Abstract
IMPORTANCE: The opioid crisis remains a global public health issue, with poor clinical outcomes, low treatment adherence, and early dropouts being major challenges. These may be driven by a pathological increase in avoidance tendencies in opioid addiction, although this has not yet been demonstrated in humans. OBJECTIVE: To determine whether opioid addiction is characterized by increased avoidance learning and whether this is also observed in individuals with chronic nonaddicted opioid use. DESIGN, SETTING, AND PARTICIPANTS: This case-control study conducted in Hamburg, Germany, used a probabilistic reinforcement learning task to investigate how individuals with chronic opioid use (both with and without opioid addiction) learn from negative outcomes. Participants were recruited from outpatient clinics for either addiction or chronic pain between December 2021 and October 2024. Data analysis was conducted between January 2025 and February 2025. Computational modeling was applied to assess whether participants learned more effectively from experiencing or avoiding financial loss and to ascertain their response tendencies. MAIN OUTCOMES AND MEASURES: The primary outcomes were computational parameters of learning from experiencing financial loss, learning from avoiding it (ie, negative reinforcement), choice consistency, and a response tendency to repeat. General and drug-related compulsivity, automatic habits, and avoidance tendencies were assessed by self-report. RESULTS: This study's sample included 88 participants (47 female participants [53%]; mean [SD] age, 44 [10.9] years). Both participants with opioid addiction and those using opioids chronically showed increased learning from negative reinforcement (F2,83 = 15.8; P < .001; mean difference [MD] = 0.071; 95% CI, 0.02-0.12) and reduced choice consistency (F2,83 = 10.4; P < .001; MD = -1.31; 95% CI, -2.03 to -0.59). Variations in avoidance learning predicted both their self-reported avoidance behavior in daily life and the severity of compulsive opioid use. The 2 opioid user groups did not differ in terms of learning from monetary loss (F2,83 = 2.91; P = .06) or their tendency to repeat prior responses (F2,83 = 1.84; P = .17). CONCLUSIONS AND RELEVANCE: This cross-sectional study supports preclinical findings emphasizing the role of negative reinforcement in opioid addiction, although this has not been shown in humans before. The results indicate that negative reinforcement is relevant even in the early stages of opioid use, suggesting potential new avenues for prevention and treatment.