The Health Belief Model and smoking cessation behaviours

Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=vach20

Journal of American College Health

ISSN: 0744-8481 (Print) 1940-3208 (Online) Journal homepage: https://www.tandfonline.com/loi/vach20

Association between perceived risk of harm and self-reported binge drinking, cigarette smoking, and marijuana smoking in young adults

Matthew Hanauer, Madison R. Walker, Kendall Machledt, Melissa Ragatz & Jonathan T. Macy

To cite this article: Matthew Hanauer, Madison R. Walker, Kendall Machledt, Melissa Ragatz & Jonathan T. Macy (2019): Association between perceived risk of harm and self-reported binge drinking, cigarette smoking, and marijuana smoking in young adults, Journal of American College Health, DOI: 10.1080/07448481.2019.1676757

To link to this article: https://doi.org/10.1080/07448481.2019.1676757

Published online: 25 Nov 2019.

Submit your article to this journal

Article views: 63

View related articles

View Crossmark data

MAJOR ARTICLE

Association between perceived risk of harm and self-reported binge drinking, cigarette smoking, and marijuana smoking in young adults

Matthew Hanauer, MPAa, Madison R. Walker, BSa, Kendall Machledt, MPHa, Melissa Ragatz, BAa, and Jonathan T. Macy, PhD, MPHb

aCenterstone’s Research Institute, Bloomington, Indiana, USA; bIndiana University School of Public Health – Bloomington, Bloomington, Indiana, USA

ABSTRACT Objectives: Evaluate the association between perceived risk of harm and self-reported binge drinking, cigarette smoking, and marijuana smoking among college students. Participants: Participants were 599 students (ages 19–28) at a large Midwestern university recruited from October 2015 to December 2017. Methods: Hurdle regression was used to test the relationship between perceived risk of harm from substance use (i.e., binge drinking, cigarette smoking, and marijuana smoking), and self- reported use. Demographic characteristics were tested as moderators of this relationship. Results: Engagement in all three substance use behaviors was less likely when perceived risk was high. Age moderated the association between perceived risk and self-reported marijuana smoking with younger participants demonstrating a stronger relationship between perceived risk of smok- ing marijuana and self-reported marijuana smoking. Conclusion: Intervention programs will be most effective when perceived risk of substance use is high. Therefore, intervention programs should aim to increase college students’ perceived risk of substance use.

ARTICLE HISTORY Received 1 September 2018 Revised 18 September 2019 Accepted 20 September 2019

KEYWORDS Binge drinking; cigarettes; college students; community-based prevention; Health Belief Model; marijuana; perceived risk of harm; substance use

Introduction

Substance use continues to be a large problem on many col- lege campuses and across adolescent and young adult popu- lations in general, with adverse physical, mental, social, and behavioral health outcomes. Substance use often co-occurs with anxiety and depressive disorders,1 and college students experience a high prevalence of both anxiety and depres- sion.2 Individuals who have an alcohol or drug use disorder are significantly more likely than the rest of the population to use another addictive substance as well, and individuals who experiment with more than one substance are more likely than nonexperimenters to use subsequent drugs.3,4

The study reported here focused on college students’ use of three substances: alcohol, cigarettes, and marijuana.

Alcohol is the most commonly-used substance by young adults and college students, closely followed by tobacco and then marijuana, the most widely-used illicit drug.5,6

Although recreational and medical marijuana use are becoming legalized in several states, in the state that this study took place, marijuana use is not legalized for either medical or recreational use. Although specific definitions vary by sex and body mass, binge drinking is commonly defined as any consumption of alcohol that increases the risk of the drinker experiencing alcohol-related problems and/or that puts others at risk of secondhand effects.7

Physical health consequences associated with binge drinking include alcohol poisoning, alcohol dependence, liver cirrho- sis, several types of cancer, preventable injuries, and increased prominence of sexually transmitted infections.8,9

Binge drinking is a dangerous behavior that contributes to a large portion of alcohol-related deaths each year, and it is the most prevalent form of alcohol consumption by college students and other young adults.10–13 In addition to the dir- ect consequences of binge drinking, individuals who engage in frequent binge drinking are also more likely to report concurrent use of cigarettes and/or marijuana.14

Tobacco use is a major leading risk factor that contrib- utes to causes of preventable death in the United States.12

One-fourth of Americans use tobacco, and nearly 20% smoke cigarettes.15 Despite recent declines in cigarette smoking across all age groups, 22% of college-aged young adults have reported smoking cigarettes in the 30 days prior to completing use surveys, and cigarette cessation is less likely when individuals begin smoking as college-aged young adults.12,16 Cigarette smoking is associated with major health consequences including cardiovascular morbidity and mor- tality and certain cancers (e.g., lung, colon, pancreas), as well as major social consequences (e.g., lower socioeconomic statuses, lower education levels).17–20 In addition, drinking alcohol can significantly increase the pleasure associated

CONTACT Matthew Hanauer [email protected] Centerstone’s Research Institute, 409 W 1st Street, Bloomington, IN 47401, USA Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/vach. � 2019 Taylor & Francis Group, LLC

JOURNAL OF AMERICAN COLLEGE HEALTH https://doi.org/10.1080/07448481.2019.1676757

with smoking cigarettes.21 Alcohol consumption and cigar- ette use often occur together, and the two substances can interact to potentially increase the risk of serious health problems.22,23

The specific consequences of marijuana use on college students are somewhat unclear due to inconsistencies reported in the literature.2 However, there is some evidence to suggest that chronic marijuana use can lead to marijuana addiction, as well as addiction to other substances.6 The prevalence of college aged marijuana users has increased in recent years to 22% according to the 2017 National Survey on Drug Use and Health.12 Among these individuals, five percent experience marijuana use disorder (defined as expe- riencing significant impairment in physical or mental health, work performance, or social interactions from repeated use of marijuana).12 Chronic and high-frequency marijuana users are more likely to report increased medical visits for physical and/or mental health issues and are more likely to report increased illnesses, injuries, and/or emotional prob- lems compared to infrequent or nonusers.24

Due to the high prevalence of substance use and the potentially associated risk factors that can lead to negative health and social consequences, there is a need to under- stand predictors of substance use. To accomplish this, researchers often use theoretical and evidence-based models. In the present study, we used the Health Belief Model (HBM). According to the HBM, perceived risk of harm is an important predictor of behavior, with perceived risk of harm defined as an individual’s subjective perception of the potential harm from engaging in those behaviors.25–29 In the present study, we focused on the HBM construct of per- ceived risk of harm by testing its relationship with three self-reported substance use behaviors: binge drinking, cigar- ette smoking, and marijuana smoking.

While there is robust support for the HBM’s efficacy with many health behaviors, evidence for the HBM regarding substance use is limited.26–28 Additionally, the limited evi- dence regarding the HBM and substance use (1) has contra- dictory results, (2) does not capture all types of substance use (e.g., marijuana use is absent), and (3) does not evaluate the effect of demographic variables as moderators of the effects of key HBM constructs on substance use behav- iors.29–33 In the present study, we captured a more compre- hensive range of substance use (e.g., marijuana use) and

evaluated the moderating influence of sex, age, race, and importance of religion to determine whether the relationship between perceived risk of harm and self-reported use of sub- stances differed by these demographic factors.

Due to the high prevalence of substance use among col- lege students and limited published findings regarding the HBM and substance use, additional research is crucial to better understand the factors associated with substance use. The current study had two aims: (1) test the relationship between perceived risk of harm from self-reported binge drinking, smoking cigarettes, and smoking marijuana and reported engagement in those behaviors, and (2) determine if those relationships were moderated by specified demo- graphic variables.

Methods

Participants

Recruitment events occurred on a rolling basis from October 2015 to December 2017 at various locations within the university campus. The recruitment events offered sub- stance use and HIV prevention and outreach services. The events were advertised through collaborations with local agencies and community nonprofit organizations, as well as through social media and flyer advertisements. The flyers were written in the 10 most commonly spoken languages at the university. Participants provided written consent before they completed the survey. Participants were 599 self- selected individuals (ages 19–28, mean ¼ 22) from a large college campus (45,000þ students) in the Midwestern United States. The demographic characteristics of the sample are shown in Table 1.

Procedure

An independent institutional review board (IRB) provided ethics approval for the study. After giving informed consent, participants completed the Government Performance and Results Act (GPRA) questionnaire, which consists of 78 structured questions regarding substance use over the 30 days prior to an individual completing the survey; the GPRA is often used to evaluate brief substance use.34,35

The GPRA was developed from two standardized substance

Table 1. Sample characteristics (n ¼ 599).a

Continuous variable Dichotomous or ordinal variables

Characteristic Mean (standard deviation) Range Number (percent) Percent missing

Age 22 (1.73) 19–28 – – Sex (female) – – 370 (61.8%) – Race – – – – White – – 436 (72.8%) 1.0% African American – – 124 (20.7%) 1.0% Asian-American – – 10 (1.7%) 1.0% Another racial identity – – 29 (4.8%) 1.0%

Religious importance 9.2% Not at all important – – 128 (21.4%) – Not too important – – 156 (26.0%) – Fairly important – – 179 (29.9%) – Very important – – 136 (22.7%) –

aData presented is for complete data set without missing data.

2 M. HANAUER ET AL.

abuse assessments (the Addiction Severity Index and the Treatment Services Review). There is evidence supporting the validity and reliability of the entire GPRA instrument.36

Since the focus of this study is to evaluate whether the HBM is applicable within the substance use field, items were chosen from the GPRA based on their relevance to the HBM (ie, self-reported perceived risk of harm and self- reported use). It took participants an average of 15 minutes to complete the GPRA, and it was completed in one sitting at the respective recruitment event. Researchers provided instructions about how to complete the GPRA and were available for questions. The research team entered the data into the Statistical Package for the Social Sciences (SPSS), and at least two researchers checked every questionnaire to identify and correct data entry errors. Additionally, the lead researcher ran frequency reports to identify mistakes to fur- ther enhance the integrity of the data. When errors were found, the researchers reviewed the original paper question- naire to correct any errors.

Measures

Demographic characteristics The following demographic variables from the GPRA were included in the analyses for the current study: (1) sex, (2) race, (3) age, and (4) the importance of religion. Sixty-two percent of the participants identified as female, 20.7% as African-American, 1.7% as Asian-American, and 4.8% as another racial identity. The importance of religion was measured using a Likert scale ranging from not at all important to very important. Twenty-one percent of the par- ticipants identified religion as not at all important (to them personally), 26% as not too important (to them personally), 30% as fairly important (to them personally), and 23% as very important (to them personally). For analyses, all the demographic characteristics besides the importance of reli- gion (ordinal) and age (continuous) were treated as dichot- omous variables (eg, white vs. nonwhite).

Perceived risk of harm Perceived risk of harm from binge drinking, smoking ciga- rettes, and smoking marijuana were self-report variables included in the GPRA.35 These perceived risks were assessed by the following questions: (1) how much do people risk harming themselves physically and in other ways when they have five or more drinks of an alcoholic beverage once or twice a week?, (2) how much do people risk harming them- selves physically and in other ways when they smoke one or more packs of cigarettes per day?, and (3) how much do people risk harming themselves physically and in other ways when they smoke marijuana once or twice a week?.35

Participants responded to each perceived risk of harm ques- tion using a four-point scale (no risk; slight risk; moderate risk; and great risk).

Outcome measures The outcome measures (i.e., reported frequencies of binge drinking, smoking cigarettes, and smoking marijuana during the 30 days prior to completing the survey) assessed in this study have been used in multiple studies as indicators of substance use.34,37,38 The outcome measures were assessed using the following questions: (1) during the past 30 days, on how many days did you have five or more drinks on the same occasion?, (2) during the past 30 days, on how many days did you smoke part or all of a cigarette?, and (3) dur- ing the past 30 days, on how many days did you use mari- juana and/or hashish?.35

Data analysis

Overall percentages of participants who reported binge drinking, smoking cigarettes, and smoking marijuana were evaluated. The relationship between the demographic varia- bles and each outcome variable was then tested using a hur- dle model because the outcome variables were not normally distributed. We confirmed the non-normality of these count variables with histograms and Q-Q plots. The hurdle model helps to account for large numbers of zeros (i.e., participants reporting no substance use) by creating two models.39 The first model is a truncated Poisson model or negative bino- mial model depending on over-dispersion (ie, variance larger than the mean for each outcome variable) used to evaluate the nonzero data. Additionally, a separate logistic model represented reported substance use versus no reported use. A truncated negative binomial model and binary logistic hurdle model were used due to the over-dispersion of the data.39 For the truncated negative binomial model, the par- ameter estimates were exponentiated, thus transforming them into incident rate ratios (IRRs) to improve interpret- ability. The IRRs show the percent change in the rate of the number of days a participant reported binge drinking, smoking cigarettes, or smoking marijuana.40 The parameter estimates were also exponentiated for the logistic model, transforming them into odds-ratios (ORs).40 Parameter esti- mates are interpreted as average change, assuming all other variables are held constant.

The moderating effect of each demographic variable on the relationship between perceived risk of harm from each substance and self-reported use of the substance was also tested. Variables were mean centered before computing inter- action terms. Significant interactions were probed following the procedure outlined by Aiken and West.41 Finally, because 15% of the data were missing, we conducted a missing com- pletely at random (MCAR) test using the MissMech package in R, which was nonsignificant (p ¼ .229). Given these results, we are reporting the results after listwise deletion.

Results

The overall mean for perceived risk of harm was 3.04 (SD 0.72) for binge drinking, 3.84 (SD 0.51) for smoking ciga- rettes, and 2.26 (SD 0.85) for smoking marijuana. We tested for demographic differences in mean perceived risk of

JOURNAL OF AMERICAN COLLEGE HEALTH 3

engaging in each behavior. For perceived risk of binge drinking, nonwhite participants (mean 3.26, SD 0.72) had significantly higher perceived risk than white participants (mean 2.95, SD 0.71), and those who reported that religion was very important (mean 3.28, SD 0.73) had significantly higher perceived risk than those who reported that religion was not all important (mean 2.92, SD 0.69), but there were no significant differences by sex or age. For perceived risk of smoking cigarettes, white participants (mean 3.87, SD 0.43) had significantly higher perceived risk than nonwhite participants (mean 3.72, SD 0.67), and females (mean 3.87, SD 0.47) had significantly higher perceived risk than males (mean 3.76, SD 0.56), but there were no significant differen- ces by age or importance of religion. Finally, for perceived risk of smoking marijuana, nonwhite participants (mean 2.42, SD 0.98) had significantly higher perceived risk than white participants (mean 2.22, SD 0.80), and those who reported that religion was very important (mean 2.60, SD 0.84) had significantly higher perceived risk than those who reported that religion was not all important (mean 2.04, SD 0.81) or not too important (mean 2.14, SD 0.85), but there were no significant differences by sex or age.

Results from the regression models are presented separ- ately below for each of the three outcomes. For each out- come variable, substance use prevalence in our sample was compared to national averages from the 2017 National Survey on Drug Use and Health.12 Main effects for risk and demographic variables for the truncated negative binomial and logistic models are reported, and any significant inter- action effects are reported. Results are presented in Tables 2 and 3.

Binge drinking

A higher percentage of participants in our sample reported binge drinking (56.6%) compared to the national average of the same population (36.9%).12 The mean number of days (out of the past 30) participants reported binge drinking was 2.53 (SD 3.87). In Table 2, for the main effects for the trun- cated negative binomial model, perceived risk of harm from binge drinking was negatively associated with the number of days a participant reported binge drinking. As perceived risk of harm increased, the rate at which participants reported binge drinking decreased. In Table 3, for the main effects of the logistic model, perceived risk of harm from binge drink- ing age, race (white), and sex (male) were also significantly related to reported days of binge drinking. As perceived risk of harm from binge drinking increased, the odds of report- ing binge drinking deceased. However, the odds of reporting binge drinking are increased for males, younger participants, and participants who identify as white. There were no sig- nificant interaction effects.

Cigarette smoking

A lower percentage of participants in our sample reported smoking cigarettes (15.2%) compared to national averages (22.3%).12 The mean number of days (out of the past 30) participants reported smoking cigarettes was 0.89 (SD 3.84). In Table 2, for the main effects of the truncated negative binomial model, only perceived risk of harm and age were significantly related to the number of days a participant reported smoking cigarettes. Those who reported high

Table 2. Results for truncated negative binomial regression models predicting substance use.

BD-30 days estimate

BD-30 days IRR

BD-p values

CS-30 days estimate

CS-30 days IRR

CS-p values

MS-30 days estimate

MS-30 days IRR

MS-p values

Demographics Age 0.016 (0.041) 1.015 0.700 0.259 (0.117) 1.296 0.027 0.246 (0.071) 1.279 >0.001 Gender 0.010 (0.139) 0.990 0.941 0.366 (0.419) 1.442 0.382 0.326 (0.176) 1.385 0.064 Race 0.324 (0.188) 0.723 0.086 0.070 (0.627) 1.073 0.911 �0.027 (0.207) 0.973 0.895 Income 0.047 (0.144) 0.954 0.743 0.156 (0.415) 1.156 0.707 �0.224 (0.177) 0.799 0.207 Religion Importance 0.058 (0.071) 0.944 0.417 0.178 (0.277) 1.195 0.521 �0.032 (0.093) 0.969 0.728

Perceived risk of harm Binge drinking 0.327 (0.096) 0.721 0.001 – – – – – – Cigarette smoking – – – 0.903 (0.412) 0.405 0.029 – – – Marijuana/hashish smoking – – – – – – �0.371 (0.117) 0.690 0.002

Significant interaction terms Perceived risk by age – – – – – – 0.176 (0.083) 1.163 0.033

Note. BD: binge drinking, CS: cigarette smoking, and MS: marijuana/hashish smoking. IRR: incident rate ratio. Demographic variables (not including age or religion importance) are dichotomized as follows: gender (0 ¼ female, 1 ¼ male); race (0 ¼ nonwhite, 1 ¼ white); income (0 ¼ low income, 1 ¼ high income).

Table 3. Results for logistic models predicting substance use over 30 days.

BD estimate (SE) BD ORs BD p values CS estimate (SE) CS ORs CS p values MS estimate (SE) MS ORs MS p values

Demographics Age 0.120 (0.052) 1.013 0.021 0.057 (0.067) 1.059 0.395 �0.206 (0.058) 0.814 <0.001 Sex 0.437 (0.185) 1.548 0.018 0.493 (0.238) 1.637 0.038 0.267 (0.190) 1.306 0.159 Race 1.129 (0.210) 3.093 <0.001 0.895 (0.336) 2.447 0.008 �0.434 (0.224) 0.648 0.052 Religion importance 0.025 (0.088) 1.025 0.781 �0.236 (0.118) 0.790 0.045 �0.258 (0.094) 0.773 0.006

Perceived risk of harm Binge drinking �0.584 (0.129) 0.550 <0.001 – – – – – – Cigarette smoking – – – �0.431 (0.032) 0.650 0.044 – – – Marijuana smoking – – – – – – �0.837 (0.123) 0.433 <0.001

Note. Means are represented for each of the three substances: BD: binge drinking, CS: cigarette smoking, and MS: marijuana/hashish smoking. SE: standard error; ORs: odds ratios.

4 M. HANAUER ET AL.

perceived risk of harm from cigarettes had lower rates of reported cigarette smoking. However, as age increased, par- ticipants had higher rates of reported cigarette smoking. In Table 3, for the main effects of the logistic model, perceived risk of harm, race (white), and religious importance were all significantly related to the reported days a participant smoked cigarettes. As both a participant’s perceived risk of harm from cigarettes and religious importance increased, the odds of reporting cigarette smoking decreased. However, the odds of reporting binge drinking increased for those who identify as male and white. For both the truncated negative binomial and logistic models, no interactions between perceived risk of harm and the demographic varia- bles were statistically significant.

Marijuana smoking

A higher percentage of participants in our sample reported smoking marijuana (36.6%) compared to the national aver- age (22.1%).12 The mean number of days (out of the past 30) participants reported smoking marijuana was 3.80 (SD 7.71). In Table 2, for the main effects of the truncated nega- tive binomial model, perceived risk of harm and age were significantly related to reported days a participant smoked marijuana. As perceived risk from harm of marijuana smok- ing increased, the rate of self-reported days of smoking marijuana decreased. For the logistic model, perceived risk of harm, age, and religious importance were all significantly negatively related to the number of days a participant reported smoking marijuana. In Table 3, for those who have higher perceived risk of harm and higher levels of religious importance, there was a decrease in the odds of self-reported marijuana smoking. The effect that age has on reported marijuana smoking interacted with perceived risk of harm from marijuana smoking (discussed below).

Age significantly moderated the effect of perceived risk of harm from marijuana smoking on days of reported mari- juana smoking. As displayed in Figure 1, younger partici- pants (i.e., one standard deviation below the mean age of 22) who perceived smoking marijuana as less risky were more likely to report smoking marijuana. However, as per- ceived risk increased, there were larger decreases in the

participants’ predicted days smoking marijuana compared to older participants (ie, one standard deviation above the mean age of 22).

Concurrent validity

Given that we used one-item constructs for self-reported perceived risk of harm and self-reported use, standard psy- chometrics such as Confirmatory Factor Analysis and Cronbach’s Alpha are not available.42 Therefore, we decided to establish the concurrent validity, which measures the cor- relation between constructs that should be related and are gathered at the same time point.42 Our main analyses estab- lished the concurrent validity of the measures, because each self-reported perceived risk of harm was statistically significantly related to its corresponding self-reported use. Therefore, there is some evidence that self-reported per- ceived harm of risk and self-reported use are valid measures.

Conclusion

Substance use is a common problem among college stu- dents. National averages for reported binge drinking, cigar- ette smoking, and marijuana smoking among college aged young adults (ages 18–25) in the 30 days prior to data col- lection were 36.9%, 22.3%, and 22.1%, respectively.12

Compared to these national averages, a smaller proportion of participants in the current study’s college student popula- tion reported smoking cigarettes (15.4%), and a larger pro- portion of participants reported smoking marijuana and binge drinking (36.7% and 56.4%, respectively). It is import- ant to note that the age range of the individuals in this study’s sample was 18 to 29, whereas the age range reported in the national data was 18 to 25. Nonetheless, the national data provide a meaningful comparison given the mean (22) and standard deviation (1.73) for age in this study’s sample. These discrepancies (ie, why our study participants reported higher frequencies of binge drinking and smoking marijuana and lower frequencies of cigarette smoking compared to national averages) are unsurprising as this sample came from a college campus, and binge drinking and smoking marijuana have been shown to be more prevalent among college students than noncollege students.42 In contrast, cig- arette smoking is more common among noncollege individ- uals than among college students.43,44 This study did not take place in a state where use of marijuana was legalized or decriminalized, so the higher marijuana use rates are not explained by such policy factors.

The primary objective of the current study was to deter- mine if perceived risk of harm from engaging in substance use behaviors was related to reported substance use among a sample of college students. According to the HBM, if an individual’s perceived risk of harm from engaging in a behavior increases, he/she is less likely to engage in that behavior.45–47 For all three behaviors, as perceived risk of harm increased, self-reported engagement in the behavior decreased, thus confirming the theorized construct of per- ceived risk per the HBM. Based on this finding, which is

Figure 1. Association between perceived risk of harm from smoking marijuana and self-reported marijuana smoking at the mean age, one standard deviation above the mean age, and one standard deviation below the mean age.

JOURNAL OF AMERICAN COLLEGE HEALTH 5

consistent across the behaviors, we suggest that substance use prevention programs delivered on college campuses should develop messages designed to increase the perceived risk of harm from engaging in these behaviors. While this study provides support for the HBM’s hypotheses in relation to binge drinking, smoking cigarettes, and smoking mari- juana in this sample of college students, previously reported research about substance use and the HBM is inconsistent. Some researchers have found similar negative relationships between perceived health risks of substance use and reported substance use,30–33 while others have found no significant relationship.29 The lack of a uniform definition of perceived risk of harm and the limited data focusing on the effect of demographic variables could lead to such discrepancies. Regardless, additional research is needed to gain a better understanding of the perceived risk of harm as a predictor of substance use.

A second objective was to evaluate whether the relation- ship between perceived risk of harm and reported behavior was moderated by demographic factors. Through our analy- ses, we identified one instance of moderation by demo- graphic characteristics: age significantly moderated the effect of risk of perceived harm from smoking marijuana on reported marijuana smoking. Younger participants were both more likely to report smoking marijuana when per- ceived risk of smoking marijuana was low, and they were more responsive to increases in perceived risk levels. In other words, younger college students experienced stronger reductions in reported days of marijuana smoked as their risk increased relative to older college students. Therefore, this finding leads us to suggest that interventions aimed at increasing risk awareness should focus on younger groups of college students to see the greatest reduction in marijuana smoking. Evidence-based practices such as Screening, Brief Intervention, and Referral to Treatment (SBIRT), Motivational Interviewing, and counseling are possible pro- grams to implement with younger college students (e.g., freshman and sophomores).48,49

Limitations

There are several limitations within this study. First, the demographic variables have broad categorizations (e.g., white vs. nonwhite), which could generalize key distinctions among groups. Second, the analyses are limited because the study was cross-sectional, thus precluding our ability to make causal inferences. Third, the data were collected using only self-report measures. While self-report can be a useful tool to track substance use, it has inherent limitations, espe- cially with college students. These limitations include misin- terpretation of questions (that can be minimized with pilot testing), biases, and under-reporting.50,51 In retrospect, a pilot study could have helped minimize any potential mis- understandings of the included measures. Future researchers can look into conducting substance use tests to validate our self-reported findings. Fourth, perceived risk of harm is likely a subjective, multi-dimensional construct rather than an objective variable.52 Due to its complex definition, it is

possible that the single survey question (for each substance use outcome) did not accurately measure an individual’s perceived risk of harm. Fifth, binge drinking is typically defined as four or more drinks for women and five or more drinks for men.53 However, the GPRA classifies binge drink- ing as five or more drinks for everyone, which may under- estimate the prevalence of binge drinking for women. Our sample may reflect this since men reported higher levels of binge drinking than women (2.94 days and 2.38 days, respectively). Sixth, when asking about marijuana use, the GPRA does not distinguish between marijuana and hashish, meaning reported marijuana use includes hashish in our analyses. Seventh, there were some missing data for partici- pant characteristics, but according to the multiple imput- ation analyses and tests of data missing completely at random, the missing data did not influence the findings. In addition to these limitations, we would like to note that it would have been ideal to conduct a pilot study first to test survey items.

Recommendations

In the future, researchers should examine the association between perceived risk of harm and biochemically verified behaviors, rather than relying solely on self-reported behav- iors. If biochemical verification is cost-prohibitive, a bogus pipeline procedure could be used to increase the validity of the self-reported data. In addition, standardized definitions and assessments of perceived risk of harm would allow for further comparisons among studies and increase fidelity in the measurements of perceived risk of harm. Based on our finding, it is possible that interventions designed to increase college students’ perceived risk of harm may result in decreases in binge drinking, smoking cigarettes, and smok- ing marijuana.

Conflict of interest disclosure

The authors have no conflicts of interest to report. The authors confirm that the research presented in this article met the ethical guidelines, including adherence to the legal requirements, of the United States and received approval from the Institutional Review Board of Indiana University - Bloomington.

Funding

This work was funded by the U.S. Department of Health and Human Services Substance Abuse and Mental Health Services Administration, Center for Substance Abuse Prevention, SP277999

References

1. Grant BF, Goldstein RB, Saha TD. Epidemiology of DSM-5 alco- hol use disorder: Results from the national epidemiologic survey on alcohol and related conditions III. JAMA Psychiatry. 2015; 72(8):757–766. doi:10.1001/jamapsychiatry.2015.0584.

6 M. HANAUER ET AL.

2. Walters KS, Bulmer SM, Troiano PF, et al. Substance use, anx- iety, and depressive symptoms among college students. J Child Adolesc Subst Abuse. 2018;27(2):1–9. doi:10.1080/1067828X.2017. 1420507.

3. Regier DA, Farmer ME, Rae DS, et al. Comorbidity of mental disorders with alcohol and other drug abuse. Results from the Epidemiologic Catchment Area (ECA) Study. JAMA. 1990; 264(19):2511–2518. doi:10.1001/jama.1990.03450190043026.

4. Agrawal A, Grant JD, Waldron M, et al. Risk for initiation of substance use as a function of age of onset of cigarette, alcohol and cannabis use: Findings in a Midwestern female twin cohort. Prev Med (Baltim). 2006;43(2):125–128. doi:10.1016/j.ypmed. 2006.03.022.

5. Bolla KI, Brown K, Eldreth D, et al. Dose-related neurocognitive effects of marijuana use. Neurology. 2002;59(9):1337–1343. doi: 10.1212/01.WNL.0000031422.66442.49.

6. Volkow ND, Baler RD, Compton WM, et al. Adverse health effects of marijuana use. N Engl J Med. 2014;370(23):2219–2227. doi:10.1056/NEJMra1402309.

7. Wechsler H, Nelson TF. Binge drinking and the American col- lege student: What’s five drinks? Psychol Addict Behav. 2001; 15(4):287–291. doi:10.1037/0893-164X.15.4.287.

8. Courtney KE, Polich J. Binge drinking in young adults: Data, definitions, and determinants. Psychol Bull. 2009;135(1):142–156. doi:10.1037/a0014414.

9. World Health Organization. Global Status Report on Alcohol and Health; 2014. http://apps.who.int/iris/bitstream/handle/ 10665/112736/9789240692763_eng.pdf?sequence=1.

10. Bala S, Marcos M, Gattu A, et al. Acute binge drinking increases serum endotoxin and bacterial DNA levels in healthy individuals. PLoS One. 2014;9(5):e96864. doi:10.1371/journal.pone.0096864.

11. Chikritzhs TN, Jonas HA, Stockwell TR, et al. Mortality and life- years lost due to alcohol: A comparison of acute and chronic causes. Med J Aust. 2001;174(6):281–284. doi:10.5694/j.1326- 5377.2001.tb143269.x.

12. Ahrnsbrak R, Bose J, Hedden S, et al. Key Substance Use and Mental Health Indicators in the United States: Results from the 2016 National Survey on Drug Use and Health. https://www. samhsa.gov/data/sites/default/files/NSDUH-FFR1-2016/NSDUH- FFR1-%0A2016.htm. Published 2016.

13. Grucza RA, Norberg KE, Bierut LJ. Binge drinking among youths and young adults in the United States: 1979-2006. J Am Acad Child Adolesc Psychiatry. 2009;48(7):692–702. doi:10.1097/ CHI.0b013e3181a2b32f.

14. Jones SE, Oeltmann J, Wilson TW, et al. Binge drinking among undergraduate college students in the United States: Implications for other substance use. J Am Coll Health. 2001;50(1):33–38. doi: 10.1080/07448480109595709.

15. NIDA. Tobacco/Nicotine and E-Cigs. National Institute on Drug Abuse website. https://www.drugabuse.gov/drugs-abuse/tobacco- nicotine-e-cigs. Accessed August 14, 2018.

16. Breslau N, Peterson EL. Smoking cessation in young adults: Age at initiation of cigarette smoking and other suspected influences. Am J Public Health. 1996;86(2):214–220. doi:10.2105/AJPH.86.2. 214.

17. Ambrose JA, Barua RS. The pathophysiology of cigarette smok- ing and cardiovascular disease: An update. J Am Coll Cardiol. 2004;43(10):1731–1737. doi:10.1016/j.jacc.2003.12.047.

18. Valdes AM, Andrew T, Gardner JP, et al. Obesity, cigarette smoking, and telomere length in women. Lancet. 2005;366(9486): 662–664. doi:10.1016/S0140-6736(05)66630-5.

19. U.S. Department of Health and Human Services. The Health Consequences of Smoking — 50 Years of Progress A Report of the Surgeon General; 2014. www.cdc.gov/tobacco. Accessed August 14, 2018.

20. Levine PB, Gustafson TA, Velenchik AD, et al. More bad news for smokers? The effects of cigarette smoking on wages. ILR Rev. 1997;50(3):493–509. doi:10.1177/001979399705000307.

21. Harrison ELR, Hinson RE, McKee SA. Experimenting and daily smokers: Episodic patterns of alcohol and cigarette use. Addict Behav. 2009;34(5):484–486. doi:10.1016/j.addbeh.2008.12.013.

22. Gubner NR, Thrul J, Kelly OA, et al. Young adults report increased pleasure from smoking cigarettes when drinking alco- hol but not when using marijuana. Addict Res Theory. 2018; 26(1):71–76. doi:10.1080/16066359.2017.1311877.

23. Miller NS, Gold MS. Comorbid cigarette and alcohol addiction: Epidemiology and treatment. J Addict Dis. 1998;17(1):55–66. doi: 10.1300/J069v17n01_06.

24. Arria AM, Caldeira KM, Bugbee BA, et al. Marijuana use trajec- tories during college predict health outcomes nine years post- matriculation. Drug Alcohol Depend. 2016;159:158–165. doi:10. 1016/j.drugalcdep.2015.12.009.

25. Janz N, Champion VL, Strecher VJ. The Health Belief Model In: K Glanz B, Rimer FL, eds. Health Behavior and Health Education: Theory, Research, and Practice. 3rd ed. San Francisco: John Wiley and Sons, Inc.; 2002:45–66.

26. Brewer NT, Chapman GB, Gibbons FX, et al. Meta-analysis of the relationship between risk perception and health behavior: The example of vaccination. Heal Psychol. 2007;26(2):136–145. doi:10.1037/0278-6133.26.2.136.

27. Edmonds E, Turner LW, Usdan SL. Osteoporosis knowledge, beliefs, and calcium intake of college students: Utilization of the health belief model. Open J Prev Med. 2012;02(01):27–34. doi:10. 4236/ojpm.2012.21005.

28. Downing-Matibag TM, Geisinger B. Hooking up and sexual risk taking among college students: A health belief model per- spective. Qual Health Res. 2009;19(9):1196–1209. doi:10.1177/ 1049732309344206.

29. Carpenter CJ. A meta-analysis of the effectiveness of health belief model variables in predicting behavior. Health Commun. 2010; 25(8):661–669. doi:10.1080/10410236.2010.521906.

30. Sharifirad G, Charkazi A, Moodi M, et al. Factors affecting cigarette smoking based on health-belief model structures in pre- university students in Isfahan, Iran. J Educ Health Promot. 2014; 3(1):23. doi:10.4103/2277-9531.127614.

31. Halpern-Felsher BL, Biehl M, Kropp RY, et al. Perceived risks and benefits of smoking: Differences among adolescents with dif- ferent smoking experiences and intentions. Prev Med (Baltim). 2004;39(3):559–567. doi:10.1016/j.ypmed.2004.02.017.

32. Hampson SE, Severson HH, Burns WJ, et al. Risk perception, per- sonality factors and alcohol use among adolescents. Pers Individ Dif. 2001;30(1):167–181. doi:10.1016/S0191-8869(00)00025-8.

33. Wolburg JM. The “Risky Business” of binge drinking among col- lege students: Using risk models for psas and anti-drinking cam- paigns. J Advert. 2001;30(4):23–39. doi:10.1080/00913367.2001. 10673649.

34. Denering LL, Spear SE. Routine use of screening and brief inter- vention for college students in a University Counseling Center. J Psychoactive Drugs. 2012;44(4):318–324. doi:10.1080/02791072. 2012.718647.

35. Substance Abuse and Mental Health Services Administration. National Minority SA/HIV Prevention Initiative Adult Questionnaire. 2016. https://www.samhsa.gov/sites/default/files/ adult-questionnaire.pdf.

36. Darby K, Kinnevy SC. GPRA and the development of perform- ance measures. J Evid Based Soc Work. 2010;7(1-2):5–14. doi:10. 1080/15433710903175833.

37. Clark TT, Belgrave FZ, Nasim A. Risk and protective factors for substance use among Urban African American adolescents con- sidered high-risk. J Ethn Subst Abuse. 2008;7(3):292–303. doi:10. 1080/15332640802313296.

38. Nasim A, Utsey SO, Corona R, et al. Religiosity, refusal efficacy, and substance use among African-American adolescents and young adults. J Ethn Subst Abuse. 2006;5(3):29–49. doi:10.1300/ J233v05n03_02.

39. Cameron AC, Trivedi P. Regression Analysis of Count Data. Cambridge: Cambridge University Press; 2013. www.cambridge. org. Accessed August 14, 2018.

JOURNAL OF AMERICAN COLLEGE HEALTH 7

40. Long JS, Freese J. Regression Models for Categorical Dependent Variables Using Stata. Third. College Station: Stata Press; 2014.

41. Aiken LS, West SG. Multiple Regression: Testing and Interpreting Interactions - Leona S. Aiken, Stephen G. West, Raymond R. Reno - Google Books; 1991. https://books-google-com.proxyiub.uits.iu.edu/ books?hl=en&lr=&id=LcWLUyXcmnkC&oi=fnd&pg=PR9&dq¼ Aiken,+L.+S.,+West,+S.+G.,+%26amp%3B+Reno,+R.+R.+(1991). þMultiple+regression:+Testing+and+interpreting+interactions.þ Thousand+Oaks,+CA:+Sage.&ots=foegXcYRZc&sig. Accessed August 14, 2018.

42. Kline P. Handbook of Psychological Testing. 2nd ed. New York City: Routledge; 2000.

43. Courtney KE, Polich J. Binge drinking effects on EEG in young adult humans. Int J Environ Res Public Health. 2010;7(5): 2325–2336. doi:10.3390/ijerph7052325.

44. Johnston LD, O’malley PM, Miech RA, et al. Monitoring the Future National Survey Results on Drug Use, 1975-2016: Overview, Key Findings on Adolescent Drug Use. Ann Arbor, MI: The University of Michigan Institute for Social Research; 2017. http://www.monitoringthefuture.org/pubs/monographs/mtf-over- view2016.pdf. Accessed August 14, 2018.

45. Rosenstock IM. Why people use health services. Milbank Q. 1966;83(4):94–124. doi:10.2307/3348967.

46. Salloway JC, PLetcher WR, Collins JJ. Sociological and social psychological models of compliance with prescribed regimen: In search of a synthesis. Sociol Symp. 1978;23:100–121.

47. Glanz K, Lewis FM, Rimer B. Health Behavior and Health Education: Theory and Practice. San Francisco: Jossey-Bass; 1990.

48. Mitchell SG, Gryczynski J, O’Grady KE, et al. SBIRT for adoles- cent drug and alcohol use: Current status and future directions. J Subst Abuse Treat. 2013;44(5):463–472. doi:10.1016/j.jsat.2012. 11.005.

49. Dunn C, Deroo L, Rivara FP. The use of brief interventions adapted from motivational interviewing across behavioral domains: A systematic review. Addiction. 2001;96(12):1725–1742. doi:10.1046/j.1360-0443.2001.961217253.x.

50. Williams RJ, Nowatzki N. Validity of adolescent self-report of substance use. Subst Use Misuse. 2005;40(3):299–311. doi:10. 1081/JA-200049327.

51. Brener ND, Billy JOG, Grady WR. Assessment of factors affect- ing the validity of self-reported health-risk behavior among ado- lescents: Evidence from the scientific literature. J Adolesc Heal. 2003;33(6):436–457. doi:10.1016/S1054-139X(03)00052-1.

52. Greene K, Brinn LS. Messages influencing college women’s tan- ning bed use: Statistical versus narrative evidence format and a self-assessment to increase perceived susceptibility. J Health Commun. 2003;8(5):443–461. doi:10.1080/713852118.

53. National Institute of Alcohol Abuse and Alcoholism. Drinking Levels Defined. https://www.niaaa.nih.gov/alcohol- health/overview-alcohol-consumption/moderate-binge-drinking. Published 2004.

8 M. HANAUER ET AL.