Nicholas Horton is a Professor of Statistics at Amherst College. His methodologic research interests are in longitudinal regression models and missing data methods. His collaborative statistical work focuses on psychiatric epidemiology and substance abuse research. Nick serves as an associate editor for CHANCEand The American Statistician.
+ Full Transcript
Bob Long: Some college students call it a Rite of Passage. You're no longer at home under the watchful eye of your parents and you love that newfound freedom. You're not old enough to legally drink alcohol, but you can easily get a fake I.D. And many college towns are also known for their special alcohol-related celebrations. Take for example here in Ohio: Halloween is party-time in Athens for Ohio University students and friends, and in Oxford Green Beer Day is the toast of the town for Miami University students the Thursday before spring break every year. Some parents point out - that's okay they drank a lot while they were in college, so what's the big deal? But is the problem larger than that today? Should we be concerned about binge drinking or the long-term impacts of alcohol abuse? I’m Bob Long. We welcome you to another edition of Stats and Stories, a program where we look at the statistics behind the stories, and the stories behind the statistics. And our topic today is alcohol abuse among college students - a major public health concern worldwide. To prepare us for our discussion today…Stats and Stories reporter Emily Hanhart talked with two researchers who have studied alcohol abuse among college students.
Emily Hanhart: It’s no surprise that binge drinking has taken hold of students on college campuses nationwide. Associate Professor of Kinesiology and Health, Rose Marie Ward, has conducted research on this topic of alcohol abuse on Miami’s campus for twelve years. She says Miami students are well above the national average in terms of the number of students who drink on a regular basis.
Rose Marie Ward: We have rates here of alcohol abuse across all years in school that are above national rates. So using national studies from Wexler, who's up in Harvard, about 41 percent of students binge drink. Well, here at Miami we're closer to 60 percent, so that’s a substantial amount.
Hanhart: Miami recently created an alcohol task force to assess how to deal with this cultural issue on campus such as tailgating at football games. Ward’s research examines how Miami’s academic policies mesh with campus culture.
Ward: There is a clear relationship between time of your first class and how much you drink, and the difficulty of your class. That was the other part of my paper that I wrote is that you could have early classes, but if they don't take attendance, or you don't need to be there or it's really easy, you're still going to drink. So it's not just the timing of the classes. You kind of sculpt your class load to fit your social life.
Hanhart: Miami has tried to add more Friday classes to model other universities that have tried to combat excessive binge drinking, and has also pushed back 8 am classes by 30 minutes. Ward says Miami and other public universities nationwide have shaped their class structure to enable certain behaviors that might lead to habitual drinking.
Ward: There's no other part of your life where you can consistently or even occasionally show up and be buzzed or hung over and still have that job. So we kind of enable a lot of things in the academic culture that are not just unique to Miami that kind of facilitate this. And so it's looking at those structures and looking at that system and seeing can we shift it a little bit and maybe put alcohol in certain controlled situations, model good drinking and then try to put other systems in to kind of keep it from spreading.
Hanhart: A researcher at the Center for Alcohol and Addiction Studies at Brown University has dedicated his career to studying the motivations of students with alcohol related issues. Brian Borsari’s most recent study involved multiple trials of 500 students at different universities who violated campus alcohol policies. After their first violation, these students participated in a brief advising session about drinking and their intake levels. After six weeks, students who violated their university policies a second time were given a more intensive motivational intervention. Borsari says the students who participated in the second session saw successful results.
Brian Borsari: When we followed those students out for nine months, we noticed that the people that got the brief motivational intervention compared to nothing - an assessment only control - that they reduced the alcohol-related problems that they had experienced.
Hanhart: Borsari also says the purpose of the interventions is to see what the motivations to drink might be for the students.
Borsari: It seems that sometimes during these sessions you can say something or comment on something or elicit something consistently that might facilitate change.
Hanhart: Borsari hopes that through further research, the issue of alcohol abuse on college campuses will diminish over time.
Borsari: You know, my interventions and others, they don't cut drinking to zero for the most part or problems to zero. But there is a significant reduction. And if you can multiply that taking a public health view over tens or hundreds of thousands of students, those interventions can have a significant effect, if implemented well.
Hanhart: Rose Marie Ward believes that this issue can be tackled at Miami. Nonetheless, she’s surprised there haven’t been more alcohol related tragedies in the twelve years that she’s been conducting research.
Ward: You can drink without being so risky that we have to worry about your life. And that's scary that every day, there's the potential for someone to die on this campus. I don't like that.
Hanhart: For Stats and Stories, I’m Emily Hanhart.
Long: Joining me for Stats and Stories today for our discussion of alcohol abuse among college students: Miami University Statistics Department chair John Bailer, our Media, Journalism and Film chair Richard Campbell is unable to be with us today, but we’re very pleased to welcome our special guest Nick Horton. Nick is a professor of statistics at Amherst College. He’s been involved in studies on a number of substance abuse issues and has also looked at a web based screening and intervention program related to alcohol abuse. Nick, we welcome you to the show today.
Nick Horton: It’s great to be here Bob.
Long: Let’s talk first of all about how you, personally, got interested in this whole area of substance abuse.
Horton: Well my background was in psychology and my interest in data and trying to make sense of it led me to a graduate degree in biostatistics. My work there led me to take some courses in a variety of areas related to psychiatric epidemiology and substance abuse research. I had the pleasure of taking a class from Henry Wechsler who was the founder of the national college alcohol study, which really gave us a much clearer sense of the impact of drinking on college campuses, both for the drinkers and for those other students who weren’t drinking. I continued my work and headed off to Boston Medical Center where I completed a training program for faculty on substance abuse prevention. And then I’ve been teaching at liberal arts colleges and collaborating with faculty at those institutions on studies of college athletes, how housing is set up in colleges and how that affects drinking, and more recently on these studies in Australia and New Zealand on electronic approaches, web-based approaches to decrease college drinking.
Long: Before we get into some of the research you’ve done I think a lot of people, when they think about alcohol abuse they think about American colleges, but what we’re talking about is more of a global type of issue.
Horton: No, it’s certainly true that the issues of drinking and drinking at university or drinking of young people is a critically important public health concern. It’s a chronic issue that has a tremendous impact on society. There’s huge costs both in terms of family violence and lost income and the like. So it really is important. Research has shown that the earlier people get involved in drinking at a heavy level, the more likely they are to be alcohol dependent later on in life so this is really an issue that’s not just a US problem.
Long: Let’s go to John Bailer for the next question.
John Bailer: So a natural question when you think about this, have you seen or does the literature suggest that there might be patterns in the amount of alcohol abuse that’s occurring? Is it going up? Is it going down? Is it relatively stable over time?
Horton: One of the things we see from the research in terms of drinking is that there is a lot of it, it does tend to be diverse and you can actually have an impact in terms of prevention and intervention. So there are some hopeful aspects and one of the areas I think has been more promising, partly because it’s cheap, are these ESBI approaches, Electronic Screening and Behavioral Intervention, where students can be provided with just more information about their drinking. That has really shown to be useful in certain approaches, in particular when you show students how much they’re spending on alcohol because that adds up in a way and there’s a lot of research that shows that if the cost of alcohol is higher, students will tend to drink less. And so thinking about these behavioral changes are hard, but there are approaches that can have an impact.
Long: As far as, I know one of the studies you were talking about was involving students, I believe in New Zealand and Australia, and I want you to kind of go into that why picking, I think it was seven universities in New Zealand for one of the studies that you were involved in.
Horton: Well one of the things that’s really important thinking about this ESBI is whether or not it can actually be rolled out in a more systematic way. So those seven universities in New Zealand are all of the universities in New Zealand. So this was a nationwide attempt to see if it was possible to decrease drinking in that manner. The way the study worked was that all the students at those universities were offered a chance to participate. They completed a questionnaire, something called the audit, to try to identify people who were at moderate or higher levels of drinking, those were the ones most perceived to be at risk. So about 3,400 of those students ended up being eligible for the study and were then immediately randomized to one of two conditions. One was to do this baseline screening and then be followed up after one and three months. The other group received a targeted website which opened up right at the end of that screening which had multiple tabs to it that compared their probability based on their responses to these questions, probability of getting into a car crash or having some other consequences to drinking, to really see how their drinking compared to other students at the university. This idea of social norms seemed to be very effective. And as I said, mention of how much they were spending on drinking every year, so that was the intervention which is very straightforward and cheap, it’s just popping up a website. They were then followed up after a couple of months and what was seen in that study is that this is not going to make an earth-shattering, big difference, but it seemed to be associated with modest decreases, 10% decreases in the duration of drinking or the number of times they drink per week, the volume consumed, which seem like part of an answer to be thinking more holistically about their drinking.
Bailer: So this was just a single intervention, a single, short intervention following this?
Horton: Yes. So just that one shot with the website, there was no follow-up and it seemed like it had some modest improvements.
Bailer: So part of the discussion to consider what could be done in the future in additional electronic interventions that might reinforce this message and is there a strategy where that might be explored?
Horton: So a lot of the other research that’s going on is smaller scale, it’s not going to be national type interventions, but really thinking about this pattern about where is this college drinking coming from, clearly in the U.S. and overseas it starts earlier in life: middle school, high school is kind of a key time where patterns of adulthood are set. So looking at drinking games, for example, looking at how different groups on campuses are drinking, what kind of effect there might be for say athletic teams versus others? Thinking about the role of acculturation of recent immigrants to the country, but to really think about ways that we could screen and identify people most at risk and then really target those folks for a particular follow-up. That approach, when we think about statistics in this area, there are clinical trials, which are great to do, but there’s a whole bunch of observational studies which may need to predate that to understand what questions, what approaches might be feasible, where we might see most impact.
Long: You’re listening to Stats and Stories, where we talk about the statistics behind the stories and the stories behind the statistics and we’re focusing today on alcohol abuse among our young people; a major public health concern, not only here in the United States, but worldwide. I’m Bob Long, our regular panelist that’s with us today Miami University Statistics Department Chair John Bailer and our special guest today Nick Horton, a professor of statistics at Amherst College who has specialized in studying this whole issue of substance abuse. Nick, kind of going on from there, we talked about New Zealand only having seven universities, how much more difficult, though, is it when you’re dealing with a country like ours where in the state of Ohio alone you’ve got, I don’t know how many public and private universities, but just across the country, thousands of different schools, and they all are having similar kinds of issues. So how much more difficult is it here in this country to deal with this issue?
Horton: Well I do think it is a challenge given the heterogeneity of our systems and structures. A lot of the research that’s happening in this area is coordinated through the National Institutes for Health and the National Institute for Alcohol Abuse and Alcoholism has put a key focus on prevention and intervention because there’s a lot of indications that early action can make a difference; that if you wait until someone has full-blown alcohol dependence in adulthood, it’s too late. It really is very difficult to treat it at that point. And so that has really led statisticians to be involved in the design of these studies to try and identify those at high risk, to identify creative solutions that can help to emolliate the problem.
Long: John Bailer.
Bailer: Can you talk a little bit about how statistics can be used to assist with the screening and identification, what are some of the strategies to use these data to try to make these identifications?
Horton: Well one of the areas where statistics plays a key role is in the psychometrics of screening. How do we identify a couple of questions that may be easy and straightforward to identify those at risk? And there are a series of these have been put together, this audit that I mentioned earlier was one where there’s essentially an eight item scale that tries to determine where there are problems. There’s another four item scale called the CAGE and to give you an example, has anyone encouraged you to cut down on your drinking? And that’s kind of if someone says yes to that and another one is an eye-opener, do you need a drink in the morning to get going? Getting two of those four is indicative of problems. And so trying to come up with a measure that is both highly specific and sensitive to find the right people at the right risk is an easy manner in which statistics plays a big role.
Long: As far as on college campuses, I know for example, Miami University has a disciplinary board, as all universities do and many times some of the issues they’re dealing with have something to do with a problem they’re seeing a student with that’s alcohol related. When you’re talking about the at risk population, is that one of the places that you can go sometimes to find, well who are the students we really need to talk to, to address these kinds of concerns?
Horton: Well I think that area of how we deal with disciplinary action, that policy related to once we identify a student, what’s the best approach? A lot of schools that have actually instituted amnesty programs, they want to make sure that if there’s a situation where there’s underage drinking and someone’s at risk, that we get that person into medical care so that they don’t die or have serious consequences. And I think there’s some indication that that’s an important strategy to protect students. My colleague at Brown University, Brian Borsari, has been very active in their programs that are essentially like that intervention. You have a situation that’s a disciplinary problem, let’s put that person into a much more intensive treatment program and that can actually often have a very positive outcome where the disciplinary wouldn’t treat the underlying problem. And so this is an area where doing randomized trials can really get a sense, at these institutions, of what the right approaches are. And that’s an area where statisticians play a big role.
Bailer: So just one thing that people might be interested in, the use of words like randomized trial, is something that might be new to folks. So could you expand on what do you mean by randomized? What do you mean by randomized trials?
Horton: So we think about randomized trials typically in terms of a drug, there’s a new drug we want to test, we compare it to the standard treatment that’s a blinded condition where the patient, the subject, in that case doesn’t know which one they’re getting, the investigator doesn’t know either. They’re followed later on and we see which is the more effective, which one has the worse side effects. In the situation that I’m talking about, it’s a little bit more subtle. You might have a situation where somebody comes into their disciplinary hearing and they’re offered the opportunity to be part of a trial of an innovative new approach to treat their issues and they would then be randomized to either perhaps standard disciplinary or this other kind of treatment program where their disciplinary might be held off. There are a lot of challenges to this because it’s no longer blinded, people know which treatment they’re in. But it’s important then to still be able to think about how to really reconstruct what it is that had the impact later on; which approaches are effective? Which ones are not? And the cost effectiveness of those approaches as we try to tackle this problem.
Bailer: So as a quick follow-up, do you worry about selection bias? When you talked about the students in New Zealand, you only got the ones that were willing to participate, when you think about these other trials, so someone’s coming in for disciplinary hearing and disciplinary action potentially, they’re opting in or opting out of these things. How do you deal with this possible selection bias with the characteristics of the people that are in your studies?
Horton: Well I think that’s one of the important parts of the randomized components of this. It’s true that the students who are participants in the New Zealand and Australian studies were the ones who were willing to participate, but they were still randomized into two groups. And so there may have been some bias or some questions about whether we can generalize to all students, but nonetheless I think it’s still possible to be comparing the groups that were randomized to the web-based screening and those not, but it is an important issue for us to keep in mind. Another important issue, we live in an imperfect world and you would imagine missing data arises a lot when you’re studying folks who are involved with substance abuse and that’s an important area where we were only able to follow-up on about 85% of those Australian and New Zealand students and so the question might arise whether or not the non-respondents were systematically different from the respondents. And there’s been a lot of innovative approaches to reach out to try to minimize that missing, try to get one answer from them, are you drinking or not? It might be that one simple question, but it’s an important role for statisticians that are part of this research team to help in understanding what conclusions you can make.
Long: I know another issue when you’re dealing with substance abuse is just the personal denial a lot of times that people go through, they don’t want to admit that they have a problem. How does that impact the kinds of things you’re trying to ask them? You’re trying to get them to give you the honest responses that you need.
Horton: Well this is another area where survey design and statisticians have been very involved. Prior to this work with students, I worked with folks coming out of a detox center in Boston. And again, these are very seriously alcohol and substance involved folks and it’s interesting because that was a study where we enrolled them in the detox, which is a depressing place just like in the movies. You might anticipate that they are in there for 3-7 days to dry out and start proceeding on to their next stage in life. If you ask them at that point whether they’ve been in detox before in the last six months, you get an answer from them about the number of times they were in detox. We were able to compare that to the billing records to the Commonwealth of Massachusetts to see how those corresponded and those reports were actually fairly good. If you then followed up with these same people six months later and said how often have you been in detox since we met since you were enrolled in this study. Now you’re in a white hospital room, social desirability bias that arises in that situation, there answers were much less accurate when compared to the same information. And so this concern about self-report bias is an important one to be thinking about and there are approaches that people have come up with in the same way that you ask people not, “Did you vote last time?” you ask them “What polling station did you vote at last time?” And it turns out that will elicit a much more honest response. But again these are all important, what I would call non-sampling error questions that we need to keep in mind.
Long: You’re listening to Stats and Stories and we’re focusing today alcohol abuse, substance abuse among students, a major public health concern not only here in America but worldwide. I’m Bob Long. Again our regular panelist with us today Miami University Statistics Department Chair John Bailer, our Media, Journalism and Film colleague Richard Campbell is unable to be with us today and our special guest, we’re very pleased to have Amherst College statistics professor Nick Horton. And Nick has been involved in studying web-based screening and intervention programs and many other studies related to substance abuse. John Bailer, we’ll go back to you for the next question.
Bailer: So can you tell us a little bit about who else is on your research team that’s investigating these problems?
Horton: So the research teams really are diverse. Statistics is a team science and thinking about improving the health of populations in this manner is true as well. We have for a lot of these studies, there would be a primary physician, a primary care physician in internal medicine might be the person leading up the project. There might be faculty in psychology because there are a lot of aspects of psychology that are related to this. Social work is important because there’s a component of trying to reach out to these populations that might be at risk. Economists would be involved in thinking about the cost effectiveness of some of the approaches. Epidemiologists trying to study the distribution and determinants of disease would be part of that team. So it really is a very diverse group of individuals and it’s important for the statisticians and the others to really know the science to be able to make those contributions.
Bailer: So what did you do in your education and your experiences that helped you get ready to function on such teams?
Horton: Well I think one of the useful parts for me were having the opportunity to be engaged in capstone projects, summer research projects, where I was really able to hear from the investigators the question they wanted to answer and then try to figure out the analysis could be put together; how the design could be structured in a way to allow them to answer those questions. So it was really that involvement early on and having a background knowledge and understanding in these areas that really made that much more possible.
Long: I want to take a broader look at what we’re talking about here. You mentioned the economists and I don’t think we think much about the overall impact if someone starts off in college and they’re abusing alcohol, now they’re out there in the workforce partying with their friends after work and things like that, what about the economics of all of this affecting our country; the problem that we’re seeing.
Horton: So the impact of alcohol abuse and dependence on the economy has been conservatively estimated in the tens of billions of dollars, that this is a growing problem that has major societal impact. This is something that the NIAAA provides estimates of as well as these estimates of, if you start drinking at 14 or 15 or 16 the probability of being alcohol involved at a level that is impairing your work or family or other engagements you have is upwards of 25 or 30-35%. And so really thinking about trying to decrease that number, to decrease the risk, figure out approaches to move people to a much more moderate level of drinking is really an important national goal.
Long: The three of us can all relate to the fact that sometimes on college campuses, people who are having alcohol issues may have attendance problems in our classrooms. We’re talking again when you get to the workforce, that exacerbates the situation there as well.
Horton: Well the inability to hold a job, the violence which often is associated with alcohol, for families I think the impact of that is really very, very serious. It’s beyond just missing some classes here and there. George Vaillant had a study of Harvard graduates from the early 20th century followed up into late adulthood and it was a fascinating glimpse; you could see the issues these men, at that point it was only men being studied, had when they were 70 and 80 were really seen when they were in their early 20s.
Bailer: If someone wanted to get involved in doing this type of work, so if I have a student that hears this program and they go, wow this is really interesting, really cool stuff. What would you recommend that they study if they’re interested in being involved a statistician or biostatistician working in this area. Preparing for this work, what would you encourage them to do?
Horton: Well for any statistics researcher to be effective, they need a number of skills. Obviously they need to have a solid understanding of statistical concepts, these questions of research design, observational randomized data, thinking about missing data, thinking about more sophisticated modeling to be able to follow people over time or repeated measures models. Thinking about multiple outcomes, how do you measure alcohol consumption? It’s not an easy yes/no question which raises questions about multiple comparisons and multiplicity. So there’s kind of the principles of statistics are really important. Certainly they need to have the ability to analyze the data, to be able to bring things together to answer the questions of that investigative team. So those computational, data related skills are really important. And it’s also important for them to have some knowledge of the subject matter area. They need to have a background to be able to communicate at a high level with these scientists so they’re not making some mistake or missing something. To be part of the team, they need that broad knowledge.
Long: I think one thing that we always mention on this show is the news tie-in, as to how this problem is reported, and based on your study, how accurate the reporting is from stories that they hear in the press and reporting on the statistical studies like the ones that you’ve done.
Horton: Well it’s always a challenge with the media because there needs to be a new component and something really novel, the “man bites dog” type of approach. One of the challenges here is that this is such a serious, chronic issue that we see that there is a tremendous number of our middle school students come into high school having drunk more than a sip of alcohol and setting patterns which then kind of proceed to college. What I think is helpful to be looking and describing innovative approaches like the electronic screening and brief intervention that’s affordable that can at least bring things to mind for students and their parents, kind of those around them. To see where their drinking relates to their peers. To see the impact of pricing on consumption. There have been some interesting studies that have been done at certain stadia for sports events have been made alcohol-free or not. Kind of using those natural experiments to see what kind of impact there can be in the communities. More broadly, there is a whole field now that’s developing of geospatial statistics. Where are the alcohol outlets? How do those change? How does that affect what’s going on? So I think it’s important to be thinking about this as a long-term and kind of be proposing approaches that people can start to understand them.
Long: John Bailer, we’ve got time for one more question for Nick Horton.
Bailer: So I’m going to channel Richard Campbell for a minute. So a question he likes to ask and I’ll ask in his absence. He says I’m involved in helping journalism students prepare for the future, reporting statistics is part of what they need to do, reporting the results of studies. What is it that you would suggest these students need to become better reporters of statistical studies and convey this to the public in a more effective way?
Horton: I think that issue of communication is a really critical one. I’ve been very impressed by a number of journals, JAMA, is one of them that provides summaries for lay people that are really intended to be the communication approaches to kind of understand the strengths and limitations of the research studies that come out because each of them, none of them is going to be the definitive answer to these questions so it’s important that students, that journalism students, that citizens have the ability to read beyond the title of a paper, to be able to go beyond the abstract and to try to see what’s being determined by that. When we look at these studies of Australia and New Zealand, we’re looking at most 10% decline at most in terms of what’s going on with drinking. That means that needs to be part of a solution. It’s not suddenly saying that boom, we have the answer.
Long: Nick Horton of Amherst College, thanks again for joining us on Stats and Stories to share your insights on this worldwide public health concern about substance abuse, we appreciate it.
Bailer: Thanks Nick.
Horton: Great to be here.
Long: If you’d like to share your thoughts on our program, just send us an email at StatsandStories@miamioh.edu. Be sure to listen for future editions of Stats and Stories where we discuss the statistics behind the stories and the stories behind the statistics.