The American Psychiatric Association (APA) has updated its Privacy Policy and Terms of Use, including with new information specifically addressed to individuals in the European Economic Area. As described in the Privacy Policy and Terms of Use, this website utilizes cookies, including for the purpose of offering an optimal online experience and services tailored to your preferences.

Please read the entire Privacy Policy and Terms of Use. By closing this message, browsing this website, continuing the navigation, or otherwise continuing to use the APA's websites, you confirm that you understand and accept the terms of the Privacy Policy and Terms of Use, including the utilization of cookies.

×
Published Online:https://doi.org/10.1176/appi.ajp.21070721

Abstract

Objective:

The diagnostic criteria for opioid use disorder, originally developed for heroin, did not anticipate the surge in prescription opioid use and the resulting complexities in diagnosing prescription opioid use disorder (POUD), including differentiation of pain relief (therapeutic intent) from more common drug use motives, such as to get high or to cope with negative affect. The authors examined the validity of the Psychiatric Research Interview for Substance and Mental Disorders, DSM-5 opioid version, an instrument designed to make this differentiation.

Methods:

Patients (N=606) from pain clinics and inpatient substance treatment who ever received a ≥30-day opioid prescription for chronic pain were evaluated for DSM-5 POUD (i.e., withdrawal and tolerance were not considered positive if patients used opioids only as prescribed, per DSM-5 guidelines) and pain-adjusted POUD (behavioral/subjective criteria were not considered positive if pain relief [therapeutic intent] was the sole motive). Bivariate correlated-outcome regression models indicated associations of 10 validators with DSM-5 and pain-adjusted POUD measures, using mean ratios for dimensional measures and odds ratios for binary measures.

Results:

The prevalences of DSM-5 and pain-adjusted POUD, respectively, were 44.4% and 30.4% at the ≥2-criteria threshold and 29.5% and 25.3% at the ≥4-criteria threshold. Pain adjustment had little effect on prevalence among substance treatment patients but resulted in substantially lower prevalence among pain treatment patients. All validators had significantly stronger associations with pain-adjusted than with DSM-5 dimensional POUD measures (ratios of mean ratios, 1.22–2.31). For most validators, pain-adjusted binary POUD had larger odds ratios than DSM-5 measures.

Conclusions:

Adapting POUD measures for pain relief (therapeutic intent) improved validity. Studies should investigate the clinical utility of differentiating between therapeutic and nontherapeutic intent in evaluating POUD diagnostic criteria.

Among U.S. adults, pain is a highly prevalent condition (1) that is often treated with prescription opioids. Between the 1990s and 2011, the number of U.S. opioid prescriptions nearly tripled (2). While the number has since declined, it remains very high (>142 million in 2020 [3]), making adverse consequences of prescribed opioids a serious public health concern. Fatal prescription opioid overdoses, the most serious adverse consequence, soared, and while they have since declined alongside declining prescription rates, more than 14,000 individuals died from prescription opioid overdoses in 2019 (4). An additional serious consequence of prescription opioid use is addiction to prescription opioids, or prescription opioid use disorder (POUD) (5, 6). However, among patients receiving prescribed opioids for chronic pain, determining the prevalence of prescription opioid addiction/POUD and trends in this prevalence is challenging due to conceptualization and measurement issues.

The terms “opioid use disorder,” “addiction,” and “dependence” are often used inconsistently and interchangeably (5, 7, 8), leading to conceptual confusion and measurement inconsistencies (9). DSM-5 did not include the term “addiction” (page 485) because of its uncertain meaning and potentially pejorative connotations (1012), but many in the field continue to use the term. In addition, provider confusion between the term “addiction” and physiological dependence (1114) may lead some to diagnose POUD erroneously. These inconsistencies are likely to have contributed to widely varying POUD prevalences (9, 15) among patients who have prescriptions for opioids for pain (1624), leaving the actual prevalence of POUD among such patients unclear.

Diagnosing opioid use disorder in patients receiving prescribed opioids for chronic pain is inherently complex and difficult. DSM-5 defines substance use disorder, including POUD, with 11 criteria (10), including tolerance and withdrawal (physiological dependence [11]) and nine behavioral or subjective criteria (e.g., craving, considerable time using). DSM-5 also provides definitions of three severity levels: mild, moderate, and severe (2–3, 4–5, and 6–11 criteria, respectively) (10, 25). The DSM-5 criteria for opioid use disorder (OUD) overlap almost entirely with the DSM-IV criteria (25) and have near-perfect agreement in case identification (26). These definitions of OUD have major importance to the field because they determine eligibility for participation in most OUD treatment trials (e.g., 27, 28), including POUD (29).

OUD criteria were initially developed in studies in heroin users at addiction clinics in the 1980s (3032), when virtually all opioid use was illicit. Recognizing the increases in use of licit opioids prescribed for pain, DSM-5 recommended not diagnosing POUD in individuals using opioids as prescribed when the only criteria met were tolerance and/or withdrawal (10, 25), the expected physiological adaptations to treatment with opioids (5, 6, 9, 11). Some have disagreed with this approach (3335), while others have proposed an entirely different syndrome for this situation, namely, “complex persistent opioid dependence” (3436), which is based on clinical observation and in-depth expert neurobiological reasoning. Complex persistent opioid dependence is not yet operationalized in a research measure, although the DSM-5 approach to tolerance and withdrawal can be readily operationalized to study its reliability and validity. Another important concern about DSM-5 as it pertains to prescribed opioids is that in the behavioral and subjective POUD criteria, the role of pain relief is not addressed.

Empirically, substance use disorders cluster within the broader domain of externalizing disorders and traits (3741) (e.g., antisocial personality disorder, sensation-seeking, impulsivity [3840]) that share early onset and etiology (4245). From this perspective, behavioral OUD criteria can be seen as phenotypic manifestations of externalizing psychopathology. However, patients receiving legal, prescribed opioids for the therapeutic (indicated) purpose of pain relief are not typically characterized by externalizing traits (23, 46, 47). Thus, while behavioral and subjective POUD criteria may appear phenomenologically similar in patients treated for pain and substance users treated for addiction, the criteria may represent different underlying phenomena if, among patients treated for pain, the observed behavioral symptoms have therapeutic intent (i.e., pain relief) rather than motives such as externalizing reasons, such as to get high (10, 48), or motives such as coping with internalizing conditions (e.g., depression, anxiety) that are common among those with pain and are associated with addiction (4952). DSM-5 clarifies that among individuals with a prescription for analgesic opioids for pain relief who are using more opioids than prescribed, OUD can be diagnosed if the signs and symptoms are present, but “not only because of persistent pain” (p. 542). Therefore, ignoring pain relief motives when assessing POUD criteria could lead to misdiagnosed POUD/addiction. Until now, patient-oriented assessment of pain relief (therapeutic intent) has not been incorporated into the measurement of DSM-5 POUD criteria.

Given the many issues related to prescription opioids, the U.S. Food and Drug Administration (FDA) held a series of meetings for pain, opioid, and addiction experts, providers, and stakeholders and issued a set of postmarketing requirements (PMRs) to study prescription opioids, including PMR #3033-5 to develop and validate a standardized measure of prescription opioid substance use disorder/addiction that incorporated patient-oriented assessment of pain relief/therapeutic intent (15, 53). For this, the clinician-administered, computer-assisted Psychiatric Research Interview for Substance and Mental Disorders, DSM-5 opioid version (PRISM-5-OP) was designed (15). The PRISM-5-OP was adapted from the PRISM (DSM-IV) (5456) and the PRISM-5 (DSM-5) (57, 58), which were used previously in reliability and validity studies (5759) and in a study of POUD in primary care (60). The PRISM-5-OP assesses POUD in three ways (15): 1) completely unadjusted: all DSM-5 criteria are considered positive if they occurred, including tolerance and withdrawal, regardless of whether opioids were used only as prescribed or not; 2) DSM-5: withdrawal and tolerance are not considered positive if patients used prescribed opioids only as prescribed; and 3) pain-adjusted: POUD behavioral/subjective criteria are not considered positive if pain relief/therapeutic intent was the only motive for the behavior described by the criterion. The validation strategy consisted of comparing the strength of the associations of these three PRISM-5-OP POUD measures with a set of 10 convergent and discriminant validators. The primary comparisons of interest were between the pain-adjusted and DSM-5 POUD measures, indicating whether the pain adjustments improved validity over the DSM-5 measure. Comparing DSM-5 and completely unadjusted POUD explored the validity of the DSM-5 adjustment for tolerance and withdrawal compared to no adjustment.

Methods

Settings and Procedures

Urban and suburban patients were recruited from two types of settings: inpatient substance treatment (two sites) and outpatient pain clinics (four sites) that utilized procedures to avoid prescribing opioids to patients for nontherapeutic reasons. These two types of settings were selected to determine whether distributions of POUD measures differed between patients with established substance use disorders and medical pain patients (an informal validity check), and to provide sufficient variance to conduct the analyses. After learning about the study from flyers or clinic staff, patients contacted study coordinators, who explained the study, screened for eligibility (≥18 years old, ever received a prescription for a ≥30-day supply of opioids to treat chronic pain, English-speaking), and obtained informed consent (see recruitment and enrollment details in Figure S1 in the online supplement). The final sample included 606 patients, of whom 348 were from pain clinics and 258 from substance treatment. Institutional review boards at New York State Psychiatric Institute, Cornell Hospital (Cornell University), Holy Name Medical Center (Rutgers University), Bellevue Hospital (New York University), and South Oaks Hospital (Hofstra University) approved the study.

Interviews, self-administered questionnaires, and electronic medical records (EMRs) provided the study data. Interviews were conducted between February 29, 2016, and April 24, 2017, by telephone (49.8%) or in person. For efficiency, all substance treatment patients were assessed in person; most pain clinic patients were assessed by telephone, based on patient or clinic preferences. In-person interviewees completed self-administered questionnaires on a tablet; telephone interviewees used a web link. Trained coordinators extracted EMR variables, entering them into a form on a tablet. Of the 606 patients, 92.4% agreed to retest reliability interviews, and by design, we reinterviewed approximately one-third of the sample (N=206; substance treatment patients, N=97; pain clinic patients, N=109). Test and retest interviews were conducted by different interviewers, with retest interviewers blinded to initial interview results. Interviews were randomly assigned to available interviewers; 60.6% of retests were completed by telephone. The mean test-retest interval was 7.2 days (SD=6.34, range=0–19). Because the purpose of the retest was to determine replicability of the measure when administered by two independent raters, a short interval was selected to reduce the impact of true change in clinical condition on agreement between the first and second interviewer.

PRISM-5-OP interviewers (N=26) had master’s degrees in social work or clinical psychology and had clinical experience (mean=3.9 years, SD=4.6, range=1–24). Interviewers received 2 days of PRISM-5-OP training (didactic, role-playing) from experienced supervisors, practiced the interview, and submitted recordings, which were reviewed by supervisors. Interviewers were certified after supervisors rated two interviews as acceptable. To maintain consistency during the study, supervisors rated recordings of randomly selected interviews (10%), using structured forms, and supervised interviewers as necessary.

Measures

PRISM-5-OP.

The PRISM-5-OP assesses DSM-5 criteria for substance use disorders and psychiatric disorders, with an expanded POUD module near the beginning of the interview. This module includes additional opioid-related items (e.g., used only as prescribed, more/other than as prescribed), and for POUD behavioral/subjective criteria, reasons the criterion occurred, classified as pain relief (therapeutic) and nontherapeutic (e.g., to feel high, for affect, for sleep, in reaction to conditioned cues).

Three POUD criteria sets: completely unadjusted, DSM-5, pain-adjusted.

Each DSM-5 POUD criterion is evaluated for occurrence within the past 12 months. In the completely unadjusted criteria set, all criteria that occurred are considered positive and are counted toward a POUD diagnosis. In the DSM-5 set, tolerance and withdrawal are coded positive only if they occurred among those using prescription opioids more than or other than as prescribed. In the pain-adjusted criteria set, tolerance and withdrawal are rated as in the DSM-5 set, while the behavioral/subjective criteria are considered positive only if they occurred for a nontherapeutic reason (i.e., other than pain relief).

Dimensional POUD measures.

By summing the criteria, dimensional measures were created for each POUD criteria set. The completely unadjusted measure was a count of all positive unadjusted criteria. The DSM-5 measure was a count of all criteria that occurred except tolerance and withdrawal, which were considered positive if they occurred only in patients using opioids in nonprescribed ways. The pain-adjusted measure included the DSM-5 adjustment and, in addition, considered as positive only the criteria reflecting behaviors that occurred for nontherapeutic reasons (i.e., for reasons other than pain relief). The completely unadjusted, DSM-5, and pain-adjusted measures had excellent internal consistency (Cronbach’s alpha, 0.92, 0.93, and 0.95, respectively). In exploratory factor analysis, all three criteria sets fit a one-factor model—that is, one eigenvalue >1.0 and the rest well below 1.0, standard fit indices within recommended values (comparative fit index ≥0.95, root mean square error of approximation <0.06), and factor loadings of 0.79–0.96 for completely unadjusted, 0.82–0.99 for DSM-5, and 0.79–0.99 for pain-adjusted criteria (see Table S1 in the online supplement).

POUD diagnosis.

Binary diagnoses were created from the dimensional measures using the DSM-5 thresholds of ≥2 criteria to indicate any DSM-5 diagnosis, and ≥4 criteria to indicate moderate to severe disorders. (Exploring other thresholds did not indicate improved validity of any novel thresholds; see the online supplement for supporting text, Tables S2 and S3, and Figures S2 and S3.)

Validators.

Ten validators were used, of which eight were convergent validators: 1) substance treatment: recruitment from substance settings, endorsing treatment in the PRISM-5-OP interview; any positive=yes; 2) family history of drug use disorders: PRISM-5-OP using validated items from general population surveys (61, 62); any parent or sibling with drug problems=yes; 3) DSM-5 other substance use disorder (alcohol, nicotine, cannabis, illicit drugs): PRISM-5-OP or EMR; any positive=yes; 4) antisocial personality disorder: PRISM-5-OP; 5) internalizing DSM-5 disorders: major or persistent depression, generalized anxiety, posttraumatic stress, PRISM-5-OP or EMR; any positive=yes; 6) tampering with prescribed opioid medication: crushing, dissolving, etc. (self-administered questionnaire); any positive=yes; 7) and 8) impulsivity, sensation-seeking (traits associated with illicit drug use), brief validated scales (63, 64) added to the self-administered questionnaire midway through the study; completed by 314 patients. The other two validators were discriminant validators: 9) worst pain, past week: Brief Pain Inventory (6567), a validated, widely used pain scale (self-administered questionnaire); and 10) prescription for legitimate reason: whether patients always obtained prescription opioids from a physician “for actual pain or a legitimate medical reason” (self-administered questionnaire).

Sociodemographic characteristics.

Sociodemographic characteristics included age categories (18–29, 30–39, 40–49, and ≥50 years), sex, race/ethnicity (non-Hispanic White, Black, and Other; Hispanic ethnicity), marital status (married, living together, neither), educational attainment (any college, yes/no), employed (yes/no), and health insurance (public, private, none).

Statistical Analysis

Reliability.

Cohen’s kappa and intraclass correlation coefficients (ICCs) with 95% confidence intervals indicated chance-corrected agreement between test and retest interviews (68) for binary and dimensional variables, respectively. Values of 0.21–0.40 indicate fair agreement, 0.41–0.60 moderate agreement, 0.61–0.80 substantial agreement, and 0.81–1.00 excellent agreement (69, 70).

Validity.

Validity was evaluated by testing differences between the pain-adjusted and DSM-5 POUD measures in the strength of their associations with the validators. Bivariate correlated-outcome regressions (71, 72) simultaneously modeled the two POUD measures as outcomes (either two dimensional or two dichotomous), taking into account their within-participant correlation using generalized estimating equations (implemented with Proc GENMOD in SAS). Separate models were fitted for each validator (predictor) to determine its association with both of the POUD measures, adjusting for covariates and testing whether one POUD measure was more strongly associated with the validator. The POUD dimensional measures were modeled using a negative binomial distribution with log link based on analyses that identified the best-fitting distribution; the dichotomous POUD measures were modeled using a logistic link. Results were expressed in mean ratios for dimensional POUD measures and odds ratios for dichotomous ones. Mean ratios and odds ratios ≥1.0 show greater severity or odds, respectively, of the POUD measure for participants with the validator present rather than absent, or at higher values of the validator, supporting validity for convergent validators; values <1.0 show greater severity or odds when the validator is absent, supporting validity for discriminant validators. Ratios of the mean ratios or odds ratios for pain-adjusted and DSM-5 POUD measures indicated whether one version showed stronger association with the validators than the other. Analyses controlled for age, sex, race, education, marital status, employment, and health insurance. Using the same models, to explore the validity of the DSM-5 adjustments for withdrawal and tolerance, we also compared the validity of the completely unadjusted and DSM-5 POUD measures.

Results

Sample Characteristics

The characteristics of the full sample and the test-retest reliability subset were similar (see Table S4 in the online supplement). About half the participants were ≥50 years of age, female, and/or unmarried; about 80% were White; about two-thirds had some college education; and about 22% were employed. Compared to patients in substance treatment, patients from pain clinics were more often older, female, college-educated, and married, and they had higher pain scores, lower sensation-seeking and impulsivity scores, and fewer substance use or psychiatric disorders.

The prevalences of completely unadjusted, DSM-5, and pain-adjusted POUD diagnoses with ≥2 and ≥4 thresholds ranged from 25.3% to 51.0% in the full sample, 53.9% to 62.0% among patients in substance treatment, and 4.0% to 42.8% among pain clinic patients (Table 1). Pain adjustments made few significant differences in POUD prevalence among patients in substance treatment, who rarely engaged in POUD criteria only for pain-relief (therapeutic) reasons (Table 1). However, pain adjustments substantially reduced prevalence among pain clinic patients. Similar patterns were observed for the individual criteria, with a significantly greater adjustment effect among pain clinic than substance treatment patients for all but one criterion (see Table S5 in the online supplement). Means of the dimensional POUD measures (Table 1) showed similar patterns.

TABLE 1. Current (past 12 months) prescription opioid use disorder (POUD) measures from PRISM-5-OP: prevalence (binary diagnostic measures) and mean scores (dimensional measures)a

MeasureFull Sample (N=606)Substance Treatment Patients (N=258)Pain Clinic Patients (N=348)
%N%N%N
Binary POUD diagnosis, ≥2 criteria
 Completely unadjusted criteriab51.030962.016042.8149
 DSM-5 criteriac44.426961.615931.6110
 Pain-adjusted criteriad30.418458.11509.834
Binary POUD diagnosis, ≥4 criteria
 Completely unadjusted criteriab31.919355.414314.450
 DSM-5 criteriac29.517954.714110.938
 Pain-adjusted criteriad25.315353.91394.014
MeanSDMeanSDMeanSD
Dimensional measure
 Completely unadjusted criteriab3.03.64.84.41.71.9
 DSM-5 criteriac2.83.64.84.41.31.7
 Pain-adjusted criteriad2.33.64.64.30.51.3

aPRISM-5-OP=Psychiatric Research Interview for Substance and Mental Disorders, DSM-5 opioid version.

bCompletely unadjusted criteria: a count of all DSM-5 opioid use disorder (OUD) criteria that occurred, regardless of whether prescription opioids were taken only as prescribed or more than or other than as prescribed.

cDSM-5 criteria: a count of all DSM-5 OUD criteria that occurred, except tolerance and withdrawal, which are counted as positive only among patients using opioids in nonprescribed ways.

dPain-adjusted criteria: a count of DSM-5 OUD criteria that includes the DSM-5 adjustment and, in addition, counted as positive only the criteria that occurred for nontherapeutic reasons (i.e., other than to treat pain, such as to get high).

TABLE 1. Current (past 12 months) prescription opioid use disorder (POUD) measures from PRISM-5-OP: prevalence (binary diagnostic measures) and mean scores (dimensional measures)a

Enlarge table

Test-Retest Reliability

The completely unadjusted, DSM-5, and pain-adjusted PRISM-5-OP POUD dimensional measures (Table 2) had excellent reliability (ICCs, 0.79–0.82), which was somewhat better among pain patients (ICCs, 0.87–0.93) than among patients in substance treatment (all ICCs=0.72). For diagnostic POUD measures, reliability was moderate for completely unadjusted and DSM-5 diagnoses (kappa values, 0.53–0.59) and moderate to substantial for pain-adjusted diagnoses (kappa values, 0.59–0.65), with reliability higher among pain patients, especially for pain-adjusted POUD (kappa values, 0.71–0.80). Two PRISM-5-OP items—taking more opioids than prescribed and using without a prescription—were used in the DSM-5 algorithm for withdrawal and tolerance, so their reliability was examined. Taking more than prescribed had substantial reliability overall (kappa=0.80, 95% CI=0.70, 0.90) and among patients in substance treatment (kappa=0.78, 95% CI=0.63, 0.94), and excellent reliability among pain patients (kappa=0.82, 95% CI=0.69, 0.95). Taking opioids without a prescription had substantial reliability overall (kappa=0.77, 95% CI=0.66, 0.88) and among patients in substance treatment (kappa=0.66, 95% CI=0.51, 0.81) and was excellent among pain patients (kappa=1.00).

TABLE 2. Test-retest reliability of current (past 12 months) prescription opioid use disorder (POUD) measures, overall and by type of setting

Full Sample (N=206)Pain Clinics (N=109)Substance Treatment (N=97)
TestRetestTestRetestTestRetest
MeanSDMeanSDICCa95% CIMeanSDMeanSDICCa95% CIMeanSDMeanSDICCa95% CI
Dimensional measures
 Completely unadjusted3.03.62.43.20.790.70, 0.861.82.21.41.70.870.82, 0.924.44.33.54.00.720.57, 0.83
 DSM-52.83.62.33.20.790.70, 0.871.42.01.21.60.870.80, 0.934.34.33.54.00.720.56, 0.83
 Pain-adjusted2.33.61.83.20.820.73, 0.890.71.70.51.30.930.81, 0.994.24.33.34.00.720.56, 0.83
TestRetestTestRetestTestRetest
%N%NKappa95% CI%N%NKappa95% CI%N%NKappa95% CI
Binary measures
 ≥2 criteriab
  Completely unadjusted50.010341.3850.550.44, 0.6743.14733.0360.600.44, 0.7557.75650.5490.480.31, 0.66
  DSM-542.78837.4770.590.47, 0.7030.33326.6290.640.48, 0.8056.75549.5480.490.31, 0.66
  Pain-adjusted32.06626.2540.650.53, 0.7611.9138.390.800.61, 0.9954.65346.4450.470.29, 0.64
 ≥4 criteriac
  Completely unadjusted32.06625.7530.590.47, 0.7117.41912.8140.680.49, 0.8748.54740.2390.460.29, 0.64
  DSM-530.16222.3460.530.40, 0.6613.8156.470.400.14, 0.6748.54740.2390.460.29, 0.64
  Pain-adjusted26.25419.4400.590.46, 0.726.473.740.710.41, 1.0048.54737.1360.440.26, 0.61

aIntraclass correlation coefficient (ICC) (calculated with the SAS intracc macro) with 95% confidence interval from 100 bootstrapped samples.

bDSM-5 threshold; includes all DSM-5 severity levels: mild (2–3 criteria), moderate (4–5 criteria), and severe (6–11 criteria).

cDSM-5 threshold that includes moderate and severe.

TABLE 2. Test-retest reliability of current (past 12 months) prescription opioid use disorder (POUD) measures, overall and by type of setting

Enlarge table

Validity: Pain-Adjusted Versus DSM-5 POUD Measures

Dimensional POUD measures.

All convergent validators (Table 3) were positively associated with the pain-adjusted (mean ratios, 1.60–5.30) and DSM-5 POUD (mean ratios, 1.26–2.30) measures. Associations were significantly stronger with the pain-adjusted measures than with the DSM-5 measures (ratios of mean ratios, 1.22–2.31). Pain-adjusted POUD was negatively associated with both discriminant validators (mean ratios, 0.96 and 0.46). DSM-5 POUD was unrelated to worst pain, but negatively associated with legitimate opioid use (mean ratio, 0.73). Pain-adjusted dimensional POUD had significantly stronger associations with both discriminant validators than DSM-5 dimensional POUD.

TABLE 3. Associations of pain-adjusted and DSM-5 PRISM-5-OP dimensional measures of prescription opioid use disorder (POUD) with convergent and discriminant validators (N=606)a

Association of Validators With DSM-5 POUD Dimensional MeasurebAssociation of Validators With Pain-Adjusted POUD Dimensional MeasurecDifference in Association of Pain-Adjustedc and DSM-5b Dimensional POUD Measures With Validators
MeasureMean Ratiod95% CIpMean Ratiod95% CIpRatio of Mean Ratiose95% CIp
Convergent validators
 Substance treatment2.301.80, 2.95<0.0015.303.77, 7.47<0.0012.311.91, 2.79<0.001
 Family history of any drug use disorderf1.261.02, 1.560.0351.601.22, 2.09<0.0011.271.13, 1.43<0.001
 Personal history of other substance use disorder1.701.34, 2.15<0.0013.842.58, 5.69<0.0012.261.73, 2.96<0.001
 Antisocial personality disorder1.391.09, 1.760.0071.911.46, 2.51<0.0011.381.25, 1.52<0.001
 Internalizing disordersg1.681.33, 2.12<0.0012.041.49, 2.81<0.0011.221.02, 1.450.0265
 Tamperingh1.731.34, 2.24<0.0012.802.07, 3.78<0.0011.611.45, 1.80<0.001
 Sensation-seekingi,j1.271.04, 1.550.0201.641.27, 2.11<0.0011.291.15, 1.45<0.001
 Impulsivityi,j1.401.10, 1.770.0061.701.29, 2.24<0.0011.221.06, 1.390.004
Discriminant validators
 Worst pain in past weekj0.9890.96, 1.010.4100.9570.93, 0.990.0090.9680.95, 0.98<0.001
 Prescription for legitimate reason0.730.57, 0.950.0180.460.34, 0.62<0.0010.620.56, 0.70<0.001

aPRISM-5-OP=Psychiatric Research Interview for Substance and Mental Disorders, DSM-5 opioid version.

bDSM-5: a count of all DSM-5 opioid use disorder (OUD) criteria that occurred, except tolerance and withdrawal, which were counted as positive only among patients using opioids in nonprescribed ways.

cPain-adjusted: a count of DSM-5 OUD criteria that includes the DSM-5 adjustment and, in addition, counted as positive only the criteria that occurred for nontherapeutic reasons (i.e., other than to treat pain, such as to get high).

dThe mean ratio is the ratio of the mean value for the dimensional measure among those with the validator and without the validator, that is, the exponentiated regression coefficient from the correlated-outcomes negative binomial regression model, controlling for covariates (age, sex, race, education, marital status, employment, and health insurance).

eThe difference in the validator effect for pain-adjusted versus DSM-5 is presented as the ratio of mean ratios. If this term is statistically significantly different from 1, differential effects are present, meaning that one criteria set shows a stronger association than the other.

fDrug use disorder indicates any substance use disorder except alcohol.

gIncludes major depressive disorder, persistent depression, generalized anxiety disorder, and posttraumatic stress disorder.

hEver tampered with prescribed opioid medication.

iN=314.

jContinuous measure; the ratio indicates change for a one-unit increase in the scale.

TABLE 3. Associations of pain-adjusted and DSM-5 PRISM-5-OP dimensional measures of prescription opioid use disorder (POUD) with convergent and discriminant validators (N=606)a

Enlarge table

POUD binary diagnostic measures.

At the ≥2 threshold, a pain-adjusted POUD diagnosis was associated with all 10 validators, and at the ≥4 threshold, it was associated with nine of the 10 validators (Table 4). The DSM-5 POUD measures at the ≥2 and ≥4 thresholds were associated with five and seven validators, respectively. At both thresholds, odds ratios were significantly stronger for pain-adjusted diagnoses than DSM-5 diagnoses with all validators but two (internalizing disorders, impulsivity).

TABLE 4. PRISM-5-OP binary diagnostic measures of current prescription opioid use disorder (POUD): associations of pain-adjusted and DSM-5 POUD with convergent and discriminant validators (N=606)a

Association of Validators With DSM-5-Adjusted POUD DiagnosisbAssociation of Validators With Pain-Adjusted POUD DiagnosiscDifference in Association Between Pain-Adjustedc and DSM-5b POUD Diagnoses With Validators
≥2 Criteria≥4 Criteria≥2 Criteria≥4 Criteria≥2 Criteria≥4 Criteria
MeasureOdds Ratiod95% CIpOdds Ratiod95% CIpOdds Ratiod95% CIpOdds Ratiod95% CIpRatio of Odds Ratiose95% CIpRatio of Odds Ratiose95% CIp
Convergent validators
 Substance treatment2.101.39, 3.19<0.0014.552.70, 7.68<0.0016.874.17, 11.34<0.00116.007.90, 32.39<0.0013.272.27, 4.69<0.0013.522.04, 6.07<0.001
 Family history of any drug use disorderf1.330.92, 1.940.1321.450.93, 2.250.0971.931.33, 2.81<0.0011.891.22, 2.910.0041.451.10, 1.900.0071.301.07, 1.590.010
 Personal history of other substance use disorder1.510.99, 2.300.0543.852.08, 7.12<0.0016.613.51, 12.47<0.00111.424.49, 29.08<0.0014.372.52, 7.59<0.0012.971.37, 6.410.006
 Antisocial personality disorder1.851.13, 3.030.0142.741.64, 4.58<0.0013.272.04, 5.24<0.0013.592.16, 5.95<0.0011.771.36, 2.30<0.0011.311.09, 1.570.004
 Internalizing disordersg1.801.21, 2.690.0042.191.36, 3.530.0012.491.61, 3.85<0.0011.931.20, 3.120.0071.380.98, 1.950.0660.880.71, 1.900.252
 Tamperingh1.821.17, 2.850.0083.031.86, 4.94<0.0014.342.74, 6.85<0.0014.922.98, 8.13<0.0012.381.87, 3.02<0.0011.621.33, 1.98<0.001
 Sensation-seekingi,j1.431.05, 1.960.0231.721.17, 2.530.0062.171.5, 3.15<0.0012.271.45, 3.55<0.0011.521.19, 1.93<0.0011.321.09, 1.600.004
 Impulsivityi,j1.570.97, 2.550.0671.250.73, 2.140.4071.811.1, 2.960.0191.390.84, 2.300.2061.150.82, 1.620.4221.100.85, 1.440.459
Discriminant validators
 Worst pain in past weekj1.010.97, 1.060.5500.970.92, 1.020.1800.930.89, 0.970.0020.940.89, 0.990.0140.920.89, 0.95<0.0010.970.95, 1.000.019
 Prescription for legitimate reason0.730.48, 1.110.1350.470.30, 0.750.0010.320.21, 0.50<0.0010.300.18, 0.48<0.0010.450.35, 0.57<0.0010.630.51, 0.77<0.001

aPRISM-5-OP=Psychiatric Research Interview for Substance and Mental Disorders, DSM-5 opioid version.

bDSM-5-adusted: a count of all DSM-5 opioid use disorder (OUD) criteria that occurred, except tolerance and withdrawal, which were counted as positive only among patients using opioids in nonprescribed ways.

cPain-adjusted: a count of DSM-5 OUD criteria that includes the DSM-5 adjustment and, in addition, counted as positive only the criteria that occurred for nontherapeutic reasons (i.e., other than to treat pain, such as to get high).

dThe odds ratio is the exponentiated regression coefficient from the correlated-outcomes logistic regression model, controlling for covariates (age, sex, race, education, marital status, employment, and health insurance).

eThe difference in the validator effect for fully adjusted versus DSM-5-adjusted is presented as the ratio of the odds ratios. If this term is statistically significantly different from 1, differential effects are present, meaning that one criterion set shows a stronger association than the other.

fDrug use disorder indicates any substance use disorder except alcohol.

gIncludes major depressive disorder, persistent depression, generalized anxiety disorder, and posttraumatic stress disorder.

hEver tampered with prescribed opioid medication.

iN=314.

jContinuous measure; the ratio indicates the change in likelihood for a one-unit increase in the scale.

TABLE 4. PRISM-5-OP binary diagnostic measures of current prescription opioid use disorder (POUD): associations of pain-adjusted and DSM-5 POUD with convergent and discriminant validators (N=606)a

Enlarge table

Validity: Completely Unadjusted Versus DSM-5 POUD Measures

For dimensional measures (see Table S6 in the online supplement), completely unadjusted POUD was associated with all validators except worst pain; DSM-5 dimensional POUD had significantly stronger associations than completely unadjusted POUD with all validators. For binary measures (see Table S7 in the online supplement), completely unadjusted POUD at the ≥2 and ≥4 thresholds was associated with six and seven validators, respectively. DSM-5 POUD at the ≥2 threshold had significantly stronger associations than completely unadjusted POUD with eight validators, and at the ≥4 threshold, with three validators.

Discussion

DSM-5 criteria for substance use disorders have been extensively validated for illicit drugs, including heroin (25), but not for opioids prescribed for pain. The FDA required a study to develop and validate a standardized measure of POUD that incorporated patient-oriented information on the patient’s pain relief/therapeutic intent in evaluating POUD criteria (15). In the instrument developed for this purpose (15), the PRISM-5-OP, pain-adjusted POUD had significantly stronger associations than DSM-5 POUD with 10 convergent and discriminant validators, suggesting greater validity of the pain-adjusted POUD measures that differentiated pain relief/therapeutic intent from other motives, including externalizing motives (e.g., to get high) and internalizing motives (e.g., to cope with negative or anxious affect).

Previous studies of POUD/addiction have produced widely varying prevalence estimates (9, 16, 17, 24), attributed to methodological inconsistencies (9). In our study, different POUD operationalizations substantially changed prevalences (range, 4%–42.8% among pain clinic patients), supporting previous concerns that inconsistent concepts or nonstandardized measures artifactually influenced rates (15). Our prevalence of pain-adjusted POUD was lower than the DSM-5 approach among pain patients; nevertheless, this prevalence was greater than the prevalence of addiction found in one-third of the studies in a widely cited review (17). Thus, while confusion between physiological dependence and addiction/POUD (11) may lead to expectations of much higher POUD prevalence than what we found, our prevalence of pain-adjusted POUD is within the range reported in previous research. Additionally, the pain-adjusted POUD measures had the strongest associations with the validators, supporting the validity of estimates in the 4%–10% range. Although rates of U.S. opioid prescriptions have declined, they remain high (>142 million in 2020 [3]). Thus, even low-end POUD prevalence estimates represent a substantial burden in the overall population (73). In addition, although high-dose opioid use, regardless of the motivation, is a major risk factor for opioid overdose, especially among pain patients who escalate their dosage, quantification of actual increases in risk due to POUD among pain patients who are opioid users remains surprisingly unclear; studies are needed to clarify this relationship.

For the DSM-5 behavioral/subjective POUD criteria, the PRISM-5-OP differentiation between pain relief and nonindicated motives could help guide clinician approaches when discussing treatment options. Patients using prescribed opioids for a high or to ameliorate depression or anxiety might benefit from a different presentation than patients whose behavioral/subjective POUD criteria are entirely for pain-relief reasons, even if the proffered treatment is the same in either case—for example, nonpharmacologic treatments, or buprenorphine, which may provide safety and analgesia benefits for those transitioning off opioids (74) and assist in treating internalizing psychiatric comorbidity (7577). In individuals with pain conditions, mood/anxiety and pain are highly related, so it can be unclear, even to the patient, what is driving medication-seeking, since chronic opioid use, regardless of the reasons, has an impact on mood, anxiety, and the functioning of the endogenous opioid system, which is involved in the regulation and functioning of multiple systems (78). Therefore, while acknowledging patients’ perspectives may help engage them in the discussion about treatment options, education about the fact that taking opioid medications to relieve emotional pain or dysphoria is a clear pathway to the development of OUD is also important. We note that others have also called for measures of prescription opioid problems that differentiate pain relief from other motives (20, 79). Research is needed to investigate the clinical utility of this differentiation.

Our findings suggest an appreciable burden of POUD in pain patients, even after incorporating the pain adjustments that lowered prevalence. A larger ongoing PMR study using the PRISM-5-OP will provide more precise estimates of prevalence and incidence among patients using opioids for chronic pain (15). In addition, our study could not address the reliability or validity of the alternative syndrome described as “complex persistent opioid dependence” (3436). Such studies should be conducted when a measure of complex persistent opioid dependence is developed.

Several limitations of this study should be noted. First, stigma or social desirability bias could affect responses about POUD criteria. However, among substance treatment patients, similar patterns of pain-adjusted and DSM-5 measures (see Table S5 in the online supplement) show that most of these patients reported using prescribed opioids for nontherapeutic reasons, while among pain clinic patients, responses indicated considerable therapeutic intent/pain relief. The consistent evidence from 10 external validators suggests that pain patient responses were truthful, since a complex set of fabrications would be needed to produce consistent results across all 10 validators. Second, patients were recruited from only two types of settings. Studies in primary care and mental health care settings (where opioid prescriptions are common [80, 81]) would be useful. Third, the PRISM-5-OP did not include pain adjustments for cannabis, which is increasingly legal for treating pain (82). Such adjustments could prove useful and should be studied in the future. Fourth, patients with dysphoria could attribute their dysphoria to pain, possibly leading to overadjustment. Investigation of this issue was outside the scope of the present study, but should be done in the future. Finally, the study design did not include prospective validation, but future studies should undertake this next logical step. Identifying POUD in chronic pain patients using opioids is an inherently difficult problem, so multiple studies are needed.

Chronic opioid use is not indicated for most pain conditions, so clinicians should always seek alternatives to opioids for chronic pain. However, given the continuing high numbers of opioid prescriptions, measures enabling study of risk factors for and consequences of POUD remain important. We found that the PRISM-5-OP measures of pain-adjusted POUD measures were reliable and more valid than standard DSM-5 concepts of substance use disorder. However, many external validators were associated with DSM-5 as well as pain-adjusted POUD measures. Studies should report both sets of results and further investigate differences between pain-adjusted and DSM-5 POUD and their consequences.

Department of Psychiatry, Columbia University Medical Center (Hasin, Shmulewitz, Aharonovich, Scodes, Wall) and New York State Psychiatric Institute (Hasin, Shmulewitz, Stohl, Greenstein, Aharonovich, Scodes, Wall), New York; Epidemiology, Pfizer, Inc., New York (Petronis); Kaiser Permanente Washington Health Research Institute, Seattle (Von Korff); Department of Anesthesiology, Rutgers University, Newark, N.J. (Datta); Department of Anesthesiology, Columbia University Medical Center, New York (Sonty, Weinberger); Department of Psychiatry, New York University, New York (Ross); Cornell University Medical College, New York (Inturrisi).
Send correspondence to Dr. Hasin ().

Presented in part at the College on Problems of Drug Dependence annual conference, San Antonio, Tex., June 16, 2019.

This study was funded by the Opioid Postmarketing Consortium (OPC) as part of a U.S. Food and Drug Administration (FDA) postmarketing requirement for opioid analgesics. The OPC member companies included the following: Allergan, Assertio Therapeutics, BioDelivery Sciences, Collegium Pharmaceutical, Daiichi Sankyo, Egalet Corporation, Endo Pharmaceuticals, Hikma Pharmaceuticals USA, Janssen Pharmaceuticals, Mallinckrodt, Pernix Therapeutics Holdings, Pfizer, and Purdue Pharma. The role of the OPC in the study was to provide input regarding the study design and review of the final study report to the FDA and review of the manuscript. The OPC had no role in the conduct of the study, in the collection, management, analysis, or interpretation of the data, in preparation or approval of the final manuscript, or in the decision to submit the manuscript for publication.

ClinicalTrials.gov identifier: NCT02660619.

Dr. Hasin has a contract with the Opioid Postmarketing Consortium to contribute supportive services to U.S. Food and Drug Administration postmarketing requirement study 3033-1 on the prevalence and incidence of misuse of, abuse of, and addiction to prescription opioid analgesics. Dr. Aharonovich has a contract with Alkermes for a study of self-reported cognitive and other functioning in subjects on buprenorphine or extended-release naltrexone treatment for opioid use disorder. Dr. Petronis was an employee of Pfizer at the time of this work and has stocks/stock options in Pfizer. Dr. Ross has received research funding from Reset Pharmaceuticals. Dr. Inturrisi has a licensing agreement with and receives consulting fees from Relmada Therapeutics for the development of d-methadone for the treatment of psychiatric symptoms, for the treatment or prevention of disorders of the nervous system and symptoms and manifestations thereof, and for cytoprotection against diseases and aging of cells. The other authors report no financial relationships with commercial interests.

References

1 Nahin RL, Sayer B, Stussman BJ, et al.: Eighteen-year trends in the prevalence of, and health care use for, noncancer pain in the United States: data from the Medical Expenditure Panel Survey. J Pain 2019; 20:796–809Crossref, MedlineGoogle Scholar

2 US National Institute on Drug Abuse: Prescription opioids and heroin research report: increased drug availability is associated with increased use and overdose. 2018. www.drugabuse.gov/publications/research-reports/prescription-opioids-heroin/increased-drug-availability-associated-increased-use-overdoseGoogle Scholar

3 US Centers for Disease Control and Prevention: US: opioid dispensing rate maps 2020. www.cdc.gov/drugoverdose/maps/rxrate-maps.htmlGoogle Scholar

4 US Centers for Disease Control and Prevention: Drug overdose deaths. 2021. www.cdc.gov/drugoverdose/data/statedeaths.htmlGoogle Scholar

5 Volkow ND, Jones EB, Einstein EB, et al.: Prevention and treatment of opioid misuse and addiction: a review. JAMA Psychiatry 2019; 76:208–216Crossref, MedlineGoogle Scholar

6 Volkow ND, Blanco C: The changing opioid crisis: development, challenges, and opportunities. Mol Psychiatry 2021; 26:218–233Crossref, MedlineGoogle Scholar

7 Blanco C, Wall MM, Okuda M, et al.: Distinguishing between physiological dependence and substance use disorder: response to Ruan et al (letter). Am J Psychiatry 2017; 174:79–80LinkGoogle Scholar

8 Ruan X, Chiravuri S, Kaye AD: Pain as a predictor for opioid use disorder. Am J Psychiatry 2017; 174:79LinkGoogle Scholar

9 Volkow ND, McLellan AT: Opioid abuse in chronic pain: misconceptions and mitigation strategies. N Engl J Med 2016; 374:1253–1263Crossref, MedlineGoogle Scholar

10 American Psychiatric Association: Diagnostic and Statistical Manual of Mental Disorders, 5th ed. Washington, DC, American Psychiatric Association, 2013CrossrefGoogle Scholar

11 O’Brien C: Addiction and dependence in DSM-V. Addiction 2011; 106:866–867Crossref, MedlineGoogle Scholar

12 Kaye AD, Jones MR, Kaye AM, et al.: Prescription opioid abuse in chronic pain: an updated review of opioid abuse predictors and strategies to curb opioid abuse: part 1. Pain Physician 2017; 20:S93–S109Crossref, MedlineGoogle Scholar

13 Ballantyne JC, Sullivan MD, Kolodny A: Opioid dependence vs addiction: a distinction without a difference? Arch Intern Med 2012; 172:1342–1343Crossref, MedlineGoogle Scholar

14 Chang Y-P, Compton P: Management of chronic pain with chronic opioid therapy in patients with substance use disorders. Addict Sci Clin Pract 2013; 8:21Crossref, MedlineGoogle Scholar

15 Coplan PM, Cepeda MS, Petronis KR, et al.: Postmarketing studies program to assess the risks and benefits of long-term use of extended-release/long-acting opioids among chronic pain patients. Postgrad Med 2020; 132:44–51Crossref, MedlineGoogle Scholar

16 Højsted J, Sjøgren P: Addiction to opioids in chronic pain patients: a literature review. Eur J Pain 2007; 11:490–518Crossref, MedlineGoogle Scholar

17 Vowles KE, McEntee ML, Julnes PS, et al.: Rates of opioid misuse, abuse, and addiction in chronic pain: a systematic review and data synthesis. Pain 2015; 156:569–576Crossref, MedlineGoogle Scholar

18 Lapham G, Boudreau DM, Johnson EA, et al.: Prevalence and treatment of opioid use disorders among primary care patients in six health systems. Drug Alcohol Depend 2020; 207:107732Crossref, MedlineGoogle Scholar

19 Fishbain DA, Cole B, Lewis J, et al.: What percentage of chronic nonmalignant pain patients exposed to chronic opioid analgesic therapy develop abuse/addiction and/or aberrant drug-related behaviors? A structured evidence-based review. Pain Med 2008; 9:444–459Crossref, MedlineGoogle Scholar

20 Boscarino JA, Withey CA, Dugan RJ, et al.: Opioid medication use among chronic non-cancer pain patients assessed with a modified drug effects questionnaire and the association with opioid use disorder. J Pain Res 2020; 13:2697–2705Crossref, MedlineGoogle Scholar

21 Boscarino J, Hoffman S, Han J: Opioid-use disorder among patients on long-term opioid therapy: impact of final DSM-5 diagnostic criteria on prevalence and correlates. Subst Abuse Rehabil 2015; 6:83–91Crossref, MedlineGoogle Scholar

22 Boscarino JA, Rukstalis MR, Hoffman SN, et al.: Prevalence of prescription opioid-use disorder among chronic pain patients: comparison of the DSM-5 vs DSM-4 diagnostic criteria. J Addict Dis 2011; 30:185–194Crossref, MedlineGoogle Scholar

23 Banta-Green CJ, Merrill JO, Doyle SR, et al.: Opioid use behaviors, mental health, and pain: development of a typology of chronic pain patients. Drug Alcohol Depend 2009; 104:34–42Crossref, MedlineGoogle Scholar

24 Degenhardt L, Bruno R, Lintzeris N, et al.: Agreement between definitions of pharmaceutical opioid use disorders and dependence in people taking opioids for chronic non-cancer pain (POINT): a cohort study. Lancet Psychiatry 2015; 2:314–322Crossref, MedlineGoogle Scholar

25 Hasin DS, O’Brien CP, Auriacombe M, et al.: DSM-5 criteria for substance use disorders: recommendations and rationale. Am J Psychiatry 2013; 170:834–851LinkGoogle Scholar

26 Livne O, Shmulewitz D, Stohl M, et al.: Agreement between DSM-5 and DSM-IV measures of substance use disorders in a sample of adult substance users. Drug Alcohol Depend 2021; 227:108958Crossref, MedlineGoogle Scholar

27 Dennis B, Bawor M, Paul J, et al.: Pain and opioid addiction: a systematic review and evaluation of pain measurement in patients with opioid dependence on methadone maintenance treatment. Curr Drug Abuse Rev 2016; 9:49–60Crossref, MedlineGoogle Scholar

28 Nunes EV, Jr., Scodes JM, Pavlicova M, et al.: Sublingual buprenorphine-naloxone compared with injection naltrexone for opioid use disorder: potential utility of patient characteristics in guiding choice of treatment. Am J Psychiatry 2021; 178:660–671LinkGoogle Scholar

29 Weiss RD, Potter JS, Fiellin DA, et al.: Adjunctive counseling during brief and extended buprenorphine-naloxone treatment for prescription opioid dependence: a 2-phase randomized controlled trial. Arch Gen Psychiatry 2011; 68:1238–1246Crossref, MedlineGoogle Scholar

30 Kosten TR, Rounsaville BJ, Babor TF, et al.: Substance-use disorders in DSM-III-R: evidence for the dependence syndrome across different psychoactive substances. Br J Psychiatry 1987; 151:834–843Crossref, MedlineGoogle Scholar

31 Rounsaville BJ, Kosten TR, Williams JB, et al.: A field trial of DSM-III-R psychoactive substance dependence disorders. Am J Psychiatry 1987; 144:351–355LinkGoogle Scholar

32 Sutherland G, Edwards G, Taylor C, et al.: The measurement of opiate dependence. Br J Addict 1986; 81:485–494Crossref, MedlineGoogle Scholar

33 Wakefield JC: DSM-5 substance use disorder: how conceptual missteps weakened the foundations of the addictive disorders field. Acta Psychiatr Scand 2015; 132:327–334Crossref, MedlineGoogle Scholar

34 Manhapra A, Sullivan MD, Ballantyne JC, et al.: Complex persistent opioid dependence with long-term opioids: a gray area that needs definition, better understanding, treatment guidance, and policy changes. J Gen Intern Med 2020; 35:964–971Crossref, MedlineGoogle Scholar

35 Ballantyne JC, Sullivan MD, Koob GF: Refractory dependence on opioid analgesics. Pain 2019; 160:2655–2660Crossref, MedlineGoogle Scholar

36 Manhapra A, Arias AJ, Ballantyne JC: The conundrum of opioid tapering in long-term opioid therapy for chronic pain: a commentary. Subst Abus 2018; 39:152–161Crossref, MedlineGoogle Scholar

37 Eaton NR, Rodriguez-Seijas C, Carragher N, et al.: Transdiagnostic factors of psychopathology and substance use disorders: a review. Soc Psychiatry Psychiatr Epidemiol 2015; 50:171–182Crossref, MedlineGoogle Scholar

38 Kotov R, Ruggero CJ, Krueger RF, et al.: New dimensions in the quantitative classification of mental illness. Arch Gen Psychiatry 2011; 68:1003–1011Crossref, MedlineGoogle Scholar

39 Vrieze SI, Perlman G, Krueger RF, et al.: Is the continuity of externalizing psychopathology the same in adolescents and middle-aged adults? A test of the externalizing spectrum’s developmental coherence. J Abnorm Child Psychol 2012; 40:459–470Crossref, MedlineGoogle Scholar

40 Keyes KM, Eaton NR, Krueger RF, et al.: Thought disorder in the meta-structure of psychopathology. Psychol Med 2013; 43:1673–1683Crossref, MedlineGoogle Scholar

41 Kendler KS, Lönn SL, Maes HH, et al.: A Swedish population-based multivariate twin study of externalizing disorders. Behav Genet 2016; 46:183–192Crossref, MedlineGoogle Scholar

42 Krueger RF: The structure of common mental disorders. Arch Gen Psychiatry 1999; 56:921–926Crossref, MedlineGoogle Scholar

43 Gustavson DE, Franz CE, Panizzon MS, et al.: Internalizing and externalizing psychopathology in middle age: genetic and environmental architecture and stability of symptoms over 15–20 years. Psychol Med 2019; 9:1530–1538Google Scholar

44 Barr PB, Dick DM: The genetics of externalizing problems. Curr Top Behav Neurosci 2020; 47:93–112Crossref, MedlineGoogle Scholar

45 Deak JD, Johnson EC: Genetics of substance use disorders: a review. Psychol Med 2021; 51:2189–2200Crossref, MedlineGoogle Scholar

46 Krueger RF, Chentsova-Dutton YE, Markon KE, et al.: A cross-cultural study of the structure of comorbidity among common psychopathological syndromes in the general health care setting. J Abnormal Psychol 2003; 112:437–447Crossref, MedlineGoogle Scholar

47 Krueger RF, Tackett JL, Markon KE: Structural models of comorbidity among common mental disorders: connections to chronic pain. Adv Psychosom Med 2004; 25:63–77Crossref, MedlineGoogle Scholar

48 Cicero TJ, Ellis MS, Surratt HL, et al.: The changing face of heroin use in the United States: a retrospective analysis of the past 50 years. JAMA Psychiatry 2014; 71:821–826Crossref, MedlineGoogle Scholar

49 Zale EL, Powers JM, Ditre JW: Cognitive-affective transdiagnostic factors associated with vulnerability to alcohol and prescription opioid use in the context of pain. Alcohol Res 2021; 41:08Crossref, MedlineGoogle Scholar

50 Pantazis CB, Gonzalez LA, Tunstall BJ, et al.: Cues conditioned to withdrawal and negative reinforcement: neglected but key motivational elements driving opioid addiction. Sci Adv 2021; 7:eabf0364Crossref, MedlineGoogle Scholar

51 Carpenter RW, Lane SP, Bruehl S, et al.: Concurrent and lagged associations of prescription opioid use with pain and negative affect in the daily lives of chronic pain patients. J Consulting Clin Psychol 2019; 87:872–886Crossref, MedlineGoogle Scholar

52 Witkiewitz K, Vowles KE: Alcohol and opioid use, co-use, and chronic pain in the context of the opioid epidemic: a critical review. Alcohol Clin Exp Res 2018; 42:478–488Crossref, MedlineGoogle Scholar

53 US Food and Drug Administration: Timeline of selected FDA activities and significant events addressing opioid misuse and abuse. 2019. www.fda.gov/drugs/information-drug-class/timeline-selected-fda-activities-and-significant-events-addressing-opioid-misuse-and-abuseGoogle Scholar

54 Hasin D, Samet S, Nunes E, et al.: Diagnosis of comorbid psychiatric disorders in substance users assessed with the Psychiatric Research Interview for Substance and Mental Disorders for DSM-IV. Am J Psychiatry 2006; 163:689–696LinkGoogle Scholar

55 Hasin DS, Trautman KD, Miele GM, et al.: Psychiatric Research Interview for Substance and Mental Disorders (PRISM): reliability for substance abusers. Am J Psychiatry 1996; 153:1195–1201LinkGoogle Scholar

56 Torrens M, Serrano D, Astals M, et al.: Diagnosing comorbid psychiatric disorders in substance abusers: validity of the Spanish versions of the Psychiatric Research Interview for Substance and Mental Disorders and the Structured Clinical Interview for DSM-IV. Am J Psychiatry 2004; 161:1231–1237LinkGoogle Scholar

57 Hasin DS, Shmulewitz D, Stohl M, et al.: Procedural validity of the AUDADIS-5 depression, anxiety, and post-traumatic stress disorder modules: substance abusers and others in the general population. Drug Alcohol Depend 2015; 152:246–256Crossref, MedlineGoogle Scholar

58 Hasin DS, Greenstein E, Aivadyan C, et al.: The Alcohol Use Disorder and Associated Disabilities Interview Schedule-5 (AUDADIS-5): procedural validity of substance use disorders modules through clinical re-appraisal in a general population sample. Drug Alcohol Depend 2015; 148:40–46Crossref, MedlineGoogle Scholar

59 Hasin D, Shmulewitz D, Stohl M, et al.: Test-retest reliability of DSM-5 substance disorder measures as assessed with the PRISM-5, a clinician-administered diagnostic interview. Drug Alcohol Depend 2020; 216:108294Crossref, MedlineGoogle Scholar

60 Von Korff M, Walker RL, Saunders K, et al.: Prevalence of prescription opioid use disorder among chronic opioid therapy patients after health plan opioid dose and risk reduction initiatives. Int J Drug Policy 2017; 46:90–98Crossref, MedlineGoogle Scholar

61 Goldstein RB, Grant BF, Huang B, et al.: Lack of remorse in antisocial personality disorder: sociodemographic correlates, symptomatic presentation, and comorbidity with axis I and axis II disorders in the National Epidemiologic Survey on Alcohol and Related Conditions. Compr Psychiatry 2006; 47:289–297Crossref, MedlineGoogle Scholar

62 Dawson DA, Grant BF: Family history of alcoholism and gender: their combined effects on DSM-IV alcohol dependence and major depression. J Stud Alcohol 1998; 59:97–106Crossref, MedlineGoogle Scholar

63 Cleland C, Magura S, Foote J, et al.: Factor structure of the Conners Adult ADHD Rating Scale (CAARS) for substance users. Addict Behav 2006; 31:1277–1282Crossref, MedlineGoogle Scholar

64 Hoyle RH, Stephenson MT, Palmgreen P, et al.: Reliability and validity of a brief measure of sensation seeking. Pers Individ Differ 2001; 32:401–414CrossrefGoogle Scholar

65 Weiss RD, Potter JS, Griffin ML, et al.: Reasons for opioid use among patients with dependence on prescription opioids: the role of chronic pain. J Subst Abuse Treat 2014; 47:140–145Crossref, MedlineGoogle Scholar

66 Griffin ML, McDermott KA, McHugh RK, et al.: Longitudinal association between pain severity and subsequent opioid use in prescription opioid dependent patients with chronic pain. Drug Alcohol Depend 2016; 163:216–221Crossref, MedlineGoogle Scholar

67 Keller S, Bann CM, Dodd SL, et al.: Validity of the brief pain inventory for use in documenting the outcomes of patients with noncancer pain. Clin J Pain 2004; 20:309–318Crossref, MedlineGoogle Scholar

68 Cohen J: A coefficient of agreement for nominal scales. Educ Psychol Meas 1960; 20:37–46CrossrefGoogle Scholar

69 Landis JR, Koch GG: The measurement of observer agreement for categorical data. Biometrics 1977; 33:159–174Crossref, MedlineGoogle Scholar

70 Davis M, Fleiss JL: Measuring agreement for multinomial data. Biometrics 1982; 38:1047–1051CrossrefGoogle Scholar

71 Fitzmaurice GM, Laird NM, Zahner GEP, et al.: Bivariate logistic regression analysis of childhood psychopathology ratings using multiple informants. Am J Epidemiol 1995; 142:1194–1203Crossref, MedlineGoogle Scholar

72 Goldwasser MA, Fitzmaurice GM: Multivariate linear regression analysis of childhood psychopathology using multiple informant data. Int J Methods Psychiatr Res 2001; 10:1–19CrossrefGoogle Scholar

73 Epstein DH, Heilig M, Shaham Y: Science-based actions can help address the opioid crisis. Trends Pharmacol Sci 2018; 39:911–916Crossref, MedlineGoogle Scholar

74 Oldfield BJ, Edens EL, Agnoli A, et al.: Multimodal treatment options, including rotating to buprenorphine, within a multidisciplinary pain clinic for patients on risky opioid regimens: a quality improvement study. Pain Med 2018; 19:S38–S45Crossref, MedlineGoogle Scholar

75 Sullivan MD, Ballantyne JC: When physical and social pain coexist: insights into opioid therapy. Ann Fam Med 2021; 19:79–82Crossref, MedlineGoogle Scholar

76 Serafini G, Adavastro G, Canepa G, et al.: The efficacy of buprenorphine in major depression, treatment-resistant depression, and suicidal behavior: a systematic review. Int J Mol Sci 2018; 19:2410CrossrefGoogle Scholar

77 Dinoff A, Lynch ST, Sekhri N, et al.: A meta-analysis of the potential antidepressant effects of buprenorphine versus placebo as an adjunctive pharmacotherapy for treatment-resistant depression. J Affect Disord 2020; 271:91–99Crossref, MedlineGoogle Scholar

78 Ballantyne JC, Sullivan MD: Discovery of endogenous opioid systems: what it has meant for the clinician’s understanding of pain and its treatment. Pain 2017; 158:2290–2300Crossref, MedlineGoogle Scholar

79 Gorfinkel L, Voon P, Wood E, et al.: Diagnosing opioid addiction in people with chronic pain. BMJ 2018; 362:k3949Crossref, MedlineGoogle Scholar

80 Owen-Smith A, Stewart C, Sesay MM, et al.: Chronic pain diagnoses and opioid dispensings among insured individuals with serious mental illness. BMC Psychiatry 2020; 20:40Crossref, MedlineGoogle Scholar

81 Seal KH, Shi Y, Cohen G, et al.: Association of mental health disorders with prescription opioids and high-risk opioid use in US veterans of Iraq and Afghanistan. JAMA 2012; 307:940–947Crossref, MedlineGoogle Scholar

82 Hasin DS, Shmulewitz D, Cerda M, et al.: US adults with pain, a group increasingly vulnerable to nonmedical cannabis use and cannabis use disorder: 2001–2002 and 2012–2013. Am J Psychiatry 2020; 177:611–618LinkGoogle Scholar