Preferred Sources For Suicide Prevention And Crisis Services Among Segments Of The US Adult Population

Abstract

New policy initiatives aim to increase use of crisis services. We conducted a nationally representative probability survey of 5,006 adults in 2023 and used latent class analysis to identify segments of U.S. adults who vary in their intentions to seek help from five different sources if experiencing suicidality/mental health crisis. Based on the clustering of responses, the segments were named: 1) “Seek Help Nowhere,” 2) “Definitely Not 988, Yes Friends and Family—Distressed,” 3) “Seek Help Everywhere,” 4) “Seek Help Most Places, But Not Religious Network,” and 5) “Relatively Indifferent—Not Distressed.” Past 30-day psychological distress levels were significantly associated with membership in some segments; awareness of the 988 Suicide and Crisis Lifeline and Democrat political party affiliation were associated with significantly higher or lower odds of membership in every segment. Communication campaigns that encourage crisis service use and help seeking may consider tailoring messages for these different audience segments.

Federal and state governments have recently launched a series of policy initiatives aimed at increasing access to, and use of, suicide prevention and crisis services. One such initiative is the 988 Suicide and Crisis Lifeline (hereafter referred to as 988), which went live nationally in July 2022 and replaced a ten-digit dialing code (Public Law No: 116–172).(1–3) Another federal initiative relates to investments in Certified Community Behavioral Health Clinics (CCBHCs), which have operated since 2016 and operate in approximately 40% of U.S. counties.(4) These clinics must provide nine categories of mental and substance use disorder services, including suicide prevention and crisis care.(5) Of the approximately 500 CCBHCs in the U.S, about 10% operate a 988 call center, and 80% have established partnerships with 988 centers.(6)

While these investments in suicide prevention and crisis service infrastructure have great potential, their impact will be influenced by the extent to which people actually reach out to 988 or mental health professionals at CCBHCs when they or a loved one are experiencing suicidality or a mental health crisis. Empirically-informed communication campaigns that promote service seeking are critical to achieving this goal;(7) and this sentiment is prominent in recently published federal suicide prevention planning documents.

For example, Goal 7 of the U.S. Department of Health and Human Services’ 2024 National Strategy for Suicide Prevention is to “Implement research-informed suicide prevention communication activities in diverse populations using best practices from communication science.”(8) The U.S. Substance Abuse and Mental Health Services Administration’s 2024 988 Quality and Services Plan emphasizes the importance of marketing about 988 and notes that “States and Territories must develop and implement a comprehensive 988 communication strategy…”.(9) While much prior research has examined determinants of seeking mental health services and evaluated interventions to encourage these behaviors,(10–14) much less has focused on help-seeking in crisis situations and very little has been conducted in the United States post-988 implementation.(15–18)

With the goal of strengthening the evidence base to inform communication campaigns about 988, CCBHCs, and other suicide prevention and crisis services, we conducted an empirical clustering audience segmentation study using data from a nationally representative probability sample of U.S. adults. The aims of the study were to:

1. Identify segments of U.S. adults who vary in terms of their intentions to seek help from different sources if they or a loved one were to experience suicidality or a mental health crisis; and

2. Identify individual characteristics that are independently associated with membership in different segments.

METHODS

We used the Ipsos KnowledgePanel to conduct a nationally representative web-based, cross-sectional survey of 5,006 adults U.S. adults in June 2023 (response rate= 55.0%). The KnowledgePanel is a probability survey that uses address-based sampling and has been widely used in public opinion research (18–26). Ipsos recruits KnowledgePanel members via post-mail and telephone and then administers surveys online. Additional details about the KnowledgePanel are in the Appendix, Exhibit A.(27)

The dependent variables were the mutually exclusive audience segments—created via latent class analysis—that we assigned respondents to the basis of their self-reported likelihood reaching out to five sources if they or a loved one were to experience suicidality or a mental health crisis. The sources were: 1) the “988 Lifeline” (which was defined in the survey) 2) “a crisis line other than the 988 Lifeline (e.g., Trevor Project, Crisis Text Line),” 3) “a psychologist, psychiatrist, social worker, or other mental health professional,” 4) “a friend or family member,” or 5) “someone in my religious network.” We randomized the list order of sources between respondents to reduce the risk of order effect bias.

Respondents rated their likelihood of reaching out to each source on 7-point scales (1= “very unlikely,” 7= “very likely”). Ratings of 6–7 were re-coded as being “likely” (yes/no) to reach out to the source in a crisis, and ratings of 1–2 were coded as being “unlikely” (yes/no). This resulted in ten dichotomous variables, two for each source, that were included in the latent class analysis. Creating dichotomous variables from scale items with unknown predictive validity is a common practice in public opinion research and audience segmentation analyses and aides the interpretation of results.(28, 29)

We used latent class analysis (LCA) to cluster survey respondents in terms of the sources they reported being “likely” and “unlikely” to reach out to if they/a loved one were experiencing suicidality/crisis. We conducted the LCA using the stepmix package in Python.(30) The number of classes (i.e., segments) was chosen to optimize the cross-validated log-likelihood and based on fit statistics such as Akaike information criterion and Bayesian information criterion.(31) The optimal cross-validated log likelihood occurred at a five-segment solution and we chose this solution based on this statistic as well as other fit parameters (see Appendix, Exhibit B.)(27)

We then calculated the weighted portion of respondents in each segment who indicated being “likely” and “unlikely” to reach out to each source. After we examined patterns in the distribution of the ten dichotomous variables across the five segments, we named each segment to concisely capture its most salient features.(32)

Independent variables were a measure of psychological distress, a range of demographic characteristics, and awareness of 988. We assessed past 30-day psychological distress using Kessler’s six-item K6 Scale.(33) K6 scores (Cronbach’s α=.93 in the sample) were calculated and respondents were categorized as having serious (score ≥13), moderate (score 5–12), or no distress (score ≤ 4) based on established cut-points.(33, 34) We included the following demographic characteristics in analyses: age category, annual household income category, highest level of education, race/ethnicity, gender, urbanicity, political party affiliation, and U.S. Census region. We also asked respondents of they had “heard of the 988 Lifeline prior to [the] survey.”

We conducted all analyses in Python, version 3.11.0. Survey-specific weights, provided by Ipsos, were applied for all analyses to generate nationally-representative estimates of the U.S. adult population. Missing data were rare (e.g., 1.9% of responses for the psychological distress score, <0.8% for each source item), and were imputed using Multiple Imputation by Chained Equations (MICE) and the Python miceforest package.

The weighted proportions of respondents with each demographic characteristic in each segment were calculated with 95% Bernoulli confidence intervals. We used multivariable weighted logistic regression models to identify respondent characteristics independently associated with membership in each segment. MICE was again used to handle missing data, and confidence intervals were pooled using Rubin’s rules.(35) Data and statistical code are available at: https://github.com/alex-dahlen/988-clustering.

Limitations

Each of the sources were assessed by a single item with unknown predictive validity. It is possible that a person’s intentions of turning to different sources in a hypothetical crisis vary from where they would actually turn. To help address this measurement limitation, we dichotomized these items at the extreme ends of the scale (i.e., 1–2 and 6–7) to focus on strong preferences that are presumably reasonably accurate proxies for future help-seeking behaviors.

Although informed by literature on crisis help seeking, the five sources listed were not exhaustive of all the sources that a person could turn to in a crisis. Furthermore, the survey items did not assess differences in preferences for help seeking in suicidal vs. non-suicidal crises or in terms of whether it was the respondent themselves or a “loved one” who was experiencing the hypothetical crisis.

Data were collected in summer 2023, one year after 988 launched nationally, and it is possible that preferences related to 988 have changed as the public has become more familiar with the crisis line. Finally, the survey was limited to adults in the United States and results may not be generalizable to youth or populations outside of the U.S.

RESULTS

Appendix Exhibit C shows the demographic composition of the sample. Exhibit 1 shows the distribution of the ten dichotomous crisis service/support source variables that clustered together to create the five mutually exclusive segments. Exhibit 2 presents the demographic characteristics of each segment and Exhibit 3 shows adjusted associations between each demographic characteristic and segment membership. The text below names and describes the most salient features of the segments.

Exhibit 1:

Clustered Segments of Crisis Help Seeking Intentions, Survey of U.S. Adults about Preferred Sources for Suicide Prevention and Crisis Services, June 2023 (N=5,006)

	Segment
	Seek Help Nowhere	Definitely Not 988, Yes Friends and Family—Distressed	Seek Help Everywhere	Seek Help Most Places, But Not Religious Network	Relatively Indifferent—Not Distressed
	%	%	%	%	%
Percent of Sample	20.7	13.2	23.4	25.5	17.1
“Likely” to Reach Out to Source
988 Suicide & Crisis Lifeline	0.6	0.0	82.2	69.7	6.1
A Crisis Line other than 988	0.1	0.0	70.5	41.9	1.5
A Mental Health Professional	3.0	22.0	82.9	81.8	18.4
A Friend/Family Member	0.0	61.3	88.6	77.0	34.5
Someone in Religious Network	0.6	23.4	72.7	0.0	10.5
“Unlikely” to Reach Out to Source
988 Suicide & Crisis Lifeline	93.0	94.6	0.9	0.8	6.4
A Crisis Line other than 988	97.1	98.2	5.3	20.8	19.6
A Mental Health Professional	90.5	40.4	1.2	6.5	9.9
A Friend/Family Member	78.2	0.0	0.4	7.2	6.3
Someone in Religious Network	95.0	55.8	0.0	93.3	42.1

Source/Notes: Authors’ analysis of survey data Ipsos KnowledgePanel survey data, data collected by Ipsos at the request of the authors.

Exhibit 2.

Past 30-Days Psychological Distress Level and Demographic Characteristics of Segments, Survey of U.S. Adults about Preferred Sources for Suicide Prevention and Crisis Services, June 2023 (N=5,006)

	Seek Help Nowhere	Definitely Not 988, Yes Friends and Family—Distressed	Seek Help Everywhere	Seek Help Most Places, But Not Religious Network	Relatively Indifferent—Not Distressed
	%	%	%	%	%
Percent of Sample	20.7	13.2	23.4	25.5	17
Has Heard of 988
Yes	29.6	31.3	43.1	52.4	47.5
No	70.4	68.7	56.9	47.6	52.5
Psychological Distress
Serious (≥ 13)	8.7	8.6	6.5	9.3	5.8
Moderate (5–12)	19.7	23.5	18.5	22.6	17.8
None (≤ 4)	71.7	67.9	74.9	68.2	76.4
Age
18–29	18.5	18.8	19.1	24.8	15.5
30–44	26.8	27.0	25.5	26.8	23.3
45–59	24.1	26.7	24.0	21.2	26.5
≥ 60	30.6	27.5	31.4	27.2	34.7
Household Income
< $10,000	7.7	2.6	2.2	1.4	3.6
$10,000– $24,999	11.6	5.5	9.3	7.4	7.5
$25,000– $49,999	17.8	15.1	17.5	12.0	18.8
$50,000– $74,999	15.9	16.8	17.1	13.8	16.6
$75,000– $99,999	11.0	13.1	13.3	13.9	13.2
$100,000– $149,999	14.9	19.4	17.7	20.3	17.9
≥ $150,000	20.9	27.6	22.8	31.3	22.3
Highest Education
Less than high school	16.7	6.5	8.4	7.3	10.6
High school	34.8	28.2	28.7	23.6	29.6
Some college	25.7	26.4	26.2	25.1	27.1
≥ Bachelor’s degree	22.8	38.8	36.7	43.9	32.7
Race/Ethnicity
Black, Non-Hispanic	12.1	9.1	14.1	7.6	16.5
Hispanic	21.5	14.6	15.5	13.9	20.0
Two or more races, Non-Hispanic	3.8	4.0	4.5	4.0	4.3
Other race, Non-Hispanic	5.9	4.2	5.5	4.6	3.6
White, Non-Hispanic	56.7	68.0	60.5	69.9	55.6
Gender
Male	46.7	49.7	49.9	52.9	56.3
Female	53.3	50.3	50.1	47.1	43.7
Political Party
Democrat	25.7	24.6	26.9	44.4	40.6
Independent	31.9	29.7	29.3	30.2	28.0
Other	13.2	10.8	8.7	8.6	10.1
Republican	29.2	35.0	35.2	16.8	21.3

Source/Notes: Authors’ analysis of survey data Ipsos KnowledgePanel survey data, data collected by Ipsos at the request of the authors.

Exhibit 3.

Adjusted Associations between Past 30-Day Psychological Distress Level, Demographic Characteristics, and Segment Membership, Survey of U.S. Adults about Preferred Sources for Suicide Prevention and Crisis Services, June 2023 (N=5,006)

	Seek Help Nowhere (Model 1)	Definitely Not 988, Yes Friends and Family—Distressed (Model 2)	Seek Help Everywhere (Model 3)	Seek Help Most Places, But Not Religious Network (Model 4)	Relatively Indifferent—Not Distressed (Model 5)
	AOR	AOR	AOR	AOR	AOR
Has Heard of 988
Yes	0.57***	0.56***	1.16*	1.47***	1.44***
No	Ref.	Ref.	Ref.	Ref.	Ref.
Psychological Distress
Serious (≥ 13)	0.92	1.49***	0.88	1.12	0.76**
Moderate (5–12)	0.96	1.44*	0.80	1.37*	0.66*
None (≤ 4)	Ref.	Ref.	Ref.	Ref.	Ref.
Age
18–29	0.83	1.18	1.00	1.62***	0.58***
30–44	0.95	1.17	0.99	1.22*	0.75**
45–59	0.95	1.21	0.97	0.99	0.97
≥ 60	Ref.	Ref.	Ref.	Ref.	Ref.
Household Income
< $10,000	1.15	0.57**	1.58**	0.76	0.99
$10,000– $24,999	0.86	0.95	1.14	0.95	1.14
$25,000– $49,999	0.96	0.86	1.44**	0.63***	1.35*
$50,000– $74,999	0.92	1.01	1.42**	0.70**	1.14
$75,000– $99,999	0.80	0.92	1.24	0.93	1.14
$100,000– $149,999	2.33***	0.70	0.78	0.34***	1.23
≥ $150,000	Ref.	Ref.	Ref.	Ref.	Ref.
Highest Education
Less than high school	2.04***	0.85	0.77**	0.79*	1.08
High school	2.70***	0.55**	0.68**	0.74*	1.11
Some college	1.65***	0.89	0.83*	0.88	1.06
≥ Bachelor’s degree	Ref.	Ref.	Ref.	Ref.	Ref.
Race/Ethnicity
Black, Non-Hispanic	1.12	0.82	1.71***	0.40***	1.57***
Hispanic	1.20	0.85	1.18	0.56***	1.55***
Two or more races, Non-Hispanic	1.50*	0.76	1.42	0.61**	0.96
Other race, Non-Hispanic	0.93	0.89	1.30	0.74	1.31
White, Non-Hispanic	Ref.	Ref.	Ref.	Ref.	Ref.
Gender
Male	0.75***	0.95	0.98	1.12	1.26**
Female	Ref.	Ref.	Ref.	Ref.	Ref.
Political Party
Democrat	0.66***	0.58***	0.45***	3.00***	1.60***
Independent	0.98	0.77*	0.61***	1.91***	1.23
Other	1.09	0.87	0.53***	1.61***	1.36*
Republican	Ref.	Ref.	Ref.	Ref.	Ref.

Source/Notes: Authors’ analysis of survey data Ipsos KnowledgePanel survey data, data collected by Ipsos at the request of the authors. Notes: ** AOR= adjusted odds ratio.

^{* =}p < 0.05,

^{** =}p < 0.01,

^{*** =}p < 0.001.

“Seek Help Nowhere” Segment

This segment comprised 20.7% of the sample. The most salient feature of the Seek Help Nowhere segment was not being “likely” to reach out to the sources of crisis service/support and expressing being “unlikely” to reach out to the sources (Exhibit 1). For each of the sources, ≤3% of respondents in this segment indicated being “likely” to reach out if they/a loved one were experiencing suicidality/crisis and ≥78.2% reported being “unlikely” to reach out. For example, only 0.6% of respondents in this segment were “likely” to reach out to 988 and 0% were “likely” to reach out to a friend/family member while 93.0% and 78.2%, respectively, expressed being “unlikely” to reach out to these sources.

Less than one-third (29.6%) of respondents in the Seek Help Nowhere segment had heard of 988. Respondent in this segment also generally had lower household income and had less formal education than respondents in other segments. For example, the proportions of respondents with annual household income less than $10,000 (7.7%) and less than a high school degree (16.7%) were twice as large as those in most other segments (Exhibit 2). After adjustment (Exhibit 3, Model 1), respondents with less than high school education had over two times higher odds of belonging to the segment than respondents with a bachelor’s degree or higher (AOR= 2.04, p< .0001) and those that had heard of 988 (AOR= 0.57, p= .0002), were male (compared to female, AOR= 0.75, p= .0001) and Democrat (compared to Republican, AOR= 0.66, p= .0001) had significantly lower odds of segment membership.

“Definitely Not 988, Yes Friends and Family—Distressed” Segment

This was the smallest segment, comprising 13.2% of the sample. Like respondents in the Seek Help Nowhere segment, the Definitely Not 988, Yes Friends and Family—Distressed segment was characterized by not being ”likely” to reach out to 988 or another crisis line if they/a loved one were experiencing suicidality/crisis. Zero percent of respondents in this segment reported being “likely” to reach out to 988 or another crisis line, and 94.6% and 98.2%, respectively, reported being “unlikely” to reach out to these sources (Exhibit 1). However, in contrast to the Seek Help Nowhere segment, the majority (61.3%) of respondents in the Definitely Not 988, Yes Friends and Family—Distressed segment indicated being “likely” to reach out to a friend or family member and 0% reported being “unlikely” to reach out to this source.

Less than one-third (31.3%) of respondents in the Definitely Not 988, Yes Friends and Family—Distressed segment had heard of 988. This was also the only segment for which membership was independently associated with significantly higher odds of having serious or moderate past 30-day psychological distress. After adjustment (Exhibit 3, Model 2), respondents with serious distress had 49% (AOR= 1.49, p= .0002) higher odds of belonging to this segment compared to those with no distress and respondents with moderate distress had 44% (AOR= 1.44, p= .02) higher odds. Having heard of 988 (AOR=0.56, p< .0001) and identifying as a Democrat (compared to Republican, AOR= 0.58, p< .0001) was associated with significantly lower odds of segment membership.

“Seek Help Everywhere” Segment

This segment comprised 23.4% of the sample. In sharp contrast to the Seek Help Nowhere and Definitely Not 988, Yes Friends and Family—Distressed segments, the core attribute of the Seek Help Everywhere segment was that relatively large proportions of the segment indicated being “likely” to reach out to all five sources and relatively small proportions reported being “unlikely” to reach out to each source. For each of the sources, ≥70.5% of respondents in this segment indicated being “likely” to reach out if they/a loved one were experiencing suicidality/crisis and ≤5.3% reported being “unlikely” to reach out (Exhibit 1). For example, 82.2% of respondents in this segment reported being “likely” to use 988 and 72.7% reported being “likely” to reach out to someone in their religious network; while 0.9% and 0%, respectively, reported being “unlikely” to reach out to these sources.

After adjustment (Exhibit 3, Model 3), respondents identifying as Black, Non-Hispanic c 1.71, p< .0001) and having less annual household income (e.g., $25,000-$49,999 compared to ≥$150,000; AOR= 1.44, p= .002) had significantly higher odds of belonging to the Seek Help Everywhere segment. Having lower levels of education (e.g., high school compared to bachelor’s degree; AOR= 0.68, p= .007) and identifying as a Democrat (compared to Republican; AOR= 0.45, p<.0001) were associated with significantly lower odds of segment membership.

“Seek Help Most Places, But Not Religious Network” Segment

This segment comprised 25.5% of the sample. Similar to the Seek Help Everywhere segment, relatively large proportions of respondents in the Seek Help Most Places, But Not Religious Network segment reported being “likely” to reach out to 988, a mental health professional, and a friend or family member if they/a loved one were experiencing suicidality/crisis; and relatively small proportions reported being “unlikely” to reach out to these sources. A distinct feature of this segment, however, was that 0% of respondents reported being “likely” to reach out to someone in their religious network, and 93.3% reported being “unlikely” to reach out to this source (Exhibit 1).

About half (52.4%) of respondents in the Seek Help Most Places, But Not Religious Network segment had heard of 988. This was the only segment for which younger age was independently associated with higher odds of segment membership. After adjustment (Exhibit 3, Model 4) and compared to respondents ages 60 or older, respondents ages 18–29 (AOR= 1.62, p< .0001) and ages 30–44 (AOR= 1.22, p= .04) both had significantly higher odds of segment membership. Having heard of 988 (AOR= 1.47, p< .0001), moderate past 30-day psychological distress (compared to no distress; AOR= 1.37, p=.01), and identifying as a Democrat (compared to Republican; AOR= 3.00, p< .0001) were also associated with significantly higher odds of segment membership. Identifying as Black, non-Hispanic (compared to White, Non-Hispanic; AOR= 0.40, p< .0001) was associated with significantly lower odds of membership in the Seek Help Most Places, But Not Religious Network segment.

“Relatively Indifferent—Not Distressed” Segment

This segment comprised 17.1% of the sample. The Relatively Indifferent—Not Distressed segment was defined by relatively small proportions of respondents being “likely” or “unlikely” to reach out each source if they/a loved one were experiencing suicidality/crisis (Appendix, Exhibit D (27)) For example, 6.5% percent of respondents in this segment indicated being “likely” to use 988 in a crisis and 6.4% reported being “unlikely” to reach out to this source (Exhibit 1). The sources that respondents in this segment were most “likely” to reach out to were a friend or family member (34.5%) and a mental health professional (18.4%), with 6.3% and 9.9%, respectively, being “unlikely” to reach out to these sources.

The Relatively Indifferent—Not Distressed segment was the only segment for which membership was independently associated with significantly lower odds of having serious or moderate past 30-day psychological distress. After adjustment (Exhibit 3, Model 5), respondents with serious distress had 24% lower odds (compared to no distress; AOR= 0.76, p= .009) of belonging to this segment and respondents with moderate distress had 34% lower odds (AOR= 0.66, p= .01). Respondent characteristics independently associated with significantly higher odds of membership in this segment included identifying as Black, non-Hispanic (AOR= 1.57, p= .0002) and Hispanic (AOR=1.55, p= .0001) compared to White, non-Hispanic and identifying as Democrat (compared to Republican; AOR= 1.66, p<.0001).

DISCUSSION

This clustered audience segmentation analysis identified a typology of U.S. adults who vary in terms of their preferences for sources of crisis services and supports if they or a loved one were experiencing suicidality or another mental health crisis. Different demographic characteristics—especially political party affiliation—past 30-day distress level, and having heard of 988 were associated with segment membership. Five statistically distinct crisis help-seeking segments were identified, a finding consistent with calls to consider population heterogeneity and diverse preferences when designing suicide prevention programs and policies.(36, 37) Two segments—Seek Help Nowhere and Definitely Not 988, Yes Friends and Family—Distressed—represent priorities for segment-tailored communication campaigns given these segments’ reluctance to seek help from most sources.

Political party affiliation was the only demographic characteristic consistently associated with significantly higher or lower odds of membership in each segment. Of note, the Definitely Not 988, Yes Friends and Family—Distressed segments were comprised of disproportionately large proportions of Republicans. While prior research has shown bi-partisan support for suicide prevention,(38) the nuanced differences in partisan preferences for crisis services and supports observed in the current study are consistent with prior research on mental health service preferences and trust in health care institutions more broadly.(39–42)

For instance, a public opinion survey fielded immediately prior to the launch of 988 found that Republicans were significantly less supportive of 988 than Democrats.(43) Republicans have also been found to have less confidence in mental health treatments and be more likely to attribute mental disorders to personal responsibility.(28, 44) Interpreted within the context of this prior work, our findings highlight partisan differences in preferences for suicide prevention services and signal the need to consider partisan asymmetries and values when designing messages about 988 and other crisis interventions.(45–47)

Efforts to tailor messages for the Seek Help Nowhere segment should consider the demographic characteristics of this segment as well as the breadth of literature on barriers to suicide prevention services and mental health treatment seeking. This segment was composed of disproportionately large proportions of respondents that had annual household income <$10,000 (7.7%, more than twice that of any other segment), less than high school education (16.7%), and identified as female (53.3%). Reviews of barriers to help-seeking among people experiencing suicidality suggest that factors related to high self-reliance, lack of perceived need for services, preference for self-management, fear of hospitalization, suicide stigma, and structural factors (e.g., lack of time, cost of services), may contribute to help seeking hesitancy among this segment.(36, 37, 48–52)

Although nearly 80% of members of the Seek Help Nowhere segment reported being “unlikely” to reach out to a friend or family member in a crisis, this was the source that the smallest proportion of members in the segment were “unlikely” to reach out to. As such, interventions that train lay individuals in suicide prevention strategies may hold potential for the Seek Help Nowhere segment.(53–55) The same holds true for the Definitely Not 988, Yes Friends and Family—Distressed segment, for which family was a primary source. Further, given that nearly one-quarter of respondents in this segment reporting being “likely” to turn to someone in their religious network, findings highlight the importance of deploying strategies that are informed by research about barriers and facilitators to implementing suicide prevention interventions in faith-based settings.(56–59)

Being aware of 988 was associated with significantly lower odds of membership in the Seek Help Nowhere as well was Definitely Not 988, Yes Friends and Family—Distressed segment. Given that awareness about 988 is readily modifiable—unlike other factors associated with membership in these segments, such as political party affiliation in income—the finding underscores the potential of tailored communication campaigns that reach and subsequently increase awareness about 988 among members of these segments.(3, 15, 18)

CONCLUSION

There is population-level heterogeneity in U.S. adults’ preferences for suicide prevention and crisis services. Our cluster analysis suggests that five distinct population segments exist, and that political party affiliation and awareness of 988 are a consistent and significant predictors of segment membership. Messaging campaigns aiming to encourage crisis help seeking from 988, CCBHCs, and other sources may consider tailoring messages for different segments.