Substance Use, Gambling, Binge-Eating, and Hypersexuality Symptoms Among Patients Receiving Opioid Agonist Therapies

Abstract

Background and Objectives:

Patients receiving opioid agonist therapies have high rates of psychiatric comorbidity. Some data suggest that comorbidity is associated with poorer treatment outcomes. The current study assessed predictors of multiple putative addictive behaviors among patients receiving opioid agonist therapies.

Methods:

Adults (N=176) recruited from an outpatient clinic providing opioid agonist therapy completed self-report measures of depression, anxiety, impulsivity, adverse childhood events, and the RAD scale, which includes 7 subscales assessing symptoms related to alcohol use, drug use, tobacco use, gambling, binge-eating, and hypersexual behavior. Linear regression and hurdle models identified significant predictors of RAD subscales. Hurdle models included logistic regression estimation for the presence/absence of symptoms and negative binomial regression for estimation of the severity of symptoms.

Results:

Most patients did not report significant symptoms beyond drug or tobacco use. However, 7% - 47% of participants reported some symptoms of other addictive behaviors (subscale score > 0). Higher impulsivity predicted the presence and/or increased severity of symptoms of drug use, gambling, binge-eating, and hypersexuality. Higher depression significantly predicted increased severity of drug use and binge-eating symptoms. Increased anxiety predicted lower severity of alcohol use and binge-eating and higher severity of smoking symptoms.

Conclusion:

A broader range of potentially addictive symptoms may be present among patients engaged in treatment for opioid use disorder.

Scientific Significance:

Few studies have assessed symptoms of binge-eating, hypersexuality, and excessive video-gaming among patients receiving opioid agonist therapy. This study contributes preliminary findings and highlights important future directions.

1. Introduction

Opioid use disorder continues to be a leading contributor to preventable death in the United States¹. Opioid agonist therapies include methadone and buprenorphine, which represent highly efficacious treatments. Opioid agonists address physiological dependence and are often combined with other treatment modalities (e.g., psychotherapy) to improve functioning and quality of life for people with opioid addiction². Growing evidence reveals high rates of psychiatric comorbidity among patients in treatment for opioid use disorder, including non-opioid substance use disorders^3,4. Substance use disorder comorbidity can be problematic due to synergistically increasing the risks of pharmacologic treatments⁵ and decreasing treatment engagement³. For example, comorbid anxiety disorders are associated with more severe clinical presentation and poorer prognosis for opioid use disorder treatment⁶. In addition, comorbid personality disorder and/or alcohol use disorder are associated with significantly greater all-cause mortality amongst individuals with opioid use disorder⁷. Comorbid alcohol use disorder is also associated with socioeconomic disadvantage, poorer psychiatric history, and higher likelihood of dependence on other substances⁸.

It is also important to consider non-substance forms of addiction when examining potential comorbidities in opioid treatment. The Diagnostic and Statistical Manual-Fifth Edition (DSM-5)⁹ recognizes a single form of non-substance addiction, gambling disorder, and identifies internet gaming disorder as a condition in need of further study. Thus far, most data point to an increased risk of problematic gambling or gambling disorder among patients engaged in treatment for opioid use disorder^10,11, and it is a risk factor for continued substance use and treatment discontinuation¹¹. The classification of internet gaming disorder has garnered considerable criticism related to the exclusion of offline games¹². Neither internet gaming nor video-gaming more generally have been characterized among patients receiving opioid agonist therapies. However, given that a number of novel interventions propose using videogame playing as a treatment delivery device, it is important to understand if those in opioid use disorder treatment are at elevated risk for excessive video-gaming. For example, gaming approaches to addressing delay discounting and other forms of impulsive decision making have been recently suggested for those addicted to heroin^13,14. While these studies offer interesting new ways to assess decision-making capacity and changes therein, if they are to be disseminated more widely, the associated risks (or benefits) for this population should first be elucidated.

Beyond gambling and internet gaming disorder, there are additional conditions or behaviors that may be related to addiction but are not formally recognized as addictive disorders, including binge-eating¹⁵ and hypersexuality. There is some evidence that overeating difficulties (e.g., binge eating, food addiction, or loss-of-control eating) are present for patients engaged in treatment for opioid use disorder and is associated with increased psychopathology in several domains¹⁶. Hypersexual behavior includes nonparaphilic repetitive and intense preoccupation with sexual fantasies or behaviors leading to clinically significant distress or impairment¹⁷. The condition is not formally recognized with the DSM-5 diagnostic system. However, it shares many common risk factors to other addictive behaviors, such as impulsivity¹⁸ and commonly co-occurs with other addictive disorders¹⁹. For patients engaged in opioid agonist treatment, sexual dysfunction have been reported as side effects of opioid agonist therapy²⁰. To date, most studies have assessed either sexual dysfunction or risky sexual behavior among patients with opioid use disorder. While potentially related to hypersexuality, risky sexual behavior doesn’t always include cognitive/affective components and distress or impairment¹⁷. No studies to our knowledge have examined whether a subset of patients receiving opioid agonist therapies experience or continue to exhibit hypersexual behavior. Given the unknown prevalence and the increased health risks associated with hypersexual behavior,²¹ there remains a need to assess hypersexual behavior among patients receiving opioid agonist therapies.

The current literature is inadequate for describing the symptomology and correlates of binge-eating, hypersexuality, video-gaming, and to a lesser degree gambling, among patients engaged in opioid agonist treatment. There are data reflecting correlates of related presentations, such as substance use disorders in isolation or more well-characterized comorbidities (e.g., opioid use disorder and tobacco use disorder). For example, adverse childhood events are associated with reporting current problematic substance use, gambling, addictive-like eating, and risky sexual behavior^22,23. Other potential transdiagnostic correlates include self-reported impulsivity²⁴, depression²⁵, and anxiety²⁶, which is robustly tied to substance and other addictive disorders. However, empirical investigation is needed to determine if these correlates predict non-substance addictions or related behaviors. Ultimately, understanding the correlates of multiple addictive behaviors may help identify potential treatment targets among this population.

To address these gaps in the literature, in a sample of patients engaged in opioid agonist treatment, we investigated correlates of 7 types of potentially addictive behaviors, including alcohol use, drug use, tobacco use, gambling, binge-eating, hypersexuality, and video-gaming. Demographic factors, depression, anxiety, impulsivity, and adverse childhood events were investigated as potential correlates. These factors, while not exhaustive, have robust associations with addiction, and represent an important preliminary step. Based on the minimal prior literature, we broadly hypothesized positive associations between the key correlates (depression, anxiety, impulsivity, and adverse childhood events) and self-reported addiction symptoms.

2. Methods

2.1. Participants.

Participants included 176 adults recruited from a single urban outpatient treatment center providing outpatient therapy and opioid agonist treatment in the Midwestern United States. Participants were recruited from May 2017 to September 2018. Participants were not consecutive cases and could enroll in the study at any point in their treatment. To increase the likelihood that all patients within the clinic had the opportunity to enroll in the present study if interested, research staff conducted study recruitment on variable days and times. The majority of patients in this setting received methadone, though a small number of patients received buprenorphine. Included participants must have been between the ages of 18 and 75 and report ability to read and write in English. No other exclusionary criteria were used. Participants completed a battery of self-report measures either electronically on a windows tablet device or using paper and pencil. To accommodate people with reading difficulties, research staff offered to read questions and use response cards indicating possible Likert responses. A research assistant was available to answer questions. For both paper and pencil and electronic versions, a response was requested for all questions, but a participant could choose not to answer. Missing data for the constructs of interest was minimal (less than or equal to 1.7%). Participants were compensated $5 for their time. All participants provided informed consent, and this study received approval from the Wayne State University School of Medicine Institutional Review Board [IRB Protocol Number: 115516B3F].

2.2. Measures

Sociodemographic variables.

Demographic and background variables included: age, race/ethnicity, gender, education, employment, and income.

Recognizing Addictive Disorders scale²⁷.

The RAD is a 35-item self-report measure assessing symptoms of 7 potential substance or other addictive disorders within the last 3 months, including alcohol use, drug use, smoking, gambling, video-gaming (on and offline), and hypersexual behavior. Every subscale has five items, with a single item corresponding to one of the following DSM-5 criteria: 1) using over a longer period of time than was intended, 2) physical or psychological problems made worse, 3) social or interpersonal problems, 4) neglecting major roles, and 5) craving. These DSM-5 criteria were chosen as the basis for item generation because of strong discriminative properties as well as a range of item difficulties based on data from National Epidemiologic Survey on Alcohol and Related Conditions²⁸. Responses for each item ranged from 0 (Does not describe me at all) to 6 (Describes me very well), and total scores can range from 0-210. The scale was developed with an online community sample and exhibited a supported factor structure and good construct validity²⁷. In the current sample, the total score (Cronbach α = .90) and subscale scores (Cronbach α_range = .83-.96) indicated good internal consistency reliability.

Adverse Childhood Experience Scale (ACES)²².

ACES is a 10-item self-report measure assessing dichotomously (absent/present) the number of adverse or traumatic events that occurred before the age of 18. Note that events from this scale are considered independent and therefore assumptions related to internal consistency do not apply. Some studies have suggested poor test-retest reliability for specific items (range = .51 - .69)²⁹, but test-retest reliability for the sum of the items is acceptable over intervals ranging from 7 weeks to an average of 9 months^30,31.

Patient Health Questionnaire (PHQ)³².

The PHQ includes a 9-item self-report measure of depression over the past two weeks based on the nine diagnostic criteria listed in the DSM-IV TR for major depressive disorder (PHQ-9), using a 4-point Likert scale: 0 = Not at all to 3 = Nearly every day. Scores on the PHQ-9 range from 0 to 27. Systematic review indicates good internal consistency, test-retest reliability, and convergent validity with other widely utilized assessments of depression³³.

The GAD-7 is a 7-item self-report measure of anxiety over the past 6-months based on DSM-IV TR criteria for generalized anxiety disorder (GAD) using a 4-point Likert scale: 0 = Not at all to 3 = Nearly every day. Systematic review indicates good construct validity based on agreement with diagnostic interviews³⁴, and a study of patients with addiction showed good internal consistency and test-retest reliability of the measure³⁵.

The PHQ-9 and GAD-7 exhibited good internal consistency reliability in the current sample, with Cronbach’s α = .87 and .92 respectively.

Shortened UPPS-P³⁶ is a 20-item self-report measure assessing multidimensional aspects of impulsivity. Subscale scores or a total score can be used (range 20-80). The shortened UPPS-P was developed by retaining four items for each of the five factors of the original UPPS-P. Participants rate the extent to which they agree with the statements on a 4-point Likert scale: 1 = Strongly agree to 4 = Strongly disagree. Psychometric analysis indicate strong internal consistency for the total score and convergent validity with other measures of impulsivity³⁶. The total scale exhibited good reliability in the current sample Cronbach’s α = .82, which supported use of the total score in the current study.

2.3. Statistical Analyses

All RAD subscales have a range of 0 to 30, though examination of the distribution of subscale scores revealed differences across subscales. The smoking and drug use subscales showed moderate variability and the skew was within normal limits (i.e., < 1.00). Linear regression was used to model predictors of the drug and smoking subscales. The video-gaming subscale exhibited extreme zero-inflation, with 91% of the sample scoring 0. Due to the lack of available data beyond 0 scores, this subscale was not modeled. The alcohol, gambling, binge-eating, and hypersexuality subscales showed significant zero-inflation. Hurdle models with robust maximum likelihood estimation were used to estimate subscale scores, which considers two distinct processes: 1) predicting a zero versus a non-zero score; and 2) if not a zero, the positive continuous score^37,38, which in the case of Likert-type responses can be interpreted as an indicator of severity. Logistic regression is used to estimate zero versus non-zero scores, while a number of approaches can be used to estimate the non-zero positive continuous scores (e.g., linear regression, Poisson regression, and negative binomial regression). Due to continued skew after truncating at zero and lack of support for equidispersion, negative binomial regression was used. See Pittman et al³⁸, for a fuller discussion on the assumptions, theory, and application of zero-inflated approaches in the context of addiction research. Outcomes for logistic regression models are reported in odds ratios (OR). Outcomes for negative binomial regression are reported as incidence rate ratios (IRR) or the expected rate of increase/decrease in the dependent variable for a 1-point increase in the independent variable, with all other variables held constant. Additionally, [1-IRR X 100] can used to report the percent increase/decrease in the expected score. Odds ratios and IRR are calculated with 95% confidence intervals, and intervals that do not contain 0 are considered statistically significant at the .05 level. IBM’s SPSS Version 26 and Mplus Version 8.4 were used for descriptive procedures and regression procedures respectively.

3. Results

3.1. Sample Characteristics

Table 1 includes a summary of demographic and clinical features of the sample. The majority of participants were female (56.8% n = 100), and Black/African American 59.1% (n = 104), and the largest proportion of the sample made less than $9,000 annually (44.7%, n = 68). The clinic from which patients were drawn serves an urban and predominately African American community and is located in an area with higher rates of poverty than the state³⁹. Consistent with other studies from our clinic^40,41, our sample contained a higher proportion of African American patients and a higher proportion of older patients than has been reported in other clinics in the United States⁴². On average, participants endorsed more than three adverse childhood events. The most commonly reported events included parental separation or divorce (62.3%), parental substance use disorders (51.4%), and verbal abuse (40.9%). The average depression and anxiety score were in the mild range.

Table 1.

Sample Clinical Features

Gender, n (%)
Male	75 (42.6)
Female	100 (56.8)
Transgender/Gender non-binary	1 (0.6)
Race/Ethnicity, n (%)
White	54 (30.7)
Black	104 (59.1)
Bi/Multiracial	15 (8.5)
Other	3 (1.8)
Education, n (%)
Less than HS	40 (22.7)
HS diploma/GED	86 (48.9)
Greater than HS	50 (28.4)
Employment, n (%)
Employed full-time	13 (7.4)
Employed part-time	20 (11.4)
Unemployed	116 (65.9)
Other	27 (15.3)
Income
≥ 9,000	68 (44.7)
10,000-24,000	52 (34.2)
≤ 24,000	32 (21.1)
Age, M(SD)	51.27 (14.99)
Adverse Childhood Events M(SD)a	3.44 (2.72)
Depression M(SD)b	7.99 (5.98)
Anxiety M(SD)c	6.37 (5.91)
Impulsivity M(SD)d	21.94 (9.15)

Notes.

^a.Min. = 0, Max. = 10

^b.Min. = 0, Max. = 27, Scores < 4,9, 14, and 19 represent minimal or none, mild, moderate, moderately severe, and severe depression respectively

^c.Min = 0, Max = 21, < 4,9, and 14 represent none, mild, moderate, and severe anxiety respectively

^d.Min = 0, Max = 60

3.2. Distribution of RAD Subscale Scores

Figure 1 presents distributional information about each RAD subscale score, including the proportion of zero scores observed in the sample. The figure shows that while drug use symptoms and smoking symptoms were relatively common, addiction symptoms associated with the other five subscales were less common. Video-gaming was extremely rare. Due to instability of estimates generated from very small samples, video-gaming was not included in subsequent analyses.

Figure 1. Distribution of RAD Subscale Scores.

RAD subscale scores can range from 0-30. The figure showed the proportion of the sample that scored 0 (i.e., degree of zero-inflation) as well as proportion of scores in the following ranges: 1-10, 11-20, and 21-30. The data table below the graph contains the sample’s proportions for each range.

3.3. Predictors of RAD Subscale Scores

Table 2 presents results of the linear regression models for drug use and smoking symptoms. People with higher scores on depression and impulsivity scored higher on the drug subscale. Women and people with higher anxiety scores had higher smoking scores. No other predictors were statistically significant.

Table 2.

Linear Regression Models Predicting RAD Drug Use and Smoking Subscales

	Drug	Smoking
	β (95% CI), p-value	β (95% CI), p-value
Non-White	−0.59 (−4.31-3.14) p = 0.758	−3.05 (−7.68-1.58) p = 0.197
Women	−1.96 (−4.63-0.71) p = 0.149	4.35 (0.57-8.13) p = 0.024
Age	−0.03 (−0.15-0.08) p = 0.570	0.07 (−0.11-0.25) p = 0.430
Depression	0.45 (0.11-0.79) p = 0.010	−0.01 (−0.41-0.4) p = 0.979
Anxiety	0.01 (−0.33-0.34) p = 0.965	0.45 (0.03-0.86) p = 0.037
Impulsivity	0.27 (0.13-0.41) p < .001	0.05 (−0.13-0.22) p = 0.622
Adverse events	0.30 (−0.2-0.81) p = 0.239	0.14 (−0.5-0.78) p = 0.664

Notes. RAD = Recognizing Addictive Disorders scale. CI = confidence interval.

Bold values indicate significant β at the p < .05 level.

Table 3 presents the results for the hurdle models, including the logistic regression (reported as odds ratios) and negative binomial regression (reported as incidence rate ratios) models. Across subscales, the variables predicting presence/absence of symptoms were often distinct from the predictors associated with the severity scores. An interpretation of the significant alcohol coefficients is reviewed first, followed by a more general discussion of the significance and directionality of predictors for the remaining hurdle models. For alcohol severity scores, people who were non-White scored 48% lower compared to White people (i.e., 1 - .52 [incidence rate ratios]). Alternatively, one could interpret the coefficients as indicating that being non-White decreased the alcohol severity score by a factor of .52. For psychological correlates of the alcohol severity score, a one-point increase in anxiety is associated with a 5% decrease in the alcohol severity score, and a one-point increase in impulsivity is associated with a 7% increase in the alcohol severity score. For gambling symptoms, being older and more impulsive was associated with increased odds of having a non-zero score. For binge-eating symptoms, being a woman, older, and more impulsive increased your odds of having a non-zero score, while higher depression and lower anxiety scores were associated with increased severity of binge-eating symptoms. For hypersexuality, being a woman decreased the likelihood of a non-zero score, while higher scores impulsivity were associated with increased severity of hypersexuality symptoms. Childhood adverse events were unrelated to outcomes for all subscales. All odds ratios and incidence rate rations describe the impact of a given variables, holding all other variables constant.

Table 3.

Zero-Inflated Hurdle Models for RAD Subscales with Odds Ratios and Incidence Rate Ratios

	Alcohol		Gambling		Binge Eating		Hypersexuality
	OR (95%CI)	IRR (95%CI)	OR (95%CI)	IRR (95%CI)	OR (95%CI)	IRR (95%CI)	OR (95%CI)	IRR (95%CI)
Non-White	0.96 (0.35-2.61)	0.52 (0.23-0.80)	1.32 (0.40-4.36)	0.82 (0.10-1.54)	0.43 (0.16-1.17)	0.78 (0.39-1.17)	1.81 (0.68-4.87)	1.39 (0.67-2.10)
Women	0.86 (0.37-1.99)	1.06 (0.47-1.66)	1.21 (0.47-3.11)	2.03 (0.74-3.32)	2.34 (1.11-4.91)	1.19 (0.74-1.63)	0.21 (0.09-0.46)	1.03 (0.66-1.40)
Age	1.01 (0.98-1.05)	1.02 (0.99-1.04)	1.06 (1.02-1.10)	1.03 (0.99-1.06)	1.05 (1.02-1.09)	1.00 (0.99-1.02)	0.99 (0.96-1.02)	0.99 (0.97-1.00)
Depression	0.98 (0.88-1.09)	1.03 (0.98-1.08)	1.11 (0.99-1.24)	1.01 (0.98-1.05)	1.06 (0.97-1.16)	1.09 (1.05-1.13)	1.02 (0.94-1.11)	1.03 (0.99-1.07)
Anxiety	1.09 (0.98-1.21)	0.95 (0.91-0.99)	0.94 (0.84-1.06)	0.98 (0.94-1.02)	0.91 (0.83-1.01)	0.94 (0.90-0.98)	1.06 (0.96-1.16)	0.99 (0.95-1.03)
Impulsivity	1.05 (1.00-1.09)	1.07 (1.04-1.10)	1.07 (1.02-1.13)	1.02 (0.98-1.06)	1.05 (1.01-1.09)	1.02 (1.00-1.05)	1.02 (0.99-1.06)	1.03 (1.01-1.05)
Adverse events	1.03 (0.88-1.20)	0.99 (0.92-1.06)	1.19 (0.99-1.43)	1.02 (0.89-1.16)	1.06 (0.93-1.22)	0.99 (0.92-1.07)	0.99 (0.86-1.13)	0.97 (0.89-1.04)

Notes. RAD = Recognizing Addictive Disorders scale. OR = Odds Ration. IRR = incidence rate ratios. CI = confidence interval.

Bold values indicate significant OR and IRR at the p < .05 level.

4. Discussion

The current exploratory study characterized symptoms of 1) conditions formally recognized as substance-related or other addictive disorders (i.e., alcohol use drugs, drug use disorders, tobacco use disorders, and gambling disorder); and 2) conditions that are not formally recognized as addictive disorders but nonetheless have phenomenological and/or neurobiological similarities (i.e., binge-eating disorder and hypersexual behavior). The data showed that most patients do not report symptoms beyond those associated with drug use or tobacco use. However, between 8% and 46%, of participants reported at least some symptoms for conditions or behaviors other than drug use or tobacco use. The results suggest a broad range of putative addictions symptoms may be present for patients engaged in treatment for opioid use disorder.

Some important demographic features emerged as related to both the presence/absence of symptoms as well as the severity of symptoms. There was a large effect for race on alcohol symptom severity, with non-White people scoring 48% lower. Few studies have investigated the influence of race on alcohol severity among patients receiving opioid agonist treatment. The broader literature may suggest the opposite pattern for some non-White racial groups⁴³. For example, epidemiologic data indicate that Black people are less likely to have an alcohol use disorder, but when they do have an alcohol use disorder it is more likely to be severe⁴³. If replication using more nuanced categorization for race and a larger sample reveals that Black patients receiving opioid agonist treatment have significantly lower alcohol use symptoms, potential protective factors should be explored. Women were more likely to report at least some symptoms of binge-eating and less likely to report at least some symptoms of hypersexual behavior. Women were also more likely to have increased smoking symptoms. These findings broadly fit with existing data, which show the binge-eating is more common among women⁴⁴, hypersexuality is less common among women⁴⁵, and tobacco use disorders can be more severe and persistent among women⁴⁶.

Depression and anxiety were inconsistently related to RAD subscales. Depression was related to more severe drug use and binge-eating symptoms. Both drug use and binge-eating are highly comorbid with depression^47,48. Depression was unrelated to alcohol use, gambling, and hypersexuality subscales, which conflicts with robust epidemiologic research^49,50. Some small clinical studies have failed to find a relationship between gambling and depression⁵¹ or hypersexuality and depression¹⁹, though conflicting findings exist^52,53. Anxiety was negatively associated with both alcohol severity and binge-eating severity. Studies with more nuanced measures of anxiety have reported significant age group differences as well as some racial differences⁵⁴. More research is needed to examine if the observed results are related to unique characteristics of the sample, unique characteristics of patients engaged in opioid agonist therapy, the non-diagnostic nature of the RAD scale, or some combination of these factors

Impulsivity was significantly associated with the majority of subscales, which is consistent with other studies⁵⁵, and highlights self-reported impulsivity as important transdiagnostic feature of multiple addictive behaviors. The current exploratory study used a single impulsivity score, but future hypothesis-driven research may wish to consider the various facets of impulsivity, such as negative urgency.

These data revealed no significant associations between adverse childhood events and RAD subscales, which may also be related to unique sample characteristics. Specifically, this is a predominantly Black, low-income sample. Some scholars have hypothesized that perceptions of adversity may vary along racial and ethnic lines⁵⁶, which may impact association with addiction symptoms. Future research using qualitative and quantitative approaches should further investigate potential differences in the perception of adversity across demographic groups.

Rates of excessive video-gaming were extremely low in this population. Replication and extension should investigate if this is observed in other samples patients receiving opioid agonist therapies or if factors such as age, gender, and socioeconomic status significantly influence the prevalence of symptoms of excessive video-game playing. Data show that video-gaming is more common among younger males⁵⁷, and this sample was majority female with an average age of 51. Lower socioeconomic status may limit purchase and use of the technologies associated with video-gaming.

Study findings should be considered within the context of several limitations. First, RAD scores are based on self-report, and the RAD scale has not yet been tested for agreement with clinical diagnosis based on gold-standard methods (e.g., semi-structured interviews). As such, it is not yet known what level of symptomology measured by the RAD is consistent with a diagnosable condition. Such validation represents an important future direction. A related measurement issue is differences in the assessment window for the self-report scales—the RAD scale assesses symptoms over 3 months, the GAD-7 over 6 months, and the PHQ-9 over 2-weeks. Understanding the symptoms of addiction, depression, and anxiety within the same time period, or more robustly over time, represents an important future direction. Second, the study was limited to the constructs reported here and additional information specific to opioid agonist treatment settings would be helpful in more fully characterizing the sample, including type and dose of opioid agonist and length of time in treatment. Other non-treatment constructs of interest for future investigation could include additional psychiatric comorbidities; oppression-based traumatic stress⁵⁸; and additional demographic covariates, including sexual orientation. Sexual orientation and considerations of intersectionality may be particularly important given the strong association between minority-stress and addiction⁵⁹. Finally, limitations with regard to generalizability should be considered. This study sample was drawn from a single urban clinic with a higher proportion of African American and older patients than typically observed in opioid agonist treatment settings. A related limitation is that some behaviors of interest showed low levels of endorsement. Confidence in interpretations are increased by use of appropriate statistical models accounting for zero-inflation, but in some cases (e.g., video-gaming), the proportion of non-zero scores was based on a small subset of the sample, which also limits generalizability. Replication in a broader of methadone treatment centers represents an important future direction. Notwithstanding these limitations, this is the first study to characterize a broad range of addictive disorders within a sample engaged in opioid agonist therapy. Additionally, these data are reflective of subgroups that are often underrepresented in clinical research (i.e., racial/ethnic minorities and older adults), and may help to improve our understanding of the clinical presentation of these priority groups. Ultimately, if replication and extension identify significant non-substance addiction symptoms among patients engaged in opioid agonist treatments, screening procedures that consider a broad range of addiction symptoms may be indicated. Longitudinal research is also needed to characterize the trajectories of substance and non-substance addiction symptoms throughout treatment.