drug interactions and prescribing errors

DrugInteractionsandPrescribingErrors

JohnR.Horn,PharmD,FCCPProfessorofPharmacy,SchoolofPharmacy

AssociateDirectorUWMedicinePharmacyServicesUniversityofWashington

Seattle,WA

FinancialDisclosureRelatedtoPresentation

JohnHornisapartnerofHanstenandHorn,LLPwhichpublishesdruginteractionreferencebookssuchasTheTop100DrugInteractions:AGuidetoPatientManagement

JohnHorndoesnotcurrentlyhaveafinancialrelationshipwithanyone,butwouldcertainlywelcomeoneasheisnearingretirement.

DDITerminology• Drug-druginteraction(DDI)

• Aclinicallymeaningfulalterationintheexposureand/orresponsetoadrugthathasoccurred asaresultoftheco-administration ofanotherdrug.

• PotentialDDI(PODDI)• Co-prescriptionorco-administrationoftwodrugsknowntointeract,andthereforeaDDIcouldoccurintheexposedpatient.

DefinitionofTerms

•ObjectDrugthedrugthatisbeingaffectedbytheinteraction

• PrecipitantDrugthedrugcausingtheinteraction

• Interactioncaneitherbeuni-directionalorbi-directional(mutual)

TypesofDrugInteractions

• PharmacokineticDrugInteractionsThoseinteractionsthatresultinachangeintheconcentration-timecourseofactivedrug/metabolitesinthecirculationand/orattheeffectortissueororgan

• PharmacodynamicDrugInteractionsThoseinteractionsinvolvingachangeinthefunctionalrelationshipbetweenthedegreeofpharmacologicresponseandthedrug/metaboliteconcentration

PharmacyResponsetoSeriousDDIs•255prescriptionsfor1of5DDIswerepresentedtocommunitypharmaciesbyreportersfromChicagoTribune

•PrimaryoutcomemeasurewasverbalindicationDDIidentifiedtopatientorMD

•8Chainpharmacies;30tests/chain•32Independentpharmacies

PharmacyResponsetoSeriousDDIs:DrugPairs

Clarithromycin– ErgotamineVerapamil– ColchicineClarithromycin– SimvastatinCiprofloxacin– TizanidineGriseofulvin– OralContraceptive

PharmacyResponsetoSeriousDDIs:CorrectlyIdentifiedbyPharmacy

40 36.7 40

63.356.7

38.5

70

56.7

28.1

01020304050607080

Percent

PharmacyResponsetoSeriousDDIs:CorrectlyIdentifiedbyDDIPair

42.9

27.5

69.5

59.3

36.2

0

10

20

30

40

50

60

70

80

Clar/Ergot Verap/Colch Clar/Simva Cipro/Tizan Griseo/OC

Percent

WhatisthePurposeofaDrugInteractionScreeningProgram?

Toidentifypotentialinteractionsandinitiatestepstopreventpossiblepatientharm

“Predictionisverydifficult.Especiallyifit’s

aboutthefuture.”NielsBohr

Prediction

Fluconazole(Diflucan)+Warfarin(Coumadin)

7peopleonwarfaringivenfluconazole100mgdailyX7dMarkedincreaseinthePTresponse(buthighvariability)Nobleedingoccurred

0

10

20

30

40

50

60

70

1 2 3 4 5 6 7

% Increase in Pro-Time

mean

Patients

Crussell-Porter LL et al. Arch Intern Med 1993;153:102-104.

Factors Influencing Drug Interaction Outcomes

CLINICALOUTCOMEOFDRUG

INTERACTIONS

PATIENTFACTORS

DRUGFACTORS

Genetics

Diseases

Diet/Nutrition

Environment

Smoking

Alcohol

Dose

Duration

DosingTimes

Sequence

Route

DosageForm

HIGHVARIABILITY Adapted from Hansten. Science & Medicine. 1998;5:16-25.

ComputerizedDIScreening:Problems

• Thesheerenormityofavailableinfo• Lackofepidemiologicalinformation•Complexityofpatients,polypharmacy•Mayrelyonliteraturereportswithoutinformedrevieworevaluation

•Mayincludeinteractionsthatare‘over’classed

PhysiciansResponsetoComputerizedDrugInteractionAlerts

• 4751DDIalerts• 3129separatepairs:Overriderate:

• Level1(Serious/substantialevidence)89.4%• Level2(Lessserious) 96.3%• Level3(Serious/someevidence) 85.4%

• MDstriedtoreentersamedrugpair26%ofthetime

Weingart SN et al. Arch Intern Med.2004;163;2625-31.

AlertOverrides:JustaBadHabit?

•Prescribersrarelyreadalertsthatareoverridden

•Timefromalertpresentationtodismissal– 8seconds – independentofacceptanceoroverrideofalert

•Morealerts=moreoverrides

McDanielRBetalJAMIA.2016;23:e138-41

ProblemsinDDIInformationProcessing

•DataGenerators: Basicandclinicalscientistsdoingcontrolledstudies(PKusually),casereports,“bigdata”mining,labeling.Hugeamountofdataofvaryingquality.

•Problem– LargevolumeofDDIdataofvaryingquality.

HornJRandHanstenPD.PharmacyTimes.May2015

CilostazolDDI:Label1/15LovastatinCo-administrationofcilostazolwithlovastatinincreaseslovastatinandß-hydroxylovastatinAUCapproximately70%.Thisismostlikelyclinicallyinsignificant.

EffectofCilostazolonCYP3A4PLETALdoesnotappeartocauseincreasedbloodlevelsofdrugsmetabolizedbyCYP3A4,asithadnoeffectonlovastatin,adrugwithmetabolismverysensitivetoCYP3A4inhibition.

DataMiningtoIdentifySevereDDIADRs

• ExtractDDI-ADRpairsfromFDAAERSforwarfarin,clopidogrel,simvastatin

• UsedSideEffectResource(SIDER– basedonpackageinserts)toIDADRsassociatedwithdrugs

• Severitygrades1(mild)to5(death)usingcommonterminologycriteriaforADR(CTCAE)

• CombinedthesesourcestoextractDDIpairsandtheADRsassociatedwiththem

JiangGetal.BioDataMining.2015;8:12

DataMiningtoIdentifySevereDDIADRs:Simvastatin

Drug1 Drug2 CTCAEGrade ADEName

Simvastatin Aspirin 5Fatal Aortic Stenosis/Coma

Simvastatin Losartan 5 CardiacArrest

Simvastatin Risiglitazone(sic)

5 LBBB/Retinopathy

Simvastatin Tegretol 4Lifethreatening Aphasia/Arrhythmia/Dermatitis

Simvastatin Amiodarone 4 Bonepain

Simvastatin Aspirin 3Severe Dermatitis

Simvastatin Viagra 3 Glucosetolerance impaired

Simvastatin Ramipril 3 Intervertebraldiscprotrusion

JiangGetal.BioDataMining.2015;8:12

ProblemsinDDIInformationProcessing

• InformationAnalysts: ReadpapersfromDataGeneratorsandwritereviews,books,andelectronicdatabases.FewspecializeinDDIinformatics;variabilityinexpertise.

•Problem– ToofewanalystswithbothgeneralandspecializedknowledgeofDDIs.

HornJRandHanstenPD.PharmacyTimes.May2015

Theophylline– Allopurinol:AMajorDrugInteraction

• ListedashighestseverityinseveralCDSdatabases,includinginstitutionallycustomizeddatabases

• TwostudiesfoundnochangeintheoPKwithallopurinol300mgdailyx7days

•Onestudyfound~25%increaseinTheoAUCandhalf-lifeafter2weeksofallopurinol300mgBIDvsday1

Theophylline– Allopurinol:AMajorDrugInteraction

Theophylline

3-Methylxanthine

1-methylurinc acid

1,3 dimethyluric acid

1-methylxanthine35%1A2 3A, 2E140%

16%1A2

Xanthine Oxidase

ProblemsinDDIInformationProcessing

• ClinicalDecisionSupportProducers: RelyonInformationAnalysts’producttomakeCDSusedbyclinicians.DecidingwhichPODDIsandhowtopresentthemisdifficultandsomedoabetterjobthanothers.AllincludePODDIsofquestionableclinicalimportance.

• Problem– InclusionoffartoomanyPODDIsinCDS

HornJRandHanstenPD.PharmacyTimes.May2015

QTcIntervalProlongationinCardiacPatients

•900ptsadmittocardiacunits•Definitions:QTcprolonged:470ms/480msformales/females

•QT-prolongingmedsfromqtdrugs.org•AdmissionQTc~460ms

TisdaleJEetal.DrugSaf.2012;35:459

QTcIntervalProlongationinCardiacPatients

• OnadmissionQTcprolongedin28%;ofptsonQTPD,31.5%hadprolongedQTc

• 18%admitswithQTc>500ms• 35%ofptswithQTcprolongationreceivedQTPD• 42%ofptswithQTc>500msreceivedQTPD;57%ofthesehadsubsequentQTcincreases>60ms

• TotalnumberofcasesofTdP=ZERO

TisdaleJEetal.DrugSaf.2012;35:459

OccurrenceofQTc-BasedDDIs

ReviewedallDDIsclassifiedasMajorinEMRDDIdatabase:N=21,944

IdentifiedDDIsthatciteprolongedQTcasthebasisoftheDDIrisk:N=5,651

26%ofallMajorDDIsareduetoQTc-basedmechanisms.

AlertFatigue

“Themuchdiscussedphenomenonofalertfatiguesuggests

thatthereistoomuchinformationtoprocessinthetypical

workflowwhenitcomestoorderprescribing.Alertshave

oftenbeenconsideredtobeduplicative,lackinginpatient-

specificcontext,orjustgenerallyspurious.”

Matuszewski KA. Pharmacy and Therapeutics. 2012;37:69.

CausesofAlertFatigue

•Refills•Lowspecificityforclinicalsignificance•Basedonnon-clinicaldata•Desiretohave“complete”listing•Legalconcerns•“It’sinthelabel”

ReducingAlertFatigue:Options

•TurnoffselectedDDIcategories•Doin-housereviewofDIdatabasetoreviseexistingalerts

•KnowledgebaseDDIprogram•PatientspecificDDIalerts

WhatNOT ToDoAboutTooManyDDIAlerts

Turnoffselectedcategories(eg,allalertsnotin“mostsevere”group)orremove“reallyirritating”alerts

Advantage– VeryeasytodoDisadvantage– Tooeasytodo

AlertOnlyMostSevereDDIs:WhatYouWillNOT See

• Amiodarone/Haloperidol– PK/PDarrhythmia• Bepridil/Clarithromycin– PK/PDarrhythmia• Colchicine/Clarithromycin– Colchicinetoxicity• Conivaptan/Ergots– Ergotism• Cyclosporine/Ketoconazole– Renaltoxicity• Cyclosporine/Rifampin– Organrejection• Simvastatin/Clarithro,Erythro,CSA– Myopathy

Horn JR, Hansten PD. Pharmacy Times. 2011;77:38

WhatYouMightDoAboutTooManyDDIAlerts

Doin-housereviewandcustomizationofsomeorallofDDIdatabase

Advantage– usuallydonebycommitteeDisadvantage– usuallydonebycommitteewithoutadequateexpertiseortraininginDDIs

InHouseDDICustomization:Identified“CriticalInteractions”

•CCBs(all)– BBs(all)[duetoAVblock]•Digoxin- Amiodarone,Macrolides,Quinidine,Verapamil,Tetracycline(Butnotazoleantifungals)

• SSRIs– Triptans• Theophylline– Allopurinol,Febuxostat

InHouseDDIDatabaseCustomization:Guidelines

• Startbydefiningwhatshouldtriggeranalert• Establishcriteriaforseriousnessclassification•Developcriteriaforassessingevidence•Developsetofrulestoguideclassificationandincreasecontinuity

InHouseDDIDatabaseCustomization:WheretoStart?

•Reviewthealertsthatarecurrentlybeingtriggered

• StartwiththeDDIscausingthemostalerts•Decreaseseriousnessofthosealertsnotmeetingcriteria

•Usepreviouslycustomizeddatabaseasstartingpoint

CustomizedDIScreeningatUniversityofWashingtonMedicalCenter

• InpreparationforCPOE,customizeDIdatabase• StartwithDIslabeledasMAJOR• About8000uniquepairsinUWMC’sdatabase(2007)• Eachpairevaluatedforappropriateclassificationindatabase(Major– Moderate– Minor)basedondataanddefinedcriteria

Horn JR et al. Am J Health-Syst Pharm. 2011;68:662-4.

CustomizedDDIDatabase

• Evaluationcriteriawasprobabilityofpatientharmandeaseofavoidance

•MAJORshouldonlyincludeDDIswhererisklikelytoexceedbenefitinmostpatients

•MAJORDDIsshouldrequirethoughtandaction:changedrug,adjustdose,and/ormonitor

Horn JR et al. Am J Health-Syst Pharm. 2011;68:662-4

CustomizedDDIDatabaseHospital TotalMAJORsin

DatabaseDIs ReducedinSeverity(%)

UWMC(2007) 7970 57

Hospital2 8404 59

Hospital3 11727 65

Hospital4 12268 77

UWMedicine(2016) 26040 71

CustomizedDDIDatabase• Reducesnumberofinappropriatealertsandpotentialforalertfatigue

• Enablesselectionofalertsbasedoninstitutionalrequirements

• Needtoeducateusersonchanges• Needtoupdate• Noopportunityforpatientspecificalerting• Labor/expertiseintensive

KnowledgeBaseDrugInteractionScreening

•Doesnotuseasimplelook-uptable• Basedondrugproperties– PK,PD

•Mechanism(ie,CYP450)based• IDsubstrate,inhibitor/inducerofCYP450s•MagnitudeofeffectonCYP450• First-Passmetabolismofobjectdrug• Therapeuticwindowofobjectdrug

Boyce E et al. IEEE Trans Info Tech Biomed, 2007;11:386-397.

Mechanism-basedDrugInteractionKnowledgeBases

Advantages:CanpredictinteractionsnotpreviouslyreportedAvoids“class”errorsPredictsinteractionsresultingfromdrugwithdrawalSupportsinteractionsbetweenthreeormoredrugsProvidesestimatesofeffectonObjectdrug

Mechanism-basedDrugInteractionKnowledgeBasesDisadvantages:MechanismsmaybeunknownPKorPDinfofordrugsmaynotbeavailableDifficulttoevaluateclinicalriskUpdatingdrugassertionsinthedatabaseEvaluatingevidenceforassertions

PatientSpecificDDIAlerts

Cleanupdatabasefirst?Createalgorithm/decisiontreewithmitigatingandriskfactorsApplyusingpatientspecificEMRdatatodecidetoalert

CommonMajorDDIAlertsUWMedicine2016:AssessmentofDDIs

ReviewtheP’col andP’kinet propertiesofobjectandprecipitantdrugsIdentifymodifying(MitigatingandRisky)factorsDrug:Dose,Duration,Route,OrderofAdministration,Co-medicationsPatient:Disease,Renalfunction,Labvalues,ECG,Pharmacogenomics,Diet,Age,Gender

CommonMajorDDIAlertsUWMedicine2016

Amiodarone/OxycodoneAmlodipine/SimvastatinCiprofloxacin/OxycodoneDiltiazem/OxycodoneFluconazole/FentanylFluconazole/OxycodoneFluconazole/Tacrolimus

Makeup58%ofMajoralerts

DoseFiltersforCommonAlerts

Amiodarone/≤40mgOxycodoneAmlodipine/≤40mgSimvastatinCiprofloxacin/≤80mgOxycodoneDiltiazem/≤80mgOxycodoneFluconazole≤200mg/FentanylFluconazole≤200mg/OxycodoneFluconazole≤200mg/Tacrolimus

AssessmentofDDIs:Fluconazole

FluconazoleCYP3A4inhibitionisdosedependent

100mg/d– 20%incr CSAAUC150mg/d– 25-50%incr MidazolamAUC100mg/d– 2-foldincr TriazolamAUC200mg/d– 4-foldincr TriazolamAUC

Fluconazole/OxycodoneDDIDecisionTree

Fluconazoledose≤200mg/day

Yes Noalert

No

ICU?No

Scheduledoxycodonedose<80mg/d

Yes

Noalert

No

Alert

Yes

Noalert

Fluconazole/OxycodoneDecisionTree• 242alertsforfluconazole/oxycodone• Flucon dose≤200mg/d:169Noalert• Flucon dose>200mg/d:73• ICU:yes– 6Noalert• ICU:no– 67•Oxydose<80mg/d:37Noalert•Oxydose>80mg/d:30Alert• ~88%reductioninalerts

Fluconazole/TacrolimusDDIDecisionTree

TacrolimusIVFormulation

Yes Noalert

No

Fluconazoledose<200mg/day

Yes

Noalert

No

Alert

Fluconazole/TacrolimusDecisionTree

•159alertsforfluconazole/tacrolimus•TacrolimusIV:39Noalert•Flucon dose<200mg/d:43Noalert•Flucon dose≥200mg/d:77Alert•~52%reductioninalerts

KCL/PotassiumSparingDiuretic

No

Yes[K]w/in48h

[K]<4.5meq

No YesAlert

Alert

Yes

Alert

No

CrCl <30ml/min

Alert

Yes

No

ACEI/ARB

KCl ≥80meq/d

No

NoAlert

Yes

Alert

EffectofDDIFilteronAlerts

Alert N Sensitivity SpecificityKincreasingDDIs;alltriggeralert

6349 100 0

K>4.0mEq/lw/in48hr priortoDDI

2226 61.1 65.7

K>4.8mEq/l48hrpriorandduringDDI

1217 74.2 95.7

Eschmann Eetal.doi:10.3233/978-1-61499-289-9-1056

76KinpatientsanalyzedforDDIsresultinginK>5.4mEq/l

AlertReductionwithFilter:Low/HighDose

0102030405060708090

100

LowDose

HighDose

%reductioninalerts

DrugInteractionAlerts:ProblemsandSolutions

SummaryandQuestions

“Never,underanycircumstances,takeasleepingpillandalaxativeonthesame

night.”

Drug Interaction to Avoid