Individuals with OUD (n=32) were recruited via flyers, referrals from a methadone clinic, and word of mouth. Inclusion criteria included: individuals who were on a stable dose of medication for OUD, aged 18-65, and able to read at an eighth-grade level. Exclusion criteria included: severe mental health issues (e.g., schizophrenia, current manic state, etc.), too impaired to complete the tasks, and pregnancy.

Neurocognitive data from non-substance-use controls (NSC, n=15) were obtained via a repository of de-identified data from pilot studies funded by the Penn Mental Health Aids Research Center. These individuals reported no alcohol or drug use on the Risk Assessment Battery (28). However, they did not complete several of the questionnaires administered to the OUD group (see below). The NSC group was matched to the OUD group on key demographic variables including age and sex to ensure comparability. All participants provided written informed consent, and the study protocol was approved by the University of Pennsylvania Institutional Review Board.

Participants completed several questionnaires to capture data on demographics: age, sex, race, income, and education. Both groups also completed the Quick Inventory Depression Scale (QIDS) (29), see Table 1 . Individuals with OUD additionally completed questionnaires on drug use, including opioids, stimulants, alcohol, and cannabis; social functioning via the PROMIS-Ability to Participate in Social Roles and Activities (PROM-SOC) (30); impulsiveness via the Barrat’s Impulsiveness Scale (BIS) (31); and prior adversity via the Extended Adverse Childhood Experiences (ACES) questionnaire (32). All questionnaires used were validated instruments with established psychometric properties.

Both groups completed the Penn Computerized Neurocognitive Battery (CNB) (33). This battery included tasks to measure episodic memory via the Face (FMT) and Word (WMT) Memory Tasks; working memory via the letter n-Back task (NBT); cognitive flexibility via the Conditional Exclusion Task (CET); complex cognition via the Spatial Line Orientation Task (LOT); and emotion recognition via the Emotion Recognition Task (ERT). The CNB also included tasks to evaluate proficiency in using a computer mouse as well as a motor task (i.e., quickly pressing the space bar) as control factors. Details and factor testing of the tasks have been reported elsewhere (34). All tasks were administered in a standardized order, and participants were given breaks as needed to minimize fatigue.

The primary performance measure for FMT, WMT, ERT, and LOT was the number of correct responses. The primary measure for NBT was the F1 score, calculated using the formula, (2*TP)/(2*TP+FP+FN), where TP = true positives, FP = false positives, and FN = false negatives. For the CET, the primary measure was an accuracy score, calculated by (Number of categories achieved + 1) * CR/(CR+ER), where CR = correct responses and ER = errors.

Potential differences between groups were assessed via independent t-tests (for continuous variables) and χ² (for nominal variables), with a threshold of p < 0.05, FDR corrected. Effect sizes (Hedge’s g) are reported for significant differences. Pearson correlations were used to examine relationships between questionnaire scores and cognitive performance within the OUD group. All analyses were performed using SPSS (v28, 2021).

There were forty-seven total individuals included in the analysis, with 15 in the NSC group and 32 in the OUD group. Average age was 46 (OUD: 45.7 (±1.6); NSC: 46.7 (±2.5); t(45) = .36, p = 0.42). There were 72.3% of individuals who were female (OUD: 78.1%; NSC: 60%; χ² = 1.68, p = 0.30) and 44.7% who were African American (OUD: 50%; NSC: 33.3%; χ² = 1.15, p = 0.36), and no significant differences between the two groups. The OUD group had significantly higher QIDS scores compared to the NSC group (OUD: 16.5 (±1.3); NSC: 4.1 (±0.8); t(44) = 5.95, p < 0.001, g = 1.9).

For the OUD group, additional information was collected. Regarding type of medication participants were taking for OUD, twenty-two (73%) reported methadone, 7 (23%) reported Suboxone, and 1 (3%) reported Naltrexone. Approximately 44% had some college or more (vs. completing high school or less); 40% were experiencing current houselessness; and 84% had at least one adverse childhood experience. Regarding drug use, the OUD group had used opioids for an average of 14 years, with approximately 47% reporting fentanyl use. Approximately 44% reported alcohol use; 38% reported cannabis use; 47% reported stimulant use; and 56% reported use of more than 1 drug.

A subgroup of NSCs completed questionnaires about education (n=13) and income (n=8), both of which differed from the OUD group. There were only 1/13 of NSCs who had some college or more (vs. completing high school or less) compared to 14/32 of the OUD group who had some college or more (χ² = 5.72, p = 0.02). In contrast, 6/8 of NSCs made more than 25k (vs. less than 25k) compared to only 4/32 of OUD who made more than 25k (χ² = 13.33, p = 0.001). Too few numbers in the NSC group prevented further investigation of the effect of income on CNB performance.

Results, FDR corrected, show that the OUD group had significantly less accuracy on the CET (t(45) = 5.26, p < 0.001) and lower F1 scores on the NBT (t(43) = 3.13, p = 0.003, g = 1.6). Uncorrected results showed fewer correct responses on the FMT (t(45) = 2.32, p = 0.03, g = .72). There was a trend towards fewer correct responses on the WMT (t(45) = 1.86, p = 0.07). See Figure 1 and Table 2 for full statistics and results. Correcting for differences in QIDS did not change results for the CET (p = 0.001) or FMT (p = 0.043) but made NBT results insignificant (p = 0.19).

The groups did not differ on reaction times. Uncorrected results showed a trend towards differing reaction time on the WMT (t(45) = 1.8, p = 0.08). See Table 3 for full statistics and results on reaction times.

There were several significant correlations between measures on mental health symptoms and behavioral variables. The QIDS and BIS were positively correlated (r = .51, p = 0.003); the QIDS and PROM-SOC were inversely correlated (r = -.51, p = 0.003); and the BIS and PROM-SOC were inversely correlated (r = -.74, p < 0.001).

Performance on the CET was inversely correlated with BIS scores (r = -.47, p = 0.006), in that lower performance was associated with more impulsiveness ( Figure 2A ). Correlations were conducted to explore the relationship of CET performance with BIS subscales ( Supplementary Table S2 ). Performance on CET was positively correlated with PROM-SOC scores (r = .46, p = 0.007) ( Figure 2B ), in that better performance was associated with higher social functioning. There were no significant correlations between mental health or behavioral variables with NBT or FMT scores.

Prior adversity and QIDs did not associate with CET ( Supplementary Table S1 ). Other factors such as polydrug use, education, age, and sex also did not associate with CET performance ( Supplementary Table S1 ).

Emerging research suggests cognitive difficulties might be improved by various means (42). For example, one study found that cognitive deficits improved with repeated testing and with increased abstinence from opioids (43). Some data suggests cognitive training improves performance on neurocognitive tasks, though it is unclear how well these improvements translate into real-world situations (44).

Executive function is predominantly regulated by cortical, particularly prefrontal cortical (PFC), regions (45), leading to downstream inhibition of subcortical regions (46); thus, bolstering PFC regions directly might improve neurocognition generally (47). Indeed, some methods, such as transcranial magnetic stimulation of lateral PFC regions have been successful in improving executive function (48, 49). Some studies have reported on the importance of the left (vs. right) dorsolateral PFC in reducing drug-seeking behaviors (50), suggesting the inhibitory impact of that area.

Pharmacological effects may also be helpful in improving both neurocognition and impulsivity, especially considering that cognition may not improve while receiving a medication for OUD (51). Psychedelics as therapeutic agents have had a resurgence in the past two decades, and studies show promise of psychedelic-assisted treatment in treating mental health issues, such as depression (52), trauma (53), and substance-use disorder (54, 55). In addition, psychedelics studies show that neurocognition, particularly cognitive flexibility, is improved post psychedelic-assisted treatment (56), which is thought to be due to changes occurring at the neural level (e.g., neuroplasticity) (57, 58).

Studies show the importance of cognition in clinical outcomes. For example, impairments in cognition have been tied to treatment non-adherence, often in older adults (59–61); impairment that are typically accelerated in those with OUD (62). and difficulties with decision-making, cognitive flexibility, and inhibition have been associated with increased drug-seeking behaviors and relapse (63–65). Thus, treatments to improve neurocognition could translate to better clinical outcomes.

The clinical significance of these findings cannot be overstated. Healthcare providers working with individuals with OUD should be aware of the potential cognitive deficits, particularly in cognitive flexibility and working memory. Routine cognitive assessments could help tailor treatment plans to individual needs. Moreover, the strong associations between impulsivity, social function, and cognitive flexibility suggest that a multifaceted treatment approach addressing these interconnected factors may be most effective in improving outcomes for individuals with OUD.

This was a pilot study with a relatively small clinical sample size, with primarily a female population and high trauma exposure. In addition, the comparator group was obtained via a repository of data from different pilot studies. Therefore, data collection methods may have varied, and several questionnaires were not collected in the NSC group. A follow-up study is in progress that aims to triple the OUD sample size as well as collect the same data from matched NSCs. In addition, demographic differences between the OUD and NSC groups, particularly in income levels, may have influenced cognitive performance. Future studies should aim to control for these variables to isolate the specific effects of OUD on cognitive function. Finally, the limited dataset also limited the statistical approaches that could be applied.

This pilot study suggests that working memory and cognitive flexibility are impaired in people with OUD and that impulsiveness and social function are key factors in cognitive flexibility impairments in OUD. The results may offer insights for larger-scale investigations and potential interventions to reduce relapse risk. Future research should focus on developing and testing interventions targeting cognitive flexibility and impulsiveness in OUD, as well as exploring the potential role of social function in treatment outcomes. By addressing these cognitive and psychosocial factors in tandem with traditional OUD treatments, we may be able to significantly improve outcomes for individuals struggling with opioid addiction.