Drinking Beyond the Binge Threshold: Predictors, Consequences, and Changes in the U.S.

      Introduction

      Binge drinking, five or more drinks on an occasion for men and four or more for women, marks risky alcohol use. However, this dichotomous variable removes information about higher, more dangerous consumption. This paper examines predictors, consequences, and changes over a decade in drinking one to two times, two to three times, and three or more times standard gender-specific binge thresholds, labeled Levels I, II, and III.

      Methods

      In 2001–2002 and 2012–2013, respectively, 42,748 and 36,083 U.S. respondents aged ≥18 years were interviewed in person in cross-sectional waves of the National Epidemiologic Survey on Alcohol and Related Conditions (response rates, 81% and 61%). Respondents were asked their past-year maximum drink consumption per day, categorized as Levels I, II, or III. Predictors and whether Levels II and III were associated with more negative consequences were analyzed in 2012–2013 data.

      Results

      In 2001–2002, 23% of adults reported past-year binge drinking, with 15% peaking at Level I, 5% at Level II, and 3% at Level III. In 2012–2013, those percentages increased significantly to 33% binging, and 20%, 8%, and 5% binging at Levels I, II, and III, respectively. After adjusting for alcohol use disorder, the strongest predictor of Level I, II, and III binging, Level III versus I and non-binge drinkers had higher odds of past-year driving after drinking and, after drinking, experiencing physical fights, injuries, emergency department visits, arrests/detentions, and other legal problems.

      Conclusions

      Level II and III—relative to Level I—binging is associated with more negative alcohol consequences and may be increasing nationally. Research needs to explore prevention and counseling interventions.

      Introduction

      Alcohol remains the most commonly used intoxicant in the U.S. The 2012–2013 National Epidemiologic Survey of Alcohol and Related Conditions (NESARC) indicates that, among 73% of adults aged ≥18 years who drank in the past year, 46% binged
      • Dawson D.A.
      • Hingson R.W.
      • Grant B.F.
      Epidemiology of alcohol use, abuse and dependence.
      (four or more drinks for women and five or more for men over 2 hours) at least once. Binging often produces blood alcohol concentrations (BACs) ≥0.08%, the legal definition of intoxication for individuals aged ≥21 years operating a motor vehicle.
      National Institute on Alcohol Abuse and Alcoholism
      Alcohol and development in youth: a multi-disciplinary overview.
      Binge prevalence is a useful population index of dangerous alcohol consumption, but being dichotomous removes information regarding heavier, more dangerous drinking and assigns identical risk to all bingers regardless of how far they exceed the threshold.
      • Patrick M.E.
      A call for research on high-intensity alcohol use.
      • Pearson M.R.
      • Kirouac M.
      • Witkiewitz K.
      Questioning the validity of the 4+/5+ binge or heavy drinking criterion in college and clinical populations.
      A study
      • White A.M.
      • Kraus C.L.
      • Swartzwelder H.S.
      Many college freshmen drink at levels way beyond the binge threshold.
      of more than 10,000 college students found 49% of men and 30% of women drank two or more times the binge threshold (i.e., eight or more drinks for women and ten or more for men) in the previous 2 weeks. A national survey 30 years ago linked consuming five or more or eight or more drinks on an occasion with higher drinking problem scores.
      • Hilton M.
      Demographic characteristics and the frequency of heavy drinking as predictors of self-reported drinking problems.
      An analysis of the 2004–2005 NESARC suggests past-year consumption of ten or more versus five or more drinks per occasion was associated with driving after drinking too much (16% vs 8%) and placing oneself in risky situations after drinking (45% vs 30%).
      • Hingson R.W.
      • Wenxing Z.
      Age of drinking onset, alcohol use disorders, frequent heavy drinking, and unintentionally injuring oneself and others after drinking.
      Another limitation of a single binge threshold is that overall binging prevalence could decline, while the prevalence of higher drinking levels stabilizes or increases. The Monitoring the Future (MTF) study, which tracks high school substance use nationally, found 10.5% of 12th graders consumed ten or more drinks and 5.6% consumed ≥15 drinks at least once in the past 2 weeks. From 2005 to 2011, the percentage who binged at five to nine and ten to 14 drinks declined but not the percentage consuming ≥15 drinks.
      • Patrick M.E.
      • Schulenberg J.E.
      • Martz M.E.
      • et al.
      Extreme binge drinking among 12th-grade students in the United States: prevalence and predictors.
      Binging three times the threshold (≥12 drinks for women or ≥15 drinks for men) is dangerous. A 135-pound female or a 160-pound male consuming these amounts over 4 hours on an empty stomach would reach BACs ≥0.30%. Even after a full meal, BACs would exceed 0.20% for men and 0.30% for women.
      • Hingson R.
      • White A.
      Trends in extreme binge drinking among U.S. high school seniors.
      The odds of blackouts, failing to remember what transpired while drinking, are 50/50 at 0.22% BAC.
      • Perry P.J.
      • Argo T.R.
      • Barnett M.J.
      • et al.
      The association of alcohol-induced blackouts and grayouts to blood alcohol concentrations.
      One report
      • Jones A.W.
      • Holmgren P.
      Comparison of blood-ethanol concentration in deaths attributed to acute alcohol poisoning and chronic alcoholism.
      found BACs averaged 0.36% among 693 people who died from alcohol poisoning. These potentially life-threatening consequences underscore the importance of characterizing the U.S. prevalence and frequency of binging at levels exceeding the threshold.
      Initial comparisons of the NESARC I and III surveys
      • Grant B.F.
      • Goldstein R.B.
      • Saha T.D.
      • et al.
      Epidemiology of DSM-5 Alcohol Use Disorder: results from the National Epidemiologic Survey on Alcohol and Related Conditions III.
      • Dawson D.A.
      • Goldstein R.B.
      • Saha T.D.
      • Grant B.F.
      Changes in alcohol consumption: United States, 2001-2002 to 2012-2013.
      indicated past-year drinking of five or more and ten or more drinks on an occasion increased from 31.0% and 11.5%, respectively, in 2001–2002 to 39.6% and 15.5% in 2012–2013.
      This paper analyzes binge drinking in the U.S. with a finer lens by examining drinking below the gender-specific binge thresholds and then one to two, two to three, and three or more times the thresholds, categorized, respectively, as Levels I, II, and III. For women, the levels correspond to one to three, four to seven, eight to 11, and ≥12 drinks on an occasion; for men, one to four, five to nine, ten to 14, and ≥15. Used in previous studies,
      • White A.M.
      • Kraus C.L.
      • Swartzwelder H.S.
      Many college freshmen drink at levels way beyond the binge threshold.
      these levels capture BACs ranging from mild to potentially deadly intoxication. This study explores predictors of peak binge level prevalence and whether it changed from NESARC I−III. Consistent with problem behavior theory
      • Jessor R.
      Problem-behavior theory, psychosocial development, and adolescent problem drinking.
      and prior research,
      • Hilton M.
      Demographic characteristics and the frequency of heavy drinking as predictors of self-reported drinking problems.
      the authors hypothesize that higher peak binge levels predict more negative alcohol-related consequences. Because previous reports suggest higher peak bingers binge more frequently
      • White A.M.
      • Kraus C.L.
      • Swartzwelder H.S.
      Many college freshmen drink at levels way beyond the binge threshold.
      and correlate with alcohol use disorder (AUD),
      • Esser M.B.
      • Hedden S.L.
      • Kanny D.
      • Brewer R.D.
      • Gfroerer J.C.
      • Naimi T.S.
      Prevalence of alcohol dependence among prevalence of alcohol dependence among U.S. adult drinkers adult drinkers, 2009–2011.
      the authors hypothesize that Level I, II, and III binging will be associated with AUD and that, after adjusting for meeting AUD criteria, Level II and III compared with Level I binging and non-binge drinking will predict experiencing more negative alcohol-related consequences.

      Methods

      Two independent, cross-sectional, nationally representative surveys sponsored by the National Institute on Alcohol Abuse and Alcoholism of U.S. adults provided data: NESARC I (2001–2002; n=43,093; response rate, 81.0%)
      • Grant B.F.
      • Kaplan K.D.
      • Shepard J.
      • Moore T.
      Source and Accuracy Statement for Wave 1 of the 2001–2002 National Epidemiologic Survey on Alcohol and Related Conditions.
      and NESARC III (2012–2013; n=36,309; response rate, 61.1%).
      • Grant B.F.
      • Amsbary M.
      • Chu A.
      • et al.
      Source and Accuracy Statement: National Epidemiologic Survey on Alcohol and Related Conditions-III (NESARC-III).
      The Census Bureau (NESARC I) and Westat, Inc. (NESARC III) conducted fieldwork. Informed consent forms provided respondents written information about survey content, data uses, voluntary participation, and response confidentiality. Both research protocols received full ethical review and approval.
      The surveys’ identical eligibility criteria were U.S. adults aged ≥18 years living in households and non-institutional group quarters. Both surveys oversampled blacks and Hispanics; NESARC I oversampled adults aged 18–24 years and NESARC III Asians/Pacific Islanders. NESARC III permitted two respondents in minority households with four or more eligible respondents. NESARC I interviewed college students in on-campus residences, NESARC III in primary off-campus residences.
      NESARC’s trained interviewers conducted in-home interviews. NESARC III but not I offered financial participation incentives.
      • Grant B.F.
      • Kaplan K.D.
      • Shepard J.
      • Moore T.
      Source and Accuracy Statement for Wave 1 of the 2001–2002 National Epidemiologic Survey on Alcohol and Related Conditions.
      • Grant B.F.
      • Amsbary M.
      • Chu A.
      • et al.
      Source and Accuracy Statement: National Epidemiologic Survey on Alcohol and Related Conditions-III (NESARC-III).

      Measures

      Both surveys asked respondents who drank: (1) During the last year, what was the largest number of drinks that you drank in a single day? (2) About how often during the last year did you drink (this largest number of drinks) in a single day? and (3) About how old were you when you first started drinking, not counting tastes or sips of alcohol?
      The NESARC III also asked respondents the age they first drank enough to feel intoxicated (your speech was slurred, you felt unsteady on your feet, or you had blurred vision). Both surveys asked respondents if, in the last year, they: (1) drove a motor vehicle while drinking; (2) drove a motor vehicle after drinking too much; (3) had a motor vehicle accident under alcohol’s influence; and after drinking; (4) drove a motor vehicle and injured themselves or someone else in an accident; (5) got into physical fights; (6) accidentally injured themselves or someone else other than in a motor vehicle accident, like a bad fall or cut; and (7) got arrested/detained or had other legal problems.
      Survey questions gauged past-year AUD at severe (six or more symptoms), moderate (four to five), or mild (two to three) levels and whether that year respondents went anywhere or saw someone for drinking-related reasons—a physician, counselor, Alcoholics Anonymous, or other community agency, including family social services, alcohol or drug detoxification centers, inpatient hospital wards, community mental health centers, halfway houses, crisis centers, religious counselors (e.g., priests or rabbis), psychologists, or other professionals. Respondents were specifically asked about alcohol-related hospitalizations and emergency department (ED) visits.
      Finally, NESARC explored respondent demographics, past-year health status, depression symptoms, job loss, moves, number of children, and lifetime drug and smoking history.

      Statistical Analysis

      Statistical analyses were performed using SUDAAN, version 11.0 in 2015–2016 accounting for complex survey design to ensure accurate estimation. Chi-square analyses examined binge level changes from NESARC I−III for the entire samples and demographic and other subgroups. NESARC I−III data were also pooled in a multinomial logistic regression model adjusting for selected demographics and a wave indicator to see if binging at each level increased significantly from NESARC I−III.
      Cross-tabulation and chi-square (an analogy of Pearson’s chi-square) analyzed significance of binge drinking differences in various demographic and substance use groups (Appendix Table 1, available online). Multinominal logistic regression with proportional-odds cumulative logit analyses identified independent, significant predictors of Level I, II, and III binging relative to non-binge drinking. Table 1 compares NESARC I and III on covariates and binge variables adjusted for in the regression analyses. The covariates were found either in previous research
      • Blazer D.G.
      • Wu L.T.
      The epidemiology of at-risk and binge drinking among middle-aged and elderly community adults: National Survey on Drug Use and Health.
      CDC
      Vital signs: binge drinking prevalence, frequency, and intensity among adults—United States, 2010.
      • Cranford J.A.
      • McCabe S.E.
      • Boyd C.J.
      A new measure of binge drinking: prevalence and correlates in a probability sample of undergraduates.
      • Kann L.
      • McManus T.
      • Harris W.A.
      • et al.
      Youth Risk Behavior Surveillance—United States, 2015.
      • Lee H.K.
      • Han B.
      • Gfroerer J.C.
      Differences in the prevalence rates and correlates of alcohol use and binge alcohol use among five Asian American subpopulations.
      • Naimi T.S.
      • Nelson D.E.
      • Brewer R.D.
      The intensity of binge alcohol consumption among U.S. adults.
      • Tucker J.S.
      • Orlando M.
      • Ellickson P.L.
      Patterns and correlates of binge drinking trajectories from early adolescence to young adulthood.

      Larimer M, Arroyo J, editors Alcohol use among special populations (special issue). Alcohol Res 2016;38(1):1–140

      • Chen C.M.
      • Slater M.E.
      • Castle I.-J.P.
      • Grant B.F.
      Alcohol use and alcohol use disorders in the United States: main findings from the 2012–2013 National Epidemiologic Survey on Alcohol and Related Conditions-III (NESARC-III).
      Substance Abuse and Mental Health Services Administration, Results from the 2013 National Survey on Drug Use and Health: summary of national findings, NSDUH Series H-48, HHS Publication No. (SMA) 14-4863.
      or this study’s bivariate analyses to predict binge drinking. A similar regression examined predictors of meeting past-year AUD criteria, including the same covariates and binge drinking as predictors of meeting AUD criteria. Non-significant variables were dropped from the final model. ORs and 95% CIs were calculated. Wald F-tests assessed statistical significance.
      Table 1Comparison of NESARC I and III on the Distribution of Covariates Used in the Study
      CovariatesNESARC I, % (SE) (n=42,748)NESARC III, % (SE) (n=36,083)p-value of χ2 test
      Age<0.0001
       18−20 years5.9 (0.18)5.1 (0.2)
       21−25 years9.0 (0.18)9.6 (0.30)
       26−30 years9.0 (0.25)8.8 (0.24)
       31−35 years9.6 (0.18)8.3 (0.22)
       36−40 years10.8 (0.19)8.4 (0.18)
       41−50 years20.2 (0.27)18.2 (0.3)
       51−60 years15.1 (0.2)18.0 (0.32)
       ≥61 years20.5 (0.38)23.6 (0.57)
      Sex0.7365
       Male47.9 (0.31)48.1 (0.4)
       Female52.1 (0.31)51.9 (0.4)
      Race/ethnicity0.076
       Black11.1 (0.71)11.8 (1.16)
       Native American2.1 (0.18)1.6 (0.15)
       Asian4.4 (0.55)5.7 (0.77)
       Hispanic11.6 (1.28)14.7 (1.43)
       White70.9 (1.68)66.2 (2.34)
      Marital status<0.0001
       Never married17.5 (0.23)19.7 (0.39)
       Divorced/widowed/separated20.9 (0.49)22.5 (0.74)
       Married61.6 (0.49)57.8 (0.71)
      Employment status<0.0001
       Unemployed7.5 (0.21)12.5 (0.44)
       Retired/homemaker22.8 (0.35)22.3 (0.51)
       Employed65.0 (0.44)58.5 (0.66)
       Student3.2 (0.17)1.5 (0.08)
       Other4.8 (0.28)1.9 (0.09)
      College enrollment<0.0001
       Yes10 (0.23)14.2 (0.41)
       No90 (0.23)85.8 (0.41)
      Highest education<0.0001
       <High school15.7 (0.52)13.0 (0.61)
       High school29.3 (0.62)25.8 (0.72)
       College43.3 (0.55)47 (0.64)
       Graduate degree11.7 (0.37)14.2 (0.61)
      Family income<0.0001
       <$20,00023.6 (0.52)22.8 (0.74)
       $20,000−49,99936.3 (0.47)32.6 (0.58)
       $50,000−99,99928.4 (0.43)27.1 (0.49)
       $100,000−149,9997.6 (0.35)10.3 (0.46)
       $150,000−199,9992.1 (0.14)3.8 (0.28)
       ≥$200,0001.9 (0.14)3.4 (0.32)
      Health status<0.0001
       Excellent30.5 (0.55)21.8 (0.58)
       Good/fair65 (0.5)74.1 (0.48)
       Poor4.4 (0.17)4.1 (0.24)
      Past-year felt depressed0.1416
       Most/all of the time6.2 (0.17)6.2 (0.23)
       Some/few times40.9 (0.42)42.2 (0.47)
       Never52.9 (0.44)51.6 (0.49)
      Past-year moved<0.0001
       Yes16.5 (0.39)22.6 (0.5)
       No83.5 (0.39)77.4 (0.50)
      Past-year lost job0.1913
       Yes6.4 (0.18)6.7 (0.17)
       No93.6 (0.18)93.3 (0.17)
      Number of children<0.0001
       116.3 (0.26)14.8 (0.31)
       214.3 (0.23)12.4 (0.27)
       35.7 (0.18)5.2 (0.17)
       ≥42.6 (0.13)2.6 (0.12)
       061.2 (0.49)65 (0.56)
      Lifetime smoking0.6516
       Current/past year27.7 (0.61)27.2 (0.8)
       Former19.2 (0.37)18.7 (0.47)
       Never53.1 (0.78)54.2 (1.03)
      Lifetime drug use
       Current/past year6.2 (0.20)13.3 (0.46)<0.0001
       Former16.6 (0.43)23.2 (0.58)
       Never77.2 (0.54)63.6 (0.83)
      Age first used drugs<0.0001
       ≤13 years2.1 (0.11)4.6 (0.19)
       14−17 years9.8 (0.30)15.9 (0.46)
       18−20 years6.2 (0.2)9.2 (0.29)
       21−25 years2.5 (0.11)3.7 (0.13)
       26−30 years0.8 (0.05)1.2 (0.07)
       ≥31 years1.2 (0.07)2.2 (0.13)
       Never77.3 (0.54)63.4 (0.83)
      Age of first drink<0.0001
       ≤13 years3.8 (0.13)5.5 (0.18)
       14−17 years23.7 (0.45)29.2 (0.61)
       18−20 years30.3 (0.43)29.4 (0.44)
       21−25 years18.6 (0.39)19.2 (0.33)
       26−30 years3.0 (0.1)2.8 (0.13)
       ≥31 years2.9 (0.11)2.7 (0.13)
       Never17.6 (0.69)11.2 (0.61)
      Binge level<0.0001
       Level III3.2 (0.15)4.9 (0.23)
       Level II5.3 (0.17)7.7 (0.23)
       Level I15.1 (0.34)20.0 (0.48)
       Drink no binge41.6 (0.55)39.9 (0.52)
       Abstain34.8 (0.71)27.5 (0.85)
      Frequency of peak binge drinking level<0.0001
       Daily2.3 (0.09)1.3 (0.07)
       1−4/week8.7 (0.20)9.0 (0.26)
       1−3/month13.4 (0.26)14.0 (0.29)
       1−11/year40.7 (0.59)48.3 (0.89)
       Abstain34.9 (0.71)27.4 (0.85)
      Note: Age first drunk was not asked in NESARC I. Boldface indicates statistical significance (p<0.05).
      NESARC, National Epidemiologic Survey of Alcohol and Related Conditions.
      Cross-tabulation with chi-square analysis also tested whether Level I, II, and III bingers were significantly more likely than abstainers and non-binge drinkers to receive counseling, visit an ED, be hospitalized for alcohol-related reasons, drive after any drinking and drinking too much, and, after drinking, be a driver in traffic crash, be in physical fights, get injured in traffic crashes or other ways, and be arrested/detained or have other legal problems.
      Logistic regression analyses with complete case analysis examined whether, compared with non-binge drinkers, Level I, II, and III bingers experienced increased likelihoods of these outcomes, adjusting for meeting AUD criteria and select regression covariates. Strongest predictors were identified by comparing log-likelihood value of models in which predictors were entered. These analyses were repeated without AUD criteria as a covariate. Pairwise comparison was performed in each model with Bonferroni adjustment to control for overall Type 1 error.

      Results

      Table 1 compares NESARC I and III on analysis covariates. In NESARC I, in the past year, 35% abstained and 23% binged once or more: 15% at Level I, 5% at Level II, and 3% at Level III. In NESARC III, 28% abstained and 33% binged (20%, 8%, and 5% at Levels I, II, and III; all increases significant, p<0.001). In NESARC I, 15% of Level I and II and 18% of Level III bingers drank those amounts at least weekly, and 36%, 37%, and 35% monthly or more. NESARC III binge frequency declined slightly, with 12% of Level I and 16% of Levels II and III bingers consuming these amounts weekly, and 31%, 34%, and 35%, respectively, monthly or more (all differences, p<0.01).
      No single demographic group accounted for most of binge Level I, II, and III increases from NESARC I-III. In some subgroups, changes were not significant. Increases were significant in all age groups except those aged 18–20 years. Both sexes, each racial/ethnic group, employed/unemployed, married, single, separated, and divorced people, those with and without children, college students, all income groups, smokers, nonsmokers, and drug and non-drug users all experienced significant increases (Appendix Tables 2A−C, available online). Multinomial logistic regression analyses adjusting for these characteristics indicated binging increased significantly from NESARC I-III (OR=1.5, 95% CI=1.43, 1.67; Wald F, p<0.0001).
      Appendix Table 1 (available online) provides characteristics of NESARC III respondents significantly (p<0.001) associated with past-year binge levels.
      Covariates identified in the Statistical Analysis section were entered into a logistic regression analysis with Level I, II, and III binging as the dependent variable. People aged 21–25 years, men, high school educated or less, childless, with earlier ages of first alcohol intoxication and drug use, past-year smoking and drug use, and, most strongly, AUD experiences all significantly, independently predicted higher-level binges. Blacks, Asians, graduate school educated, and homemakers or retired individuals were less likely to binge at these levels (Table 2).
      Table 2Significant Predictors of Drinking at Levels I, II, and III among NESARC III Drinkers
      Outcome binge level is a four-level ordinal variable. Multinomial Logistic Regression with Proportional-Odds Cumulative Logit Model is fitted controlling for variables listed in the table plus non-significant variables, including college enrollment, family income, health status, past-year felt depressed, past-year moved, past-year lost job, lifetime drug use, and age of first drink.
      (n=25,110)
      PredictorsAOR (95% CI)p-value
      Based on Wald F-tests, all predictors are significant at p<0.01.
      Age<0.0001
       18–20 years0.7 (0.5, 0.8)
       21–25 years1.0 (1.0, 1.0)
       26–30 years0.9 (0.8,1.1)
       31–35 years0.8 (0.7, 0.9)
       36–40 years0.7 (0.6, 0.8)
       41–50 years0.6 (0.5, 0.7)
       51–60 years0.3 (0.3, 0.4)
       ≥61 years0.2 (0.3, 0.3)
      Sex<0.0001
       Male1.4 (1.3, 1.5)
       Female1.0 (1.0, 1.0)
      Race/ethnicity<0.0001
       Black0.6 (0.5, 0.6)
       Native American1.2 (0.9, 1.5)
       Asian0.7 (0.6, 0.8)
       Hispanic1.2 (1.1, 1.4)
       White1.0 (1.0, 1.0)
      Marital status0.0025
       Never married1.1 (0.95, 1.2)
       Divorced/widowed/separated1.1 (1.01, 1.3)
       Married1.0 (1.0, 1.0)
      Employment status0.0009
       Unemployed0.9 (0.8, 1.0)
       Retired/homemaker0.8 (0.8, 0.9)
       Employed1.0 (1.0, 1.0)
       Student1.1 (0.9, 1.3)
       Other0.9 (0.7, 1.1)
      Highest education<0.0001
       <High school1.4 (1.3, 1.6)
       High school1.1 (1.1, 1.2)
       College1.0 (1.0, 1.0)
       Graduate degree0.8 (0.7, 0.9)
      Number of children0.0004
       10.8 (0.8, 0.9)
       20.9 (0.8, 0.9)
       30.8 (0.7, 0.96)
       ≥40.8 (0.6, 1.04)
       01 (1.0, 1.0)
      Lifetime smoking<0.0001
       Current/past year1.8 (1.6, 1.9)
       Former1.1 (1.03, 1.3)
       Never1 (1.0, 1.0)
      Age first used drugs<0.0001
       ≤13 years1.4 (1.2, 1.7)
       14–17 years1.3 (1.2, 1.6)
       18–20 years1.3 (1.2, 1.5)
       21–25 years1.2 (1.04, 1.5)
       26–30 years1.1 (0.8, 1.4)
       ≥31 years0.9 (0.7, 1.3)
       Never1 (1.0, 1.0)
      Age first drunk<0.0001
       ≤13 years5.9 (4.7, 7.5)
       14–17 years7.3 (6.4, 8.3)
       18–20 years6.2 (5.4, 7.1)
       21–25 years4.8 (4.2, 5.5)
       26–30 years4.5 (3.6, 5.6)
       ≥31 years4.9 (3.9, 6.1)
       Never1 (1.0, 1.0)
      Past-year DSM-V AUD
      Pairwise comparisons also shows that the odds of higher binge level significantly differ between all levels of AUD, p<0.05.AUD, alcohol use disorder; NESARC, National Epidemiologic Survey of Alcohol and Related Conditions. AUD, alcohol use disorder; NESARC, National Epidemiological Survey of Alcohol and Related Conditions.
      <0.0001
       Severe (≥6 symptoms)17.8 (15.2, 20.8)
       Moderate (4–5 symptoms)10 (8.5, 11.1)
       Mild (2–3 symptoms)6 (5.4, 6.6)
       No AUD1 (1.0, 1.0)
      Note: Boldface indicates statistical significance (p<0.05).
      a Outcome binge level is a four-level ordinal variable. Multinomial Logistic Regression with Proportional-Odds Cumulative Logit Model is fitted controlling for variables listed in the table plus non-significant variables, including college enrollment, family income, health status, past-year felt depressed, past-year moved, past-year lost job, lifetime drug use, and age of first drink.
      b Based on Wald F-tests, all predictors are significant at p<0.01.
      c Pairwise comparisons also shows that the odds of higher binge level significantly differ between all levels of AUD, p<0.05.AUD, alcohol use disorder; NESARC, National Epidemiologic Survey of Alcohol and Related Conditions.AUD, alcohol use disorder; NESARC, National Epidemiological Survey of Alcohol and Related Conditions.
      Logistic regression analyses adjusting for variables outlined in the Statistical Analysis section indicated the strongest predictor of meeting past-year AUD criteria was binging (Level III: OR=52.5, 95% CI=43.7, 63.2; Level II: OR=20.4, 95% CI=17.6, 23.6; Level I: OR=6.9, 95% CI=6.0, 8.0; non-binge drinkers: OR=1.0, 95% CI=1.0, 1.0) (Appendix Table 3, available online, lists all significant predictors). In turn, the strongest independent predictor of receiving alcohol-related counseling, ED visits, or hospitalizations was meeting AUD criteria (Table 3). Overall, 19% of drinkers met past-year AUD criteria. More (45%) reported one or more past-year binge episode. Among past-year bingers, 38% met past-year AUD criteria (24% at Level I, 51% Level II, and 74% Level III). Binging Levels I, II, and III each independently predicted past-year, alcohol-related ED visits (Table 3). Further, Level III and II versus Level I bingers significantly more often reported alcohol-related ED visits.
      Table 3Seeking Counseling, ED Visits, and Being Hospitalized for an Alcohol-Related Problem Predicted by Binge Levels I, II, III and DSM-V AUD Level
      Covariates used in modeling are age, sex, race/ethnicity, marital status, employment status, college enrollment, highest education, family income, health status, past-year felt depressed, past-year moved, past-year lost job, number of children, lifetime smoking, lifetime drug use, age first used drug, age first drink, age first drunk, binge level, past-year DSM-V alcohol use disorder, and frequency of peak binge drinking level.
      Binge/AUD levelsAlcohol counseling,
      Three outcome variables are dichotomous. Separate logistic regression models were fit for the three outcome measures respectively.
      OR (95% CI) (n=25,482)
      p-value for pairwise comparisonED visit,
      Three outcome variables are dichotomous. Separate logistic regression models were fit for the three outcome measures respectively.
      OR (95% CI) (n=25,545)
      p-value for pairwise comparisonHospitalization,
      Three outcome variables are dichotomous. Separate logistic regression models were fit for the three outcome measures respectively.
      OR (95% CI) (n=25,463)
      p-value for pairwise comparison
      Past-year DSM-V alcohol use disorder
       Severe (≥6 symptoms)18.5 (13.2,25.9)36.6 (9.5, 140.6)27.9 (8.4, 92.6)
       Moderate (4−5 symptoms)4.7 (3.2, 6.9)2.8 (0.5, 15.1)4.3 (1.1, 17.0)
       Mild (2−3 symptoms)3.1 (2.1, 4.6)2.2 (0.4, 11.2)1.6 (0.4, 7.3)
       No AUD1.0 (1.0, 1.0)1.0 (1.0, 1.0)1.0 (1.0, 1.0)
      Binge
       Level III
      Corresponding predictor is not statistically significant at p=0.05, comparison between levels were not performed, and the covariate was dropped from the final model. AUD, alcohol use disorder; ED, emergency department.
      9.9 (1.6, 61.7)
      Corresponding predictor is not statistically significant at p=0.05, comparison between levels were not performed, and the covariate was dropped from the final model. AUD, alcohol use disorder; ED, emergency department.
       Level II
      Corresponding predictor is not statistically significant at p=0.05, comparison between levels were not performed, and the covariate was dropped from the final model. AUD, alcohol use disorder; ED, emergency department.
      13.1 (2.1, 82.5)
      Corresponding predictor is not statistically significant at p=0.05, comparison between levels were not performed, and the covariate was dropped from the final model. AUD, alcohol use disorder; ED, emergency department.
       Level I
      Corresponding predictor is not statistically significant at p=0.05, comparison between levels were not performed, and the covariate was dropped from the final model. AUD, alcohol use disorder; ED, emergency department.
      5.1 (1.0, 26.9)
      Corresponding predictor is not statistically significant at p=0.05, comparison between levels were not performed, and the covariate was dropped from the final model. AUD, alcohol use disorder; ED, emergency department.
       No binge
      Corresponding predictor is not statistically significant at p=0.05, comparison between levels were not performed, and the covariate was dropped from the final model. AUD, alcohol use disorder; ED, emergency department.
      1.0 (1.0, 1.0)
      Past-year DSM-V alcohol use disorder
       Severe vs moderate3.9 (2.8, 5.5)<0.000113.1 (4.0, 43)0.00016.5 (2.5, 17.7)0.0002
       Severe vs mild6.0 (4.2, 8.5)<0.000116.6 (6.1, 47)<0.000117.4 (5.5, 55.2)<0.0001
       Moderate vs mild1.5 (1.03, 2.2)0.04131.3 (0.3, 5.6)0.90162.7 (0.6, 11.7)0.1990
      Binge
       Level III vs II
      Corresponding predictor is not statistically significant at p=0.05, comparison between levels were not performed, and the covariate was dropped from the final model. AUD, alcohol use disorder; ED, emergency department.
      Corresponding predictor is not statistically significant at p=0.05, comparison between levels were not performed, and the covariate was dropped from the final model. AUD, alcohol use disorder; ED, emergency department.
      0.8 (0.4, 1.5)0.7800
      Corresponding predictor is not statistically significant at p=0.05, comparison between levels were not performed, and the covariate was dropped from the final model. AUD, alcohol use disorder; ED, emergency department.
      Corresponding predictor is not statistically significant at p=0.05, comparison between levels were not performed, and the covariate was dropped from the final model. AUD, alcohol use disorder; ED, emergency department.
       Level III vs I
      Corresponding predictor is not statistically significant at p=0.05, comparison between levels were not performed, and the covariate was dropped from the final model. AUD, alcohol use disorder; ED, emergency department.
      Corresponding predictor is not statistically significant at p=0.05, comparison between levels were not performed, and the covariate was dropped from the final model. AUD, alcohol use disorder; ED, emergency department.
      1.9 (1.03, 5.0)0.0342
      Corresponding predictor is not statistically significant at p=0.05, comparison between levels were not performed, and the covariate was dropped from the final model. AUD, alcohol use disorder; ED, emergency department.
      Corresponding predictor is not statistically significant at p=0.05, comparison between levels were not performed, and the covariate was dropped from the final model. AUD, alcohol use disorder; ED, emergency department.
       Level II vs I
      Corresponding predictor is not statistically significant at p=0.05, comparison between levels were not performed, and the covariate was dropped from the final model. AUD, alcohol use disorder; ED, emergency department.
      Corresponding predictor is not statistically significant at p=0.05, comparison between levels were not performed, and the covariate was dropped from the final model. AUD, alcohol use disorder; ED, emergency department.
      2.6 (1.03, 6.4)0.0143
      Corresponding predictor is not statistically significant at p=0.05, comparison between levels were not performed, and the covariate was dropped from the final model. AUD, alcohol use disorder; ED, emergency department.
      Corresponding predictor is not statistically significant at p=0.05, comparison between levels were not performed, and the covariate was dropped from the final model. AUD, alcohol use disorder; ED, emergency department.
      Note: Boldface indicates statistical significance (p<0.05).
      a Covariates used in modeling are age, sex, race/ethnicity, marital status, employment status, college enrollment, highest education, family income, health status, past-year felt depressed, past-year moved, past-year lost job, number of children, lifetime smoking, lifetime drug use, age first used drug, age first drink, age first drunk, binge level, past-year DSM-V alcohol use disorder, and frequency of peak binge drinking level.
      b Three outcome variables are dichotomous. Separate logistic regression models were fit for the three outcome measures respectively.
      c Corresponding predictor is not statistically significant at p=0.05, comparison between levels were not performed, and the covariate was dropped from the final model.AUD, alcohol use disorder; ED, emergency department.
      Given the strong correlation between AUD and Levels I, II, and III binging, when repeat regressions removed AUD as a covariate, compared with non-binge drinking, Levels I, II, and III binging most strongly, independently predicted past-year alcohol counseling (Level III: OR=7.4, 95% CI=4.7, 11.5; Level II: OR=4.0, 95% CI=2.6, 6.2; and Level I: OR=2.1, 95% CI=1.4, 3.1), ED visits (Level III: OR=93.1, 95% CI=23.4, 371.5; Level II: OR=70.2, 95% CI=18.2, 271.0; and Level I: OR=13.1, 95% CI=3.1, 55.0), or hospitalizations (Level III: OR=7.6, 95% CI=2.2, 25.8; Level II: OR=2.1, 95% CI=0.5, 8.1; Level I: OR=1.5, 95% CI=0.4, 5.5).
      Table 4 displays drinking consequences significantly associated with Level I, II, and III binging and AUD after adjusting for all selected characteristics. Adjusting for AUD, Level I, II, and III binging compared with non-binge drinking all predicted higher odds of each consequence except driving under the influence accidents. Odds of having alcohol-related consequences increased with AUD severity. Further, Level III versus I bingers significantly more often drove after any drinking and drinking too much and, after drinking, were in physical fights, injured, arrested/detained, or had a legal problem. Level II versus I bingers were also significantly more likely to drive after drinking and drinking too much. In repeat regressions without AUD as a covariate, Level III binging, followed in order by Level II and I binging, most strongly predicted driving while drinking, driving under the influence accidents after drinking too much, and experiencing physical fights, injuries, arrests/detentions, and other legal problems after drinking (Appendix Table 4, available online, lists the results of this regression).
      Table 4Six Adverse Consequences Predicted by Binge Levels I, II, III and DSM-V AUD Level
      Covariates used in modeling are age, sex, race/ethnicity, marital status, employment status, college enrollment, highest education, family income, health status, past-year felt depressed, past-year moved, past-year lost job, number of children, lifetime smoking, lifetime drug use, age first used drug, age first drink, age first drunk, binge level, past-year DSM-V AUD, and frequency of peak binge drinking level.
      Adverse consequencesDrive after drinking
      Six outcome variables are dichotomous. Separate logistic regression models were fit for the six outcome measures, respectively.
      (n=25,213)
      Drive after drinking too much
      Six outcome variables are dichotomous. Separate logistic regression models were fit for the six outcome measures, respectively.
      (n=25,147)
      Driver in traffic crash
      Six outcome variables are dichotomous. Separate logistic regression models were fit for the six outcome measures, respectively.
      (n=25,456)
      In physical fight
      Six outcome variables are dichotomous. Separate logistic regression models were fit for the six outcome measures, respectively.
      (n=25,447)
      Injured
      Six outcome variables are dichotomous. Separate logistic regression models were fit for the six outcome measures, respectively.
      (n=25,469)
      Arrested or legal problem
      Six outcome variables are dichotomous. Separate logistic regression models were fit for the six outcome measures, respectively.
      (n=25,459)
      OR (95% CI)p-value
      For pairwise comparison.
      OR (95% CI)p-value
      For pairwise comparison.
      OR (95% CI)p-value
      For pairwise comparison.
      OR (95% CI)p-value
      For pairwise comparison.
      OR (95% CI)p-value
      For pairwise comparison.
      OR (95% CI)p-value
      For pairwise comparison.
      Past-year DSM-V AUD
       Severe (≥6 symptoms)23.2 (18.8, 28.9)24.8 (18.2, 32.4)31.1 (18.1, 53.3)60.7 (35.5, 103.8)29.6 (19.4, 45.3)18.2 (8.7, 38.1)
       Moderate (4-5 symptoms)13.1 (10.4, 16.5)14.0 (10.5, 18.6)12.6 (6.9, 22.9)23.4 (13.7, 40.0)8.2 (4.9, 13.8)5.0 (2.4, 10.8)
       Mild (2-3 symptoms)7.7 (6.4, 9.2)6.8 (5.2, 8.8)6.7 (4.0, 11.2)11.4 (6.7, 19.4)5.3 (3.2, 8.8)3.4 (1.4, 7.9)
       No AUD1.0 (1.0, 1.0)1.0 (1.0, 1.0)1.0 (1.0, 1.0)1.0 (1.0, 1.0)1.0 (1.0, 1.0)1.0 (1.0, 1.0)
      Binge level
       III3.6 (2.8, 4.7)5.2 (3.3, 8.1)
      Corresponding predictor is not statistically significant at p=0.05, comparison between levels were not performed, and the covariate was dropped from the final model.
      2.9 (1.7, 5.0)7.0 (3.2, 15.3)6.1 (2.4, 15.9)
       II3.4 (2.7, 4.3)5.0 (3.2, 7.7)
      Corresponding predictor is not statistically significant at p=0.05, comparison between levels were not performed, and the covariate was dropped from the final model.
      1.9 (1.1, 3.3)4.3 (2.0, 9.2)3.6 (1.3, 10.2)
       I2.7 (2.2, 3.5)4.0 (2.7, 5.8)
      Corresponding predictor is not statistically significant at p=0.05, comparison between levels were not performed, and the covariate was dropped from the final model.
      1.8 (1.1, 2.9)3.8 (1.8, 7.9)3.1 (1.2, 8.0)
       No binge1.0 (1.0, 1.0)1.0 (1.0, 1.0)
      Corresponding predictor is not statistically significant at p=0.05, comparison between levels were not performed, and the covariate was dropped from the final model.
      1.0 (1.0, 1.0)1.0 (1.0, 1.0)1.0 (1.0, 1.0)
      Past-year DSM-V AUD
       Severe vs moderate1.8 (1.4, 2.2)<0.0001
      Simultaneously significant at p=0.05 with Bonferroni adjustment. AUD, alcohol use disorder.
      1.8 (1.4, 2.2)<0.0001
      Simultaneously significant at p=0.05 with Bonferroni adjustment. AUD, alcohol use disorder.
      2.5 (1.6, 3.8)0.0001
      Simultaneously significant at p=0.05 with Bonferroni adjustment. AUD, alcohol use disorder.
      2.6 (2.0, 3.3)<0.0001
      Simultaneously significant at p=0.05 with Bonferroni adjustment. AUD, alcohol use disorder.
      3.6 (2.6, 4.9)<0.0001
      Simultaneously significant at p=0.05 with Bonferroni adjustment. AUD, alcohol use disorder.
      3.6 (2.3, 5.6)<0.0001
       Severe vs mild3.0 (2.5, 3.7)<0.00013.6 (2.9, 4.4)<0.00014.6 (3.0, 7.3)<0.00015.3 (3.8, 7.4)<0.00015.6 (4.0, 7.8)<0.00015.3 (3.3, 8.7)<0.0001
       Moderate vs mild1.7 (1.4, 2.1)<0.00012.1 (1.7, 2.5)<0.00011.9 (1.2, 3.2)0.02032.1 (1.5, 2.9)<0.00011.5 (1.05, 2.3)0.02461.5 (0.8, 2.6)0.2259
      Binge level
       III vs II1.1 (0.9, 1.3)0.51721.0 (0.8, 1.3)0.7437
      Corresponding predictor is not statistically significant at p=0.05, comparison between levels were not performed, and the covariate was dropped from the final model.
      Corresponding predictor is not statistically significant at p=0.05, comparison between levels were not performed, and the covariate was dropped from the final model.
      1.5 (1.1, 2.1)0.00471.6 (1.2, 2.2)0.00191.7 (1.1, 2.6)0.0012
       III vs I1.3 (1.1, 1.6)0.03081.3 (1.01, 1.7)0.0407
      Corresponding predictor is not statistically significant at p=0.05, comparison between levels were not performed, and the covariate was dropped from the final model.
      Corresponding predictor is not statistically significant at p=0.05, comparison between levels were not performed, and the covariate was dropped from the final model.
      1.6 (1.6, 1.1)0.00451.8 (1.3, 2.5)0.00021.9 (1.3, 3.0)<0.0001
       II vs I1.2 (1.1, 1.4)0.04511.2 (1.02, 1.6)0.0409
      Corresponding predictor is not statistically significant at p=0.05, comparison between levels were not performed, and the covariate was dropped from the final model.
      Corresponding predictor is not statistically significant at p=0.05, comparison between levels were not performed, and the covariate was dropped from the final model.
      1.0 (0.7, 1.4)0.79871.1 (0.8, 1.6)0.46661.2 (0.7, 2.2)0.5593
      Note: Boldface indicates statistical significance (p<0.05).
      a Covariates used in modeling are age, sex, race/ethnicity, marital status, employment status, college enrollment, highest education, family income, health status, past-year felt depressed, past-year moved, past-year lost job, number of children, lifetime smoking, lifetime drug use, age first used drug, age first drink, age first drunk, binge level, past-year DSM-V AUD, and frequency of peak binge drinking level.
      b Six outcome variables are dichotomous. Separate logistic regression models were fit for the six outcome measures, respectively.
      c For pairwise comparison.
      d Corresponding predictor is not statistically significant at p=0.05, comparison between levels were not performed, and the covariate was dropped from the final model.
      e Simultaneously significant at p=0.05 with Bonferroni adjustment.AUD, alcohol use disorder.
      Numerous other significant covariates predicted negative consequences after drinking, but their ORs were uniformly lower than odds predicted by Level I, II, and III binging (data available upon request).

      Discussion

      High-intensity binge drinking is prevalent in the U.S. In 2012–2013, 33% of those aged ≥18 years binged (20% peaking at Level I, 8% Level II, and 5% Level III). Approximately 12% who binged at Level I and 16% at Levels II and III drank at those levels at least weekly and approximately one third monthly or more. Consistent with other reports, Level I, II, and III binging was particularly common among young adults. MTF data
      • Johnston L.D.
      • O’Malley P.M.
      • Bachman J.G.
      • Schulenberg J.E.
      • Miech R.A.
      Monitoring the Future National Survey Results on Drug Use, 1975−2015.
      suggest 13% of college students consumed ten or more drinks in a row at least once in the prior 2 weeks and 5% consumed ≥15 drinks. Similarly, among non-college respondents, 12% had ten or more and 5% had ≥15. Another study
      • Blazer D.G.
      • Wu L.T.
      The epidemiology of at-risk and binge drinking among middle-aged and elderly community adults: National Survey on Drug Use and Health.
      reported 12.4% of adults aged 25–26 years consumed ten or more drinks once in the past 2 weeks.
      Level I, II, and III binging increased significantly from NESARC I-III, though binge frequency declined slightly. These increases were significant in most demographic subgroups, except those aged 18–20 years (below the legal drinking age). Previous studies of high school students, college students, and other young adults either found no changes or past-decade declines in high peak consumption levels.
      • Johnston L.D.
      • O’Malley P.M.
      • Bachman J.G.
      • Schulenberg J.E.
      • Miech R.A.
      Monitoring the Future National Survey Results on Drug Use, 1975−2015.
      • Terry-McElrath Y.M.
      • Patrick M.E.
      Intoxication and binge and high-intensity drinking among U.S. young adults in their mid-twenties.
      MTF data from 2005–2010 and 2011–2015 suggest declines in young men’s past 2-week consumption of ten or more (23.8% to 18.3%) and ≥15 (10% to 7.8%) drinks. No significant changes were noted for women (6.8% to 6.9% for ten or more and 1.7% to 1.7% for ≥15). By contrast, NESARC I−III comparisons suggests an increase in past-year ten or more drinking for men aged 18–29 years from 33.9% to 38% and eight or more drinking for women from 14.7% to 22.1%.
      • Grant B.F.
      • Goldstein R.B.
      • Saha T.D.
      • et al.
      Epidemiology of DSM-5 Alcohol Use Disorder: results from the National Epidemiologic Survey on Alcohol and Related Conditions III.
      Possible differences in the patterns observed in MTF and NESARC resulted from MTF asking about high peak drinking in the previous 2 weeks, whereas NESARC asked about the past year. The increases in past-year Level I, II, and III binging among people aged ≥18 years in NESARC is consistent with U.S. per capita consumption increases over the past 15 years,
      • Haughwout S.P.
      • LaVallee R.A.
      • Castle I.-J.P.
      Apparent per capita alcohol consumption: national, state, and regional trends, 1977–2014.
      and recent national increases in alcohol-related ED admissions.
      • Mullins P.M.
      • Mazer-Amirshahi M.
      • Pines J.M.
      Alcohol-related visits to U.S. emergency departments, 20012011.
      Binging at levels beyond the threshold raises concern because it increases risk of blackouts, overdoses, and other serious consequences. Also, high peak bingers more often consume drugs, which combined with alcohol, further heighten risk.
      • White R.
      • Hingson R.
      • Pan I.-J.
      • Yi H.Y.
      Hospitalizations for alcohol and drug overdoses in young adults aged 1824 in the United States, 19992008: results from the Nationwide Inpatient Sample.
      Level III bingers were more likely than Level I bingers to drive after drinking too much, and, after drinking, experience physical fights, injuries, arrests/detentions, and other legal problems. Level II versus I bingers were more likely to drive after any drinking and drinking too much.
      Level I, II, and III binging heightened the likelihood of meeting AUD criteria. Meeting AUD criteria most strongly predicted past-year alcohol-related counseling, ED visits, or hospitalizations. Level I, II, and III binging further predicted alcohol-related ED visits (Levels II and III versus Level I significantly more). Alcohol screening and brief counseling interventions for adult primary care are recommended.
      • Solberg L.I.
      • Maciosek M.V.
      • Edwards N.M.
      Primary care intervention to reduce alcohol misuseranking its health impact and cost effectiveness.
      • Jonas D.E.
      • Garbutt J.C.
      • Amick H.R.
      • et al.
      Behavioral counseling after screening for alcohol misuse in primary care: a systematic review and meta-analysis for the U.S. Preventive Services Task Force.
      • Moyer V.A.
      Preventive Services Task Force
      Screening and behavioral counseling interventions in primary care to reduce alcohol misuse: U.S. Preventive Services Task Force recommendation statement.
      Meeting AUD criteria triggers clinician willingness to treat or recommend alcohol treatment. However, many more drinkers binge than meet AUD criteria. Because, independent of meeting AUD criteria, Level I, II, and III binging predicts many behaviors placing individuals at risk of harming themselves or others, healthcare providers should consider questions about peak drinking levels when deciding whether further assessment or interventions are warranted.
      Consistent with the prevention paradox,
      • Rose G.
      Strategy of prevention: lessons from cardiovascular disease.
      though higher percentages of Level III bingers experienced negative alcohol-related consequences, many more drinkers binged at Levels I and II. Consequently, Level I and II bingers experienced more occurrences of each explored negative alcohol-related consequence than Level III bingers. For example, 12% of Level III bingers experienced past-year injuries after drinking, compared with 5% and 2% of Level II and I bingers. However, 269 Level I and II bingers were injured compared to 221 Level III bingers. Level I and II bingers also accounted for more counseling episodes, ED visits, and hospitalizations than Level III bingers. Nonetheless, Level II and III bingers combined accounted for more negative consequences and counseling episodes than more numerous Level I bingers (data available upon request).

      Limitations

      Reliance on self-report is a major limitation of this study. Some may be reluctant to report peak binging levels, and drinking-induced blackouts could reduce recall. Second, the survey did not ask whether alcohol-related consequences (e.g., traffic injuries) occurred on peak drinking days or the frequency of these consequences. Future studies should do so. Third, given the large sample size, often ORs close to 1 were significant but not very meaningful. Fourth, small data collection changes occurred between NESARC I and III, and NESARC III response rates were lower. NESARC III but not NESARC I college students were interviewed at home, possibly reducing reported binging at the three levels.
      Despite survey design and data collection method differences, both samples are nationally representative. Some of the differences between the samples reflect population change. Post-survey weighting adjustment partially compensated for different non-response rates. However, differential non-response between the two surveys in some population and drinking level subgroups could have affected results. Finally, this analysis focused on national, cross-sectional surveys and acute consequences. Longitudinal studies of predictors and acute and chronic disease consequences of higher level binging are needed.

      Conclusions

      Findings suggest Level I, II, and III binging is prevalent and perhaps increasing. Level III versus I binging was significantly associated with more adverse health and social consequences, like alcohol-related ED visits, AUD, and, after drinking, experiencing fights, injuries, traffic crashes, arrests/detentions, and other legal problems. Level II versus I binging was associated with alcohol-related ED visits and driving after any drinking and drinking too much. Focusing simply on overall binge drinking prevalence obscures important information about higher, more dangerous Level II and III versus level I binging. There is a clear need to identify interventions to reduce such high-intensity drinking and related harms. Research is needed to determine whether questions about peak consumption levels are valuable in screening instruments and assess gender-specific risk factors for engaging in, and suffering consequences from, drinking at high peak levels.

      Acknowledgments

      The research presented in this paper is that of the authors and does not reflect the official policy of the National Institute on Alcohol Abuse and Alcoholism. Dr. Ralph Hingson is responsible for the design and conduct of the research; data analysis and interpretation; the drafting of the abstract, paper, and figures; and the coordination of the completion of the article. Dr. Wenxing Zha is responsible for drafting the methods section of the article, conduct of the statistical analyses, and interpretation of the results. Dr. Aaron White is responsible for aspects of the design of the research, the conception of some of the statistical analyses, the drafting of sections of the article, and overall editing of the article. Drs. Timothy Heeren, Boston University School of Public Health, and I-Jen Castle, CSR, Incorporated, offered useful recommendations during our revisions.
      No financial disclosures were reported by the authors of this paper.

      Appendix A. Supplementary material

      References

        • Dawson D.A.
        • Hingson R.W.
        • Grant B.F.
        Epidemiology of alcohol use, abuse and dependence.
        in: Tsuang M. Tohen M. Jones P. Textbook in Psychiatric Epidemiology. 3rd ed. John Wiley & Sons, Chichester, United Kingdom2011: 361-379https://doi.org/10.1002/9780470976739.ch21
        • National Institute on Alcohol Abuse and Alcoholism
        Alcohol and development in youth: a multi-disciplinary overview.
        Alcohol Res Health. 2004/2005; 28: 175
        • Patrick M.E.
        A call for research on high-intensity alcohol use.
        Alcohol Clin Exp Res. 2016; 40: 256-259https://doi.org/10.1111/acer.12945
        • Pearson M.R.
        • Kirouac M.
        • Witkiewitz K.
        Questioning the validity of the 4+/5+ binge or heavy drinking criterion in college and clinical populations.
        Addiction. 2016; 111: 1720-1726https://doi.org/10.1111/add.13210
        • White A.M.
        • Kraus C.L.
        • Swartzwelder H.S.
        Many college freshmen drink at levels way beyond the binge threshold.
        Alcohol Clin Exp Res. 2006; 30: 1006-1010https://doi.org/10.1111/j.1530-0277.2006.00122.x
        • Hilton M.
        Demographic characteristics and the frequency of heavy drinking as predictors of self-reported drinking problems.
        Br J Addict. 1987; 82: 913-925https://doi.org/10.1111/j.1360-0443.1987.tb03912.x
        • Hingson R.W.
        • Wenxing Z.
        Age of drinking onset, alcohol use disorders, frequent heavy drinking, and unintentionally injuring oneself and others after drinking.
        Pediatrics. 2009; 123: 1477-1484https://doi.org/10.1542/peds.2008-2176
        • Patrick M.E.
        • Schulenberg J.E.
        • Martz M.E.
        • et al.
        Extreme binge drinking among 12th-grade students in the United States: prevalence and predictors.
        JAMA Pediatr. 2013; 167: 1019-1025https://doi.org/10.1001/jamapediatrics.2013.2392
        • Hingson R.
        • White A.
        Trends in extreme binge drinking among U.S. high school seniors.
        JAMA Pediatr. 2013; 167: 996-998https://doi.org/10.1001/jamapediatrics.2013.3083
        • Perry P.J.
        • Argo T.R.
        • Barnett M.J.
        • et al.
        The association of alcohol-induced blackouts and grayouts to blood alcohol concentrations.
        J Forensic Sci. 2006; 51: 896-899https://doi.org/10.1111/j.1556-4029.2006.00161.x
        • Jones A.W.
        • Holmgren P.
        Comparison of blood-ethanol concentration in deaths attributed to acute alcohol poisoning and chronic alcoholism.
        J Forensic Sci. 2003; 48: 874-879https://doi.org/10.1520/JFS2002420
        • Grant B.F.
        • Goldstein R.B.
        • Saha T.D.
        • et al.
        Epidemiology of DSM-5 Alcohol Use Disorder: results from the National Epidemiologic Survey on Alcohol and Related Conditions III.
        JAMA Psychiatry. 2015; 72: 757-766https://doi.org/10.1001/jamapsychiatry.2015.0584
        • Dawson D.A.
        • Goldstein R.B.
        • Saha T.D.
        • Grant B.F.
        Changes in alcohol consumption: United States, 2001-2002 to 2012-2013.
        Drug Alcohol Depend. 2015; 148: 56-61https://doi.org/10.1016/j.drugalcdep.2014.12.016
        • Jessor R.
        Problem-behavior theory, psychosocial development, and adolescent problem drinking.
        Br J Addict. 1987; 82: 331-342https://doi.org/10.1111/j.1360-0443.1987.tb01490.x
        • Esser M.B.
        • Hedden S.L.
        • Kanny D.
        • Brewer R.D.
        • Gfroerer J.C.
        • Naimi T.S.
        Prevalence of alcohol dependence among prevalence of alcohol dependence among U.S. adult drinkers adult drinkers, 2009–2011.
        Prev Chronic Dis. 2014; 11: 140329https://doi.org/10.5888/pcd11.140329
        • Grant B.F.
        • Kaplan K.D.
        • Shepard J.
        • Moore T.
        Source and Accuracy Statement for Wave 1 of the 2001–2002 National Epidemiologic Survey on Alcohol and Related Conditions.
        National Institute on Alcohol Abuse and Alcoholism, Bethesda, MD2003
        • Grant B.F.
        • Amsbary M.
        • Chu A.
        • et al.
        Source and Accuracy Statement: National Epidemiologic Survey on Alcohol and Related Conditions-III (NESARC-III).
        National Institute on Alcohol Abuse and Alcoholism, Bethesda, MD2014
        • Blazer D.G.
        • Wu L.T.
        The epidemiology of at-risk and binge drinking among middle-aged and elderly community adults: National Survey on Drug Use and Health.
        Am J Psychiatry. 2009; 166: 1162-1169https://doi.org/10.1176/appi.ajp.2009.09010016
        • CDC
        Vital signs: binge drinking prevalence, frequency, and intensity among adults—United States, 2010.
        MMWR Morb Mortal Wkly Rep. 2012; 61: 14-19
        • Cranford J.A.
        • McCabe S.E.
        • Boyd C.J.
        A new measure of binge drinking: prevalence and correlates in a probability sample of undergraduates.
        Alcohol Clin Exp Res. 2006; 30: 1896-1905https://doi.org/10.1111/j.1530-0277.2006.00234.x
        • Kann L.
        • McManus T.
        • Harris W.A.
        • et al.
        Youth Risk Behavior Surveillance—United States, 2015.
        MMWR Surveill Summ. 2016; 65: 1-174https://doi.org/10.15585/mmwr.ss6506a1
        • Lee H.K.
        • Han B.
        • Gfroerer J.C.
        Differences in the prevalence rates and correlates of alcohol use and binge alcohol use among five Asian American subpopulations.
        Addict Behav. 2013; 38: 1816-1823https://doi.org/10.1016/j.addbeh.2012.11.001
        • Naimi T.S.
        • Nelson D.E.
        • Brewer R.D.
        The intensity of binge alcohol consumption among U.S. adults.
        Am J Prev Med. 2010; 38: 201-207https://doi.org/10.1016/j.amepre.2009.09.039
        • Tucker J.S.
        • Orlando M.
        • Ellickson P.L.
        Patterns and correlates of binge drinking trajectories from early adolescence to young adulthood.
        Health Psychol. 2003; 22: 79-87https://doi.org/10.1037/0278-6133.22.1.79
      1. Larimer M, Arroyo J, editors Alcohol use among special populations (special issue). Alcohol Res 2016;38(1):1–140

        • Chen C.M.
        • Slater M.E.
        • Castle I.-J.P.
        • Grant B.F.
        Alcohol use and alcohol use disorders in the United States: main findings from the 2012–2013 National Epidemiologic Survey on Alcohol and Related Conditions-III (NESARC-III).
        (NIH Publication No. 16-AA-8020)U.S. Alcohol Epidemiologic Data Reference Manual. Volume 10. National Institute on Alcohol Abuse and Alcoholism, Bethesda, MD2016
      2. Substance Abuse and Mental Health Services Administration, Results from the 2013 National Survey on Drug Use and Health: summary of national findings, NSDUH Series H-48, HHS Publication No. (SMA) 14-4863.
        Substance Abuse and Mental Health Services Administration, Rockville, MD2014
        • Johnston L.D.
        • O’Malley P.M.
        • Bachman J.G.
        • Schulenberg J.E.
        • Miech R.A.
        Monitoring the Future National Survey Results on Drug Use, 1975−2015.
        Volume II, College Students and Adults Ages 1955. Volume II. Institute for Social Research, University of Michigan, Ann Arbor, MI2016
        • Terry-McElrath Y.M.
        • Patrick M.E.
        Intoxication and binge and high-intensity drinking among U.S. young adults in their mid-twenties.
        Subst Abus. 2016; 37: 597-605https://doi.org/10.1080/08897077.2016.1178681
        • Haughwout S.P.
        • LaVallee R.A.
        • Castle I.-J.P.
        Apparent per capita alcohol consumption: national, state, and regional trends, 1977–2014.
        Surveillance Report #104. National Institute on Alcohol Abuse and Alcoholism, Rockville, MD2016
        • Mullins P.M.
        • Mazer-Amirshahi M.
        • Pines J.M.
        Alcohol-related visits to U.S. emergency departments, 20012011.
        Alcohol Alcohol. 2016; https://doi.org/10.1093/alcalc/agw074
        • White R.
        • Hingson R.
        • Pan I.-J.
        • Yi H.Y.
        Hospitalizations for alcohol and drug overdoses in young adults aged 1824 in the United States, 19992008: results from the Nationwide Inpatient Sample.
        J Stud Alcohol Drugs. 2011; 72: 774-786https://doi.org/10.15288/jsad.2011.72.774
        • Solberg L.I.
        • Maciosek M.V.
        • Edwards N.M.
        Primary care intervention to reduce alcohol misuseranking its health impact and cost effectiveness.
        Am J Prev Med. 2008; 34: 143-152https://doi.org/10.1016/j.amepre.2007.09.035
        • Jonas D.E.
        • Garbutt J.C.
        • Amick H.R.
        • et al.
        Behavioral counseling after screening for alcohol misuse in primary care: a systematic review and meta-analysis for the U.S. Preventive Services Task Force.
        Ann Intern Med. 2012; 157: 645-654https://doi.org/10.7326/0003-4819-157-9-201211060-00544
        • Moyer V.A.
        • Preventive Services Task Force
        Screening and behavioral counseling interventions in primary care to reduce alcohol misuse: U.S. Preventive Services Task Force recommendation statement.
        Ann Intern Med. 2013; 159: 210-218https://doi.org/10.7326/0003-4819-159-3-201308060-00652
        • Rose G.
        Strategy of prevention: lessons from cardiovascular disease.
        Br Med J (Clin Res Ed). 1981; 282: 1847-1851https://doi.org/10.1136/bmj.282.6279.1847