Identification of cannabis use disorder using the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM‐5) versus DSM‐5‐proxy measures: Differences by socio‐demographic characteristics

Abstract

Background and Aims

Cannabis use disorder (CUD) is a pressing public health concern in the United States, and understanding trends in prevalence requires considerations of how changes in measurement influence identification of CUD. Starting in 2021, the National Survey on Drug Use and Health (NSDUH) assessed CUD using all 11
Diagnostic and Statistical Manual of Mental Disorders
, Fifth Edition (DSM‐5) criteria in their entire sample; before then, studies could use a nine‐criteria DSM‐5‐proxy measure based on shared DSM‐IV criteria. We aimed to identify sociodemographic characteristics associated with additionally identified CUD when using the full DSM‐5 measure compared with the DSM‐5‐proxy measure.



Design
Observational study using nationally representative data.


Setting
United States.


Participants

39 973 participants ages 12 + who reported past‐year cannabis use in the 2021–2023 NSDUH (weighted
N
 = 57 872 556).



Measurements
Additionally identified mild, moderate, or severe CUD was defined as meeting 2–3, 4–5, or 6 + of the 11 DSM‐5 criteria and not meeting the same thresholds with the nine DSM‐5‐proxy criteria. Multinomial logistic regression models compared characteristics of people with additionally identified mild (vs. none), moderate (vs. mild), and severe (vs. moderate) CUD. We also calculated the percentage of people meeting each of the 11 criteria who had additionally identified CUD.


Findings
Among people who reported past‐year cannabis use, 30.5% had DSM‐5 CUD, including 7.9% with additionally identified mild CUD not previously identified using the DSM‐5‐proxy measure, 6.9% with additionally identified moderate CUD and 3.4% with additionally identified severe CUD. People who were younger (vs. 35–49); multiracial (vs. white); non‐Hispanic American Indian or Alaska Native (vs. white); or publicly insured or uninsured (vs. privately insured only) were more likely to have additionally identified CUD and/or CUD severity. People who were older (vs. 35–49); Hispanic (vs. white); were non‐Hispanic Asian, Native Hawaiian, or Pacific Islander (vs. white); reported female sex (vs. male); or had annual income greater than $75 000 (vs. <$20 000) were less likely to have additionally identified CUD and/or CUD severity. Additionally identified CUD was most common among those meeting “craving” (25.0%), “withdrawal” (20.3%), and “spending time” (14.8%) criteria.


Conclusions

The full
Diagnostic and Statistical Manual of Mental Disorders
, Fifth Edition (DSM‐5) measure of cannabis use disorder (CUD) additionally identified CUD differentially across sociodemographic groups in the United States, compared with the DSM‐5‐proxy measure. Some groups (e.g., people younger than 35 years old, people with any public or no insurance, and people who are multi‐racial or non‐Hispanic American Indian or Alaska Native) may need more CUD‐related services and support than previously expected.