Cannabis Screening Test Reveals Two Distinct Dimensions of Risk, Swiss Study Finds

A widely used questionnaire for detecting cannabis use disorder may be telling clinicians two different stories at once, according to a new analysis of survey data from more than 3,400 people who use cannabis in Switzerland. Researchers at the Psychiatric University Hospital and the University of Zurich report that the Cannabis Use Disorders Identification Test, revised version known as the CUDIT-R, is best understood not as a single scale but as a two-dimensional instrument that separates how much people use cannabis from how aware they are of the problems that use creates. The findings, published in Advances in Drug and Alcohol Research, could change how early screening for problematic cannabis use is conducted and how interventions are tailored to the people who need them most.

Cannabis use has been rising across Europe despite its illegal status in many countries, and in Switzerland roughly 5.5 percent of men and 2.5 percent of women aged 15 to 64 reported using cannabis within the past month. Heavy use carries a documented risk of developing cannabis use disorder, a condition that can become chronic and is associated with anxiety, depression, and psychosis. A review of the eleventh revision of the International Classification of Diseases has emphasized that current treatment systems tend to target advanced stages of substance use disorder, leaving a gap at the earlier stages where secondary prevention could still halt progression. Identifying problematic use early is therefore considered crucial, yet it remains difficult in practice because cannabis is illegal in many places, because people who use it may have limited awareness of the risks, and because the population of users is highly diverse.

The CUDIT was developed to meet this need. The original ten-item version, created by Adamson and Sellman, showed good psychometric properties but was validated only on a small sample of 53 people with alcohol dependence. A later revision using a clinical sample of 144 participants identified a two-factor structure for the original scale and a one-factor structure for a shortened eight-item version. Subsequent studies supported the one-factor solution for the short form, but they relied on college students with low levels of cannabis use. A team led by Annaheim then replaced three poorly performing items to produce the ten-item CUDIT-R with improved properties in high-risk populations, treating the scale as unidimensional. The ambiguity over the instrument’s underlying structure, combined with the small and specific samples used in earlier work, left open the question of what the CUDIT-R actually measures when applied to a broad population of cannabis users.

To resolve this, the Zurich team analyzed data from 3,454 people who use cannabis, collected through an anonymous online survey in the German- and French-speaking parts of Switzerland between November and December 2016. Participants ranged in age from 10 to 99 years, with a mean of 30.31 years, and 20.5 percent were women. To qualify, respondents had to have used non-medical cannabis within the past month and expressed interest in a future study on cannabis regulation. The researchers split the sample into four demographic groups by gender and by age relative to the median of 27 years, a cut-off chosen because it aligns with evidence on neurodevelopmental maturation, declining cannabis prevalence with age, and the end of emerging adulthood. The team then applied principal component analysis with oblique rotation to explore the factor structure, followed by confirmatory factor analysis to test whether that structure held across the four groups.

The statistical results were striking. Two components with eigenvalues above Kaiser’s criterion of one explained 50.1 percent of the variance in the ten CUDIT-R items. The first factor, labeled Use Intensity, captured how often and how heavily participants used cannabis, including frequency of use, intoxication lasting more than six hours, and habitual consumption. The second factor, labeled Awareness of Problematic Use, reflected the negative consequences of cannabis on personal, social, and working life, as well as concerns expressed by people around the user. Both factors showed acceptable internal consistency, with Cronbach’s alpha values of 0.71 and 0.72 respectively. The correlation between the two factors was statistically significant but modest at 0.52, suggesting they genuinely represent different concepts. Two items, difficulty in quitting and morning use for recovery after heavy use, loaded on both factors and were counted half toward each when calculating factor scores.

Confirmatory factor analysis confirmed an acceptable model fit using the comparative fit index, root mean square error of approximation, and standardized root mean square residual, all of which are more reliable than the chi-square statistic when sample sizes are large. Testing for measurement invariance across the four demographic groups revealed partial metric invariance: factor loadings were comparable between groups except for the item on difficulties at school or work, which loaded most strongly on Awareness of Problematic Use among younger women. When the analysis compared only gender or only age groups, full metric invariance was achieved, indicating stronger equivalence in these simpler splits. This matters because measurement invariance is the statistical precondition for meaningfully comparing scores across groups; without it, differences in scores could reflect differences in how the instrument functions rather than genuine differences in cannabis-related risk.

The demographic findings added nuance to what was previously known. When the CUDIT-R was treated as a single total score, younger men scored significantly higher than all other groups. But the two-factor view told a different story. Younger men and younger women were significantly more aware of the negative effects of their cannabis use than older participants, while younger women scored significantly lower on Use Intensity than everyone else. The authors suggest this pattern may reflect the social and employment turbulence common in emerging adulthood, such as frequent changes in relationships and jobs, which could make the consequences of use more visible to younger users. It may also reflect the fact that neurological effects of heavy cannabis use, including deficits in attention, learning, and memory, are more severe and persistent in younger users. Older users, by contrast, may have more stable lives, may have learned to avoid negative effects, or may simply be the survivors of a process in which those who experienced problems stopped using.

The study also examined motives for cannabis use, ranking 22 motives by average endorsement and selecting the mid-50 percent for analysis because motives endorsed by nearly everyone or nearly no one provide little discriminating information. Structural equation modeling showed that using cannabis to forget worries, reduce sadness or nervousness, alleviate pain, improve appetite, or enhance a party experience was positively related to one or both CUDIT-R factors. Coping-related motives in particular were linked to higher use intensity and greater awareness of problems, consistent with previous research showing that coping motives reliably predict more frequent use and worse consequences. Interestingly, using cannabis for higher creativity and openness to new experiences was associated with lower scores on one or both factors, while using it for better concentration was associated with higher awareness but lower intensity. The motives explained approximately 17.3 percent of the variance in awareness and 25.5 percent of the variance in intensity, and crucially, these relationships held across all four demographic groups.

The practical implications are considerable. The authors argue that simple variables such as age, gender, and use motives, which are commonly collected in non-clinical questionnaires, could serve as an efficient first-stage screen to identify people who use cannabis and may be at risk, who could then receive more detailed assessment with the full CUDIT. The two-factor structure also suggests different intervention entry points: people with high awareness of problems may be more receptive to direct help and personalized interventions, while those with low awareness may first need their risk perception raised. People with high use intensity might benefit from interventions focused on reducing consumption, regaining control, and safer-use practices. Because coping motives were consistently linked to worse outcomes, offering alternative strategies for managing worries, sadness, and nervousness could reduce both use and harm, and users who turn to cannabis for concentration should be informed that the drug does not enhance and may impair cognitive performance.

The study has limitations that the authors acknowledge. The sample was recruited through snowball sampling and was directed at German and French speakers in Switzerland with past-month cannabis use and interest in a cannabis regulation pilot study, so it is not representative of all cannabis users in the country. The cross-sectional design means causal relationships and changes over time cannot be examined, and collapsing participants into two broad age groups limits the ability to detect nuanced or non-linear age effects. Social desirability bias may also have influenced responses despite the anonymity of the survey. Future research, the authors suggest, should treat the CUDIT-R as a two-factor instrument, weight items differently for different demographic groups, explore finer age categories, and investigate additional variables such as education, employment status, and social environment. Longitudinal studies will be needed to clarify how motives, demographics, and cannabis use disorder interact over time, but for now the message is clear: a single number on a screening test may be hiding two very different stories about cannabis risk.

Subject of Research: Factor structure of the Cannabis Use Disorders Identification Test-Revised and its associations with demographic factors and cannabis use motives

Article Title: A two-factor structure for cannabis use disorder identification test and its associations with demographic factors and cannabis use motives

Article References: A two-factor structure for cannabis use disorder identification test and its associations with demographic factors and cannabis use motives. (n.d.). https://doi.org/10.3389/adar.2026.16094

Image Credits: AI Generated

DOI: 10.3389/adar.2026.16094

Keywords: cannabis use disorder, CUDIT-R, factor analysis, screening, cannabis use motives, gender differences, age differences, Switzerland, structural equation modeling, secondary prevention, psychometrics, addiction research