Slutske WS. Alcohol Use Disorders Among US College Students and Their Non–College-Attending Peers. Arch Gen Psychiatry. 2005;62(3):321–327. doi:10.1001/archpsyc.62.3.321
Heavy/binge drinking among college students has become a major public health problem. There is consistent evidence suggesting that young adults in college are drinking more than their non–college-attending peers, but it is still not clear whether they are more likely to suffer from clinically significant alcohol use disorders.
To compare the prevalence of alcohol use disorders and alcohol use disorder symptoms in college-attending young adults with their non–college-attending peers within the same study in a large and representative US national sample.
Civilian, noninstitutionalized US population.
Young adults (n = 6352) from the 2001 National Household Survey on Drug Abuse (19-21 years of age, 51% female, 66% white, 14% African American, 14% Hispanic).
Main Outcome Measures
Lifetime, past-year, and past-month drinking, past-year and past-month weekly drinking, past-month weekly binge drinking, past-month daily drinking, typical quantity consumed in the past month, and past-year DSM-IV alcohol dependence and abuse diagnoses.
Eighteen percent of US college students (24% of men, 13% of women) suffered from clinically significant alcohol-related problems in the past year, compared with 15% of their non–college-attending peers (22% of men, 9% of women; overall odds ratio = 1.32). The association between past-year alcohol use disorder and college attendance was stronger among women (odds ratio = 1.70) than men (odds ratio = 1.14). College students were more likely to receive a diagnosis of DSM-IV alcohol abuse than their peers not attending college; despite the fact that those in college were drinking more, they were not more likely to receive a diagnosis of DSM-IV alcohol dependence.
College students suffer from some clinically significant consequences of their heavy/binge drinking, but they do not appear to be at greater risk than their non–college-attending peers for the more pervasive syndrome of problems that is characteristic of alcohol dependence.
Highly publicized reports have raised public awareness about the high rates of heavy/binge drinking on college campuses.1 The US Surgeon General and the US Department of Health and Human Services2,3 have identified binge drinking among college students as a major public health problem,3 and some experts4 have concluded that binge drinking is the most serious public health problem facing US colleges. A recent report commissioned by the National Institute on Alcohol Abuse and Alcoholism3 estimated that alcohol is involved in approximately 1400 student deaths, 500 000 injuries, 600 000 assaults, and 70 000 sexual assaults each year on college campuses.5
The phenomenon of college student drinking is not solely explained by the higher rates of alcohol involvement among young adults in general. Studies conducted since 1988, when a uniform national legal drinking age of 21 years was established in the United States, have tended to find that college students are more likely than their non–college-attending peers to be engaged in heavy/binge drinking.5- 7 A natural question one might ask is “Are college students also more likely to be alcoholic8 ?”
Few direct comparisons have been made of the prevalence of alcohol use disorders of college students with their non–college-attending peers within the same study. In a sample of 608 Missouri-born young adult women (identified for a population-based female twin study), there were no differences between the women in college and those not in college in the prevalence of either past-year alcohol abuse (6% vs 6%) or dependence (9% vs 7%).7 A report from the 2002 National Survey of Drug Use and Health stated that the rate of combined past-year alcohol abuse or dependence was higher among 22-year-old male and female full-time college students (25%) than non-students (20%).9 The scant evidence from these 2 studies is inconsistent concerning whether college students are more likely to have a diagnosis of alcohol abuse or dependence. It is surprising, given the relatively consistent evidence that college students are drinking more than their non–college-attending peers, that the results of at least one study suggest that they do not have higher rates of alcohol use disorders.
The purpose of this investigation was to compare the prevalence of alcohol use disorders and alcohol use disorder symptoms in college-attending young adults with their non–college-attending peers within the same study, in an attempt to reexamine the earlier paradoxical findings of Slutske et al7 in a larger and more representative US national sample, and to explore whether there may be differences between men and women in the association between college attendance and alcohol involvement.
The data for this study were obtained from the public use data set of the 2001 National Household Survey on Drug Abuse10 (NHSDA), an interview survey of 68 929 individuals drawn from the civilian, noninstitutionalized US population (including residents of group quarters, such as college dormitories). The participation rate for the interview was 73%. To protect the confidentiality of the participants, the public use file contained data from 55 561 individuals that were randomly sampled from the original data set. Other steps were also taken to ensure confidentiality, such as excluding particular variables from the public use data set and collapsing response categories for certain questions. The data collection procedures were approved by the Committee for the Protection of Human Subjects at the Research Triangle Institute.
Participants aged 19 to 21 years who completed the NHSDA interview were selected for this study. The choice of this age range was based on several considerations. Because the focus of the study was past-year alcohol use disorders, 18-year-old participants were excluded to minimize capturing any alcohol-related problems that occurred prior to college in the college-attending subsample. Those older than 21 years were excluded to minimize the inclusion of any college graduates in the non–college-attending subsample. Finally, in an attempt to replicate the results of Slutske et al,7 the same age range was included. There were 6352 respondents aged 19 to 21 years in the NHSDA survey who were identified as either currently attending college full- or part-time (n = 3184, 51%), or who were identified as not currently enrolled in school (3168, 49%). The mean age of the sample was 20 years (SE = 0.02). Participants in college were slightly but significantly younger than those who were not in college (in college, mean age = 19.9 years; not in college, mean age = 20.1 years; t = −5.63, P<.001). Forty-nine percent of the participants were male and 51% were female. Other demographic characteristics of the sample are summarized in Table 1.
Structured computer-assisted interviews for the NHSDA interview were conducted by a nationwide staff of approximately 900 field interviewers. Most of the interview, including all of the questions about alcohol involvement and alcohol use disorders, was done using audio computer-assisted self-interviews. Respondents listened to the questions read out loud through headphones while reading the question text on a computer screen. Respondents then keyed their numeric answers into the computer. Prior to the interviews, respondents were trained by the field interviewer in the use of the audio computer-assisted self-interview system, and the field interviewer remained in the room with the respondent throughout the interview. This system was used to provide respondents with privacy and to increase the likelihood that they would feel comfortable revealing sensitive information. A portion of the interview, including the questions about demographic characteristics, was administered as a computer-assisted personal interview, in which the field interviewer read the questions out loud from the computer screen and keyed in the responses. There were 223 (3.9%) interviews conducted in Spanish. Informed consent for participation was obtained from all respondents prior to the survey.
Seven different dichotomous indicators of frequency of alcohol use were examined: lifetime, past-year and past-month use of any alcohol, past-year and past-month drinking at least once a week (on average), past-month binge drinking at least once a week (on average), and past-month daily drinking. The past-year and past-month measures of weekly or daily drinking were derived from questions about the number of days in the past year (0-365) or the past month (0-30) that the respondent drank any alcohol. Drinking at least once a week in the past year was based on the respondent reporting consuming alcohol on at least 52 days of the year. Drinking at least once a week in the past month was based on the respondent reporting consuming alcohol on at least 4 days of the month. The past-month measure of weekly binge drinking was derived from a question about the number of days in the past month that the respondent drank at least 5 drinks on one occasion. Binge drinking at least once a week in the past month was based on the respondent reporting consuming at least 5 drinks of alcohol on one occasion on at least 4 days in the past month. Respondents who had consumed alcohol in the past month were also asked about the typical quantity they consumed on the days that they drank in the past month. This continuous indicator was log transformed to reduce the positive skew of the distribution of responses.
Symptoms of past-year DSM-IV11 alcohol dependence and abuse were assessed. Diagnoses of alcohol dependence required that at least 3 of 7 dependence symptoms occurred in the past year. Diagnoses of alcohol abuse required that at least 1 of 4 abuse symptoms occurred in the past year (in other words, a diagnosis of alcohol abuse only required experiencing a single symptom). Diagnoses of alcohol abuse were made only for individuals who did not meet the criteria for past-year alcohol dependence. A combined diagnosis of alcohol use disorder (DSM-IV alcohol dependence or abuse) was also examined. The estimates of DSM-IV alcohol dependence or abuse obtained in the present study (16.6% overall, 22.5% among men, and 11.0% among women) were very similar to those obtained among the 18- to 29-year-old participants in the National Longitudinal Alcohol Epidemiologic Survey (16.2% overall, 22.4% among men, and 10.1% among women) and the National Epidemiologic Survey on Alcohol and Related Conditions (15.9% overall, 22.1% among men, and 9.8% among women).12
All of the analyses were conducted using the survey estimation techniques in Stata Release 8.013 to account for the stratified sampling employed in the NHSDA and to incorporate sampling weights associated with each observation. The sampling weights adjusted the final sample so that it was demographically matched to the US census, and accounted for unequal participation rates of certain demographic groups (including college dormitory residents). All of the sample percentages presented are based on the weighted data, and therefore will not match percentages calculated from the unweighted sample sizes provided.
For categorical outcomes, the differences between those in college and not in college were tested with logistic regression, predicting the alcohol outcome from college attendance. For the continuous outcome (typical quantity consumed), differences were tested with ordinary least squares regression. Sex differences in the strength of the association between an alcohol outcome and college attendance was indicated by a significant interaction of college attendance by sex in the regression model. This method yields the same result as the Breslow-Day test for the homogeneity of the odds ratio, but has the benefit of allowing for the inclusion of control variables such as age in the analysis. Because of the well-documented differences between men and women in their levels and rates of drinking and prevalence of alcohol use disorders, the comparison results of college students with their non–college-attending peers were controlled for sex. Given the slight but significant difference in ages of the participants in college and not in college, all of the comparison results were also controlled for age.
College students were no more or less likely than their same-age peers not in college to be lifetime abstainers from alcohol, but they exceeded their non–college-attending peers on all of the other indicators of yearly, monthly, and weekly alcohol use, as well as weekly binge drinking (Table 2 and Table 3). Daily drinking, on the other hand, was more common among the young adults not in college.
Among young adults who drank any alcohol in the past month, the average amount of alcohol consumed on a day when they drank was 5.9 drinks (SE = 0.23; median = 5) for men and 3.8 drinks (SE = 0.12; median = 3) for women. There was not a significant difference between those in college and those not in college in the average number of drinks consumed, both prior to data transformation (t = -0.97, 1 df, P = .33) as well as after a log transformation of the scores (t = 0.36, 1 df, P = .72).
Formal tests of sex differences revealed that the associations between college attendance and any alcohol use in the past month (t = 4.27, 1 df, P<.001) and drinking at least once a week in the past year (t = 2.05, 1 df, P = .045) and in the past month (t = 2.19, 1 df, P = .033) were significantly stronger among women than men (Table 3). The associations of the remaining indicators of alcohol use with college attendance did not differ for men and women; there were no instances in which alcohol use was more strongly associated with college attendance among men than women.
College-attending young adults were significantly more likely to be diagnosed with an alcohol use disorder (dependence or abuse) than their non–college-attending peers (18.0% vs 15.1%; t = 3.25, 1 df, P = .002). However, there was not a significant difference between college-attending and non–college-attending young adults in the prevalence of past-year alcohol dependence (6.1% vs 6.6%; t = -0.25, 1 df, P = .80; Table 2 and Table 4). Although there was not a difference in the prevalence of alcohol dependence as a whole, young adults in college were significantly more likely than those not in college to report 2 of the 7 symptoms of alcohol dependence: tolerance and spending a great deal of time over a period of a month obtaining, using, or getting over the effects of alcohol. Withdrawal was one symptom of alcohol dependence that the young adults in college were significantly less likely to report than those not in college.
Young adults in college were significantly more likely to be diagnosed with alcohol abuse than their non–college-attending peers (11.9% vs 8.5%; t = 6.86, 1 df, P<.001), and they were also more likely to have experienced 2 of the 4 symptoms of alcohol abuse in the past year: having serious problems at home, work, or school that were caused by alcohol, and regularly using alcohol and then participating in an activity that might be physically dangerous.
Formal tests of sex differences revealed a significantly stronger association between college attendance and the combined diagnosis of alcohol dependence or alcohol abuse (t = 2.06, 1 df, P = .04) and the alcohol abuse symptom of having serious problems at home, work, or school that were caused by alcohol (t = 2.31, 1 df, P = .02) among women than men. The associations of the remaining diagnoses and symptoms with college attendance did not differ for men and women; there were no diagnoses or symptoms that were more strongly associated with college attendance among men than women.
Even if the heavy drinking at college contributes to the development of alcohol dependence, the problems may take several years to develop. Therefore, differences between college students and their non–college-attending peers may not become evident until later years in school. This possibility was evaluated by examining whether the prevalence of alcohol dependence was higher among college students in their later years in school compared with those in their first year. Between both college-attending men and women, the prevalence of alcohol dependence was nearly identical across the college years (men: t = 0.10, 1 df, P = .92; women: t = -0.18, 1 df, P = .86).
Failure to find a difference in the prevalence of alcohol dependence between those in college and those not in college may be due to the fact that the individuals suffering from alcohol problems are more likely to drop out of college, and therefore not remain in the college student subsample. Among the young adults who were not currently attending college, alcohol dependence was not more prevalent among the 239 men who had previously attended some college compared with the 1297 men with no previous college (t = -0.07, 1 df, P = .94), and among the 293 women who had previously attended some college compared to the 1339 women with no previous college (t = -1.38, 1 df, P = .18).
The failure to find a difference in the prevalence of alcohol dependence between those in college and those not in college might be due to the inclusion of nontraditional part-time students in the college-attending subsample. Among both the men and women who were currently attending college, alcohol dependence was not more prevalent among the full-time vs the part-time students (men: t = -0.46, 1 df, P = .65; women: t = 0.61, 1 df, P = .54), and there still were not significant differences between college-attending and non–college-attending men (odds ratio = 0.79, t = -1.24, df = 1, P = .22) or women (odds ratio = 1.32, t = 1.37, df = 1, P = .18) in the prevalence of past-year alcohol dependence when part-time students were excluded from the college-attending subsample.
In a large representative national sample, 18% of US college students (24% of men, 13% of women) suffered from clinically significant alcohol-related problems in the past year, compared with 15% of their non–college-attending peers (22% of men, 9% of women). College students were more likely to receive a diagnosis of DSM-IV alcohol abuse than their peers not in college, but despite the fact that those in college were drinking more, they were not more likely to receive a diagnosis of DSM-IV alcohol dependence.
Several possible explanations were explored for the lack of a difference in the prevalence of alcohol dependence among those in college vs those not in college. The lack of a difference could not be explained by the fact that some college students in their earlier years in school may not have had adequate time to develop problems, nor could it be explained by differential drop-out rates of those with alcohol dependence from college, or by the inclusion of nontraditional part-time students in the college student subsample.
The results of this study are consistent with the suggestion of Slutske et al7 that college attendance may be associated with experiencing some consequences of drinking to a greater extent than others and that it may not necessarily be related to a more pervasive syndrome of problems that is required for a diagnosis of alcohol dependence. This might be because certain symptoms are not as relevant or noticeable against the backdrop of the college environment, or because college provides a buffer against experiencing certain alcohol-related problems, compared to life outside of the college context. In addition, there were some symptoms that were actually less common among those in college than their non–college-attending peers, which also may have contributed to the lack of a difference for alcohol dependence as a whole.
The emerging picture of the relationship between alcohol involvement and educational status and educational attainment is complex. As previously illustrated by results from the Epidemiologic Catchment Area study,14 there may not be a simple linear relation between years of education and the probability of drinking or having an alcohol use disorder. For example, during high school, those who are college bound drink less than those who do not go on to college7,14- 17; substance use disorders are associated with the failure to enter college among high school graduates.18 However, college students drink more than their non–college-attending peers during the college-age years.5- 7 Lifetime alcohol use disorders are higher among those who attended some college but did not graduate compared with those who graduated from high school but had no college14 and those who graduated from college.14,19 Those who attend college drink more5- 7 and have higher rates of alcohol abuse than their non–college-attending peers during college (present study), but college graduates have the lowest rates of lifetime alcohol use disorders.14,19,20
This complicated array of findings suggests another possible explanation for the lack of a difference in the prevalence of alcohol dependence among those in college vs those not in college. Samples of college students do not include two groups most at risk for alcohol use disorders, those who fail to graduate from high school and those who do not enter college. Thus, there may be opposing forces at work: the association of alcohol use disorders and the failure to graduate from high school or to enter college, and the association of alcohol use disorders and heavy drinking in college. These combined opposing forces may sum to an overall lack of differences in alcohol dependence between those who attend college and those who do not,7 and may also explain why daily alcohol use and some alcohol use disorder symptoms, such as withdrawal, are actually more common among those not in college than those in college.
The results of this study raise several important questions that warrant further investigation. First, why do college students have higher rates of alcohol abuse than their non–college-attending peers? There is some evidence to suggest that the association between college attendance and heavy/binge drinking can be explained by systematic differences between college students and their non–college-attending peers on factors such as race, family income, marital status, or living arrangement.7,21 Other indicators of college student drinking (such as the frequency of getting drunk,7 the maximum quantity of alcohol consumed in a single day,7 and driving under the influence of alcohol22) cannot be completely explained by such factors. For example, in a study of 73 young adult female twin pairs discordant for attending college (a study design that provides an ideal control for any family background characteristics that might explain the association between college attendance and alcohol involvement), the college-attending twin exceeded her non–college-attending co-twin by nearly one half of a standard deviation on the maximum amount of alcohol that she had consumed in a single day in the past year.7 Such results are consistent with the conclusion that some aspect of the college experience may be an important environmental risk factor for certain aspects of drinking among young adults.7 It will be important to determine whether the same is true for alcohol abuse. Nearly two thirds (62%) of recent high school graduates enrolled in college in 2001.23 If some aspect of the college experience is a risk factor for the development of clinically significant alcohol use disorders, then it is a risk factor to which many of the young adults in the United States are potentially exposed.
Second, why is the increased risk of alcohol use disorder associated with college attendance higher among women than men? It is unclear whether this can be interpreted as a more potent effect of the college environment on women than on men, whether it is better explained by greater differences between college-attending vs non–college-attending women than between college-attending vs non–college-attending men on drinking-related background characteristics, or whether it is better explained by the roles that relatively more non–college-attending women than non–college-attending men have assumed by their college-age years that might be related to reduced drinking (such as marriage or parenthood).
Third, are the alcohol use disorders that occur among college students more or less transitory than those experienced by their non–college-attending peers? Although it is generally assumed that the alcohol-related problems among young adults are developmentally limited,24 it is clear that a substantial fraction continue to have problems beyond the college-age years. Among the 18- to 19-year-old young adults from a college-based and a community-based longitudinal study who were diagnosed with an alcohol use disorder, 43%24 and 55%,25 respectively, had an alcohol use disorder when followed up at age 25 years. To date, there are no longitudinal studies that have examined the relative persistence of alcohol use disorders of college students vs those of their non–college-attending peers.
The results of this study provide a more encouraging message about the consequences of college drinking than many of the recent reports—although college students suffer from some clinically significant consequences of their heavy/binge drinking, they do not appear to be at greater risk than their non-college attending peers for the more pervasive syndrome of problems that is characteristic of alcohol dependence. Unfortunately, the alcohol-related problems that are associated with college often lead to some of the most dire consequences. For example, about 75% of the 1400 alcohol-related deaths among college students each year are due to the alcohol abuse symptom of being under the influence of alcohol when it is physically hazardous, that is, while driving a car.5,22 Furthermore, such hazardous use does not just imperil the life of the college student, but also the lives of other passengers, and anyone else sharing the road. The consequences of college student alcohol use disorders extend well beyond the college campus.
Correspondence: Wendy S. Slutske, PhD, Department of Psychological Sciences, University of Missouri-Columbia, 210 McAlester Hall, Columbia, MO 65211 (firstname.lastname@example.org).
Submitted for Publication: November 14, 2003; final revision received July 30, 2004; accepted August 20, 2004.
Funding/Support: Preparation of this article was supported in part by National Institutes of Health grants AA013989 and AA011998. The National Household Survey on Drug Abuse was sponsored by the Office of Applied Studies within the Substance Abuse and Mental Health Services Administration and was conducted by Research Triangle Institute, Research Triangle Park, NC.