Taxonicity of Cannabis Use Disorder: Findings from Nationally Representative Community Sample and an Inpatient Clinical Sample

Operational diagnostic definitions of drug addiction have evolved considerably over the last twenty years, including both a categorical one (substance dependence; ICD-11) and predominantly dimensional one (substance use disorder [SUD]; DSM-5). Although cannabis use disorder (CUD) is among the most prevalent SUDs, few studies have directly evaluated whether its underlying structure is categorical or dimensional or investigated this question with a clinical sample. Three taxometric procedures, Mean Above-Minus Below a Cut [MAMBAC], Maximum Eigenvalue [MAXEIG)], Latent Mode [L-Mode], were conducted in two datasets: (1) participants who reported cannabis use in the National Epidemiological Study and Alcohol and Related Conditions III (NESARC-III), a large, nationally representative sample of U.S. community adults (N=3623) and (2) patients reporting pre-admission cannabis use in an inpatient SUD treatment program in Ontario, Canada (N=621). Comparison curve fit indices (CCFI) for the NESARC-III analyses supported dimensional structure: MAMBAC=0.48; MAXEIG=0.30; L-Mode=0.43; Mean CCFI=0.40. CCFI coefficients for the clinical dataset also supported dimensional structure: MMAMBAC=0.09, MAXEIG=0.16, L-Mode=0.29, Mean CCFI=0.18. These results suggest that the latent structure of cannabis addiction is a dimensional construct along a spectrum of severity rather than a dichotomous categorical construct. These findings are more consistent with the DSM-5 conceptualization compared to ICD-11.