Researchers pooled longitudinal data from 829 participants originally enrolled in seven separate pharmacologic clinical trials for cannabis use disorder, each of which independently tracked weekly cannabis, alcohol, and cigarette use as part of its own study design.

The team used mixed-effects models to ask whether a given week’s cannabis use or craving predicted the following week’s alcohol and cigarette use, and whether being randomized to active medication changed that relationship, an approach that lets a single analysis draw on trial data never originally designed together for this specific question.

Weeks with both higher cannabis craving and higher cannabis use frequency predicted a small increase in the following week’s alcohol quantity, a pattern the authors describe as consistent with complementary use, where cannabis and alcohol use rise and fall together rather than substituting for each other.

This effect did not extend to how often participants drank, only to how much, and the statistical signal was modest, with a p-value of 0.0497, just inside the conventional threshold for significance.

Higher cannabis quantity in a given week predicted slightly lower cigarette quantity the following week across the full sample, a pattern more consistent with substitution than complementary use.

Among participants specifically assigned to active pharmacotherapy, higher cannabis craving was linked to fewer smoking days the following week, a treatment-specific signal that the authors flag as an initial, exploratory finding rather than an established effect.

Across the pooled analysis, no significant three-way interactions between treatment assignment, cannabis measures, and substance outcomes emerged, meaning active pharmacotherapy did not broadly reshape how cannabis use related to alcohol or tobacco use.

The one treatment-specific finding, for cigarette frequency and craving, was described by the authors as limited evidence, a signal worth following up rather than a confirmed treatment effect.

This was an exploratory secondary analysis of data collected for other purposes, pooled across seven pharmacologically distinct trials rather than a single study designed to test substitution effects, and it relied on self-reported substance use rather than biological verification.

The p-values for the key findings sat close to the conventional 0.05 threshold, and the analysis does not describe correction for the number of comparisons tested, both of which mean these results should be read as hypothesis-generating rather than confirmatory.

This analysis adds to a mixed and long-running literature on substance substitution during addiction treatment, where some studies find that reducing one substance leads patients to increase use of another, and others find no such relationship or even simultaneous declines across substances.

The authors frame their findings as initial insights meant to prompt closer monitoring of alcohol and tobacco use during cannabis use disorder treatment, not as a settled account of how cannabis cessation affects other substance use.

This is exactly the kind of study I want to see more of: a real attempt to look past the single substance a trial was designed to treat and ask what happens to a patient’s broader substance use pattern. Addiction rarely stays confined to one substance, and treatment research that ignores that misses something clinically important.

That said, I would not change how I counsel patients based on p-values this close to the 0.05 line in an exploratory, pooled analysis of trials that were not designed to test this question. What I take from this is a reminder to ask patients in cannabis use disorder treatment about their alcohol and tobacco use directly, rather than assuming one will improve as the other does. The signal here is worth watching, not yet worth building a clinical protocol around.

A Four-Step Reading Frame