( ){)12-j l0 I si()r l l l Ir I 4 -, :1!)' l ' t rr ,

. . lor t i )rAl. { .}1, ' N 0ttvot.s,txt r Nf i :xt,r t , l . }rst: , tsl :(iolrvrighl iL) 2{)(.til by LipJrin<:cilt lVillialts & Wilkins

Treatment of Acute PsychosisOutcomes From

If is notatrle that, lJO years after its initial designand implementation ancl 17 years since completiorlof'data collection, the Soteria project is sti l l produc-ing irifonnat,ion I'eler.'ant to today's management ofpsychosis. Soteria's original aim was to assesswhether a specially designecl intensive psychosocialtreatment, a relationship-locused therapeutic milieuincorlrorating minimal use of antipsychotic medica-tions tor 6 weeks, coulcl produce equivalent or betteroutr:omes in treating newly diagnosecl patients withschizophrenia conrpared vvith general hospital psy-chiatric warrl treatment rsllh antipsychotic nredica-tions. Sol,eria also inteucled to reduce the proportion

' l)ep:rr' lnrent ol'Soci:tl Work, [,rriversity of Southern C]alifor-nia, NlllF-!12, Los ,A,ngeles. California 900E9-0411. Sencl reprinLletlrrests to I)r. IJola.

' Sotet' ia Associates, San I)iegt.r, Calit 'omia.Tlris researcrh rvas supporlt'd by grants fronr the Nati<lnal

Institrrl.e of Merital I lealth.I)rc:sentecl in paft at the World Psychiatric Assoc:iation's Intet-

rrat,ir;tral Cr-rtigress in Nlaclrit l, Spinn, October"2, 2001.Ttre arrl.hr-rrs t,irank Leonarcl S. Miller. [ ']r.f)., Lr(l Berkeley, antl

,f im trl intz, Ph.l)., Ir(]I,A, lor their help r+'ith statistical consulta-tions, anrl John NI. L)avis, M.D., Llniversit.v oI l l l inois, lor crit iqLreol'an ea.rlier ,, 'eLsion <lf lLris nr:rnusc'r' ipt.

\ ,o1 . l ! )1 . No. . l

l )r i nl t ' t l i rr / ' .^\ ' . '1.

Without Neuroleptics: T$o-Yearthe Soteria Project

of patients rnaintained on antiltsychotic mer{icatiorrs(thereby recltrcing exposure to dnrg-incluced toxici-ties) and to re(luc:e the rate at which early-episo(leclients became chronic users of mental health ser-vices. This study is unique in errnploying a relativelylalge sample (1/ : 179) of clients newlir diagnclseclwith DSM-II ^schizophrenia (cliagnoses were strbse-quently converted to DSM-IV schizophrenia andschizophrenifbnn disorder) in a rluasiexlrerimentalresearch design conrpadng multiple otrtconres at 2years.

F'or many years, antipsychotic medications havel,reen the treatment of r:hoice for patie'nt,s with earlyepisode psychotic: disorriers (APA, 1997; Cole et al.,1966; Lehmzur and Stein\A'achs, 1998). I lowever, pl 'e-scription of conventional antipsychotics c:arries sub-stantial risk of ch'ug toxicities (Poptrl ancl Trezza,1998) and structural brain change.s (e.9., N'fatisen et&1., 1998), While atyprical antipsychotics exlribit amorr) benign shr-rrt-term sidereffect prohlc (Worlelctal., 2000), there has r-rot yet treen adequzrte t,irrle'obseling their etl'ects to nrle out t-.r-nergeltce oI-adtlitional long-t,erm toxicities. F'or example, tlre ru:-cently reporterl asso<:iation oI' atypical iurtipsyc:hot-

JOHN R. BOLA, Prr.D.,r ancl I,OREN R. MOSHER. M.D.!

Tire Soteria project (1971-19811) <:ornpared resiclent.ial treaturenl. irr the r:omnrunityand rninitrtal use of antipsychotir: nredication with "usurll" hospital treatment torpatients witir early episode schizophrenia spectrurn psychosis. Newly diagnoseclDSM-II schizopltre'nia subjects were assigned c:onsec:utir.'ely ( 1971 to 1976, l/ : 7tl)or randontly 1197{i to 1979, t\ : 100) to the }rospital or Soteria ancl followed tor 2years. Atlutission diagnoses were subsequently converled to DSM-N schizophreniaand scltizopirrettif omi disorder. Mult,ivadate analyses er,.ah-ratetd hypotheses of erlualor better outcotnes in Soteria on eight individual outcorne measures and a conpos-ite outcoute scale in three ways: for endlloint subjects (.V : 160), for contplet,ingsulrjects (l{ : 129), and for completing subjects correctecl for diff'erential attrition(tV : 12!)). Endlroint sub.jects exhibitecl snrall to rnediunr effect size trends lavoringexperintetrtal treal,tnent. Completing subjects hacl significerntly better c:onrposit.eoutcomes of a nredittnt eftec:t size at Soteria (+.47 SD,,1r : .03). Corlpleting sulrjectswith schiznphrenia exhibited a large effect size benefit n'ith Soteria treatment ( +.81SD, 1; : .02), particularly in clornaius of psychopatholog.y, wor-k, and social l\rnc-tioning. Soteria treattrtent resulted in tretter'2-year outcclmes fi:r patients with newlvdiagnosed schizophrenia spec:trum psyr:hoses, partictrlarly for completing subjet:tsantl for those wittr schizopltrenia. In adclition, rinly 58% of Soteria su[jer:ts rer:r,rivedar-rtipsychotic medicatious during the follow-up periocl, and only 19/o were contin-uouslv maintained on antipsychotic medications.

-,1 I',ert: IVIenl Di,.s l9L:2lg*229.2003

219

220 I }O[,A ANI) NI0S}I t iR

ics w-ith tliabetes nrellitus (Sernyak et al., 2002) is(:ause for sonre concenl.

In developing the Soteria approach to treatment,the clesire to nrinirnizer rnedication-induced toxicitiesconverged with three additional factors: the recog-nition of significant rates of recovery withotrt drugtreatment in early episode llsvchosis, the obsenra-tion that many patients dcl not benefit from metlica-tions (through cimg treatnrent resistanc'e and non-compliance), aud a valuing of interpersonal care andtreatment of mentally ill patient,s.

Rates of recovery without meclications are signit-i<:ant, partic:ularly frrr those with early episode psy-chosis. l'or example, placeb<l recovery in the acutephase of the early NIMTI rnultisite trial wa.s appl'ox-imately 37% (Clole et al., 1964), arcl the placebo-treatedgroup had ferwer rehospitalizations at I year postdi,s-charge (Schooler et al., 1967). Estimates of placebcrrespons€] rates fbr patients with acute schizophreniarange flom 1fflo to 4fflo (Davis et aJ., 1989; Thornley etal., 2001), with a median of 2l:o/o (Di-xon et al., 1995).Long-tenrr frrllow-up stucties coucluctecl prior to thewidesl-l'eacl use of ilntipsychotic drugs report, func-tiorul reco\()ry ratcs greater than 500/o (Rleuler, 1978;Oionrpi, 1980; Huber et al., 1980).

Not all psychotic patients benefit from dmg treat-ment. 'l'reatment resistance to cclnventional antipsy-ch<>tic agents is estimated to be ZU/o to 40o/o(Ilellewell, 1999). Nonc:om1:lianc:e w'ith conventionalarrtipsychotics is estimated to be 4Io/oLo 55o/o (F'entortet al., 1997). lurproved compliance with at,ypical an-tipsychotics is often assumed but has not yet beenestablished (Wahlbeck et aI., 2001).

"'I'raitement moral," a hulnAnistic trend in the c:u'eanti treat,ment of persons witli mental illness, c'an betlacred t<l Pinel's removing chains fi'onr the nren inParis' Bi<,:etre' Ilospital in 1797. Following in thehruuanistic tleatment tradit,iott, Soteria incorpot'atedaspect,s of moral treatmetrt (Rockhovell, 196:l),Sullivan's (1962) interpersonal theory and speciallydesigned milieu at SheJrarcl-Praff, Hospital in the1920s, and the "der,,eloprnental crisis" notion thatgrowth nray be possible from psychosis (Laing, 1967;Ilenninger, 1959; Perry, 1974).

'I'his is the first report, li'om the entire Soteriasarnple using multivariate methods to test hypothe-ses of cclmlrarable outcomes over a Z-year period.We use l,wo-tailed tests to evaluate hypotheses forea<rh outcorne in three ways: for endpoint, sub.iects(N : 160), for cc'rmpleting sub.jects (/t' : 129'), andI'or c'ompleting sub.lects ac{iust,ed for different,ial at-trition (,\ : 129; Heckrnan, 1979). Subsequently, wecronclur:teci tests lor schizopht'etria and schizophreni-lorm subjects separately.

Methods

St,tlrly De,sigrt,

TI-re Sote ria pr<lj e ct emp loye d a rlua^sie x1r erime r-rtaltreatrnent comparison using consecntivc space-available treatmernt assignment in tire fir'st ccihofl(1971 to 1976, /V : 79) and an experimental <lesignwith ranclom assignment in the st:conci cohort (1976to 1979, .A/ : 100).

Subjer:ls

Sub,ject,s were recruitecl from two county l-rostrrital

trrsyr:hiatric energency rooms in the San F ranciscoBay Area. Nl persons meeting the follou'ing crit,erialyere zrsked to participate: initial rliagnosis of sr:hizo-phrt:nia by threre indeperndent clinir:ians (per DSM-

D; at least four of seven cardinal synrlrtoms ofschizophrenia (thinking or speech clisturbances)catat,clnic motor l-rehavior, paranoici icleation, hallu-cinations, delusional thinking other than paranoid,bluntecl or inappropriate emotion, clisturtrance of'social behavior and interpersonal relations); .juttgedin need of hospitalization; no lnore than one previ-ous hospitalization for 4 weeks or krss with a diag-nosis of schizolrhrenia; agecl 15 to i32 years; ancl nc>tmanied. These criteria were intencied to produce arelativel.y poor prognosis g.roup, at heightene.d riskfor a chronic corlrse, through the exclusion of olderand manied patient,s (Strauss itnd Llarpenterr, 197.9).After description of the study to the sub.jects, writ-ten informecl consent wa^s obtained fronr lratientsancl their families, if available.

Emelgency r'oom stzrff psychiatrists made initialdiagnoses. An irtdepenilent research team trained t<-rmaintain interrzrter reliability (Kappa) of .80 or bet-ter on all measrtres made subsetluent assessments.Merasures were takern at ently, 72 hours (clesigrred toscreen out drug-ittdu<:ed psychoses), 6 u'eeks, Iyear, ancl 2 years postadmission. Most lbllow'-upmeasures were face valicl (e.9., v!'ork, Iir,'ing arrange-rttents, r'ehospitalization, etc. ).

The ethnicity of subjects (I/ : 171) was 80% Eu-ropeiur American, 9o/o Alrican American, and ll't/aother ethnic gr(rups. Sixty-four percent (A' : 179)were male and.16% wel'e temale. The rnean age \4ras21.7 years (range, 15 to 32 years; SD : 3.4; rV: 179),with the average client coming frorn Hollingheacl's(1957) lower.mitlclle class (higher score is lowersocial class: cla-ss III is ZfJ to 4i3; rnean SIIS score :

42.3; SD : 16.1; range, 11 to 77; A : 159).

Treahrnnts

Soteria providecl pleclominantly extramerli<:altreatment, employing a cleveloJrmental crisis ap-

SO'l' l iRIA:'I 'WO-Y I'IAR OI l1'( l( ) i\ ' IES 221

proach to recovely fr'onl psychosis. Treatment in-volved a slnall, homelike, intensive, iute4rersonallyfcrcusecl theratrleutic milieu with a nonprofessiclnalsta.ff that exlrected ret:ovely and relatecl with clients"in ways that do rtot result in tlte invalidation of theexperient'e of madness" (Mosher and Menn, 1978a,

fi 716). Experimental treaturent was provicled at twofar:ilitiers: at Soteria ancl a replication facility,Eman<ln. Artipsychotic medications were ordinarilynot used cluring the first 6 weeks of treatntent. How-erver, there were explicit criteria for their short-termuse rltrring this period; 760/o (62 of 82) received noantipsychotic rneclic'ation.s during the initial 45-clayperiod. AJter 6 weeks, medication presctiption deci-sions lvere made at a treatment conference thatincludecl the client, stalT, and the c:onsulting psychi-atrist^ A manual describing Soteria treatment ingreatr-'r' det,aii has been published in [iermart(Mosher et al., 1994).

Control facilities \4'ere well-staffed general hospital psychiatrii: units gealed towat'd "rapid evaluationancl placement in other parts of the county's treat-rnerrt nc,twork" (Mosher ancl Menn, 1978a, p 717). Inthr.se'unit"s, virtually all sr-rb.jects (94%,85 of 90) weretreatcd with continuous courses of antipsychoticme-dication (average 700 mg chlotpromazine equir.'-alents per cla;y), and nearly all were prescribed post-rlischargt. nredications. On discharge, subjects werereferred to an extensive array of outpatient services.

X'Ieasurr:s

F)ight outcome rneasures were used: readmissionto 24-hour care (yes or no), number of readtnissions,rlays in reaclmission(s), a glotral psycltopathologyscale ('N'losher et al., 1971; 1 to 7, higher is nroresyrnptomati<:), a global imtrtrovemet'rt scale (Mosheret al., 1!i71; codecl I to 7, 1 : much improventent,4 : lro cliange, 7 : utucl-r worse), living indepen-dently or with peers (yes or no), an ordit"ral measureof working (none, part-tinre, full-tinte), and the so-c:ial functiouitrg stttrscale of the tsrief F ollow-up Rat-ing (tlFR; Sokis, 1970). [.'or cornpleting .subjects(i/ : 129), obsen'ations to the 2-yeat'follow-up eval-tuatiou were nsed. Endlroint analyses (I/ : 160) usedobservations to tire last postdischarge obsetvation.

flonrlrosit,e orttc'ome sca^les were created ftrr end-point aucl r:onrplcting subjects frotn the eight out-c:onle measures by conveft,ing each to standardized(:) scores oriented with lrositive values for betteroutcornes and summing. Missittg values were set tothe sub.ject's rnean score on available standardizedrneasrrres 1-ctr 5o/o of mis.sing endpoint and 8% ofrnissing completet' infonnation. Cronbach's alphawas .77 and .74 for the entlpoint scale and the conl-

pleter s<:ale, respectively. Composite sc'akls \4rerethen restandarclized, allowing subsequent analysestci be interl,n'eted in stattdard cleviation (effect size)units (Neter et ai., 199ti).

DSM-II schizophrenia patients with symptonrs forat least ti montlm were rediagnosecl with schizophre-nia (,129/0, 71 of 1ii9) because the addition of thiscriterion was the primaty change frorn the DSM-II tot)SM-lII and has been <:arriecl lbrward into the DSM-IV. Sub.iects not meeting t,his criterion were rediag-nosetl with schizophreniform disorcler (58o/o, {i8 of169). A variabie approximating clays of antipsychoticruse during the follow-up period (betweert thr. end ofexperimental control clf meclication at ,{5 days andthe c-rbservation at, 2 years) wa^s created as the pro-portion clf use (0 : llo use, .33 : occasional rtse,.b7 : f requent use, I : continuous use ) t irnes thelengtli (in days) of the observation perriod atrdstrnrmecl (mean [SD] : 327.5 pTrtl; range, 0 to ti85).

Slat ist ic:al Armlysis

In main effect anal;zses, thc influence of expeli-mentzrl treatnrent, on corntrrosi[e outcorne and ot-teach outcome rnca^sure was estinratecl in three ways:lbr endpoint sub,jects (1/ : 160), fclr c:ontpletingsubjects (1V : 129), and for completing sutrjectsstatistically adjusted for difft-.rential attrition (Sote-ria nonat,trition is 83% [68 of 82] u.s. hospital nonat-trition of 6ll%' [61 of 97]; chi-square : 8.86, rlJ' : 1,p : .00). This presents a range of platnible trt'at-ment etfect estim:rtes for each outcome. 'I'hese anal-yses employed control variables ftrr the protrlortiotrof su[jects diagnosed with schizophreuia (47'Yo [32 of681 in Soteria us. 2lIYo [7 of 58J of hospital compl-etrrrs; chi-sclual'e : 4.75, dJ': l, 'p : .04) ancl for thelength of time in the posttlischarge follow.up perioclbecause Soteria's clesign allowecl longer initial treat-nrent, stays (mean : 548 postdist:harge days for'Soteria completers us. 677 tor hospil,al completers;t : 5.89, df : L28,,p : .00).

Due to ditf'erential attrition across treatmeutgroups, Heckman'.s (1979) procedure for correctittgattrition biers was usecl in one set of trealrnent effectestimates for comtrlleters (Tables I througli 13, r:ol-umn 4). This procedure involr,'es three steps: esti-mating a probit moclel on rtonattritiott frour bix;eiinevariables, calculating a function of the probaliilitythat a sulrject was ttot, lost to follow-up (tite iuversemills ratio), ancl usir-rg this tunction as a covariate inmultivariate estimates of treatntent t-.ffects. l'he in-verse nrills ratio fi'otn tlte probit rnodel on notrattri-tion was assessed fbr collinearity with other t:ontrolvariables (Stolzenberg ancl Relles, 1997; scltizophre-nia, clays in the follow-u1r periocl); none n'as found

222 BOLA ANI) N{()SHER

'fAtsLF] IiIt t'u irr u l E,l,fect s rtf' Ilt.7tn'i, m

Complc.tt,ls Oom'eclul.frtr Atlrili,on ('N ,,, I29)IJS) uul

( ) t t tc:orrte Virr iable F)rtdllt.rittt "'' (lotttplt:1.nrs" ' ( 'o rnp lc t t ' r ' s Ar [ j r rs t t ' t l ' ' '

Oonrpositr olltc{ )lue'rSocial frrnctioning"(llulltrl psv('l lopal ltolog.v /

lmlrrovetnt'nt in pslichopalhology1'

Working"AnyI.'trIl-tiur.e

L iv ing a lo r re o r w i l l t pee ls 'Reilrlrt issir>n /

N ulutlrr o1' readmissions k

Days in reaclrnissionl

. 1 7

.!0

.05

.09

.01

.02

.18' i : : i- . 1 0-.:10-.9;J

. ? i 1 ,

. 1 8

.21*'r

. 1 7 +

'05.04. l9 i '

--.Oit;

- .11-4 .6

.;ll't'r

.08

.20r"r'1'7rl

.08

.07

. t 7- .1 t ;1 .-.gg{,*,

-2:j.(j

" l:stiuates control for schizophrenia./schizophrenifomr disorder and nulnber of days between initial disdlaxge &i' Estjnr{tes conhd for the schizophrenitschizophreuifoml disor

dilTerential atlrition by treatnent group." Signiti(ance tests are trro-tailed: *p <-: .10, **p .i .06.'/ Dilferenm in the cotnlosite outcome for Soteda sub,iects (in standard deviation lmits).'' Differcnce in the probability of membership in the two best categodes (hal'ing little or tro p8ychopathologs).J Ilifference ill the probability of ntembe-Fhip in the two best categories (having excellent or very good impi)ve-rBent in psychopatholosr).'r l)ilfen:lxre in the prubability of the event occurring (readmission)." Differcnce in the expected value Glumber of readmissions).' llifference ir the expe{rted \'alue (da}s in readmission)./ Differen{ie in lhe pmbirbility of ttre ei?nt occuning (living al('le or u'ifl pecJ:t).i I)itTerence in the prcbability of the e\crts occurring (any work, firiltime work).' Difference in sociai ftrnetioning (on a ii point scale).

T.q.tsLE 2

(N : 49), atul (:ornpltters (''rtn'rrted..fbr Attritiott f N : 49)( )utt:otne Vitriir"trle Endpoint"' ' C-lornpleters"' (-lor1 trr1,, rr't r Adiustrrd""

Oonrposite ttrrtcouledSor:ial [\ rrrctir-rniug't i lob;r l psycl tupal I to lugy'Inrproveurenl in psychopafholog-vg

Workingr'ar l 'I,\rll-tirne

Lir, ' ing alone or lvit lt !r(,els?Iteaclnrissir-rn '

N Lrmtler of reacft-nissitltts r

l)a.r's in rea(llnission/

.39

.64**

.34*'i

.34r't'

. 1 8

.11 )

. 1 9

.05

.36a l ' )

.38

.67j"k

.44|-

.,1!lf i'

,) |. , ) f

- L t )

.27

. l !

.il83.1.fl

.81 'k ' l '

. irl)*''14'r"i'..19r,r,

.4(Ji'*

.211't+

.28-.?I't '- .92

-:1.83

" Ilstinrates control for nunrber of dal,s betw'een rnitiitl discharge anrl Z-year fbllow.up.

'S ig t r i f l can< :e tes l s a re tw* ta i l e t l : *p ' , : . 10 , * *p . . - .05 .'/ DilTerence in the compclsite out('c)lne f'or Sotena treatment (in standzrrd devi:rtion units,).'

I)jftt'rence in ttre prohatrility of rnetnbership in ttre two best t'ategories (htrving littie or no trrsyclro1ratlrology)./ Difference itt the probatii l i tv of trtentbetship itr t lre tw'o best r:at,egories (having excellent or very good inrprovetutnl).1/ I)ifference in the probability of the event occuning (readrnission).' ' Difl ir lence in tlre e'r1rt,r:tecl vahrt' (rrrurrber of rearlnrissir.rns).' [)ifference in tlte expected value (ciays in readmission)./ []ift'erenc't- in the probability olthe event r-rcrr:urring (ii\,ing alone or n'ith peers).I' I)ifferenc,. in the lrrobability of the events occuring (iury work, full-time *'ork).t I)ift 'erence in soc'ial fun<rtiorring (on a l3 pclint s(:ale).

(the largest correlation was with schizophrenia,Pearsr-ln r' : -.06, NS). The two-stage fleckmanprocedure resulLs irr a small clistortion of standarderror (lstinrates through uscl of an estimated ratherthan an ol)senred inverse mills ratio. lnitial efforts to

coll'ect the st,andard en'ols resulted in onl5r slightp-value chariges (in the third decinral place); t,here'-fore, they were left uncorrected.

These analytic procedures werer r€)peated sepa-rately lor endlloint (It' : 63) and conlpleting sub-

SOI'FIRIA: ] 'W{)'Y FIAR Ol rf( l( )I\ 'IFIS

TAtsLF] 3i\'lot'rf irLll

Oorrrpleting Suhjects (N ,.= ,3{//, i lt ld (:otrtplelers Cctnr.tctcd.f'rtr Atlrit ict}u, f N ='. 8{/)

223

FlntlJ;oinLo' ( 'o t t tp l t ' t r ' r s " ' (,lornplcl crs Adjtrsl t:ti""

Oonrposit,e c)utcolre'rSocial l iructionitrg'(ikrbal psychollatholt>gy/Inrprovernenl in psyc'hopatholog//

lVorkingt'Anyl\rll-tirne

l , i v i r rg ak rne o r u i l l t I ' ee rs tlte;xlntission'N rrnrtlt'r' of re:rclnrissiotts/'l)ays in roacltnission/

' 1 f l- . l t )

.03

.06

- . 0 I- . 01

. t 7- . 1 9 *- .59+

- 16.,{

. r . ) r )

- . 1 5

.08

.05

- .0 t )-.Oti

I ' t

- . l ( i* .gg*+

-30.c)

. ' r+- . 22

.07

.06

- .09* .u8

. t 2- . : 0 1 '

- L24*1- .71 . { t

" llst-inrates contnrl for nurnber' of days between initial discharge and 2-year tbllovr.up.

' Signifir-'an<'e tests are tw'o-tailed: *p '.< .10, **p < .05.'1 Diffr'rence in conrposite orrtcorne for Sotcria treatnrent (in stturdard deviation unitsl.' I)ifferenc't' in sociai fllni'tioning 1on a i3 trloint scale)./ I.)i lference in the probability of memberchip in the tw'o best categories (travrng litt le cn'no ;lsychopathology).I ' l) ifference in tht'probability of menrbership in the tlvo best categt.rnes thaving excellent or very goorl irnprroveur(iut).

" lJiftirrenct, in tlre Jrrotratrility of the events ocr:urring (any r,r'ork, ltll-tirne rvork).t l)ifference in the probatrility of the event occurring (living alone or rr,.ith peers).i Differenct- in the probability of thc event occllning (readmissiorr),A Difft'rc'nr.'e in the expet:l,e<i vahte (reatlnrissions).' f)if l 'ererrce in the e-xper.:terJ value ldays in readrnission).

jects (l/ : 49) with insi<lious-c)nset schizoplu'enia,?lnd lor endpoint (rV : 97) ancl completing subjects(i/ : 80') r,r'ith schizophl'enifornr disorder. Subgroupanalyses rme(l the saure control variables, omittingonlv the inrli<:ator variable for schizophrellia.

In each analysis, estinrat,es were made with themultivariate statistical proceclure appropriate forthe level of measurement of the dependent variable:or(linaly least squares (OLS) r'egression for intervallneasures (compositc outcome scale, social firnc-tioning), a maximum likelihood probit for binarycategori(:al variables (readrnission, Iir,ing indelren-dently), an ordered probit for orderecl categoricalvariables (McKelvey tlnd Zavoina, 1975; global psy-chopathology, improvemeut in psychopathology,working), and a classical tobit for lower tmncatedinten'al nreasures (Tobin's probit, Tobin, 1958; num-ber of readmissions, days in readmission).

Treatment effects from probit nloclels report th€lrlifference in the probability of the clbsen'ed out-come (readmission, living independently) for exper-imental subjects. Experimental treatment estimatesfrom orclered probability rnodels reporf, the clift'er-en(:€l in the combined probability of membership irrthe two best categories of the dependent variable.Effect estimates on work functioning are presentedas the change in the probability of working full-timeand as the clralge in the probability of working at all(working full-tinre plus working part-time) lbr exper'-irnental subjects. F'or trturcated interal measures

(nunrber of rearhnissions, tlays in readmission), esti-mates represent the change in the e4recterl value ofthe clependernt variable associated with experirnelrtaltreatnrent (see Breen, 1996, p 27, Eq. 2.18 for the slrec-ification). fuialyses were conductetl using the statisti-cal software packages SPSS and LIX,IDBP (LlMiteclDEPendent variables: (]reene. 1998),

Results

Mairt Effer't,s

Main effect results for enclpoint subjects (.n/ : 160;Table 1, coiumn 2) inclicate that experimentallytreated sutrjects hacl a n<lnsignificant two tenths of astandard der,' iation bertt,er outcom€)s (+.17 SD, I :

1.07, d,f : 149, NS, all statistical tests are two taile.cl).Experimentally treated endpoint strbjects had signif'-icantly better outconres on one of the eight out-comes: an 18% higher probabiii{y of living alone orwith peers (+.18, z : 1.94, dt f : 147, t r r : .05).

Results for completers, unaqiusted lor attlition(l/ : 129; Table 1, column ll'), indicate that experi-rlrentally treated strb.iects had one thircl of a stanclarddeviation better composite outcomes (statisticaltrr- 'nd, +.35 SD, / : 1.73, df : 124,p : .09), incluclirrgsignificantiy bett,er outc:omes on one of eight nrea-sures: a 21% higher probability of har,ing no or verylow psychopatlrology scale scores (+.21, z : -2.5:1,

d J ' : 1 0 3 , p : . 0 1 ) .

224 B0LA ANI)

Main eftect results for completers adjusted forattrition (N : l2t); l'able l, colnmn 4) indicate thatex1-rerimentally treateci sub.jects had nearly one halfstandarrl deviation better composite <lutcomes(^+.47 SD, t - - 2.20, dl ' : 123, p : .03) and signi f i -cantl.y better outconres on two of eight rneasrires: a20%, higher probability of meurbership in the lowesttwo psychopathology c'ategories (*.20, z : -2.22,

rlJ' : 1gr,,p : .0il) and uearly one l.evver readmissiorr('-0.98, z : - 2.37, 4f : 123, P : .02).

S r:h i e r ry ht'tt t, itt St t Qi e: ct s

Ilndpoint schizophrenia subjects (// = 63; Table 2,colurnn 2) had four tenths of a standard deviationtretter r:onrposite outcontes in Soteria (not statisti-cal ly s igni f icant; +.39 SI) , r l : 1.42, dJ ' :60,p : .16).This inclucle's significantly bett,er outcomes on tltreeof eigirt mea^snres a 31o/o higher probability of har'ingno or neailv no psychopa.thology (+.34, ; : -2.74,

4l' : l'>8, p : .01), a34o/o higlier probability of mem-hrt-.rship irt the two best psychopathology improve-nrerrt categclries (+.34, z : -2.16, df : 58, p : .03)ancl six tenths of a point (on a 3-point scale) betterscrcrial functioning (+.64, f : 2.i34, df -- 45, p : .02).

{lnadiusled for attrition, schizophrenia compl-eters ('I'able 2, colurnn l)) treated at Sotena had anonsigniticant four tenths of a st,anclard deviatiorrbetter <rutcome (+.38 SD, f : 1.19, 4t ' : 46, NS) anclstatistically significant benefits on three of eightotrtcorrres'. a 44% higher probability of being in thelcrwest two lrsychopathology categodes (+.44, z :-2.13, df : i36, p : .03), a49o/o highel probabil ity ofbeing in the best two psy<:hopathology improvementcategor ies (+.49, z : -2.75, r IJ ' : 116, p : .01), anclfrvo thirds of a point better social outcomes (f .67,t : Z.lsi), tl,f' : i37, tt : .(')2).

Adjusted f<-rl differential attrition, conrpletingschizophrenia subjects (l/ : 49; Tabie 2, column 4)had eigltt tenths of a standard deviation better col"n-pr-rsite out,cornes when treated at Soteria (+.81 SD,t : 2.42, 4f : 4ls, 7t - .02) ancl sigruficantly betteroutc'ornes on fcrur of eigltt measures: a M(% higherlikelihood of havirtg uo or nearly no psychopathol-o g y ( + . 4 1 , 2 : - 2 . 1 1 , , { l ' : [ ) 5 , p : . 0 4 ) , a 1 9 { X t h i g h e rlikelihood of having excellent, or very good psycho-pathology improvement (+.48, z : -2.67, df : 34,p : .01), and a 40o/o higher probabil ity of working(+.,40, z : 2.30, t l f ' : 40, p : .02).

,!i t:l t i z o 1t h.r'er t tfo nn Suhj ec ts

Sc:hizophrenifcln endpoint sub,jects (1/ : 97; Ta-lrle 3, coiurnn 2) had a nonsignifi<:ant two tenths of a

N{OSHI.]R

standarcl deviation better outcome at Soteria ( +. 19SD, t : .92, df : 94, NS).

Llnac{iusted for attrition, c:ompleting schizolrhreni-tbmr subjects (rV : 80; Table 3, column il) had anonsignificant one tlrild standarcl cleviat,ion trett,ercorn;rosite outc<lme at Sot,eria (+.33 SD, t : 1.28,df : 77, p : .20) that includes one statisticailysignificant findiug, approxinrately one fewer read-nrission to 24-hour care (-0.98 readntits, z : - 1.98,d , l " : 7 4 , p : . 0 5 ) .

Ac{ustecl for attrition, comp}eting schizophreni-form subject.s (l/ : B0; 'l'able l-1, column 4) hact anonsignificant one third stanclarcl der,'iation bettercomposite outct>me at Soteria (+.114 SD, I : 1,22,d-f' : 76, NS), incltrding signilicantly better outcomeson one of eight, nleasures, an average clf one anci onequarf er fewer reaclmissions ( - L.'24 reaclmits, e :-2.36, d,l : 7b, 7t : .02).

Post ltoc Artulqsr:s

Post hoc analysis comparing enclpoint subjecLslater lost to follow-up (9 Soteria and 22 hospitalsubjects) founcl no conrposite out,come ditferences(-0.18 L).s. : -0.23, f : .89, r lJ ' :29, NS), indicat ingthat loss of a lriglt-functioning subgl'oup of hospitalsubjects is nr>t a plausible explanatior-r ftir observedSoteria treatment benefits.

Investigating whether Soteria acled to reclucemedication for all subjects or only for those notrrtedicaterd during the follow-up period, nonmecli-cated comple[ers (29 of 68 Soteria arrcl 2 of 61 hos-pital subjects) \4.ere excluclecl in a comparison ofmeilication use. This cornparison founcl no betll'een-group clifferences (Soteria nrean : 421days rrs. hos-pital mean : 4!37 days; I : -0.42, df : 96. NS),indicating that experirnental tre'atment does not re-duce the duration of rneclicatiorl use for those re-ceir,'ing medicatiorts l-rut only re.duces the proportionof patienls medicated.

Comparison of the proport,ions of Sot,eria-treatedschizophrenia vel'sus scltizophreniforru sub.ject,s notreceiving antipsychotic niedications during the fol-low-up period found no signific:ant differenccr: 44% ofsclrizophrenilonn (16 of lJ6) versus 47o/o of schizo-pl'rrenia subjects (1i) of 32) were not drug treatecl(chi-square : .10. d,f : l , NS), indicating Soteria wasequally effective in reclucing ant,ipsychotic medica-tion use in both groups.

Discussion

tuIuirt Fittd'lngs

Despite some treatntent crossoverr during the fol-low-up periorl, strikingly trerreficial effects of Soteria

treatment are still evitlent at tl're 2-year follow-u1rperiocl. This is particularly notable because an ear-lier leport, of 2-year out,comes from the fir'st c'ohortof this study descritretl more moclest benefits(Nlo.slrer and Menn, 197fla). These results externd antlrefine previous reports by including both cohortsancl conclucting rnultivariate enclpotnt and two com-pleting sutr.ject trnalyses. Recall that previously re-

lrortetl (Mosher ancl Menn, 1978b; Mosher et al.,1995) separate cohort analyses of 6-week datashorved signilicant and comparable synrptomaticirnpror"'ement for both groups clespite marked cliffer-ences irr neruroleptic treatment.

Three sets of treatment elfect estimates show apatten'r of small to medium effect size trenelits forSoteria that, are larger for completing than fclr end-point subjects. This may be partly due to completershaving the full 2-year period in whir:h to recover'. Thepossibility thal, a gr'oup of higlter funr:tioning controlsr"rbjer:ts may harre been lost to lollow-up betweenenclpoint and cclmtrrletion turns out not to be anexlrlanation frrr Soteria benefits because endpointsubjercts lost to follt-rw-up had comparable outcomesin trclth treatnrent, gt'cltt1rs. flowevet', due to higherattriti<ln among the hospital-treated subjects, espe-cially alnong hospital-treated schizoprhrenia sub-

.jr:r:ts, elfect estimates for con-rpleting subjects unad-

.justecl for attrition are likely to contain a bias.'fhe'refore, the third set of treat,rnent effect estirnatesuses a statistical procedure developed by Nobel Lau-rt:ate,James Heckmau (2000 in Economics) to rrtoreac<'urately estimate the eff'ec'ts of Soteria treatmentoll ir nelv sanrple of similar clients. A{usting forclifferential attrition, colntrlleting subjects treated at,Soteda had nearly oue half of'a standarcl deviationbetter composite' outcome scores than the usuaitreatment group (T'able 1, column 4), a "medium"effect size (Cohen, 1987) that is statistically and

1x'acticerlly significant. Soteria-treated sub.jects alsohad lou'er psychopathology scores and fewer rearl-missions than ltospital-t,reated subjects.

When consiclering schizophrenia subjects sel)a-rately, results inclicate even more favorable out-collles in the Soteria-treat,erl grc)up. Adiusted fordiff'erential attrition, these subjects have signifi-cantly better corrrposite outcomes of a large effectsize (Cohen, 1987; "large" effect size : .80) despitenot lrt:ing lllol'e frequently meciicaterl in Soteria thanschizophreniform subj ects.

IVh e l,4r' c:our L[s .lor' 7h ese lri ntlirLgs,)

These favorable finrlings flom Soteria call forsorne explanation. Therefore, we exarnined possible,exlilanations in t,hree areas: analytic rnethorlolclgy,

22it

components of treatment, and similarities betweenSoteria ancl factors hypothesized as responsible fkrrfavorable developing country outcomes in WorldHealtli Organization stuclies (Jablensky et al., 1992:Leff et al., 1992).

An.o.['yl, i,r, fuIe l hodol,o g.g

The nrore favorable results in the present analysesseem partly due to the larger sarnple and mrlre con-temporary st,atistical methods. We have notecl sev-eral inlportant variables relaterl to outc<lmes ancldifferent acl'oss treat,rnent groups (scliizophrerria,length of ttre postdischarge follow-rip period, andattdtion). The contrast between these ancl pre'"'i-ously reported results highlights the import,ance ofincluding rn statistical analyses relevant control l,'ari-ables that are a) scientitically related to sttrcly design(length of foilclw-r"rp peri<xl; Wyatt, 1991), b) theoret-icaily related to outcome (schizophrenia; Lrohen andCohen, 1983), or c) rnay affect tlie generalizability ofresulLs (differential attrition rates; Heclcnan, 1979).

To illustrate this point, control variables fi'om there.gression on composite outcome were adcled oneat a time using t:ornpleting subjects (rV : 129; Table1, compare with row 1). When only experimentaltreatment was included in the regression, the effectsize estiniate (regression coefficient, in stzurdard de-viation units) for experimental treatrnent was .12and not statistically significant (t : .66, dJ' : 127,p : .51). This is analogous to the cornmonl.V used,and lterhaps overly simplist.ic, two-group l-test usedir-r earlier reports" When the r.ariable fbr lengt,h of thefollow-up per"iod was added, the effect size estirrratcfbr experimental treatment ber:ame ..')2 (f : 1.63,

4 t ' : 126 , p : .11) ,Add ing the var iab le fo r sch izo-ptrrenia (and its missing value inclicator'), the elfectsize estimate for experimental treatment became .ili>(Table 1, r 'ow 1, c:olumn j-]; / : 1.73, 4f': I24, p :

.08). Finally, adding the indicator lor the probabilityof nonattritiou, the eff'ect size estimate for experi-merrtal treatment became, a.s reportecl here, .47 (Ta-ble l, row 1, colttmn 4; t : 2.20, df : I23,7t : .0ll).Thus, it appears that previous reports fronr Soteriahave untlerestimated the benefits of experimentaltreatment through omission of important controlvilriables. In sum, we view these mult.ivariate ana-lytic methocls as more approJrriate than previouslyusecl bivariate' merthods and as producing relativelyunbiixed estimates of the effec:tir,'eness of Sclteriatreatment.

Treu ttrtenL C o n upon erut s

A nuurber of therrapeutic ingredients in Soteriatreatrnent have been suggestecl by Mosher (2001) a^s

SOl'ltRIA:'l 'W()-YIIAR OII'f(l( )i l lFls

22(i [}OLA ANI) IIIOSHt.]R

likely sources <lf benefit, including a) the milieu, b)attitudes of staff and residents, c) quality of relation-sliips, ancl d) supportive social processes.

XLilLeu, Differen<:es between experimental andhospital urilieus were assessed with the Moos WardAtmosphere (WAS) and (lommunity Oriented Pro-grarn llnvironment Scales (COPES; Moos, 1974). Sig-nificant, differcnces were found on 8 of 10 subscalcs,notably favoring the exlrerimental milieu <in involve-rnent, support., ancl splontar-reity (Wendt et al., 1983).

Al.liturles Soteria staff wix significantly moreintuitive, intro','ertecl, flexible, and tolerant than hos-

Jrital staff (Hirshleld et al., 1977). Soteria's atmo-spherre was imbued with the expectation that recov-ery florn psvchosis was to be expectecl (Mosher,200r) .

Th,erapeutic Relutionsh,'i,'rts. Perhaps the most im-poftant therapeutic ingredient in Soteria emergedfrom the quality of relationships that formecl, in p&rt,because of tlie additional t,rrratment tinre allowed.Within staff-resident relationships, an integrativecontext was cr€)atecl to promote understanding andthe discovery of meaning r,r'ithin the subjective ex-perience of psychosis. Resitlents were encouagecl toaclcnowleclge precipitating events ancl emotions ancl tcldiscuss an<l eventually place thern into perspectivewithin the continrritv of their life zurd social network.

Soc:'iul i{elutrn'ks. Tlte role of social net,works inpror.'irling direct suJrtrrort and buffering stress forpatient,s witlt psychotir: disordels has been well docrunented ('Iluchanan, 1991-r). Social strppott has beenpositively con'elatecl with fuvorable outcomes(Erickson et al., 1998). Psychotic patients tencl alscrto har.e diminishing social support networks (Coheuanrl Sokolovsk-v, 1978). To address this deficit, tlreSoteria project providecl a surrogate fanrilv for cli-ents in residence, and a client-centered postclis-charge social netn'ork grew up de no\ro. The resultwas [)eer support for conuntutity reintegration (e.9.,peers helped to organize housing, education, work,antl a social life) and alr ongoil-rg source of socialsul,rport.

Su,trtport'tr:t: Sor,rlal ltr^occsses. Social plocesseswere influencetl by a number of aspects of the pro-gram (Mosher, 2001): the creation clf a family-likeatrnosphere, an egalitarian approach to relationshipsancl househokl func'ti<lning, and an enl'ironment thatrcspected a:rcl toleratecl indir,iclual differences andaut<lnomy.

(.tulhral Fa,c{ors

Ilvicient contlasts between Soteria and hospitaltreaturent cultures brings t,o rnincl the superior out,-comes in developing countries for patients withfirst-episode schizophrenia in World Ilealth Organi-zation (WHO) stuclies (.lablensky et al., 1992; Lelf etal., 1992; Whitaker',2002'). There are' nlany plausiblesirnilarities between Soteria and the support,ive andcollectir,'ist social processes trequently hypothesizedas responsible for bett,er developing country out.-conles. The second WHO study aiso reportecl a 43Volower ploportion of pat,ients maintaint:d on ant,ipsv-chotic meclicrations in cleveloping countries (16%'ls.developed countries, 59%; .Iablensky et al., 1992).

St r Lrl,y Li m i,tat'[on,s

Tliis study has a number of limitations that re-strict the validity and generalizability of these find-ings. One limitation arises from the inclusion ofsome second-episode clients (lYo%, 63 of 179 hacltreen prer,'iously hospitalized) and requiring bothpoor prclgnosis characteristics of young age anci utl-rurarried. Thus, this sanrple can be considered to beof somewl"rat poorer prognosis than one representa-tive of only filst-episode sc:hizophrenia spectrunrclisorders.

Another limitation arises fi'om tht. lack of explicitcomparability between t,he rediagnosis of schizo-phrerria used here (DSM-II schizophrenia atrcl aninsidious onset of sympt,orns fclr 6 month.s or nrore )ancl a DSNI-IV diagnosis of schizophrenia. DSM-IValso requirers diurinished functioning. lloweve'r, cli-agnostic c:ritetia in the Soteria study were quiterigorous, requiring agreement from tlu'ee inclt-.penclentclirdcians, and since a]l werc cleemed in need of hos-pitalization, imJraired functioning carl be assulne(l.

Attrition ot28tr/o gives rise to concerrr fol the sanl-

llle's representativeness. This is accolnpanied b.vconcern lor a possible bias in the treatment effectestinrat,e due to greater attrition in the hospit,algroup (37W than in the experimental gronp (17%).As cletailecl abr.rve, statistical methocls to control fbrattrition bias ilr estimating treatment effects lvereused (Heckman, 1979), but these methods certainlydo not leplace rnissing subjects.

It, is possible. that the use of inclepenclent ratersnot blincl to treatment could have introduced a mea-surerlrent bias. Wrile financial lirnitations precludeclthe use of blind reliewers, raters were inclepr:ndenlof ther pro.ject, rotat,ed acl'oss condit,iorrs, and wel'etrained to maintain high interrater reliability on thefew meastrres that, required rater judgment.

Ar aciclitional lirnitation derives from the quasiex-perimental nature of the str.rcly. While second cohort

subjects were randomly it-ssigned to treatment, firstcohort subjects were assigned using a colmecutivespace ;rl'ailable decision rule. This raises the ques-tion of group comparability. Althougl'r we clid notfirnd st,atistical evidence of het,ween-group differ-ences at baseline, there were cleeuly some differ-L.nccs. I)ifl'erences tencled to favnr the hospital group,parlicularly with an initially lowel proportion of insid-ious-onset schizclphlenia sutljecls that became signifi-cantly clifferent lry fcrllow-up evaluation (addressc.d r,'iastatistical control). Ilowever, sinrilar results have beenn<lted in conrparing fincting.s tiom experimental andrluasiexperimental designs (Shadish and ltagsdale,1!l9fl), especially when controlling for betrn'een-grouJ)tliff'erenr:es.

( )l i.t t ical I nL ltli r:rtt, io n s

On the whole. these data argue that a relationallyfocused t,herapeutic milieu with minimai use of an-tiJrsyc:hotic clrugs, rather thzur drug treatment in thehospital, should be a pref'erred treatment for per-sons nern4v cliagnosed ll'ith schizophrenia spectrumclisorrler. We think that the balance of risks andirenefits a*ssociatecl with the commorl pract,ice ofnredicating nearly all early elrisocles of psychosisshould be reexamined. Itt additiotr, the search, be-gun earlier. for treatment response subtypes inschizophrenia spect.rum disorclers (Carpenter ancll leinrichs, l98l), particularly for pat,ienls not bene-fiting fxrnr antips.ychotic medications, should be re-sttmetl." 'n

In marty mintls, and in cliuical practice guidelinesfor schizopl^u'enia (APA, 1997; Irrauces et al., 1996;l,eluran iurd Steinwachs, 1998), ther question ofrn,hether to administer antipsychotics for all patientswith early episocles is answerecl affirmatively, anddiscussion of alterriate irrteruentions is therebyclclsecl. We regarrcl this c:losure of inquirv iLS prerna-ture. Current Scandinar.ian projects involrdng in-honre family crisis inten'euti<-rn, avoiding use of hos-pitals ancl neuroleptics, ancl lrror,iding continuity ofteams and approach over an extended period havesliown highly promising result,s for the treatment ofpatients with newly diagnosecl psychosis (Lehtinenct al., 2000; Cullberg et al., 240D. 'Ihe St:andinar,'ianresults, Cliompi's Soteria replication (Ciomlri et al.,1992, 1993), and thcr finclings reported hele indicatet,hat,, contrary to lropular r,tews, mininral use rif an-

' ' Bola Jll, Nlosher LIt (in press) I)redicting chug-free treatnrent

l'{lsp()rrse iu ar:ute psyc'lrosis fron-r lhe Soteria prcr.je<1. Srltizophljhrll.

l l lr-rla JIl, Nlosher LIt (in press) L)lzrshing ider,rlogies ol sr:ieu-1ific rl iscorrrse'/ Sl'/ i i,zrtnltr Bull.

227

tipsychotic medications combined with specially de-signed llsyi:hosocial ir-rterventi<ln lor patients new-lvicientified lr.ith schizophrenia sJrectrum disorders isnot harmfrrl but appears to be aclvantageous.

In a well-known reattalysis of mr-rstly firsf,-episodeschizophrenia spectrum stuclies comparirrg antipsy-chotic medications vel'sus psvchosocial or urilieutreatment, Wyatt (1991) concludetl: "e:rly intelen-tion with neuoleptics in first-break schizophrenic pa-tienls increa*ses the likelihoocl of an improl'ecl lclng-terrn course" (p 325). ThLs conchsicln has cont,ributedto ertthusiasm for efforts to plevent psychosis tluouglr"early inten'ention" in the proclrorne, often with low-dose at,lpical antipsychotic medications. [{owel'er,most of the studies reviewerl by Wyatt (1991) n'ere ofa preexpedmenta.l (mirrorimage) design that did not,control nrany threats to internal r.alidity (Oa4tenter,1997). In t-act, a preponderance of the few availablequasiexperirnental or t,xlrerinrentally clesigned t.ar{y tepisode studies in u'hich olre group was initially not trnetlic:ated (Czrllenter et aL, 7977; Ciompi et nl., L992,199i ; Lehtinen et al., 2000; Mc.rsher zur<l Menn, 1978a;Rappaport et al., 1978; Schooler, 1967) show betterlong-temr outcomes for the unrnedicateci subjects. [nconcert u'ith the fuller presentatiotr of Sotena resultsheLe, these shrclies suggest that specially designed psy-chosocial iuter"u'tntion contbhred with minimal r-nedi-caticln use may be an effective treatrnent strategy forpafients with ein'ly episoile schiz<lpfuenizr spectrrurrpsychosis.

References

Ameri<:an Psyc'hiatric Associat ion (1 997) Pr':rr:t ico gui<lelines firrthe treatnrent of patients u'it lr sc:hizophrenia.,Ant. ,l Ps.t1r:lt itrtt-yI I ' r4: l -6 i ] .

IJleuler MIt (l!f78) TI're long-turrn course ol' sthizophrenic ps.v-c'hosis^ In LC Wyrrne, RL ('romvvell, S \{;rtt l iysse (lL<ls), 7'h,lttrrt?t.n; r2l' sthi,zoTtlt.rettin: 'Yeto upprort<'ltes to rcseo.rr:lt. u.trltreutrtettt (pp {j31*ti i}6). Neu' Yorl<: .lcihn Wiley antl Sons.

Bocklroven (1961]) X'Itttwl. t.t 'r 'utm.ntl itt Atnerir:an psl/c)lt. iutt '.!J.Npw' York: Springer.

l lreen R (1996) Regrerssiou rnodels: (lenst-rred, sarnplt- selectecl,trr trultcatt'cl data. Qu,urttitatiue apTtlicalion.s tn the sociu,lscie'rrr:s (pp 80). Beverl;v l l i l ls: Sage.

I.Juclianau .l (1995) Social supporl ancl sr.:hjzoplirenia: A ler.' iew' o1'the literature,. At'clt Psyt:hialr Ahrrs 9:(i8-76.

(- )arpenter W' l ' , I le indchs l ) l \ ' ( i9 iJ l ) ' f rpatnlent-re l tvanl st r t r -L.vpes of sc'hizophrtnia. ./ l let'u Xlent l l is lt i9:l l3*11!).

Carpenter \!T, N{cGlzsltan TH, Strauss JS (19771 The tleatmentr-rl' ac'ute schizophrenia ll'ithoLrt clrugs: .fur investigatir-rn ofsorlre current assurnptions. Arz .l Ps'gcltiuh y l:- ' i-1:14*?0.

(lzirpenler \.\,T (1$97) The risk of nredication-free resealc'h.lith i zophr B ul.l 23:ll-\8.

Cionrpi L (1980) Clattuuttestic long{ertn stutl.! ' on t,}re course oflife anct zrging ol sr:hizr-rplrrerrics. ̂!r'lr iz olsh r B ull 6:60(i - G tB.

(i iornpi L, Durvalclel H-P, Maier C, Aebi !1, 'fmtsch K, KLrpper Z,

Itulishauser Lr t.1!)l)l) The pilot proje<:t "Sot.eria Belne:" (' l ini-t ':rl experientes antl results. Ilt ' ,I Psyr:lt iutt",U 7(i1:145*153.

SOI'FIRIA:'I 'WO-Y FIAR Ot rT(lOlllF)S

K

:28 T}0IA ANI)

{' ir iurpi L, Kupper Z, Ael.ri E, Durvalder HP, Hul-rscluniclt T.'fmtsch K, RLritsharsel C (199.i) [The pilot ploject "SoteriaIlern" in tleatrnc'nt of acuter scltizophrenia patients: IL ResLrltsof a courparative prospective Iirl low-up pet'iod over'2 yeu'sl.I)ir Pilot-l 'ro.jt 'kt "Soteria lJern" zur trehandlnng akut s<:hizt>-pherrer: II. l lrgebnisse der vergleichenrlen prospektiven ver-lirrrfsslrrclie uber zrvei .iaht'e. i\'k"n;qrutr'l z (i4:440-450.

( 'ohen ( ' I , Sokolovsky , I (1978) Schizopl ' r renia ar t t i socia l net-wrrlks: F)x-patients itt the intrer cit,v". ,Sr,'ftreophr l lu,Ll "1:546-ir60.

(l<rJror-r .I (1f|87) S(ot,istir:ol p(nret' analSlsis .l in' the btlutt,ioretls c i e r rc e s ( tter,-iserl). I lillsdiile, N,J : lawr ence lirl trautn,{qsociates.

(lcrlrerr .1, (loiren I) (lt)8;]) .47tplirtl mullipk, reqre.ssion/t'on'r:l.rr-l ion a.nu,ll ls'r,s.l?rr" tlte beha'uitn'tt l sciettr:rt.s (2nd ecl, pp 545).I{i l lsdale. NJ: I-awrenr:e Erlbautn Associertes.

(loler J(), Klprrnrur (i1,, Goldberg S(1, NIMII PSCI Group (196{)I)lrtnrrllriiv,ine t,re:rtntt'nt in aclrt,t. scliizophrenia. At'tJt GenI'syr'h, i a tn1 10:24(l -20 L

(role.l( ), (ioldberg $(r, f)ar,is JIl (i90ti) Drr-rgs in the l.reatrnenl of

1rs3,<'lrosis: (lontlollerl studies. Itt P. Solornr>n (Erl;, Pstlthiulr-itrlttrrls: Iltttt'eerlittgs oj tt t't,:st.ttt't:h cottfo'etu'e lLeLd irt Brtsl,ott(pp 1l>ii-I,90). Nerv York: (irLrne ancl Slratton.

(irrl lbt'rg J, l,evzndt)r S, Ilolmquist R. N,lattson 1\{, Weiselgran l-N{(2002) One-year outcr)rne in l irst t 'piso<ie psychosis palients inl.he Swe<lisli Palac:ltute Pro.jer:t. Arto [)stlthi.utrico Stondi-t Lu tti r: i l l0(i:27ti -2|lr.

I)avis Ji\{, Barter..} 'f. Kane Jl'I (1989) Arttipsychotic di'ugs. In HiKaplarr, Iil Sadock (Eds), Corrtptvlrctxiur,: tefibook o.l"rtst1chi-t t t ty \pp l i i9 l -1626). I la l t i rnort : Wi l l ianrs and Wi lk ins.

I)i-xon Lli, Lehmtrn AF, Ler' ' iue J (199i:) Oonventional antipsy-chotic nrerlications for schizoplrrenia. Sr:lti iolth.t' Ru Ll ZI:it(t7 -

57iFlri<rkson [)II, Beiser I\ '1, Iacono WG (199U) Social support pre-

clirt.s ii-.year outcclrie in lirst-episode schizopluetria. J Abnot"tttPs grlwl. 1 07:68 1 -- 6ti5.

Fenlon WS, ISlvler' OR, lleinssen RK (1997) l)ett.nninants ofnrcrlicatiou compliau<:e in schizophrenia: Iitnpiric'al itnd clini-t-af l inil ings. &'h izophr Iltrl l 2i):6i]7*651.

l '-r 'an<:es A. l)o<:her1.y .IP, Kahtr DA (1996) The expert (:olrsensusgirit lelines series:

'I 'realment of schizoplu'enia. ,I Clirt Ps'gchi-

tttt 'u rt7:5-58.( i reene WII (191)t i ) LM(l ) i l | ' : Vers iot t , 7.0 user 's nta,nu,ol . (Re-

visecl, pp l)3ir). Plainview, N)': iJconouretric Soltwar"e Iuc.llec'knran .l.J (19?9) Siunple selection biirs as :r specilicatiou

en'or'. Eto trnur( t ' iut 47 :l ir3-161.l lellervell .JS (1999) Tleat,ment-resistanl schizophrenia: Revieu--

ing tire options zurci iclent.il'ving lhe wa.y fonvard. ,l Clirr Ps'g-c l t i u l t ' g 60 :1 .1 -19 .

ll i lschl'eld RMA, lt:rtt l iews SNI, Mosher LR, Menn AZ (1977)I3ering with rnatJness: l)erst)n:rlitv c'hruit<:teristi<'s of tlu'ee freat-nierrt stirf fs . fI o s 7t ( i o t n n t u n i.t u P s !tr: h i a t n1 28:26i -27 :,).

I loll irrgshearl AI3 (191-r7'l T'no .fttclot' [,nde,r: o.l ' sorial positiorL.Nen' l laven: August Hr-rl l ingshead.

Hul-rer Ci, Gross G, Sc:hrrttler R, Littz t'I (1980) Longitudinal stud-ie-^ ol' sch izop ltt'enic p atie'nt s.,5'ch i z o pl t, r Ifu l.l 6:it!)t- (t0t>.

.lablensky A, Saflorius N, FJrnberg (i, Anker NI. Xorten A, Cloopt'r ', lFl. I)av ll, I lerlelsen A (1!)1]2) Sc'hizolrhrenia: Manifestations,in<:ident.'e, (::rn ('ourse irr difl 'ertttt t 'ullures. A \\rorl<l l lealthOrganiz:rtion ien-countr.v stucly. I 'suclwl XIed frIonogr Suppl2 ( f : 1 -97 .

l;rirrg III) (lgiin The pulit'ir:s ol etytnieru?. New Yot'k: tsilliurtine.l,ell 'J. Sartorius N, Jablensky A, l(orlon A {1992)

' l '}re interna-tional pikit stutly of s<:li izophrenia: I"ive-year follow-up find-irrgs. Psryrlt til X'l r:d, 22 1:)1-l,I5.

Lehnran ,{F, Sl.einwachs DNI (19!)tl) The schizo;ihrenia pal,ientoutt'onres lesearch tearn ( 1'}()R'l) lreatnrent leconrlnendations.SLh izrtph r Ilu,ll 24:I-10.

lchtinerr V, Aaltonern J, l{ofl 'ert l ' , Rakkolaiuen V, SXvalahti F)(2f)00)'l 'rvt>year ortlr:ome in fir.st-eglisode psyr'ltosis treated:rccr.rrcling to :rn integrated tnotlel. Is itntnecliate neurolept.isa-tiori alrvays nee<letl? E ut' I)sgth iut t'.t1 l5:ilI2-320.

N{OSHI.]R

N{zrclsen AL, Keidliug N, Karle A, FJsbjerg S, Hemnringsen Il (1998JNeuroleptics in progressive struclural ltrain alrnonnzrlities irrpsychiatric illness [Researclt I etters]. La t'Lic et : J52 : 7ti.4 *781-i.

N{cKelve.y Rl), Zavoina W (1975) A statistical rnoclel lbr lhe:rnii lysis of or<linal level dependent variables. ./ Math SocirtlJ :10 ; l -120 .

tr'Ienrringer K (1959) A ps'qclti.ut,ri,st's 'u,rtt'kl: Tlte' ,selet'ted pttltrt"so.l Kurl l letttt ' inger'(pp 931), New York: Viking l)ress.

N{ot ts RI I (197.1) EuuLuol i t tg t realrn.et t t . enuintntn,er i l .s : ,4 ,sr tc iu lccoloylical tpprou'h. New York: Jt-rhn Wiley.

Mosher l, lt 42001) Soteria-Czrliftnnia iurd its su(:cess()rs; 'f lrera-

peuti<: ingredients. In l, Ciornpi, II lftrff'mann, I\{ I}roccarcl(Erls), W'ir:so tr:itkr: Srtlr riu?-eitt urryr:'tt,olt,nlicllr: schizupltt 'r,-tt iebtlutulLuttg unte.r tler lupr; (lUh:q tlr-tes Sotet' iu tuolk-urt.ttt L tt,s u rtl s t'h i z uph ret t i t l h e'nt'p g1 u,r t d e r c.uu m i t t u, t t,ot t) 1pp 13-,11). New York and lJerne, Switzerland: I{uber.

N{os}rer l,R, Ilendricks V, Parlicipants SI' (l9tl,1) [)abeisein: I)asmanual zur praxis in der Soteria ('I 'reatrnent at Soteria l loust':A rnanuiil lbl tlie pr':rclit'e of inte4rersonal psluronrenokrgy). lnD lrcrrcl lErl), P.si1r'ftrtsosoz'iule Ailteitsh Ll.l?,n

'?. Berlin: I'syr:hia-

trie-\ierlag.l\{oslier LR, Ntenn r\Z (.1971Ja) Cornnrunity residential t leatnrenl.

lcrr schizophrenia; 'l'rvo-yt

ar lilllou'-rrp. l {osp Oorn tnutti l tl ['sy,thi,atrv 29:7[,t-723.

Ntosher l,Il, il'Ienu M, (1978h) Ir)nhancing psyr:hosocial cr;rnprrtence in sr:hizopluerria: Prelitninary results of the Soteria prrrjecrt,In trT) Fzuur. IC Carc:a:i, A Pokorney, et zrl. (Eds;. IlNtbrrtettrioq.tJanrl hvalnrcnl uf sr:hizrtplLrutn.ia (pp ;-l7t-j186). Jiuuacia, NY:Spectrlm l>ublications.

Nloslrer l,R, I 'oll in W, Stabenau R ( 1Si71) lclentical t,rvins discor-dant for sr:hizophreuia; Neurologic' f indings. Arch. Gen Psur:lt i-u(.t'y 24:122-4!30.

lkrsher LR, Valk>tre R. Nletut A (1995) The treat,rnenl of ac'u1e

1-rsyt:ltosis wilhout treuroleptics: Sh vt,eek psyc:hopalltr-rklgyorr[conre datzr l i 'r,rm the Soteria project. lrtt J Soc Psythiotrv.11 : l57 - i 7 l l .

Neter ,J, Kutner b,Il l, Nachtsheirn (-,I, Wassennan W (191)(i) ,,11r-plie'd l i.rtear stutisti.r:nl nutd.els (.lth ed). Ohir::rgo: lnn'in.

PerryJW (1974,1 TIU:.lttt ' si,de of ntrtrLru:ss (pp 177) Englewor-rdLlliffs. NJ: f)renticre-Ha]l.

I)opp SM, ' lrezza

tlR (1998) Sicle efl 'et'ts of and leactious topsychotr:opic nrerlications. ltr IrM Kleespies {[)d), l lnrcrgctrciesin menlql heal t l t pract , ice (pp 271)-311; . Nerv York: ( iu i l lo ld.

Rappapor-t J\l. I lopkins IIh, l lall K, Ilellez:r'f, Silvemr:rn.l (1978)fu'e there s<:hizophenics for whonr drugs may be ume('ess:rryor <:ontt'aindi<:al ed? [ n ( Plttt rrn a ('ttp,sJJ(h i,u.l r"g 13:100 - 1 1 1.

Schooler NIl, Goklberg SC, iloothe H. (role JO (19ti7) One yearalter disclrarge: Conlltrurrity erclrrstnietrt r-rl 'sr-' i i izoplrrenic' prt-tients. Attt J Psyeh,iut.r"y 12:J:986*t)!)5.

Sernyak M.l, Leslie I)1,, Alitrcor-r Il l), [,os{rtrczy N,[F, Rosenheck li(2002) Associilt i<iu of rl iabetes nrell itus with use of at11ricii lneurolell l ics in the tre:rttnent of sc'lt izophrenia. Attt ,I Psur:lt, i-nInt 75!) : l ( j l -5[ i ( i .

Shadish WR, Ragsdale K (l9flti) Itanclorrr versus nonranclonr as-sigunrent in r-ontrr-rl led exper.itntnts: l)o .volr get. the satneanswer",' . I Co tt s ult (:L i,n Ps?Jchol ti,l: I l9tJ * I l]0;"r.

Sokis DA (1970) A brief fblk>w-uJr rating. (,-otttltt ' [)s3yhiu,tt"11I l:-14J-r-4t-r9.

Slolzenberg Rl\,{, Relles IIA (19!}7t Tools frrr inluil, ion abor.rl sarn-ple selection bias ancl its con'eclion. Atn Sr-u.' iri. I leu 62:49'1-507.

Strauss,lS, (larpr:nter WT ( 11)7tl) ' l 'he prognosis o1'sc:hizophreniit:I i irt ionale tbr a multit l inrensional (:on(:ept. SrhizoplLt' Bu,ll-1:f-r6 * 67.

Srrll ivan lls (1962) ,Srltizoltht'r:nid es u h?tltton pt-orrcs.s. Nen'York: Norton.

' l.homley 13, Adams (rFl, Awad G (2001) (lhiorpronrazine \'ersus

placebr-r lbr schizophrenia. Ccu'lt,rone I)ntabuse ,lqst Bcl2:( l l)0t)018.1.

Tr-rbin.l (1958) Bstirnal.ion of relationsldlrs f<>r l irniterl deperidenlvafialrles. Etrt ttrnrtetricn 2(i:J4-;l(1.

Wahlbeck K, 'Iuurtainen A. Ahokas A, Leucht S (2001) Dropout

rates in rarrrlomized iurl i l lsychotic <lurg trials. F'st1c ho ph arvm,-{:rtls11 O I fi5:3;10 -23J.

\\ 'endt I{J, I{osher LIl, Matttreu's SM, N1eutt AZ (1983) A contpar-ison of two lreattnenl enr-irontrtents ftrr sc'ltizophrenia. In JG(irrntlersr>n, f)A \Vil l, LII Mosher (Etls;, ' l ' |rc prittt: iples and

1tt'actic:es o.f nil ieu llterop'y (pp 17-ltl l). New Yr:t 'k: JasonAronson Inc'.

229

Mritaker It (2C)02) Motl i'rL Arnaricu: Rqd, sr.:ir:ttt'r:. Ltu.d rnedirittr',und lh.e endtoirtl l tni,stt 'eut,rntni cll th,e nutttu.l lry i l l. ( 'ani-

briclge, MA; Perseus }Jooks.Worrel JA, N{arken PA, Becknran SE. Ilr"reltter VL (2Cf00) Atypic'al

zurlipsy<:lrol.ic agents: A critical review. Am .I Hca,ltlt, SystPlta nn ;.-17;21]8 -Ji>8.

W.yatt RJ (Il)91) Neut'oleptics and llte tratural course oI'schizo-

lrhretriir. S<' I t iz r t y t I t r IJ ul I 17 :;i2it-lJb L

SOl'[iRIA:'l 'W0-YIJAR OI ]11(l( )i\ '1HS