https://orcid.org/0000-0003-0773-9644
Abstract
Contemporary theories of attention-deficit/hyperactivity disorder (ADHD) and alcohol use disorder (AUD) emphasize core dysfunctions in reward-related processes and behaviors as pathognomonic characteristics. However, to date, it is unclear which domains of reward functioning are unique to ADHD versus AUD symptom dimensions, and which represent underlying shared correlates.
The current study employed secondary data analyses from a large community sample of emerging adults (N = 602; 57.3% female) and novel transdiagnostic modeling (i.e. bi-factor confirmatory factor analyses and structural equation modeling) of ADHD, AUD and shared symptom dimensions to identify unique and common reward-related dimensions: environmental suppressors, reward probability, hedonic capacity, proportionate substance-related reinforcement and delay discounting.
The presence of environmental suppressors was the only reward-related construct that correlated with the underlying ADHD-AUD shared dimension. The AUD symptom dimension was uniquely associated with proportionate substance-related reinforcement, whereas the ADHD symptom dimension was uniquely associated with limited reward probability. No significant associations were found for delay discounting or hedonic capacity.
These novel findings highlight specific aspects of reward-related functioning in ADHD, AUD and shared symptom dimensions. In so doing, this work meaningfully advances theoretical conceptualizations of these two commonly co-occurring presentations and suggests future directions for research on transdiagnostic correlates. Future longitudinal studies should include clinical samples with diagnoses of AUD and ADHD to further identify underlying correlates over time.
Rewards are central in motivating approach behaviors and learning (Verharen et al., 2020). Rewards also impact emotions by inducing pleasure with receipt, thereby increasing the likelihood of further engagement (i.e. reinforcement; Berridge and Kringelbach, 2008). Indeed, reward anticipation and receipt generally activate a repertoire of emotions and behaviors that impact everyday decisions, preferences and the cognitive resources that we allocate to accessing rewarding stimuli.
Given the prominent role of rewards in guiding behavior, researchers have focused on various reward-related domains in understanding the etiology, development and treatment of commonly co-occurring mental health symptoms that are typified by aberrant reward-related behavior. Transdiagnostic modeling of mental health symptoms, inspired by the Research Domain Criteria (RDoC) framework, is useful to this end. In part, this agenda emphasizes the use of advanced statistical approaches to identify shared mechanisms among multiple psychiatric phenotypes and unique mechanisms specific to psychiatric symptoms that reflect markers of differential risk (Sanislow et al., 2019).
This approach may clarify the co-occurrence of attention-deficit/hyperactivity disorder (ADHD) and alcohol use disorder (AUD). Converging evidence shows that ADHD, a neurodevelopmental disorder with onset in childhood, is a developmental risk factor for alcohol problems, which often begin in adolescence or emerging adulthood. Relative to those without ADHD, individuals with ADHD experience more alcohol-related negative consequences and evidence higher rates of AUD symptoms (e.g. Rooney et al., 2012). As many as 15–30% of adolescents and 35–55% of adults with ADHD have an alcohol/substance use disorder; conversely, ADHD has been documented in anywhere between 10% and 75% of treatment-seeking and general population samples of adolescents and adults with alcohol/substance use disorders (Wilens et al., 2008). The co-occurrence of ADHD and AUD is associated with poorer treatment prognosis for both disorders, yet evidence-based recommendations for treatment sequencing for co-occurring ADHD and AUD remain scarce. Therefore, identifying shared and transdiagnostic correlates of ADHD and AUD symptoms, consistent with translational research initiatives, is an essential step in developing and optimizing tailored treatments (Hershenberg and Goldfried, 2015).
Contemporary models of both ADHD and AUD emphasize dysfunctions in reward-related functioning (Carey et al., 2017). However, it is unclear which aspects of reward functioning are unique to ADHD versus AUD, and which represent shared correlates. Perhaps the most compelling evidence of a shared transdiagnostic reward-related correlate comes from delay discounting studies among people with ADHD or AUD symptoms. People with ADHD or AUD tend to show a strong preference for smaller-sooner rewards over larger-later rewards (i.e. delay discounting; Jackson and MacKillop, 2016). However, no studies to date have disaggregated ADHD and AUD symptoms into unique and shared components to evaluate the respective associations among ADHD, AUD and delay discounting. Thus, it is unclear if this pattern is associated with ADHD and AUD independently, or if it is indicative of a shared underlying dimension.
Other facets of reward-related processes that have never been examined as possible correlates of ADHD, AUD or their shared symptom dimension include the presence of deterrents to accessing rewards in the environment (i.e. environmental suppressors; Carvalho et al., 2011), availability of potentially reinforcing events that are rewarding and one’s perceived ability to access them (i.e. reward probability; Carvalho et al., 2011), strength of the reward experience once attained (i.e. hedonic capacity; Snaith et al., 1995) and a summary of the relative level of activity participation and enjoyment related to substance use versus substance-free activities (i.e. proportionate substance-related reinforcement; Acuff et al., 2019). Importantly, these constructs are conceptually linked, yet capture separable reward-related domains. For example, although positive reinforcers are typically experienced as pleasurable, environmental events that are unpleasant/aversive in form can also operate as positive reinforcers (e.g. increase behavior as a function of pleasure or freedom from aversive stimuli; Abreu and Santos, 2008, Carvalho et al., 2011). People with AUD exhibit difficulties accessing and deriving pleasure from natural stimuli, increasing the reinforcing efficacy of alcohol use. In a recursive loop, continued use of alcohol and other addictive substances can diminish the availability of substance-free rewarding stimuli, in part because frequent use can result in reduced sensitivity to reward in response to other stimuli (i.e. exercise and food) and diminished capacity to regulate behavior to attain delayed rewards. Additionally, the social/behavioral consequences of alcohol use may reduce access to reward (e.g. losing a job or damaging an interpersonal relationship).
ADHD symptomatology may be associated with similar reward-related processes. Given the constellation of symptoms and impairment, ADHD is often associated with aversive experiences and reduced access to reward (Garcia et al., 2012). For example, in their experience sampling study of adults, Knouse et al. (2008) showed that ADHD symptoms were associated with indices of general distress, including less positive and more negative mood, and less activity satisfaction in daily living. These findings support the possibility that adults with ADHD experience limited access to rewarding experiences, perhaps related to the presence of environmental suppressors to rewards or a limited capacity to attain rewards, but this remains unexamined. Relatedly, ADHD symptoms are thought to manifest, in part, from dysfunction in the brain reward cascade, which contributes to hypo-dopaminergic traits (Mereu et al., 2017). Some propose that these traits are consistent with high-risk drug-seeking behavior, given that drugs activate, among other neurochemicals, a dopamine release. Perhaps low hedonic capacity reflects a shared correlate of ADHD-AUD comorbidity.
These reward-related processes may be evident in a summary of a person’s pattern of activity participation and enjoyment related to substance use versus substance-free activities. Generally, people with AUD devote more time to, and derive more pleasure from, their alcohol use relative to substance-free activities (Acuff et al., 2019). ADHD symptoms are especially costly to multiple substance-free domains (e.g. academics and relationships). Many substance-free activities involve planning and organization, follow-through and regulated behavior—among core impairments of ADHD (Barkley, 1997). Drinking, in contrast, is often readily available in most young adult social circles and can be immediately rewarding (e.g. intoxication and social connection) with little effort/persistence. Perhaps experiencing more reinforcement from alcohol use than substance-free stimuli represents a shared underlying correlate of ADHD-AUD. However, no known research has included multiple reward-related constructs in a single model with symptom dimensions. Consequently, it is unknown whether these reward variables are associated with ADHD versus AUD symptoms, or a shared underlying ADHD-AUD symptom dimension.
THE CURRENT STUDY
There is limited research on specific aspects of reward functioning in ADHD and AUD symptom presentations, limiting theory development. Indeed, researchers have commented that reward functioning tends to be oversimplified in the literature due to limited attention to the multi-faceted nature of this construct (e.g. Joyner et al., 2016). In these secondary data analyses using data from a larger community-based study on emerging adult alcohol use (citation omitted for masked review), we used bi-factor modeling to separate and quantify (a) variance unique to ADHD symptoms, (b) variance unique to AUD symptoms and (c) common variance between ADHD-AUD symptoms. Importantly, this approach isolated the unique ADHD and AUD symptom constructs and any commonality shared among the two. Subsequently, we explored associations between these three latent factors (i.e. ADHD, AUD and shared dimensions) with measures of reward.
We selected reward constructs empirically or conceptually associated with AUD or ADHD symptoms in prior research (Barkley, 1997; Knouse et al., 2008; MacKillop et al., 2011; Garcia et al., 2012; Joyner et al., 2016; Acuff et al., 2019). Specifically, we employed measures of environmental suppressors, reward probability, hedonic capacity, proportionate substance-related reinforcement and delay discounting. Given that this is the first study to systematically disaggregate symptoms and possible reward correlates, hypotheses were largely exploratory. As stated, the strongest support from prior theory and research comes from the delay discounting literature. Consistent with prior work independently linking delay discounting to ADHD and AUD symptoms, we hypothesized that delay discounting would be associated with the underlying shared variance of ADHD-AUD symptoms.
METHOD
Participants
Participants were 602 emerging adults (57.3% female; Mage 22.63 years) from a mid-size city in the Southern United States: 47% identified as White/European Ancestry, 41.5% as Black/African Ancestry, 3.8% as Asian or Middle Eastern, 5% as Multiracial and 2.7% as another race. Among the participants, 36.4% reported having an associate degree or less, 25.6% were currently enrolled in a 4-year degree program, 21.3% had completed a bachelor’s degree but were not seeking post-baccalaureate education and 16.8% were enrolled in graduate school/had obtained an advanced degree.
Procedures
Participants were recruited through flyers posted in the community, social media advertisements, research pools, email screeners to university students and via in-person announcements at social events [citation omitted for masked review]. Participants were eligible if they were between the ages of 21.50 and 24.99 and reported drinking at least 3 or 4 alcoholic drinks for women or men, respectively, on at least two occasions in the past month. Participants were ineligible if they were currently in alcohol/drug use treatment, reported a psychotic disorder or were not fluent in English. Eligible participants attended a two-hour session in a university-based research laboratory where they completed measures assessing demographics, alcohol and drug use, and psychosocial risk and protective factors. Participants were compensated $40 for their time following assessment. Procedures were approved by the University’s Institutional Review Board [citation omitted for masked review].
Measures
Demographics
Participants reported on their race, sex at birth (male = 0, female =1), age and education (dichotomized for analyses: associates degree or less = 0; completed or currently seeking bachelor’s degree or higher = 1). Sex was included in all models as a covariate.
Adult ADHD self-report scale – V1.1 (ASRS)
The ASRS is an 18-item self-report questionnaire assessing current ADHD symptoms in people >18 years old (Kessler et al., 2005). The ASRS items are based on the World Health Organization Composite International Diagnostic Interview ©2001 and DSM criteria, with items modified to reflect developmentally appropriate symptom expression in adults. Items are ranked on a 5-point Likert scale to indicate the frequency of occurrence of symptoms (0 = never; 1 = rarely; 2 = sometimes; 3 = often; 4 = very often). The number of symptoms that occur ‘often’ or ‘very often’ is summed to create the number of clinically elevated symptoms ranging from 0 to 18, with higher scores indicating more severe ADHD symptoms. The ASRS has high internal consistency and concurrent validity (Adler et al., 2006).
AUD symptom checklist
The AUD checklist is an 11-item self-report checklist capturing the 11 DSM-5 AUD symptoms (World Health Organization, 2004). Participants indicate by marking ‘yes’ or ‘no’ whether they experienced each AUD symptom in the past year, with example items included under response options. Responses are summed to create a total AUD symptom count, with higher scores indicating more AUD symptoms. This checklist has demonstrated validity with emerging adult (Joyner et al., 2016) and adult clinical samples (Levitt et al., 2021).
Reward probability index (RPI)
The RPI is a 20-item self-report questionnaire measuring the probability of obtaining rewards from one’s environment (Carvalho et al., 2011). The RPI consists of two subscales: (a) reward probability, measuring the availability of rewards in one’s environment and perceived ability to attain rewards (e.g. ‘I feel a strong sense of achievement.’; ‘there are many activities that I find satisfying.’) and (b) environmental suppressors, measuring aversive/unpleasant experiences that prevent access to rewards (e.g. ‘people have been mean or aggressive toward me’; ‘changes have happened in my life that have made it hard to find enjoyment’). Items are rated using a 4-point Likert scale from 0 = ‘Strongly disagree’ to 3 = ‘Strong Agree.’ Higher scores on the RPI total score and reward probability subscale indicate more perceived experiences of reward. A higher score on the environmental suppressors subscale reflects the presence of more suppressors to reward (i.e. lower reward access). The RPI is valid and reliable (Carvalho et al., 2011). The internal consistency in present sample was good (e.g. reward probability = 0.87, environmental suppressors = 0.84).
Snaith–Hamilton pleasure scale (SHAPS)
The SHAPS is a self-administered measure of hedonic capacity (Snaith et al., 1995). Participants rate the extent to which they agree with each of 14 statements using 4 response categories (Definitely Agree, Agree, Disagree and Strongly Disagree; e.g. ‘I would enjoy reading a book, magazine or newspaper’ and ‘I would enjoy being with my family or close friend’). The total score on the SHAPS reflects a sum of the 14 items, ranging from 0 to 14. A higher total SHAPS score indicates higher levels of present state of anhedonia. The SHAPS has high internal consistency and construct validity (Nakonezny et al., 2015).
Activity level questionnaire (ALQ)
The ALQ is a self-report measure of relative reinforcement from substance use (Meshesha et al., 2020). Participants report the frequency (0—0 times to 4—More than once a day) and enjoyment (0—Unpleasant or neutral to 4—Extremely pleasant) related to 37 activities. They complete ratings for these activities under the influence of any substance(s) and sober. A cross-product reflecting substance-related (Cronbach’s alpha [α] = 0.93) and substance-free (Cronbach’s alpha [α] = 0.88) reinforcement was calculated by multiplying the frequency and enjoyment ratings for substance-related and substance-free activities. A reinforcement ratio (r-ratio) was calculated to index substance-related reinforcement relative to total reinforcement (substance-related reinforcement/[substance-free reinforcement + substance-related reinforcement]). Similar measures of reinforcement ratio have shown good reliability and validity among young adult drinkers (Acuff et al., 2019).
5-trial adjusting delay task (ED50)
In the delay task, participants were presented with a series of questions between some amount of a delayed monetary reward and half that amount, available immediately (e.g. $100 in 4 months vs. $50 now) (Koffarnus and Bickel, 2014). Monetary amounts remained stable, with delay to the larger amount adjusted to index the Effective Delay 50% (ED50) value, or the point at which the reinforcer is worth half of what it was at the immediate price. Consistent with prior research, the ED50 variable was log-transformed to correct for non-normal distribution, and participants who failed quality check were excluded from analyses.
Data analysis
Analyses were conducted using Mplus (version 8) software. Outliers (i.e. any value greater than 4 SD from the mean) were changed to one unit above the next highest non-outlying value. Given that binary scores were used as indicators for each latent factor, we used the mean and variance-adjusted weighted least squares (WLSMV) extraction for all the CFA analyses. WLSMV estimation does not assume that variables are normally distributed and is a robust estimator recommended for CFA with categorical scores (Beauducel and Herzberg, 2006). Model fit was evaluated on three indices: Comparative Fit Index (CFI; Bentler, 1990), Standardized Root Mean Square Residual (SRMR) and Root Mean Square Error of Approximation (RMSEA; Steiger, 1990). Conventional thresholds of these indices indicate that CFI of 0.90 is acceptable (Schweizer, 2010) and 0.95 (Hu and Bentler, 1998) is a good fit. SRMR values less than 0.08 are considered a good fit (Hu and Bentler, 1998). RMSEA values below 0.08 are considered acceptable (MacCallum et al., 1996) and values below 0.06 suggest good fit (Hu and Bentler, 1998).
We first evaluated a bi-factor measurement model wherein ADHD and AUD latent variables were comprised of their unique variances, with a general underlying factor capturing the shared variance between the two (Fig. 1). Covariances between the latent factors were constrained to zero, and the error variance of each latent factor was fixed to 1. All factor loadings were allowed to freely vary. To evaluate general factor strength for the bi-factor measurement model, we computed the percent uncontaminated correlations (PUC; Rodriguez et al., 2016) and the explained common variance (ECV; Reise et al., 2013). These indices provide information about parameter bias. The PUC for this model was 0.49 and the ECV for the general factor in this model was 0.60, supporting the presence of a strong general factor dimension (Reise et al., 2013).
Bi-factor model schematic of ADHD, AUD and their shared factor.
Once the factors were established, we ran our main analyses in the structural model that included associations among reward measures (i.e. r-ratio, SHAPS, RPI environmental suppressors subscale, RPI reward probability subscale and delay discounting) and the three latent factors. Reward variables were modeled as cross-sectional predictors of the factors.
RESULTS
Descriptive statistics and correlations
Descriptive statistics and correlations among key study variables are presented in Table 1. About 40.40% of participants reported two or more symptoms of AUD and 29.70% of participants endorsed five or more inattentive or hyperactive/impulsive symptoms, DSM-5 cut-offs for diagnosis.
Bivariate correlations among study variables (N = 602)
| %, M (SD) | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1. Age | 22.63 (1.03) | |||||||||||
| 2. Sex (% female) | 57.3% | −0.07 | ||||||||||
| 3. College status (% non-college) | 36.4% | 0.00 | −0.03 | |||||||||
| 4. Race (% White) | 47% | −0.01 | 0.08 | −0.05 | ||||||||
| 5. ADHD Sx | 4.67 (4.12) | −0.03 | 0.02 | 0.13** | −0.08 | |||||||
| 6. AUD Sx | 1.93 (2.39) | −0.03 | −0.04 | −0.13** | −0.02 | 0.32** | ||||||
| 7. SHAPS | 1.38 (1.92) | −0.01 | −0.01 | −0.10* | 0.04 | 0.18** | 0.20** | |||||
| 8. R-Ratio | 0.36 (0.18) | −0.01 | −0.06 | −0.29** | 0.07 | 0.06 | 0.26** | 0.10 | ||||
| 9. RPI Total | 57.52 (7.10) | 0.04 | −0.05 | 0.16** | −0.02 | −0.31** | 0.05 | −0.33** | −0.21** | |||
| 10. RPI RP | 36.40 (5.90) | 0.04 | 0.04 | 0.05 | 0.04 | −0.31** | −0.27** | −0.33** | −0.21** | −0.31** | ||
| 11. RPI ES | 21.17 (6.20) | −0.05 | 0.11** | −0.21** | 0.05 | 0.15** | 0.08* | 0.23** | 0.08* | −0.80** | 0.15** | |
| 12. ED50 | 0.05 (0.11) | 0.01 | 0.03 | −0.15 | −0.02 | 0.00 | 0.07 | 0.07 | 0.25** | −0.05 | 0.05 | 0.13** |
| %, M (SD) | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1. Age | 22.63 (1.03) | |||||||||||
| 2. Sex (% female) | 57.3% | −0.07 | ||||||||||
| 3. College status (% non-college) | 36.4% | 0.00 | −0.03 | |||||||||
| 4. Race (% White) | 47% | −0.01 | 0.08 | −0.05 | ||||||||
| 5. ADHD Sx | 4.67 (4.12) | −0.03 | 0.02 | 0.13** | −0.08 | |||||||
| 6. AUD Sx | 1.93 (2.39) | −0.03 | −0.04 | −0.13** | −0.02 | 0.32** | ||||||
| 7. SHAPS | 1.38 (1.92) | −0.01 | −0.01 | −0.10* | 0.04 | 0.18** | 0.20** | |||||
| 8. R-Ratio | 0.36 (0.18) | −0.01 | −0.06 | −0.29** | 0.07 | 0.06 | 0.26** | 0.10 | ||||
| 9. RPI Total | 57.52 (7.10) | 0.04 | −0.05 | 0.16** | −0.02 | −0.31** | 0.05 | −0.33** | −0.21** | |||
| 10. RPI RP | 36.40 (5.90) | 0.04 | 0.04 | 0.05 | 0.04 | −0.31** | −0.27** | −0.33** | −0.21** | −0.31** | ||
| 11. RPI ES | 21.17 (6.20) | −0.05 | 0.11** | −0.21** | 0.05 | 0.15** | 0.08* | 0.23** | 0.08* | −0.80** | 0.15** | |
| 12. ED50 | 0.05 (0.11) | 0.01 | 0.03 | −0.15 | −0.02 | 0.00 | 0.07 | 0.07 | 0.25** | −0.05 | 0.05 | 0.13** |
Notes: Race = Non-White (0) and White (1); ADHD Sx = Attention-Deficit/Hyperactivity Disorder Symptoms; AUD Sx = Alcohol Use Disorder Symptoms; SHAPS = Snaith–Hamilton Pleasure Scale; R-ratio = Proportionate Substance-Related Reinforcement; RPI Total = Reward Probability Index Total Score; RPI RP = Reward Probability Index Reward Probability Subscale; RPI ES = Reward Probability Index Environmental Suppressors Subscale; ED50 = Effective Delay—50.
*Correlation is significant at the 0.05 level (two-tailed).
**Correlation is significant at the 0.01 level (two-tailed).
Bivariate correlations among study variables (N = 602)
| %, M (SD) | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1. Age | 22.63 (1.03) | |||||||||||
| 2. Sex (% female) | 57.3% | −0.07 | ||||||||||
| 3. College status (% non-college) | 36.4% | 0.00 | −0.03 | |||||||||
| 4. Race (% White) | 47% | −0.01 | 0.08 | −0.05 | ||||||||
| 5. ADHD Sx | 4.67 (4.12) | −0.03 | 0.02 | 0.13** | −0.08 | |||||||
| 6. AUD Sx | 1.93 (2.39) | −0.03 | −0.04 | −0.13** | −0.02 | 0.32** | ||||||
| 7. SHAPS | 1.38 (1.92) | −0.01 | −0.01 | −0.10* | 0.04 | 0.18** | 0.20** | |||||
| 8. R-Ratio | 0.36 (0.18) | −0.01 | −0.06 | −0.29** | 0.07 | 0.06 | 0.26** | 0.10 | ||||
| 9. RPI Total | 57.52 (7.10) | 0.04 | −0.05 | 0.16** | −0.02 | −0.31** | 0.05 | −0.33** | −0.21** | |||
| 10. RPI RP | 36.40 (5.90) | 0.04 | 0.04 | 0.05 | 0.04 | −0.31** | −0.27** | −0.33** | −0.21** | −0.31** | ||
| 11. RPI ES | 21.17 (6.20) | −0.05 | 0.11** | −0.21** | 0.05 | 0.15** | 0.08* | 0.23** | 0.08* | −0.80** | 0.15** | |
| 12. ED50 | 0.05 (0.11) | 0.01 | 0.03 | −0.15 | −0.02 | 0.00 | 0.07 | 0.07 | 0.25** | −0.05 | 0.05 | 0.13** |
| %, M (SD) | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1. Age | 22.63 (1.03) | |||||||||||
| 2. Sex (% female) | 57.3% | −0.07 | ||||||||||
| 3. College status (% non-college) | 36.4% | 0.00 | −0.03 | |||||||||
| 4. Race (% White) | 47% | −0.01 | 0.08 | −0.05 | ||||||||
| 5. ADHD Sx | 4.67 (4.12) | −0.03 | 0.02 | 0.13** | −0.08 | |||||||
| 6. AUD Sx | 1.93 (2.39) | −0.03 | −0.04 | −0.13** | −0.02 | 0.32** | ||||||
| 7. SHAPS | 1.38 (1.92) | −0.01 | −0.01 | −0.10* | 0.04 | 0.18** | 0.20** | |||||
| 8. R-Ratio | 0.36 (0.18) | −0.01 | −0.06 | −0.29** | 0.07 | 0.06 | 0.26** | 0.10 | ||||
| 9. RPI Total | 57.52 (7.10) | 0.04 | −0.05 | 0.16** | −0.02 | −0.31** | 0.05 | −0.33** | −0.21** | |||
| 10. RPI RP | 36.40 (5.90) | 0.04 | 0.04 | 0.05 | 0.04 | −0.31** | −0.27** | −0.33** | −0.21** | −0.31** | ||
| 11. RPI ES | 21.17 (6.20) | −0.05 | 0.11** | −0.21** | 0.05 | 0.15** | 0.08* | 0.23** | 0.08* | −0.80** | 0.15** | |
| 12. ED50 | 0.05 (0.11) | 0.01 | 0.03 | −0.15 | −0.02 | 0.00 | 0.07 | 0.07 | 0.25** | −0.05 | 0.05 | 0.13** |
Notes: Race = Non-White (0) and White (1); ADHD Sx = Attention-Deficit/Hyperactivity Disorder Symptoms; AUD Sx = Alcohol Use Disorder Symptoms; SHAPS = Snaith–Hamilton Pleasure Scale; R-ratio = Proportionate Substance-Related Reinforcement; RPI Total = Reward Probability Index Total Score; RPI RP = Reward Probability Index Reward Probability Subscale; RPI ES = Reward Probability Index Environmental Suppressors Subscale; ED50 = Effective Delay—50.
*Correlation is significant at the 0.05 level (two-tailed).
**Correlation is significant at the 0.01 level (two-tailed).
Factor loadings for the bi-factor measurement model
Table 2 shows model fit indices for the bi-factor model, and Supplemental Table 1 presents the standard factor loadings of the ADHD, AUD and general factors. Factor loadings in the ADHD model are consistent with prior work (Goh et al., 2020). The general factor contained both variances from the ADHD and AUD factors. Taken together with the model reliability indices, these data support a strong general factor encompassing shared variance among ADHD and AUD symptoms.
Model fit indices
| χ2 (df), P-value | χ2 Diff. test | RMSEA (90% CI) | CFI | SRMR | |
|---|---|---|---|---|---|
| Bi-factor model | 625.67 (348), <0.001 | 0.036 (0.032, 0.041) | 0.951 | 0.077 | |
| Predictor model | 815.88 (504), <0.001 | 0.033 (0.029, 0.037) | 0.932 | 0.088 |
| χ2 (df), P-value | χ2 Diff. test | RMSEA (90% CI) | CFI | SRMR | |
|---|---|---|---|---|---|
| Bi-factor model | 625.67 (348), <0.001 | 0.036 (0.032, 0.041) | 0.951 | 0.077 | |
| Predictor model | 815.88 (504), <0.001 | 0.033 (0.029, 0.037) | 0.932 | 0.088 |
Notes: Df = degrees of freedom; RMSEA = Root Mean Square Error of Approximation; CFI = Comparative Fit Index; SRMR = Standardized Root Mean Squared Residual.
Model fit indices
| χ2 (df), P-value | χ2 Diff. test | RMSEA (90% CI) | CFI | SRMR | |
|---|---|---|---|---|---|
| Bi-factor model | 625.67 (348), <0.001 | 0.036 (0.032, 0.041) | 0.951 | 0.077 | |
| Predictor model | 815.88 (504), <0.001 | 0.033 (0.029, 0.037) | 0.932 | 0.088 |
| χ2 (df), P-value | χ2 Diff. test | RMSEA (90% CI) | CFI | SRMR | |
|---|---|---|---|---|---|
| Bi-factor model | 625.67 (348), <0.001 | 0.036 (0.032, 0.041) | 0.951 | 0.077 | |
| Predictor model | 815.88 (504), <0.001 | 0.033 (0.029, 0.037) | 0.932 | 0.088 |
Notes: Df = degrees of freedom; RMSEA = Root Mean Square Error of Approximation; CFI = Comparative Fit Index; SRMR = Standardized Root Mean Squared Residual.
ADHD, AUD and shared factors in the structural model
Table 3 presents a summary of results from main structural model and Fig. 2 illustrates associations. Environmental suppressors were positively associated with AUD and the shared factor, but not with ADHD. Reward probability was negatively associated with ADHD, but not with the shared factor or with AUD. Hedonic capacity was not significantly associated with ADHD, AUD or the shared factors. R-ratio was positively associated with AUD only, but not with the ADHD or shared factors. Delay discounting was not associated with ADHD, AUD or the shared factors. No significant associations among sex and these constructs were evident.
Summary of results from analytic models
| Omnibus model | ||
|---|---|---|
| b (SE) | B | |
| Effects on ADHD symptom factor | ||
| Environmental suppressors | 0.01 (0.02) | 0.05 |
| Reward probability | −0.06** (0.01) | −0.31 |
| Hedonic capacity | −0.03 (0.04) | −0.05 |
| R-ratio | −0.67 (0.36) | −1.11 |
| Delay discounting | 0.01 (0.07) | 0.01 |
| Sex | −0.12 (0.12) | −0.06 |
| Effects on AUD symptom factor | ||
| Environmental suppressors | 0.05** (0.01) | 0.27 |
| Reward probability | 0.01 (0.01) | 0.03 |
| Hedonic capacity | 0.04 (0.03) | 0.07 |
| R-ratio | 1.47** (0.37) | 0.23 |
| Delay discounting | 0.03 (0.06) | 0.03 |
| Sex | −0.21 (0.11) | −0.09 |
| Effects on shared symptom factor | ||
| Environmental suppressors | 0.06** (0.01) | 0.35 |
| Reward probability | 0.01 (0.01) | 0.04 |
| Hedonic capacity | 0.05 (0.03) | 0.08 |
| R-ratio | 0.12 (0.33) | 0.02 |
| Delay discounting | −0.08 (0.04) | −0.08 |
| Sex | 0.01 (0.10) | 0.004 |
| Omnibus model | ||
|---|---|---|
| b (SE) | B | |
| Effects on ADHD symptom factor | ||
| Environmental suppressors | 0.01 (0.02) | 0.05 |
| Reward probability | −0.06** (0.01) | −0.31 |
| Hedonic capacity | −0.03 (0.04) | −0.05 |
| R-ratio | −0.67 (0.36) | −1.11 |
| Delay discounting | 0.01 (0.07) | 0.01 |
| Sex | −0.12 (0.12) | −0.06 |
| Effects on AUD symptom factor | ||
| Environmental suppressors | 0.05** (0.01) | 0.27 |
| Reward probability | 0.01 (0.01) | 0.03 |
| Hedonic capacity | 0.04 (0.03) | 0.07 |
| R-ratio | 1.47** (0.37) | 0.23 |
| Delay discounting | 0.03 (0.06) | 0.03 |
| Sex | −0.21 (0.11) | −0.09 |
| Effects on shared symptom factor | ||
| Environmental suppressors | 0.06** (0.01) | 0.35 |
| Reward probability | 0.01 (0.01) | 0.04 |
| Hedonic capacity | 0.05 (0.03) | 0.08 |
| R-ratio | 0.12 (0.33) | 0.02 |
| Delay discounting | −0.08 (0.04) | −0.08 |
| Sex | 0.01 (0.10) | 0.004 |
**P < 0.01.
Summary of results from analytic models
| Omnibus model | ||
|---|---|---|
| b (SE) | B | |
| Effects on ADHD symptom factor | ||
| Environmental suppressors | 0.01 (0.02) | 0.05 |
| Reward probability | −0.06** (0.01) | −0.31 |
| Hedonic capacity | −0.03 (0.04) | −0.05 |
| R-ratio | −0.67 (0.36) | −1.11 |
| Delay discounting | 0.01 (0.07) | 0.01 |
| Sex | −0.12 (0.12) | −0.06 |
| Effects on AUD symptom factor | ||
| Environmental suppressors | 0.05** (0.01) | 0.27 |
| Reward probability | 0.01 (0.01) | 0.03 |
| Hedonic capacity | 0.04 (0.03) | 0.07 |
| R-ratio | 1.47** (0.37) | 0.23 |
| Delay discounting | 0.03 (0.06) | 0.03 |
| Sex | −0.21 (0.11) | −0.09 |
| Effects on shared symptom factor | ||
| Environmental suppressors | 0.06** (0.01) | 0.35 |
| Reward probability | 0.01 (0.01) | 0.04 |
| Hedonic capacity | 0.05 (0.03) | 0.08 |
| R-ratio | 0.12 (0.33) | 0.02 |
| Delay discounting | −0.08 (0.04) | −0.08 |
| Sex | 0.01 (0.10) | 0.004 |
| Omnibus model | ||
|---|---|---|
| b (SE) | B | |
| Effects on ADHD symptom factor | ||
| Environmental suppressors | 0.01 (0.02) | 0.05 |
| Reward probability | −0.06** (0.01) | −0.31 |
| Hedonic capacity | −0.03 (0.04) | −0.05 |
| R-ratio | −0.67 (0.36) | −1.11 |
| Delay discounting | 0.01 (0.07) | 0.01 |
| Sex | −0.12 (0.12) | −0.06 |
| Effects on AUD symptom factor | ||
| Environmental suppressors | 0.05** (0.01) | 0.27 |
| Reward probability | 0.01 (0.01) | 0.03 |
| Hedonic capacity | 0.04 (0.03) | 0.07 |
| R-ratio | 1.47** (0.37) | 0.23 |
| Delay discounting | 0.03 (0.06) | 0.03 |
| Sex | −0.21 (0.11) | −0.09 |
| Effects on shared symptom factor | ||
| Environmental suppressors | 0.06** (0.01) | 0.35 |
| Reward probability | 0.01 (0.01) | 0.04 |
| Hedonic capacity | 0.05 (0.03) | 0.08 |
| R-ratio | 0.12 (0.33) | 0.02 |
| Delay discounting | −0.08 (0.04) | −0.08 |
| Sex | 0.01 (0.10) | 0.004 |
**P < 0.01.
Correlates of ADHD, AUD and the shared factor. Note. Bolded lines denote statistically significant associations, P < 0.05. RPI = Reward Probability Index; SHAPS = Snaith–Hamilton Anhedonia Pleasure Scale; ED50 = Effective Delay—50; R-ratio = Proportionate Substance-Related Reinforcement.
DISCUSSION
To date, it is unclear which domains of reward functioning are unique to ADHD versus AUD symptoms, and which represent underlying shared correlates, thus limiting theory development. Using a large cross-sectional community-based sample of young adults, this is the first study to identify both unique and shared reward correlates of ADHD-AUD within a single model. In so doing, this work meaningfully advances theoretical conceptualizations of this commonly co-occurring presentation.
The bi-factor model revealed a strong general factor, containing both variances from the ADHD and AUD symptom factors. This novel finding suggests that there is a shared, common factor among ADHD and AUD symptoms. Indeed, this is the first known study to use transdiagnostic modeling to parse the continuous symptom dimensions of ADHD and AUD and isolate a shared dimension among these multiple co-occurring symptoms—an essential step in developing theoretical models of comorbidity that can guide assessment and treatment efforts.
Both the AUD and the shared ADHD-AUD shared symptom dimensions were significantly associated with the presence of environmental suppressors. Young adult drinking often enhances the availability of some rewards (e.g. social activity) and can inadvertently reinforce escape from aversive/unpleasant experiences in the short term. Environmental suppression of reward could also make efforts to obtain natural rewards burdensome and effortful, and ADHD is associated with difficulties recruiting effort for goal attainment (Winter et al., 2019). Perhaps those with ADHD are vulnerable to poorly regulated drinking in the context of easy access to alcohol, which is characteristic of many young adult social circles. This possibility expands upon Joyner et al. (2016), who showed that college students with more environmental suppressors also experienced more alcohol-related problems, controlling for depression levels and alcohol consumption. This finding supports emerging intervention approaches for hazardous alcohol use in young adults with ADHD, which pair behavioral activation strategies and organizational skills to help youth with ADHD access rewarding and value-driven activities that may be limited by aversive conditions (Meinzer et al., 2021).
Only the ADHD symptom domain correlated with low reward probability. Importantly, reward probability refers to two complementary phenomena: the presence of potentially reinforcing events that are rewarding and a person’s perceived ability to deploy adaptive behaviors to experience rewards (Hill et al., 2017). Perhaps the symptom constellation of ADHD results in fewer experiences of reward over time. Individuals with ADHD demonstrate more difficulties in task persistence, effort regulation and social skills compared to non-ADHD controls (Barkley, 1997). Thus, people with ADHD might experience limited interest in potentially reinforcing activities that require effort and also display limited confidence in deploying the strategies necessary to achieve effortful rewards. This possibility aligns with cognitive–behavioral models of ADHD, which illustrate a cascading effect of ADHD symptoms on failure experiences, quality of life impairments and comorbidity (Safren et al., 2005).
Proportionate substance-related reinforcement was positively associated with the AUD symptom domain only, but not with the ADHD or ADHD-AUD shared factors. This variable has shown robust associations with alcohol problems even in models that control demographic and other alcohol-related risk factors such as delay discounting and alcohol demand (Acuff et al., 2019). There is also evidence that brief alcohol interventions that include a focus on increasing substance-free activity participation are associated with reductions in alcohol use and related problems and that change in these variables is mediated by change in proportionate substance-related reinforcement (Murphy et al., 2019). The current study expands upon this work in suggesting that this variable is uniquely related to AUD symptoms and is not associated with either ADHD symptoms in isolation or a shared ADHD-AUD dimension. In connection with our findings on reward probability and environmental suppressors, perhaps limited access to rewarding experiences and more perceived suppressors of environmental rewards is a feature of both AUD and ADHD, whereas a relative predominance of substance-related activity participation and enjoyment is a unique feature of AUD. This is consistent with the idea that severe AUD involves a narrowing of the behavioral repertoire where alcohol and other drug use gradually become the predominant source of reward.
Contrary to our hypothesis, delay discounting was not associated with the AUD, ADHD or shared symptom dimensions. This finding is inconsistent with prior research and exposes several possible explanations, each warranting additional empirical attention. It is possible that delay aversion was undetected in the brief task wherein individuals were not given monetary reward for their choices. Effect sizes are larger in lengthier measures of this construct (MacKillop et al., 2011). Additionally, value allocation is influenced by contextual variables (Ashe and Wilson, 2020). People might make different decisions while failing to ‘maximize utility’. For example, young adults in the middle of an academic semester may anticipate an end-of-the-year party and plan to consume a large amount of alcohol at that point; thus, they may select delayed receipt (i.e. money for alcohol, later). Also, immediate choices may be more likely made in conditions of deprivation and uncertainty; future research on additional variables (e.g. income and chronic stress) is warranted. Finally, delay discounting may be less relevant in lower severity/young adult samples (MacKillop et al., 2011).
Hedonic capacity, or the trait predisposition underlying an individual’s baseline range of ability to feel pleasure, was not significantly associated with ADHD, AUD or the shared dimension. It has been theorized that those with a low hedonic capacity require higher ‘doses’ of positive experiences to feel euthymic, with some speculating that this is accomplished by pursuing external stimulation (i.e. through substance abuse or risky behaviors) to maximize pleasure (Sternat and Katzman, 2016). Our preliminary results do not support this possibility in the context of ADHD and AUD.
LIMITATIONS AND FUTURE DIRECTIONS
This study is strengthened by a large and racially diverse sample with a large portion of female participants, who were both college and non-college attending. We used a robust statistical approach to tease apart the variance attributable to AUD, ADHD and the shared symptom dimension. Despite these strengths, findings should be interpreted in the context of limitations. Analyses were cross-sectional and thus cannot support causal models. Future studies should include longitudinal analyses to examine etiological risks. This study also employed a community-based sample with continuous measures of symptomatology, so findings cannot necessarily be generalized to treatment seeking and formally diagnosed samples. Future research should also incorporate a multimodal assessment to minimize reporting biases. Additionally, ADHD and AUD are comorbid with conduct/antisocial personality disorders; future research on these comorbidities is necessary to clarify effects. Finally, linguistic and cultural constructs may influence reward functioning, and these variables should be considered in future work.
CONCLUSION
The current study identified shared and unique reward-related correlates of ADHD and AUD symptoms, which, to date, have never been systematically examined using transdiagnostic modeling. These novel findings advance existing theory on these two commonly co-occurring disorders and meaningfully expose future directions for research on transdiagnostic correlates.
ACKNOWLEDGEMENTS
The data underlying this article may be shared upon reasonable request to the corresponding author.
Funding
This project was supported by grants from National Institute on Alcohol Abuse and Alcoholism (R01 AA024930, James MacKillop & James G. Murphy; F31 AA027937, Lauren Oddo; F31 AA027140, Samuel Acuff).
Conflicts of interest statement
The authors declare that they have no conflicts of interest.
