Anxiety and suicidality in the college student population

Abstract

Objective

This study examined symptoms of anxiety, depression and suicidality in a national sample of college students.

Participants

Using national survey data from the Healthy Minds Study (HMS), a random sample from 184 U.S. campuses from fall 2016 to spring 2019 was analyzed (N=119,875).

Methods

Prevalence rates were examined with the Generalized Anxiety Disorder 7-Item scale, Patient Health Questionnaire-9 and suicidality questions. The relationship between anxiety, depression and suicidality was assessed through Spearman’s correlations, Kruskal-Wallis H test and logistic regressions.

Results

Screening only for depression would pick up 23% of suicidal ideation, increasing to 35% when also screening for anxiety. Those with anxiety and no to minimal depression had the second highest likelihood of suicide attempt, following those with anxiety and depression. The symptom “feeling afraid something awful might happen” doubled the odds of suicidal ideation.

Conclusions

College campuses may benefit from assessing particular anxiety symptoms in relation to suicide.

According to the World Health Organization¹, anxiety disorders are the most common mental health problem in the United States, with rates continuing to climb. Contributing to this rise are factors such as an increased tendency to report, reduced stigma, increased recognition of symptoms, and greater comfort with utilizing services². Other factors associated with increased anxiety rates include psychosocial vulnerabilities and increased use of technology and social media³. Additionally, consumer culture has shifted towards individualism, with society emphasizing the importance of pursuing extrinsic goals, leading to individuals setting higher expectations that are often less attainable⁴. As the price of education continues to rise, the job market is becoming increasingly competitive and school acceptance rates are declining⁵. Many of these challenges are hitting young adults hard, potentially contributing to the increased psychological distress in this population.

Epidemiological studies indicate that 75% of the onset of lifetime mental illness occurs by age 24⁶, with the average age of anxiety onset from early adolescence to young adulthood. College student populations represent a particularly important group to study. Over half of each birth cohort enrolls in college each year, with over 20 million in U.S. postsecondary education⁷. In addition to the enhanced risk associated with young adulthood, college students are exposed to unique stressors that place them at risk for mental illness.⁸ Some unique factors that shape mental health risk for college students include student debt⁹; uncertainties about job prospects,¹⁰ academic pressures,¹¹ “culture shock”, especially for first-generation students¹² and separation from the family unit, which can be especially challenging for students from particular cultures.¹³ Thus, college students represent a crucial population for screening, prevention, early intervention, and implementation of effective treatments.

Anxiety and Depression in College Populations

As of 2019, an annual report by the American College Health Association–National College Health Assessment (ACHA-NCHA) revealed that a large portion of students were experiencing distressing symptoms¹⁴. Students reported feeling: overwhelmed by all they had to do (87%), very sad (71%), and overwhelming anxiety (66%). Furthermore, 13% reported seriously considering suicide. Lipson and colleagues¹⁵ analyzed data from the 2012–2015 Healthy Minds Study (HMS) samples – a public health initiative collecting data on college mental health. The results from 43,375 students revealed that 18% were experiencing anxiety, measured as a score of ≥ 10 on the Generalized Anxiety Disorder 7-Item Scale (GAD-7). Eisenberg and colleagues² reported on data from the 2016–17 HMS sample and found that out of 1,347 students, 39% were struggling with symptoms related to their mental health. The authors reported that 4% of undergraduates and 4% of graduate students endorsed symptoms of anxiety, while 14% of undergraduates and 11% of graduate students endorsed symptoms of depression, measured by the Patient Health Questionnaire-9 (PHQ-9). Not only are these rates alarming, but rates have been increasing over time. Duffy and colleagues examined trends in mental health outcomes among U.S. college students from 2007 to 2018 across the ACHA-NCHA and HMS and found a worsening of anxiety and depression¹⁶. In fact, rates of moderate to severe anxiety rose from 17.9% in 2013 to 34.4% in 2018 and rates of moderate to severe depression rose from 23.2% in 2007 to 41.1% in 2018.¹⁶ These findings underscore the importance of focusing on the college student population to be able to better prioritize and focus intervention and treatment.

Suicidality and Anxiety in College Students

Just as anxiety is increasing in college populations, rates of suicide are also on the rise¹⁷. Suicide is the second leading cause of death in the college student population¹⁷. Whereas prevalence rates and associations between depression and suicidality have been a focus of prior research¹⁸, less is known about the correlates, prevalence or treatment of suicidal behavior among individuals with anxiety¹⁹. Auerbach and colleagues²⁰ reported on data from 14,348 first year college students across 19 institutions and found that students with high diagnostic comorbidity, particularly those with GAD, may be at elevated risk for attempting suicide. Studies have found relationships between panic disorder (PD), GAD and suicide risk^21,22. Khan and colleagues²³ studied 20,076 anxious patients that were participating in clinical trials assessing new anti-anxiety medications. The researchers found that individuals with anxiety disorders were at 10 times greater risk of suicide attempt than individuals in the general population. Studies have also found associations between anxiety and suicidality in non-clinical samples. A study by Abdel-Khalek and Lester²⁴ found that suicidal ideation was positively and significantly associated with increased anxiety in undergraduate students. Several studies have also found symptoms of anxiety correlated with suicidal thinking and suicide rumination^{25, 26, 27}.

Although these findings are intriguing, researchers have speculated that co-occurring mood disorders may be accounting for the increase in suicidality seen across anxious samples,^{21, 22} as anxiety disorders and major depression are highly comorbid, with rates of co-occurrence as high as 60%.²⁸ Depression appears to play a causal role in suicidal behavior, with evidence from a wide range of studies demonstrating strength of association and controlling for confounding factors^{29, 30}. For instance, Lamis and Jan³¹ reported a strong relationship between hopelessness, negative affect and suicidality. From a developmental perspective, transitional periods put individuals at a higher risk, and many who are in college and have fewer social supports, psychological resources, and higher negative affect are at a higher risk for suicidal thoughts and behaviors^{31, 32}.

Referencing findings that the relationship between anxiety and suicide risk may instead be due to comorbid depression, Sareen and colleagues³³ studied suicidality in individuals without a history of mood disorders. Data collected over three years from 7,076 individuals participating in a prospective population-based survey revealed that anxiety disorder diagnosis, without history of a mood disorder, was in fact significantly associated with increased risk of suicidal ideation and suicide attempt. Norton, Temple and Pettit³⁴ replicated and extended these findings in a sample of 166 college students and found a significant association between anxiety and suicide risk after controlling for depression symptoms.

Given the significant overlap between anxiety and depression symptoms, it is difficult to ascertain if the relationship between depression and suicidality is purely accounted for by depression or if also anxiety plays a role. It is possible that the association between anxiety and suicidality could be weak once parsing out the covariance of anxiety and depression^{35, 36} Although there is a clear association between depression and suicidality, not every depressed person is suicidal. Likewise, suicide attempts do not always happen when a person is in a depressive episode³⁷. Much of the work focused on understanding the interplay of anxiety, depression and suicidality has been conducted in clinical populations; focusing on a general college population may aid in understanding this interplay in an at-risk group.

Research Aims

This study had several aims related to better understanding the relationship between anxiety and suicidality. First, we sought to investigate the relationship between suicidality and anxiety, both with and without depression, in a large college sample. Second, we sought to examine which specific symptoms of anxiety are most predictive of suicide risk. Our aim with studying anxiety at the symptom level was to provide further insight into more nuanced ways of understanding the relationship between anxiety and suicide, which could ultimately lead to better, more targeted identification and treatment.

Materials and Methods

Study Design

This study utilized national survey data obtained from HMS. The present study is an analysis of HMS data, which are made publicly available for research purposes. Colleges elect to participate in the Healthy Minds Study and there are no exclusionary criteria for institutional enrollment. Recruitment was open to all types of institutions. In the present study, we analyze data from three waves (2016–2019). There were 54 participating institutions in 2016–17, 60 in 2017–18 and 79 in 2018–19, for a total of 184 unique institutions included in this analysis. Recruitment was open to all types of institutions. The sample of colleges and universities includes a diversity of institutions, broadly representing all types of higher education across the United States. The HMS has been completed by both urban and rural institutions, public and private, of all varying sizes. The HMS annual web-based survey assesses factors such as student demographic information, mental health status, service utilization and barriers to treatment among undergraduate and graduate students. Data were collected through Qualtrics, and the study was approved by the Institutional Review Board at all campus sites. Additionally, the study was covered by a National Institutes of Health Certificate of Confidentiality. Further details about the study design of HMS have been reported on widely and additional information is available on the HMS website: https://healthymindsnetwork.org/hms/.

Student Recruitment

At each institution with ≥ 4,000 students, a random sample of 4,000 students ages 18 or older was invited to participate in the online survey; on smaller campuses, all students were invited to participate. The only inclusion criteria was that students had to be at least 18 years of age. Recruitment was conducted via email across all sites. A total of $2000 worth of incentives were distributed each year. Students were informed that they would be entered into a drawing to win one of several prizes (ten $100 and two $500 gift cards per year). All randomly sampled students on participating campuses were eligible for the incentives each year. Participants provided informed consent. Survey response rates were 24% in 2016–17, 24% in 2017–18, and 16% in 2018–19. Sample probability weights were constructed to adjust for potential differences between responders and nonresponders. Administrative data on students’ sex, race/ethnicity, degree level, and grade point average were obtained from participating schools. These data were used to construct response weights, equal to one divided by the estimated probability of response, using a logistic regression to predict the likelihood of response associated with each variable.

Measures

Generalized Anxiety Disorder 7-Item Scale (GAD-7)

The GAD-7³⁸ is a 7-item screening tool that assesses symptoms of anxiety over the past two weeks. In completing this valid and reliable measure,³⁹ participants rate each question on a Likert scale from 0 (not at all) to 3 (nearly every day). Scores range from 0–21. Following standard scoring protocol, participants were placed into four groups based on their GAD-7 scores: none to minimal (score <5), mild (5–9), moderate (10–14) and severe anxiety (>14). Further, using a standard cut-off of ≥10, participants were placed into anxiety or no to minimal anxiety groups. Finally, the relationship between individual GAD-7 items and suicidality was examined. For this analysis, the items were transformed into binary variables, whereby 0 was coded as no and any score >0 was coded as yes for endorsing that symptom.

Patient Health Questionnaire-9 (PHQ-9)

The PHQ-9⁴⁰ was used to measure depression over the past two weeks. PHQ-9 scores demonstrate high internal consistency and are highly correlated with clinical diagnosis^{41, 42}. Subjects rate each question on a Likert scale from 0 (not at all) to 3 (nearly every day). Following standard scoring protocol, PHQ-9 scores were divided categorically into five groups, consisting of: none to minimal (score <5), mild (5–9), moderate (10–14), moderately severe (15–19) and severe (>19). Participants were also grouped dichotomously into depression or no to minimal depression groups based on a standard cut-off of ≥ 10.

Suicidality Questions from the National Comorbidity Survey (NCS-R)

Past-year suicidal ideation and suicide attempt were assessed using yes/no questions originally developed for the NCS-R⁶. The first question was “In the past year, did you ever seriously think about attempting suicide?” If the response was “yes,” then participants were asked: “In the past year, did you attempt suicide?” These questions have been used to assess for suicidality in many other HMS papers^{15, 43}.

Demographic Characteristics

Demographics were assessed through self-report measures (See Table 1). Subjects reported their current age, race/ethnicity, nationality, gender identity and undergraduate or graduate status.

Table 1.

Sample Demographics

	N (total=127,555)	Weighted %
Age
18–19	39,229	30.8
20–21	44,408	34.8
22–25	30,556	24.0
26–30	13,364	10.5
Race/Ethnicity*
African American/Black	10,958	8.6
American Indian or Alaskan Native	2,345	1.8
Asian American/Asian	16,825	13.2
Hispanic/Latino	13,654	10.7
Middle Eastern, Arab or Arab American	2,667	2.1
Native Hawaiian or Pacific Islander	980	0.8
White	90,980	71.3
Self-Identify	2,621	2.1
Nationality
International	9,437	7.4
Gender Identity
Male	51,451	40.3
Female	72,034	56.5
Trans male	559	0.4
Trans female	188	0.1
Genderqueer/Nonconforming	2,192	1.7
Self-Identity	993	0.8
Degree Status*
Graduate	30,105	23.6
Undergraduate	94,728	74.3
Other	2,816	2.2

^*Race/ethnicity and Degree Status select all that apply so total over 100%

Data Analysis

For each of the demographic characteristics, we calculated and reported on prevalence rates and frequencies, which were weighted using the non-response weights described above (See Table 1). For each of the four levels of anxiety and five levels of depression, we reported on the proportions of suicidal ideation and attempt. We ran Spearman’s Rho correlations on anxiety and depression overall scores to understand the correlation between the diagnoses. In conducting group comparisons, we used the Kruskal-Wallis H test. Finally, we conducted two logistic regressions to understand which items of the GAD-7 are most predictive of suicidal ideation and suicide attempt. We reported odds ratios (ORs), 95% confidence intervals (CIs), standard errors, standardized and unstandardized betas and p-values. All analyses were conducted using SPSS Version 24.

Results

Sample Characteristics

Our study analyzed data from 119,875 students on 184 U.S. campuses who completed the X survey from fall of 2016 to spring of 2019. Given the young adult age of onset for anxiety, the analytic sample was restricted to survey respondents ages 18–30 years (M=21.26 SD=2.99). The sample was 56.6% female, 71.3% White, 92.6% U.S. citizen. See Table 1 for additional demographic information.

Clinical Characteristics

Descriptive statistics were run to understand the prevalence of anxiety, depression, suicidal ideation and suicide attempt in the population. The mean anxiety level was mild (GAD-7=7.2, SD=5.7). Results revealed that 40.1% of students had no to minimal anxiety (GAD-7<5), 28.9% had mild anxiety (GAD-7=5–9), 17.6% had moderate anxiety (GAD-7=10–14) and 13.4% had severe anxiety (GAD-7>14). The mean depression level of the population was mild (PHQ-9=8.5, SD=6.3). Results also revealed that 32.0% of students had no to minimal depression (PHQ-9<5), 31.5% had mild depression (PHQ-9=5–9), 18.8% had moderate depression (PHQ-9=10–14), 10.7% had moderately severe depression (PHQ-9=15–19) and 7.0% had severe depression (PHQ-9>19). In terms of suicide risk over the past year, 14.3% of students reported suicidal ideation and 1.4% attempted suicide.

Relationships between Depression, Anxiety and Suicidality

A Spearman’s Rho correlation was run to assess the relationship between anxiety and depression. Results showed that depression and anxiety total scores were significantly correlated at r_s=0.76, p<0.001. Figure 1 presents the percentage of individuals with varying levels of anxiety and depression who also reported suicidal ideation and attempt. Of those with no to minimal anxiety and no to minimal depression (N=70,2777), 4.6% reported ideation and 0.3% reported an attempt. Of those with anxiety and no to minimal depression (N=8,795), 9.5% reported ideation and 0.8% reported an attempt. Of those with depression and no to minimal anxiety (N=14,016), 22.5% reported ideation and 2.0% reported an attempt. Finally, of those with anxiety and depression (N=26,787), 34.5% reported ideation and 4.0% reported an attempt.

Figure 1.

Anxiety, Depression and Suicidality

The Kruskal-Wallis H test showed a statistically significant difference in suicidal ideation between the different groups, χ²(3) = 16924.36, p<0.01. Specifically, those with depression and anxiety had significantly higher ideation than all other groups (Mean rank = 76646.71); those with depression and no to minimal anxiety (Mean rank = 68863.49) were higher than those with anxiety and no to minimal depression (Mean rank = 60534.79); those with both no to minimal anxiety and depression were lower than all other groups on ideation (Mean rank = 57439.73)

Similarly, Kruskal-Wallis H test showed a statistically significant difference in suicide attempt between groups, χ²(3) = 70.39, p<0.01, but with a varying pattern of differences. Specifically, those with both anxiety and depression (Mean rank = 9131.66) differed from all other groups as they were the most likely to report a suicide attempt. Those with depression and no to minimal anxiety (Mean rank = 8851.83), those with those with anxiety and no to minimal depression (Mean rank = 8854.74) and those with both no to minimal anxiety and depression (Mean rank = 8712.7) did not significantly differ from one another in risk for suicide attempt.

Anxiety Subscale Items

To investigate the relationship between individual items of the GAD-7 and suicide risk, we ran logistic regressions related to suicidal ideation and attempt (See Table 2). When examining individual symptoms of anxiety, as measured by the GAD-7, six of seven items were significant predictors of suicidal ideation. Item 3 (“worrying too much about different things”) was not a significant predictor. Endorsing item 7 (“feeling afraid as if something awful might happen”) was associated with two times higher odds of suicidal ideation (OR=2.22, 95% CI=2.13–2.31, B=0.80, p<0.001). For suicide attempt, items 4 (“trouble relaxing;” OR=1.45, 95% CI=1.19–1.78, B=0.37, p<0.001) and 5 (“being so restless that it is hard to sit still;” OR=1.35, 95% CI=1.19–1.52, B=0.30, p<0.001) were significant. Item 1 “Feeling nervous, anxious, or on edge” was also significant, however this was in the negative direction (OR=0.76, 95% CI=0.62–0.94, B= −0.28, p=0.01).

Table 2.

Anxiety Symptoms and Suicidality

		B	S.E.	Sig.	Exp(B)	Lower CI	Upper CI
Anxiety Items and Suicidal Ideation	Item 1	0.24	0.03	0.00	1.28	1.20	1.36
	Item 2	0.46	0.03	0.00	1.59	1.50	1.69
	Item 3	0.03	0.04	0.34	1.03	0.97	1.11
	Item 4	0.45	0.03	0.00	1.57	1.48	1.65
	Item 5	0.42	0.02	0.00	1.52	1.46	1.57
	Item 6	0.37	0.03	0.00	1.45	1.38	1.52
	Item 7	0.80	0.02	0.00	2.22	2.13	2.31
Anxiety Items and Suicide Attempt
	Item 1	−0.28	0.11	0.01	0.76	0.62	0.94
	Item 2	0.20	0.11	0.06	1.22	0.99	1.51
	Item 3	0.03	0.12	0.78	1.04	0.82	1.31
	Item 4	0.37	0.10	0.00	1.45	1.19	1.78
	Item 5	0.30	0.06	0.00	1.35	1.19	1.52
	Item 6	−0.09	0.08	0.23	0.91	0.78	1.06
	Item 7	0.06	0.07	0.40	1.06	0.93	1.20

^aAny endorsement of each item was considered a “yes” for each anxiety symptom.

^bGAD-7 Items Endorsed Over Past 2 Weeks:

Q1: Feeling nervous, anxious, or on edge

Q2: Not being able to stop or control worrying

Q3: Worrying too much about different things

Q4: Trouble relaxing

Q5: Being so restless that it’s hard to sit still

Q6: Becoming easily annoyed or irritable

Q7: Feeling afraid, as if something awful might happen

Discussion

This is one of the largest known studies to examine the relationship between anxiety and suicidality in college students – a highly vulnerable population with rising rates of mental health concerns^{2, 5}. Several important findings emerged and include: 1) high rates of anxiety and depression across the college population; 2) anxiety as an added risk factor for suicide above and beyond depression alone; and 3) assessing for specific symptoms of anxiety may increase accuracy in identifying suicide risk.

Our findings reflect alarming rates of anxiety, depression and suicidality in the college population, which is consistent with prior literature^14,15 Indeed, although the mean anxiety and depression levels were reported to be mild, over 17% of students reported moderate anxiety and over 13% reported severe anxiety. In addition, over 18% of students reported moderate depression, over 10% reported moderately severe depression and 7% reported severe depression. Over that prior year, more than 14% of students reported suicidal ideation and over 1% reported an attempted suicide.

An important implication of these findings is that if we were to only screen for depression, we may identify approximately 23% of suicidal ideation in students, whereas if we also screen for anxiety, we would identify approximately 35% of students with suicidal ideation. Not surprisingly, depression remained a significant risk factor for suicidal ideation in this sample. However, students with co-occurring positive screens for depression and anxiety had approximately double the rates of suicide attempts than those with depression alone.

When suicidal ideation and attempt were compared across groups, those with anxiety and depression had the highest risk of suicidal ideation and attempt. Those with depression and no to minimal anxiety had the second highest likelihood of suicidal ideation. However, those with anxiety and no to minimal depression had the second highest likelihood of suicide attempt. In addition to the increased risk of anxiety in those who are also depressed, it turns out anxiety may be an important flag for who may attempt suicide even in those with no to minimal depression.

Another important aim of this research was to study anxiety at the symptom level, to improve our understanding of which symptoms may be most indicative of suicidality. Almost every anxiety item was correlated with suicidal ideation. The item “feeling afraid as if something awful might happen” was most predictive of suicidal ideation. This finding is important, as it can be used for prediction purposes when assessing for suicide risk. This item may reflect feelings of pessimism or hopelessness about the future. Psychiatry has long recognized that these negative future oriented thoughts may be more predictive of suicide than depression itself.⁴³ Endorsement of this symptom may trigger further evaluation of pessimism and hopelessness.

“Trouble relaxing” was positively associated with suicide attempt. Prior literature has found that subjective complaints of restlessness⁴⁵ can lead to an energy to act on impulses that had been suppressed⁴⁶. Studies have suggested this may lead to de novo suicidal ideation⁴⁵ and can worsen pre-existing suicidal ideation,⁴⁷ as well as case reports have found close temporal association between restlessness and attempted suicide⁴⁵. At the same time, “Feeling nervous, anxious, or on edge” was significantly and negatively correlated with suicide attempt. One possible explanation could be that anxiety sensitivity and attempting suicide appear to be negatively related.⁴⁸ Anxiety sensitivity is when an individual notices physical sensations more often and is more likely to interpret those symptoms are harmful or dangerous. Future research should carefully examine these two symptoms to better understand the differences between feeling nervous, anxious or on edge and symptoms of restlessness.

Limitations should be noted when interpreting these findings. First, single binary measures for suicidal ideation and suicide attempt do not fully and adequately capture suicidal behavior. Additionally, the GAD-7 only asks about anxiety over the past two weeks and the suicidality questions ask over the past year. Thus, some relationships between suicidality and episodic anxiety may not be captured within this assessment window. Another limitation is that the cross-sectional nature of these data precludes the possibility of clearly inferring the directionality of the relationship between anxiety and suicidality. Although anxiety may increase risk for subsequent suicidality, it is also possible that the experience of suicidality in the past may increase anxiety or that a third variable may increase risk for both outcomes. Future longitudinal research tracking the emergence of suicidal ideation and anxiety over time is needed to address this relationship more clearly. Additionally, the response rate of 16–24% should be noted. Although this is a typical response rate for online surveys⁴⁹, this does raise the potential question of response bias. As described above, we adjusted estimates with nonresponse weights along known characteristics, although there may be differences between responders and nonresponders on unobserved characteristics. It would also be important to replicate these findings in a non-collegiate young adult population. Perhaps this developmental period more broadly lends itself to increased anxiety, depression and suicidality, rather than just the college environment. It would therefore be useful to study these relationships in a more general population. Another limitation of our study was that there were nine institutions that participated two or more times in the years examined. Inclusion of all participating campuses during this time is consistent with prior publications of HMS data. Finally, as we shift towards a more transdiagnostic approach to understanding underlying symptoms, future research should encompass all anxiety symptoms, as the GAD-7 only technically assesses for GAD, rather than all anxiety disorders or associated symptoms. Evidence that particular anxiety symptoms increase risk for suicidality also points to the need to move away from studying diagnoses and towards understanding risk at the symptom level across the general population.

Conclusions

Our findings suggest the importance of considering anxiety symptoms as potential indicators for suicide risk. Although depression is the most prominent predictor of suicidal thoughts and behavior,²² our findings highlight the importance of assessing for anxiety as a compounding factor for suicide risk, and even suggest that anxiety alone could be a risk factor for suicide attempt. From a prevention angle, there are benefits for colleges to screen for both anxiety and depression to reach high-risk students in need of help and to identify those at greatest risk for suicidal thoughts and behaviors. Studying anxiety at the symptom level may provide greater insight as to who is more at risk, leading to better referral options and safety.

Similar to other studies investigating college mental health, the results of this study emphasize the importance of increasing the number of available programs on college campuses that address anxiety and suicide risk (e.g., Active Minds, National Alliance on Mental Illness, the Reflect Organization)⁵⁰. Trainings could also be implemented at the beginning of college to make students aware of signs of anxiety and how to properly seek help. This approach to intervention and treatment would be in accordance with newer research and treatment models emphasizing the need to focus on profiles or symptoms of disorders rather than a specific diagnosis (e.g., Process-Based Therapy; Unified Protocol).^{51, 52} It is important for colleges to continue working towards identifying high-risk students and offering indicated prevention services, early intervention, and effective treatment options.