Skip to main content
Full access
Articles
Published Online: 4 December 2019

Engagement With Crisis Text Line Among Subgroups of Users Who Reported Suicidality

Abstract

Objective:

Despite the interest in technology-enhanced preventive interventions for suicidality, there is minimal peer-reviewed research on conversations of text message hotlines. In this large-scale study, the authors explored distinct classes of users of the Crisis Text Line who reported suicidality. Classes were based on texters’ presenting psychosocial issues and were explored across frequency of hotline use and conversation number.

Methods:

Data included 153,514 conversations from 122,909 individuals collected by the Crisis Text Line (2013–2017). Analyses were restricted to conversations from users who mentioned current or previous suicidality and excluded texters who discussed a third party. The authors used latent class analysis to identify distinct classes of texters on the basis of crisis counselor–assigned issue tags and across subgroups reflecting whether the conversation was the first or last contact with the hotline and frequency of use (one-time, two-time, or ≥3-time texters).

Results:

Three classes emerged in all subsamples. The largest class, “lower distress,” had the lowest prevalence of all issues. The second largest class, “anxious distress,” had the highest prevalence of anxiety or stress and elevated depression. The smallest class, “relational distress,” had the highest prevalence of depression and self-harm and higher probability of endorsing relational indicators.

Conclusions:

Psychological and relational issues mostly distinguished the three classes. Despite differing frequency of hotline usage, most texters who reported suicidal ideation and behaviors endorsed similar issues, and these issues did not seem to vary across conversations. Yet there appeared to be distinct subgroups of texters with different presenting issues, which may inform how crisis counselors tailor strategies for both low- and high-volume texters.

HIGHLIGHTS

This is one of few studies to explore distinct classes among hotline users reporting suicidality on the basis of their presenting psychosocial issues and across frequency of hotline use and conversation number.
Conversations from the Crisis Text Line provide an opportunity to examine how psychosocial issues among hotline users who report suicidality may vary by frequency of hotline usage and by users’ first and last conversations.
Using latent class analysis, the authors identified three user classes: lower distress, anxious distress, and relational distress.
Across frequency of hotline usage and conversation subsamples, most texters with suicidality endorsed similar issues; yet, distinct subgroups of texters appeared on the basis of presenting psychosocial issues.
Suicide remains a leading cause of death in the United States, and suicide rates have risen in nearly every state from 1999 to 2016 (1). Experiencing suicidal thoughts and attempting suicide are also pervasive across the U.S. population. In 2016, 9.8 million adults (ages 18 and older) reported having serious thoughts of killing themselves, and 1.3 million adults reported attempting suicide during the past 12 months (2).
Regarding this public health crisis, it is important to link people who are experiencing suicidal thoughts and behaviors to accessible and immediate resources. Since the second half of the 20th century, suicide hotlines have played an important role in U.S. suicide prevention efforts by de-escalating people in crisis and referring them to community-based resources (35). In 1999, the National Hopeline Network became the first national suicide hotline (6). Subsequently, the National Suicide Prevention Lifeline was launched in 2005 and is currently the largest suicide hotline in the United States (7). Calls are redirected to local call centers, where volunteer crisis counselors provide crisis intervention and service referrals (8). The National Suicide Prevention Lifeline received 12 million calls from 2005 to 2017 and expects to receive the same number of calls within the next 4 years (7). This high call volume emphasizes the role of empirical research in understanding hotline users.
Research indicates that commonly mentioned problems by users of U.S.-based suicide hotlines include mental and general health problems, financial instability, relationship problems, and housing problems (3, 9). These psychosocial issues correspond with the contributing factors identified by the Centers for Disease Control and Prevention for suicide among individuals with and without psychiatric diagnoses (10). Yet, studies have often included samples from select call centers and have not reported on whether the types of issues vary by the number of calls per user (3, 9), even though repeated callers account for most hotline calls (11). This method obscures potential differences and severity of caller issues by varying frequency of usage. Also, available research has not measured how the issues of repeat callers may change by conversation (1113). Because repeat callers and callers who report suicidal ideation or related behaviors require great amounts of hotline resources (7, 12), it is important to identify these callers’ issues as a first step to inform crisis counseling interventions (13).
Hotlines utilize new technology in an effort to increase help seeking among people who wish to remain anonymous or want to use different modalities (14, 15). Crisis Text Line is a text-based platform supporting a network of crisis support organizations in an effort to increase help seeking for distressed individuals. This free hotline can be accessed via text message, either by phone or Facebook messenger. Users send a text message to the help number and are then connected to a crisis counselor based at one of the network’s crisis centers. The service was originally based in the United States and has now broadened its reach to Canada and the United Kingdom.
The demographic characteristics of the hotline’s users are diverse. Approximately 75% of users are younger than age 25 (16). Of the texters, 6% have identified as Native American, and 19% have identified as Latinx or Hispanic (16)—two ethnic groups that have been traditionally less likely than white non-Latinx residents to seek help via phone hotlines (17, 18). Additionally, 20% of the hotline’s user volume comes from the lowest 10% of income zip codes in the United States (16).
Texters can decide to have their anonymous conversations scrubbed from the hotline’s database. Texter data may inform literature on hotline services and advise training and interventions conducted by crisis counselors. Because the hotline uses a machine-learning algorithm to help high-risk texters first (i.e., those who report suicidal ideation, previous suicide attempts, or both; 19), it is imperative that crisis counselors understand the needs and issues of these texters.
Considering the interest in technology-enhanced crisis hotlines (8) and the use of text messages in suicide attempt prevention (20), more peer-reviewed studies on large-scale conversations of text message hotlines are needed (2123). Because research suggests that individuals who report suicidality are often repeated users of text-based hotlines (22), in this study we aimed to increase knowledge by exploring distinct classes of texters who report suicidality by frequency of Crisis Text Line use and by whether the conversation was the first contact with the hotline, one of two contacts, or the last of several. Usage frequency highlights differences between individuals who have potentially acute versus ongoing issues, whereas conversation number among repeated texters allows insight into whether latent subgroups change by interaction with the hotline. Identifying subgroups of texters and their issues may inform text message hotline counselors on how to communicate with texters who report suicidality and alleviate crises (11, 24, 25).

Methods

The original data set included 211,258 Crisis Text Line conversations in which users mentioned current or previous suicidality from 2013 to 2017. Data scientists determined suicidality through a word search of all conversations approved for research purposes during the hotline’s initial years of service. Data scientists queried the message-level data using machine-learning techniques to find all conversations that included associated terms and word combinations (i.e., phrases that expressed previous or current suicidality, such as “overdose”). The machine-learning model was programmed to predict suicide risk by reading variability in text messages and understanding context (19). The model can identify 86% of people at severe imminent risk for suicide in their first conversations. Additional queries ensured that all selected conversations mentioned phrases related to suicide and that the hotline users and crisis counselors were engaged in actual conversations rather than in “dropped” or incomplete conversations.

Subsamples by Usage Frequency and Number of Conversations

All conversations were from the U.S.-based branch of the hotline. For this analysis, subsamples were created on the basis of the users’ frequency (i.e., total number of text conversations from the same phone number) of accessing the service and conversation number (for frequent texters). Cases with a third-party issue tag were removed from the sample (N=5,690) because this tag indicated that the hotline user was seeking advice for another person rather than for him- or herself; therefore, the suicidal person was not participating in the conversation. Each texter received a unique number after initial contact with the hotline, and the frequency of the texter identification number within the data set indicated whether a texter had called once, twice, or three or more times. One-time users included 92,304 participants, composing 44.9% of the full sample’s conversations (N=205,568). Two-time users (N=15,999) contributed 31,998 conversations (15.6%). Texters contacting the hotline three or more times (N=14,606) contributed 81,266 conversations (39.5%), but only the first and last conversations were included in this analysis to evaluate whether texter issues were different between the initial contact with the hotline and the final conversation. The final analytic sample comprised 153,514 conversations from 122,909 individuals.
This study was approved by the internal review boards from the respected universities as exempt from human subject research. The data set did not include any identifying information of the texters or crisis counselors (e.g., names or specific locations). Access to the aggregate-level texter data were granted by the hotline as part of its initial data enclave program (26).

Measures

Indicators of psychosocial issues were related to four domains: psychological health, relational life, abuse, and environmental stressors. Crisis counselors “tagged” or reported psychosocial issues at the completion of each conversation on the basis of its content. Crisis counselors are volunteers who have completed 34 hours of supervised training and who have been vetted by the hotline to respond to and document calls. Crisis counselors selected from a list of 35 issue tags that had been established by the hotline. Seven tags were excluded from our analysis because they did not inform the latent construct of psychosocial issues, were redundant for our sample (i.e., suicidal ideation), and had been identified by the hotline’s research scientists as highly biased: suicidal ideation, third-party caller, medication issues, none, other, psychiatric hospitalization, and problems with therapist or psychiatrist.
Twenty-eight issues were selected to measure the psychosocial issue construct, including seven psychological health tags (depression, self-harm, anxiety, mental health issues, stress, eating disorders, substance use), seven relational tags (family, friends, relationships, bereavement, isolation, bullying, LGBTQ [lesbian, gay, bisexual, transgender, and questioning]), five abuse tags (physical, sexual, emotional, domestic violence, sexual assault), and nine environmental tags (medical, financial, work, educational testing, school, military, homelessness, homicide, and local and national elections). Tagging was based on the discretion of the crisis counselor. Demographic variables were excluded from the analysis because of the high amount of missing data (approximately 92%). Missing or incomplete caller or texter demographic data are not uncommon in hotline research (11, 27, 28) given that hotlines are an anonymous crisis service. Yet, the amount of missing data in our study exceeded that of other studies (with values ranging from approximately 1% to 64%).

Analysis

The five subsamples based on frequency of use and conversation number were described according to the 28 issue tags informing the latent construct. To facilitate theoretical interpretation of the tags and our findings, we categorized tags into one of four broad issue domains: psychological, relational, abuse, and environmental. We then used latent class analysis (LCA; 29, 30) with an expectation-maximum algorithm to classify texters from subpopulations or “classes” with similar psychosocial issues during their first and last conversation by subsample of users grouped by frequency of use and conversation number. The categorization of issues by psychosocial tag did not influence modeling of the latent classes because all 28 issues were included in the analyses.
LCA is an appropriate statistical technique for our study because it identifies two or more unobserved subgroups of participants on the basis of covariation between multiple observed indicator variables (31, 32). Class enumeration occurred within each frequency and conversation subsample, guided by fit indices and considerations of parsimony (32). Within each subsample, we implemented one- to five-class LCA models and evaluated the following fit statistics: Akaike information criterion, Bayesian information criterion, sample size–adjusted Bayesian information criterion, and Lo-Mendell-Rubin likelihood ratio test (33).
Starting with a one-class model (the baseline comparison for all subsequent models), enumeration continued until empirical underidentification or convergence problems signaled that the inclusion of additional classes was not needed or feasible (see online supplement). Entropy was also estimated to evaluate quality of class separation (34), with values approaching 1 indicating clearer distinction between classes on the basis of the indicators (35). Following model selection, class-specific probabilities of having an item tagged were graphically displayed within each frequency and conversation subsample. Analyses were conducted with Mplus, version 8.1 (36).

Results

Table 1 displays the prevalence of the 28 psychosocial issue tags for each subsample (one-time users, first and second conversation among two-time users, and first and last conversation among ≥3-time users) Across the overall sample, depression (41.9%) and family concerns (25.0%) were the two most frequent issues. Among the one-time users, depression (42.2%) and family concerns (26.5%) were also the two most frequent issues. Among the two-time users, 45.0% described depression as an issue in their first conversation versus 37.7% in the second conversation. Family concerns were again a frequent issue in their first (25.5%) and second (21.9%) conversations. Among users who contacted the Crisis Text Line three or more times, depression (47.7%) and self-harm (25.7%) were the most frequent issues in the first conversation; depression (35.5%) and family concerns (19.2%) were the most frequent issues in the second conversation.
TABLE 1. Prevalence of psychosocial issue tags among users of the Crisis Text Line who expressed suicidality, by frequency of use and conversation numbera
     2-time texters (N=15,999)≥3-time texters (N=14,606)
 Overall sample (N=153,514)1-time texters (N=92,304)First conversationSecond conversationFirst conversationLast conversation
Issue tagN%N%N%N%N%N%
Psychological concern            
 Depression64,32341.938,94842.27,19845.06,03137.76,96847.75,17835.5
 Self-harm25,36716.513,12414.23,24420.32,57716.13,75225.72,67018.3
 Anxiety24,23815.815,01616.32,67216.72,13413.32,61717.91,79912.3
 Mental10,0806.65,9776.51,1137.01,0146.31,0317.19456.5
 Stress24,97016.315,65617.02,74017.12,11313.22,78319.11,67811.5
 Eating disorder3,8152.52,0782.34472.83692.35333.63882.7
 Substance use4,4582.92,9303.24302.73962.53432.33592.5
Relational concern            
 Family38,42525.024,46026.54,08225.53,49921.93,57924.52,80519.2
 Friends15,94910.49,57810.41,79811.21,5659.81,69411.61,3149.0
 Relationships25,09716.316,70018.12,53215.82,31914.51,90113.01,64511.3
 Bereavement4,5282.92,9683.24502.83712.34493.12902.0
 Isolated20,16613.112,22213.22,27114.21,90111.92,18114.91,59110.9
 Bullying6,1114.03,4013.78445.36133.88045.54493.1
 LGBTQ4,1012.72,2472.45363.43792.45874.03522.4
Abuse            
 Physical2,1091.41,2211.32661.71581.03322.3132.9
 Sexual4,1882.72,3712.64502.84022.55733.93922.7
 Emotional2,7331.81,7821.92601.62721.71891.32301.6
 Domestic violence637.4448.548.357.444.340.3
 Sexual assault977.6576.688.699.671.51431.0
Environmental concern            
 Medical4,3332.82,7873.04262.73712.34242.93252.2
 Financial4,3172.83,0473.33822.43372.12872.02641.8
 Work4,0272.62,7603.03642.33422.12721.92892.0
 Testing90.180.15.01.03.01.0
 School6,9254.54,1234.58275.26854.36874.76034.1
 Military323.2239.322.126.215.121.1
 Homeless1,165.8827.979.5114.767.578.5
 Homicide198.1111.112.129.216.130.2
 Election390.3282.332.238.222.216.1
a
Data were from 153,514 Crisis Text Line conversations from 2013 to 2017 (N=122,909 individuals).
Class enumeration results and fit statistics for the unconditional latent class models are presented in Table 2. On the basis of statistical and substantive results, a three-class solution was chosen for each of the five subgroups. Entropy (range 0.73–0.82) indicated that the most probable class assignment was good (34).
TABLE 2. Model fit indices for final latent class models among users of the Crisis Text Line who expressed suicidality, by frequency of use/conversation number subsamplea
  Free Information criteriaLMR likelihood ratio teste 
SubsampleClassesparametersLog likelihoodAICbBICcaBICdValuepEntropy
1 time         
 Single386−508,3131,016,7981,017,6091,017,3366,258.49<.001.758
2 times         
 First386−89,028178,228178,888178,6151,340.22<.001.776
 Second386−80,983162,137162,798162,524901.11<.001.753
≥3 times         
 First386−82,756165,683166,336166,0631,436.56<.001.816
 Last386−70,912141,996142,649142,375841.43.038.732
a
Data were from 153,514 Crisis Text Line conversations from 2013 to 2017 (N=122,909 individuals). Texters were grouped into five subsamples (one-time users; two-time users, first conversation; two-time users, second conversation; ≥3-time users, first conversation; and ≥3-time users, last conversation).
b
Akaike information criterion.
c
Bayesian information criterion.
d
Adjusted Bayesian information criterion.
e
LMR, Lo-Mendell-Rubin.
Similar classes of texters according to their psychosocial issue tags were found across all five frequency/conversation subgroups (Figures 1, 2, and 3). The largest class (approximately 75% of each subsample) was named “lower distress” because the prevalence of all issues, regardless of domain, was lower in this class than in the other classes. However, theses texters expressed nonnegligible levels of depression and self-harm. Other salient issues among this class were relational in nature and pertained to family, friends, or relationships. The second largest class (11%–15% of each subsample) was composed of texters with “anxious distress,” named because they had the highest prevalence of anxiety or stress. They also were more likely to report depression than texters in the lower distress class.
FIGURE 1. Probability of endorsing 28 psychosocial issue tags among one-time texters to the Crisis Text Line who expressed suicidality, by latent classa
aData were from Crisis Text Line conversations with 92,304 one-time users from 2013 to 2017. Latent class analysis identified three classes of texters on the basis of endorsement of psychosocial issue tags: lower distress, anxious distress, and relational distress.
FIGURE 2. Probability of endorsing 28 psychosocial issue tags among two-time texters to the Crisis Text Line who expressed suicidality, by latent class and conversation numbera
aData were from A) the first and B) the second Crisis Text Line conversation with 15,999 two-time users from 2013 to 2017. Latent class analysis identified three classes of texters on the basis of endorsement of psychosocial issue tags: lower distress, anxious distress, and relational distress.
FIGURE 3. Probability of endorsing 28 psychosocial issue tags among ≥3-time texters to the Crisis Text Line who expressed suicidality, by latent class and conversation numbera
aData were from A) the first and B) the last Crisis Text Line conversation with 14,606 ≥3-time users from 2013 to 2017. Latent class analysis identified three classes of texters on the basis of endorsement of psychosocial issue tags: lower distress, anxious distress, and relational distress.
Finally, the smallest group was the “relational distress” class (range 7%–11%). These texters were classified as having the highest prevalence of depression and self-harm. Although they had elevated levels of the majority of indicators in the relational domain, the most salient issues were related to family, friends, and isolation. In addition, school issues emerged as prominent concerns. These texters were classified as having medium levels of anxiety and stress compared with the other two classes. These three classes (lower distress, anxious distress, and relational distress) were found in all five subsamples.

Discussion

In this large-scale study, the authors explored distinct classes of users of the Crisis Text Line who reported suicidality. Classes were based on texters’ presenting psychosocial issues and were explored across frequency of hotline use and conversation number. We found three general classes of texters who reported suicidal ideation or behaviors, which did not change greatly by frequency of hotline use or conversation number. Classes were distinguished mostly by prevalence of psychological and relational issues, as opposed to abuse and environmental issues, reported by crisis counselors during text message interactions. Whereas the most common subgroup across hotline use and conversation displayed fewer issues of depression and self-harm, two smaller subgroups demonstrated problems with stress, anxiety, friends, and school in addition to higher levels of depression and self-harm. Our results suggest that, despite differing frequency of hotline usage, most texters who reported suicidality endorse similar issues, and these issues did not seem to vary by conversation. However, there seemed to be distinct subgroups of texters on the basis of their presenting issues, which can inform hotline strategies.
These findings support the emerging research on the psychosocial issues of users of text message–based hotlines who report suicidal ideation. Sindahl et al. (22) reported that self-harm, bullying, and interpersonal problems were co-occurring issues for Danish texters who reported suicidal ideation. Mental health problems, issues with family or partners, and rumination were the most prominent co-occurring issues for users of the Dutch 113Online suicide prevention crisis chat service (27). Among callers from eight local telephone hotlines in the United States, Gould et al. (37) identified similar psychosocial issues, including exposure to interpersonal problems, mental health concerns, and difficulty meeting basic needs (e.g., shelter, food). More research is needed to better understand the subgroups of users of text message hotlines who report suicidality across frequency of use and conversations, and how they may compare with telephone hotline users.
Our findings have implications for crisis counselor training. The LCA results suggest that, in addition to depression, suicidal texters primarily struggle with stress and anxiety management and interpersonal relationships. Crisis counselors should be prepared to use emotion regulation techniques to support texters until they receive community-based services. Specifically, crisis counselors should understand evidence-based coping skills suited for people with urges to self-harm, anxiety, and interpersonal issues outside of abuse.
Crisis counselors may also tailor approaches to the three distinct classes of texters that were found in the overall sample of texters who reported suicidality. Understanding that these subgroups exist may help crisis counselors address issues not initially presented by the texter. For example, a crisis counselor who engages a texter expressing school-related issues may also want to use open-ended questions to inquire about other issues known to be prevalent in the “relational” class (i.e., the class with the highest probability of school issues, such as problems with family, friends, and feelings of isolation. Because the classes present similarly across frequency and conversation subsamples, how the crisis counselor approaches the individual is not necessarily informed by a texter’s history with the hotline. In contrast to studies on frequent users (1113), our findings suggest that knowing whether a texter is a repeat caller versus a one-time caller does not enhance the crisis counselor’s ability to address the psychosocial issues at hand.
To validate the tagging of texters’ psychosocial issues, the hotline counselor may ask for texters to self-report their most pressing issues. The storage of message-level data may also provide future opportunities to examine the validity and reliability of issues tags from conversations and to improve the accuracy of issue tagging. Such information may illuminate why texters who report suicidal ideation and related behaviors and who are tagged with similar issues display different frequencies of hotline usage and how to encourage community-based help seeking.
Several limitations should be considered when interpreting these findings. As described in the Methods section, data were collected during the hotline’s initial years of service, and there are significant missing data on texter demographic characteristics (i.e., age, gender, and ethnicity). Therefore, we cannot apply the findings to specific populations. Recently, the hotline has used strategies to reduce missing data regarding user demographic characteristics and user experience, such as imputing survey responses for nonrespondents.
Additionally, the crisis counselors tagged texter conversations from a list of prevalent psychosocial issues versus collecting texters’ self-reported problems. Tags were at the discretion of one crisis counselor per conversation, and interrater reliability could not be calculated. Research scientists regularly perform quality checks of the issue tags and have removed highly biased tags from the list (see the Methods section). As previously mentioned, the inclusion of message-level data in future studies is another strategy to further validate the issue tags labeled by the crisis counselors. Last, individuals who use the hotline are a self-selecting population who are seeking help. The findings do not speak to psychosocial issues that non-help-seeking suicidal individuals endure.
Typically, hotline studies are naturalistic studies and do not include a control sample because callers are affected by various life circumstances and may engage in various help-seeking activities (38). Although we did not have access to a comparison sample, future studies may include a sample of texters without suicidality so that differences and similarities in presenting psychosocial issues of texters can be evaluated.

Conclusions

Text message hotlines are one strategy to reach diverse groups of people with suicidal ideation and related behaviors on a nationwide basis. Yet, there is a disparity in peer-reviewed research that explores the meaningful subgroups of texters reporting suicidality. Findings suggest that most texters who report suicidality struggle with psychological and relational issues. However, distinct subgroups of texters face a unique constellation of issues. Crisis counselors may benefit from additional training on emotion regulation and interpersonal problem solving as well as from understanding which issues tend to co-occur. This training will allow crisis counselors to inquire about additional but unexpressed issues. Findings also have implications for how the hotline may supplement collection of texter information through user identification of psychosocial issues and message-level data. This information may inform how the severity and types of issues differ among subgroups of texters with suicidality and inform future implementation of hotline services.

Footnote

These views represent the opinions of the authors and not necessarily those of the National Institutes of Health. Research is reflective of the first and third authors’ ongoing collaboration with the Crisis Text Line.

Supplementary Material

File (appi.ps.201900149.ds001.pdf)

References

1.
Vital Signs: Suicide Rising Across the US. Atlanta, Centers for Disease Control and Prevention, 2018. www.cdc.gov/vitalsigns/suicide/index.html. Accessed Oct 8, 2019
2.
2016 National Survey on Drug Use and Health: Methodological Summary and Definitions. Rockville, MD, Substance Abuse and Mental Health Services Administration, 2017
3.
Evans WP, Davidson L, Sicafuse L: Someone to listen: increasing youth help-seeking behavior through a text-based crisis line for youth. J Community Psychol 2013; 41:471–487
4.
Litman RE, Farberow NL, Shneidman ES, et al: Suicide-prevention telephone service. JAMA 1965; 192:21–25
5.
Dew MA, Bromet EJ, Brent D, et al: A quantitative literature review of the effectiveness of suicide prevention centers. J Consult Clin Psychol 1987; 55:239–244
6.
Spencer-Thomas S, Jahn DR: Tracking a movement: US milestones in suicide prevention. Suicide Life Threat Behav 2012; 42:78–85
7.
Kim V: As Calls to the Suicide Prevention Lifeline Surge, Under-Resourced Centers Struggle to Keep Up. Arlington, VA, PBS, 2018. www.pbs.org/newshour/nation/suicide-prevention-lifeline-centers-calls.Accessed Oct 8, 2019
8.
Crosby Budinger M, Cwik MF, Riddle MA: Awareness, attitudes, and use of crisis hotlines among youth at-risk for suicide. Suicide Life Threat Behav 2015; 45:192–198
9.
Ramchand R, Jaycox L, Ebener P, et al: Characteristics and proximal outcomes of calls made to suicide crisis hotlines in California. Crisis 2017; 38:26–35
10.
Suicide Rising Across the US: More Than a Mental Health Concern [Graphic 3: Many Factors Contribute to Suicide Among Those With and Without Known Mental Health Conditions]. Atlanta, Centers for Disease Control and Prevention, 2018. www.cdc.gov/vitalsigns/suicide/infographic.html#graphic3. Accessed Oct 8, 2019
11.
Middleton A, Pirkis J, Chondros P, et al: The health service use of frequent users of telephone helplines in a cohort of general practice attendees with depressive symptoms. Adm Policy Ment Health 2016; 43:663–674
12.
Pirkis J, Middleton A, Bassilios B, et al: Frequent callers to telephone helplines: new evidence and a new service model. Int J Ment Health Syst 2016; 10:43
13.
Middleton A, Woodward A, Gunn J, et al: How do frequent users of crisis helplines differ from other users regarding their reasons for calling? Results from a survey with callers to Lifeline, Australia’s national crisis helpline service. Health Soc Care Community 2017; 25:1041–1049
14.
Gould MS, Lake AM, Galfalvy H, et al: Follow-up with callers to the National Suicide Prevention Lifeline: evaluation of callers’ perceptions of care. Suicide Life Threat Behav 2018; 48:75–86
15.
Predmore Z, Ramchand R, Ayer L, et al: Expanding suicide crisis services to text and chat. Crisis 2017; 38:255–260
16.
Crisis Trends: FAQs. New York, Crisis Text Line, 2018. crisistrends.org/#faq. Accessed Oct 8, 2019
17.
Draper J, McCleery G, Schaedle R: Mental health services support in response to September 11: the central role of the Mental Health Association of New York City; in 9/11: Mental Health in the Wake of Terrorist Attacks. Edited by Neria Y, Gross R, Marshall RD, et al. Cambridge, England, Cambridge University Press, 2006
18.
Halpern P: An AI/AN Suicide Prevention Hotline: Literature Review and Discussion With Experts. Washington, DC: Office of the Assistant Secretary for Planning and Evaluation, 2009
19.
Gupta A: Detecting Crisis: An AI Solution. New York, Crisis Text Line, 2018. www.crisistextline.org/blog/ava2. Accessed Oct 8, 2019
20.
Berrouiguet S, Baca-García E, Brandt S, et al: Fundamentals for future mobile-health (mHealth): a systematic review of mobile phone and web-based text messaging in mental health. J Med Internet Res 2016; 18:e135
21.
Althoff T, Clark K, Leskovec J: Large-scale analysis of counseling conversations: an application of natural language processing to mental health. Trans Assoc Comput Linguist 2016; 4:463–476
22.
Sindahl TN, Côte L-P, Dargis L, et al: Texting for help: processes and impact of text counseling with children and youth with suicide ideation. Suicide Life Threat Behav (Epub ahead of print, Nov 23, 2018)
23.
Dinakar K, Chen J, Lieberman H, et al: Mixed-initiative real-time topic modeling and visualization for crisis counseling; in Proceedings of the 20th International Conference on Intelligent User Interfaces. Atlanta, ACM Press, 2015
24.
Grigorash A, O’Neill S, Bond R, et al: Predicting caller type from a mental health and well-being helpline: analysis of call log data. JMIR Ment Health 2018; 5:e47
25.
Ohtaki Y, Oi Y, Doki S, et al: Characteristics of telephone crisis hotline callers with suicidal ideation in Japan. Suicide Life Threat Behav 2017; 47:54–66
26.
Pisani AR, Kanuri N, Filbin B, et al: Protecting user privacy and rights in academic data-sharing partnerships: principles from a pilot program at Crisis Text Line. J Med Internet Res 2019; 21:e11507
27.
Mokkenstorm JK, Eikelenboom M, Huisman A, et al: Evaluation of the 113Online suicide prevention crisis chat service: outcomes, helper behaviors and comparison to telephone hotlines. Suicide Life Threat Behav 2017; 47:282–296
28.
Gould MS, Greenberg T, Munfakh JLH, et al: Teenagers’ attitudes about seeking help from telephone crisis services (hotlines). Suicide Life Threat Behav 2006; 36:601–613
29.
Goodman LA: Exploratory latent structure analysis using both identifiable and unidentifiable models. Biometrika 1974; 61:215–231
30.
Lazarsfeld PF, Henry NW: Latent Structure Analysis. New York, Houghton Mifflin, 1968
31.
Masyn KE: Latent class analysis and finite mixture modeling; in The Oxford Handbook of Quantitative Methods in Psychology: Statistical Analysis, vol. 2. Edited by Little TD. New York, Oxford University Press, 2013
32.
Nylund KL, Asparouhov T, Muthén BO: Deciding on the number of classes in latent class analysis and growth mixture modeling: A Monte Carlo simulation study. Struct Equ Modeling 2007; 14:535–569
33.
Lo Y, Mendell NR, Rubin DB: Testing the number of components in a normal mixture. Biometrika 2001; 88:767–778
34.
Celeux G, Soromenho G: An entropy criterion for assessing the number of clusters in a mixture model. J Classif 1996; 13:195–212
35.
Asparouhov T, Muthen B: Variable-Specific Entropy Contribution. Los Angeles, Muthén & Muthén, 2018
36.
Muthén LK, Muthén BO: Mplus User’s Guide, 8th ed. Los Angeles, Muthén & Muthén, 1998
37.
Gould MS, Kalafat J, Harrismunfakh JL, et al: An evaluation of crisis hotline outcomes. Part 2: suicidal callers. Suicide Life Threat Behav 2007; 37:338–352
38.
Mishara BL, Chagnon F, Daigle M, et al: Which helper behaviors and intervention styles are related to better short-term outcomes in telephone crisis intervention? Results from a silent monitoring study of calls to the US 1-800-SUICIDE network. Suicide Life Threat Behav 2007; 37:308–321

Information & Authors

Information

Published In

Go to Psychiatric Services
Go to Psychiatric Services
Psychiatric Services
Pages: 319 - 327
PubMed: 31795857

History

Received: 21 March 2019
Revision received: 20 September 2019
Accepted: 1 October 2019
Published online: 4 December 2019
Published in print: April 01, 2020

Keywords

  1. Suicide and self-destructive behavior
  2. Crisis intervention
  3. Hotline
  4. Texting

Authors

Details

Hannah Selene Szlyk, Ph.D., L.C.S.W. [email protected]
Brown School, Washington University in St. Louis, St. Louis (Szlyk, Roth); Campus Puente El Común, Universidad de La Sabana, Bogotá, Colombia (García-Perdomo).
Kimberly Beth Roth, Ph.D., M.H.S.
Brown School, Washington University in St. Louis, St. Louis (Szlyk, Roth); Campus Puente El Común, Universidad de La Sabana, Bogotá, Colombia (García-Perdomo).
Víctor García-Perdomo, Ph.D.
Brown School, Washington University in St. Louis, St. Louis (Szlyk, Roth); Campus Puente El Común, Universidad de La Sabana, Bogotá, Colombia (García-Perdomo).

Notes

Send correspondence to Dr. Szlyk ([email protected]).

Competing Interests

The authors report no financial relationships with commercial interests.

Funding Information

This research was supported by National Institute of Mental Health grant T32-MH-019960 (to Drs. Szlyk and Roth). The University of Texas at Austin Health Communications Scholars Program supported the early phase of this study (Drs. Szlyk and García-Perdomo).

Metrics & Citations

Metrics

Citations

Export Citations

If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download.

For more information or tips please see 'Downloading to a citation manager' in the Help menu.

Format
Citation style
Style
Copy to clipboard

View Options

View options

PDF/EPUB

View PDF/EPUB

Get Access

Login options

Already a subscriber? Access your subscription through your login credentials or your institution for full access to this article.

Personal login Institutional Login Open Athens login
Purchase Options

Purchase this article to access the full text.

PPV Articles - Psychiatric Services

PPV Articles - Psychiatric Services

Not a subscriber?

Subscribe Now / Learn More

PsychiatryOnline subscription options offer access to the DSM-5-TR® library, books, journals, CME, and patient resources. This all-in-one virtual library provides psychiatrists and mental health professionals with key resources for diagnosis, treatment, research, and professional development.

Need more help? PsychiatryOnline Customer Service may be reached by emailing [email protected] or by calling 800-368-5777 (in the U.S.) or 703-907-7322 (outside the U.S.).

Media

Figures

Other

Tables

Share

Share

Share article link

Share