We examined distributions of key demographics of the total adult sample (aged ≥18) and the prevalence of CU and CUD. We calculated the CUD prevalence in the sample and among past-year cannabis users (conditional probability of CUD given use), respectively, by survey year for each racial/ethnic group. We conducted logistic regression analysis of the total adult sample to determine racial/ethnic and yearly variations in the prevalence of CU and CUD, respectively. Among past-year cannabis users, we examined racial/ethnic differences in weekly CU, monthly CU, and CUD. We conducted logistic regression analyses to estimate associations of race/ethnicity with weekly CU, weekly CU, and CUD among cannabis users, while adjusting for age, sex, education, and survey year to lessen for their confounding effects. All analyses took into account the NSDUH’s complex designs, such as weighting and clustering (StataCorp, 2013). All results are weighted except for sample sizes.
