The Substitution Mechanism: Between-Person Patterns, Not Daily Fluctuations
Last Updated: February 2026 | Reading Time: 4 minutes
When researchers find evidence of cannabis substitution for alcohol, the natural question is: how does it work? Does consuming more cannabis on a given day directly reduce alcohol consumption that same day? Or does the effect operate through a different mechanism, one that is less immediate but more durable? The 2026 Ottawa CMAPS study by Goulet-Stock et al. provides a methodologically rigorous answer, with significant implications for how future cannabis substitution programs are designed and evaluated.
The Ottawa Findings: Two Levels of Analysis
The Goulet-Stock, Hacksel, Scandiuzzi, Boyd, Pauly & Stockwell (2026) study, published in the International Journal of Drug Policy (147: 105083), used hierarchical mixed-effects models to examine two distinct sources of cannabis-alcohol association in a managed alcohol program (MAP) setting: between-person effects (do people who use more cannabis on average also drink less overall?) and within-person effects (when the same person uses more cannabis on a given day, do they drink less that same day?).
The between-person effect was statistically significant and practically meaningful: participants who used more cannabis on average consumed significantly less alcohol overall. The specific estimate was 2.43 fewer mean daily standard drinks per additional 0.4-gram joint, containing approximately 15.2 standard THC units or 76 mg THC.
The within-person effect was not significant. Short-term, day-to-day fluctuations in cannabis consumption did not predict concurrent changes in alcohol use in the same individual. A participant who consumed more cannabis on Tuesday was not meaningfully more likely to drink less on Tuesday.
What This Means for the Substitution Mechanism
The finding that the cannabis substitution effect operates between persons over stable patterns, rather than within persons across days, has a specific implication for the mechanism. It rules out a simple pharmacological displacement model: cannabis does not acutely reduce alcohol craving or consumption in real time. Instead, the pattern that produces the harm reduction effect is a stable individual relationship established over time.
This is consistent with what Vancouver MAP participants described qualitatively in the CMAPS cannabis substitution research: cannabis as a tool for gradual moderation, not immediate cessation. The qualitative findings from the Vancouver program (Bailey et al., in peer review at Drug & Alcohol Review) document participants describing cannabis as something that allowed them to slow their alcohol use over time, rather than something that mechanically replaced a drink each time it was used. The Ottawa quantitative finding and the Vancouver qualitative finding are telling the same story from different angles.
Implications for Program Evaluation
The between-person mechanism has direct implications for how cannabis substitution programs should be evaluated. Measuring daily or weekly cannabis and alcohol use and looking for same-day trade-offs will not capture the substitution effect, because that is not where the effect operates. Evaluations that look only at short-term fluctuations risk concluding that substitution is not occurring, when in fact the program may be producing stable between-person differences that only become visible over longer time windows.
The Ottawa study used two years of longitudinal data and a mixed-effects modelling approach that explicitly separated within-person from between-person effects. Programs designing cannabis substitution evaluations for alcohol use disorder populations should plan for equivalent time horizons and analytic approaches. Short-term pilots assessed against acute measures will systematically underestimate program impact.
Implications for Dosing and Supply Design
If the mechanism is stable, individual patterns rather than acute displacement, program design implications follow. Consistent, predictable supply matters more than the availability of higher doses on specific high-risk days. The consistency of the cannabis supply, rather than any single day’s dose, is the structural condition under which the between-person substitution pattern developed in the Ottawa MAP cohort.
This has practical implications for how institutions and harm reduction programs structure cannabis supply logistics. Irregular or unpredictable supply, even at equivalent average quantities, may not reproduce the stable individual pattern that the evidence suggests is the operative mechanism.
The Time-Independent Alcohol Decline
A second finding from the Ottawa study deserves attention for researchers designing similar cannabis substitution programs: alcohol use declined over time as an independent trend, separate from the cannabis substitution effect. The mixed-effects model decomposed these effects and found both were present and independent. Cannabis substitution reduced alcohol use through stable between-person differences; program participation and time also reduced alcohol use through a separate pathway.
For researchers evaluating cannabis substitution programs targeting alcohol use disorder, this means both effects need to be modelled. Programs that measure only overall alcohol decline risk, attributing the time-based trend to cannabis substitution, and inflating the cannabis effect. Programs that control too aggressively for time may suppress the independent alcohol decline and underestimate total program impact. The Goulet-Stock et al. approach, explicitly modelling both and reporting them separately, is the methodological template for future evaluations.
This article summarizes peer-reviewed research findings for an academic and research audience. It does not constitute medical advice.
References
- Goulet-Stock, Hacksel, Scandiuzzi, Boyd, Pauly & Stockwell (2026). International Journal of Drug Policy, 147: 105083. DOI: 10.1016/j.drugpo.2025.105083