Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2025 Nov 1.
Published in final edited form as: Soc Sci Med. 2024 Sep 26;361:117369. doi: 10.1016/j.socscimed.2024.117369

Assessing access: Texting hotline app provides mental health crisis care for economically deprived youth

Douglas Tharp a,*, Brent M Kious b,c,d, Amanda Bakian b, Simon Brewer a, Scott Langenecker b,e, Mindy Schreiner b,f, Andrey Shabalin b, Hilary Coon b, Robert C Welsh b,g, Richard M Medina a
PMCID: PMC11772012  NIHMSID: NIHMS2046894  PMID: 39369499

Abstract

Background:

Due to rapidly increasing youth suicides in the U.S state of Utah, the legislature funded creation of a 24/7 texting-based smartphone app in Spanish and English targeting Utah’s school aged population. Recent research elsewhere (in the Netherlands) suggests cost inhibits help seeking among the economically disadvantaged. We evaluate the relationship between poverty and app usage during the onset of the COVID-19.

Method:

Local demographics, social determinants of health and COVID-19 infection rates were modeled using a Bayesian spatio-temporal approach examining usage rates.

Results:

When controlling for generally researched suicide crisis covariates, app usage is shown to vary depending on economic status of the population, with the largest relative increases in use among disadvantaged youth.

Discussion:

This bilingual Spanish/English, texting (SMS) based, smart phone app crisis hotline proved effective at providing adolescents from certain populations access to mental health care. The groups discussed are in Census Block Groups (CBGs – neighborhoods) with higher poverty, and/or lower population density (rural areas). The usage of the crisis hotline by these populations increased relative to the overall population as the COVID-19 pandemic unfolded. However, adolescents from areas of higher mobility (our proxy for housing insecure) and those in areas with larger non-White populations had a relative decrease in usage.

Keywords: Suicide prevention, Adolescent mental health, Crisis, Disadvantaged populations, Human geography, Covid-19

1. Introduction

Currently, suicide is the second leading cause of death for those aged 10 to 24 nationally, and the leading cause in Utah. According to the National Center for Health Statistics, the age-adjusted suicide rate in the United States has increased by 33% from 1999 to 2017, with the largest increases among females aged 10–14 years (240% increase) and 15–24 years (93% increase) (Meza and Bath, 2021).

Research has shown that most commonly occurring serious mental health disorders begin in childhood and adolescence (Kessler and Wang, 2008). Additional to the general concern for increasing social stressors and suicidal ideation among youth, the COVID-19 pandemic has been a period of heightened stress and anxiety. Multiple authors have called for research into the effects of the pandemic on suicidality (e.g., Zortea et al., 2021). A concerning signal of increased youth suicide in the UK has been cited as one reason for increased suicide prevention prioritization during the pandemic. Multiple groups have called for increased suicide prevention measures and research on impacts during the pandemic (John et al., 2020). A recent international JAACAP survey of mental health professionals and researchers found that understanding the effects of the COVID-19 pandemic on the mental health of children and adolescents should be a research priority (Novins et al., 2021). That broad survey of pediatric mental health professionals indicated one widely supported strategy for minimizing the impact of the pandemic on children and adolescents is to ensure reliable access to mental health care and social support, especially in underserved communities. It has been suggested that, among other approaches, increasing access to suicide helplines would be helpful in attenuating increases in suicide risk during the pandemic (Gunnell et al., 2020).

Recent research on mental health apps (Parrish et al., 2022) suggests that not all apps are evidenced based in design. We examine the population reach of one specific app designed for crisis. Utah has employed a mobile phone app based, real-time crisis intervention center since 2017 available to adolescents 24/7, in English and Spanish. It is staffed by trained mental health professionals who interact anonymously through the app with those who perceive themselves to be in a mental health crisis and initiate contact (Fig. 1). The mental health professionals provide crisis counseling, supportive listening, and triage to additional medical services. This app is marketed in all public schools across Utah.

Fig. 1.

Fig. 1.

Open in a new tab

Mobile interface.

Residents of Utah have been found to use text-based crisis help centers more frequently than residents of any other state (Larsen et al., 2019). Despite this, the rate of youth suicides in the state has been increasing since well before the COVID-19 pandemic began. While the US rate of youth suicide rose 21% in the first half of the 2010–2020 decade, in Utah the rise was 141% (Utah Board of Education, 2021). In response, the Utah legislature funded prevention programs, including the SafeUT program.

While it is difficult to identify correlates to suicide and suicidal ideation with certainty, some important social factors are identified in the research literature (Durkheim, 1951 and many others). Social fragmentation, or those affected by an inequitably fractured society (e.g., the factors of unemployment, housing insecurity, etc.), and social deprivation, or that which reduces an individual’s connection to society (e.g., poverty, lacking education, and low socioeconomic status) (Evans et al., 2004; Congdon, 2013) have both been associated with increased suicide incidence (Marco et al., 2018). A 4-year longitudinal study in Germany showed a negative non-linear relationship between suicide and income, and a positive and non-linear relationship between suicide and unemployment (Helbich et al., 2007–11). Further, a systematic review of 221 separate analyses dating from 1897 to 2004 found that both increased poverty levels and unemployment levels were directly related to increased suicide risk (Rehkopf and Buka, 2006). While the population studied is generally not yet in the labor force, small area unemployment was included in our analysis. The relationship between unemployment and suicide risk is well established in adults (Stuckler et al., 2009).

As suggested above, there is an established body of literature using Bayesian methods to examine the relationship between socioeconomic factors and actual suicide (Congdon, 2013; Hsu et al., 2015). There are also examples of a Bayesian probabilistic approach in example neighborhood characteristics driving traditional telephone crisis call center volume (Marco et al., 2018). We are extending these approaches to usage of a texting (SMS) based system while examining social deprivation among a specific population, adolescents in Utah.

A general concern, as always with social resources, is underserved population access. In the United States, access to mental health care is rationed primarily through either employment related health insurance or by a person’s ability to pay. Financial barriers have been shown to reduce mental health care help seeking for adolescents in the Netherlands context (Lopes et al., 2022). While suicide risk in the U.S. is generally greatest for Non-Hispanic Whites and American-Indian/Alaska Natives (CDC, 2021), research has clearly shown racial and ethnic disparities in mental health care (McGuire and Miranda, 2008; Meza and Bath, 2021; Nelson, 2003). Furthermore, groups other than Whites have been disproportionately impacted by the pandemic in both disease and psychological harms (Marroquína et al., 2020; Webb Hooper et al., 2020a; Webb Hooper et al., 2020b; Ibrahimi et al., 2020).

Additional factors impacting suicide and suicidal ideation include: (1) social isolation, which has been hypothesized to increase suicidal ideation (Marroquína et al., 2020). Research suggests that college students experienced increased anxiety with distanced learning and social isolation during the pandemic (Fruehwirth et al., 2021). (2) Rurality, which has also been found to be positively associated with suicide risk (Congdon, 2013; McCarthy et al., 2012). (3) Connections between seasonality and suicidality in spatiotemporal research has been illustrated (Marco et al., 2017). (4) Housing insecurity, which research has shown to be related to increased levels of depression (Gilman et al., 2003; Kushel et al., 2006; Meltzer and Schwartz, 2016). (5) Availability of traditional mental health care workers also matters (Hoffmann et al., 2023).

Our research objective was to analyze spatiotemporal variations in text-message contacts to the app across socioeconomic groups and identify how the COVID-19 pandemic impacted the usage by populations presumed to be disadvantaged. Specifically, we focused on adolescents living in poverty and examined how poverty impacted SafeUT usage and how rising infection rates changed that usage. Due to the spatio-temporal variation in both SafeUT usage and the COVID-19 pandemic, we use a Bayesian statistical approach to construct a model based on variable relationships as outlined above. The research question considered is whether this anonymous SMS (texting) based telehealth smart phone app provided mental health care access to Utah’s disadvantaged adolescents during the COVID-19 period of heightened stress. The presumption of disadvantage in the United States’ means based health care system rests on neighborhood (CBG) poverty levels.

2. Data

We analyze two years of data between 2019 and 2020. This time period corresponds to the beginning of the COVID-19 outbreak in the United States. During these two years, there were approximately 51,000 unique users who texted, with half contacting the center more than three times. Thirty percent of the 105,000 unique encounters were labeled “crisis” by the crisis workers. Approximately half of all encounters came from users who have location services (latitude and longitude) enabled on their smartphone app. This fraction began to decrease substantially in December 2020 due to default smart phone setting changes, leading to an apparent decrease in crisis contacts. In November and December of 2020, the iOS (Apple iphones) and Android operating systems introduced the system changes that prompted users to select location enabling status (Hicks, 2021; Coombes and Hicks, 2021, Android Developers, n.d.). However, when we separately accounted for the decline in location enabled proportion, we saw no decrease in the fraction of contacts that were crisis. The location-enabled data were treated as a proxy for all encounters and for the prevalence within the target population. Ultimately, our research included approximately 18,500 crisis contacts over the two-year period.

The encounter labeling process is conducted by the medical health professionals/counselors who are social workers with at least a master’s degree. These professionals categorize each encounter into at least one of approximately 150 “disposition” codes. These codes were developed based on counselors’ clinical judgments about the causes and outcomes of encounters witnessed over the first 3.5 years of the program’s history. For analysis, we combined multiple disposition codes, including “Suicidal Ideation”, “Mobile Crisis Outreach Team Dispatched”, several versions of “Active Rescue”, “Self-Harm”, “Suicidal Thoughts”, and similar categories into “Crisis”.

Crisis contacts are therefore the count of geo-referenced contacts (“encounters”) labeled “Crisis” that include adolescents aged 10 to 19. For geo-referencing, each encounter was assigned by latitude and longitude to one of Utah’s 1690 2010 Census Block Groups (CBGs). We collapsed encounters’ multiple text exchanges, sometimes spanning days, into one record per encounter.

Socioeconomic data used are sourced from the 2017 five-year US Census, American Community Survey (ACS) estimates. These variables include measures of education, mobility, race, unemployment and poverty rate. The 2020 count of licensed mental health providers by county was obtained from the Robert Wood Johnson Foundation (RWJ). “Mental health providers are defined as psychiatrists, psychologists, licensed clinical social workers, counselors, marriage and family therapists, mental health providers that treat alcohol and other drug abuse, and advanced practice nurses specializing in mental health care.” (Robert Wood Johnson Foundation, 2021). One county has no licensed mental health providers, so we substituted 3 for the Z score (RWJ gives counties with fewer mental health providers higher Z scores).

Utah is a geographically large state with dispersed population centers. Therefore, we used county level COVID-19 infection rates as a covariate in modeling the change in crisis contacts. Statewide counts were presumed to be irrelevant in remote, smaller population areas. Some of Utah’s counties have relatively small population counts and therefore have daily counts that are too small to report for data privacy reasons. Utah Department of Health and Human Services marker for infections between zero and five was converted to the midpoint, three. As the temporal scale of this analysis is months, we used the maximum infections reported for each month.

All socioeconomic data were collected at the CBG level, except for the counts of mental health providers and COVID-19 infections, as those two variables are not available and are not necessary for this analysis. The measures of infection rates and mental health care providers are only captured at the county level and not more granularly. Therefore, we applied the county level metric to every census block group within the county. We assume that the spatial variation in infection and health care provision metrics within a county is immaterial. County level measures should be adequate in the context of heightened anxiety (rising infections) and mental health care provision. For these two variables we assigned the county level measure to each CBG in a nested design. CBG is the most granular level for which the US Census Bureau’s demographic variables are reported.

To understand the extent to which SafeUT serves the population, especially the disadvantaged (populations presumed to have less access to mental health services), we examined the interaction between poverty and infection rates. We created an interaction term of poverty (CBG’s percent of households below the poverty level) crossed with infection rates to account for differential impacts based on economic disadvantage while also assembling the controls for measures of social fragmentation (unemployment, mobility), educational attainment (at or above 4-year college degree), racial/ethnic composition (proportion of Whites), rurality (population density), seasonality (binary summer indicator) and traditional mental health services (per capita availability of mental health providers).

3. Methods

As usage of SafeUT changed during the COVID-19 pandemic with variation across geographies, an analysis that considers the changes across space and time is warranted. Thus, we selected a Bayesian modeling approach with spatio-temporal random effects. Following Moraga (2020), we approached this as a small area, relative rate model by examining the change in relative rates of crisis encounters at the CBG level from the beginning of 2019 to the end of 2020. Specifically, we explore whether these encounters increased during the COVID-19 era (see Fig. 2), and how poverty level affected this increase by creating an interaction term between infection rates and poverty.

Fig. 2.

Fig. 2.

Open in a new tab

Crisis and total contacts by month relative to the 24 Month study period.

Small area models are increasingly applied to examine the rate of uncommon, or rare conditions in space and/or time relative to an underlying population. Our approach relies on spatial-temporal dependency to “borrow” observations across neighboring geographical units and time, shrinking any extreme values that result from low population size or rare event occurrence (Gelfand et al., 2010; Lawson, 2009). The sums of spatial and temporal structures are assumed to have similar risk. (Gelfand et al., 2010; Lawson, 2009). Spatial and temporal autocorrelation and overdispersion are addressed with this approach (Khana et al., 2018). The model was fit using an Integrated Nested Laplace Approximation (INLA) (Rue et al., 2017), deployed in R (R Core Team, 2021 Version 4.0.5, RStudio Version 1.3.1093). Analogous to traditional frequentist statistical significance metrics, this Bayesian approach yields credibility intervals indicating whether an association is “credible” (Mills and Strawn, 2020). Priors for penalizing model complexity were left at default (Simpson et al., 2017). The sum of contacts per CBG location per month is modeled as count data, assumed to be conditionally independently Poisson distributed. In our model, the expected count is estimated as a statewide per capita standard rate multiplied by the size of the 10 to 19-year-old population at that location. The standard rate is based on the statewide total of SafeUT counts divided by the statewide total 10 to 19-year old population. For four CBGs, we substituted one for the value of zero for the 10 to 19-year-old age population to enable the complete neighborhood structure required for ‘borrowing’ observations across CBGs. Finally, the relative rates are estimated using the full set of covariates:

Equation=f(education+mobility+Non−White%+MentalHealthproviders+unemployment+virtualschooldistrict+populationdensity+summermonths+COVID−19Infectioncount+povertyrate+COVID−19infectioncountpovertyinteraction)+ε

All covariates were transformed to z-scores and are interpreted here as the percentage change in usage for a one standard deviation change in the underlying variable (while holding constant all other measures in this multivariate model). Any finding broadly straddling 1 is considered not credible. Any credible coefficient below 1 is negatively associated with the crisis count.

When calculated for rare diseases or on small populations, the resulting measures may be less reliable. This bias can be reduced by incorporating spatial-temporal dependencies, allowing information to be borrowed from neighboring areas and time periods (Aregay et al., 2017; Moraga, 2020). To account for spatio-temporal dependencies, the model error is decomposed into three terms:

ϵij=ui+vj+wij

Where ui is a spatial random effect, modeled using as a conditional autoregressive distribution a Besag-York-Mollié model (Besag et al., 1991); vj is a temporal random effect modeled as an AR1 process; and wij is an unstructured noise term.

4. Results

Posterior estimates of model coefficients are shown in Table 1 (credible results shaded). For each coefficient, we used the posterior marginal distribution to calculate the probability of that coefficient exceeding 1 (i.e., showing a positive relationship). Values above 0.95 or below 0.05 are taken as showing credible evidence for a positive or negative relationship, respectively. Descriptive statistics of the independent variables included in our model are summarized in the Appendix.

Table 1.

Exponentiated Independent Variable Coefficients, Credibility Intervals and Exceedance probability.

Crisis Encounter Count
mean (0.025, 0.975) P > 1
Ed Attainment 0.951 (0.866, 1.04) 0.146
Mobility 0.922 (0.865, 0.982) 0.005
Non-White 0.839 (0.770, 0.913) 0.000
MH Provider 0.908 (0.780, 1.058) 0.107
Unemployment 0.966 (0.909, 1.027) 0.136
SLC School District 1.072 (0.699, 1.644) 0.624
Population Density 0.784 (0.718, 0.854) 0.000
Summer 0.620 (0.494, 0.778) 0.000
Max Daily Infections 1.041 (1.005, 1.079) 1.000
HH Poverty 1.040 (0.962, 1.122) 0.839
Infection:HHPoverty 1.031 (1.004, 1.067) 0.989
Open in a new tab

4.1. Factors associated with increased relative usage of SafeUT for crisis encounters

Our variable of interest, usage of the SafeUT app by disadvantaged adolescents, has a nuanced relationship with the onset of the COVID-19 pandemic. As with interaction terms generally, we report the impact of both the base terms (poverty and infection rates) as well as their interaction. We found a credibly positive relationship for crisis encounters with both the base infection rate over time (zRR = 1.04, CI = 1.01–1.08, p < 0.001), and the interaction of the infection rate with poverty rates (zRR = 1.03, CI = 1.00–1.07, p = 0.010). The infection rate is positively related to crisis contacts while the base poverty rate itself is non-credible, indicating that the poverty relationship with encounter count is not meaningful outside of the rising infection rate period. Expressed differently, those living in CBGs with higher poverty rates showed a relative increase in crisis contacts as infection counts rose over time (Fig. 3).

Fig. 3.

Fig. 3.

Open in a new tab

Crisis contact usage relative rate by the interaction of poverty and infection rates.

Our (inverse) proxy for rurality, population density, is credibly negative (zRR = 0.784, credible interval (CI) = 0.72–0.85, p < 0.001), meaning students in more rural CBGs showed relatively higher rates of crisis contacts.

4.2. Factors associated with lower relative usage of SafeUT for crisis encounters

We found that the social fragmentation variable, housing mobility (a proxy for increased housing insecurity) (zRR = 0.92, CI = 0.87–0.98, p = 0.005), the Non-White race variable (zRR = 0.84, CI = 0.77–0.91, p < 0.001) and the out-of-school period variable (zRR = 0.620, CI = 0.49–0.78, p < 0.001) were credibly negatively related to crisis encounter counts. Crisis contacts fell markedly in both summers of the study period, as they have during the other summers of the program’s existence.

4.3. Factors associated with no change in relative usage of SafeUT for crisis encounters

That unemployment is not credibly related to crisis contacts can plausibly be attributed to Utah’s second lowest unemployment rate in the U.S. during this period. The one school district (of 44) that remained on-line throughout 2020 had, contrary to expectation, no credible difference in crisis contacts. The composite educational attainment social deprivation metric, the standardized count of local (same county) mental health providers, the social fragmentation metric unemployment, are all not credibly related to crisis contacts.

5. Discussion

Increasing COVID-19 infection rates in Utah during 2020 were associated with an increase in crisis contacts. Fear of contagion has been identified as a driver in help line contacts in other settings, e.g., in Taiwan (Hwang et al., 2022). Evidence for whether the SafeUT resource was used by the less advantaged at greater rates is positive. For those facing the deprivation of poverty the results suggest increased usage. This implies that the free-to-users SafeUT service is successful at providing support during the onset of the pandemic to the more socioeconomically vulnerable populations that are presumed to have less access to mental health resources than students living in CBGs with lower poverty rates.

Our controls show mixed relationships to usage. SafeUT appears to have been successful in providing services to those in rural settings where there are fewer mental health care resources available. The relationship between lower usage and housing insecurity is possibly an indicator of lack of access or resources. Our results suggest that parental education may not be related to youth help seeking in the context of the growing COVID-19 pandemic.

The negative association between relative SafeUT use and race/ethnicity other than Non-Hispanic White can be interpreted in several ways. This result potentially supports the idea of a reduction in the thwarted belongingness from Interpersonal Suicide Theory (Van Orden et al., 2010). Do adolescents of color feel increased belonginess when at home? Possibly those living in more racially/ethnically diverse areas are less likely to be culturally predisposed to ask for help (Ward et al., 2013). Or, possibly, help seeking is associated with increased stigma, or there is less trust in institutional or government-provided health care resources, or perhaps there is simply less awareness of SafeUT, which might suggest a need in marketing and outreach efforts. It could also indicate that disadvantaged populations are less likely to have the privacy or technological resources required to use the app. Alternatively, it could signify greater cultural (or local) emphasis on self-reliance, more reliance on faith-based care, or a greater ability to rely on family or community networks. Finding the explanations and addressing these barriers clearly warrants additional research.

The non-credibility of the always-virtual school district variable is understood as indicating that in-person versus virtual classroom status was irrelevant to students’ decisions to seek crisis mental health support via SafeUT. If SafeUT usage roughly tracks distress levels in the community, then these results suggest the transition to virtual schooling did not significantly increase or decrease distress in the district.

6. Limitations

The SafeUT smartphone app is specifically designed to provide anonymity. For that reason, our study conclusions reflect the demographics of the location from which texts to the crisis center originated, rather than individual characteristics. While any individual case may vary from its group (the ecological fallacy), general patterns emerging from the tens of thousands of texts are the basis of our analysis. The app is distributed (promoted for download) in all of Utah’s schools targeting pupils aged 10 to 19. Thus, while we cannot differentiate responses by age, we assume they overwhelmingly come from school-age adolescents in this age range. We infer “average” race and ethnicity from location. While gender issues do appear in the content of some texts, we have not attempted to identify any patterning based on gender.

We further assume that students have smartphones available for downloading and using the SafeUT app, but were unable to control for rates of youth cellphone ownership. As SafeUT does not require a personal computer, no requirement for internet access beyond cellular connectivity was considered.

7. Conclusion

The SafeUT marketing team has since targeted those populations studied that saw relative falls in usage – students who live in areas with relatively higher rates of Non-White populations or those who are housing insecure as well as targeting geographies of lower usage.

A universal form of social deprivation, poverty, that cuts across other social and race categories is assumed to be associated with those who have relatively limited access to mental health care services. This work demonstrates that the gap in coverage is at least somewhat addressed via our smart phone-based texting initiative when promoted through public schools. We hope that this effort can serve as a model for other states.

Appendix

Descriptive Statistics

min mean median max sd
All Encounters 0 0.23 0 15 0.58
Crisis Encounters 0 0.1 0 6 0.34
Educational Attainment 0.02 0.33 0.3 0.98 0.17
Mobility/Housing Insecurity 0 0.56 0.08 23.86 1.56
Non-White 0 0.19 0.14 1 0.17
MH Provider Std Score −2.98 −0.4 −0.04 5.98 0.75
Unemployment 0 0.04 0.04 0.26 0.04
SLC (Virtual) School District 0 0.08 0 1 0.28
Rurality (Population Density) 0.03 1630.7 1458.5 16050 1539.2
Monthly Max Daily Infections 0 0.24 0 4.89 0.49
HH Poverty 0 0.11 0.08 1 0.11
CBG Total Population 110 1774.7 1538 18752 1151.4
CBG Population Aged 10 to 19 0 287.38 229 2814 240.39
Open in a new tab

Footnotes

No ethics issues to report

No ethics issues to report.

CRediT authorship contribution statement

Douglas Tharp: Writing – review & editing, Writing – original draft, Methodology, Formal analysis, Data curation, Conceptualization. Brent M. Kious: Writing – review & editing, Supervision, Investigation. Amanda Bakian: Writing – review & editing, Project administration. Simon Brewer: Writing – review & editing, Visualization, Methodology, Formal analysis, Conceptualization. Scott Langenecker: Writing – review & editing, Resources, Project administration, Funding acquisition. Mindy Schreiner: Investigation. Andrey Shabalin: Investigation. Hilary Coon: Writing – review & editing, Supervision. Robert C. Welsh: Writing – review & editing, Investigation. Richard M. Medina: Writing – review & editing, Methodology, Data curation.

Data availability

The data that has been used is confidential.

References

  1. Android Developers. (n.d.). Privacy changes in Android 11: Location. Android Developers. Retrieved June 22, 2024, from https://developer.android.com/about/versions/11/privacy/location. [Google Scholar]
  2. Aregay M, Lawson AB, Faes C, Kirby RS, 2017. Bayesian multi-scale modeling for aggregated disease mapping data. Stat. Methods Med. Res 26 (6), 2726–2742. 10.1177/0962280215607546. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Besag J, York J, Mollié A, 1991. Bayesian image restoration with two applications in spatial statistics (with discussion). Annals of the Institute of Statistical Mathematics 43 (1), 1–59. 10.1007/BF00116466. [DOI] [Google Scholar]
  4. CDC, 2021. Retrieved Fri. https://www.cdc.gov/nchs/data/hestat/suicide/rates_1999_2017.pdf.
  5. Congdon P, 2013. Assessing the impact of socioeconomic variables on small area variations in suicide outcomes in England. Int J Environ Res Public Health 10 (1), 158–177. 10.3390/ijerph10010158. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Coombes L, Hicks M, 2021. iOS 14 tips and tricks: how to make the most of Apple’s latest iPhone software. TechRadar. Retrieved June 22, 2024, from. https://www.techradar.com/how-to/ios-14-tips-and-tricks. [Google Scholar]
  7. Durkheim E, 1951. In: Spaulding JA, Simpson G (Eds.), Suicide, a Study in Sociology. Routledge, London: (Original work published; 1897). [Google Scholar]
  8. Evans J, Middleton N, Gunnell D, 2004. Social fragmentation, severe mental illness and suicide. Soc Psychiatry Psychiatr Epidemiol 39 (3), 165–170. 10.1007/s00127-004-0733-9. [DOI] [PubMed] [Google Scholar]
  9. Fruehwirth JC, Biswas S, Perreira KM, 2021. The Covid-19 pandemic and mental health of first-year college students: examining the effect of Covid-19 stressors using longitudinal data. PLoS One 16 (3), e0247999. 10.1371/journal.pone.0247999. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Gelfand AE, Diggle PJ, Guttorp P, Fuentes M, 2010. Handbook of Spatial Statistics. Chapman & Hall/CRC, Boca Raton, Florida. ISBN 978–1420072877. [Google Scholar]
  11. Gilman S, Kawachi I, Fitzmaurice G, Buka S, 2003. Socio-economic status, family disruption and residential stability in childhood: relation to onset, recurrence and remission of major depression. Psychol. Med 33 (8), 1341–1355. 10.1017/S0033291703008377. [DOI] [PubMed] [Google Scholar]
  12. Gunnell D, Appleby L, Arensman E, Hawton K, John A, Kapur N, Khan M, O’Connor RC, Pirkis J, Caine ED, Chan LF, Chang S. Sen, Chen YY, Christensen H, Dandona R, Eddleston M, Erlangsen A, Harkavy-Friedman J, Kirtley OJ, et al. , 2020. Suicide risk and prevention during the COVID-19 pandemic. Lancet Psychiatr 7 (Issue 6), 468–471. 10.1016/S2215-0366(20)30171-1. Elsevier Ltd. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Helbich M, Plener PL, Hartung S, Blüml V, 2007–11. Spatiotemporal suicide risk in Germany: a longitudinal study. Sci. Rep 7 (1), 1–9. 10.1038/s41598-017-08117-4, 2017. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Hicks M, 2021. How to Turn iPhone Location Services on or off. TechRadar. Retrieved June 22, 2024, from. https://www.techradar.com/how-to/how-to-turn-iphone-location-services-on-or-off. [Google Scholar]
  15. Hoffmann JA, Attridge MM, Carroll MS, Simon NE, Beck AF, Alpern ER, 2023. Association of Youth Suicides and County-Level Mental Health Professional Shortage Areas in the US. JAMA Pediatr 177 (1), 71–80. 10.1001/jamapediatrics.2022.4419. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Hsu CY, Chang SS, Lee ES, Yip PS, 2015. Geography of suicide in Hong Kong: spatial patterning, and socioeconomic correlates and inequalities. Social Science & Medicine 130, 190–203. [DOI] [PubMed] [Google Scholar]
  17. Hwang I-T, Fu-Tsung Shaw F, Hsu W-Y, Liu G-Y, Kuan C-I, Gunnell D, Chang S-S, 2022. ‘I can’t see an end in sight’ how the COVID-19 pandemic may influence suicide risk: a qualitative study. Crisis J. Crisis Interv. Suicide Prev 10.1027/0227-5910/a000877. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Ibrahimi S, Yusuf KK, Dongarwar D, Maiyegun SO, Ikedionwu C, Salihu HM, 2020. COVID-19 devastation of african American families: impact on mental health and the consequence of systemic racism. International journal of MCH and AIDS 9 (3), 390–393. 10.21106/ijma.408. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. John A, Pirkis J, Gunnell D, Appleby L, Morrissey J, 2020. Trends in suicide during the covid-19 pandemic. BMJ 371. 10.1136/bmj.m4352. [DOI] [PubMed] [Google Scholar]
  20. Kessler RC, Wang PS, 2008. The descriptive epidemiology of commonly occurring mental disorders in the United States. Annu. Rev. Publ. Health 29, 115–129. 10.1146/annurev.publhealth.29.020907.090847. [DOI] [PubMed] [Google Scholar]
  21. Khana D, Rossen LM, Hedegaard H, Warner M, 2018. A bayesian spatial and temporal modeling approach to mapping geographic variation in mortality rates for subnational areas with R-inla. J. Data Sci 16 (1), 147–182. 10.6339/JDS.201801_16(1).0009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Kushel MB, Gupta R, Gee L, Haas JS, 2006. Housing instability and food insecurity as barriers to health care among low-income Americans. J. Gen. Intern. Med 21 (1), 71–77, 0.1111/j.1525-1497.2005.00278.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Larsen ME, Torok M, Huckvale K, Reda B, Berrouiguet S, Christensen H, 2019. Geospatial suicide clusters and emergency responses: an analysis of text messages to a crisis service. Proc Annu Int Conf IEEE Eng [DOI] [PubMed] [Google Scholar]
  24. Lawson AB, 2009. Bayesian Diesease Mapping: Hierarchical Modeling in Spatial Epidemiology. Chapman & Hall/CRC, Boca Raton, Florida, pp. 6109–6112. 10.1109/EMBC.2019.8856909. ISBN 978–1584888406 Med Biol Soc EMBS. 2019. [DOI] [Google Scholar]
  25. Lopes FV, Riumallo Herl CJ, Mackenbach JP, Van Ourti T, 2022. Patient cost-sharing, mental health care and inequalities: a population-based natural experiment at the transition to adulthood. Soc. Sci. Med 296. 10.1016/j.socscimed.2022.114741. [DOI] [PubMed] [Google Scholar]
  26. Marco M, López-Quílez A, Conesa D, Gracia E, Lila M, 2017. Spatio-temporal analysis of suicide-related emergency calls. Int. J. Environ. Res. Publ. Health 14 (7). 10.3390/ijerph14070735. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Marco M, Gracia E, López-Quílez A, Lila M, 2018. What calls for service tell us about suicide: a 7-year spatio-temporal analysis of neighborhood correlates of suicide-related calls. Sci. Rep 8 (1). 10.1038/s41598-018-25268-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Marroquína B, Vine V, Morgan R, 2020. Mental health during the COVID-19 pandemic: effects of stay-at-home policies, social distancing behavior, and social resources. Psychiatr. Res 293. 10.1016/j.psychres.2020.113419., 113419. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. McCarthy JF, Blow FC, Ignacio RV, Ilgen MA, Austin KL, Valenstein M, 2012. Suicide among patients in the Veterans Affairs health system: rural-urban differences in rates, risks, and methods. Am J Public Health 102 (Suppl. 1). 10.2105/AJPH.2011.300463. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. McGuire TG, Miranda J, 2008. New evidence regarding racial and ethnic disparities in mental health: policy implications. Health Aff 27 (2), 393–403. 10.1377/hlthaff.27.2.393. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Meltzer R, Schwartz A, 2016. Housing affordability and health: evidence from New York city. Hous Policy Debate 26 (1), 80–104. 10.1080/10511482.2015.1020321. [DOI] [Google Scholar]
  32. Meza JI, Bath E, 2021. One size does not fit all: making suicide prevention and interventions equitable for our increasingly diverse communities. Journal of the American Academy of Child and Adolescent Psychiatry 60 (2), 209–212. 10.1016/j.jaac.2020.09.019. Elsevier Inc. [DOI] [PubMed] [Google Scholar]
  33. Mills JA, Strawn JR, 2020. Antidepressant tolerability in pediatric anxiety and obsessive-compulsive disorders: a bayesian hierarchical modeling meta-analysis. Journal of the American Academy of Child and Adolescent Psychiatry 59 (11), 1240–1251. 10.1016/j.jaac.2019.10.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Moraga P, 2020. Geospatial Health Data: Modeling and Visualization with R-INLA and Shiny. Chapman & Hall/CRC Biostatistics Series. [Google Scholar]
  35. Nelson AR, 2003. Unequal Treatment: Report of the Institute of Medicine on Racial and Ethnic Disparities in Healthcare Ann Thorac Surg 76:S1377–81. [DOI] [PubMed] [Google Scholar]
  36. Novins DK, Stoddard J, Althoff RR, Charach A, Cortese S, Cullen KR, Frazier JA, Glatt SJ, Henderson SW, Herringa RJ, Hulvershorn L, Kieling C, McBride AB, McCauley E, Middeldorp CM, Reiersen AM, Rockhill CM, Sagot AJ, Scahill L, et al. , 2021. Note and special communication: research priorities in child and adolescent mental health emerging from the COVID-19 pandemic. Journal of the American Academy of Child and Adolescent Psychiatry 60 (Issue 5), 544–554.e8. 10.1016/j.jaac.2021.03.005. Elsevier Inc. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Parrish EM, Filip TF, Torous J, Nebeker C, Moore RC, Depp CA, 2022. Are mental health apps adequately equipped to handle users in crisis? Crisis 43 (4), 289–298. 10.1027/0227-5910/a000785. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. R Core Team, 2021. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. URL. https://www.R-project.org/doi:10.1017/S003329170500588X. [Google Scholar]
  39. Rehkopf DH, Buka SL, 2006. The association between suicide and the socio-economic characteristics of geographical areas: a systematic review. Psychol. Med 36 (2), 145–157. 10.1017/S003329170500588X. [DOI] [PubMed] [Google Scholar]
  40. Robert Wood Johnson Foundation, 2021. County health rankings. Clinical care, access to care, mental health providers. https://www.countyhealthrankings.org/explore-health-rankings/measures-data-sources/county-health-rankings-model/health-factors/clinical-care/access-to-care/mental-health-providers. [Google Scholar]
  41. Rue H, Riebler A, Sørbye SH, Illian JB, Simpson DP, Lindgren FK, 2017. Bayesian computing with INLA: a review. Annual Reviews of Statistics and Its Applications 4 (March), 395–421. 10.1111/j.1467-9868.2008.00700.x. [DOI] [Google Scholar]
  42. Simpson D, Rue H, Riebler A, Martins TG, Sørbye SH, 2017. Penalising model component complexity: a principled, practical approach to constructing priors. Stat. Sci 32 (1), 1–28. 10.1214/16-STS576. [DOI] [Google Scholar]
  43. Stuckler D, Basu S, Suhrcke M, Coutts A, McKee M, 2009. The public health effect of economic crises and alternative policy responses in Europe: an empirical analysis. Lancet 374 (9686), 315–323. 10.1016/S0140-6736(09)61124-7. [DOI] [PubMed] [Google Scholar]
  44. Utah Board of Education, available online at: https://www.schools.utah.gov/prevention/suicide.
  45. Utah Department of Health, Retrieved Wed, 22 December 2021 from the Utah Department of Health, Indicator-Based Information System for Public Health Web site: http://ibis.health.utah.gov.
  46. Van Orden KA, Witte TK, Cukrowicz KC, Braithwaite SR, Selby EA, Joiner TE Jr., 2010. The interpersonal theory of suicide. Psychol. Rev 117 (2), 575–600. 10.1037/a0018697. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Ward EC, Wiltshire JC, Detry MA, Brown RL, 2013. African American men and women’s attitude toward mental illness, perceptions of stigma, and preferred coping behaviors. Nurs. Res 62 (3), 185–194. 10.1097/NNR.0b013e31827bf533[. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Webb Hooper M, Nápoles AM, Pérez-Stable EJ, 2020a. COVID-19 and racial/ethnic disparities. JAMA 323, 2466–2467. 10.1001/jama.2020.8598. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Webb Hooper M, Nápoles AM, Pérez-Stable EJ, 2020b. COVID-19 and racial/ethnic disparities. JAMA 323, 2466–2467. 10.1001/jama.2020.8598. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Zortea TC, Brenna CTA, Joyce M, McClelland H, Tippett M, Tran MM, Arensman E, Corcoran P, Hatcher S, Heisel MJ, Links P, O’Connor RC, Edgar NE, Cha Y, Guaiana G, Williamson E, Sinyor M, Platt S, 2021. The impact of infectious disease-related public health emergencies on suicide, suicidal behavior, and suicidal thoughts: a systematic review. Crisis J. Crisis Interv. Suicide Prev 42 (6), 474–487. 10.1027/0227-5910/a000753. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

The data that has been used is confidential.

RESOURCES