Skip to main content
. Author manuscript; available in PMC: 2017 Jun 6.
Published in final edited form as: J Pain. 2015 Feb 4;16(5):389–411. doi: 10.1016/j.jpain.2015.01.009

Table 1.

Psychometric data for the evaluated instruments

Instrument Evaluated (less recent to most recent) Reference Population (n) Measurement Properties
Patient-Reported Instruments (self-report)
Non-specific
DAST Skinner, 1982 [52] Pts seeking help for inappropriate drug or alcohol use (n=223) Internal consistency: α = 0.92. Statistically significant differences in DAST scores between participants with (1) only drug problems, (2) those with drug and alcohol problems, and (3) only alcohol problems, with participants in group 1 demonstrating the highest DAST scores, followed by group 2, then group 3. DAST scores positively and significantly correlated with frequency of drug use over 12 months and psychopathology.
Skinner & Goldberg, 1986 [53] Pts seeking help for inappropriate opiate use (n=105) DAST-20: Concurrent correlation between DAST and self-identified narcotics problem: r = 0.30. Principal components analysis yielded 5 factors, accounting for 55% of the variance.
Gavin et al., 1989 [25] Pts seeking help for drug or alcohol use (n=501) DAST-20: Strong concurrent correlations between DAST and current and lifetime DSM-III drug addiction diagnoses = 0.75 and 0.74, respectively, and between DAST and number of DSM-III drug-related symptoms in the past month. Moderate concurrent correlations between DAST and number of drug use days in the past week and number of drugs used in the past week. Small concurrent correlations between DAST and psychological health. Small and generally negative concurrent correlations with measures of alcohol use and abuse. ROC analysis yielded AUC of 0.93 (criterion is DSM-III drug abuse/dependence diagnosis). ROC analysis indicated cutoff score of “5/6” was optimal, yielding sensitivity of 96% and specificity of 79%.
Staley & El-Guebaly, 1990 [55] Pts in psychiatric treatment programs (n=250) Internal consistency: α = 0.94. Statistically significant differences in DAST scores between participants (1) with DSM-III substance abuse diagnoses in an outpatient substance abuse program and those in the (2) inpatient psychiatric program, (3) outpatient anxiety disorder program, and (4) day hospital. 75% of the participants in (1) had a DAST score of ≥ 6, compared to 28% in (2), 13% in (3), and 18% in (4). Principal components analysis yielded 5 factors accounting for 64% of the variance.
Kemper et al., 1993 [30] Mothers of children < 6 in pediatric clinics (n=507) DAST-20: Mothers who indicated they used drugs within the past 24 hours were not identified by the DAST.
Saltstone et al., 1994 [50] Females on probation or incarcerated (n=615) DAST-20: Internal consistency: α = 0.88. Initial principal components analysis yielded 5 factors; 1 factor accounted for <1% of the variance, leading to a principle components analysis constrained to 4 factors that accounted for 56% of the variance. Including DAST items and alcohol abuse items into a principal components analysis demonstrated that DAST items loaded independently.
El-Bassel et al., 1997 [20] Participants in an employee assistance program (n=176) Internal consistency: α = 0.92. 2 week test-retest reliability in 20 participants: r = 0.85.
Principal components analysis (using 26 of 28 items) yielded 5 factors, accounting for 63% of the variance. 93% of self-reported current drug users scored ≥ 6 on the DAST; 74% of self-reported former users scored ≥ 6; 6% of self-reported non-users scored ≥ 6.
Cocco & Carey, 1998 [15] Psychiatric outpatients (n=97) DAST-10: Internal consistency: α = 0.86. 3–10 day test-retest reliability: ICC = 0.71. Principle components analysis yielded 3 factors, accounting for 64% of the variance. Moderate concurrent correlations with questions related to drug use. Cutoff scores between 1 or 2 and 3 or 4 were associated with acceptable to high sensitivity levels, but low to acceptable specificity levels (criterion is DSM-III-R drug use disorder diagnosis).
DAST-20: Internal consistency: α = 0.92. 3–10 day test-retest reliability: ICC = 0.78. Principle components analysis yielded 6 factors, accounting for 71% of the variance. Moderate concurrent correlations with questions related to drug use. Individuals with current drug abuse diagnosis had significantly higher DAST scores than those with a prior drug abuse diagnosis or no history of drug abuse. Cutoff scores between 2 or 3 and 5 or 6 were associated with acceptable to high sensitivity levels, but low to acceptable specificity levels (same criterion).
Maisto et al., 2000 [32] Outpatients with serious persistent mental illness (n=162) DAST-10: A cutoff score of 2 resulted in sensitivity of 85% and specificity of 78% (criterion is current drug use disorder diagnosis using DSM-IV criteria).
Martino et al., 2000 [36] Adolescent psychiatric inpatients (n=194) DAST-A: Internal consistency: α = 0.91. 1 week test-retest reliability: r = 0.89. Adolescents with drug dependency diagnoses scored significantly higher on the DAST-A than adolescents who abused drugs, abused or were dependent on alcohol, or had no substance use diagnoses. Principle components analysis yielded 1 “meaningful” factor which accounted for 32% of the variance. A cutoff score of > 6 was associated with sensitivity of 70–79% (criterion is DSM-III-R and DSM-IV drug-related disorders, respectively) and specificity of 82–84% (DSM-III-R and DSM-IV, respectively).
McCann et al., 2000 [37] Pts at an ADHD clinic (n=143) Internal consistency: α = 0.92. Individuals with current or past drug use disorders scored significantly higher on the DAST than individuals without a history of problematic drug use. A cutoff score of 6 was associated with 85% sensitivity and 71% specificity (criterion is drug abuse or dependence using DSM-IV criteria).
Carey et al., 2003 [11] Inpatients in an Indian psychiatric hospital (n=1349) DAST-10: Internal consistency: α = 0.94. Factor analysis supported 1 factor which accounted for 94% of the variance. Patients receiving treatment for addiction had significantly higher DAST scores than other patients.
Cassidy et al., 2008 [12] Individuals experiencing their first psychotic episode (n=112) DAST-20: Internal consistency: α = 0.99. Median DAST scores were significantly higher among patients diagnosed with drug abuse or dependence and misuse within the past year than among patients without a drug use disorder diagnosis. The conventional cutoff score of 6 was associated with 55% sensitivity and 86% specificity (criterion is drug misuse diagnosis using DSM-IV criteria), although a cutoff score of 3 was associated with 85% sensitivity and 73% specificity. ROC analysis yielded AUC of 0.83 (same criterion).
Møller & Linaker, 2010 [43] Individuals with mental illness involving psychosis in Norway (n=48) DAST-20: A cutoff score of 5 was associated with sensitivity of 86% and specificity of 67% using ICD-10 substance abuse diagnosis as the criterion (phi-coefficient = 0.41). A cutoff score of 5 was associated with 74% sensitivity and 66% specificity using the staff-reported Clinical Drug Use Scale (DUS) as the criterion (phi-coefficient = 0.34).
Summary – DAST

  1. No information on input from patients, experts, or literature review. Not all items and time frames applicable to evaluating current inappropriate medication use events.

  2. 2 week test-retest data [20].

  3. Relationship between DAST and history drug and alcohol problems and diagnoses [20,37,52,55] and frequency of drug use [52].

  4. No data.

  5. No data.

  6. No data.


Summary – DAST-10

  1. No information.

  2. 3–10 day test-retest data [15].

  3. Relationship between 10-item DAST and drug use questionnaires, drug use disorder diagnoses, and addiction treatment [11,15,32].

  4. No data.

  5. No data.

  6. No data.


Summary – DAST-20

  1. No information.

  2. 3–10 day test-retest data [15].

  3. Relationship between 20-item DAST and problematic drug use, drug use disorder diagnoses, and clinician rated problematic drug use [12,15,25,30,43,53]. Distinction between 20-item DAST items and alcohol abuse items [50].

  4. No data.

  5. No data.

  6. No data.


Summary – DAST-A

  1. No information.

  2. 1 week test-retest data [36].

  3. Relationship between DAST-A and drug-related diagnoses [36].

  4. No data.

  5. No data.

  6. No data.
Opioid-specific
COMM Butler et al., 2007 [7] Chronic noncancer pain pts from hospital and pain management centers currently taking opioids (n=227) Experts who work with chronic pain patients (i.e., primary care physicians, nurses, psychologists, pain specialists, and addiction specialists) brainstormed about signals that a patient on opioids is exhibiting aberrant opioid use behaviors. This list was reduced by asking an independent group of experts to sort and rate the importance of each signal, the data from which were entered into a multidimensional scaling software program. The importance and wording of individual items was then evaluated by a 3rd group of experts. The final 17 items were identified by examining the psychometric properties among the 227 current pain patients taking opioids. Internal consistency: α = 0.86. 1 week test-retest reliability: ICC = 0.86. Two ROC analyses yielded AUC from 0.81–0.92 (criterion is ADBI). Cutoff score of 9 yields sensitivity of 94% and specificity of 73%; cutoff score of 10 yields sensitivity of 84% and specificity of 82% (same criterion). Reassessed 86 participants 3 mo after initial assessment to look at responsiveness; COMM detected 29 of 31 who were misusing (same criterion).
Butler et al., 2010 [6] Pts recruited from pain management centers taking opioids (n=226) Internal consistency: α = 0.83. ROC analysis yielded AUC of 0.79 (criterion is ADBI). Cutoff score of ≥ 9 yields sensitivity of 71% and specificity of 71% (same criterion).
Meltzer et al., 2011 [41] Primary care pts with chronic pain (n=238) Significantly higher COMM score in pts with prescription drug use disorder than those without. ROC analysis yielded AUC of 0.84 (criterion is DSM-IV prescription drug use disorder). Cutoff score of ≥ 13 yields sensitivity of 77% and specificity of 77% (same criterion).
Finkelman et al., 2013 [23] Same data as Butler 2007, Butler 2010 Various stopping rules result in sensitivity ranging from 69–70% and specificity ranging from 70–72% (criterion is ADBI); sensitivity ranging from 96–100% and specificity ranging from 99–100% (criterion is full COMM).
Summary

  1. Instrument created with input from experts regarding items and wording [7]; no input from pain patients taking opioids. Not all items applicable to evaluating current inappropriate medication use events.

  2. 1 week test-retest [7].

  3. Relationship between COMM scores and ADBI [6,7,23] and prescription drug use disorder [41].

  4. No data.

  5. Responsiveness to change in misuse status after 3 mo [7].

  6. No data.

PODS-CS Banta-Green et al., 2010 [4] Pts prescribed opioids for chronic pain (n=1144) Items identified from prior interviews with pain patients regarding problems and concerns about their chronic opioid therapy, input from 2 expert clinicians, and literature review. Internal consistency (original 0–5 scale): αs range from 0.75–0.79. Internal consistency (recoded scoring): αs range from 0.63–0.65.
Sullivan et al., 2010 [56] Same data as Banta-Green 2010 PODS-CS not significantly related to past 3 mo average pain intensity, although there was a small association with pain interference (r = 0.09) and a moderate relationship with depression. Odds ratio of prior drug use disorder diagnosis significantly higher in pts with mid-range or high scores on PODS-CS than those with low scores.
Summary

  1. Instrument created with input from patients and experts regarding items, as well as literature review [4]. Not all items and time frames applicable to evaluating current inappropriate medication use events.

  2. No data.

  3. Relationship between PODS and pain intensity, pain interference, depression, drug abuse/drug dependence diagnoses [56].

  4. No data.

  5. No data.

  6. No data.

[no name] 20 aberrant drug-related behaviors Hansen et al., 2011 [26] HIV pain pts on chronic opioid therapy (n=296) Items identified from a literature review and input from clinicians.
Summary

  1. Instrument created using a literature review and expert input [26]; no input from HIV pain patients taking opioids. Items appear to be content valid for evaluating inappropriate medication use, although a 90-day time frame may not be “current,” nor is information pertinent to classifying inappropriate use events captured.

  2. No data.

  3. No data.

  4. No data.

  5. No data.

  6. No data.

[no name] 8-item opioid analgesic misuse instrument Jeevanjee et al., 2013 [28] Homeless HIV pain pts prescribed antiretroviral drugs (n=258) Opioid misuse in past 90 days was associated with significantly lower adherence to antiretroviral medication.
Summary

  1. No information on input from patients, experts, or literature review. Items appear to be content valid for evaluating inappropriate medication use, although a 90-day time frame may not be “current,” nor is information pertinent to classifying inappropriate use events captured.

  2. No data.

  3. Relationship between misuse and antiretroviral adherence [28].

  4. No data.

  5. No data.

  6. No data.

Clinician-Reported Instruments
Non-specific
ASI, 5th edition McLellan et al., 1992 [38] Pts in detoxification and drug rehabilitation programs (n=42) Pilot tested the new additions in the ASI 5th edition to ensure items and instructions were understood.
Butler et al., 2001 [9] Pts in substance abuse treatment (n=202) Multimedia version (ASI-MV): Drug use composite score has good 3–5 day test-retest reliability and good criterion validity against 5th edition ASI. Convergent and discriminant validity for drug use composite score generally superior to 5th edition ASI drug use composite score.
Mäkelä, 2004 (review of 37 studies) [33] Analysis of ASI performance in 37 studies In a review of 37 studies, the drug use composite score demonstrates low inter-rater and test-retest reliabilities, low criterion validity, and variable internal consistency, sensitivity, and specificity.
Cacciola 2007 Pts in substance abuse treatment from an outpatient clinic and a methadone maintenance clinic (n=145 + 50) Shortened “lite” version (ASI-L-VA): Internal consistency was low for the drug use composite score in both the ASI-L-VA and the ASI 5th edition.
Summary - ASI 5th edition

  1. Patient comprehension of 5th edition revisions tested [38]. No information on input from experts or literature review. Items and time frames not applicable to evaluation of current inappropriate medication use events.

  2. Inter-rater and test-retest reliabilities for ASI [33].

  3. Relationship between ASI and other measures used for sensitivity and specificity tests [33].

  4. No data.

  5. No data.

  6. No data.


Summary – ASI-MV

  1. No information.

  2. 3–5 day test-retest reliability for ASI-MV [9].

  3. Relationships between ASI-MV drug use score and conceptually related and less related constructs [9].

  4. No data.

  5. No data.

  6. No data.

Prescription drug-specific
AIA Adams et al., 2006 [1] Chronic pain pts beginning treatment with tramadol, hydrocodone, or NSAIDs (n=11,352) Items identified from addiction behaviors in American Academy of Pain Management (AAPM), American Pain Society (APS), and American Society of Addiction Medicine (ASAM) consensus statement, DSM-IV-TR abuse and dependence classifications, with literature review, input from steering committee, and pilot testing with 30 patients. Hydrocodone pts had significantly more positive scores on the AIA than pts taking tramadol or NSAIDs
Summary

  1. Several items based on indicators of addiction from AAPM, APS, and ASAM statement, abuse and dependence classifications in DSM-IV-TR, along with a literature review and steering committee input. Tested in patients to ensure comprehension and brevity [1]. Not all items and time frames applicable to evaluating current inappropriate medication use events.

  2. No data.

  3. Relationship between AIA and analgesic treatment [1].

  4. No data.

  5. No data.

  6. No data.

Opioid-specific
POAC Chabal et al., 1997 [13] Pain pts enrolled in a pain clinic on chronic opioid therapy (n=76) Experts working with chronic pain patients developed the instrument. Inter-rater reliability = 0.90.
Summary

  1. Instrument created by chronic pain experts (i.e., physicians, fellows, and psychologists; [13]). No information on input from literature review or patients. Overall, items appear content valid for evaluating inappropriate medication use, except for a portion of item 3 that may not indicate inappropriate use. Further, time frame to be assessed not identified, nor is information pertinent to classifying inappropriate use events captured.

  2. Inter-rater reliability [13].

  3. No data.

  4. No data.

  5. No data.

  6. No data.

PDUQ Compton et al., 1998 [17] Pain pts receiving chronic opioid therapy referred from a multidisciplinary pain center for “problematic” medication use (n=52) Items based on literature review, review of medical records of chronic pain patients with addiction. Designed to meet ASAM and DSM-IV criteria for addiction. Internal consistency: α = 0.81. Nonaddicted participants had significantly lower PDUQ scores than substance abusing and substance dependent participants.
Butler et al., 2007 [7] Chronic noncancer pain pts currently taking opioids from hospital and pain management centers (n=227) Internal consistency: α = 0.79.
Wasan et al., 2009 [62] Pain pts currently on chronic opioid therapy (from pain management centers) (n=455) Pts who reported craving had significantly higher PDUQ scores than those who reported no craving.
Compton et al., 2008 (self-report version, PDUQp) [18] Pts with chronic pain from a multidisciplinary VA chronic pain clinic (n=135) Concurrent correlation btw PDUQp and PDUQ = 0.64. Baseline PDUQp demonstrated moderate predictive validity with PDUQ at 4, 8, and 12 mo. PDUQp has moderate test-retest reliability at 4, 8, and 12 mo. Cutoff score of ≥ 10 has sensitivity of 66.7% and specificity of 59.7% in predicting discontinuation of study due to medication agreement violation.
Banta-Green et al., 2009 (15-item PDUQ) [3] Pain pts on chronic opioid therapy from an HMO database - general medical setting (n=704) Original PDUQ: Internal consistency is poor (Cronbach’s α = 0.56).
15-item PDUQ: Factor analysis shows the 15 items loaded onto 3 factors which were distinct from items measuring DSM-IV dependence and abuse.
Summary - PDUQ

  1. Items based on addiction criteria from ASAM and DSM-IV, along with a literature review and review of patient medical records [17]. No information on input from patients or experts. Not all items and time frames applicable to evaluating current inappropriate medication use events.

  2. No data.

  3. Relationship between PDUQ and addiction status [17] and craving [62].

  4. No data.

  5. No data.

  6. No data.


Summary - PDUQp

  1. No information.

  2. Test-retest reliability [18].

  3. No data.

  4. Relationship between PDUQp and PDUQ and medication agreement violation [18].

  5. No data.

  6. No data.


Summary – 15-item PDUQ

  1. No information.

  2. No data.

  3. Distinction between 15-item PDUQ items and items diagnostic of DSM-IV dependence and abuse [3].

  4. No data.

  5. No data.

  6. No data.

Screening evaluation tool for controlled substance abuse Manchikanti et al., 2003 [35] Pts prescribed controlled substances in a pain management center (n=500) Items identified using a literature review. Total score ≥ 2 on screening tool correctly identified 90% of pts with history of drug abuse; only 3% of pts without drug abuse history scored ≥ 2.
Manchikanti et al., 2004 [34] Pts prescribed controlled substances in a pain management center (n=150) Total score ≥ 2 on screening tool correctly identified 79% of pts with controlled substance abuse (regardless of illicit drug use); only 2% of pts without substance abuse scored ≥ 2.
Summary

  1. Items based on a literature review, but no information about expert or patient input [35]. Overall, items appear content valid for evaluating current inappropriate medication use events, except time frame to be assessed is not reported, nor is information pertinent to classifying inappropriate use events captured.

  2. No data.

  3. Relationship between screening tool and history of drug and controlled substance abuse [34,35].

  4. No data.

  5. No data.

  6. No data.

POTQ Michna et al., 2004 [42] Chronic pain pts taking opioids (n=145) Significantly higher POTQ scores in participants at high risk of drug and alcohol abuse than low risk participants
Wasan et al., 2009 [62] Pain pts currently on chronic opioid therapy (from pain management centers) (n=455) Pts who reported craving had significantly higher POTQ scores than those who reported no craving.
Summary

  1. No information on input from patients, experts, or literature review. Overall, items appear content valid for evaluating inappropriate medication use, except that the time frame to be assessed is not identified, nor is information pertinent to classifying inappropriate use events captured.

  2. No data.

  3. Relationship between POTQ scores and risk of drug and alcohol abuse [42] and craving [62].

  4. No data.

  5. No data.

  6. No data.

PADT Passik et al., 2004 [48] Pain pts on chronic opioid therapy (n=388) Items identified using literature review, input from pain and addiction experts, and feedback from implementing clinicians.
Summary

  1. Items based on a literature review, as well as input from experts and clinicians who pilot tested the PADT [48], but no information about patient input. Not all items and time frames applicable to evaluating current inappropriate medication use events.

  2. No data.

  3. No data.

  4. No data.

  5. No data.

  6. No data.

ABC Wu et al., 2006 [63] Veterans in a Veterans Affairs Chronic Pain Clinic on chronic opioid therapy (n=136) Items based on literature review; designed to meet AAPM, APS, ASAM criteria for addiction. Inter-rater reliability: rs from 0.94–0.95. Pts whose clinician categorized them as inappropriately using medications scored significantly higher on the ABC than pts whose clinician categorized them as appropriately using medications. Correlation with the PDUQ: r = 0.40. Cut-off score of ≥ 3: Sensitivity = 87.5% specificity = 86.1% in predicting clinician categorization as inappropriate or appropriate medication user. Participants who were discontinued from study due to inappropriate medication use had significantly higher mean ABC score at last study visit than participants who remained in the study.
Summary

  1. Items based on addiction definition from AAPM, APS, and ASAM, along with a literature review [63]. No information on input from patients or experts. Not all items applicable to evaluating current inappropriate medication use events.

  2. Inter-rater reliability [63].

  3. Relationship between ABC and PDUQ and clinician-identified appropriateness of medication use [63].

  4. No data.

  5. Responsiveness to change in inappropriate use at visit before discontinued from study.

  6. No data.

POMI Knisely et al., 2008 [31] Pain pts prescribed OxyContin & pts treated for OxyContin addiction (n=74) ROC analysis comparing POMI score and DSM-IV opiate diagnosis yielded an AUC of 0.89. POMI cutoff score of ≥ 2 demonstrates sensitivity = 82% and specificity = 92% (same criterion).
Summary

  1. No information on input from patients, experts, or literature review. Overall, items appear content valid for evaluating inappropriate medication use, except that the time frame to be assessed is not identified, nor is information pertinent to classifying inappropriate use events captured.

  2. No data.

  3. Relationship between POMI and DSM-IV opiate diagnosis [31].

  4. No data.

  5. No data.

  6. No data.

Composite Instruments
ADBI Wasan et al., 2009 [62] Pain pts currently on chronic opioid therapy (from pain management centers) (n=455) Pts who scored positively on the ADBI were significantly more likely to report craving than to report no craving.
Summary

  1. No information on input from patients, experts, or literature review. Not all items and time frames applicable to evaluating current inappropriate medication use events.

  2. No data.

  3. Relationship between ADBI and craving [62].

  4. No data.

  5. No data.

  6. No data.

DMI Jamison et al., 2010 [27] Chronic back or neck pain pts from a pain management clinic currently on opioid therapy (n=228) At the end of the 6 month study, 73.7% of pts at high risk for inappropriate opioid use who did not undergo medication counseling scored highly on the DMI, compared to 26.3% of high-risk pts who underwent counseling and 25.0% of low-risk control participants.
Summary

  1. No information on input from patients, experts, or literature review. Not all items and time frames applicable to evaluating current inappropriate medication use events.

  2. No data.

  3. Relationship between DMI and risk for inappropriate opioid use [27].

  4. No data.

  5. No data.

  6. No data.

Abbreviations: AUC = area under curve; Pts = patients; ROC = receiver operating curve.

Measurement properties: (1) content validity; (2) cross-sectional reliability (test-retest, inter-rater); (3) cross-sectional construct validity; (4) longitudinal construct validity; (5) longitudinal ability to detect change; (6) longitudinal determination of clinically meaningful change or responder definitions.

Articles that did not provide psychometric data are not included in this table.