Smoking and Suicidal Ideation among College Students: Smoking Expectancies as Potential Moderators

Abstract

Objective:

In the present study, we sought to establish a link between suicidal ideation (SI) and smoking in college students, determine whether psychopathology accounted for the association, and determine whether smoking expectancies were moderators.

Participants:

Participants (N = 607) were identified as nonsmokers, exsmokers, infrequent smokers, or daily smokers.

Methods:

Participants were assessed for smoking patterns, smoking expectancies, psychopathology, SI, and past suicide attempts.

Results:

Daily smokers had the highest level of SI. There was a dose-response relationship between smoking and SI, such that the higher the daily smoking rate, the greater the SI, even when controlling for depression, alcohol use, and drug use. Trend-level results indicated that at lower smoking rates, elevated smoking-related negative affect reduction (NAR) expectancies were associated with lower SI, while elevated NAR expectancies in combination with higher smoking rates were associated with greater SI.

Conclusion:

Smoking cessation programs for college students should screen for SI.

Introduction

Cigarette smoking remains the leading cause of preventable death in the United States and is associated with a myriad of adverse health outcomes, including cardiovascular disease, chronic obstructive pulmonary disease, and lung cancer.¹ A less well known and poorly understood association with smoking is morbidity due to suicide. Research shows that current smoking is associated with death by suicide, and the higher the daily smoking rate, the higher the likelihood for completed suicides.^2,3 In fact, a recent meta-analysis indicated that the odds of suicide death were 28% higher for former smokers and 81% higher for current smokers, as compared to nonsmokers.⁴

For obvious reasons, the association between suicidal ideation (SI), versus completed suicides, and smoking is more practical for investigation related to potential interventions for both smoking and suicide. SI is defined as the thoughts of harming or killing oneself and has been identified as one of the major predictors of attempted or actual suicide.^5–8 Related to smoking, past research shows that SI is highest among current smokers, followed by former smokers, and never smokers.⁹ Additionally, current smokers are at risk of experiencing SI as compared to other groups regardless of age, gender, history of psychological symptoms, and alcohol abuse/dependence,⁹ with smoking being identified as a contributing factor to suicide.¹⁰ The association between smoking and SI has not previously been investigated with college student populations, yet suicide is currently the second most common cause of death among college students,¹¹ with depression and substance use identified as risk factors.^12,13 In addition, cigarette smoking is prevalent among college students, with one-third of college-aged students using tobacco products in some way, whether it be cigarettes, chewing tobacco or cigars.^14,15 It is therefore important to establish whether the association between smoking and suicide exists in this population.

Potential mechanisms have yet to be identified that would account for the established association between smoking and suicide. In an overview of the research linking smoking and suicide, Hughes¹⁶ described the three most plausible explanations for the association: (1) smokers have pre-existing, comorbid conditions that increase their risk of suicide; (2) smoking causes painful and debilitating conditions that might lead to suicide; and (3) smoking decreases serotonin and monoamine oxidase (MAO) levels.

The first of these explanations appears compelling, as smoking is prevalent in populations at greater risk for suicide, including individuals diagnosed with substance use or other psychiatric disorders.¹⁷ Indeed, suicidal ideation is a diagnostic feature of Major Depressive Disorder (MDD) as per the Diagnostic and Statistical Manual(DSM),¹⁸ and among other psychiatric disorders, has uniquely predicted suicidal ideation in adults.¹⁹ Previous research has established that one of the underlying mechanisms between smoking and SI is that there is a strong relationship between smoking and depression, as well as smoking and substance use.^20–23 However, as Hughes¹⁶ points out, if pre-existing conditions in smokers accounted for the risk for suicide, that risk should persist even when smoking has stopped. Data from existing studies indicate that acute nicotine withdrawal can be associated with increases in negative affect,^24,25 but that prolonged cessation is associated with reduced negative affect.²⁶ Other studies indicate that suicide risk is lower for former smokers as compared to current smokers, but that former smokers remain at higher risk for suicidal ideation than nonsmokers.^4,9

It does seem that if pre-existing, comorbid disorders accounted for the association between current smoking and suicidal ideation, there should be evidence of self-medicating negative mood with smoking as a maintaining factor of smoking.

The second possible explanation for the association between suicidal ideation and smoking proposed by Hughes¹⁶ is that smoking causes painful and debilitating conditions that might lead to suicide. For example, long-term smoking is associated with increased risk of cancer, which is painful, and there is evidence of cancer being associated with an increased risk of suicide.^27,28 However, there is only limited evidence for smoking as a causal factor in chronic pain conditions (e.g., lower back pain and rheumatoid arthritis).²⁹ Collectively, the painful conditions linked to smoking appear to be distal effects associated with long-term smoking, and therefore it is unclear if this relationship will exist among younger populations. However, similar to the pattern established with adult smokers, smoking among adolescents is also related to SI and suicide attempts.^30–32

The third possible explanation for the association between SI and smoking proposed by Hughes¹⁶ is that smoking decreases serotonin and monoamine oxidase (MAO) levels, and these decreases can lead to depression, impulsivity, and aggression—all of which are related to suicide. Active smoking is associated with lower levels of MAOs as compared to nonsmoking.³³ This relationship is particularly troubling as lower serotonin function is associated with impulsive/aggressive traits, which in turn are often related to suicidal acts.³⁴ However, recent research has identified the MAO inhibiting (MAOI) content of cigarette smoke and its well-established role in mood regulation, most notably its antidepressant effects. Specifically, the beta-carboline alkaloids in tobacco smoke have been found to act as inhibitors of human MAO.³⁵ Although the degree of MAOI activity in cigarette smoke varies with tobacco type,³⁶ the general antidepressant content in cigarette smoke is consistently found, and animal research in this area may shed light on problematic processes in the clinical course of smoking, such as relapse.³⁷

In addition to these postulations, smoking outcome expectancies, the anticipated reinforcing and punishing effects of smoking, may influence the association between smoking and SI. Past research has found strong associations between cigarette use and positive/negative reinforcement outcome expectancies, ^38–40 with a strong belief among smokers that smoking will reduce negative affect. These beliefs about negative affect reduction are elevated in smokers with comorbid depression.⁴¹ In fact, in a study with college student smokers, negative affect regulation expectancies fully mediated the positive relationship between depressive symptoms and level of smoking.⁴¹ It therefore seems likely that this belief would be present in smokers who were self-medicating negative affect related to a pre-existing diagnosis as stated in Hughes’ first postulation. Expectancies for reduction of negative mood would also likely be present in Hughes’ third potential reason in that physiological variables cause negative mood, and smokers smoke in response to negative mood. For the second possible reason proposed by Hughes,¹⁶ knowledge of painful and debilitating conditions related to smoking should be reflected in expectancies regarding smoking-related health risks. Existing research with college students and other adult smokers has established an association with smoking and beliefs about the health risks associated with smoking.^38,39

The present study sought to: (1) establish a link between SI and smoking in college students; (2) determine whether mental health issues (e.g., depression/substance use disorders) account for the association; and (3) determine whether smoking outcome expectancies act as moderators that influence the link between smoking and suicide. It was hypothesized that: (1) SI and smoking would be positively correlated; (2) the positive association between SI and smoking would remain significant while statistically controlling for diagnoses of major depressive disorder (MDD), alcohol use disorder (AUD), and drug use disorder (DUD); (3) smoking outcome expectancies for negative affect reduction and health risks would moderate the influence of smoking on SI. Specifically, we predicted that stronger beliefs that smoking relieves negative affect and heavier smoking patterns would positively predict greater SI, and that lower beliefs in the health risks of smoking and heavier smoking patterns would be associated with greater SI.

Materials and Methods

Participants

Participants (N=607) were recruited through the online registration system for the psychology experiment participant pool and granted course credits for participation. The inclusion criteria included that participants were between 18–24 years old and currently enrolled as an undergraduate student at the university at which the study was being conducted. Participants were primarily female (82.4%) and White (75.1%), with a mean age of 20.1 (SD = 1.9) years.

Procedure

In the present study, we assessed college undergraduates (N = 607) for smoking patterns, smoking-related outcome expectancies, diagnoses of major depression and alcohol/drug use disorder, and SI. Participants who met preliminary inclusion criteria provided informed consent and completed all measures via a secure online survey engine. Due to the psychological sensitivity of the assessment measures, we provided mental health referral information to all participants at the outset of the study. This study and its procedures were approved by the Institutional Review Board of Louisiana State University. Data was collected in 2017–2018 from a large Southeastern university.

Measures

Smoking Status Questionnaire

This form included demographic information (age, sex, and race), and smoking-related variables, such as current and past smoking patterns (“Do you currently smoke cigarettes?” “Do you currently smoke every day?” “How many years have you been smoking?” “Did you used to smoke cigarettes?” “If so, how long has it been since you stopped smoking?” and smoking cessation attempts (i.e., “How many times have you made a serious attempt (at least 24 hours) to quit smoking?” This form also included the Fagerström Test for Nicotine Dependence (FTND) to assess nicotine dependence level in daily smokers.⁴² The FTND was administered to daily smokers only in the present study in order to determine their level of physiological and psychological dependence on nicotine.

Psychiatric Diagnostic Screening Questionnaire (PDSQ)

The PDSQ is a brief, self-report scale designed to screen for the most common DSM Axis I disorders encountered in outpatient mental health settings.⁴³ The PDSQ consists of 126 questions assessing the symptoms of 13 DSM disorders in 5 areas and has demonstrated reliability and validity in assessing DSM disorders.⁴³ The MDD, AUD, and DUD scales were used in the current study because of the association between these disorders and suicidality. The MDD scale comprises 21 items covering the DSM symptom inclusion criteria of the disorder, including 6 items that address suicidality. The AUD scale includes 6 items addressing an individual’s self-perceived problems with drinking, thoughts about cutting down or limiting drinking, and alcohol-related psychosocial problems. The DUD scale also includes 6 items, and they assess self-perceived problems with drug use, thoughts about cutting down or limiting drug use, and drug-related psychosocial problems. Each of these scales yield total scores that range according to the number of items on the scale.

Beck Depression Inventory, 2nd Edition (BDI-II)

The BDI-II is a 21-item, self-report questionnaire for evaluating the severity of depressive symptoms, including SI, in normal and psychiatric populations.⁴⁴ Additionally, the BDI-II has been validated as a reliable measure of self-report depression among college students.⁴⁵ The BDI-II was included in the present study as an additional measure of depression (in addition to the diagnostic information yielded by the PDSQ for MDD), as well as our primary outcome measure of SI per item #9.

Beck Scale for Suicide Ideation (BSS)

The BSS is a 19-item instrument that evaluates the presence and intensity of suicidal thoughts in a week before evaluation.^46,47 The BSS items correlate highly with items that relate to suicide on the BDI-II. For the purposes of the current study, we were interested in the BSS items regarding past suicide attempts and the total score.

Brief Smoking Consequences Questionnaire-Adult (BSCQ-A)

This questionnaire measures smoking outcome expectancies, anticipated rewarding and punishing consequences from smoking a cigarette.⁴⁸ The BSCQ-A comprises 10 factors derived from principal components analysis. The BSCQ-A scales have demonstrated good internal consistency construct validity.⁴⁸ The Negative Affect Reduction (NAR) and Health Risks (HR) subscales were used for the current study. The NAR scale includes items such as, “Smoking calms me down when I feel nervous” and others reflecting smoking to alleviate negative mood. The HR scale contains the items, “By smoking I risk heart disease and lung cancer” and “The more I smoke, the more I risk my health.” The NAR and HR scale scores were used in the moderation analyses in the present study, because these beliefs were hypothesized to moderate the relationship between smoking and SI.

Data Analysis

Those participants who reported never smoking (n = 433) were termed nonsmokers, participants who reported past smoking for at least one year were termed exsmokers (n = 30), participants who reported current non-daily smoking were termed infrequent smokers (n = 111), and participants who smoked daily were termed smokers (n = 33). We conducted analyses of variance (ANOVAs) on baseline measures to identify differences across smoking status groups on continuous outcome variables. We conducted chi-square analyses for dichotomous and categorical outcome variables with the smoking status groups as factors. These variables included the demographics of age, sex, and race in order to establish comparability on these baseline measures among the smoking status groups prior to examining our primary hypotheses. Any baseline differences in demographic variables were subsequently treated as covariates in the regression analyses. Hierarchical linear regression analyses were used to determine whether daily smoking rate would predict SI, while controlling for MDD, AUD and DUD. On the first step, we entered age and race, on the second step, we entered MDD, AUD, and DUD (via PDSQ), and on the third and final step, we entered daily smoking rate. SI (as measured by the BDI-II SI item) was the dependent variable. In order to determine whether smoking expectancies moderated the association between daily smoking rate and SI, we conducted two separate hierarchical linear regression analyses. In the first step of the first regression we entered age and race, on the second step we entered MDD/AUD/DUD, on the third step we entered daily smoking rate and the NAR scale of the BSCQ-A, and on the fourth and final step, we entered the interaction term which was the product of daily smoking rate and the BSCQ-A NAR scale. In the first step of the second regression testing for moderation, we entered age and race, on the second step we entered MDD/AUD/DUD, on the third step we entered daily smoking rate and the HR scale of the BSCQ-A, and on the fourth and final step, we entered the interaction term which was the product of daily smoking rate and the BSCQ-A HR scale.

Results

Most participants (87%) reported no SI, 10.3% reported ideation, 2.0% reported a desire to kill themselves, and .7% reported they would, “kill myself if I had the chance” as measured by responses to the BDI-II. See Table 1 for details regarding cigarette frequencies and dependency (via FTND score).

Table 1.

Participant Characteristics

	Overall (N = 607)	Nonsmokers (n = 433)	Exsmokers (n = 30)	Infrequent smokers (n = 111)	Daily smokers (n = 33)	p
Mean Age (SD)	20.1 (1.9)	20.0 (1.6)a	21.8 (4.0)abc	20.1 (1.6)b	20.5 (2.5)c	<0.001
Female N (%)	500 (82.4%)	366 (84.5%)	22 (73.3%)	87 (78.4%)	25 (75.8%)	0.152
White N (%)	456 (75.1%)	313(72.3%)ab	21 (70.0%)c	92 (82.9%)a	30 (90.9%)bc	0.016
BDI Total
Mean (SD)	10.4 (13.1)	9.1 (12.6)ab	9.0 (7.5)c	14.5(15.1)ac	14.0 (12.8)b	0.001
BDI SI item
Mean (SD)	0.17 (0.48)	0.14 (0.42)ab	0.14 (0.36)c	0.24 (0.56)a	0.42 (0.75)bc	0.003
BSS Total
Mean (SD)	1.6 (2.9)	1.5 (2.5)a	1.5 (1.8)	1.9 (3.7)	2.6 (4.9)a	0.141
Past attempt
Once N (%)	26 (4.6%)	12 (3.0%)a	6 (21.4%)ab	5 (4.6%)b	3 (9.1%)	<0.001
> Once N (%)	7 (1.2%)	3 (0.7%)	1 (3.6%)	2 (1.9%)	1 (3.0%)
MDD N (%)	119(20.6%)	68 (16.7%)ab	5 (17.2%)	36 (32.7%)a	10 (30.3%)b	0.001
AUD N (%)	160(27.8%)	88(21.7%)ab	8 (27.6%)	48 (44.0%)a	16 (48.5%)b	<0.001
DUD N (%)	76 (13.2%)	33 (8.1%)ab	5 (17.2%)	25 (22.9%)a	13 (39.4%)b	<0 .001
BSCQ-A NAR	2.55 (2.95)	1.52 (2.46)abc	5.83 (2.04)ad	4.32 (2.58)bde	6.71 (1.60)ce	<0 .001
BSCQ-A HR	6.33 (3.58)	5.74 (3.95)abc	7.20 (2.08)a	7.84 (1.91)b	7.80 (1.24)c	<0.001

Note. BDI-II = Beck Depression Inventory-II; BSS = Beck Scale for Suicide Ideation; MDD = Major Depressive Disorder; AUD = Alcohol Use Disorder; DUD = Drug Use Disorder. BSCQ-A = Brief Smoking Consequences Questionnaire-Adult; NAR = negative affect reduction; HR = health risks. Subscript letters indicate post-hoc analysis was significant between groups at least at the p < .05 level

Significant differences were observed among participants in age and race across smoking status groups. Smoking status groups significantly differed on BDI-II total scores, such that infrequent and daily smokers reported greater depressive symptomatology. There was also a significant difference across smoking status groups on the BDI-II item regarding SI, such that daily smokers reported the highest level of SI, followed by infrequent smokers, followed by ex- and nonsmokers. Groups showed significant difference on the BSS item regarding past suicide attempts, in that exsmokers were the most likely to report a previous attempt. Finally, infrequent and daily smokers were more likely to meet diagnostic criteria for MDD as compared to ex- and nonsmokers. Rates of AUD were also significantly higher in the two smoking groups (infrequent and daily smokers), as were rates of DUD. There were significant differences across smoking status groups on the BSCQ-A NAR scale, such that daily smokers had the highest scores, followed by exsmokers, infrequent smokers, and then nonsmokers. Finally, there were also significant differences across smoking status groups on the BSCQ-A HR scale, such that daily, infrequent, and exsmokers had the highest scores, followed by nonsmokers. See Table 1 for complete results.

Prediction Analyses with Regression

When all variables were entered, the model was significant, F(6, 606) = 39.74, p <.001, R Square = .284, Adjusted R Square = .277, with results indicating that race, MDD, and daily smoking rate were significant predictors of SI. See Table 2 for summary statistics and results for first two steps of the model, which were also significant.

Table 2.

Daily Smoking Rate Predicts Suicidal Ideation

	Step 1	p	Step 2	p	Step 3	P
Age	−0.038	.347	−0.004	.900	−0.008	.821
Race	0.098*	.016	0.090	.010	0.095	.006
MDD			0.491	<0.001	0.489	<0.001
AUD			0.003	.932	−0.003	.937
DUD			0.075	.051	0.063	.099
Daily smoking rate					0.089	.012
F (df)	3.31(2, 606)	.037	46.01(5, 606)	<0.001	39.74(6, 606)	<0.001
R 2	0.011		0.277		0.284
Adjusted R2	0.008		0.271		0.277

Note. Values in each step are β’s for each variable included within that model. MDD = Major Depressive Disorder; AUD = Alcohol Use Disorder; DUD = Drug Use Disorder.

Expectancies as Moderators

The overall model for the hierarchical linear regression to determine whether NAR expectancies moderated the influence of smoking rate on SI was significant [F(8, 606) = 30.79, p < .001, R Square = .292, Adjusted R Square = .282]. Race, MDD, and daily smoking rate were identified as significant predictors of SI in this population, and the NAR scale * daily smoking rate interaction term approached significance as a predictor (β = −.267, p = .092), indicating a trend for NAR expectancies to moderate the influence of smoking on SI. The moderating influence of NAR expectancies was such that higher daily smoking rates and higher NAR expectancies predicted higher SI. However, at lower daily smoking rates, NAR expectancies predicted lower SI. See Table 3 for complete results, including statistics related to the first three steps of the model.

Table 3.

Hierarchical Regression to Test Negative Affect Reduction Expectancies as a Moderator of Daily Smoking Rate on Suicidal Ideation

	Step 1	p	Step 2	p	Step 3	p	Step 4	P
Age	−0.038	.347	−0.004	.900	0.000	.999	−0.005	.876
Race	0.098*	.016	0.090	.010	0.091	.009	0.089	.010
MDD			0.491	<0.001	0.504	<0.001	0.503	<0.001
AUD			0.003	.932	−0.005	.893	−0.013	.745
DUD			0.075	.051	0.069	.074	0.065	.090
Daily smoking rate					0.113	.003	0.374	.019
BSCQ-A NAR scale					−0.071	.068	−0.068	.080
BSCQ-A NAR *							−0.267	.092
Daily smoking rate	3.313(2, 606)	.037	46.01(5, 606)	<0.001	34.68(7, 606)	<0.001	30.79(8, 606)	<0.001
F (df)	0.011		0.277				0.292
R 2	0.008		0.271		0.280		0.282
Adjusted
R2

Note. Values in each step are β’s for each variable included within that model. MDD = Major Depressive Disorder; AUD = Alcohol Use Disorder; DUD = Drug Use Disorder; BSCQ-A NAR = Brief Smoking Consequences Questionnaire-Adult= Negative Affect Reduction.

The overall model for the hierarchical linear regression to determine whether HR expectancies moderated the influence of smoking rate on SI was significant [F(8, 606) = 30.77, p <.001, R Square = .292, Adjusted R Square = .282]. Race, MDD and BSCQ-A HR expectancies were identified as significant predictors. The BSCQ-A HR * daily smoking rate interaction term was not significant. See Table 4 for complete results.

Table 4.

Hierarchical Regression to test Health-related Expectancies as a Moderator of Daily Smoking Rate on Suicidal Ideation

	Step 1	p	Step 2	p	Step 3	p	Step 4	P
Age	−0.038	.347	−0.004	.900	−0.001	.978	−0.001	.987
Race	0.098	.016	0.090	.010	0.081	.020	0.082	.020
MDD			0.491	<0.001	0.499	<0.001	0.499	<0.001
AUD			0.003	.932	−0.004	.923	−0.003	.937
DUD			0.075	.051	0.059	.124	0.060	.118
Daily smoking rate					0.097	.006	0.044	.806
BSCQ-A HR scale					−0.087	.015	−0.088	.014
BSCQ-A HR *							0.054
Daily smoking rate	3.313(2, 606)	.037	46.01(5, 606)	<0.001	35.21(6, 606)	<0.001	30.77(8, 606)	<0.001
F (df)	0.011		0.277		0.291		0.292
R 2	0.008		0.271		0.283		0.282
Adjusted
R2

Note. Values in each step are β’s for each variable included within that model. MDD = Major Depressive Disorder; AUD = Alcohol Use Disorder; DUD = Drug Use Disorder; BSCQ-A = Brief Smoking Consequences Questionnaire-Adult; HR = Health Risks.

Discussion

In summary, the present findings provide further evidence that current and former smoking is associated with a variety of comorbid psychiatric symptoms, including SI, past suicide attempts, MDD, AUD, and SUD. Consistent with existing literature, smoking status was also associated with smoking outcome expectancies for NAR and HR. Other findings include that smoking rate predicted SI above and beyond the variance explained by MDD, AUD, and SUD, and a statistical trend for NAR expectancies to moderate the influence of smoking on SI whereby they interacted with lower and higher smoking rates differently.

Our first goal in the present study was to establish a link between SI and smoking among college students. We specifically predicted that SI and smoking would be correlated. This hypothesis was supported in that smoking status groups significantly differed in SI, such that daily smokers reported the highest level of SI, followed by infrequent smokers, followed by ex- and nonsmokers. Also, smoking status groups showed significant difference on past suicide attempts, with exsmokers being the most likely to report a previous attempt. Finally, in regression analyses, daily smoking rate was a significant predictor of SI, a relationship which has previously been found in other populations of smokers.^49–52 Therefore, the present study illustrates an important relationship between SI and smoking among college students, findings which can now be added to the many studies of various smokers shown to be at higher risk for SI and suicide.

Our second goal in the present study was to determine whether mental health issues (e.g., depression/substance use disorders) account for the association between SI and smoking, and we specifically predicted that the association between SI and smoking would remain significant while statistically controlling for diagnoses of major depressive disorder (MDD), alcohol use disorder (AUD), and drug use disorder (DUD). Consistent with our prediction, we found that the association between SI and smoking remained significant while statistically controlling for MDD and SUDs. Thus, this initial evidence among college student smokers indicates that risk of SI may not be confined to only those individuals meeting diagnostic criteria for MDD and SUDs. Previous research has established that one of the underlying mechanisms between smoking and SI is that there is a strong relationship between smoking and depression, as well as smoking and substance use.^20–23 Furthermore, Hughes¹⁶ hypothesized that smoking decreases serotonin and monoamine oxidase (MAO) levels, which have been related to impulsivity, a trait linked to both suicidality and SUDs.³⁴ However, in the present sample of college students, while MDD explained variance in SI, it did not account for the link between smoking and SI. These results illustrate that there are potentially unique mechanisms accounting for the association between SI and smoking among college student smokers that differ from other adult smoking populations. Interestingly, previous research has demonstrated that SI among college students is not limited to individuals with severe depression.⁵³ Given this, additional research is needed to elucidate and understand other predictors outside of psychiatric symptomatology that may contribute to SI risk among college smokers.

These findings demonstrate the need to incorporate SI assessment/screening into smoking cessation interventions among college student smokers. Typically, SI assessment can be triggered by elevation on self-report screenings for MDD among college students⁵⁴; however, these results demonstrate that regardless of the presence of symptomatology related to MDD or SUD, college student smokers may endorse SI. Given the observed differences between the relationship between smoking and SI among college students versus other adults, future research should investigate unique, underlying mechanisms that account for such differences. Additional research would identify significant risk factors that are observed among this younger population and thus could be incorporated into college smoking cessation programs to screen for elevation of SI in this population.

Our third goal was to determine whether smoking outcome expectancies act as moderators that influence the link between smoking and SI. We specifically predicted a moderating role for smoking outcome expectancies for negative affect reduction and health risks among college students. Trend-level results indicated that at lower smoking rates, elevated NAR expectancies were associated with lower SI, while elevated NAR expectancies in combination with higher smoking rates were associated with greater SI. It should be noted that since these results were found only at the trend level, they should be interpreted with caution. College student smokers have been shown to have lower levels of nicotine dependence, as a majority report nondaily smoking⁵⁵ and thus may smoke more for the positive effects of nicotine (e.g. enhancement of mood and mental and physical functioning) as opposed to nicotine-dependent adults, who may be more likely to smoke for negative affect reduction or pharmacological symptoms (e.g., withdrawal).⁵⁶ Accordingly, college students who are less nicotine dependent and smoking at lower rates, may be less likely to smoke for NAR. For those participants smoking at higher daily rates in combination with high NAR expectancies, however, risk for SI was greater. In this relationship, there is evidence suggesting that smokers may be self-medicating (albeit ineffectively) their symptoms of depression or subclinical negative affect. These smokers would also be more nicotine dependent due to their daily smoking rate and therefore more closely resemble nicotine-dependent adult smokers. Additional research is needed to explore such relationships to improve cessation interventions and outcomes among college student smokers.

Results indicated that while there was a main effect for HR expectancies predicting SI whereby lower HR expectancies predicted higher SI, HR expectancies did not moderate the association between smoking and SI in the current sample. College student smokers have relatively limited smoking histories, thus limiting exposure to painful smoking-related conditions (e.g., cancer, COPD).⁵⁷ Furthermore, evidence suggests that college student smokers report that smoking has minimal effect on their health.⁵⁸ Thus, college smokers who report lower HR expectancies may be less concerned about their current/future health and more susceptible to experiencing SI. Future research should explore the relationship between HR expectancies among college smokers, given their limited smoking histories, as well as the potential link between SI and health in this population.

Limitations

While the present study established a unique relationship with smoking and SI among college students that warrants further exploration, it has several limitations that warrant mention. First, data was collected from a large Southeastern university that may not be generalizable to other geographical areas. Second, participant responses were collected via an online survey. Due to the methods of data collection the diagnoses of depression and substance use were survey-based and thus do not include the comprehensive diagnostic rigor of a clinician interview. However, the measures used in the survey have been established as reliable and valid measures of psychopathology and SI and are widely-used self-report measures.^45,46,59 Finally, the current study was cross-sectional in design, and future studies would be improved by implementing a longitudinal design in which variables were assess across time.

Conclusions

The current study identified a unique association between smoking and SI among college students. Evidence suggests that this relationship is independent of MDD, AUD, and DUD, in that an increased daily smoking rate predicted SI, after controlling for MDD, AUD and DUD. Furthermore, a trend was found such that elevated NAR expectancies in combination with higher smoking rates were associated with greater SI. Additionally, while lower HR expectancies were found to predict SI, they did not moderate the relationship between SI and smoking. The present results indicate that among college student smokers, it may be beneficial to screen for SI at the outset and during smoking cessation interventions, even in the absence of depressive symptoms or SUDs. Additional research is warranted to explore the underlying mechanisms of smoking and SI in order to identify important risk factors for SI that may be unique to college student smokers. Such research might examine other types of psychopathology (e.g., personality disorders or anxiety disorders) that were not included in the current study and identify their potential relation with smoking and SI. In addition, future research should investigate whether a similar association exists between e-cigarette use and SI in this population, given the prevalence of e-cigarette use among college students.