Actuarial versus SPJ Risk Instruments with SOMMI - WI-ATSA · Actuarial versus SPJ Risk Instruments...

Preview:

Citation preview

ActuarialversusSPJRiskInstrumentswithSOMMIDAVID THORNTON

PRESENTAT ION AT THE ATSA 36 TH ANNUAL RESEARCH AND TREATMENT CONFERENCE , OCTOBER 26 TH

2017 , KANSAS C I T Y, M I SSOUR I

FinancialDisclosureIhavenodirectfinancialinterestsinthetopicscoveredbythispresentation

Imayonoccasionbepaidfeesforprovidingtrainingonrelatedtopics

Goals/ObjectivesParticipantswillbeabletodescribethepredictiveaccuracyoftheStatic-99/Static-99RwithSOMMIbyusingtheAreaUndertheCurve(AUC)statisticandtheprevalenceofSOMMIintheStatic-99Rnormativesamples.

ParticipantswillbeabletoprovidetheAUCrangeforstructuredandSPJmeasures

ParticipantswillbeabletodescribethreewaysMMIsymptomscanaffectsexualoffendingrisk.

BackgroundProfessionalsmaybecalledtoassesstheriskpresentedbymenwithahistoryofsexualoffendingandmajormentalillness(SOMMI)foranumberofreasons◦ Tofacilitateconcentratingresourcesonthosewhopresentthegreatestrisk◦ Todeterminewhethersomeabsoluteriskthresholdismet◦ Tofacilitateriskmanagement

Thetaskischallengingbecauseexistingtoolsdonotfitverywell◦ SOMMIareonlyaverysmallpartofthepopulationofmenadjudicatedforsexoffendingsotoolsdesignedforthatgroupmaynotfitSOMMI

◦ Similarly,toolsdesignedforthosewithMMI,eventhosewithMMIandahistoryofcrime,maynotfitwellsincethekindsofcrimeinvolvedaremoreusuallynon-sexualviolence

PotentialInfluenceofMMIonRiskFollowingKelley&ThorntonKelley,S.&Thornton,D.(2015)Assessingriskofsexoffenderswithmajormentalillness:integratingresearchintobestpractices.JournalofAggression,ConflictandPeaceResearch,7,258-274.

WecandistinguishthreepotentialinfluencesofMMIonsexualrecidivismrisk◦ MMIsymptomsreduceriskbydisruptingabilitytopursueoffendingintentions◦ MMIsymptomsareirrelevant◦ MMIsymptomscanworsenrisk

Actuarialinstrumentsdesignedtoassesssexualrecidivismriskpresently◦ DonottakeintoaccountthepresenceofMMI◦ Donotdistinguishthedifferentkindsofinfluenceitmayhave

ImplicationsThissuggeststhatactuarialinstrumentsmaybelessusefulwithSOMMIandraisesthepossibilityofSPJallowingbetterprediction

SPJallowsmoreflexibilityandSPJinstrumentsliketheSVR-20includeanitemforMMI

ThislineofthoughtmayexplainwhySPJismorecommonlyusedinforensicmentalhealthsettings

Buthowwellfoundedisthis?

Outline◦ ReviewofStatic-99/RpropertieswithSOMMI

◦ SOMMIinnormativesamples?◦ AUCinsampleswithmoreSOMMI

◦ ReviewofSPJpropertieswithSOMMI◦ AUCinsampleswithmoreSOMMI◦ AUCgenerally

◦ Implications

PrevalenceofMMIin2009Static-99RNormativeSamples:Routine

STATICMeasuresinNormativeSamplesØOriginalStatic-99(Hanson&Thornton,2000):◦ Basedon4samples;onepsychiatricsampleincludedpsychosisasprimarydiagnosis◦ Frequenciesofdxnotdescribed,but…◦ Static-99predictedriskinpsychiatricsampleequallywhencomparedwiththeotherthreesamples(AUC=.67vsAUCs=0.65-0.73)

ØCurrentStatic-99R:◦ Natureofincludedpsychiatricdisorderswasnotspecifiedorreported◦ PrevalenceofSOMMIinnormativesampleofStatic-99RwaseitherunspecifiedorknowntobelowinsomeHRHNsamples(under15%).

◦ Static-99R(aswellasStatic-2002R)bothpredictiveofrecidivism(AUCsin0.7s)inindividualswithpsychiatrichistoryasdefinedbyDSPproject

(Helmus,2012)

SummaryofActuarialResultsInpopulationsidentifiedasMIorhavingahigherconcentrationofMI:◦ Static-99AUC0.67(original)◦ Static-99AUC0.65Craissati &Blundell(2013) mentallydisorderedsexoffendersplacedincommunity◦ STABLE-2007AUC0.63Craissati &Blundell(2013)◦ Static-99R0.74DSP(Helmus,2012)psychiatrichospitalization◦ Static-2002R0.73(Helmus,2012)psychiatrichospitalization

◦ AUCsrangedfrom0.63to0.74forstaticanddynamicactuarialinstruments

SPJThreeSPJmeasuresthatmayberelevanttoassessingriskintheSOMMIpopulation:

ØSexualViolenceRiskManagement- 20(SVR-20)(Boer,etal,1997)

ØRiskforSexualViolenceProtocol(RSVP)(Hart,etal,2003)◦ Bothofaboveintegratedynamicandstaticrisk;allowforanexplicitassessmentofindividual’smentalillness;andallowscliniciantogiveweighttomentalillnessinmakingfinalriskassessment

ØAssessmentandRiskManageabilityforIndividualswhoOffendSexually(ARMIDILO-S)(Boer,etal,2004)◦ Firstofitskindtoidentifyandassessindividualaswellenvironmentalfactors

NumbersofSPJRecidivismStudiesforSOMMISVR-20◦ 3sampleswithsubstantialproportionofSOMMI(56%,19%+,43%;2ofthe3studiesinvolvedcasesreferredforNGIassessment)

RSVP◦ NostudiesbutishighlycorrelatedwithSVR-20sopropertiesprobablytransfer

ARMIDILO-S◦ NostudieswithSOMMIbutseemstoworkwellwithDDsamples

AUCsforSPJInstruments

SummaryofSPJResultsRangeofAUCsforSVR-20withSOMMI◦ 0.48– 0.80◦ Median0.52

MedianAUCforallSVR-20Studiesis0.63◦ 0.80,0.83◦ 0.74,0.71◦ 0.63,0.66,0.68◦ 0.59,0.58,0.52,0.58◦ 0.48,0.49,

SPJsseemtogivehighlyvariableresultsingeneralandwithSOMMI

ConcludingThoughtsWeproperlyhaveconcernaboutthevalueofexistinginstrumentswiththemostsevereMMIpresentationswheresymptomsimpactriskindirectlyordirectly

TodatestaticactuarialinstrumentsseemtodookayinsampleswithsignificantnumbersofSOMMIbutstudiesofsevereSOMMIareabsent

SPJinstruments(oratleasttheSVR-20),althoughtheyexplicitlytakeintoaccountMI,seemtobelesssuccessfulinassessingsexualrecidivismrisk,eveninMIpops

Wemaywellneedtoaddsomethingtostaticactuarials forthemoresevereSOMMIbutSPJinstrumentsliketheSVR-20aremorelikelytomisleadthantoassistusinthis

IntroductiontoSOMMIProject

GoalsCharacterizetheSOMMIpopulation◦ DefinedasindividualswithahistoryofsexualoffendingandMajorMentalIllnessmeaningBipolarIorPsychoticDisorders

FindwaysofreliablyidentifyingdifferencesbetweenSOMMIsothatriskmanagementcanbebetterindividualized

Thisincludesspecifically:◦ DeterminewaysofreliablyidentifyingLTVsinSOMMIpopulation◦ StudyingtherelationshipofMMIsymptomstosexualoffending◦ Developingwaysofreliablycharacterizingthedifferentformsthisrelationshipcantake

Ultimategoalistobehelpfultocliniciansworkingwiththispopulation

SOMMIPopulation(s)WewillneedgoodsizedsamplesdrawnfromSOMMIpopulations

WewillneedtodrawfromSOMMIpopulationsaccumulatedinthedifferentinstitutionalcontextsinwhichcliniciansmightencounterthem

Thisisanongoingproject

SelectionofIndividualswithahistoryofSOMMIfromWISVPProgramAllWISVPsarere-evaluatedeveryyearandtheseevaluationsincludeDSMdiagnoses.TheSRSTCResearchUnitmaintainsspreadsheetscontainingthisdiagnosticinformation.WeidentifiedindividualsforthestudybasedonDSM-IVTR/DSM-5diagnosesindicatingBipolarIorPsychoticDisorders.

Wetriedtobemoreinclusive(someindividualshaddiagnosesthatvariedovertheyears)

WhenthecasefileswerebeingratedtheratingpsychologistidentifiedcaseswheretherewasquestionaboutwhethertheyreallywereSOMMI(aboutafifthwerequestionable)

Some55caseshavebeenratedwith30ofthesecasesbeingratedbytwoindependentraters

NotethatexactNsmayvaryforspecificanalysesdependingonmissingdata

SelectionfromtheBridgewaterPoolThesewereapoolofindividualswhoreceivedtreatmentatBridgewaterandforwhomRayKnighthadaccumulatedelectronicfiledata

DSM-IIIAxis1diagnoseshadbeenmadeandwereviewedthosetodeterminewhichparticulardiagnosesbestcorrespondedtomodernconceptionsofPsychosisorBipolarI

WhenthecasefileswerebeingratedtheratingpsychologistidentifiedcaseswheretherewasquestionaboutwhethertheyreallywereSOMMI(aboutathirdwerequestionable)

Some25caseshavebeenratedsofarwith20ofthesecasesbeingratedbytwoindependentraters

Whatfollowsareprogressreportsonthedifferentfocioftheproject