Assessing Changes in Alcohol Use and Alcohol Use Disorder Prevalence in the United States: Evidence From National Surveys From 2002 Through 2014

Grant et al¹ have previously reported what they call “substantial increases” in 12-month prevalence estimates of alcohol use and alcohol use disorders, as defined by the DSM-IV based on a comparison of 2 surveys of the US adult population conducted about 10 years apart. They found that the prevalence of alcohol use increased from 65.4% to 72.7% of the surveyed population between the periods 2001-2002 and 2012-2013; the prevalence of alcohol use disorders increased from 8.5% to 12.7% of the surveyed population in the same time frame. We evaluated evidence of this trend in a concurrent series of 13 independently conducted, annual probability surveys of the US population from 2002 through 2014 on these same points and examined if and when any substantial increase in alcohol use and alcohol use disorder prevalence estimates occurred.

As part of the National Surveys on Drug Use and Health (NSDUH), 13 independent, nationally representative probability samples conducted from 2002 through 2014 on noninstitutionalized US civilians were assessed.² The field staff of NSDUH used institutional review board–approved protocols and audio computer-assisted self-interviews to collect data on alcohol use, alcohol dependence, and nondependent alcohol abuse as defined by DSM-IV diagnostic criteria. The NSDUH public use files include data on 492 831 adults 18 years and older.

The institutional review board at Michigan State University ruled that this analysis did not constitute human subjects research because all data came from deidentified files made available for public use. The data sources were US Department of Health and Human Services, Substance Abuse and Mental Health Services Administration, and Center for Behavioral Health Statistics and Quality.

Data were used to estimate year-specific prevalence estimates of each specified health behavior or condition. Estimates from the NSDUH online data analysis tool account for analysis weights, standard errors, and 95% confidence intervals. Statistical summaries were prepared using the Metacum modules of Stata, version 14.1 (StataCorp); logistic regressions were prepared using Joinpoint, version 4.5 (National Cancer Institute).³

From 2002 through 2014, year-specific sample sizes varied from 36 370 to 41 671 individuals. Between 82% and 91% of sampled dwelling units consented to participate and 71% to 79% of invited individuals consented to participate. The Figure shows relatively stable 12-month prevalence estimates for the 3 outcomes over the 13-year study period. Notably, there was a possible decline in the prevalence of DSM-IV–defined nondependent alcohol use after 2005. The Table summarizes the 2002 estimates, followed by conventional cumulative meta-analysis summary estimation that treats each successive estimate as an additional study result. For all 3 alcohol-related outcomes, the final meta-analysis summaries were not appreciably different from the 2002 estimate. When comparing the meta-analytic summary estimate for 2012 and 2013 with the 2002 estimate, we found no appreciable differences in prevalence estimates for alcohol drinking and alcohol dependence. For nondependent alcohol abuse as defined by the DSM-IV, a smaller prevalence estimate was observed for 2012 and 2013 compared with 2002.

Joinpoint analysis indicated monotonic trends for all 3 outcomes with modest changes from 2002. For alcohol-drinking prevalence, the estimated annual percentage increment was 0.2% (95% CI, 0.1% to 0.3%) with an estimated 10-year cumulative increment of 2.0%. No robust change occurred in alcohol dependence estimates; for this variable, the annual percentage change was −0.5% (95% CI, −1.2% to 0.1%). We also observed a 12-month decline in nondependent alcohol abuse prevalence estimates, with an annual percentage decrement of −1.9% (95% CI, −2.5% to 1.3%) with an estimated 10-year cumulative decrement from approximately 4.3% to approximately 3.7%.

In 1961, the Group for the Advancement of Psychiatry⁴ drew attention to what they called “problems of estimating changes in frequency of mental disorders,” which highlighted potential influences of the research approach. The approaches used by NSDUH for sampling, recruiting, and assessing participants and creating analysis weights were highly consistent and aimed to reduce problems of shifting estimates throughout the 12-year data collection period, consistent with principles outlined by the Group for the Advancement of Psychiatry. We conclude that prevalence estimates derived from these NSDUH data do not confirm the epidemiological hypothesis that the prevalence of alcohol use or alcohol use disorder (as defined by the DSM-IV) substantially increased among US adults during the first decade of the 21st century.

Differences in epidemiological survey estimates can sometimes occur from seemingly minor method variations, including alteration in participation levels, incentives to participate, and survey diagnostic assessment approaches. These variations deserve special attention when increases or declines are seen in study populations or study subgroups (such as older adults⁵), with consistency checks whenever additional surveys provide useful evidence. Resolution of contradictory results of the type seen here calls for the collaboration of epidemiologists and survey methodologists.

Corresponding Author: James C. Anthony, PhD, Department of Epidemiology and Biostatistics, Michigan State University, B601 W Fee Hall, 909 Fee Rd, East Lansing, MI 48823 (janthony@msu.edu).

Published Online: December 27, 2017. doi:10.1001/jamapsychiatry.2017.4008

Author Contributions: Drs Cheng and Anthony 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.

Study concept and design: Cheng, Anthony.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: All authors.

Critical revision of the manuscript for important intellectual content: Cheng, Breslau, Anthony.

Statistical analysis: All authors.

Obtained funding: Anthony, Breslau.

Administrative, technical, or material support: Anthony.

Supervision: Cheng, Breslau, Anthony.

Conflict of Interest Disclosures: None reported.

Funding/Support: This work was supported by the National Institute of Drug Abuse (grant T32DA021129 to Dr Cheng and grant K05DA015799 to Dr Anthony), the National Institute on Minority Health and Health Disparities (grant R01MD010274 for Dr Breslau), and Michigan State University (Drs Cheng and Anthony and Mr Kaakarli).

Role of the Funder/Sponsor: The funding organizations had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The content is the sole responsibility of the authors and does not necessarily represent the official views of Michigan State University, RAND Corporation, the US National Institute on Drug Abuse, or the US National Institutes of Health.

Additional Contributions: Public use files are made available by the US Department of Health and Human Services, Substance Abuse and Mental Health Services Administration, and Center for Behavioral Health Statistics and Quality via this website: https://www.icpsr.umich.edu/icpsrweb/ICPSR/series/64. These institutions did not provide funding to this study, and no compensation was received from a funding sponsor for such contributions.