https://orcid.org/0000-0001-8455-9757
-
PDF
- Split View
-
Views
-
Cite
Cite
Xin Zhao, Stephen M Schueller, Jeongmi Kim, Nicole A Stadnick, Elizabeth Eikey, Margaret Schneider, Kai Zheng, Dana B Mukamel, Dara H Sorkin, Real-World Adoption of Mental Health Support Among Adolescents: Cross-Sectional Analysis of the California Health Interview Survey, Journal of Pediatric Psychology, Volume 50, Issue 1, January 2025, Pages 20–29, https://doi.org/10.1093/jpepsy/jsad082
Close - Share Icon Share
Abstract
We aim to examine: (a) the extent to which patterns of adoption of counseling services and digital mental health interventions (DMHIs) shifted in recent years (2019–2021); (b) the impact of distress on adoption of mental health support; and (c) reasons related to adolescents’ low adoption of DMHIs when experiencing distress.
Data were from three cohorts of adolescents aged 12–17 years (n = 847 in 2019; n = 1,365 in 2020; n = 1,169 in 2021) recruited as part of the California Health Interview Survey. We estimated logistic regression models to examine the likelihood of using mental health support as a function of psychological distress, sociodemographic characteristics, and cohorts. We also analyzed adolescents’ self-reported reasons for not trying DMHIs as a function of distress.
The proportion of adolescents reporting elevated psychological distress (∼50%) was higher than those adopting counseling services (<20%) or DMHIs (<10%). A higher level of distress was associated with a greater likelihood of receiving counseling (OR = 1.15), and using DMHIs to connect with a professional (Odds ratio (OR) = 1.11) and for self-help (OR = 1.17). Among those experiencing high distress, adolescents’ top reason for not adopting an online tool was a lack of perceived need (19.2%).
Adolescents’ main barriers to DMHI adoption included a lack of perceived need, which may be explained by a lack of mental health literacy. Thoughtful marketing and dissemination efforts are needed to increase mental health awareness and normalize adoption of counseling services and DMHIs.
Introduction
Unmet mental health needs remain a longstanding problem in pediatric settings, despite significant advances in evidence-based interventions, such as modular and brief interventions (Frederick et al., 2023; McDanal et al., 2022). A meta-analysis found that less than 10% of adolescents with mental health needs receive care in pediatric primary care settings (Duong et al., 2021). On the other hand, access to and use of technology is high (95% reported in the 2018 Pew Survey) (Schaeffer, 2019). Many adolescents are regular smartphone users, making digital mental health interventions (DMHIs) particularly relevant for pediatric populations and service settings (Bell et al., 2022). Due to their potential to provide mental health care in an accessible, minimally stigmatizing, and scalable way (Lattie et al., 2022; Wies et al., 2021), DMHIs are a promising solution to unmet mental health needs. However, for DMHIs to be truly impactful on a population level, high uptake is necessary (Kodish et al., 2023). Adoption, also referred to as uptake, is a key implementation outcome and has been defined as “the intention, initial decision, or action to try or employ an innovation” (Proctor et al., 2023). The objective of this article was to examine factors related to adolescents’ naturalistic adoption of DMHIs, that is, their use of these resources in real-world settings.
DMHIs are broadly defined as digital translations of psychosocial support and/or protocolized treatments that may be provided through mobile applications or other digital formats (Lehtimaki et al., 2021). Controlled trials and open trials provide compelling evidence for the efficacy and effectiveness of DMHIs in addressing various mental health conditions, such as pediatric depression and anxiety (Khanna & Carper, 2022). Supported DMHIs have moderate to large effect sizes, but even unsupported DMHIs tend to have consistent, albeit small, effect sizes (Linardon et al., 2019), although these small effect sizes may still be meaningful from a population level if these resources were widely adopted.
Over recent years, an increase in mental health demands and massive shifts in service delivery models have led to growing interest in DMHIs. The coronavirus disease 2019 (COVID-19) pandemic played a role in this, as it forced service providers to shift to telehealth and reduced in-person capacity. This had both negative and positive effects, but it did lead to the rapid development of DMHIs and attempts to integrate DMHI into diverse practice settings at the federal, state, and organizational levels (Lattie et al., 2022; Palinkas et al., 2021). Recent studies of service delivery revealed youth’s generally positive attitudes toward virtual services (Nicholas et al., 2021; Zolopa et al., 2022). Researchers reported increased therapy attendance of adolescents via telehealth formats in 2020 compared to 2019 (Nicholas et al., 2021).
Despite the infrastructure changes and growing interest in DMHIs, DMHI implementation remains challenging. Adoption and engagement rates of DMHIs in real-world contexts among adolescents are low (Bell et al., 2022; Garrido et al., 2019). A nationally representative survey of teenagers and young adults (14–22 years of age) found that only 19% had used a mental health app, although it was higher among those with moderate-to-severe depressive symptoms at 46% (Rideout et al., 2021). Acceptability, such as positive attitudes (Zolopa et al., 2022) and acceptance of virtual services (e.g., Hawke et al., 2021), did not lead to adoption (i.e., intention of trying or actual usage of DMHIs). A recent survey reporting on early pandemic data in Australia revealed that although 77% of youth from the general population endorsed the belief that DMHIs were helpful, only half of them used a DMHI (Bell et al., 2022). Thus, it is important to understand factors associated with naturalistic adoption of DMHIs and how these factors may be similar to and/or different from their naturalistic adoption of traditional services.
Barriers to adolescents’ adoption of counseling services have been extensively studied and are multi-level (Aguirre Velasco et al., 2020; Radez et al., 2021). At the familial and individual levels, negative beliefs about healthcare systems, low levels of mental health literacy, and a desire for autonomy and independence have been reported as adolescents’ help-seeking barriers (Aguirre Velasco et al., 2020; Radez et al., 2021). Paradoxically, help avoidance may be related to higher levels of psychological distress, meaning that those most in need may fail to seek services (Aguirre Velasco et al., 2020). At the system level, challenges, such as persisting provider shortages (Hoffmann et al., 2023), make it difficult to keep pace with the growing mental health needs of adolescents. At the societal level, mental health stigma and structural racism contribute to inequitable service utilization; Black, Asian American, and Latinx adolescents are less likely to seek or receive services, in part due to mistrust in the mental health system (Alvarez et al., 2022).
Barriers to adolescents’ DMHI adoption have been less studied, though they also appear multifaceted (Pretorius et al., 2019). Similar to adoption of professional services, at the individual level, a lack of perceived need for services, inadequate mental health literacy, and mistrust have been identified as barriers to adoption of DMHIs among adolescents and young adults (Kodish et al., 2023; Pretorius et al., 2019). In a German sample, youth with COVID-19-related psychological distress were more likely to report positive attitudes toward and actual use of DMHIs (Rauschenberg et al., 2021). Other reasons for not being willing to use DMHIs included a lack of human connection, privacy issues, and technical concerns (e.g., poor internet connection) (Hawke et al., 2021). Individuals from racial/ethnic minority groups, with chronic illnesses and disabilities, may face additional barriers to DMHI access. For example, compared to their typically developing peers, adolescents with intellectual disability were less likely to own a digital device, such as a smartphone, and they experienced more difficulty searching for and understanding online information (Alfredsson Ågren et al., 2020), which limited their access to DMHIs. In addition to access to devices and information, many existing DMHIs were designed by and for middle or high socioeconomic status (SES), non-disabled, heterosexual, and non-Latinx white populations (Ramos et al., 2021), which limited their usability and helpfulness for marginalized communities. These barriers are often understudied, demonstrating the structural inequities within DMHI research and practice efforts, and a need for additional research.
The Current Study
Considering the transformative changes in service delivery models, service utilization patterns, and attitudes toward DMHIs that took place due to COVID-19 lockdowns, understanding factors related to adolescents’ service adoption in recent years is needed. Our aims included: (a) the extent to which patterns of naturalistic adoption of counseling services and DMHIs shifted in recent years (2019–2021) among adolescents; (b) the association between psychological distress and naturalistic adoption of counseling services and DMHIs; and (c) adolescents’ reasons for not adopting DMHIs when experiencing psychological distress in real-world settings. Given what is known about changing beliefs and behaviors around mental health services among adolescents (Bell et al., 2022; Nicholas et al., 2021) and the barriers to service adoption (Aguirre Velasco et al., 2020; Pretorius et al., 2019), we investigated the hypotheses that (a) higher rates of using counseling services and online tools would be observed in recent cohorts; (b) adolescents with higher levels of psychological distress would be more likely to use online tools due to their mental health need but may avoid seeking professional support due to their high levels of distress; and (c) when experiencing psychological distress, adolescents’ main reasons for not trying online tools for mental health support would include lack of perceived need and lack of perceived helpfulness.
Methods
We used data from the California Health Interview Survey (CHIS). CHIS employs an address-based sample design and geographical stratification to create a representative sample of the noninstitutionalized population in California. Surveys were available in six languages, including English, Spanish, Chinese (Mandarin and Cantonese dialects), Korean, Vietnamese, and Tagalog. Further information about CHIS methodology is available elsewhere (UCLA Center for Health Policy Research, 2021, 2022). Households that had eligible adolescents (aged 12–17 years) were invited to have one randomly selected adolescent complete the survey online or by telephone after parental consent was obtained. These interviews and surveys offer opportunities to understand population-level patterns of health behaviors of adolescents. We used cross-sectional data from the adolescent sub-sample of the CHIS that included three annual cohorts (n = 847 in 2019; n = 1,365 in 2020; n = 1,169 in 2021), resulting in a total N of 3,381. Variables related to using DMHIs were only available for 2019–2021 at the time of data analyses.
Variables
Psychological Distress
The CHIS assesses psychological distress by asking questions on the Kessler-6 (Kessler et al., 2002). K6 assesses six symptoms: felt nervous, hopeless, restless or fidgety, worthless, depressed, or felt everything was an effort. K6 was demonstrated to have satisfactory psychometric properties during its initial development and its validation in the CHIS sample (Kessler et al., 2002; Prochaska et al., 2012). Response options included “none” (0), “a little” (1), “some” (2), “most” (3), or “all” (4), “refused,” and “don’t know.” The sum scores (“K6”) range from 0 to 24, with higher scores indicating higher levels of psychological distress. A categorical variable that had three levels, including none to low (0 ≤ K6 ≤ 5), medium (6 ≤ K6 ≤ 12), and high (K6 ≥ 13) levels of psychological distress was also generated; these cut-off values were based on prior psychometric work using receiver operating characteristic curve analyses (Prochaska et al., 2012). Past analyses using data from the CHIS adult sample suggested that individuals with medium (6 ≤ K6 ≤ 12) or high (K6 ≥ 13) levels of psychological distress were more likely to report experiencing functional impairment and needing treatment (Prochaska et al., 2012).
Use of Professional Help for Mental Health Problems
To assess the use of mental health support, all respondents were asked the following question: “In the past 12 months, have you received any psychological or emotional counseling.” Counseling may be either in-person or virtual.
Use of Online Tools for Mental Health Support
To assess the use of an online tool specifically for connecting with a professional, all respondents were asked: “In the last 12 months, have you used online tools to find, be referred to, contact, or connect with a mental health professional?” To assess the use of an online tool for self-help, respondents were asked “In the past 12 months, have you tried to get help from an online tool, including mobile apps or texting services, for problems with your mental health, emotions, nerves, or your use of alcohol or drugs?” Those who responded “no” to this question were asked about the “main reason they did not try to get support from an online tool, including mobile apps or texting services.” Adolescents were asked to choose from a list of 12 reasons, including “got better/no longer needed,” “wanted to handle problem on own,” “don't own a smartphone or computer,” “didn't know about these apps,” “don't trust mobile apps,” “concerns about privacy and security of the data,” “don't think it would be helpful or work,” “cost,” “don’t have time,” “received traditional/face-to-face services,” “I don't think I needed it,” and “don't have enough space to download new apps.” Adolescents can also specify their own reasons, refuse to answer, or select “don’t know.”
Race and Ethnicity
To collect information about ethnicity, respondents were asked to answer, “Are you Latino or Hispanic?.” We use “Latinx” for consistency, gender inclusivity, and simplicity in this paper. Adolescents were asked to select one or more response options to describe their racial backgrounds; response options included “White,” “Black or African American,” “Asian,” “American Indian or Alaska Native,” and “Pacific Islander or Native Hawaiian.” Adolescents can specify their own reasons, refuse to answer, or select “don’t know.” For simplicity’s sake, we use “Black” in this paper. “American Indian or Alaska Native” and “Pacific Islander or Native Hawaiian” were recoded into “other.”
Analyses
Weighting procedures used for CHIS compensated for selection bias and respondent bias, adjusted for survey design and administration factors, and reduced the variance of the estimates by using auxiliary information (Sherr et al., 2022). Weighted means and standard errors were reported for continuous variables. Weighted frequencies and proportions were reported for categorical variables. We ran three multivariable logistic regression models, each model contains the same set of a priori determined predictor variables, informed by the literature, and a unique dependent variable linked to our research questions and hypotheses. The predictor variables for each model included psychological distress (continuous variable), cohort (categorical variable; 2019 was the reference level), the interaction term between distress and cohort, race/ethnicity, sex, age, and country of birth (whether the respondent was born in the US). The dependent variables were receiving counseling for psychosocial or emotional problems for model 1, using an online tool to connect with a professional for model 2, and using an online tool for self-help for model 3. For those who reported having not used an online tool for mental health, we also analyzed weighted frequency and percentages stratified by categorical distress variable to demonstrate adolescents’ main reasons for not trying to get support from an online tool.
UCLA CHIS Data Access Center (DAC) performed analyses using SAS 9.4. The UCLA South General IRB has approved DAC to conduct analyses on confidential CHIS data (IRB#11-002227).
Results
Sample Characteristics
The mean age across three annual cohorts (2019–2021) was 14.61 years (SD = 0.02). The sample included 43.9% Latinx, 34.4.% White, non-Latinx, 12.1% Asian, non-Latinx, and 5.2% Black, non-Latinx adolescent respondents. Approximately half of the sample was female (49.0%). Most adolescents in the analytic sample were born in the US (92.2%). Additional sociodemographic information by annual cohort is reported in Table I.
| Overall sample . | 2019 . | 2020 . | 2021 . | |||||
|---|---|---|---|---|---|---|---|---|
| Weighted n or M . | Percent or SE . | Weighted n or M . | Percent or SE . | Weighted n or M . | Percent or SE . | Weighted n or M . | Percent or SE . | |
| Age (M, SD) | 14.6 | 0.02 | 14.6 | 0.04 | 14.6 | 0.03 | 14.6 | 0.03 |
| Race/ethnicity (n, %) | ||||||||
| Black, non-Latinx (n, %) | 161,822 | 5.2 | 52,350 | 5.0 | 62,044 | 5.9 | 47,429 | 4.6 |
| Asian, non-Latinx (n, %) | 377,639 | 12.1 | 133,542 | 12.8 | 134,056 | 12.8 | 110,041 | 10.6 |
| Latinx (n, %) | 1,372,505 | 43.9 | 418,578 | 40.1 | 420,527 | 40.1 | 533,401 | 51.5 |
| White, non-Latinx (n, %) | 1,076,172 | 34.4 | 397,105 | 38.0 | 396,655 | 37.9 | 282,412 | 27.2 |
| Other (n, %) | 141,102 | 4.5 | 43,311 | 4.2 | 34,403 | 3.3 | 63,388 | 6.1 |
| Female (n, %) | 1,532,309 | 49.0 | 513,332 | 49.1 | 514,513 | 49.1 | 504,464 | 48.7 |
| US born (n, %) | 2,885,462 | 92.2 | 974,624 | 93.3 | 958,319 | 91.5 | 952,518 | 91.9 |
| Psychological distress (continuous: M, SD)a | 6.5 | 0.1 | 6.7 | 0.2 | 6.1 | 0.2 | 6.8 | 0.2 |
| Psychological distress (categorical)a,b | ||||||||
| None to low (n, %) | 1,625,482 | 51.9 | 512,461 | 49.0 | 579,136 | 55.3 | 533,886 | 51.5 |
| Medium (n, %) | 1,060,446 | 33.9 | 400,791 | 38.4 | 330,393 | 31.5 | 329,262 | 31.8 |
| High (n, %) | 443,313 | 14.2 | 131,634 | 12.6 | 138,156 | 13.2 | 173,523 | 16.7 |
| Received psychological or emotional counseling (n, %) | 584,519 | 18.7 | 198,971 | 19.0 | 169,818 | 16.2 | 215,730 | 20.8 |
| Used an online tool to connect with a professional (n, %)b | 219,458 | 7.1 | 55,190 | 5.4 | 70,229 | 6.7 | 94,039 | 9.1 |
| Used an online tool for self-help (n, %) | 244,595 | 7.8 | 70,452 | 6.8 | 80,673 | 7.7 | 93,470 | 9.0 |
| Overall sample . | 2019 . | 2020 . | 2021 . | |||||
|---|---|---|---|---|---|---|---|---|
| Weighted n or M . | Percent or SE . | Weighted n or M . | Percent or SE . | Weighted n or M . | Percent or SE . | Weighted n or M . | Percent or SE . | |
| Age (M, SD) | 14.6 | 0.02 | 14.6 | 0.04 | 14.6 | 0.03 | 14.6 | 0.03 |
| Race/ethnicity (n, %) | ||||||||
| Black, non-Latinx (n, %) | 161,822 | 5.2 | 52,350 | 5.0 | 62,044 | 5.9 | 47,429 | 4.6 |
| Asian, non-Latinx (n, %) | 377,639 | 12.1 | 133,542 | 12.8 | 134,056 | 12.8 | 110,041 | 10.6 |
| Latinx (n, %) | 1,372,505 | 43.9 | 418,578 | 40.1 | 420,527 | 40.1 | 533,401 | 51.5 |
| White, non-Latinx (n, %) | 1,076,172 | 34.4 | 397,105 | 38.0 | 396,655 | 37.9 | 282,412 | 27.2 |
| Other (n, %) | 141,102 | 4.5 | 43,311 | 4.2 | 34,403 | 3.3 | 63,388 | 6.1 |
| Female (n, %) | 1,532,309 | 49.0 | 513,332 | 49.1 | 514,513 | 49.1 | 504,464 | 48.7 |
| US born (n, %) | 2,885,462 | 92.2 | 974,624 | 93.3 | 958,319 | 91.5 | 952,518 | 91.9 |
| Psychological distress (continuous: M, SD)a | 6.5 | 0.1 | 6.7 | 0.2 | 6.1 | 0.2 | 6.8 | 0.2 |
| Psychological distress (categorical)a,b | ||||||||
| None to low (n, %) | 1,625,482 | 51.9 | 512,461 | 49.0 | 579,136 | 55.3 | 533,886 | 51.5 |
| Medium (n, %) | 1,060,446 | 33.9 | 400,791 | 38.4 | 330,393 | 31.5 | 329,262 | 31.8 |
| High (n, %) | 443,313 | 14.2 | 131,634 | 12.6 | 138,156 | 13.2 | 173,523 | 16.7 |
| Received psychological or emotional counseling (n, %) | 584,519 | 18.7 | 198,971 | 19.0 | 169,818 | 16.2 | 215,730 | 20.8 |
| Used an online tool to connect with a professional (n, %)b | 219,458 | 7.1 | 55,190 | 5.4 | 70,229 | 6.7 | 94,039 | 9.1 |
| Used an online tool for self-help (n, %) | 244,595 | 7.8 | 70,452 | 6.8 | 80,673 | 7.7 | 93,470 | 9.0 |
Note. SE = standard error. The complex design of the sample requires proper weighting and variance calculation of the estimates; population weights were applied to calculate weighted frequency.
2019 versus 2020, p < .05.
2019 versus 2021, p < .05.
| Overall sample . | 2019 . | 2020 . | 2021 . | |||||
|---|---|---|---|---|---|---|---|---|
| Weighted n or M . | Percent or SE . | Weighted n or M . | Percent or SE . | Weighted n or M . | Percent or SE . | Weighted n or M . | Percent or SE . | |
| Age (M, SD) | 14.6 | 0.02 | 14.6 | 0.04 | 14.6 | 0.03 | 14.6 | 0.03 |
| Race/ethnicity (n, %) | ||||||||
| Black, non-Latinx (n, %) | 161,822 | 5.2 | 52,350 | 5.0 | 62,044 | 5.9 | 47,429 | 4.6 |
| Asian, non-Latinx (n, %) | 377,639 | 12.1 | 133,542 | 12.8 | 134,056 | 12.8 | 110,041 | 10.6 |
| Latinx (n, %) | 1,372,505 | 43.9 | 418,578 | 40.1 | 420,527 | 40.1 | 533,401 | 51.5 |
| White, non-Latinx (n, %) | 1,076,172 | 34.4 | 397,105 | 38.0 | 396,655 | 37.9 | 282,412 | 27.2 |
| Other (n, %) | 141,102 | 4.5 | 43,311 | 4.2 | 34,403 | 3.3 | 63,388 | 6.1 |
| Female (n, %) | 1,532,309 | 49.0 | 513,332 | 49.1 | 514,513 | 49.1 | 504,464 | 48.7 |
| US born (n, %) | 2,885,462 | 92.2 | 974,624 | 93.3 | 958,319 | 91.5 | 952,518 | 91.9 |
| Psychological distress (continuous: M, SD)a | 6.5 | 0.1 | 6.7 | 0.2 | 6.1 | 0.2 | 6.8 | 0.2 |
| Psychological distress (categorical)a,b | ||||||||
| None to low (n, %) | 1,625,482 | 51.9 | 512,461 | 49.0 | 579,136 | 55.3 | 533,886 | 51.5 |
| Medium (n, %) | 1,060,446 | 33.9 | 400,791 | 38.4 | 330,393 | 31.5 | 329,262 | 31.8 |
| High (n, %) | 443,313 | 14.2 | 131,634 | 12.6 | 138,156 | 13.2 | 173,523 | 16.7 |
| Received psychological or emotional counseling (n, %) | 584,519 | 18.7 | 198,971 | 19.0 | 169,818 | 16.2 | 215,730 | 20.8 |
| Used an online tool to connect with a professional (n, %)b | 219,458 | 7.1 | 55,190 | 5.4 | 70,229 | 6.7 | 94,039 | 9.1 |
| Used an online tool for self-help (n, %) | 244,595 | 7.8 | 70,452 | 6.8 | 80,673 | 7.7 | 93,470 | 9.0 |
| Overall sample . | 2019 . | 2020 . | 2021 . | |||||
|---|---|---|---|---|---|---|---|---|
| Weighted n or M . | Percent or SE . | Weighted n or M . | Percent or SE . | Weighted n or M . | Percent or SE . | Weighted n or M . | Percent or SE . | |
| Age (M, SD) | 14.6 | 0.02 | 14.6 | 0.04 | 14.6 | 0.03 | 14.6 | 0.03 |
| Race/ethnicity (n, %) | ||||||||
| Black, non-Latinx (n, %) | 161,822 | 5.2 | 52,350 | 5.0 | 62,044 | 5.9 | 47,429 | 4.6 |
| Asian, non-Latinx (n, %) | 377,639 | 12.1 | 133,542 | 12.8 | 134,056 | 12.8 | 110,041 | 10.6 |
| Latinx (n, %) | 1,372,505 | 43.9 | 418,578 | 40.1 | 420,527 | 40.1 | 533,401 | 51.5 |
| White, non-Latinx (n, %) | 1,076,172 | 34.4 | 397,105 | 38.0 | 396,655 | 37.9 | 282,412 | 27.2 |
| Other (n, %) | 141,102 | 4.5 | 43,311 | 4.2 | 34,403 | 3.3 | 63,388 | 6.1 |
| Female (n, %) | 1,532,309 | 49.0 | 513,332 | 49.1 | 514,513 | 49.1 | 504,464 | 48.7 |
| US born (n, %) | 2,885,462 | 92.2 | 974,624 | 93.3 | 958,319 | 91.5 | 952,518 | 91.9 |
| Psychological distress (continuous: M, SD)a | 6.5 | 0.1 | 6.7 | 0.2 | 6.1 | 0.2 | 6.8 | 0.2 |
| Psychological distress (categorical)a,b | ||||||||
| None to low (n, %) | 1,625,482 | 51.9 | 512,461 | 49.0 | 579,136 | 55.3 | 533,886 | 51.5 |
| Medium (n, %) | 1,060,446 | 33.9 | 400,791 | 38.4 | 330,393 | 31.5 | 329,262 | 31.8 |
| High (n, %) | 443,313 | 14.2 | 131,634 | 12.6 | 138,156 | 13.2 | 173,523 | 16.7 |
| Received psychological or emotional counseling (n, %) | 584,519 | 18.7 | 198,971 | 19.0 | 169,818 | 16.2 | 215,730 | 20.8 |
| Used an online tool to connect with a professional (n, %)b | 219,458 | 7.1 | 55,190 | 5.4 | 70,229 | 6.7 | 94,039 | 9.1 |
| Used an online tool for self-help (n, %) | 244,595 | 7.8 | 70,452 | 6.8 | 80,673 | 7.7 | 93,470 | 9.0 |
Note. SE = standard error. The complex design of the sample requires proper weighting and variance calculation of the estimates; population weights were applied to calculate weighted frequency.
2019 versus 2020, p < .05.
2019 versus 2021, p < .05.
Patterns of Adopting Counseling Services and DMHIs
Rates of receiving counseling for psychosocial and emotional concerns (<20%), rates of using an online tool to connect with a professional (<10%), and rates of using an online tool for self-help (<10%) were far below the percentages of adolescents with elevated levels of distress (∼50%) from 2019 to 2021 (Table I). Minimal cohort effects were detected and a post hoc sensitivity analysis suggested that with our sample size at α = 0.05 and 80% power, we had sufficient power to detect parameter estimates as small as 0.28 for most estimates. In our regression analyses (Table II), rates of receiving counseling were not significantly different by cohort, ps > .05 in model 1. Rates of using an online tool to connect with a professional (model 2) or for self-help (model 3) were not significantly different by cohort, ps > .05.
Correlates of Use of Mental Health Support Based on Three Separate Logistic Models
| Received psychosocial or emotional counseling model 1: N = 3,381 . | Used an online tool to connect with a professional model 2: N = 3,374 . | Used an online tool for self-help model 3: N = 3,378 . | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimates . | SE . | t . | p . | Estimates . | SE . | T . | p . | Estimates . | SE . | t . | p . | |
| Parameter estimates | ||||||||||||
| Intercepts | −2.04 | 0.31 | −6.6 | <.001 | −3.88 | 0.4891 | −7.94 | <.001 | −3.71 | 0.52 | −7.18 | <.001 |
| Year (reference = 2019) | ||||||||||||
| 2020 | 0.14 | 0.27 | 0.51 | 0.61 | −0.19 | 0.42 | −0.46 | .649 | 0.72 | 0.42 | 1.69 | .092 |
| 2021 | 0.21 | 0.25 | 0.81 | 0.416 | 0.52 | 0.35 | 1.46 | .145 | 0.7 | 0.42 | 1.65 | .101 |
| Psychological distress | 0.14 | 0.02 | 6.62 | <.001 | 0.1 | 0.03 | 4.02 | <.001 | 0.16 | 0.03 | 4.77 | <.001 |
| Distress × year | ||||||||||||
| 2020 | −0.03 | 0.03 | −1.23 | 0.221 | 0.06 | 0.03 | 1.7 | .09 | −0.04 | 0.04 | −1.06 | .291 |
| 2021 | 0 | 0.03 | −0.11 | 0.91 | 0.01 | 0.03 | 0.26 | .793 | −0.04 | 0.04 | −1 | .318 |
| Race/ethnicity (reference = White, non-Latinx) | ||||||||||||
| Asian, non-Latinx | −1.98 | 0.28 | −7.06 | <.001 | −1.53 | 0.36 | −4.21 | <.001 | −0.24 | 0.28 | −0.87 | .384 |
| Latinx | −1.06 | 0.13 | −8.35 | <.001 | −0.81 | 0.19 | −4.19 | <.001 | −0.36 | 0.16 | −2.28 | .024 |
| Black, non-Latinx | −0.15 | 0.3 | −0.52 | 0.606 | −0.12 | 0.49 | −0.24 | .814 | −1.36 | 0.59 | −2.28 | .023 |
| Others | −0.4 | 0.18 | −2.22 | 0.028 | −0.40 | 0.29 | −1.42 | .158 | −0.07 | 0.29 | −0.25 | .803 |
| Sex (reference = female) | −0.23 | 0.13 | −1.84 | 0.067 | −0.21 | 0.17 | −1.22 | .222 | −0.37 | 0.15 | −2.42 | .016 |
| Age | 0.07 | 0.04 | 1.76 | 0.079 | 0.14 | 0.06 | 2.29 | .023 | 0.23 | 0.05 | 4.45 | <.001 |
| US born (reference = non-US born) | 0.21 | 0.23 | 0.89 | 0.375 | 0.76 | 0.35 | 2.18 | <.001 | 0.01 | 0.33 | 0.04 | .966 |
| F | Num df | Den df | p | F | Num df | Den df | p | F | Num df | Den df | p | |
| Model statistics and fit | ||||||||||||
| Likelihood ratio | 55.22 | 10.75 | 2,590.76 | <.001 | 23.6 | 10.27 | 2,474.24 | <.001 | 24.58 | 9.83 | 2,368.71 | <.001 |
| Score | 25.14 | 12 | 241 | <.001 | 10.23 | 12 | 241 | <.001 | 12.38 | 12 | 241 | <.001 |
| Wald | 21.2 | 12 | 241 | <.001 | 14.96 | 12 | 241 | <.001 | 11.83 | 12 | 241 | <.001 |
| Received psychosocial or emotional counseling model 1: N = 3,381 . | Used an online tool to connect with a professional model 2: N = 3,374 . | Used an online tool for self-help model 3: N = 3,378 . | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimates . | SE . | t . | p . | Estimates . | SE . | T . | p . | Estimates . | SE . | t . | p . | |
| Parameter estimates | ||||||||||||
| Intercepts | −2.04 | 0.31 | −6.6 | <.001 | −3.88 | 0.4891 | −7.94 | <.001 | −3.71 | 0.52 | −7.18 | <.001 |
| Year (reference = 2019) | ||||||||||||
| 2020 | 0.14 | 0.27 | 0.51 | 0.61 | −0.19 | 0.42 | −0.46 | .649 | 0.72 | 0.42 | 1.69 | .092 |
| 2021 | 0.21 | 0.25 | 0.81 | 0.416 | 0.52 | 0.35 | 1.46 | .145 | 0.7 | 0.42 | 1.65 | .101 |
| Psychological distress | 0.14 | 0.02 | 6.62 | <.001 | 0.1 | 0.03 | 4.02 | <.001 | 0.16 | 0.03 | 4.77 | <.001 |
| Distress × year | ||||||||||||
| 2020 | −0.03 | 0.03 | −1.23 | 0.221 | 0.06 | 0.03 | 1.7 | .09 | −0.04 | 0.04 | −1.06 | .291 |
| 2021 | 0 | 0.03 | −0.11 | 0.91 | 0.01 | 0.03 | 0.26 | .793 | −0.04 | 0.04 | −1 | .318 |
| Race/ethnicity (reference = White, non-Latinx) | ||||||||||||
| Asian, non-Latinx | −1.98 | 0.28 | −7.06 | <.001 | −1.53 | 0.36 | −4.21 | <.001 | −0.24 | 0.28 | −0.87 | .384 |
| Latinx | −1.06 | 0.13 | −8.35 | <.001 | −0.81 | 0.19 | −4.19 | <.001 | −0.36 | 0.16 | −2.28 | .024 |
| Black, non-Latinx | −0.15 | 0.3 | −0.52 | 0.606 | −0.12 | 0.49 | −0.24 | .814 | −1.36 | 0.59 | −2.28 | .023 |
| Others | −0.4 | 0.18 | −2.22 | 0.028 | −0.40 | 0.29 | −1.42 | .158 | −0.07 | 0.29 | −0.25 | .803 |
| Sex (reference = female) | −0.23 | 0.13 | −1.84 | 0.067 | −0.21 | 0.17 | −1.22 | .222 | −0.37 | 0.15 | −2.42 | .016 |
| Age | 0.07 | 0.04 | 1.76 | 0.079 | 0.14 | 0.06 | 2.29 | .023 | 0.23 | 0.05 | 4.45 | <.001 |
| US born (reference = non-US born) | 0.21 | 0.23 | 0.89 | 0.375 | 0.76 | 0.35 | 2.18 | <.001 | 0.01 | 0.33 | 0.04 | .966 |
| F | Num df | Den df | p | F | Num df | Den df | p | F | Num df | Den df | p | |
| Model statistics and fit | ||||||||||||
| Likelihood ratio | 55.22 | 10.75 | 2,590.76 | <.001 | 23.6 | 10.27 | 2,474.24 | <.001 | 24.58 | 9.83 | 2,368.71 | <.001 |
| Score | 25.14 | 12 | 241 | <.001 | 10.23 | 12 | 241 | <.001 | 12.38 | 12 | 241 | <.001 |
| Wald | 21.2 | 12 | 241 | <.001 | 14.96 | 12 | 241 | <.001 | 11.83 | 12 | 241 | <.001 |
Note. df = degree of freedom; SE = standard error.
Correlates of Use of Mental Health Support Based on Three Separate Logistic Models
| Received psychosocial or emotional counseling model 1: N = 3,381 . | Used an online tool to connect with a professional model 2: N = 3,374 . | Used an online tool for self-help model 3: N = 3,378 . | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimates . | SE . | t . | p . | Estimates . | SE . | T . | p . | Estimates . | SE . | t . | p . | |
| Parameter estimates | ||||||||||||
| Intercepts | −2.04 | 0.31 | −6.6 | <.001 | −3.88 | 0.4891 | −7.94 | <.001 | −3.71 | 0.52 | −7.18 | <.001 |
| Year (reference = 2019) | ||||||||||||
| 2020 | 0.14 | 0.27 | 0.51 | 0.61 | −0.19 | 0.42 | −0.46 | .649 | 0.72 | 0.42 | 1.69 | .092 |
| 2021 | 0.21 | 0.25 | 0.81 | 0.416 | 0.52 | 0.35 | 1.46 | .145 | 0.7 | 0.42 | 1.65 | .101 |
| Psychological distress | 0.14 | 0.02 | 6.62 | <.001 | 0.1 | 0.03 | 4.02 | <.001 | 0.16 | 0.03 | 4.77 | <.001 |
| Distress × year | ||||||||||||
| 2020 | −0.03 | 0.03 | −1.23 | 0.221 | 0.06 | 0.03 | 1.7 | .09 | −0.04 | 0.04 | −1.06 | .291 |
| 2021 | 0 | 0.03 | −0.11 | 0.91 | 0.01 | 0.03 | 0.26 | .793 | −0.04 | 0.04 | −1 | .318 |
| Race/ethnicity (reference = White, non-Latinx) | ||||||||||||
| Asian, non-Latinx | −1.98 | 0.28 | −7.06 | <.001 | −1.53 | 0.36 | −4.21 | <.001 | −0.24 | 0.28 | −0.87 | .384 |
| Latinx | −1.06 | 0.13 | −8.35 | <.001 | −0.81 | 0.19 | −4.19 | <.001 | −0.36 | 0.16 | −2.28 | .024 |
| Black, non-Latinx | −0.15 | 0.3 | −0.52 | 0.606 | −0.12 | 0.49 | −0.24 | .814 | −1.36 | 0.59 | −2.28 | .023 |
| Others | −0.4 | 0.18 | −2.22 | 0.028 | −0.40 | 0.29 | −1.42 | .158 | −0.07 | 0.29 | −0.25 | .803 |
| Sex (reference = female) | −0.23 | 0.13 | −1.84 | 0.067 | −0.21 | 0.17 | −1.22 | .222 | −0.37 | 0.15 | −2.42 | .016 |
| Age | 0.07 | 0.04 | 1.76 | 0.079 | 0.14 | 0.06 | 2.29 | .023 | 0.23 | 0.05 | 4.45 | <.001 |
| US born (reference = non-US born) | 0.21 | 0.23 | 0.89 | 0.375 | 0.76 | 0.35 | 2.18 | <.001 | 0.01 | 0.33 | 0.04 | .966 |
| F | Num df | Den df | p | F | Num df | Den df | p | F | Num df | Den df | p | |
| Model statistics and fit | ||||||||||||
| Likelihood ratio | 55.22 | 10.75 | 2,590.76 | <.001 | 23.6 | 10.27 | 2,474.24 | <.001 | 24.58 | 9.83 | 2,368.71 | <.001 |
| Score | 25.14 | 12 | 241 | <.001 | 10.23 | 12 | 241 | <.001 | 12.38 | 12 | 241 | <.001 |
| Wald | 21.2 | 12 | 241 | <.001 | 14.96 | 12 | 241 | <.001 | 11.83 | 12 | 241 | <.001 |
| Received psychosocial or emotional counseling model 1: N = 3,381 . | Used an online tool to connect with a professional model 2: N = 3,374 . | Used an online tool for self-help model 3: N = 3,378 . | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimates . | SE . | t . | p . | Estimates . | SE . | T . | p . | Estimates . | SE . | t . | p . | |
| Parameter estimates | ||||||||||||
| Intercepts | −2.04 | 0.31 | −6.6 | <.001 | −3.88 | 0.4891 | −7.94 | <.001 | −3.71 | 0.52 | −7.18 | <.001 |
| Year (reference = 2019) | ||||||||||||
| 2020 | 0.14 | 0.27 | 0.51 | 0.61 | −0.19 | 0.42 | −0.46 | .649 | 0.72 | 0.42 | 1.69 | .092 |
| 2021 | 0.21 | 0.25 | 0.81 | 0.416 | 0.52 | 0.35 | 1.46 | .145 | 0.7 | 0.42 | 1.65 | .101 |
| Psychological distress | 0.14 | 0.02 | 6.62 | <.001 | 0.1 | 0.03 | 4.02 | <.001 | 0.16 | 0.03 | 4.77 | <.001 |
| Distress × year | ||||||||||||
| 2020 | −0.03 | 0.03 | −1.23 | 0.221 | 0.06 | 0.03 | 1.7 | .09 | −0.04 | 0.04 | −1.06 | .291 |
| 2021 | 0 | 0.03 | −0.11 | 0.91 | 0.01 | 0.03 | 0.26 | .793 | −0.04 | 0.04 | −1 | .318 |
| Race/ethnicity (reference = White, non-Latinx) | ||||||||||||
| Asian, non-Latinx | −1.98 | 0.28 | −7.06 | <.001 | −1.53 | 0.36 | −4.21 | <.001 | −0.24 | 0.28 | −0.87 | .384 |
| Latinx | −1.06 | 0.13 | −8.35 | <.001 | −0.81 | 0.19 | −4.19 | <.001 | −0.36 | 0.16 | −2.28 | .024 |
| Black, non-Latinx | −0.15 | 0.3 | −0.52 | 0.606 | −0.12 | 0.49 | −0.24 | .814 | −1.36 | 0.59 | −2.28 | .023 |
| Others | −0.4 | 0.18 | −2.22 | 0.028 | −0.40 | 0.29 | −1.42 | .158 | −0.07 | 0.29 | −0.25 | .803 |
| Sex (reference = female) | −0.23 | 0.13 | −1.84 | 0.067 | −0.21 | 0.17 | −1.22 | .222 | −0.37 | 0.15 | −2.42 | .016 |
| Age | 0.07 | 0.04 | 1.76 | 0.079 | 0.14 | 0.06 | 2.29 | .023 | 0.23 | 0.05 | 4.45 | <.001 |
| US born (reference = non-US born) | 0.21 | 0.23 | 0.89 | 0.375 | 0.76 | 0.35 | 2.18 | <.001 | 0.01 | 0.33 | 0.04 | .966 |
| F | Num df | Den df | p | F | Num df | Den df | p | F | Num df | Den df | p | |
| Model statistics and fit | ||||||||||||
| Likelihood ratio | 55.22 | 10.75 | 2,590.76 | <.001 | 23.6 | 10.27 | 2,474.24 | <.001 | 24.58 | 9.83 | 2,368.71 | <.001 |
| Score | 25.14 | 12 | 241 | <.001 | 10.23 | 12 | 241 | <.001 | 12.38 | 12 | 241 | <.001 |
| Wald | 21.2 | 12 | 241 | <.001 | 14.96 | 12 | 241 | <.001 | 11.83 | 12 | 241 | <.001 |
Note. df = degree of freedom; SE = standard error.
Association Between Psychological Distress and Use of Mental Health Support
Experiencing higher levels of psychological distress was associated with a higher likelihood of receiving counseling (model 1: OR = 1.15, 95% Confidence interval (CI) [1.11, 1.20], p < .001). Experiencing higher levels of psychological distress was associated with a higher likelihood of using an online tool to connect with a professional (model 2: OR = 1.11, 95% CI [1.04, 1.17], p < .001) and self-help (model 3: OR = 1.17, 95% CI [1.11, 1.25], p < .001). The interaction effect between psychological distress and cohort was not significant in all models (models 1–3), meaning that the relationship between psychological distress and mental health support did not differ by cohort, ps > .05.
Other Sociodemographic Predictor Variables in Models
As shown in model 1, compared to White non-Latinx adolescents, adolescents who identified as Latinx, OR = 0.14, 95% CI [0.08, 0.24], p < .001, Asian non-Latinx, OR = 0.35, 95% CI [0.27, 0.45], p < .001, and other unspecified racial groups, OR = 0.67, 95% CI [0.47, 0.96], p = .028, were less likely to receive counseling. The male–female and age differences were not significant for receiving counseling, ps > .05. The difference between US- and non-US-born participants was not significant for receiving counseling, p > .05.
As shown in model 2, compared to White non-Latinx adolescents, adolescents who identified as either Latinx (OR = 0.44, 95% CI [0.30, 0.65], p < .001) or Asian, non-Latinx (OR = 0.22, 95% CI [0.10, 0.45], p < .001) were less likely to use an online tool to connect with a professional. The male–female difference was not significant for using an online tool to connect with a professional, p > .05. Older adolescents were more likely to use an online tool to connect with a professional, OR = 1.15, 95% CI [1.02, 1.30], p = .023. Compared to adolescents who were not born in the US, adolescents who were born in the US were more likely to use an online tool to connect with a professional, OR = 2.15, 95% CI [1.08, 4.27], p < .001.
As shown in model 3, compared to White non-Latinx adolescents, adolescents who identified as either Latinx (OR = 0.70, 95% CI [0.51, 0.96], p = .024) or Black non-Latinx adolescents (OR = 0.26, 95% CI [0.08, 0.84], p = .023) were less likely to use an online tool for self-help. Male participants were less likely to use an online tool for self-help compared to female participants, OR = 0.69, 95% CI [0.51, 0.93], p = .016. Older adolescents were more likely to use an online tool for self-help, OR = 1.26, 95% CI [1.14, 1.39], p < .01. The difference between US- and non-US-born participants was not significant for receiving counseling, p > .05.
Adolescents’ Reasons for Not Using an Online Tool for Mental Health Support
Adolescents’ reasons for not using an online tool for mental health problems are presented in Table III. Reasons for not using online tools for mental health support varied by respondents’ level of psychological distress, X2 = 642.81, adjusted F(24, 240) = 24.22, p < .001. Among adolescents with high levels of psychological distress, the most common reasons for not using an online tool were a lack of perceived need (i.e., 19.2% “I do not think I need it”), desire for handling the problems themselves (i.e., 16.4% “I wanted to handle problems on my own”) and a lack of perceived utility (i.e., 15.7% “I do not think it would be helpful or work”). Among adolescents with medium levels of psychological distress, the most common reasons included were a lack of perceived need (38.9%), improvement in mental health (12.9% reported “got better/no longer needed it”), and their desire for handling the problems themselves (12.8%). Among adolescents reporting none to low levels of psychological distress, most reported a lack of perceived need (68.5%), and small subgroups reported improvement in mental health (11.4%) and desire for handling the problems themselves (4.6%). Post hoc analyses revealed racial/ethnic and sex differences in adolescents’ reasons for not using online tools for mental health support (Supplementary Appendices 1 and 2). When experiencing medium or high levels of psychological distress, a higher percentage of Latinx (17.3%), Black, non-Latinx (17.0%), and Asian, non-Latinx (15.3%) adolescents reported wanting to handle problems on their own compared to White, non-Latinx adolescents (8.8%). When experiencing medium or high levels of psychological distress, a higher percentage of male adolescents (39.7%) reported that they did not think they needed it than female adolescents (27.9%).
Adolescents’ Reasons for Not Using an Online Tool, Including Mobile Apps or Texting Services for Mental Health
| High distress . | Medium distress . | None to low distress . | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Frequency . | Weighted frequency . | Percent . | Frequency . | Weighted frequency . | Percent . | Frequency . | Weighted frequency . | Percent . | |
| Do not think they needed it | 76 | 85,291 | 19.2 | 441 | 412,163 | 38.9 | 1,218 | 1,114,121 | 68.5 |
| Wanted to handle problem on one’s own | 86 | 72,526 | 16.4 | 130 | 136,245 | 12.8 | 90 | 69,500 | 4.3 |
| Do not think it would be helpful | 76 | 69,672 | 15.7 | 101 | 70,928 | 6.7 | 27 | 25,321 | 1.6 |
| Received traditional/face-to-face services | 35 | 36,106 | 8.1 | 69 | 75,501 | 7.1 | 48 | 34,446 | 2.1 |
| Did not know about apps | 28 | 27,138 | 6.1 | 39 | 44,641 | 0.1 | 61 | 75,301 | 4.6 |
| Got better/no longer needed | 23 | 15,487 | 3.5 | 137 | 137,156 | 12.9 | 192 | 185,792 | 11.4 |
| Do not trust mobile apps | 13 | 13,796 | 3.1 | 25 | 29,349 | 2.8 | 17 | 7,447 | 0.5 |
| Concerns about privacy and data security | 12 | 8,794 | 2.0 | 25 | 26,952 | 2.5 | 12 | 11,370 | 0.7 |
| Cost | 13 | 8,399 | 1.9 | 8 | 5,580 | 0.5 | 9 | 11,461 | 0.5 |
| Do not have time | 9 | 6,176 | 1.4 | 13 | 8,390 | 0.8 | 11 | 10,281 | 0.6 |
| Do not own a device | N/A | N/A | N/A | 6 | 1,113 | 0.1 | N/A | N/A | N/A |
| Do not have enough space to download | 0 | 0 | 0 | 5 | 3,456 | 0.3 | 5 | 6,634 | 0.4 |
| Other | N/A | N/A | N/A | 7 | 7,648 | 0.7 | N/A | N/A | N/A |
| Not applicable | 117 | 97,235 | 21.9 | 126 | 101,323 | 9.6 | 53.00 | 52,549 | 0.6 |
| Total | 491 | 443,313 | 100.0 | 1,132 | 1,132 | 100.0 | 1,758 | 1,625,482 | 100.0 |
| High distress . | Medium distress . | None to low distress . | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Frequency . | Weighted frequency . | Percent . | Frequency . | Weighted frequency . | Percent . | Frequency . | Weighted frequency . | Percent . | |
| Do not think they needed it | 76 | 85,291 | 19.2 | 441 | 412,163 | 38.9 | 1,218 | 1,114,121 | 68.5 |
| Wanted to handle problem on one’s own | 86 | 72,526 | 16.4 | 130 | 136,245 | 12.8 | 90 | 69,500 | 4.3 |
| Do not think it would be helpful | 76 | 69,672 | 15.7 | 101 | 70,928 | 6.7 | 27 | 25,321 | 1.6 |
| Received traditional/face-to-face services | 35 | 36,106 | 8.1 | 69 | 75,501 | 7.1 | 48 | 34,446 | 2.1 |
| Did not know about apps | 28 | 27,138 | 6.1 | 39 | 44,641 | 0.1 | 61 | 75,301 | 4.6 |
| Got better/no longer needed | 23 | 15,487 | 3.5 | 137 | 137,156 | 12.9 | 192 | 185,792 | 11.4 |
| Do not trust mobile apps | 13 | 13,796 | 3.1 | 25 | 29,349 | 2.8 | 17 | 7,447 | 0.5 |
| Concerns about privacy and data security | 12 | 8,794 | 2.0 | 25 | 26,952 | 2.5 | 12 | 11,370 | 0.7 |
| Cost | 13 | 8,399 | 1.9 | 8 | 5,580 | 0.5 | 9 | 11,461 | 0.5 |
| Do not have time | 9 | 6,176 | 1.4 | 13 | 8,390 | 0.8 | 11 | 10,281 | 0.6 |
| Do not own a device | N/A | N/A | N/A | 6 | 1,113 | 0.1 | N/A | N/A | N/A |
| Do not have enough space to download | 0 | 0 | 0 | 5 | 3,456 | 0.3 | 5 | 6,634 | 0.4 |
| Other | N/A | N/A | N/A | 7 | 7,648 | 0.7 | N/A | N/A | N/A |
| Not applicable | 117 | 97,235 | 21.9 | 126 | 101,323 | 9.6 | 53.00 | 52,549 | 0.6 |
| Total | 491 | 443,313 | 100.0 | 1,132 | 1,132 | 100.0 | 1,758 | 1,625,482 | 100.0 |
Note. N/A: cells with very low frequencies are masked by the CHIS DAC team as NA to protect respondent confidentiality. An adjacent cell is also masked along with the low-frequency cell to prevent back-calculation. The complex design of the sample requires proper weighting and variance calculation of the estimates; population weights were applied to calculate weighted frequency.
Adolescents’ Reasons for Not Using an Online Tool, Including Mobile Apps or Texting Services for Mental Health
| High distress . | Medium distress . | None to low distress . | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Frequency . | Weighted frequency . | Percent . | Frequency . | Weighted frequency . | Percent . | Frequency . | Weighted frequency . | Percent . | |
| Do not think they needed it | 76 | 85,291 | 19.2 | 441 | 412,163 | 38.9 | 1,218 | 1,114,121 | 68.5 |
| Wanted to handle problem on one’s own | 86 | 72,526 | 16.4 | 130 | 136,245 | 12.8 | 90 | 69,500 | 4.3 |
| Do not think it would be helpful | 76 | 69,672 | 15.7 | 101 | 70,928 | 6.7 | 27 | 25,321 | 1.6 |
| Received traditional/face-to-face services | 35 | 36,106 | 8.1 | 69 | 75,501 | 7.1 | 48 | 34,446 | 2.1 |
| Did not know about apps | 28 | 27,138 | 6.1 | 39 | 44,641 | 0.1 | 61 | 75,301 | 4.6 |
| Got better/no longer needed | 23 | 15,487 | 3.5 | 137 | 137,156 | 12.9 | 192 | 185,792 | 11.4 |
| Do not trust mobile apps | 13 | 13,796 | 3.1 | 25 | 29,349 | 2.8 | 17 | 7,447 | 0.5 |
| Concerns about privacy and data security | 12 | 8,794 | 2.0 | 25 | 26,952 | 2.5 | 12 | 11,370 | 0.7 |
| Cost | 13 | 8,399 | 1.9 | 8 | 5,580 | 0.5 | 9 | 11,461 | 0.5 |
| Do not have time | 9 | 6,176 | 1.4 | 13 | 8,390 | 0.8 | 11 | 10,281 | 0.6 |
| Do not own a device | N/A | N/A | N/A | 6 | 1,113 | 0.1 | N/A | N/A | N/A |
| Do not have enough space to download | 0 | 0 | 0 | 5 | 3,456 | 0.3 | 5 | 6,634 | 0.4 |
| Other | N/A | N/A | N/A | 7 | 7,648 | 0.7 | N/A | N/A | N/A |
| Not applicable | 117 | 97,235 | 21.9 | 126 | 101,323 | 9.6 | 53.00 | 52,549 | 0.6 |
| Total | 491 | 443,313 | 100.0 | 1,132 | 1,132 | 100.0 | 1,758 | 1,625,482 | 100.0 |
| High distress . | Medium distress . | None to low distress . | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Frequency . | Weighted frequency . | Percent . | Frequency . | Weighted frequency . | Percent . | Frequency . | Weighted frequency . | Percent . | |
| Do not think they needed it | 76 | 85,291 | 19.2 | 441 | 412,163 | 38.9 | 1,218 | 1,114,121 | 68.5 |
| Wanted to handle problem on one’s own | 86 | 72,526 | 16.4 | 130 | 136,245 | 12.8 | 90 | 69,500 | 4.3 |
| Do not think it would be helpful | 76 | 69,672 | 15.7 | 101 | 70,928 | 6.7 | 27 | 25,321 | 1.6 |
| Received traditional/face-to-face services | 35 | 36,106 | 8.1 | 69 | 75,501 | 7.1 | 48 | 34,446 | 2.1 |
| Did not know about apps | 28 | 27,138 | 6.1 | 39 | 44,641 | 0.1 | 61 | 75,301 | 4.6 |
| Got better/no longer needed | 23 | 15,487 | 3.5 | 137 | 137,156 | 12.9 | 192 | 185,792 | 11.4 |
| Do not trust mobile apps | 13 | 13,796 | 3.1 | 25 | 29,349 | 2.8 | 17 | 7,447 | 0.5 |
| Concerns about privacy and data security | 12 | 8,794 | 2.0 | 25 | 26,952 | 2.5 | 12 | 11,370 | 0.7 |
| Cost | 13 | 8,399 | 1.9 | 8 | 5,580 | 0.5 | 9 | 11,461 | 0.5 |
| Do not have time | 9 | 6,176 | 1.4 | 13 | 8,390 | 0.8 | 11 | 10,281 | 0.6 |
| Do not own a device | N/A | N/A | N/A | 6 | 1,113 | 0.1 | N/A | N/A | N/A |
| Do not have enough space to download | 0 | 0 | 0 | 5 | 3,456 | 0.3 | 5 | 6,634 | 0.4 |
| Other | N/A | N/A | N/A | 7 | 7,648 | 0.7 | N/A | N/A | N/A |
| Not applicable | 117 | 97,235 | 21.9 | 126 | 101,323 | 9.6 | 53.00 | 52,549 | 0.6 |
| Total | 491 | 443,313 | 100.0 | 1,132 | 1,132 | 100.0 | 1,758 | 1,625,482 | 100.0 |
Note. N/A: cells with very low frequencies are masked by the CHIS DAC team as NA to protect respondent confidentiality. An adjacent cell is also masked along with the low-frequency cell to prevent back-calculation. The complex design of the sample requires proper weighting and variance calculation of the estimates; population weights were applied to calculate weighted frequency.
Discussion
To our knowledge, this is the first study to examine the rates of naturalistic adoption of counseling services along with rates of adoption of DMHIs to connect with a professional and for self-help among adolescents. Past studies have examined the engagement of DMHIs and the adoption of telehealth services or DMHIs alone (Hawke et al., 2021; Nicholas et al., 2021; Rauschenberg et al., 2021). In our large state representative sample, we explored elevated psychological distress as a correlate and reasons related to low adoption of mental health support among adolescents in real-world settings. Across all distress levels, adolescents’ main reasons for not adopting DMHIs included a lack of perceived need and a desire to handle the problems themselves, which were especially salient among adolescents from racial/ethnic minority backgrounds. These findings were consistent with prior studies reporting on adolescents’ and young adults’ barriers to DMHI adoption (Kodish et al., 2023; Liverpool et al., 2020). Among adolescents with a high level of psychological distress, another common barrier to adolescents’ DMHI adoption was a belief that DMHIs would not be helpful, while among those with low or medium levels of psychological distress, roughly 10% reported that they did not use mental health apps because their symptoms had improved.
Despite the shift from in-person to telehealth service delivery models (Palinkas et al., 2021), rates of adopting mental health support among adolescents are not markedly higher than those reported decades ago in real-world settings (Kataoka et al., 2002; Merikangas et al., 2011), although methodologic differences and geographical variations cannot be ruled out as an explanation. In our analyses, rates of adopting counseling services or DMHIs remained far below the percentage of adolescents reporting elevated psychological distress in our sample (∼50%), although it is worth noting that elevated psychological distress on a screener might not equate to the need for interventions in some cases. Compared to adolescents in the 2019 cohort, the rates of using online tools for mental health problems and to connect with a professional in the 2020 and 2021 cohorts were not higher while also including psychological distress and sociodemographic variables in our models. These cross-sectional observations raise questions about the possible slow pace of progress in increasing access to mental health support for adolescents with mental health needs at the population level. Parameter estimates associated with cohort effects for receiving counseling and using an online tool to connect with a professional for the 2020 versus 2019 comparison were smaller than others and our post hoc sensitivity analyses suggested that we may have not had sufficient power to detect effect sizes of those magnitudes.
Our findings about the association between psychological distress and DMHI adoption were consistent with some past research. Specifically, we found that psychological distress was associated with an increased likelihood of using an online tool to connect with a professional or for self-help; a nationally representative survey revealed that 46% of adolescents and young adults with moderate-to-severe depressive symptoms (which overlapped with some items on K6) reported using DMHIs (Rideout et al., 2021). More recently, researchers also found significant associations between psychological distress and DMHI adoption in cross-sectional data collected during the COVID-19 pandemic (Rauschenberg et al., 2021). Other studies also found psychological distress as a barrier to the adoption of professional services (Aguirre Velasco et al., 2020). We did not find a negative association between levels of psychological distress and use of online tools for mental health problems and to connect with a professional. Possibly, in some cases, aspects of psychological distress measured by K6 (e.g., feeling nervous) were considered situationally appropriate during the COVID-19 pandemic and, thus, the related service adoption was normalized. It is not surprising that higher levels of psychological distress were associated with a higher likelihood of adopting counseling services. This may be because K6 items were mostly symptoms of depression and anxiety.
The presented results highlighted the continued racial/ethnic disparity related to naturalistic service adoption. Specifically, we found that compared to White non-Latinx adolescents, Asian American adolescents and Latinx adolescents were less likely to receive counseling and use online tools to connect with professionals. This may be because Asian and Latinx cultures prioritize family and interpersonal relationships and prefer seeking help from their family and friends rather than from professionals for mental health concerns (Kim & Lee, 2022; Wang et al., 2020). In our analyses, compared to White non-Latinx adolescents, Latinx adolescents and Black non-Latinx adolescents were less likely to use online tools for self-help. It is well-documented in the literature that racially/ethnically minoritized populations are underrepresented in DMHI studies and underserved in the communities (Ramos et al., 2021; Schueller et al., 2019). These consistent results around racial/ethnic disparity should be viewed along with inequitable source allocation, systematic racism, and the digital divide. Notably, Asian American adolescents did not differ significantly in their use of DMHI for self-help compared to their White non-Latinx peers in this sample; possibly, strong preferences for self-help in part due to the “model minority stereotype” increased the likelihood of Asian Americans using DMHIs for self-help (Kim & Lee, 2022). These results indicated the importance of nurturing a culturally competent support system, normalizing help-seeking for mental health (e.g., self-help via DMHIs, professional help), and tailoring services to meet the needs of diverse populations (Friis-Healy et al., 2021; Kodish et al., 2023; Ramos et al., 2021).
Our findings also offer insights into factors that may lead to low naturalistic adoption rates of DMHIs among adolescents, such as adolescents’ perceived lack of need for mental health support and the helpfulness of DMHIs. Although DMHIs are very different from counseling and are often considered less stigmatizing because of the anonymity features, barriers (e.g., mental health literacy, beliefs, and autonomy) to adoption of human support (Aguirre Velasco et al., 2020) may apply to adolescents’ adoption of DMHIs for self-help. It is not surprising that most adolescents experiencing none to low levels of psychological distress did not think they needed mental health support. However, the most endorsed reason by adolescents experiencing elevated psychological distress for not using online tools for self-help was also related to a lack of perceived need. The lack of perceived need for help despite elevated psychological distress may be related to low levels of mental health literacy (Aguirre Velasco et al., 2020; Pretorius et al., 2019); multiple studies have found insufficient levels of mental health literacy, including difficulty with problem recognition and a lack of awareness of available resources, pose barriers to adopting formal and informal mental health support (Aguirre Velasco et al., 2020). Adolescents may deny their mental health service needs and normalize their psychological distress to avoid help-seeking. Perceived need for help with mental health problems is the initial stage of the help-seeking process and guides subsequent decision-making about seeking, selecting, and utilizing services (Zhao et al., 2022). Adolescents’ second and third most endorsed reasons for not trying online tools for mental health support varied by distress level. It is not surprising that the second most endorsed reason among adolescents with low and medium levels of distress was their improvement in their mental health. The second most endorsed reason among adolescents with high psychological distress was a desire to handle problems themselves; this was also mentioned as the third most endorsed reason among adolescents with medium levels of distress. Prior research has reported the desire for autonomy as a barrier to adopting professional services among adolescents (Aguirre Velasco et al., 2020). Rather than perceive DMHIs as aligning with a preference for self-reliance (Pretorius et al., 2019), some adolescents preferred not to use DMHIs due to their need for autonomy and independence. The third most endorsed reason among adolescents with high levels of psychological distress was a lack of perceived helpfulness, which might be an accurate perception as professional support and intensive care are needed when symptoms are severe. Consistent with our results, past studies reported that adolescents were more likely to use a DMHI when they perceived it as credible and useful (Liverpool et al., 2020). Relatedly, adolescents expressed distrust about online health information (Freeman et al., 2020) in part due to the large variability in the quality of commercially available DMHIs (Neary & Schueller, 2018; Wang et al., 2023). To improve trust and usability, implementation strategies focused on addressing innovation-level determinants, such as fit, adaptability, and design features are needed (Graham et al., 2020; Zhao et al., 2023). It may be helpful for DMHI development teams (i.e., developers, design specialists, content experts) to follow human-centered design principles (Stiles-Shields et al., 2022) and partner with adolescents (Stiles-Shields et al., 2023), especially those from historically marginalized communities, in the design of DMHI itself and implementation strategies. Relatedly, Kodish et al. (2023) recommended ensuring the representation of individuals of color across marketing and recruitment materials and codesigning DMHI with youth of color (Kodish et al., 2023).
Several study limitations are worth noting. First, although we have multiple survey time points (i.e., 2019, 2020, 2021), due to the cross-sectional nature of the CHIS data, no conclusions about causal associations between psychological distress and service utilization can be drawn. Second, our findings may not generalize outside California. However, the data were from a large, diverse, and representative sample of adolescents, with the surveys available in multiple languages (UCLA Center for Health Policy Research, 2021, 2022). Third, parental consent was required prior to adolescents’ participation. Although this is a common practice, parental consent is a barrier to engaging marginalized youth, such as adolescents experiencing homelessness and identifying as sexual minority (Cavazos-Rehg et al., 2020). Our sample may be biased to those whose parents were allowed to engage in research. It is worth noting that in California, a minor who is 12 years of age or older may consent to mental health treatment. Thus, adolescents receiving mental health treatment and those whose parents allow them to participate in research may be non-overlapping groups. Fourth, our measures were based on self-report responses from adolescents. Despite past evidence on the clinical validity of self-reported K6 (Prochaska et al., 2012), elevated psychological distress did not equate to a clinical diagnosis or a need for interventions. Additionally, adolescents may not be accurate in reporting their symptoms and service utilization. Fifth, our analyses relied on the assumption that CHIS sampling methods and survey weights sufficiently corrected for non-responses and other types of missing data (Sherr et al., 2022). Adolescents who were most impacted and experienced the highest levels of psychological distress during the early pandemic time may have decided not to participate in interviews and surveys. Sixth, although we found that unconditional rates of using an online tool to connect with a professional in 2021 were significantly higher than in 2019 (Table I), in our regression models that included all predictor variables, we found no significant cohort differences (Table II). Models that used different sets of predictors, however, could reach different conclusions. Seventh, only one survey question was about adolescents’ reasons for not using online tools for mental health support, including mobile apps or texting services; thus, we cannot examine the differences in reasons by service type. Some listed reasons do not apply to texting-based tools, such as the lack of space to download an app and not owning a smartphone.
Despite the study’s limitations, our findings have several clinical and research implications. Thoughtful marketing and dissemination efforts are needed to increase mental health awareness and normalize mental health help-seeking and service adoption. For example, it may be particularly helpful to promote mental health awareness and market and deploy DMHIs in real-world settings where adolescents most frequently receive care such as primary care centers, community clinics, and schools (Mohr et al., 2017). Physicians, teachers, and caregivers should be provided with up-to-date information to recommend and guide the use of science-informed DMHIs. To support providers and caregivers, Psihogios et al. (2020) proposed a three-step decision-making framework (i.e., narrow, explore, and contextualize) for selecting appropriate DMHIs for adolescents presenting to pediatric specialty clinics or primary care (Psihogios et al., 2020). Different resources are publicly available to help providers and consumers narrow the target problem, end user, and potential DMHI options and explore the scientific evidence, usability, and privacy features of a DMHI (Camacho et al., 2022; Neary & Schueller, 2018). For example, One Mind PsyberGuide contains expert ratings of different DMHIs based on three metrics (credibility, transparency, and user experience). The M-Health Index and Navigation Database (MIND) also provides information about apps regarding origin/accessibility, privacy/security, clinical foundation, features and engagement, inputs and outputs, and interoperability (Camacho et al., 2022). Note neither of the app-rating platforms has information about usability and efficacy for different racial/ethnic groups or has dedicated sections for cultural appropriateness of contents; thus, future work is needed to support clinicians in contextualizing and evaluating the fit between a DMHI and their patient’s personal, familial, and cultural values and preferences. Clinical judgment can help guide adolescents’ decisions around adopting counseling services and/or DMHIs, as they may represent different levels of symptom severity or intensities of care needed. However, given the national shortage of providers (Hoffmann et al., 2023), facilitating adoption of DMHIs can be hugely helpful at the population level. Information and support through DMHIs may be the only option when a provider is not available and even DMHIs that had not been clinically validated, demonstrated small yet consistent effects (Linardon et al., 2019).
Conclusion
This study employed population-based data to investigate the respective associations among psychological distress, annual cohort, and naturalistic adoption of mental health support (i.e., professional help, DMHIs) and reported on adolescents’ reasons for low adoption of DMHIs. As we enter the post-pandemic time and millions of adolescents’ mental health service needs remain unmet, it is urgent to increase adoption of clinician-delivered and DMHI-based support. Our findings identified adolescents’ main reasons for low DMHI adoption and suggested a need for increasing mental health awareness and normalizing help- and service-seeking behaviors through marketing and dissemination efforts, so adolescents are empowered to choose the types of tools that can be helpful. Due to the lack of longitudinal data, further research is needed to recommend more specific actionable items in clinical practice and policymaking. Meaningfully crafted dissemination to adolescents about different, evidence-based DMHIs may increase adoption rates of mental health support.
Supplementary Data
Supplementary data can be found at: https://dbpia.nl.go.kr/jpepsy.
Acknowledgments
The information or content and conclusions presented here are those of the authors and should not be construed as the official position or policy of, nor should any endorsements be inferred by, the participating Help@Hand Counties or CalMHSA. The authors wish to thank the members of the California Health Interview Survey (CHIS) team at UCLA, who contributed to the data analyses of this work.
Author Contributions
Xin Zhao (Conceptualization [lead], Formal analysis [supporting], Methodology [lead], Writing – original draft [lead], Writing – review & editing [lead]), Stephen M. Schueller (Conceptualization [supporting], Methodology [supporting], Writing – original draft [supporting], Writing – review & editing [supporting]), Jeongmi Kim (Formal analysis [lead], Writing – review & editing [supporting]), Nicole Ashley Stadnick (Conceptualization [supporting], Writing – review & editing [supporting]), Elizabeth Eikey (Writing – review & editing [supporting]), Margaret Schneider (Conceptualization [supporting], Writing – review & editing [supporting]), Kai Zheng (Writing – review & editing-equal), Dana B. Mukamel (Funding acquisition [supporting], Writing – review & editing [supporting]), and Dara H. Sorkin (Conceptualization [supporting], Funding acquisition [lead], Methodology [supporting], Writing – review & editing [supporting])
Funding
This work was funded by the Help@Hand Project (agreement 417-ITS-UCI-2019), a project overseen by the California Mental Health Service Authority (CalMHSA). This work was also supported by the Institute for Clinical and Translational Sciences (ICTS) under Grant (UL1TR001414).
Conflicts of interest
S. M. Schueller has received consulting payments from Otsuka Pharmaceuticals and Trust (K Health) and is a member of the Headspace Scientific Advisory Board, for which he receives compensation. X. Zhao has received consulting payments from FirstThen Inc for work unrelated to this manuscript. The authors have no further interests to declare.
Data Availability
The publicly available data that support the findings of this study are available at https://healthpolicy.ucla.edu/chis/data/pages/getchisdata.aspx. Access to confidential CHIS data through the CHIS Data Access Center (DAC) requires a research application, renewal, and approval.
