiiza»
`tion. Resuits: Risk of iiospitaiization was signifieanfl‘y correiated with
`campiiance. ‘With a“ definitions, lower compliance was assnciateri with
`a greater risk oi" haspitaiization over and abcave any other risk factors
`for imspitaiizatinn. For exampie, the presence of any gap in medicatinn
`coverage was associated with increased risk of hospitalization, inciuding
`gaps as smali as (me to ten days (odds ratin [031:198). A gap (Sf 11 to
`30 days was associated with an, OR of 2.81, and a gap of mere than 30
`
`days 1 'as associated with an OR of 3.536. ( 1w?
`nns‘ This study shower}
`:1 direct correlation between estimated Partial campiianee and hospitaL
`ization risk among patients with sailizophrtnia across a continuum 0f
`compiiance behavior. {Psychiatric Services 552886—6913 98041)
`
`
`
`edication compliance, or
`adherence. among patients
`. with scijimpln'enia has ()1
`ten been reported as an allornotifinn‘
`behavior: the patient eitl161 is compli—
`ant, or is not. This notion oi noncom—
`
`pliance as complete. willful cessation
`of all antipsychotic medications is not
`an accurate representation of actual
`medication takinv below/1m“ among
`outpatient Impulations with schiznr
`
`pinenia. Rec
`" studies using more
`
`
`
` Di': Wkidan is nflilinfie‘r] with the mum]: ’
`
`093'! (#3011?P?” service aftim State University Of
`mind
`
` if?! New Yin": in Emails-n. Dr. anuma is .
`with the
`Uniting: (1f Pharmarw (1f the: Univ
`01’: Cm? and D1". Lac19121" are wi
`he welcomes
`
`
`
`r‘s' y ofSamiiz, Car-3157mm Comm m
`
`
`
`
`
`eépDI'I-
`ruse/arc}: division ijzmssm Medical Afiai \ In [Hurt/(Zia? [V:w
`f‘wey Semi L707?
`,1125 7'1Mon Hmbomion Road.
`(Zena; to Dr Loc‘Hearaijamsrm Moriicai Afiaim, 11111
`
`
`Titusu27:76,?»"we”; I;
`51 08560(e—maiijloc'kld@‘nrizisjig.0(21n.). A 7‘
`, "Sign nffhix‘ 'pfi'pfii' was
`
`presented at [he mnuczi amazing ofziw Cali
`
`J‘f Psi/chitin
`.yzti«molt/gm Pfla.ma-
`Cisfix, Edd May 7 to 4,. 2005i in Charleston, Snntii Cardin/i.
`386
`PSi‘Ctii/iFREC SERVICES
`
`snphisticaiod assessments have Found
`that a majority of patients with schiz-
`
`
`L'lirenia who are considerenl
`to be
`cmnplianl with their antipsychotic
`medication regi mens actually Show a
`range of compliance behaviors, prob—
`ably for many (iii/@156: reasons. The
`full range [if cumpiiaucc-spectrnm
`behasior becomes apparent when pa
`tient
`seif—repnrt
`is
`(contrasted with
`other. more quantitative. measures.
`such as the Medicatiun Event Muni—
`
`toring System (MEMS) (1), Or when
`compliance is determined by bloozi
`ed
`samples taken during unscli
`home Visits (2}.
`Thus the term “partial compiiance”
`
`seems pm"
`able to “noncornpliam
`
`in that ti] / firmer explicitly alt-knowi—
`
`.1 yes the common situation in which
`
`65 8011167 but not ail, minis
`a persnn t2
`or her prescribed medication. Partial
`cmnpliance may take several forms.
`includingr taling an amount
`that
`is
`can sistcntlv it:ss than rmcmmnndfi.
`irregular (on-and-oit) dosing be-
`havior. and having discrete gaps in an—
`iipsyclintic therapy—461' exanipie. in
`the case of patients whu are unwi Hing
`or unable to ref’ll a prescription.
`It is important
`to note
`that pedtJal
`cumpiiance refers only to compliance
`behavior and does not reflect eitlie r the
`
`efficacy of the treatnimit 01‘ {no por—
`sons attitude tow-curl
`taking metijca-
`tion. For exanipie. partiai compliance
`can be due to efficacy problems (such
`as engnitivc dystunctinn that toads to
`ibrgetting to pick up a refill}, systems
`barners (for example. a PIESCl‘ipl'iDI] is
`not rciilicd because insurance cover-
`
`http://pspsychia trynnltncorg
`
`August, 2.004 Vol. 55 N0. 8
`
`LATUDAO4357214
`
`2
`
`

`

`age has run out). or an intentional deci—
`sion to stop talung medication.
`A number of studies have. shown
`
`that most patients with schiroph rcni a
`are partially conipliai
`.
` 90
`ty and colleagues {3)
`touiid that
`percent of patients with schizophre—
`nia had some degree of partial conr
`
`pli an e. in this overall sample of 675
`
`patients. medications were not avail—
`able for 36 percent ot’ patient—days of
`medication exposure. McCombs and
`colleagues (4‘) found that 92 percent
`of a sample of 2,010 patients wi .1.
`schzophrenia had at least one disrup-
`tion in antipsychotic medication cov—
`
`erage during; the course of :1 ye- r and
`that the mean duration ot’therapy was
`only i412 days per year (4)., Another
`study rcported that among patients
`with early--
`'sode schizophrenia. (53
`
`percent oi‘a sample of 182 had at least
`
`7 OVE’I a (H'tG-YCET
`one gap in ther
`period. with illOSt of these gaps eX3
`tending over a 11‘1011tl1i5) However;
`comparison oil partial compliance
`rates between studies is difficult, be~
`CAUSE l'lit’ l'tthli'Jliqtlt’S used it) tilt’ufiilt't’,
`
`compliance, as well.
`is the. definitions
`of compliance.
`vary from study to
`study.
`it
`is well known that medication
`
`noncompliance is one ot‘the in ost im--
`portant modiliable risk factors for re—
`lapse among patients with schizo—
`phrenia (6.7). Estimates suggest that
`nonconiplianee causes about 40 per—
`
`cent of relapse 18:‘1.A review 01.
`. .ven
`studies demonstrated that noncoiu
`
`to
`pliant patients had a six—montl’i
`two—year relapse risk that ‘was about
`.3 7 times that of compliant patients
`(9) Firstrepisode patients, who poi
`tcntially have the most to lose from
`repeated relapse, are similarly likely
`to experience relapse when their
`
`treatment. is i iterrupted
`l0). How—
`
`ever. these stuc
`“ used the, tradition 3
`
`that
`al definition of noncompliance
`'11tinuation of a11—
`
`is. complete (1
`
`ti1;
`‘ycliotic medication. Thus the rela-
`
`tionship between partial compliance
`and relapse rislr is not known. An un—
`:lerstandind ot the role ot partial com--
`pliance in relapse will help deline the
`threshold between the eXtent of par
`ti al compliance and risk of rclapsc.
`One approach for examining the el—
`iect of partial compliance on outcome
`is to use phannacy claims data as :1
`WWlliA’lRiC SthlCLS
`
`tittp://ps.psychiatryoutinccrg
`
`August 3004 Vol. 5'5 No. 8
`
`conservative proxy measure for comr
`pliance behavior. Analysis olipharina—
`
`cycclaims has been iiied succeshill},
`ior other chronic diseases
`tor csamr
`
`
`pie. hyqxu e1
`ion and epil
`
`show relatioi
`ships between partial
`and
`hospitalization
`compliance
`(11714). Claims records in adminis3
`tratiVe databases (3:2.n be used to as ess
`
`Whether patients discontinue their
`medication therapy {stop taking their
`medication) or retill medications in
`consistently (slop doses) {iii—2(2).
`The primary objectives oi’the, analy—
`sis reported here were to determine
`the association be ween estimates of
`
`partial compliance and outcome. with
`the hypothesis that the lower the com--
`pliance. the greater the risk of hospir
`tal ization, and to ovaluatc the quaint ita—
`tive characten‘ stics that detine any po-
`tential relationship between partial
`compliance and liiospi taliza ti on .
`
`Methods
`
`5:1in desigfi and population
`A “20 percent
`random sample of
`”99732001 Calitornia Medicaid data
`was used to eval1.1ate the association. be—
`
`tween partial comp iance and hospital--
`ization. To be included. patients with
`
`schizophreniaithat
`‘
`. patients with
`an {CD424’if 1.ode (1i295XX------had to
`
`
`have at least two disp
`sing events for
`
`anti svchotic medications (111 11.11 g a 5"—
`month enrollment period (luly l
`to
`December 3L 19.99). Qualifying prev
`scription claims included claims for all
`approved oral antiosychotic mediczr
`
`
`tions. includ rig ne.
`r antipsychotics
`available before january l 2000.
`t ach patient was assidned an index
`date. defined as the date of the par
`ticnts tirst prescription during the en—
`rollment period. Because it is possible
`that patients with new diagnoses
`would have signilticantlv ditierent
`compliance issues while beinO‘ stahir
`
`lized with medication therapy: the}
` elore.
`was to study patie its who we]i
`t.
`
`receiving ant"
`
`)sycu'liot " lhe'
`patients were also required to have at
`lrast one prescript ion in the six
`months before their index date the
`
`study was not examined by an institu
`tional review board. because all per'3
`sonal
`identifiers were removed and
`
`the investigators were not aware ol‘the
`patients’ identities at anytime.
`Data var rc C‘ibtainr.d tor l2 months
`
`
`
`after each natient’s indeX date (the ob—
`
`sor1ationpcnod Because compliance
`patterns might be attected when pa
`pa—
`ticnts arc about to lose eligibility.
`.sinc
`ill
`1
`
`
`.d in the stud 7 we
`who remained eligible tor California
`Medicaid for an additional
`three
`
`months after the obscwation period.
`Patients were excluded if they were
`younger than 1.8 years at the start of
`the study period if they had long—
`
`term care 11'.
`s (becauc of the possi-
`
`bility ofincomplctc rccordsl or if their
` )er of
`calculated medication use (111
`units dispensed divided by days’ sup-
`ply was ten or more. The reason For
`the unitsrperrday restriction was that a
`use of ten or more times per day sug-
`~gests data entry errors.
`Patients who had a claim coded
`
`with an lCDut/J—CM code tor bipolar
`disorder (296%. 296.iX. or 296.4—
`195.8) at any point within the avail—
`able data set were dropped from the
`analysis on the grounds that these pa—
`
`tients nught 1
`drugs other than an—
`tipsychotics {for example. lithium) as
`a primary therapyithus violating nec~
`
`essarv
`sumptions.1.01 calmilation of
`compliance vanableS/ Patients who
`were receiving long—acting antipsy—
`chotics ihaloperidol decanoate or
`tluphenazine decanoate) were eX
`eluded because of inconsistencies in
`
`the number oil d ays’ supply recorded.
`
`Measures ofparn'al compliance
`
`Four measures oi compliance Wt e
`evaluated: gaps in medication the ra-
`py,
`iner. cation consistency. medics
`
`tion persistence. and a in
`tion
`possession ratio (M PB}, Because the
`results from all measures we 1, simi—
`
`
`3.1 is
`lar tht primary locus in this
`on gaps in medication. a measure that
`is conceptually straightforward and
`
`easiest to use in cli
`ical practice. For
`this studv, medication gap was de
`
`line 1 as the longest pcriod (.lurinfi
` Whi
`d to be
`
`1....o 111'dication appee. e
`available. Ci
`in
`xiitiguous periods
`which no medication appeared to be
`available were based on dispensing
`date and record
`l days supply for
`each antipsychotic prescription. Four
`categories based on each. patients
`
`maximum gap in therapy were (le-
`
`iinect: 3ro days, one to ten days. ii to
`30 days. and more than 30 days. The
`mean mimhcr of gaps per patient and
`887
`
`LATUDAO4357215
`
`3
`
`

`

`claim) were summed. The MFR was
`
`calculated by di‘ading this sum hy the
`number of ambulatory days in the
`study period. By evaluating the perv
`s over a ti.
`.d period,
`
`ments of
`
`skippinU' doses (consistency) and dis
`
`continuation ( ‘7
`' .ce) into a sin
`
`gle composite n. .asure.
`
`Mention of hospitalization
`A marker was created to indicate
`
`least one
`whether a patient had at
`“mental, health hospitalization" dur—
`ing the one-year. postindex observa-
`tion period. Mental health hospitalr
`izations were identified hy using;
`“mental health" ICTD—Q-CM diagnosis
`codes in the first (priniaiy) diagnosis
`ticld. The t‘ollowing diagnosis codes
`were used: schi; iphrenia (2.95m),
`
`depression (236.
`296.1); 296%.
`30(14):. 3309.0X. or Stirs}, an);
`' y
`
`(300.05g,
`3002):.
`300.3X,
`306.54X,
`7
`308xx 309.2X 3094);, or 309.9X).
`other psychoses
`{297.Xx, 298.“.
`299.“, 300. ix. 1302.8x, or 1507.9xl, and
`dementia (29(3.xx. 29i.:2x. Sittfix. or
`33th). Use ot‘ a broad definition oi
`psychiatric hospitalization ensured
`that no relevant psychiatric hospitalr
`izations related to the, in dex diagnosis
`of schizophrenia were missed.
`
`Stamtim! azizaéi’sis
`The primary analysis evaluated the
`relationship between compliance and
`the presence of at least one mental
`health hospitalization during the one
`year toilow—up period. Logistic rev
`
`ting presence
`gression models pre
`of at least one hospitalization in the
`postindex year were developed for
`each ot‘ the cor‘npliancn measures.
`Medication gap models predicted
`hospitalization by using four gap cater
`
`to days. one to ten days. it
`to 3'0 days, and more than 30 days.
`
`Logistic models for consistency, perv
`
`s" stenc
`. and the MFR predic ed hos—
`pitalization lJV using continuous
`nieasu res. lnteractirms were included
`
`only i tthcy added significantly to the
`e planatory power of the model; vari—
`ah
`
`as were dropped iron] the models
`lk they were insignificant and had a
`negative impact on model
`tit. Corn--
`pliance was also categorized and evair
`uated by using chi square tests.
`t7or descriptive analyses. consisten-
`t’St‘tItii/tit‘ltifl SERVICES
`
`a i
`
`Tflgakj
`
`Characteristics of a sample of 4.325
`outpatients for whom antipsychotics
`W'C ‘
`e prescribed for
`treatment of
`
`schizoph reni a
`
`NVariable "/0
`(\
`cars)
`
`MeaniSD
`
`
`
`
`
`African American
`Asian
`
`Hispanic
`Other
`Unimowu or mi?
`
`
`
`Medicare eligibility
`Yes
`No
`l l ospital ized
`Yes
`NG
`
`2,114.
`2,21 i
`
`65/!
`3.577
`
`48.9
`51 .t
`
`15.].
`84.9
`
`the mean gap duration (across all.
`
`therapy
`traps) were also catculated.
`Medication consistencyis a measure
`of whether patients shipped doses
`when medication shouid have been
`
`available—dint is, between the dates
`
`ot’ the tirst and last presciiptions a pa—
`tient had tilled. Consistency was calcu
`
`lated. using a rnoditied definition from
`the literature (21). as the percentage
`oi. time a patient appears to have
`medication available divided hy the
`period during which the patient
`should have tl‘ieoretically used all the
`available medication. A weighted ave
`cragc was taken across antipsychotic
`therapies. Medication persistence
`captures wl’iether a patient discontin—
`ued all
`therapies. This <.‘al<."ulation.
`represents the numher ot‘ days he.
`tween the iirst and last prescription,
`
`divided by the ii 'ed number of days
`in the study petiod.
`The MFR was calculated in a mare
`
`'
`.
`. '\
`.
`.
`V
`ner similar to that used tor therapy
`gaps and is a modification of the liter--
`ature—hased t‘oriiiula (15,16). The
`nuinher of d: ys a patient was not hose
`
`pitalizcd and showed evidence of use
`..
`tion
`oi
`any antipsycl/iotic
`inedica
`rs
`(based on dispensing date and ct
`supply recorded on the. prescription
`388
`
`cy, persistence, and MFR mean scores
`as well as categorical frequencies are
`presented. For consistency and the
`MFR, the txnnpliancc categories de—
`
`
`in
`ti an ce
`“d less than 70 perc ,nt con":
`as noncompliant and
`at least 70 per-
`cent compliance as compliant. Air
`though, no standard is available tor
`identifying compliance categories the
`literature suggests that a 70 percent
`cutoti is reasonable (33,2223). The
`MPH was calculated in a manner such
`
`that no values exceeded 100 percent.
`The categories tor persistence were
`
`less than 90 percent compliance and at
`least 90 percent compliance.
`
`Results
`Patient disposition and
`demographic characteristics
`
`A total of 4,325 patients met the se
`lection criteria. Patient characteristics
`
`are summarized in Table l. The pa—
`tients’ mean age was 44.2 years; 58.5
`percent were nien, and 55.6 percent
`were white. Approximately half we e
`also Medicare eiigihle (48.9 percent).
`
`A total ot‘65ri patients (i591 perce'
`‘
`had at least one psychiatric hospitai—
`ization. Analysis of the crude data also
`showed that age, ethnicity, and i
`sur—
`ance status were associated with he
`
`on. Hospitaliza-
`lihood of hospital
`tion was less likely with increasing age
`
`but more likel
`among patients who
`were African American and among
`patients who were
`eligible
`for
`
`Me
`are {Table 2).
`
`Partial compliance
`and hospitalization
`The patients in the study were on the
`more compliant end of the compli—
`ance mntinuum, as indicated by the
`mean compliance variables, shown in
`Table, 3. Only 26" patients (6.2 per—
`cent} had a persistence level of less
`than 90 percent, yielding an average
`
`persistence of 97 percent. Durir " the.
`. year observation period patients
`had a ineanzSD nuni her of dispens-
`
`" events of litt- ‘r.9 for l.65:..87
`
`rent drug entities.
`Figure 1 shows the percentage of
`patients hospitalized as categorized
`
`by medication. gap within. a Lute—year
`period. Ali pairwi so comparisons with
`the reference group were signiiicant
`tip/4.905). As the maximum gap in—
`creased,
`the percentage of patients
`
`http://pspsychta tryonltncorg
`
`August 2.004 Vol. 55 No. 8
`
`LATUDAO4357216
`
`4
`
`

`

`
`n—
`hospitalizeti increased. For consist
`cy anti
`the MPH. significant differ—
`ences (p<.t)t)l) in the percentages of
`
`Tobie 2
`
`Grids of hospitalization for a sample of 4.32:3 outpatients for whom antipsychotics
`
`patients with at toast one hospitaliza— tion were found. Partialty compliant
`
`Variable
`
`
`
`
`
`
`
`i
`.
`~
`.
`_
`Duration at maximum gap (\daysw
`1 to 10
`11 to (30
`MOW than 30
`
`1,98
`2,81
`3.96
`
`1 9‘7_3‘25
`L’s—4.64
`2.54765
`
`.004,
`<30]
`«0m
`
`
`
`‘76‘35
`
`“nib/ear ‘mmaw m age
`Race:
`‘
`
`
`
`“Um
`~82
`.026
`4,94
`1.31
`.725
`.12
`.96
`,894
`.02
`1‘03
`
`l58
`25.7
`<: .901
`“' Zero is t"
`‘
`rent.
`t) VVtii .
`
`9 Not ts'lcdicarc t
`to is the referent.
`
`
`ld‘
`‘
`
`
`
`patients had signiti *antly hi gher rates
`of hospitalization. Patients who were
`less than 70 percent compliant by the
`MPH had higher rates (it hospitalizau
`tion than those who were at least 70
`
`pe
`ant compliant (22.3 percent. and
`i3 8
`resnectively )< an it
`ercent
`‘1 ’
`P" ’
`‘
`1."
`
`Similar rcsnlts Vv’Ct ., observed lor eoi’i—
`sistency (24.5 percent compared with
`12.8 percent. p<:.t)01}.
`in addition.
`patients who were identified as being
`less than 99 percent compliant in? the
`persistence measure had hi gher rates
`
`of hospita ization than those who
`were 'irit'21‘ltiiicrl as hcing at least 90
`percent persistent (25.1 percent anti
`14,5 percent.
`respectively. p<.00l)
`(Table 32:).
`
`Compiiame as a
`predictor ofbospz'ttziéza firm
`Having a maximum gap in use of
`medication that was as smail as one
`
`to ten days in a one—year period was
`associated with a significantly iii--
`creased rislt of hospitalization (crisis
`
`ratio [OR]: 98) (Table 2}. Come
`pared with patients who did not h ave
`
`gaps it
`cation therapy. patients
`who had a one— to ten—day in aximuni
`gap had almost
`twice the (raids of
`hospitalization. As the gap increased
`to it to 30 tiays and more than 30
`
`‘_.8l
`tin-zit
`days. Otis increased to
`3. 96 . re s ne ctive ly.
`Logistic regression results (or the
`other three compliance measures
`were similar to the results for metie
`
`ication gaps. Vv’ith a Ht percent im—
`provement
`in consistency. persist--
`ence, or the MFR. the odds of hos—
`pitaiization were iowereri by factors
`of t 6" percent. 9 percent. and 23 per
`cent,
`respectiveiy (13401). These
`models are consistent with the re—
`sults observed tor
`the maximum
`
`medication gap models. Grids of
`hospitalization were also significant—
`ly ait‘ectett by Medicare eligibility.
`depending on the cornpiiance vari—
`able (Tattle 4). hitetncare eligihiiity
`and an increase in age oi~ ten years
`were significant factors in the mod—
`ets for the MPH. consistence. anti
`persi stcn cc.
`PSYCI'E tA’i‘tttC SERVICES
`
`Discussion
`
`
`The major tinting of this study was
`the direct reiationship between meas~
`tires of partial compliance and rislt of
`of
`hospitalization: the lower the ie‘
`compliance,
`the greater the risli of
`hospitalization, Vie emphasize that
`
`this finding is not as i]: Jinve as it
`might appear. Most puhiisheti studies
`showing the iinlc between ilUiiCUitiplt'
`ance and relapse define nonconipli:
`ance as persistent and complete dis--
`continuation oi‘antipsycl‘iotic iiietiicar
`tion. From that per
`ctivc the co—
`
`
`hort in our anahs was mostly comrr
`pliant. and even then a relationship
`was obsewet’l between partial compli—
`ance and hospitalization risk. This 33..
`sociation behaves more like a continr
`
`uons function than, a categorical hinc—
`
`tion—that is, once any tie}? ee of par—
`
`tial compliance was it
`ed by the
`data. there did not seem to he any
`lOW'EDd cutoff below which hospital—
`ization risk reverted to that of the ret—
`
`erence cohort t no indication ot‘partial
`compliance), This observation held
`for ail
`tour compiiance measures.
`These results suggest that retati'v‘ety
`sniall changes in overali compliance
`are meaningiully associated with
`changes in the risk of hospitaiization.
`Partial compliance seems to he as—
`sociated with increasing risk of re--
`lapse in the lorrg~tei'in treatment of
`schizophrenia. “/0 tbund that
`incri—
`s as small as one to ten
`ication g .
`
`continuous (lays in a one-year period
`were associated with a twoioid in
`
`91799.36
`
`
`in hospitalization risk. Ti
`
`.Figum 1
`
`Percentage of patients with schizophrenia who were rel’iospitaiizeti, by maximum
`gap in the any“
`25 —
`
`n
`:3
`R? "J
`9.4
`aso
`{J
`it
`Qo
`3
`d
`$4
`
`
`
`
`
`20 _
`
`v
`to —e
`
`u
`o v
`
`4) ~
`
`
`
`i 1730
`
`>30
`
`Total
`
`
`rum gap {days within one year)
`
`
`a All pa’
`se, comparisons were, Siginificant it putt”?
`
`tittp://ps.psychian‘yontincorg
`
`August 3004 Vol. 5'5 No. 8
`
`LATUDAO4357217
`
`5
`
`

`

`Taoism?
`
` tonsinu sample oi /t325 outpatirmts
`Conniliztnee and 11st: ol prescription met
`tor wlimn antip‘ ‘ "l
`e preserihcd tor tre- tnmnt oi ‘eliizt
`(1hr\nia
`
`
`
`Variable
`
`
`
`
`,de i .2
`
`ilk/Evil
`
`21
`203
`188
`242
`
`73
`+44
`
`
`
`6.4
`11.9
`16.],
`4.8
`
`21.5
`12,8
`
`21".l
`ME
`
`306
`LEO?
`978
`880
`
`5332
`37018
`
`20!)
`3,471
`
`93.6
`881
`83.9
`78.4
`
`75.5
`87.2
`
`74.9
`
`' 5.5
`
`
`Gap in medication th
`y
`
`lX/lai'iinunt gap C day”,
`,
`.
`
`0 (la)
`b
`l to ll) days
`l l to 30 davsd
`trier; than 30 days
`Consistency”
`<70 percent compliant
`270 percent compliant
`“
`,7
`
`<90 per ‘ ant compliant
`
`290 percent compliant
`Medication possession ratiog
`.861} 8
`77.7
`522
`223
`150
`mt compliant
`
`
`
`
`
`504 113.8 3,149‘ 86.2
`
`
`
`~ for it {it y; versus ll to 30 Clays7xi:ttl.97, dt’:l; tor
`‘ys verSiS t to lil d-ws x—
`3t! dag
` := 111,61 dirt; for t to it} (lays versus more than :30 days,
`ll] 1, (it: l,
`.t. unless the patient had at least three dispensing
`
`l, p<_Mill
`
`
`results are consistent with those of
`other studies that have demonstrated
`
`the negative consequences of partial
`compliance. Valenstein and col—
`leagues {23) found that patients who
`had poor compliance were "2.4 times
`
`as ii ,ely to have inpatient admissions
`eompared with patients who had
`good compliance.
`in ad dition pa—
`tients who had poor co, oliance had a
`greater total number of psychiatric
`inpatient days (a mean of 33 days per
`
`year) compared with patients who
`had good compliance Ca mean of 24
`days per year).
`
`Another
`tudy found that partially
`compliant patients were 49 percent as
`likely as cornpliant patients to have an
`inpatient hospitalization (24) Most
`recentlyfl(lilmer and conorl<ers C25),
`in an analysis oi a California Medicaid
`database also found that rates of psy—
`chiatric hospitalization were lowest
`among patients who were compliant
`
`with their a
`‘isyehotic medication
`
`
`7‘1. These data suggest
`regimens C‘
`
`that patients who do not achieve
`,7
`
`isi‘aetory responses to treatment may
`be experiencing partia
`l compliance
`
`problem. ‘ather than rnedieation eiii-
`cacy problems Steps taken to ink
`prove compliance otter an important
`treatment option that should he con-
`
`sidered along with other options
`or switching an—
`such as combining
`tipsyehotic medication s.
`Several
`limitations of this study
`should he noted. One of the most im—
`
`portant limitations is that pharmacy
`claims data do not provide insights
`into the reasons for partial compli—
`ance. Partial compliance is a behav-
`ioral finding with no attributable un—
`dm‘lying cause. For esatnple. there, is
`no way to know whether partial com-
`pliance in our study sample reflected
`
`an, i1’1tentional difl
`on to stop taking
`medication or was unintentional, per,
`haps due to service barriers such as
`di continuity of care (26).
`li'urtl'terinore the causality ot‘ the
`association between partial eor’npli~
`anee and hospitalization has not heen
`estahlisliecl For example it is possihle
`that patients Who do not iullv respond
`to their medication would he more
`
`likely to have inedieati on g:{as and
`that the common shared trio
`is lit£oni—
`
` plete medication t
`not attempt a test of temporal conti:
`guity between noncompliance and
`
`ization, but such a test could
`hospital
`
`
`he eons.
`,red for a future analysis
`The database we used had teelinirr
`cal limit; Must The use of Medicaid
`
`
`claiins data as a proxy for partial com-
`pliance relies on minimal coding er—
`
`
`
`
`
`
`
`
`
`
`325 outpatients tor whom antipswhoties were preseri hed tor treatn‘icnt ot' schizo-
`Odds of hospitalization for a sample oi
`
`phrenia, based on compliance modelsa
`
`M edimtion possession ratio
`OR
`OR
`OR
`
`E n rlpoin t
`est ima te
`22
`d ll
`est in: a to
`X2
`(if
`estimate
`352
`d i
`
`Cnnsistency
`
`Persistence
`
`i
`48.7
`l
`9.07
`.Ql'k
`t
`86.42
`.8; W
`it) percent improved compliance
`l
`2778
`i
`29.01
`.81 M
`l
`22%
`.53”
`Ten-Near increase in age
`
`—-----
`-------
`l
`8.1.9
`1L1
`-----------
`-----—
`African Ant:
`ican versus white
`
`i
`23.84
`l
`238]
`2]. 3:” 14 5”
`1.55""
`Medicare c igihlc
`2'
`a Only variables that were significant in the model are shown.
`p<.01
`twt<
`,
`Kuwaiti
`
`.87“:
`{“22”
`
`3%
`
`PSt‘Ctli/it‘lttt} SERVICES
`
`http://pspsyehia tryonltncorg
`
`August, 2.004 Vol. 55 No. 8
`
`LATUDAO4357218
`
`6
`
`

`

`1015135291 PatienES' discontinuation
`of use of 1\"1e(1icai(1 5911111135 may 1101
`be reflected 111 a 1111131y 1115111011 [2281:
`1V'11t11icaid 1131101119 Hay 1101 111: 1111119—
`sen 12111116 011116 11010111311011 as 21 11711016;
`and 012111115 (121121 1112111 1101 111C111de pi)"
`1121111211 0011101111(1015, 511011 215 111E11111
`11131.012111(1111951‘1719 121111015 1
`C2112
`
`1111121110113 101’ 001111011 21111113 1211112111199 are
`dependent .011 21001113.10 (12111115 111101—
`
`1111111011}
`taining 10 dates days 511].)—
`11111 (11 11131110311011 1110 pahentsbeing
`61180111211371.(1 (1111111‘13111119 111(3(11<.1(111011 as
`
`a1. adm ".011, and 1111? 2‘15511111ption 111211
`111610 are no other sources (111111edicau
`
`11011 <11pp1y~10r tax-2111111113. 111ed'103111011
`52111110199. These 131101 111011101.11s 0011M
`1821(1 10 2111 (11760931111121th ()1 1.116 {15110--
`
`13.11011 111311114011 11111111030011 1,21115 211111
`1109pil211i7atirm. A1111011g11 111911911 (11
`{3112111112113};
`(13.11115; met11011010gy 1121s
`1111:- poter’rtia1 advantage (11111556551110111
`()1 Large (10110112;
`()1 13111161115 over
`1011ger periods, 130101111111 sources 01
`error 110 111 111C 2159111111111011 111211 (11211118
`data are a proxy 101‘
`[121111211 C(imp11--
`a11ce. A1t1‘1uug11 111656 13101015 may
`have 111110111101111 1112115 11110 our 211111113
`1213 (.1
`515.
`1111?:
`1112111
`(1112111231 g
`.
`.
`sources (11‘ (11111118111131 error 11311116011
`
`1111:: reference gmup 311d 1111». 1711111211
`0.011111111511101. group.
`11
`however.
`Given these 111010151,
`511011111 be noted 111211 1110(‘110511 01311115
`
`211111051 certai111y memsiiumte
`(19.13
`90111111131109 because 012111115 (12113 can
`1111011 011117 11119111er the 111030111111011
`
`was 11.11051. 1101 W 1191‘ 1110 m (11.3.
`
`11011 was taken correcfly 01‘ a1 2111 1.281.
`Ftir111ermn1‘e,
`the ref re 1(‘e group
`111(e1y 111011.1d€:<.1
`5011113
`11211191115 W110
`
`
`0011111 be mare accum V described
`
`215 par112111y (10111111111111. 111115 the 1101117
`31 1"01311'0113111p 11(‘twm111 113111211 (50111—
`p1i2111ce 21nd 11(1sp11'2111'22111011 is 1i1<e1y to
`be s1r011ge1‘ 1112111 could be. identified
`1:10.111 our 1.121121. Factors other than.
`
`()1 which
`111-2111}!
`[111111211 C(111’1p11'aime.
`11/011111 1101 be captured in 3 (1.1311113
`database. indudijlg so '
`1 support——
`
`111‘
`131014
`111ere01------211115.
`5111151211106
`
`abuse 02111 2111001 1103131121112311011 121135
`
`{70111111310115
`
`(1011'1p112111ee 011 other aspects of
`11111
`schizophrenia,
`911011 as 1110 (1111811011
`211111 severity (111111111535. U111111ate1y, 311--
`1131101111 (301111111 121.1100, 11110115111 11111110171111
`
`1.1112111112113010
`'
`(1111f; (19111132131 11110111111—
`1i (ms 01‘ 1111311 311101211 11110 mentions, 0011151
`
`reduue i111) 1011 01‘ re1apse 1111(1 f61105p1'
`1111122111011 215500111190 11711.11 the1o1’1g21erm
`1132111110111 01' 8(‘1‘11 201111 rem-21.
`
`Acknawiedgment
`
`9.111(111 was 511131301135 by janssen
`T1111;
`1’11311113113111111021 Preducts, 1.11).
`
`Re 'erenees
`1.
`NF. Matsui 1),Herm1111n (1'
`
`
`
`
`1g 01‘ 1111”pg/011011905311113111111epres—
`'
`V111t111111e.111163137.1991
`
`An—
`1’, G mg $1.1, Kozma L
`
`1111311121.
` (- 21111111211 111ee11ng o1
`
`
`.-
`)1 Neumpsy 10131132
`1112111. P11011031“) 1.70138 1.52
`
`V11;(j0mbs_18
`1101 1
`
`
`
`
`
`14.
`
`C1631
`
`
`associated 1105p
`271159221166. 19811
`
`8161118111“ Keepsefl 11.) F1111 813,61. 2
`
`7
`1(1 01(
`(1 scription
`14432:}.
`
`
`
`
`
`15. 501211DA,Sk;1(.rTL
`
`mnsd2
`a1de11v
`
`S1K6111L p)"
`91(supp1 1A1
`.
`16. 50131 DA,31(361'1‘1
`
`
`tw.ai1sder111a1 dehverys‘v'st
`
`
`11‘21111111yp8111
`2
`
`' VB t11f314py. 111 [\1 U
`1 1011111211 (>1~ Mcde
`
`S. 1991
`1: 911supp1 11/115753
`
`
`
`
`
`erxdent d1abetr
`
`
`Phammq 211111 1112.11113311119: 1.
`
`299, 199-3
`1.9.
`
`:20..
`
`
`
`
`911. S.
`.infl‘ 11 1710011212
`
`mag records:
`121111 compha
`11101110119, V3111
`
`10115 journal 01'
`
`111‘s 1—511111155
`. 7116, 1997
`cz11Ep1d611
`C11
`1 L 5‘3;337.-
`
` 11111011 of (01
`.
`.
`1’39 11:1
`19311.1. P51111111
`32. {ogzirgv GE: P11:
`
`
`
`. 101111131 01 '
`
`112187723. 1993
`1‘5. Avign- (£1111
`“r
`
`
`11111110119111};
`
`
`311011 1011r“€1 01C1in1cn1
`
`try 59(5unp12112212.5, 1998
`
`
`
`
`IIIBD ‘
`
`baa/10.1111 He:111.11 1V12ifif11: 11.16 1.10. .1997
`9. 179111.011 V1215, 1111191111 heinssen 111%.: DE191'
`
`
` 13111191111 2316372851, 19117
`
`
`10. Robinson D. V‘v’bcr
`.1
`
`Predictors of relapse
`
`(1121 or 5011120311
`episode 01
`111201111
`'
`'
`8 disorder. Are 1.
`
`
`
`£17. 19119
`
`M G, A1v'
`
`
`
`‘QT1
`
`
`
`:1T11e e1"-
`.
`11y M. Munch P, 812
`11
`
`
`13101101'11011184'
`‘
`70111111111111
`
`sour‘cr L1
`2 11011 1'11 a 50111. 1111101110 21110 bipo—
`
`
`111‘ 1101011111
`111 Prater 111656111611 3.1.1the anrtua]
`1..
`
`
`'1u 011111 Newi
`ical D1110 L\31112111011
`, 13012. 11211011.
`11
`
`
`.111112 12, L71111L’
` ‘ CK. 118C117 11’, 01
`
`
`.111 W11 1) 211111 psychotic
`
`1‘61’1
`15111101101111
` .
`
`99, 211011
`
`
`[iont-omp1iance and 1110111121 11681111 [30111:
`1.
`
`
`Co“ ‘1131109 ”1d 1116 Trent
`‘111 Alliance.
`
`wood, 1996
`11 P. 0115011 M: Cost of re 31
`
`0131
`113..
`SC11‘1Z0P1118‘J12’1
`131‘
`425, 111115
`
`
`25’. 1.1:1(11‘1‘1 PAA1/0r1’1 I 1171/
`r12
`
`
`(1318 a; a 2» ‘011r’9 1(1r (“1:111
`
`11 31.1011
`’“1’1171113 211'
`pidemi110,3}? 42:‘13791111.18!)
`
`29. 16220111 L1: Assessing! quahty using adm 1152
`11
`five (
`121.
`1‘
`
`
`
`12718 pt 2 .
`
`1
`
`331(1112111'1‘1111‘31511111153
`
`1111,1111pr
`
`15.11;:1151300/1 VOL 5'5
`
`17-10. 8
`
`LATUDAO4357219
`
`T110 10511115 01111's 51u11\1511;1<1'es’( that
`11111121110111g 1.1.1111 1,111,211 11.1: 2111.11111:; 1111;. 121111312;
`011133111311 C(11Tip1iance 110112111015; 02111 111--
`d‘uee the risk 011105131121112'2111011 among
`
`
`p211
`"115
`111111
`$01.11
`11.11111?1112..
`17110111:
`511111102; wi11 111<p1rm1 1110 1111111101 (11 par--
`
`
`,111‘9111‘y0111i11cx
`
`
`
`[O
`
`
`
`1111‘ measuring 211111
`1110(1s
`Farmer KC:
`1110111101"
`med; 3.11011 regj m :11 adherence
`
`.
`('1ir11.1
`'.. C1111 .,
` 1
`1empm11ics 21:107110911. 11199
`13. 1
`E‘J‘OHCC‘ RF, {3112311 LS