The Addiction Behaviors Checklist: Validation of a New ... - Westat

342 Journal of Pain and Symptom Management Vol. 32 No. 4 October 2006Original ArticleThe Addiction Behaviors Checklist: Validationof a New Clinician-Based Measure ofInappropriate Opioid Use in Chronic PainStephen M. Wu, PhD, Peggy Compton, RN, PhD, Roger Bolus, PhD,Beatrix Schieffer, PhD, Quynh Pham, MD, Ariel Baria, MSN, Walter Van Vort, MD,Frederick Davis, MD, Paul Shekelle, MD, and Bruce D. Naliboff, PhDGreater Los Angeles Veterans Affairs Healthcare System (S.M.W., B.S., Q.P., A.B., W.V.V., F.D., P.S.,B.D.N.); Acute Care Section, School of Nursing at UCLA (P.C.); and Center for Neurovisceral Sciencesand Women’s Health (S.M.W., R.B., B.S., B.D.N.), Department of Medicine (A.B.), and Departmentof Psychiatry and Biobehavioral Sciences (B.D.N.), David Geffen School of Medicine at UCLA, LosAngeles, California, USAAbstractThis study introduces the Addiction Behaviors Checklist (ABC), which is a brief (20-item)instrument designed to track behaviors characteristic of addiction related to prescriptionopioid medications in chronic pain populations. Items are focused on observable behaviorsnoted both during and between clinic visits. One hundred thirty-six consecutive veterans ina multidisciplinary Veterans Affairs Chronic Pain Clinic who were receiving long-termopioid medication treatment were included in this study. This study represents one of the firstto follow a sample of chronic pain patients on opioid therapy over time, using a structuredassessment tool to evaluate and track behaviors suggestive of addiction. Interrater reliabilityand concurrent validity data are presented, as well as a cut-off score for use in determininginappropriate medication use. The psychometric findings support the ABC as a viableassessment tool that can increase a provider’s confidence in determinations of appropriate vs.inappropriate opioid use. J Pain Symptom Manage 2006;32:342e351. Ó 2006 U.S.Cancer Pain Relief Committee. Published by Elsevier Inc. All rights reserved.Key WordsChronic noncancer pain, opioid medications, substance abuse, addiction, medication misuseThis study was supported by VA Health Services Researchand Development.Address reprint requests to: Bruce D. Naliboff, PhD,Center for Neurovisceral Sciences and Women’sHealth, VAGLAHS, Building 115, Room 223,11301 Wilshire Boulevard, Los Angeles, CA 90073,USA. E-mail: naliboff@ucla.eduAccepted for publication: May 2, 2006.Ó 2006 U.S. Cancer Pain Relief CommitteePublished by Elsevier Inc. All rights reserved.IntroductionDiagnosing addictive disease in patients withchronic nonmalignant pain has proven to bea clinical challenge. 1e3 Standardized diagnosticcriteria for opioid addiction 4 in pain-freepopulations have proven to be less than validor difficult to apply in the context of chronicpain and therapeutic opioid prescription. 5e9Acknowledging the difficulties inherent inidentifying addictive disease in patients for0885-3924/06/$esee front matterdoi:10.1016/j.jpainsymman.2006.05.010

Vol. 32 No. 4 October 2006 Addiction Behaviors Checklist343which opioids are therapeutically prescribed,a consensus statement was developed by theAmerican Academy of Pain Medicine, theAmerican Pain Society, and the American Societyof Addiction Medicine in 2001 that providedinitial definitions for addiction inchronic pain populations. Addiction is definedin this public policy statement as ‘‘a primary,chronic, neurobiological disease, with genetic,psychosocial, and environmental factors influencingits development and manifestations’’and ‘‘characterized by behaviors that includeone or more of the following: impaired controlover drug use, compulsive use, continued usedespite harm, and craving.’’ 10Thus, in the context of chronic pain treatment,addiction is determined by the presenceof certain observable patient behaviors, the descriptionof which are nearing clinical agreement.For example, ‘‘behaviors suggestive ofaddiction’’ identified in the above consensusstatement include an inability to adhere tothe prescription schedule, insistence on certainforms or routes of medication, and resistanceto other nonopioid treatments. 10Portenoy and Payne 11 listed 10 aberrantdrug-related behaviors suggesting an addictiondisorder: selling prescription drugs; forgingprescriptions; stealing drugs; injecting oral formulations;obtaining drugs from nonmedicalsources; concurrently abusing alcohol or otherdrugs; evidence of deterioration in function;and failing to comply with dosing regimen, losingmedications, and/or seeking prescriptionsfrom other medical sources on multipleoccasions.In their analysis, Chabal et al. 8 focused onmedication-related issues, including an overwhelmingfocus on opioid issues, using supplementalsources of opioids, and patterns ofearly refills, multiple telephone calls, unscheduledvisits, and episodes of lost or stolenprescriptions, as behaviors indicative of problematicopioid use. These medication-relatedbehaviors are consistent with those predictingfuture opioid medication misuse in theconcept mapping analysis of Butler et al. 12Modeled after the consensus statement criterianoted above, Savage 3 categorized many of thesame behaviors as either due to the consequences,loss of control or compulsivity, andcraving or preoccupation related to medicationuse, with an emphasis on behaviors thatcan be detected during the visit, such as sedatedappearance, a preference for opioidswith high reward value, and reported poor tolerancefor nonopioid and nonpharmacologicalinterventions. Adding patients’perceptions of their medication use, Adamset al. 13 found that patients who believe theyneed a higher dose of medication than thatprescribed, report having difficulty gettingthe medication they need from the physician,and worry that they may be ‘‘too dependent’’on the medication were those most likely tosuffer addiction.The development of valid instruments fortracking behaviors characteristic of addictionin the context of opioid medications prescribedfor patients with chronic pain is animportant step in improving clinicalcare. 1e3,13e15 The Addiction Behaviors Checklist(ABC) was, therefore, developed in an effortto organize and provide an operationaltool for use by clinicians to monitor ongoingand current behaviors characteristic of addictionin patients with chronic pain and treatedwith opioid analgesics on a continuing basis.The majority of tracked behaviors includedin the ABC are consistent with those behaviorssuggested within the consensus definition bythe American Academy of Pain Medicine, theAmerican Pain Society, and the American Societyof Addiction Medicine. 10The aims of this study were as follows: 1) introducethe item development and the specificitems of the ABC, 2) examine interrater reliability,3) examine indicators of validity forthe sum score from the items as well as individualitems, and 4) introduce results regardingthe sensitivity and specificity for cut-off scoresfor the ABC. This study improves upon previousefforts in that ongoing, prospective datafrom both the ABC and other validity indicatorsare collected across multiple time points,enabling cross-validation of the ABC withother measures of problematic behaviors.MethodsParticipantsThis study included 136 chronic pain patientswho were being prescribed long-termopioid medications to treat their chronic painproblem. Participants were veterans recruitedfrom a chronic pain clinic at the Greater Los

344 Wu et al.Vol. 32 No. 4 October 2006Angeles Veterans Affairs (VA) HealthcareSystem.This study was approved by the local institutionalreview board. Data were collected as partof a larger data collection effort following patientswho were receiving long-term opioidmedication treatment over the course of oneyear. Participants visited the chronic pain clinicon a monthly basis in order to refill their opioidprescriptions. During the course of thesevisits, participants received various assessmentsreviewing their levels of pain over the course ofthe month and other areas related to potentiallyproblematic medication use. Participantswere asked to give signed consent prior to enrollmentinto this research study.MeasuresAddiction Behaviors Checklist. The developmentof ABC items was based upon an extensivereview of the relevant literature, 3,7,8,10,16,17 withan overriding aim to keep the tool concise,easy to complete, and otherwise useful ina busy clinical setting. Integrating observablebehaviors identified in and across the work ofpublished experts on indicators of addictionin chronic pain patients resulted in the synthesisof items for the ABC (Appendix). The fullset of 20 items initially selected was used inall the analyses presented in this paper. Allbehaviors tracked in the ABC reflect thosesuggested in the consensus document by theAmerican Academy of Pain Medicine,the American Pain Society, and the AmericanSociety of Addiction Medicine. 10The tool was designed to be administered inan interview format and scored based uponthe participant’s responses to questions, the interviewer’sobservations of displayed behaviorsduring the session, and pertinent informationgathered from medical chart review. Itemswere classified into two major categories: 1) addictivebehaviors noted between visits (e.g.,‘‘patient running out of medications early,’’‘‘receiving narcotics from other providers’’)and 2) addictive behaviors observed withinthe visit (e.g., ‘‘patient appearing sedated,’’‘‘patient expressing concern about future availabilityof narcotic’’). In addition to these twomajor categories, another question wasincluded that could be used if the patient’sfamily members were present in session(‘‘significant others express concern overpatient’s use of analgesics’’). Family memberswere rarely present within the setting of thecurrent data collection, therefore producinginsignificant influence on study results. Thisitem was retained in the instrument, however,because this information might be more availablewithin other medical settings. A total of 20items were included in the instrument, with dichotomousresponse scoring. Each affirmativeresponse was counted as one point, and pointswere added to calculate the total score, consequentlyresulting in scores ranging from 0 to20. While some of the items on the ABC areclearly based on patients’ reports and henceare subject to reporting bias, the primary emphasisof the scale and item scoring is on observablebehaviors or objective informationderived from the patient or other sources toindicate opioid misuse. Participants wereassessed on a monthly basis using the ABC.Prescription Drug Use Questionnaire. To evaluateconstruct validity for the ABC, the PrescriptionDrug Use Questionnaire (PDUQ) 16 wasadministered to all subjects at the 4-monthclinic visit. Unique in that scores have beencross-validated with clinical expert diagnosisof addiction, the PDUQ consists of 39 itemsevaluating five different domains of problematicopioid analgesic use in chronic pain patients,including the characteristics of thepain condition, opioid use patterns, social/family factors, familial/personal history of substanceabuse/addiction, and psychiatric history.Initial analyses of PDUQ scores providesuggested cut-off values indicative of the diagnosisof substance abuse and substance dependence.16 Since it includes historical data, thePDUQ was developed primarily to assess addiction(or addiction risk) at a single time pointand, therefore, is not a redundant measurewith the ABC. However, both measures dooverlap in their aim to measure the constructof current addiction behaviors and thusshould show moderate positive correlations.Global Clinical Judgment. A common criterionvariable in studies of opioid misuse is an expertclinician’s global judgment. 14,16 For thecurrent study, subjects were rated at each visitby their treating pain clinician on a dichotomousglobal question (‘‘Do you think patient

Vol. 32 No. 4 October 2006 Addiction Behaviors Checklist345is using medications appropriately?’’) assessingeither appropriate or inappropriate opioidmedication use over the past month. It shouldbe noted that the rating clinicians were familiarwith both the patients (for the most partpatients were routinely seen by only one ortwo clinicians over the course of the study)and with current concepts of addiction, pseudoaddiction,physical dependence, and tolerance.Participants were assessed on a monthlybasis using the global clinical judgment.Opioid Medication Discontinuation for Noncompliance.An additional outcome measure wasthe occurrence of major medication contractviolations leading to opioid discontinuationover the course of the one-year study. Discontinuationdecisions were made as a group bythe multidisciplinary pain clinic treatmentteam based on results of urine toxicologyscreens, medication contract violations, andother factors, such as alcohol intoxication orinappropriate behaviors in clinic. The primaryreasons for discontinuing opioid medicationswere as follows: 1) alcohol or illicit substanceabuse, 2) noncompliance with prescribed medications(e.g., repeatedly running out of medicationsearly, seeking opioid medications fromother sources), and 3) noncompliance withclinic procedures (e.g., refusing to submiturine toxicology screens, repeatedly missingappointments, and expecting medication refillsvia phone request). Opioid medicationdiscontinuation decisions were not based specificallyon scores of the ABC, PDUQ, orGlobal Clinical Judgment evaluations.Visual Analog Scales (Pain Scores). At eachmonthly visit, patients received an assessmentof both their usual level of pain and their worstlevel of pain since the last visit. This was measuredusing a 10 cm visual analog scale, rangingfrom 0 (no pain) to 10 (highest level ofpain imaginable).Interrater Reliability ProceduresInterrater reliability estimates for the ABCwere determined for two separate samples ofpatients during the study (Sample 1, n ¼ 23;Sample 2, n ¼ 19). For each sample, the samepair of raters completed the ABC for thesame individuals during their visit, but eachrater was blind to the other’s scoring. Visitsto be rated were based on consecutive clinicappointments without preselection of patientsto be rated and regardless of position withinthe subject’s study year.ResultsPatient DemographicsMean age was 53 years (range 25e65 years).The study sample included 8 female and 128male participants. The high male to female ratioof study participants is typical of VA populations.Based on initial general categorization,105 participants had a primary musculoskeletalpain problem, 26 participants had a primaryneuropathic pain problem, and the remaining5 participants had multicategory pain problemsor it was unclear what category the primarypain problems fit into. Regardingmarital status, 53 participants (39%) were marriedor living with their partner, 19 participants(14%) were never married, 55 participants(41%) were divorced or separated, and 8 participants(6%) were widowed. Concerning employment,17 participants (13%) were workingfull time, 8 (6%) were working part time, andthe remainder (107 participants, approximately81%) were not working. Of the 107 participantswho were not working, approximately46% (48 participants) indicated that they wereunemployed or unable to work due to pain.ReliabilityInterrater Reliability. ABC scores for Sample 1(Raters 1 and 2) ranged from 0 to 10(mean ¼ 2.74, SD ¼ 2.53), and for Sample 2(Raters 2 and 3) ranged from 0 to 9(mean ¼ 1.93, SD ¼ 2.13). The Pearson correlationcoefficient between Raters 1 and 2 was0.94 (n ¼ 23, P < 0.01), and between Raters 2and 3 was 0.95 (n ¼ 19, P < 0.01). These resultsshow strong interrater reliability for the ABC.ValidityIndividual Item Validity. Table 1 displays correlationsbetween item responses and providerglobal clinical judgment of problematic opioiduse. These correlations were based on ABC assessmentstaken following the first four to fivemonthly visits, providing enough time for providersto more accurately make determinations

346 Wu et al.Vol. 32 No. 4 October 2006Table 1Correlation Between ABC Individual Itemsand Concurrent Global Clinical Judgmentat Approximately 4 MonthsAfter Start of Study (n ¼ 101)ABC ItemPearson Correlation Coefficient1 0.182 0.30 a3 0.54 a4 0.44 a5 0.64 a6 0.31 a7 0.0689 0.0210 0.0711 0.52 a12 0.25 b13 0.35 a14 0.25 b15 0.54 a16 0.21 b17 0.25 b18 0.51 a19 0.58 a20 0.04a P < 0.01.b P < 0.05.of patients’ opioid use patterns. As noted, Items2e6 and 11e19 displayed significant correlationswith the interviewer’s global clinical judgment.It should be briefly noted that Table 1does not include information on Variable 8,‘‘patient bought meds on the street,’’ as thisitem was not endorsed across all subjects duringthis initial period of follow-up visits. Over theyear follow-up in general, this item shows highlyinfrequent endorsement (see Discussion).Concurrent ValiditydGlobal Clinical Judgment.Using the provider’s global clinical judgmentas a criterion variable to operationalize appropriatemedication use, initial t-test resultsindicated that the mean ABC scores can significantlydistinguish between patients whodisplay appropriate use of opioid medicationsand patients who display inappropriate use ofopioid medications. Mean ABC scores werehigher for inappropriate users (mean ¼ 5.31,SD ¼ 2.96) than appropriate users (mean ¼ 1.00,SD ¼ 1.27, t(16) ¼ 5.75, P < 0.001).Concurrent ValiditydPDUQ. Correlations weredetermined between mean ABC scores andPDUQ total scores. Mean ABC scores are theaverage ABC score across the initial 4e5months. The Pearson correlation coefficientbetween mean ABC scores and the PDUQ totalscore at this time point was 0.40 (P < 0.01).Sensitivity, Specificity, and Determinationof Cut-Off ScoreUsing the global clinical judgment as the criterionmeasure for inappropriate opioid use,measurements of sensitivity and specificitywere determined comparing the cross-tabulationmeasurements of ABC total score vs. globaljudgment rating. Fig. 1 shows a graphical representationof sensitivity [true positives/(true positivesþ false negatives)] and specificity [truenegatives/(true negatives þ false positives)]measurements for different ABC total scores usingglobal clinical judgment as the criterionmeasurement of inappropriate opioid use. Becauseof a relatively small number of endorsedopioid misuse behaviors at any specific visit,ABC monthly data were combined across allvisits (n ¼ 998) in order to maximize calculationof sensitivity and specificity. These results indicatethat a cut-off score of 3 or greater on theABC shows optimal sensitivity and specificity indetermining whether a patient is displaying inappropriateopioid use. To evaluate the impactof combining data, calculations of sensitivityand specificity at the initial visit time pointwere also performed and reproduced the samefindings of optimal sensitivity/specificity ata cut-off score of 3 or greater (e.g., using ABCdata from initial visit only, sensitivity ¼ 87.50%and specificity ¼ 86.14%).To further evaluate the utility of this cut-offscore, t-tests were performed comparingPercent120100806040200SensitivitySpecificity0 1 2 3 4 5 6 7 8 9ABC Cutoff ScoreFig. 1. Sensitivity vs. specificity based on varyingABC total score. Criterion variable ¼ Interviewer’sGlobal Judgment (n ¼ 998). ABC data collectedacross all visits.

Vol. 32 No. 4 October 2006 Addiction Behaviors Checklist347differences in PDUQ scores between participantswho received mean scores of 3 or greateron the ABC across the first 4e5 months andthose with mean scores below 3. Mean PDUQscores were higher for those participants witha mean ABC score greater than or equal to 3(mean ¼ 11.77, SD ¼ 3.99) than those withmean ABC scores less than 3 (mean ¼ 8.52,SD ¼ 4.05, t(86) ¼ 2.97, P ¼ 0.004). A similart-test was performed comparing participantswho received any ABC score greater than orequal to 3 over the first 4e5 months with thosewho consistently had scores less than 3 duringthis period. Mean PDUQ scores were againhigher for those participants who had at leastone ABC score greater than or equal to 3(mean ¼ 10.17, SD ¼ 4.35) than those who consistentlyhad scores less than 3 (mean ¼ 8.02,SD ¼ 3.80, t(86) ¼ 2.46, P ¼ 0.016).Trends Over Final Four Visits of StudyOver the course of the one-year data collectionfollow-up period, 38 participants had theiropioid medication prescriptions discontinueddue to objective misuse criteria establishedwithin the clinic setting. These objective criteriaincluded receiving a ‘‘bad’’ urine toxicologyscreen (e.g., positive urine findings of drugs ofabuse, numerous occurrences of absence ofprescribed medications within urine toxicologyscreens) or refusing to comply with themedication contract used within the clinic(e.g., failing to submit to random pill counts,repeatedly deviating from prescribed opioidmedication schedule). It should be notedthat the clinic used a sophisticated urinescreening methodology, including high-performanceliquid chromatography of all samples,which was both sensitive and specific for individualopioid compounds. Given the initial developmentof the ABC, participants did nothave their opioid medications discontinueddue to elevated scores on the ABC, but wereonly discontinued from their medicationsdue to the previously mentioned objectivemeasures of opioid misuse. Of these 38 participants,11 participants had their opioid medicationsdiscontinued due to alcohol or illicitsubstance abuse, 21 due to noncompliancewith prescribed medications, and 6 due tononcompliance with clinic procedures.It might be expected that objective signs ofopioid misuse and opioid prescriptiondiscontinuation would be preceded by an increasingtrend in misuse behaviors. To examinethis, the ABC scores were analyzed forthe final three to four visits for each patient.Comparisons were made between those participantswho were dropped from the study forone of the above-mentioned offenses vs. thoseparticipants who either completed the study orhad to drop out of the study for nonproblematicreasons (i.e., need for surgery, movedout of town). It should be noted that the lastvisit in this data set could either be the participant’scompletion of the one-year follow-upvisits or the point at which the patient wasdropped from the study, due to either problemsin opioid use or other reasons. Fig. 2 displaysa graphical representation of the trendsseen between the mean ABC scores for thesetwo groups in the last four visits of the study.As can be seen in Fig. 2, the ABC mean scoregradually increased, approaching a totalmean score of 3, as the participants nearedthe final visit in which they were droppeddue to problematic opioid medication use.ABC mean scores for participants who eithercompleted the study or were dropped due tononproblematic reasons tended to remainfairly stable around a mean score of about1.2e1.3. Based on t-test analysis, equalMean ABC Score3.02.52.01.51.00.50.03 Visits PriorCompleted or Non-problematic dropDropped due to a problem2 Visits Prior1 Visit Prior*Final Visit* P < 0.05 for scores at final visitFig. 2. Comparison of mean ABC scores over thefinal four visits in the study (n ¼ 136). Participantswho were discontinued due to opioid misuse problem(e.g., problem displayed in urine toxicology,noncompliance with clinic procedures) displayedan elevated ABC score (P < 0.05) at final visit ascompared to participants who completed or discontinuedthe study due to nonproblematic reasons(i.e., need for surgery).

348 Wu et al.Vol. 32 No. 4 October 2006variances not assumed, there was a significantdifference between these two groups at thefinal visit (t(47) ¼ 2.40, P ¼ 0.021).Correlations Between ABC Scoresand Pain ScoresIn the presence of pseudoaddiction, onemight expect that ABC scores would tend tocovary with pain scores (i.e., as patient receivesadequate pain relief and pain scores decrease,pseudoaddictive behaviors would decrease aswell). Table 2 displays correlation coefficientsbetween mean ABC scores and both usualand worst pain scores at the quarterly visit.This analysis found no significant correlationsbetween ABC scores and pain scores. This findinglends support to the concept that increasesin ABC scores would be indicative of inappropriatebehaviors suggestive of addiction ratherthan a pseudoaddiction process.DiscussionThe goal of this study was to test the psychometricproperties of the ABC, a new instrumentfor tracking behaviors suggestive ofaddiction related to opioid medications inchronic pain populations. The ABC showsstrong interrater correlations across three separateinterviewers, supporting the reliabilityand objective nature of the individual itemsand total score. The ABC total score alsoshowed good concurrent validity in terms ofTable 2Pearson Correlation CoefficientsBetween Mean ABC Scores and Usual Pain Scoreat Quarterly Visit PointsABC Mean ScoreInitialVisit4-MonthVisit8-Month 12-MonthVisit VisitUsual pain, initial visit 0.12 0.05 0.01 0.05Usual pain,0.12 0.04 0.04 0.024-month visitUsual pain,0.1 0.08 0.01 0.058-month visitUsual pain,0.12 0.09 0.00 0.1212-month visitWorst pain, initial visit 0.12 0.10 0.09 0.10Worst pain,0.16 0.06 0.04 0.024-month visitWorst pain,0.13 0.04 0.05 0.058-month visitWorst pain,12-month visit0.23 0.06 0.10 0.13its relationship with global clinical judgmentsof appropriate opioid use, as well as validitywith an opioid misuse risk assessment instrument,the PDUQ, which itself has shown verygood preliminary external validity. 16 The sensitivityand specificity analysis indicates thata cut-off score of 3 or more on the ABC providesa good general estimate of appropriatevs. inappropriate opioid use. This cut-off scoreshowed initially positive results in differentiatingopioid abuse as assessed by the PDUQ.Comparison with Other MeasuresThe ABC can be distinguished from otherrecently developed instruments in this areain that it is specifically focused on longitudinalassessment and tracking of problematic behaviorsand not determination of risk. There arenow several published measures for use withchronic pain patients to assess risk of opioidmisuse or addiction. 13,14,16 These measures includehistorical information (e.g., history ofsubstance abuse), current nonopioid abuse behaviors(e.g., nicotine craving), and opioidmedication-related behaviors. The focus onhistorical information is critical for assessmentof risk since previous substance abuse can bepredictive of future opioid abuse. 15,18 The natureof these scales, however, makes them lesslikely to change over time and they are notdesigned to track signs of misuse or flagbehaviors that signal increasing medicationproblems. Additionally, even though certainpredictors have suggested an increased riskfor future opioid abuse, the data indicatethat the prediction models are not highly robustand presence of these predictors doesnot provide certainty that chronic pain patientswould misuse their opioid medicationsin the future. For example, Compton et al. 16found that in a sample of 52 chronic pain patientsevaluated for opioid addiction, 27%and 50% of those who did not meet addictioncriteria still had a positive personal history andfamily history of addictive disease, respectively.This suggests that history variables, while cautionary,may not be clear enough indicatorsto withhold or modify opioid treatment decisions.On the other hand, the ABC representsthe first measure specifically designed and validatedfor longitudinal assessment of behaviorssuggestive of addiction. Because addictivedisease is essentially a disease of behavior, we

Vol. 32 No. 4 October 2006 Addiction Behaviors Checklist349assembled an assessment tool relying on behavioralobservations that are both logicallyconsistent with accepted definitions of opioidaddiction and relatively objective (the behavioris either present or not present).Importance of Specific Behaviorsin AddictionAlthough there is general consensus regardingthe primary behaviors that make up addictionin chronic pain patients using opioids,there is as yet no diagnostic algorithm for compilingthese behaviors. In this context it is interestingto examine how individual ABC itemsrelate to expert global clinical judgments in orderto provide initial direction for the weightingof certain behaviors in the process ofidentifying addiction. The items that weremost frequently endorsed and most associatedwith the global clinical judgment included ‘‘difficultywith using medication agreement,’’ ‘‘increaseduse of narcotics (since last visit),’’‘‘used more narcotics than prescribed,’’ and‘‘patient indicated that s/he ‘needs’ or ‘musthave’ analgesic meds.’’ These same behaviorshave been variously emphasized in the frameworksof Savage, 3 Chabal et al., 8 and Portenoyand Payne, 11 and were likely to also be endorsedon the PDUQ. 16 Further, each of thesebehaviors is consistent with consensus criteriafor addiction in pain patients. These indicatorsof loss of control over opioid use and cravingfor opioids in chronic pain patients appear towarrant increased and/or emphasized concernin addiction assessment. Although the abovementioneditems could be considered forheavier weighting when assessing for behaviorscharacteristic of addiction, a weighting systemwas not incorporated into the current analysis.A future direction for the ABC would be to explorepossible weighting systems or further-developedscoring paradigms that might giveadded emphasis to more serious opioid misusebehaviors predictive of addiction.Since the ABC is a checklist, it may also be importantto include low-frequency items eventhough they did not display a significant bivariaterelationship with the global clinical judgmentin the current sample. An extremeexample of this is the item assessing if the patientbought medications on the street. This informationis very difficult to gather fromobservable behavior or the medical record andis usually only available from a patient reportor trusted collateral report. Though this itemshowed extremely low endorsement, we wouldargue that it should be left in the checklist sincea positive endorsement would be a clear warningsign to providers regarding continued prescriptionof opioid medications. Other itemsmight have low endorsement rates that are populationspecific. For example, the items regardingsignificant others’ concern about use ofanalgesics and receiving medications frommore than one provider also showed a low correlationwith the global clinical judgment inthis study. Both of these items had low endorsementrates, but this is hypothesized to be sampledependent. In the VA population, families arenot as frequently involved as in other settings,and the HMO nature of care and the computerizedpatient record effectively limit a patient’sability to receive opioid medications from multipleproviders. These items may well be morefrequent and useful in other contexts.Another issue in the interpretation of an instrumentdesigned to track behaviors indicativeof potential addiction is the concept ofpseudoaddiction. 19e22 Pseudoaddiction includesbehaviors that are similar to addictionbut that are due to the undertreatment ofpain. Therefore, it is important to consider ifthe behaviors tracked with the ABC are indeedrepresentative of behaviors suggestive of addictionor due to the undertreatment of pain.Though it is an important theoretical consideration,the actual extent of pseudoaddiction hasnot been empirically validated. Within thechronic pain clinic at the Greater Los AngelesVeterans Affairs Healthcare System, physiciansand providers are cognizant of the iatrogenicissues involved in pseudoaddiction and arecareful to not make decisions regarding opioidmedication discontinuation or addiction basedsolely on patients’ requests or behaviors thatare due to an increased medication need dueto poorly controlled pain.Use of the ABCAlthough further validation and refinementis important, the current data suggest severalareas of utility for the ABC. It can be used clinically,as it was in this study, to track abuse behaviorsat clinic visits. Scores of 3 or aboveshould flag for more careful monitoring of patients,including more frequent urine screens

350 Wu et al.Vol. 32 No. 4 October 2006or increased medical chart review. Fig. 2, for example,shows some evidence for increasingabuse behaviors in visits preceding a majorproblem that led to removal of opioids. TheABC may be especially important in trainingclinics to cue clinicians to the range of possibleproblem behaviors and the need to examinespecific signs of misuse. The ABC may also bea good longitudinal outcome measure for researchstudies on effectiveness and risks of opioidmedications for chronic pain. A significantdifficulty with the existing literature is the lackof standardized measures of behaviors suggestiveof addiction, making comparisons acrosspopulations, medications, and settings very difficult.An advantage of the ABC in this regard isthat, in addition to reporting the total score,the pattern of behaviors suggestive of addictioncan be easily reported in tabular form.Study LimitationsThe present findings are based on data collectedin a single tertiary pain center in a largeurban VA Medical Center. While the sampledoes present with common pain problemsseen in most pain clinics, the results need replicationand extension in other contexts (primarycare settings, other pain clinics outside of theVA). One limitation of the study design wasthat not all patients who entered the pain cliniccould be included in the research study. A smallgroup of patients at the initial clinic assessmentwere found to already show significant problemswith opioid use. These patients were notrecommended to continue chronic opioidtreatment and were usually weaned off opioidmedications or voluntarily sought care elsewhere.The use of a sample of patients deemedappropriate for continued opioid use may haveresulted in fewer problems than the use of otherless restricted samples. Another limitation concernsthe use of a global clinical judgment asa major outcome criterion. While this measurehas high interrater reliability and care was takento make sure that raters were very familiar withthe patient’s past and present behaviors, it includesindividual clinician judgments and,therefore, may be prone to bias and both positiveand negative influence. In addition, anybias may be common to both the global judgmentand the ABC score leading to an inflatedvalidity estimate. The fact that the ABC was significantlyrelated to not only the global clinicaljudgment but also the discharge criterion andPDUQ moderates this limitation.SummaryIn summary, the ABC shows initial promise asan assessment instrument to quantify opioidmedication misuse in patients with chronicpain. It can be rated reliably across clinicians usinginformation typically available duringa clinic visit. The total score is significantly relatedto other misuse criteria including a globalclinical judgment and validated risk interview(PDUQ). The ABC can serve as a tool for providersto alert themselves when further interventionis needed, such as collecting randomurine toxicology screens or pill counts. It allowsfor more objective tracking of opioid medicationuse behaviors and changes in these behaviorsfor both clinical and research purposes.AcknowledgmentsThe authors thank Teresa Olivas for hereditorial assistance and the staff and patientsof the VAGLAHS pain clinic for their participationin this study. This study was supportedby the VA Health Services Research and Developmentprogram.References1. Portenoy RK. Opioid therapy for chronic nonmalignantpain: a review of the critical issues.J Pain Symptom Manage 1996;11:203e217.2. Passik SD, Kirsh KL, Whitcomb L, et al. A newtool to assess and document pain outcomes inchronic pain patients receiving opioid therapy.Clin Ther 2004;26:552e561.3. Savage SR. Assessment for addiction in pain--treatment settings. Clin J Pain 2002;18:S28eS38.4. American Psychiatric Association. Diagnosticand statistical manual of mental disorders (4thed.). Washington, DC: American Psychiatric Association,1994.5. Miotto K, Compton P, Ling W, Connolly M. Diagnosingaddictive disease in chronic pain patients.Psychosomatics 1996;37:223e235.6. Fishbain DA, Rosomoff HL, Rosomoff RS. Drugabuse, dependence, and addiction in chronic painpatients. Clin J Pain 1992;8:77e85.7. Sees KL, Clark HW. Opioid use in the treatmentof chronic pain: assessment of addiction. J PainSymptom Manage 1993;8:257e264.8. Chabal C, Erjavec MK, Jacobsen L, Mariano A,Chaney E. Prescription opiate abuse in chronic

Vol. 32 No. 4 October 2006 Addiction Behaviors Checklist351pain patients: clinical criteria, incidence, and predictors.Clin J Pain 1997;13:150e155.9. Harden RN. Chronic opioid therapy: anotherreappraisal. APS Bull 2002;12.10. American Academy of Pain Medicine, AmericanPain Society, American Society of Addiction Medicine.Definitions related to the use of opioids forthe treatment of pain. Glenview, IL, 2001.11. Portenoy RK, Payne R. Acute and chronic pain.In: Lowinson J, Ruiz P, Millman R, Langrod J, eds.Substance abuse: A comprehensive textbook. Baltimore,MD: Williams and Wilkins, 1997: 563e590.12. Butler SF, Budman SH, McGee MD, et al. Addictionseverity assessment tool: development of a self--report measure for clients in substance abusetreatment. Drug Alcohol Depend 2005;80:349e360.13. Adams LL, Gatchel RJ, Robinson RC, et al. Developmentof a self-report screening instrument forassessing potential opioid medication misuse inchronic pain patients. J Pain Symptom Manage2004;27:440e459.14. Butler SF, Budman SH, Fernandez K,Jamison RN. Validation of a screener and opioid assessmentmeasure for patients with chronic pain.Pain 2004;112:65e72.15. Michna E, Ross EL, Hynes WL, et al. Predictingaberrant drug behavior in patients treated forchronic pain: importance of abuse history. J PainSymptom Manage 2004;28:250e258.16. Compton P, Darakjian J, Miotto K. Screeningfor addiction in patients with chronic pain and‘‘problematic’’ substance use: evaluation of a pilotassessment tool. J Pain Symptom Manage 1998;16:355e363.17. Robinson RC, Gatchel RJ, Polatin P, et al.Screening for problematic prescription opioid use.Clin J Pain 2001;17:220e228.18. Schieffer BM, Pham Q, Labus J, et al. Pain medicationbeliefs and medication misuse in chronicpain. J Pain 2005;6:620e629.19. Weissman DE, Haddox JD. Opioid pseudoaddictiondaniatrogenic syndrome. Pain 1989;363e366.20. Gajraj N, Hervias-Sanz M. Opiate abuse orundertreatment? Clin J Pain 1998;14:90.21. Chabal C, Jacobson L, Mariano A, Chaney E.Opiate abuse or undertreatment? Clin J Pain 1998;14:90e91.22. Kirsh KL, Whitcomb LA, Donaghy K, Passik SD.Abuse and addiction issues in medically ill patientswith pain: attempts at clarification of terms and empiricalstudy. Clin J Pain 2002;18:S52eS60.AppendixAddiction Behaviors ChecklistInstructions: Code only for patients prescribed opioid or sedative analgesics on behaviors exhibited‘‘since last visit’’ and ‘‘within the current visit’’ (NA ¼ not assessed)Addiction behaviorsdsince last visit1. Patient used illicit drugs or evidences problem drinking a Y N NA2. Patient has hoarded meds Y N NA3. Patient used more narcotic than prescribed Y N NA4. Patient ran out of meds early Y N NA5. Patient has increased use of narcotics Y N NA6. Patient used analgesics PRN when prescription is for time contingent use Y N NA7. Patient received narcotics from more than one provider Y N NA8. Patient bought meds on the streets Y N NAAddiction behaviorsdwithin current visit1. Patient appears sedated or confused (e.g., slurred speech, unresponsive) Y N NA2. Patient expresses worries about addiction Y N NA3. Patient expressed a strong preference for a specific type of analgesic orY N NAa specific route of administration4. Patient expresses concern about future availability of narcotic Y N NA5. Patient reports worsened relationships with family Y N NA6. Patient misrepresented analgesic prescription or use Y N NA7. Patient indicated she or he ‘‘needs’’ or ‘‘must have’’ analgesic meds Y N NA8. Discussion of analgesic meds was the predominant issue of visit Y N NA9. Patient exhibited lack of interest in rehab or self-management Y N NA10. Patient reports minimal/inadequate relief from narcotic analgesic Y N NA11. Patient indicated difficulty with using medication agreement Y N NAOther1. Significant others express concern over patient’s use of analgesics Y N NAa Item 1 original phrasing: (‘‘Patient used ETOH or illicit drugs’’), had a low correlation with global clinical judgment. This is possibly associatedwith difficulty in content interpretation, in that if a patient endorsed highly infrequent alcohol use, he or she would receive a positive rating onthis item, but not be considered as using the prescription opioid medications inappropriately. Therefore, we include in this version of the ABCa suggested wording change for this item that specifies problem drinking as the criterion for alcohol use.