EVIDENCE FROM A CANNABIS USE DISORDERS MULTIPLE CAUSES MODEL FOR THE CANNABIS-TOBACCO COMBINATION KNOWN AS ‘BLUNTS’ Background What is a cannabis ‘blunt’?  A method of tobacco-cannabis co-administration where tobacco from a cigar is removed and replaced with cannabis.  Similar to other forms of co-use such as “spliffs” or “mulling” (mixing tobacco-cannabis in a joint), or “chasing” (smoking tobacco immediately after cannabis) 1 . How common is blunt smoking and who are primarily affected?  Lifetime blunt use may be as high as 10% among youths 2  In the US, as many as half of cannabis users have smoked a blunt recently (within the past 30 days) 2 . .  Blunt smokers are disproportionately male, African-American, older youths, and urban 3,4 . Why might the study of blunt smoking be important?  Rates of blunt use have been increasing among younger cohorts and differences between racial/ethnic groups are diminishing 5 .  Blunt smokers might experience more cannabis and nicotine dependence, but this finding needs to be further replicated 4 . Reference: 18-25 Reference: Females Reference: White CFI = 0.932 TFI = 0.932 RMSEA = 0.004 HPCU e 1 cd1 e 2 cd2 e 3 cd3 e 4 cd4 e 5 cd5 e 6 cd6 e 7 ca1 e 8 ca2 e 9 ca3 e 10 ca4 e 11 Blunts e 12 male 12-17 26-34 35-49 50+ Black Hispanic Other .21* .05 .30* -.24* -.15 -.95 .04 -.02 .08 .12* -.47* -.31* -1.54* -1.20* .44* .03 -.06 .77* .66* .74* .64* .80* .89* .87* .69* .67* .81* .19 .12 .10 -.02 -.12 -.05 -.15 -.10 -.12 -.18 B.J. Fairman and J.C. Anthony Department of Epidemiology, Michigan State University, East Lansing, MI USA Contact Information: Brian J. Fairman Room B601 West Fee Hall 909 Fee Road bfairman@epi.msu.edu Objectives What is this research about?  Explore the factor structure of harmful or problematic cannabis use (HPCU) as measured by items designed to tap DSM-IV cannabis use disorders (CD, CA).  Evidence from most, but not all prior studies of this type have favored a single underlying trait 6 .  DSM-5 combines most dependence and abuse criteria.  Estimate the degree to which blunt smoking is associated with the HPCU level, holding constant sex, age, and race/ethnicity.  We expected blunt smoking to be associated with a greater level of HPCU after covariate adjustment.  Assess whether blunt smokers experience individual CD/CA features differently than non-blunt cannabis users, adjusting for background characteristics and level of HPCU. Methods Design and Sample  Data: US National Surveys on Drug Use and Health (NSDUH) 2004-2005. Nationally representative annual cross-sectional surveys of non-institutionalized US residents 12 years of age or older; participation level 70%. Sample focused on respondents who used cannabis on six or more days in the past year. Sample size: 2004 data (n = 7,054), 2005 data (n = 6,853). Variables  Harmful/Problematic Cannabis Use (HPCU): Ten features (No vs. Yes) based on DSM-IV clinical criteria for cannabis dependence and abuse (see Figure 2).  Lifetime blunt smoking (Blunts): No vs. Yes.  Sex: Male vs. Female.  Age at interview: 12-17, 18-25, 26-34, 35-49, 50 years or older.  Race/ethnicity: White, African-American/Black, Hispanic, All others. Analysis  In a random half sample of 2004 data (S1: n = 3,534), exploratory factor analysis (EFA) was performed to reveal the underlying structure of HPCU items. Data from other half sample (S2: n = 3,520) was used for confirmatory factor analysis (CFA) comparing two factors (DEPEND and HARM) with a single factor (HPCU). Analysis was replicated using the 2005 data (n = 6,853). Factor loadings, scree plot of eigenvalues, comparative fit index (CFI), Tucker-Lewis index (TFI), root mean square error of approximation (RMSEA), and residual correlation values among factors were examined to evaluate factor structure and model fit.  The unidimensional continuous latent HPCU factor scores were regressed on blunt smoking to estimate the unadjusted association. Next, multiple indicators, multiple causes models (MIMIC) relating blunt smoking, HPCU, and background covariates were estimated: 1) To control for background influences in blunt-HPCU estimate; 2) To test if blunt smoking was directly related to individual HPCU items, controlling for HPCU level. All analyses accounted for sampling weights and complex survey design. Mplus v. 5.0 used for data analysis. Results Figure 3. Exploratory Factor Analysis Loadings and Scree Plot of Eigenvalues. NSDUH 2004 (S1: n=3534) Figure 5. Multiple Indicators, Multiple Causes Model of Relationship Between Blunt Use, HPCU, and Other Factors. NSDUH 2005 (n=6853) Strengths and Limitations Strengths:  Large, representative samples as found here promoted statistically precise estimates, and results that could be generalized to the majority of the US population.  The ability to directly replicate findings and check for estimate stability across independent samples, with reduced variability due to consistent sampling designs and data collection procedures. Limitations  Using cross-sectional data, the temporal sequence of blunt smoking and onset of HPCU features was ambiguous, thus inferring causality is cautioned.  This study was limited to past-year prevalence of HPCU features, and blunt smoking might be associated with duration/ reoccurrence of HPCU, rather than incidence.  Covariate adjustment was limited to background factors. Main Findings  The tobacco-cannabis combination called blunts is the rule rather than exception in the US among past-year cannabis users; prevalence was as high as 90% in some cannabis-smoking groups.  We determined that harmful or problematic cannabis use (HPCU) features best fit a unidimensional latent model, confirming prior findings by others 6 .  Blunt smokers experienced greater levels of HPCU, even after accounting for the influence of sex, age, and race/ethnicity.  Blunt smokers did not experience any individual HPCU feature over-and-above that which could be explained by level of HPCU and other model covariates.  Post-estimation finding of no direct link from being male or African-American with level of HPCU, given adjustment for blunt smoking, was unexpected, and warrants further inquiry into role blunt use play in cannabis use disorders for these at-risk groups. Figure 1. Prevalence of Blunt Smoking Among Past- Year Cannabis Users. Figure 2. Prevalence of Harmful or Problematic Cannabis Use by Blunt Smoking. NSDUH 2004 (n = 7,054) * p<0.05 **Estimates of paths depicted with black arrows are standardized coefficients.  Proportion of past-year cannabis users who have smoked a blunt was high (>70%) in both NSDUH 2004 and 2005 samples.  Rates were especially elevated for cohorts born after 1970/71 (younger than 35 yrs), African-Americans, and Hispanics.  Few male-female or year-to-year differences were observed.  ‘Spending more time’ getting/using cannabis, and tolerance were the most frequently experienced; ‘trouble with the law’ the least (range).  Blunt smokers experienced a higher proportion of HPCU items than non-blunt cannabis users, except for ‘problems with family/friends’.  Scree plot and ratio of 1 st and 2 nd eigenvalues (5.7/1.3 = 4.4) indicated a single, dominant factor.  Giving up important activities, serious role impairment, and problems with family/friends due to cannabis were the most discriminating items in the one-factor solution.  In the two-factor solution, CA items, along with CD5 and CD6 loaded on a single factor, with other CD items loading on a second. Depend cd1 e 1 cd2 e 2 cd3 e 3 cd4 e 4 cd5 e 5 cd6 e 6 Harm ca1 e 7 ca2 e 8 ca3 e 9 ca4 e 10 .73* .60* .69* .60* .74* .89* .92* .63* .75* .78* .86 CFI = 0.96 TFI = 0.96 RMSEA = 0.009 HPCU cd1 e 1 cd2 e 2 cd3 e 3 cd4 e 4 cd5 e 5 cd6 e 6 ca1 e 7 ca2 e 8 ca3 e 9 ca4 e 10 .77* .60* .74* .59* .79* .89* .87* .68* .65* .80* CFI = 0.953 TFI = 0.947 RMSEA = 0.015 Figure 4. Two-Factor and One-Factor Confirmatory Factor Analysis of Harmful or Problematic Cannabis Use. NSDUH 2004 (S2: n = 3520).  Both models fit the data well (CFIs>0.95; TFIs≥0.95; RMSEAs<0.05), two-factor model showed strong factor correlation (0.86). Based upon these results, we favored the more parsimonious model of a single latent dimension of HPCU.  Blunt smoking (blue arrow) was associated with a greater level of HPCU (0.21), adjusting for sex, age, and race/ethnicity (unadjusted est.= 0.57, not shown in figure).  Estimated probability of smoking blunts was greater for males, African-Americans, and 18-25 year olds (green arrows). Adjusting for blunt use, a younger age was associated with a higher level of HPCU, but no differences were observed by sex or race/ethnicity (p>0.05).  There was no direct relationship between blunt status and individual HPCU items (red arrows) after accounting for level of HPCU, sex, age, and race/ethnicity. References 1. Agrawal, A., Budney, A. J., & Lynskey, M. T. (2012). The co-occurring use and misuse of cannabis and tobacco: a review. Addiction, 107(7), 1221–1233. 2. Golub, A., Johnson, B. D., & Dunlap, E. (2005). The growth in marijuana use among American youths during the 1990s and the extent of blunt smoking. Journal of Ethnicity in Substance Abuse, 4(3-4), 1–21. 3. Substance Abuse and Mental Health Services Administration (SAMHSA) (2007). Use of Marijuana and Blunts Among Adolescents: 2005. The NSDUH Report. Via http://www.oas.samhsa.gov. 4. Timberlake, D. (2009). A Comparison of Drug use and Dependence Between Blunt Smokers and Other Cannabis Users. SUBSTANCE USE & MISUSE, 44(3), 401–415. 5. Timberlake, D. S. (2013). The changing demographic of blunt smokers across birth cohorts. Drug and alcohol dependence, 130(1-3), 129–134. 6. Wu, L., Woody, G. E., Yang, C., Pan, J., Reeve, B. B., & Blazer, D. G. (2012). A dimensional approach to understanding severity estimates and risk correlates of marijuana abuse and dependence in adults. International Journal of Methods in Psychiatric Research. Supported by NIDA grants R01DA016558/K05DA015799 (JCA), and T32DA021129 (BJF) Disclaimer: The content is the sole responsibility of the authors and does not necessarily represent the official views of Michigan State University, the National Institute on Drug Abuse (NIDA), or the National Institutes of Health No conflicts of interest to report. *p<0.05 Standardized estimates Implications & Future Directions Study Implications  Blunt smoking (or similar forms of tobacco-cannabis co-use) may be an important cannabis using sub-culture to consider in studies on the etiology of harmful or problematic cannabis use.  Given the relatively high prevalence of blunt smoking in the US, public health officials concerned with cannabis smoking and/or tobacco control may wish to focus greater attention on cigar consumption, marketing, and surveillance. Future Directions  Future work that builds upon these findings will focus on variation across blunt use frequency, adjustment for other factors (e.g., nicotine dependence), constraining sequence of blunt smoking and HPCU, and replication using more recent NSDUH data.  Prospective studies are needed to definitively separate cause and effect between blunt smoking and HPCU.