Beyond Consequences: Longitudinal Evidence That Gambling Harms Reinforce Dependence

Background and aims: Gambling harms are traditionally conceptualized as consequences of gambling dependence, yet the reverse pathway remains empirically untested. This study examined the bidirectional longitudinal relationships between gambling harms and gambling dependence within individuals over time.Design and setting: A prospective four-wave longitudinal survey study conducted in the United Kingdom, with data collected via the Prolific online research platform. The study was pre-registered on OSF (https://osf.io/n6qrp) prior to data collection.Participants: 956 UK residents scoring 3 or above on the Problem Gambling Severity Index (PGSI) at baseline (42.6% female; mean age = 35.53 years). Participants completed self-report questionnaires at four time-points, with the final wave (T4) occurring six months post-baseline.Measurements: The PGSI was used at baseline to screen for moderate-risk or problem gambling. Gambling harm and dependence were measured at each wave using the Problem and Pathological Gambling Measure (PPGM), adapted to a four-point Likert scale. A Random Intercept Cross-Lagged Panel Model (RI-CLPM) estimated within-individual cross-lagged effects across the four waves. Two sensitivity analyses were conducted: one constraining cross-lagged paths to equality across waves to test temporal invariance, and one incorporating lag-2 paths to assess delayed effects not captured in the primary model. Findings: In the primary analysis, gambling dependence at T3 significantly predicted higher gambling harms at T4 (β = 0.185, 95% CI = 0.063–0.306, p = 0.003). When cross-lagged paths were constrained to equality across waves, past dependence consistently predicted subsequent harm across all waves (β = 0.100, 95% CI = 0.007–0.193, p = 0.036). In the lag-2 sensitivity analysis, prior gambling dependence predicted harm at the subsequent wave from Wave 2 onward (β = 0.200–0.248, p = 0.002–0.006). Additionally, past gambling harm consistently predicted higher gambling dependence two waves later (β = 0.187–0.209, p = 0.015–0.018). Conclusions: There is a reciprocally reinforcing relationship between gambling harms and symptoms of dependence. This study provides the first within-person longitudinal evidence that gambling harms play an active role in the development and escalation of dependence. These findings challenge conventional models that treat harm as solely downstream of dependence and support the integration of harm reduction as a core component of prevention and early intervention for gambling problems.