This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Mental Health, is properly cited. The complete bibliographic information, a link to the original publication on http://mental.jmir.org/, as well as this copyright and license information must be included.
High rates of mental illness, stress, and suicidality among teens constitute a major public health concern in the United States. However, treatment rates remain low, partially because of barriers that could be mitigated with tech-based telemental health (TMH) resources, separate from or in addition to traditional care.
This study aimed to analyze TMH resource usage by high school students to establish current user characteristics and provide a framework for future development.
A total of 2789 students were surveyed regarding demographics, recent anxiety and depression symptoms, suicidality, and stress; people with whom they could openly and honestly discuss stress or problems, and prior TMH use. Logistic regression models and a general linear model were used to test relationships between variables.
Overall, 30.58% (853/2789) and 22.91% (639/2789) of students reported moderate to severe anxiety and depression symptoms, respectively, in the past 2 weeks; 16.24% (414/2550) had seriously considered suicide in the past year, consistent with national averages. Meanwhile, 16.03% (447/2789) of students had previously used at least 1 of 4 types of TMH resources (ie, self-help, anonymous chat, online counselor, or crisis text line). Teens reporting depression symptoms, higher stress, or suicidality were less likely to talk to a parent about stress or problems and more likely to tell no one. Suicidality was related to the use of all 4 types of TMH resources. Depression symptoms were related to the use of anonymous chat and crisis text line, and those with higher stress were more likely to have used an online counselor. Those reporting anxiety symptoms were less likely to have no one to talk to and more likely to have used a self-help resource.
Youth struggling with mental health symptoms, some of whom lack real-life confidants, are using existing TMH support, with resource preferences related to symptoms. Future research should consider these preferences and assist in the creation of specialized, evidence-based TMH resources.
With rates rising over the last decade [
As with all health care, there are barriers to mental health service access. Less than 30% of teens experiencing suicidality and 40% of teens experiencing major depression seek professional treatment [
Youth may instead seek out informal sources of help, such as friends or the internet [
Researchers have suggested that digital health interventions could be particularly successful among younger individuals because of their frequent technology usage [
At present, data regarding effectiveness and perceptions of TMH resources are both mixed and somewhat limited [
Due to the novelty and ever-evolving nature of communication technology, ongoing research of TMH resources is necessary to direct continual development [
Participants were from 4 northwestern Indiana high schools (2 suburban, 2 rural). These schools were representative of the general area of study, which contains 1 midsize city with much suburban sprawl, surrounded by several large rural counties. Surveys were conducted during educational assemblies at each school in February and March 2017. Schools provided detailed study information to parents at least 2 weeks before each event. Parents either passively consented or opted out on their child’s behalf; students also completed an age-appropriate consent or assent process directly before the survey.
Students assembled into their school’s auditorium or gymnasium and connected their tablet, laptop, or mobile phone to the secure local Wi-Fi network provided by the study team. Survey questions were presented via prerecorded video, integrated within an hour-long media-rich educational presentation. In addition to the large screen displayed to the group, questions were shown on students’ devices, on which they responded confidentially. Anonymous aggregate responses were displayed to the group after all answers for each question had been recorded, contributing to a larger message of stigma reduction within the presentation. See
A more detailed description of these events is available in our methodology paper regarding the use of immediate response technologies to gather health data from youth [
The original sample included 3412 high school students. We removed responses from 168 students who only completed the practice questions, 434 students who stopped the survey before the 30th question, and 27 students who responded “prefer not to answer” or did not respond for 80% or more of the questions, leaving a final sample of 2789. To maintain representativeness in the sample, participants who provided partial data were retained where possible (n=1667), and analysis was performed with pairwise deletions, resulting in varying sample sizes across the results.
The 35-question survey began with demographic questions, including age, race, and gender and included the measures below. Due to the sensitive nature of some measures, participants could select “prefer not to answer” or skip any question, with the exception of an initial question regarding age (ie, minor status) to determine consent versus assent.
Example screen images from the live survey events.
Summary statistics were calculated for demographics, suicidality, PHQ-4 anxiety and depression scores, stress, and previous TMH use. To explore bivariate relationships between predictors and outcomes, zero-order correlations were computed (tetrachoric and polychoric for categorical variables, and Pearson coefficients for continuous variables). A total of 3 series of statistical models were tested. First, to understand how covariates were related to mental health outcomes, demographic characteristics (age, gender, race) were entered into separate logistic regression models predicting suicidality, moderate or severe anxiety, and moderate or severe depression; for stress, the same set of covariates were tested using a general linear model. Second, demographics and mental health outcomes were related to individuals with whom students felt they could openly and honestly discuss problems—parents, friends, or no one—using separate logistic regression models. Third, separate ordinal logistic models were used to relate demographics, suicidal ideation, anxiety, depression, and stress to students' previous use of the aforementioned 4 types of TMH resources (3=
As shown in
In preliminary analyses examining zero-order correlations for demographics with mental health outcomes, all correlations were low (
High school sample characteristics.
Characteristic | n (%) | |
13-14 | 235 (8.43) | |
15-16 | 1514 (54.30) | |
17-19 | 1039 (37.27) | |
White | 1746 (62.60) | |
Black | 342 (12.26) | |
Hispanic American or Latino | 211 (7.57) | |
Other | 339 (12.15) | |
Prefer not to answer | 151 (5.41) | |
Male | 1266 (45.39) | |
Female | 1442 (51.70) | |
Other | 81 (2.90) | |
Seriously considered suicide in the last 12 months | 414 (16.24) | |
Moderate or severe anxiety symptoms last 2 weeks | 853 (30.58) | |
Moderate or severe depression symptoms last 2 weeks | 639 (22.91) | |
0-3 | 516 (20.32) | |
4-7 | 1131 (44.55) | |
8-10 | 892 (35.13) |
Histogram of number of responses or scores for suicidality, Patient Health Questionnaire (PHQ) anxiety, PHQ depression, stress, and website, anonymous online chat, online counselor, and crisis text line use for telemental health.
Binary logistic regression analyses for combined demographics predicting suicidality, anxiety, and depression and generalized linear model analysis for combined demographics predicting stress.
Predictor | Seriously considered suicide in the last 12 months (n=2427) | Moderate or severe anxiety in the last 2 weeks (n=2637) | Moderate or severe depression in the last 2 weeks (n=2637) | Average stress during past month (n=2403) | |||||
Ba (SEb) | ORc (95% CId) | B (SE) | OR (95% CI) | B (SE) | OR (95% CI) | B (SE) | Wald 95% confidence limit | ||
Intercept | –0.80 (0.76) | —e | –1.94f (0.60) | — | –1.87f (0.63) | — | 5.50f (0.73) | 4.06, 6.94 | |
Age | –0.04 (0.05) | 0.96 (0.88-1.06) | 0.11f (0.04) | 1.12 (1.04-1.20) | 0.06 (0.04) | 1.06 (0.98-1.14) | 0.09f (0.05) | 0.005, 0.18 | |
Not white | –0.06 (0.12) | 0.94 (0.74-1,19) | –0.48f (0.10) | 0.62 (0.51-0.75) | 0.14 (0.10) | 1.15 (0.94-1.39) | –0.29f (0.11) | –0.52, –0.07 | |
Male versus female | –0.77f (0.12) | 0.46 (0.37-0.59) | –1.34f (0.10) | 0.26 (0.22-0.32) | –0.84f (0.10) | 0.43 (0.36-0.53) | –1.88f (0.11) | –2.10, –1.67 | |
Other versus female | 1.54f (0.29) | 4.68 (2.65-8.28) | 1.08f (0.29) | 2.94 (1.68-5.14) | 1.03f (0.27) | 2.81 (1.65-4.78) | 0.91f (0.39) | 0.14, 1.68 |
aUnstandardized parameter estimate.
bSE: standard error.
cOR: odds ratio.
dCI: Confidence Interval.
eNot available.
fRepresents significant findings,
More students indicated that they could openly and honestly discuss stress or problems with friends (1874/2682, 69.87%) than with parents or guardians (1204/2682, 44.89%); teachers, guidance counselors, or school staff (360/2682, 13.42%); other adults (270/2682, 10.07%); health professionals (204/2682, 7.60%); or someone else not listed (281/2682, 10.48%). Unfortunately, 19.35% (519/2682) of students reported that they could talk with no one about their stress or problems. The largest zero-order correlation coefficients were observed between talking with a parent and previous suicidality (
Binary logistic regression analyses for combined demographics, suicidality, depression, anxiety, and stress predicting talking with friends, parents, or no one (n=2168).
Predictor | Talk to friend | Talk to parent or guardian | Talk to no one | ||||
Ba (SEb) | ORc (95% CId) | B (SE) | OR (95% CI) | B (SE) | OR (95% CI) | ||
Intercept | 0.96 (0.67) | —e | 0.90 (0.62) | — | –2.07f (0.79) | — | |
Age | 0.02 (0.04) | 1.02 (0.94-1.11) | –0.001 (0.04) | 1.0 (0.93-1.08) | –0.03 (0.05) | 0.98 (0.89-1.07) | |
Race: not white | –0.64f (0.10) | 0.53 (0.43-0.64) | –0.22f (0.10) | 0.81 (0.67-0.98) | 0.28f (0.12) | 1.32 (1.05-1.68) | |
Male versus female | –0.37f (0.11) | 0.69 (0.56-0.85) | –0.32f (0.10) | 0.73 (0.60-0.88) | 0.49f (0.13) | 1.63 (1.28-2.09) | |
Other versus female | –0.55 (0.34) | 0.58 (0.30-1.12) | –0.83g (0.42) | 0.44 (0.19-1.00) | 1.31f (0.34) | 3.70 (1.89-7.22) | |
Suicide: Yes | –0.13 (0.15) | 1.14 (0.85-1.53) | –.59f (0.15) | 1.80 (1.34-2.41) | 0.31g (0.16) | 1.37 (1.00-1.87) | |
Depression | –0.10f (0.05) | 0.90 (0.82-0.99) | –0.24f (0.05) | 0.79 (0.72-0.86) | 0.28f (0.06) | 1.32 (1.18-1.48) | |
Anxiety | 0.005 (0.05) | 1.01 (0.91-1.11) | –0.06 (0.04) | 0.95 (0.87-1.03) | –0.13f (0.06) | 0.88 (0.79- 0.99) | |
Stress | 0.04g (0.02) | 1.05 (1.00-1.09) | –0.05f (0.02) | 0.95 (0.91-0.99) | 0.05f (0.03) | 1.06 (1.00-1.12) |
aUnstandardized parameter estimate.
bSE: standard error.
cOR: odds ratio.
dCI: Confidence Interval.
eNot available.
fRepresents significant findings,
gRepresents marginally significant findings,
Overall, 447 students reported using 1 or more of the 4 TMH tools, with most (318/447, 71.1%) of this group using only 1 type. Anonymous online chat (189/2523, 7.49%) and self-help apps or websites (191/2616, 7.30%) were the most common, followed by the crisis text line (158/2615, 6.04%) and online counselor (92/2652, 3.46%). In review of zero-order correlations involving TMH use, suicidality had the largest correlations with all 4 types (TMH website:
Ordinal logistic regression analyses for combined variables predicting previous telemental health tool use (app or website, anonymous online chat, online counselor, crisis text line).
Model and predictors | B coefficient (SEa) | Odds ratio (95% CIb) | ||
Intercept 1 | –2.42c (0.14) | —d | ||
Intercept 2 | –3.02c (0.15) | — | ||
Male versus female | –0.32c (0.14) | 0.73 (0.55-0.96) | ||
Other versus female | 0.59e (0.30) | 1.80 (1.00-3.26) | ||
Suicide: Yes | 0.46c (0.16) | 1.59 (1.17-2.16) | ||
Anxiety | 0.25c (0.05) | 1.28 (1.17-1.41) | ||
Intercept 1 | –2.46c (0.13) | — | ||
Intercept 2 | –3.12c (0.14) | — | ||
Male versus female | –0.15 (0.14) | 0.86 (0.66-1.13) | ||
Other versus female | 1.06c (0.29) | 2.89 (1.64-5.07) | ||
Suicide: Yes | 0.36c (0.17) | 1.43 (1.03-1.99) | ||
Depression | 0.28c (0.05) | 1.33 (1.20-1.47) | ||
Intercept 1 | –3.83c (0.31) | — | ||
Intercept 2 | –4.38c (0.32) | — | ||
Race: Not white | 0.53c (0.19) | 1.70 (1.18-2.47) | ||
Male versus female | 0.19 (0.20) | 1.21 (0.81-1.80) | ||
Other versus female | 1.49c (0.39) | 4.42 (2.04-9.57) | ||
Suicide: Yes | 0.69c (0.23) | 2.00 (1.29-3.12) | ||
Stress | 0.09c (0.04) | 1.09 (1.01-1.18) | ||
Intercept 1 | –2.70c (0.14) | — | ||
Intercept 2 | –3.58c (0.15) | — | ||
Race: Not white | 0.66c (0.13) | 1.94 (1.51-2.50) | ||
Male versus female | 0.27c (0.13) | 1.31 (1.01-1.70) | ||
Other versus female | 0.98c (0.32) | 2.67 (1.43-5.00) | ||
Suicide: Yes | 0.88c (0.16) | 2.41 (1.74-3.32) | ||
Depression | 0.14c (0.05) | 1.15 (1.04-1.27) |
aSE: standard error.
bCI: Confidence Interval.
cRepresents significant findings,
dNot available.
eRepresents marginally significant findings,
Youth in this high school sample reported high rates of anxiety and depression symptoms and suicidal ideation that were consistent with national averages among teenagers (31% [
Prior research has found similarly low TMH usage rates: in 1 recent study of patients who were older than our sample (average age=57.5 years), but
Depression and anxiety symptoms occurred at higher rates among female students, consistent with national trends [
In general, teens who reported issues with their mental health were more likely to have utilized TMH resources than those who did not, indicating that those who are struggling are attempting to find help and TMH has potential to serve these groups. Teens experiencing suicidality were more likely to have used all 4 categories of TMH resources, those with depressive symptoms were more likely to have used anonymous chat and the crisis text line, and those with higher stress were more likely have used an online therapist. A related theme that emerged, reflecting previous findings [
In contrast, teens reporting anxiety, as well as those who did not report any mental health symptoms, were more likely to talk to parents or friends about their problems and less likely to talk to no one. Overall, they had not used the full scope of TMH resources compared with those struggling with depression and suicidality, although they were more likely to have tried self-help apps. Together, this collection of preferences related to symptom type and available confidants represents an important consideration for researchers and developers of TMH services. Future research could further evaluate these unique preferences among symptom groups, and, assuming preferences remain, engage these target end users in participatory design and usability testing related to the specific types of apps they have shown preferences for. For example, those designing services with the goal of helping with depression and stress management could focus on evidence-based services that allow anonymous disclosure (ie, online therapy and anonymous chat) and allow these potential users to give feedback on their experiences using prototypes of the service before it is released. Services designed to reach those with anxiety could use a self-help model, and all services could potentially include information about resources (both tech-based and in local communities) to help those considering suicide.
Nearly a third of the students in our study who reported recent anxiety or depression symptoms, high stress, or suicidal ideation also reported prior use of a TMH resource. However, we did not investigate in detail whether they found TMH helpful, although this is an important area of consideration for future study. In asking about prior use, we allowed students to select “yes, and it was helpful” or “yes, but it was not helpful,” which were collapsed into 1 answer of “yes” for most analyses. These responses were split nearly evenly overall (49/51%, respectively), with little variation across resource categories. However, these data alone do not provide much insight without further information regarding exact services participants were using, how they hoped to benefit from them, and why they were unsatisfied with their experience. Other research has suggested that insufficient personalization of resources is likely a factor affecting use but emphasized the difficulty of drawing these conclusions merely from survey data [
Even when effective TMH resources are created, there should be continued work to integrate them into the health care system, likely in conjunction with face-to-face care, as this has not yet been sustainably executed on a large scale [
TMH also has the potential to play an important role in suicide prevention—youth in this study who had considered suicide were more likely to have used all categories of resources. Given the disparity between urban and rural suicide rates [
Finally, participants in prior research have cited data protection, information security, and anonymity as important concerns related to health care delivered via apps or websites [
Our sample included only students enrolled in and present at school, whose parents had not opted out. Several variables (ie, male or other gender, minority race, higher depression level) related positively to nonresponse, potentially limiting data for topics particularly sensitive to these individuals. The setting of the survey (ie, an auditorium or gymnasium in which students were seated near each other and could not be truly prevented from speaking) could have contributed to nonresponse or skewed responses toward lower reporting of stigmatized topics. Conversely, the group setting may have kept other students engaged. In addition, the themed presentation may have influenced students to respond either positively or negatively to the TMH-related questions, if they assumed we were seeking positive response and chose to fulfill or deny this expectation. However, we do not believe these limitations would necessarily create strong trends in any specific direction, nor would they be entirely mitigated in a different setting (ie, classroom or computer lab).
In addition, the number of students reporting their gender as “other” was small, but gender-nonconformity was associated with higher rates of depression and anxiety symptoms, suicidal ideation, stress, inability to discuss stress or problems openly and honestly with a parent (or anyone), and prior use of all 4 TMH categories. We did not record sexual orientation, but given the consistency of gender nonconformity as a predictor, future research should explore both gender and sexual minority status in relation to mental health help-seeking. These 2 groups, which sometimes overlap, face similar sets of social and familial challenges and may uniquely benefit from the lack of required parental involvement involved in TMH help-seeking.
This high school survey was part of a larger study about TMH usage that also included college students; thus, the PHQ-4 was used for consistency among both groups, although it has not been specifically validated with adolescents. However, over 37% of our sample for this study was aged 17 to 19 years (for whom the PHQ-4 has been validated [
Overall, our results indicate that teenagers experiencing mental distress are utilizing existing TMH resources at a moderate rate consistent with extant literature. Type of resource usage correlated with mental health and demographic variables, providing a framework for future research and targeted resource development. Suicidality and gender nonconformity predicted use of all 4 categories of resources; depression, anxiety, and stress all predicted use of at least 1 unique type. In addition, suicidality, depression, and stress were correlated with lacking confidants with whom to discuss stress or problems, whereas those with anxiety were less likely to report this. As the mental health field progresses toward electronically-based care, it is important to consider findings such as these to provide appropriate interventions that target specific populations for effective and tailored care or supplementation of care [
Patient Health Questionnaire
telemental health
Support for this research was provided by the Robert Wood Johnson Foundation (grant #73055). The views expressed here do not necessarily reflect the views of the foundation. The authors would like to acknowledge and thank the other members of their study team who made this study possible, including a number of people in the community. They would also like to thank the high school administrators and students that participated in this research.
None declared.