Mental Health Treatment Seeking Among Veteran and Civilian Community College Students

Onset of most mental disorders typically occurs before age 24 (1). If left untreated, these disorders impair academic achievement (2). About a third of college students have a mental disorder, yet only a third of those with a disorder receive counseling or medications (3). Nearly half of all college students are enrolled in two-year community colleges (4). Community college students experience greater financial stress and have substantially lower annual household incomes compared with students of four-year colleges (5). They also enter college less academically prepared, with lower high school grade point averages and college admission test scores (5). Only 16% of two-year community college students receive a degree within three years of enrollment (5). Although college campuses represent an opportunity to detect and treat mental disorders, few community colleges offer on-site clinical services.

Recently returning veterans are a particularly vulnerable group of community college students. Over one-third of veterans using Post-9/11 GI Bill benefits are enrolled in community colleges (6). Veterans face challenges transitioning from highly structured military life to relatively unstructured, self-directed campus life (7). When combined with other stressors faced by veterans reintegrating to civilian life, transitioning to campus life can aggravate symptoms of mental illness. Veteran community college students are significantly more likely than their civilian counterparts to suffer from depression, posttraumatic stress disorder (PTSD), and suicidal ideation and are significantly more likely to perceive a need for treatment (8).

The health beliefs model posits that treatment seeking depends on perceptions about need for, barriers to, and effectiveness of treatment (9). Among students at traditional four-year colleges, greater perceived need for, barriers to, and effectiveness of mental health treatment are important predictors of service use (10,11). Among active-duty military personnel, personal stigma and unfavorable beliefs about treatment effectiveness have been reported as barriers to care (12). Students from racial-ethnic minority groups and those with lower socioeconomic status at four-year colleges are less likely than white students and those with higher socioeconomic status to use mental health services (10).

Because there is virtually no information about treatment seeking among community college students, we examined correlates of mental health service use in this population. Identifying significant correlates of treatment seeking provides critical information needed to develop engagement interventions targeting this highly vulnerable population. We hypothesized that students with greater perceived need, greater perceptions of treatment as being effective, and lower personal stigma would be more likely to seek treatment. We also expected that veterans and students from racial-ethnic minority groups would be less likely to use services when the analysis controlled for other sociodemographic and clinical characteristics.

Methods

Our sampling, recruitment, and weighting methods have been described in a previous publication (8). The study was approved by the University of Arkansas for Medical Sciences Institutional Review Board. Our sampling frame included students enrolled for the 2012 spring semester in 11 community colleges in Arkansas. Veteran status (as reflected by receipt of Post-9/11 GI Bill benefits) was initially determined by the registrar’s office. We contacted all veterans and a random sample of civilians. Stratification weights were specified as the inverse probability of being sampled.

Students were contacted via an opt-in letter (in postal mail) and e-mail reminders. Written informed consent was obtained online. The response rate to this Web-based survey was 31% for veterans and 32% for civilians. Poststratification weights were calculated to account for survey nonresponse by using demographic data available from the registrar’s office. The stratification weight was multiplied by the poststratification weight to generate an overall weight. All numbers and percentages are weighted. Survey respondents self-reported whether they had served in the military, and 74 students who were receiving Post-9/11 GI Bill benefits reported not serving in the military (that is, they were spouses of veterans) and were reclassified as civilians. A total of 765 students (211 veterans and 554 civilians) responded to the survey.

We measured sociodemographic and clinical characteristics with items from the Healthy Minds Study (3). Current mental disorders were assessed by using validated screening instruments for depression (nine-item Patient Health Questionnaire [PHQ-9]), generalized anxiety disorder (seven-item Generalized Anxiety Disorder screener), and posttraumatic stress disorder (Primary Care PTSD screen). Binge drinking, illegal drug use, and nonlethal self-injury were assessed with single items. Suicidal ideation in the past two weeks was assessed with the ninth item of the PHQ-9, and suicidal intent in the past year was assessed with a single item. Students with a positive screen were considered to have a probable mental disorder. Perceived need for and effectiveness of treatment were measured with single items. Personal stigma (how one views others with a mental disorder) and public stigma (how others are thought to view people with a mental disorder) were measured with a scale adapted from Link and colleagues (13). Participants were considered to have used mental health services if they reported receiving counseling from a mental health professional (psychiatrist, psychologist, or social worker) or taking any psychotropic medications in the past year. Separate survey items asked about each class of psychotropic medication.

The analytic sample (N=511) was defined as the subset of students who screened positive for a current mental disorder or reported a perceived need for treatment (149 veterans and 362 civilians). Those who did not report a need for treatment were excluded because we did not want significant correlates of prevalence to confound identification of correlates of treatment seeking. In addition, because our long-term goal is to design engagement interventions, community college students without a need for treatment would not have contributed relevant information. The logistic regression analyses specified use of psychotropic medications and receipt of psychotherapy as the dependent variables. Explanatory variables included sociodemographic variables, clinical characteristics, and perceptions about need, treatment effectiveness, and stigma. Because so few students used psychotherapy, we limited the number of explanatory variables to the number of observed events divided by five (14). Standard errors were adjusted for stratified sampling by college. Weights were used in the statistical analysis to ensure results were generalizable to the student population in the 11 community colleges participating in the study.

Results

Almost half (N=240, 47%) of the 511 participants were under age 23, 371 (73%) were female, 146 (29%) were from racial-ethnic minority groups (black, N=82, 16%; other, N=64, 13%), 147 (29%) were married, 320 (63%) were employed full- or part-time, 278 (54%) were insured, and 241 (47%) reported that they struggled financially and another 223 (44%) reported that finances were tight.

The unweighted percentage of veterans in the sample was 29% (N=149), and the weighted percentage (reflecting the proportion in the student population of the 11 community colleges participating in the study) was 3.7%. Three-quarters of veterans (N=114, 77%) reported having been deployed. Compared with civilians, veterans were more likely to be age 23 or older (N=187, 52%, versus N=137, 92%, p<.001), less likely to be female (N=270, 75%, versus N=33, 22%, p<.001), and more likely to be married (N=100, 28%, versus N=84, 56%, p<.001).

A large proportion of the 511 participants screened positive for depression (N=165, 32%), generalized anxiety disorder (N=152, 30%), PTSD (N=112, 22%), binge drinking (N=182, 36%), illegal drug use (N=84, 16%), suicidal ideation in the past two weeks (N=94, 18%), acute suicidal ideation in the past year (N=70, 14%), and nonlethal self-injury (N=75, 15%). Compared with civilians, veterans were more likely to screen positive for depression (N=115, 32%, versus N=70, 47%, p<.01), PTSD (N=77, 21%, versus N=55, 37%, p<.01), and binge drinking (N=127, 35%, versus N=76, 51%, p<.01).

Over half of the 511 participants (N=290, 57%) reported a perceived need for mental health treatment in the past year. Two-thirds (N=338, 66%) thought that psychotherapy would be helpful, and 301 (59%) thought that psychotropic medications would be helpful. Veterans were less likely than civilians to perceive that medications were helpful (N=70, 47%, versus N=216, 60%, p=.03) and more likely to perceive public stigma (p<.01).

Thirty percent (N=151) reported taking psychotropic medications in the past year. Among sociodemographic variables, only age was a significant predictor of medication use (Table 1). Compared with students ages 18–22, those ages 23–30 (odds ratio [OR=4.51], p=.029), 31–40 (OR=4.85, p=.035) and 41 and older (OR=8.99, p<.004) were significantly more likely to report taking psychotropic medications. None of the clinical screeners were significant predictors of medication use. However, perceived need was a significant and substantial predictor of medication use (OR=7.81, p<.001). In addition, the perceived effectiveness of psychotropic medications was a significant predictor (OR=3.38, p=.012). Neither personal-stigma nor public-stigma were significant predictors of medication use.

A sensitivity analysis added psychotherapy as an independent variable to the regression equation predicting use of psychotropic medications, and its effect was significant and substantial (OR=4.05, p=.03), suggesting that the two treatments (psychotherapy and medications) are complementary. In a second sensitivity analysis that did not use weights, insurance status and screening positive for generalized anxiety disorder and PTSD were positive and significant (p<.05) predictors of medication use.

Fifty-six (11%) of the participants reported receiving psychotherapy in the past year. Of the sociodemographic variables, only poorer financial status was a significant predictor. Among the clinical screeners, only having a positive PTSD screen was a significant predictor of psychotherapy use (OR=2.78, p=.037). Perceived need for treatment, perceived effectiveness of psychotherapy, and personal or public stigma were not significant predictors. A sensitivity analysis added use of psychotropic medications as an independent variable to the regression equation predicting psychotherapy use, but it was not a significant predictor. In a second sensitivity analysis that did not use weights, veteran status, male gender, perceived need, and screening positive for generalized anxiety disorder were positive and significant (p<.05) predictors.

Discussion

To the best of our knowledge, this study is the first to identify predictors of mental health treatment seeking among community college students and to compare treatment seeking between veteran and civilian community college students. We focused on students with a probable mental disorder or a perceived need for treatment in order not to confound correlates of prevalence with correlates of treatment seeking and because our goal was to inform the design of engagement interventions for students in need of services. Results indicated that community college students with a probable mental disorder or perceived need for treatment had low rates of mental health service use. A few modifiable risk factors were found to correlate with treatment seeking and could be targeted by engagement interventions.

The Healthy Minds Study, which surveyed students of four-year colleges and universities, found that race-ethnicity was an important predictor of psychotropic medication use (3). However, this variable was not a significant predictor in our sample of community college students of lower socioeconomic status (464 [91%] reported that their current financial situation was tight or a struggle). Because there is less variation in socioeconomic status among community college students than among students of four-year colleges, racial-ethnic differences may have been less important in our sample. When the analysis controlled for sociodemographic and clinical characteristics, veteran students did not differentially engage in pharmacotherapy or psychotherapy compared with civilian students. We previously reported that in a sample of community college students with and without mental disorders, veterans were more likely than civilians to receive psychotherapy when the analysis controlled for age, gender, and race-ethnicity (8). However, when the analysis in the study reported here controlled for clinical characteristics, veterans were not more likely than civilians to use psychotherapy. Our more recent finding suggests that the previous findings were driven by higher prevalence rates or severity of mental disorders among veterans (8).

In contrast to the Healthy Minds Study (3), our study did not find that the screening instruments predicted medication use. These contrasting findings may be due to the fact that the Healthy Minds Study included students who screened negative for mental disorders, whereas our study excluded students without a probable mental disorder or perceived need for treatment. Similar to the Healthy Minds Study, we found that students with greater perceived need for treatment and treatment effectiveness were more likely to use psychotropic medications (10). Stigma was not a significant barrier to either psychotropic medication use or receipt of psychotherapy.

These findings can inform the design of engagement interventions for community college campuses that promote treatment seeking for mental and substance use disorders. Fortunately, most nonmodifiable sociodemographic factors, such as race-ethnicity and veteran status, were not found to be strong predictors of treatment seeking. Modifiable characteristics, such as perceived need for treatment and effectiveness of psychotropic medications, were strong predictors of pharmacotherapy use and should be targeted by engagement interventions. In particular, these beliefs should be targeted among younger students, who were less likely to use pharmacotherapy.

Results should be interpreted with the following limitations in mind. First, findings may not generalize to other geographic regions. Second, as in many Web-based surveys, our response rate was low, which increased the risk of nonresponse bias. We attempted to mitigate this risk by using nonresponse weights. Third, the low rate of psychotherapy use reduced the statistical power to detect clinically meaningful predictors.

Conclusions

Both civilian and veteran community college students had high rates of unmet need. Engagement interventions that focus on increasing the perceived need for and effectiveness of treatment may be helpful in encouraging younger community college students to seek treatment. Because most community colleges lack any on-site mental health services (15), engagement interventions should focus on linking students with off-site services.