Methods
Using data from two waves of the National Epidemiologic Survey of Alcohol and Related Conditions (NESARC), Hasin and colleagues recently reported that the past-year prevalence of marijuana use in the United States has more than doubled in a decade (2002/2003–2012/2013), alongside a sharp increase in the past-year prevalence of DSM-IV marijuana use disorder.1 We analyzed data from the adult samples (ages 18+) of the 2002–2013 National Survey on Drug Use and Health (NSDUH; N=451,160) to directly compare prevalences and trends in marijuana use and marijuana use disorder from that survey with those from the NESARC. The NSDUH is an annual survey representative of the household-dwelling population of the United States and is a primary source of information about prevalence and trends in drug use. Similar methods have been used annually since 2002 with typical response rates around 75%.2–4
Results
Annual past-year prevalences of marijuana use, DSM-IV marijuana use disorder, and marijuana use disorder among past-year users (conditional prevalence) are plotted in the Figure. Although the trends are not necessarily linear, we fitted each series to a trend line to evaluate the average annual change between 2002 and 2013. The trend for prevalence of marijuana use was positive and significant (β=0.021; 95% CI: 0.017, 0.025; p<0.001), reflecting a relative increase of 19%. The trend for prevalence of marijuana use disorder was flat (β=0.006; 95% CI: −0.015, 0.003; p=0.23). There was a net decrease in the conditional prevalence of marijuana use disorder (β=−0.018; 95% CI: −0.027, −0.009; p <0.001). The Table lists prevalence estimates for each outcome for the years 2002 and 2013, along with comparable NESARC results.
Figure.
Past-year prevalence estimates for marijuana use (triangles), marijuana use disorder (squares) and marijuana use disorder among past-year users (circles), 2002–2013. Lines represent fits to linear trend models and are not meant to describe the precise functional form of the trends. Error bars represent 95% confidence intervals. Trends were analyzed using the “surveylogistic” procedure in SAS version 9.4, modeling each outcome variable as a function of year of assessment.
Table.
Past-year prevalences of marijuana use, marijuana use disorder and marijuana use disorder among past-year marijuana users, estimated from the 2001–02 NESARC, the 2012–13 NESARC3, the 2002 NSDUH and the 2013 NSDUH.
| Marijuana Use | Marijuana Use Disorder | Marijuana Use Disorder Among Users | ||||
|---|---|---|---|---|---|---|
|
| ||||||
| Prevalence | 95% CI | Prevalence | 95% CI | Prevalence | 95% CI | |
| NSDUH | ||||||
| 2002 | 10.5 | (9.9, 11.0) | 1.6 | (1.4, 1.8) | 14.9 | (13.5, 16.4) |
| 2013 | 12.5 | (12.0, 13.0) | 1.5 | (1.3, 1.6) | 11.6 | (10.4, 12.8) |
| Relative Change | +19% | (12%, 22%) | −8% | (−22%, 8%) | −23% | (−13%, −35%) |
| NESARC | ||||||
| 2001–2002 (Wave 1) | 4.1 | (3.8, 4.4) | 1.5 | (1.3, 1.7) | 35.6 | (32.8, 38.4) |
| 2012–2013 (Wave 3) | 9.5 | (9.0, 10.0) | 2.9 | (2.6, 3.2) | 30.6 | (28.5, 32.7) |
| Relative Change | +131% | (117%, 146%) | +93% | (73%, 113%) | −14% | (−5%, −23%) |
NSDUH prevalence estimates were estimated from public use files using the “surveyfreq” procedure in SAS version 9.4. NESRAC and NEARC3 estimates are from Hasin et al.1 Confidence intervals for relative change estimates were calculated from pooled standard errors.
Discussion
In contrast to results from two waves of the NESARC covering roughly the same time period, NSDUH estimates suggest a more modest increase in marijuana use and no increase in the prevalence of marijuana use disorder. It is well known that individuals under-report socially proscribed behaviors in face-to-face interviews like those utilized by the NESARC.5 In contrast, the NSDUH uses audio-computer administered self-interview and other methods known to enhance privacy and reduce social desirability bias.2,4,5 Accordingly, a previous comparison of drug use prevalence estimates from NSDUH and NESARC (Wave 1) showed that NSDUH estimates were two- to five-times higher for all drugs.4 Because marijuana has become more socially acceptable, people may now be more willing to disclose use to an interviewer than they were in 2002. This likely led to a partial closing of the gap in reported prevalence of use between NSDUH and NESARC (Table).
If marijuana use was under-reported in the NESARC, this likely led to bias in the prevalence estimate of marijuana use disorder, because only those who report past-year use are assessed for marijuana use disorder. If under-reporting was more prevalent in the first wave of the NESARC than in the most recent wave (Wave 3), this could explain the apparent increase in the prevalence of marijuana use disorder. Trends in the past-year conditional prevalence of marijuana use disorder were similar across surveys, though NESARC estimates were substantially higher, suggesting that its diagnostic assessment is more sensitive than the one used in the NSDUH.1,4
There were also methodological differences between the two NESARC waves. For example, in Wave 1, interviews were conducted by U.S. Census Bureau employees, whereas Wave 3 utilized a private company.1,4 It is possible that individuals are less likely to disclose illegal behaviors to government employees than to private sector interviewers.
In summary, changes in the social acceptability of marijuana use and methodological changes likely account for much of the apparent doubling in prevalences of marijuana use and disorders reported by Hasin and colleagues.1
Acknowledgments
This work was supported by grants from the National Institute on Drug Abuse: DA23668, DA32573 and DA4041 and DA031288. The funding agency had no role in the design, conduct, collection, management, analysis or interpretation of data, or the preparation, review or approval of the letter. Public use NSDUH data were obtained from the Interuniversity Consortium for Social and Political Research. RAG had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. LJB is listed as an inventor on Issued U.S. Patent 8,080,371,“Markers for Addiction” covering the use of certain SNPs in determining the diagnosis, prognosis, and treatment of addiction.
References
- 1.Hasin DS, Saha TD, Kerridge BT, et al. Prevalence of Marijuana Use Disorders in the United States Between 2001–2002 and 2012–2013. JAMA Psychiatry. 2015 Oct;:1. doi: 10.1001/jamapsychiatry.2015.1858. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Kennet J, Gfroerer JC. Evaluating and Improving Methods Used in the National Survey on Drug Use and Health. Department of Health and Human Services, Substance Abuse and Mental Health Services Administration, Office of Applied Studies; 2005. [Google Scholar]
- 3.Substance Abuse and Mental Health Services Administration. Results from the 2012 National Survey on Drug Use and Health: Summary of National Findings (No. NSDUH Series H-46, HHS Publication No.(SMA) 13-4795) 2013 http://archive.samhsa.gov/data/NSDUH/2012SummNatFindDetTables/Index.aspx.
- 4.Grucza RA, Abbacchi AM, Przybeck TR, Gfroerer JC. Discrepancies in estimates of prevalence and correlates of substance use and disorders between two national surveys. Addict Abingdon Engl. 2007;102(4):623–629. doi: 10.1111/j.1360-0443.2007.01745.x.. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Tourangeau R, Rips LJ, Rasinski K. The Psychology of Survey Response. Cambridge University Press; 2000. [Google Scholar]