Unhealthy Behavior Clustering and Mental Health Status in United States College Students

Abstract

Objective:

Examine the association of health risk behavior clusters with mental health status among U.S. college students.

Participants:

105,781 U.S. college students who completed the Spring 2011 National College Health Assessment.

Methods:

We utilized latent class analysis to determine clustering of health risk behaviors (alcohol binge drinking, cigarette/marijuana use, insufficient physical activity, and fruit/vegetable consumption), and chi-square analysis and ANOVA to examine associations between class membership and mental health (mental health diagnoses, psychological symptoms, self-injurious thoughts/behaviors).

Results:

Three classes were identified with differing rates of binge drinking, substance use, and insufficient physical activity, but similar rates of insufficient fruit/vegetable consumption. Students classified with the highest rates of binge drinking and cigarette/marijuana use had the highest rates across all mental health variables compared to other classes.

Conclusions:

Students who reported engaging in multiple health risk behaviors, especially high alcohol and cigarette/marijuana use, were also more likely to report poorer mental health.

Introduction

The transition to higher education presents a risky period for young adults as there are increased opportunities for initiating and establishing unhealthy behaviors.^1,2 A large proportion of college students do not meet health recommendation guidelines for multiple health behaviors.³ College students often demonstrate a high prevalence of smoking,⁴ sedentary behavior,⁵ binge drinking,⁶ substance use,⁷ and poor dietary habits.⁸ For instance, among college students, nearly one-quarter of students reported using tobacco products in the last 30 days;⁹ up to one-half reported not meeting criteria for adequate physical activity;^5,10 over one-third reported binge drinking in the past two weeks;¹¹ over one-quarter reported marijuana use in the last 30 days;¹² and over three-quarters of students reported not consuming the recommended five or more cups of fruits and vegetables per day during the last month.⁵ Furthermore, the tendency to engage in unhealthy behaviors during college are likely to continue as students transition out of college into adulthood. For instance, in a sample of college students who were surveyed at the beginning of college and then again at follow-up four years later, 87% of daily smokers and 50% of occasional smokers were found to still be smoking cigarettes.¹³ Students with tobacco use early in college were also found to have heavier alcohol use up to 4–6 years later.¹⁴ Therefore, college is an important time to address these unhealthy but modifiable lifestyle behaviors among young adults.

Health risk behaviors and mental health status

College is also an especially vulnerable period for the onset of mental health problems.^15,16 In an international survey conducted by the World Health Organization, 20% of college students reported having a mental health disorder in the prior 12 months.¹⁷ Individuals with mental health problems experience higher rates of medical problems compared to the general population with contributing unhealthy behaviors, such as tobacco use and poor diet, often originating during adolescence or young adulthood.¹⁸ We review below the literature examining associations between multiple health risk behaviors (i.e., substance use, physical activity, poor diet) and mental health outcomes among college students.

Substance use and mental health.

College students are at a particularly high risk of substance use,¹⁹ with alcohol being the most commonly used substance.²⁰ In general, higher rates of substance use has been associated with poorer mental health.^21,22 For example, alcohol use has been frequently linked with greater likelihood for depression and mental health difficulties.^23–25 Tobacco use has also been associated with increased risk for mental health problems.²⁶ A meta-analysis of cross-sectional studies studying cigarette use and depression found cigarette smokers were nearly twice as likely to have depression than never or former smokers.²⁷ Among college students, tobacco product users have been shown to be more likely to report a mental health diagnosis, increased stress, as well as greater depressive symptoms than non-users.²⁸ Marijuana use is also becoming increasingly relevant as a health risk behavior among college students due to its increasing prevalence among U.S. college students and association with poor mental health.²⁹ For instance, frequent marijuana use has been associated with depression, anxiety, and substance use problems among college students.^30,31

Physical activity and mental health.

Among two samples of U.S. college students, participants who engaged in moderate to vigorous physical activity were less likely to report poorer mental health, including depression and stress.^32,33 A systematic review that examined the association between sedentary behavior and mental health among adolescents also found a strong consistent relationship between low rates of physical activity and psychological distress.³⁴

Diet and mental health.

College students with poor diet are consistently associated with having increased mental health issues. For instance, among college students in the United Kingdom (UK), greater consumption of “unhealthy foods”, including snacks, sweets, and fast food, was associated with higher depressive symptoms and reported stress.³⁵ Additionally, in a sample of college students across three European countries, decreased intake of fruits and vegetables were associated with increased depressive symptoms.³⁶ Similar results were found in a sample of college students in China, in which decreased consumption of fresh fruit was found to be correlated with increased depressive symptoms and stress.³⁷

Health risk behavior clustering and mental health

Generally, these health risk behaviors frequently occur together to contribute to increased risk of chronic illnesses and premature mortality,³⁸ including cancer.³⁹ Multiple unhealthy behaviors can be more detrimental to one’s health than a single behavior.^40,41 Although individual health behaviors are associated with mental health problems, there is limited research addressing the association between patterns of multiple unhealthy behavior and mental health outcomes in college students, even though health risk behaviors often co-occur and overlap.^38,42,43 Three studies have examined the clustering of health risk behaviors and mental health status among college samples: one from China,⁴⁴ one from the UK,⁴⁵ and one from Canada.⁴⁶ Findings from these studies showed that students with greater numbers of unhealthy behaviors had increased risks for depression and anxiety as well as greater self-perceived psychological stress.^44–46 However, these studies were geographically limited and consisted of small samples (i.e., 410 to 2,422 students) from one or two institutions that reduced generalizability of results. To date, no previous research has examined the relationship between the clustering of unhealthy lifestyle behaviors and mental health problems in a large national sample of students attending college in any country.

Using the American College Health Association National College Health Assessment (ACHA-NCHA), we examined the association between health risk behavior clustering and mental health status in a large, national sample of U.S. college students. Consistent with past studies,^9,44–46 we hypothesized that clusters of students engaging in more health risk behaviors will be more likely to report increased mental health distress. An important strength of the NCHA is its focus on multiple health behaviors: physical activity; fruit and vegetable consumption; and alcohol, cigarette, and marijuana use. In addition, the ACHA-NCHA enabled evaluation beyond general psychological stressors that have been reported in previous studies,^44,45 providing an opportunity to examine mental health diagnoses, multiple psychological symptoms, and suicidal thoughts or behaviors among a diverse sample.

Methods

Study Description

This study used data from the National College Health Assessment (NCHA) collected by the American College Health Association (ACHA; http://www.acha-ncha.org/). The NCHA is a national survey that assesses attitudes, behavior, and health among U.S. college students. The NCHA has been administered twice a year since 2000, and is established in its reliability and validity among U.S. college students by the ACHA.⁴⁷ Human subjects research approval was obtained through each of the institutions who participated in the survey. The present analysis included data gathered during the Spring 2011 wave, which consisted of 105,781 students from 129 institutions across the U.S. We chose to utilize the Spring 2011 wave due to availability of the dataset and being the second largest sample size in the history of NCHA administration.

Measures

Demographics.

Participant demographics were obtained through self-report including gender, age, race, ethnicity, enrollment status (e.g., full-time), and their class year. Institution data was also gathered by the ACHA regarding campus size, the region of the U.S. that the college was in, and whether the college was public or private.

Health Risk Behavior.

We focused on five health behaviors measured by the NCHA: cigarette and marijuana use, alcohol binge drinking, physical inactivity, and insufficient fruit and vegetable consumption. Cigarette use, binge drinking, physical inactivity, and insufficient fruit and vegetable consumption were chosen for the current study due to their known associations with chronic illness and disease.^38,40 Marijuana use was included given the rapidly increasing prevalence of marijuana use among U.S. college students.^12,29

Cigarette and marijuana use.

Using the question stem of “within the last 30 days, on how many days did you use,” participants reported their cigarette and marijuana use (i.e., pot, weed, hashish, or hash oil) separately. Response options included: (1) Never used; (2) Have used, but not in the last 30 days; (3) 1–2 days; (4) 3–5 days; (5) 6–9 days; (7) 10–19 days; (8) 20–29 days; or (9) used daily. Responses were coded as positive for cigarette use or marijuana use if they reported using at least 1–2 days during the last month. Participants who had never used or not used within the last 30 days were considered as non-users respectively for each substance.

Alcohol binge drinking.

Participants were asked: “Over the last two weeks, how many times have you had five or more drinks of alcohol at a sitting,” where one drink of alcohol was defined in the survey as “a 12 oz. can or bottle of beer or wine, 4 oz. glass of wine, or 1.5 oz. shot of liquor straight or in a mixed drink.” One standard drink of alcohol contains about 14 grams of pure alcohol based on the National Institute on Alcohol Abuse and Alcoholism definitions.⁴⁸ Response options included: (1) N/A, don’t drink; (2) None; (3) 1 time; (4) 2 times; (5) 3 times; (6) 4 times; (7) 5 times; (8) 6 times; (9) 7 times; (10) 8 times; (11) 9 times; and (12) 10 or more times. In our examination of binge drinking, we included three categories: People who do not consume alcohol (i.e. N/A); people who consume alcohol, but did not report binge drinking in the last 2 weeks (i.e. none); and people who reported binge drinking in the last 2 weeks (i.e. at least 1 or more times). This analysis focused on binge drinking specifically rather than general alcohol use as it is uniquely relevant to college students and campuses as well as incurs greater risk for poor health outcomes when compared to light or moderate alcohol use.⁶

Physical inactivity.

Participants were asked: “On how many of the past 7 days did you: (1) Do moderate-intensity cardio or aerobic exercise (caused a noticeable increase in heart rate, such as a brisk walk) for at least 30 minutes? (2) Do vigorous-intensity cardio or aerobic exercise (caused large increases in breathing or heart rate, such as jogging) for at least 20 minutes?” Response options ranged individually from zero to seven days. We categorized physical inactivity dichotomously with insufficient physical inactivity as either having less than three days of vigorous exercise for at least 20 minutes or less than five days of moderate exercise for at least 30 minutes, based on current U.S. physical activity guidelines published by the U.S. Department of Health and Human Services.⁴⁹

Fruit and vegetable consumption.

Participants were asked about nutrition with the question: “How many servings of fruits and vegetables do you usually have per day?” Participants were provided with a definition of serving size in the survey, with one serving equal to “1 medium piece of fruit; ½ cup of chopped, cooked, or canned fruits or vegetables; ¾ cup of fruit or vegetable juice; 1 cup salad greens; or ¼ cup of dried fruit.” Response options included: (1) 0 servings per day, (2) 1–2 servings per day, (3) 3–4 servings per day, and (4) 5 or more servings per day. We examined fruit and vegetable consumption as a dichotomous variable with insufficient fruit and vegetable intake categorized as less than five servings of fruits and vegetables per day, based on U.S. national dietary guidelines.⁵⁰

Mental Health.

We examined several mental health variables separately, including mental health diagnoses, psychological symptoms, and the presence of suicidal or self-injurious behaviors.

Mental health diagnosis.

A single-item question asking participants whether a professional had diagnosed or treated them for a mental health disorder during the past 12 months was used to measure this dependent variable. Disorders included anxiety, bipolar, depression, and substance-related disorders. The question was phrased as: “Within the last 12 months, have you been diagnosed or treated by a professional for any of the following?” Response options for each condition included: (1) No; (2) Yes, diagnosed but not treated; (3) Yes, treated with medication; (4) Yes, treated with psychotherapy; (5) Yes, treated with medication and psychotherapy; and (6) Yes, other treatment. A mental health condition was considered to be present if one or more diagnoses by a professional (treated or untreated) were reported.

Psychological symptoms during past 30 days and past 12 months.

Eight self-report items were used for the two dependent variables measuring psychological symptoms during the past 30 days and past 12 months. Using the question stem of “Have you ever felt,” participants reported whether they had experienced eight psychological symptoms during the past 30 days and again for the past 12 months. Examples of items included: “things were hopeless, overwhelmed by all you had to do, very lonely, so depressed that it was difficult to function, and overwhelming anxiety.” A response of yes was coded 1 and a response of no was coded 0. A mean score (0–8) was then calculated for both time intervals with higher scores indicating the student self-reported experiencing a greater number of psychological symptoms.

Suicidal and self-injurious behaviors.

Participants responded to three separate items about suicidal or self-injurious behaviors using the time periods responses as mentioned above for the psychological symptom score. We examined each question separately for suicidal ideation (i.e., “Have you ever seriously considered suicide”), suicidal attempt (i.e., “Have you ever attempted suicide”), and one item examining self-injurious behaviors (i.e., “Have you ever intentionally cut, burned, bruised, or otherwise injured yourself”).

Statistical Analysis

Latent class analysis (LCA) is a multivariate statistical method which organizes multiple dimensions of behavior to allow for further analysis and interpretation of higher order interactions among multiple risk behaviors to identify distinct risk subgroups or clusters, providing information that may not be possible with traditional statistical methods.^51,52 LCA has been used in numerous studies to identify latent health behavioral patterns, and identifies subgroups unobserved in the population but manifest in data to help explain risk behaviors.^53,54 Thus, LCA can be used to reduce complex multivariate variables into meaningful parsimonious variables or latent classes in order to optimize an assumed probability model,⁵⁵ and allow us to understand complex behavior patterns and identify subgroups that are most at-risk within a population, as we have done previously using ACHA-NCHA data.⁹ To determine the number of clusters, we began with a two-class model and successively increased the number of classes by one. At each step, we fitted a new LCA model to the data, and continued adding classes until we identified the simplest model that provided an adequate fit. The number of clusters were determined using the Bayesian Information Criterion (BIC) and the entropy value.^55–57 The BIC corresponds to the model which is the most plausible for the data and is ideal for large samples.⁵⁵ Entropy is an indication of the delineation of classes within the data.⁵⁷ In LCA, local independence is assumed (i.e., health behaviors are independent from each other within each class).

We conducted all analyses in SPSS and MPlus software.^58,59 First, descriptive analyses of the five health behaviors were conducted to characterize the overall sample, including student demographics and institution characteristics. Second, LCA was used to identify distinct clusters of the five health behaviors. Several of the variables of interest, including all health risk behaviors and mental health diagnosis, were dichotomized in order to achieve the most parsimonious model. Third, after identifying classes, chi-square analyses were used for dichotomous variables (e.g., gender, race), while analysis of variance (ANOVA) was used for continuous variables (e.g., age) to examine demographic difference between the identified classes. For significant differences found between classes, post-hoc analyses using z-test with Bonferroni adjusted p-values were used. If the ANOVA analysis showed a significant result, post-hoc analyses were conducted using Bonferroni correction to determine where these differences lay. The Phi correlation coefficient (φ) was used as a measure of effect size for chi-square analysis, where .10 is a small, .30 is a medium, and .50 is a large effect size.⁶⁰ For ANOVAs, effect size was measured by Eta Squared (η²) for ANOVAs and standard conventions are .01 for small, .06 for medium and .14 for large effect sizes.⁶¹

Results

Clustering of Health Risk Behaviors

We compared model fit indices for one to five latent classes to select the most appropriate model of health risk behaviours. The solution involving five latent classes could not be sufficiently identified, and was therefore deemed inappropriate. The BIC indices decreased from the two-class solution (571, 534.20) to the four-class solution (569, 212.68) suggesting a slightly superior fit for four-classes,⁶² however, when examining entropy values there was a decrease from three to four classes (0.756 and 0.535). Entropy values approaching one indicate a clear delineation of classes,⁵⁷ therefore the three-class model was more easily identified and selected as the most appropriate fit for the data.

Students belonging to Class 1 (23.5% of the sample) had the highest rates of binge drinking and the highest rates of cigarette smoking and marijuana use (“High Alcohol and Drug Use”). In Class 2 (41.2% of the sample), none of the students met criteria for sufficient levels of physical activity (i.e., “Low Physical Activity”). Additionally, students in the Low Physical Activity class reported low rates of alcohol binge drinking and low rates of cigarette smoking and marijuana use. Students belonging to Class 3 (35.3% of the sample) also presented with low rates of binge drinking and drug use; additionally, all students met criteria for sufficient physical activity (i.e., “Low Risk Behavior”). However, notably, the students in the Low Risk Behavior class did present with similar rates of insufficient fruit and vegetable consumptions to the other two classes (all classes over 90%). The significant differences between classes for binge drinking and drug use had large effect sizes, whereas the physical activity had a small effect size (Table 1).

Table 1.

Health risk behaviors overall and by class

	Overall	Classes			χ2	Effect size (Φ)	Post Hoc
		1 (High Alcohol and Drug Use)	2 (Low Physical Activity)	3 (Low Risk Behavior)
N (%)	105,781	24,824 (23.5%)	43,607 (41.2%)	37,350 (35.3%)
Insufficient Fruit and Vegetable Consumption (%)	93.8	94.2	96.7	90.3	1411.53*	.11	3<1, 2; 1<2
Binge Drinking (%)	34.0	66.2	20.3	28.5	18262.72*	.42	2,3<1; 2<3
Insufficient Physical Activity (%)	53.7	54.5	100	0	79456.00*	.15	1<2
Cigarette Use (%)	15.1	62.3	0.8	0.4	56321.79*	.40	1<2, 3; 3<2
Marijuana Use (%)	15.7	66.1	0.3	0.2	62109.24*	.43	1<2, 3; 3<2

Notes: Insufficient Fruit and Vegetable Consumption: less than 5 servings of fruits/vegetables per day; Binge Drinking: number of times with five or more drinks of alcohol in a single sitting during the prior two weeks; Insufficient Physical Activity: less than three days of vigorous exercise for at least 20 minutes, or less than five days of moderate exercise for at least 30 minutes. 0.04% in Class 1, 41.4% in Class 2 and 27.1% in Class 3 don’t drink alcohol.

^*p<.001

Demographic Differences among Classes

The High Alcohol and Drug Use class had the greatest proportion of men, and students identifying themselves as multiracial, upperclassman (i.e., 3^rd and 4^th year or more) undergraduates, on campuses with 10,000 to 19,999 students, from the Northeast or South region of the U.S., and from private college students, as compared to the other classes. The Low Physical Activity class had the greatest proportion of women, racial minorities (i.e., Asian, Black, and Hispanic), part-time students, and from campuses with over 20,000 students, as compared to the other classes. The Low-Risk Behavior class had the greatest proportion of students from campuses with less than 10,000 students, freshman (i.e., 1^st year) students, and from the Midwest region, as compared to the other classes. Table 2 also shows that the size of the effect stemming from differences in demographic characteristics were considered small (≤0.15).

Table 2.

Demographic characteristics overall and by class

	Overall	Classes			F or χ2 (Effect size)	Post hoc
		1 (High Alcohol and Drug Use)	2 (Low Physical Activity)	3 (Low Risk Behavior)
N (%)	105,781	24,824 (23.5%)	43,607 (41.2%)	37,350 (35.3%)
Gender (%)					670.74* (Φ= .10)
Female	65.4	59.2	69.1	65.2		1,3<2 1<3
Male	34.6	40.8	30.9	34.8		2,3<1 2<3
Age in years (Mean, SD)	22.67 (5.90)	22.12 (4.80)	23.32 (6.68)	22.27 (5.90)	425.66* (η2=.70)
Race/ethnicity (%)					2210.34* (Φ= .15)
White	68.6	73.2	62.4	72.7		2<1,3
Asian	9.9	6.0	13.8	8.0		1,3<2 1<3
Black	4.8	3.1	6.2	4.2		1,3<2 1<3
Hispanic	5.8	5.4	6.6	5.2		1,3<2
American Indian	0.7	0.7	0.7	0.7		NS
Multiple race	6.9	8.3	6.5	6.5		2,3<1
No response	1.8	1.8	2.2	1.4		1,3<2
Other	1.4	1.5	1.6	1.3		3<2
Enrollment status (%)					270.77* (Φ= .06)
Full-time	91.6	93.0	90.2	92.5		2<1,3
Part-time	7.7	6.4	9.0	7.0		1,3<2 1<3
Other	0.7	0.6	0.8	0.6		1,3<2
Campus size (%)					262.27* (Φ= .08)
<10 000	37.1	36.1	23.4	39.1		1,2<3
10 000 to 19 999	23.4	25.1	22.1	23.9		2,3<1 2<3
20 000+	39.5	38.6	42.1	37.0		1,3<2 3<1
Class year (%)					685.310* (Φ= .08)
1st year	20.9	20.4	19.8	22.6		1,2<3
2nd year	20.0	21.0	18.9	20.8		2<1,3
3rd year	19.8	21.0	19.3	19.5		2,3<1
4th or more year	21.1	23.3	20.8	20.1		2,3<1
Graduate/ professional	17.2	13.6	20.2	16.1		1<2,3 3<2
Other	0.9	0.7	1.1	0.9		1<2,3
Region (%)					476.16* (Φ= .07)
Northeast	37.0	41.1	37.5	33.8		2,3<1 3<2
Midwest	17.2	15.0	16.7	19.1		1,2<3 1<2
South	22.6	23.1	22.1	22.7		2,3<1
West	23.3	20.6	23.8	24.4		1<2,3
College type (%)					51.90* (Φ= .02)
Public	59.0	57.0	59.3	59.8		1<2,3
Private	41.0	43.0	40.7	40.2		2,3<1

Notes: NS: Not significant. Information on “Other” not available in survey.

^*p<.001

Between Class Differences in Mental Health Functioning

The proportion of individuals who reported mental health diagnoses, symptoms, and history of self-injurious thoughts or behaviors differed significantly among classes (see Table 3).

Table 3.

Differences in mental health overall and by class

	Overall	Classes			F or χ2 (Effect size)	Post hoc
		1 (High Alcohol and Drug Use)	2 (Low Physical Activity)	3 (Low Risk Behavior)
N (%)	105,781	24,824 (23.5%)	43,607 (41.2%)	37,350 (35.3%)
Mental health diagnosis (%)
Past 12 months	21.3	30.1	19.4	17.7	1453.730* (Φ= .12)	2,3<1
Psychological Symptom Score (mean, SD)
Past 30 days	2.95 (2.45)	3.37 (2.57)	2.93 (2.44)	2.68 (2.35)	575.09* (η2= .02)	2,3<1 3<2
Past 12 months	2.67 (2.74)	3.45 (3.04)	2.66 (2.70)	2.28 (2.53)	333.34* (η2= .02)	2,3<1 3<2
Self-Injurious Thoughts or Behaviors (%)
Intentionally cut, burned, bruised, or otherwise injured yourself					1694.60* (Φ= .08)
No, Never	83.0	74.6	85.2	86.1		2,3<1; 2<3
No, not in the last 12 months	11.8	16.9	10.5	9.9		2,3<1;3<2
Yes, in the last 30 days	2.2	3.5	1.9	1.8		2,3<1
Yes, in the last 12 months	2.9	5.0	2.4	2.2		2,3<1
Seriously considered suicide					1243.02* (Φ= .11)
No, Never	80.7	73.3	81.8	84.3		1,2<3; 1<2
No, not in the last 12 months	12.9	17.0	12.4	10.8		2,3<1; 3<2
Yes, in the last 30 days	2.3	3.5	2.1	1.7		2,3<1; 3<2
Yes, in the last 12 months	4.1	6.1	3.7	3.3		2,3<1; 3<2
Attempted suicide					666.59* (Φ= .06)
No, Never	92.5	88.9	93.3	94.0		1,2<3; 1<2
No, not in the last 12 months	6.5	9.2	5.9	5.3		2,3<1; 3<2
Yes, in the last 30 days	0.5	0.6	0.3	0.3		2,3<1
Yes, in the last 12 months	0.7	1.3	0.5	0.5		2,3<1

^*Notes: p<.001

^**p<.05

NS = Not Significant

Students in the High Alcohol and Drug Use class reporting the highest prevalence rates across all mental health functioning categories. Post-hoc analyses showed that students in the High Alcohol and Drug Use class had significantly greater likelihood of reporting a mental health diagnosis in the past 12 months as compared to the other two classes. Psychological symptom scores in the past 30 days and past 12 months were significantly higher among the High Alcohol and Drug Use class and Low Physical Activity class when compared to the Low Risk Behavior class. Post-hoc analyses following chi-square test showed that High Alcohol and Drug Use class had a significantly greater proportion of students reporting self-injurious behavior (3.5% vs. 1.8–1.9%), seriously considering suicide (3.5% vs. 1.7–2.1%), and suicide attempts (0.6% vs. 0.3%) within the last 30 days. As shown in Table 3, the effect sizes for these differences were considered small (≤0.12).

Comment

To our knowledge, no prior studies have examined how the prevalence of mental health status may vary based on health risk behavior clustering among U.S. college students. The current study provided the first examination of the relationship between the clustering of unhealthy lifestyle behaviors and mental health problems in a large national sample of U.S. college students. Using latent class analysis, this study identified three distinct classes of students. The High Alcohol and Drug Use class had high rates of alcohol binge drinking and cigarette/marijuana use; the Low Physical Activity class had high rates of insufficient physical activity, but low levels of alcohol binge drinking and cigarette/marijuana use; and the Low Risk Behavior class showed low rates of all health risk behaviors. While alcohol binge drinking, cigarette smoking, marijuana use, and physical activity varied across classes, all classes presented with similarly low levels of fruit and vegetable intakes (90.3–96.7% not achieving sufficient consumption). These results further emphasize the likelihood of poor diet among college students potentially contributing to future health problems, especially as higher consumption of fruits and vegetables has been associated with lower risk of cancer, stroke, and cardiovascular disease.^63–65

Upon examination of mental health status presented among each class, results demonstrated patterns between clusters of health risk behaviors and mental health status with greater prevalence of health risk behaviors related to higher psychological symptom scores, greater prevalence of mental health diagnoses, and greater prevalance of self-injurious or suicidal behaviors. The High Alcohol and Drug Use class had the greatest proportion of students reporting poor mental health, including mental health diagnoses and self-injurious thoughts or behaviors, and the highest rates for psychological symptoms when compared to the other two classes. However, physical activity may be a protective factor for mental health, as the Low Risk Behavior class, which contains students who all meet criteria for recommended levels of physical activity, also presents with the lowest rates of other health risk behaviors and mental health difficulties. This aligns with previous evidence demonstrating that physical activity may be protective against the development of mental health difficulties,³² especially depression and anxiety.^66,67 Additionally, it should be noted that while the High Alcohol and Drug Use class, which has the highest amount of unhealthy and risky behaviors, is the smallest class, this class still contains almost 25,000 students. As such, it is still important to continue to engage all college students in interventions and education targeting the prevalence and consequences of engaging in health risk behaviors in order to increase disease risk reduction and both physical and mental health promotion.

Health practitioners treating college students not only need to recognize these clusters of health risk behaviors, but also how mental health may play a significant role in the pattern of unhealthy lifestyle behaviors. As a significant number of young adults enroll in post-secondary education, this period may be a critical and opportune time to intervene with treatments that target associated risk factors in individuals experiencing mental illness or distress to result in improvements in both physical and mental health. Educating college students on how health risk behaviors can often present in clusters and the physical and mental consequences of these risky behaviors could make interventions more effective as well as result in improvements in both physical and mental health. Utilization of interventions that are effective in promoting both mental and physical behavior change could help potentially health practitioners, and consequently students, target and manage these frequently co-occurring conditions.⁶⁸ For instance, behavioral activation has been shown effective for the treatment of depressed mood as well as smoking cessation.^69,70 Therapeutic lifestyle interventions targeting both health behaviors and psychological stress management can also be effective in contributing to overall health and well-being,⁷¹ however, these interventions are often underutilized by health professionals.⁷² As individuals with mental health difficulties are more likely to transition from substance use to dependency,⁷³ it is important to target students who present with poorer mental health and increased health risk behaviors, especially as these health behavior patterns may still not yet be established.⁷⁴ While many studies have sought to examine the effectiveness of interventions targeting multiple health risk behaviors,⁷⁵ future studies should examine how health behavior interventions may also improve psychological symptoms and disorders.

Although this study cannot infer the direction of causality between association between health risk behaviors clusters and mental health status due to the cross-sectional nature of the ACHA-NCHA survey, there is evidence that engaging in two or more health risk behaviors moderately increased the risk for poor psychological health at 2-year follow up among adolescents.⁷⁶ For instance, related to smoking and poor mental health, there is evidence for both depression and anxiety being associated with future smoking behavior and vice versa,^27,77,78 especially among adolescents.^79,80 Similarly, individuals may also have increased risky health behaviors to help cope with mental health distress, such as individuals that increase alcohol use to help manage the symptoms of depression.⁸¹ Therefore, future studies utilizing longitudinal assessment could help to clarify the directionality between the presentation of multiple health risk behaviors and mental health diagnoses as well as potential adverse outcomes, including cardiovascular/cancer risk or depressive symptoms.^75,82

Limitations

First, although colleges self-select to participate in the ACHA-NCHA, results from the survey have previously been found comparable to various large national databases of college students,⁸³ and consisted of a large sample of students from 129 institutions across all regions of the U.S. Second, measures were gathered through self-reported surveys and not an objective, quantitative assessment, including the self-report of mental health diagnoses. While no studies have examined the accuracy of self-reported mental health diagnoses among this population, other studies have shown high validity of self-reported diagnoses of depression compared to evaluation through a structured clinical interview by a clinician.⁸⁴ Future studies could also consider obtaining objective measurements, such as utilizing activity trackers or pedometers to monitor activity or collecting biological measures that may provide a more objective measure of the health of the student population. Additionally, as the current study focused on colleges located in the U.S., future studies should examine where there are similar trends in the clustering of the health behaviors and mental health in other countries. Third, a limitation of this study is the small effect size of the significant differences between each class. However, large sample sizes, such as in the present study, are commonly associated with small effect sizes. Despite the small effect sizes, the present study demonstrates important health differences in college students and is a key contribution to the literature. Fourth, the binge drinking measure available through the survey did not allow us to examine binge drinking according to gender-specific definitions between women (i.e., ≥ 4 drinks) and men (i.e., ≥ 5 drinks) by the National Institute of Alcohol Abuse and Alcoholism guidelines.⁸⁵

Conclusions

With more than 20 million individuals enrolled in higher education institutions in the United States and an estimated 34% gross enrollment in tertiary education worldwide,⁸⁶ it is important to address risky health behaviors and mental health issues among this population group. As previous studies have been limited by sample size, geographical area, as well as limited number and relevance of health behaviors and mental health factors studied,^44,45 this study extends current understanding of risky health behavior clustering by examining mental health status among U.S. college students. Our findings reveal a significant relationship between clustering of health behaviors with poorer mental health outcomes; specifically, students with greater likelihoods of engaging in binge drinking, cigarette smoking, and marijuana use were more likely to report having mental health diagnoses, symptoms, and self-injurious thoughts or behaviors. By understanding how poorer mental health may appear in the context of multiple unhealthy lifestyle behaviors in conjunction, these findings highlight the potential for these institutions to create targeted interventions for these specific student groups that may be at higher risk for mental distress.