This was a retrospective cohort evaluation of patients presenting for their first visit with an outpatient mental health clinician between January 1, 2024, and October 31, 2025, in the Headway mental health clinician network. Headway is the largest behavioral health network in the country, comprising 60,000 clinicians who have more than 1 million patient visits per month [6]. We excluded patients without commercial insurance (ie, Medicaid or Medicare coverage or self-pay).

We collected all patient, -clinician, and visit information from the Headway digital platform. Patient information included patient age, sex, home zip code, clinical condition, and insurance type. Clinician type was classified as psychotherapist or psychiatric clinician.

To assess SI at presentation, we extracted responses on the Patient Health Questionnaire-9 (PHQ-9) at participants’ first visit [7]. Notably, we did not require the use of the PHQ-9 by clinicians; rather, we reported on the outcomes collected by clinicians and provided via the Headway platform. The ninth question of the PHQ-9 has been routinely used to screen for suicidal thoughts [8-10] and asks, “Over the last two weeks, how often have you been bothered by thoughts you would be better off dead or of hurting yourself in some way?” Responses are scored 0 for not at all, 1 for several days, 2 for more than half the days, and 3 for nearly every day [7].

We assigned a Social Vulnerability Index (SVI) score to each participant based on their home zip code according to the Centers for Disease Control and Prevention (CDC) [11]. The SVI is presented as a percentile from 0 (least vulnerable) to 1 (most vulnerable) that allows geographies to be directly compared based on 16 measures across 4 categories per geography: socioeconomic status, household characteristics, racial and ethnic minority status, and housing type and available transportation [11].

We compared the prevalence of SI (as defined by a nonzero response on question 9 of the PHQ-9) among patients presenting for initial outpatient care by age, sex, geography, and clinician type using univariate analyses. We used regression analysis to evaluate which of these characteristics were most associated with SI.

For geographic analyses of SI in the United States, we sought to evaluate the burden of SI in the population compared with the availability of mental health clinicians in each state. Specifically, we computed an SI Care Gap Index, which we calculated as the rate of SI in each state (from our analyses) divided by the number of mental health clinicians per population as reported by the Kaiser Family Foundation (KFF) [12]. An SI Care Gap Index of 1.0 indicates a state with average SI prevalence and average mental health workforce; values>1.0 indicate a dual burden of elevated SI prevalence and mental health workforce shortage; values approaching 2.0 indicate the most acute dual burden.

This study was determined to be exempt by the WIRB-Copernicus Group Institutional Review Board (Headway.001; July 17, 2025). We followed the cross-sectional STROBE (Strengthening the Reporting of Observational Studies in Epidemiology; Checklist 1) [13] reporting guidelines.

189,225 commercially insured individuals were included in this study. The mean age of the population of 185,913 for whom age was available was 33.60 (SD 11.33) years. Among those reporting sex, 68% (34,168/50,378) identified as female, 29% (14,491/50,378) as male, and 3% (1719/50,378) as nonbinary. Most patients presenting for outpatient mental health care had a diagnosis of anxiety (56,176/189,225, 29.7%), depression (50,075/189,225, 26.5%), or adjustment disorders (40,102/189,225, 21.2%). Nearly three-quarters of patient visits (138,259/189,225, 73%) were with a psychotherapist (50,966/189,225, 27% were with a psychiatric prescriber).

The mean PHQ-9 score among the total presenting population was 9.69 (SD 6.25; Table 1). Of these, 81.4% (154,010/189,225) reported no SI (mean PHQ-9 score 8.24, SD 5.46), while 18.6% (35,215/189,225) indicated some level of SI (mean PHQ-9 score 16.03, SD 5.50). Specifically, 13.5% (25,499/189,225) indicated SI for several days, 3.1% (5819/189,225) for more than half the days, and 2.1% (3897/189,225) nearly every day.

Across all age groups, SI was highest among teenagers (aged 13‐17 years) with 32.8% (2107/6426) reporting any suicidality and, importantly, 4.8% (306/6426) reporting SI nearly every day (P<.001; Table 1 and Figure 1). The population with the greatest SI comprised individuals identifying as nonbinary (644/1719, 37.5% vs 3129/14,491, 21.6% for male individuals vs 6,429/34,168, 18.8% for female individuals P<.001; Table 1). Notably, SI was higher among female individuals younger than 18 years but then consistently higher for male individuals older than that (Figure 2). Among patients presenting with SI, depression was the most common disorder (14,977/35,215, 42.5%), with anxiety (7411/35,215, 21%) and adjustment disorder less common (4670/35,215, 13.3%; Multimedia Appendix 1). The underlying mental health condition among suicidal patients varied by age: 11.8% (80/677) of children (aged 6‐12 years) with SI had attention-deficit/hyperactivity disorder (Multimedia Appendix 1). In contrast, posttraumatic stress disorder and stress-related disorders were the primary conditions among 9.1% (04/4453) of adults aged 45 to 64 years.

SI increased as a function of patients’ social vulnerability (P<.001; Table 1 and Multimedia Appendix 2). With increasing social vulnerability, patients had more SI and were more likely to report more frequent, including nearly daily SI (Multimedia Appendix 2). Analysis of the subcategories of social vulnerability demonstrated that SI was statistically significantly (P<.001 for all associations) associated with increased socioeconomic vulnerability (eg, poverty, unemployment, housing cost burden, and lower educational attainment) and living in at-risk households (eg, those with low English-language proficiency, single parents, and civilians with disabilities).

SI varied by geography with the District of Columbia and Rhode Island having the lowest rates of SI (163/1153, 14.1% and 29/191, 15.2%, respectively) and West Virginia and Utah having the highest (38/125, 30.4% and 274/924, 29.7%, respectively; Figure 3). When comparing the states with high rates of SI to the KFF Mental Health Care Health Professional Shortage Areas, we found that West Virginia and Alaska had the greatest SI care gap (Figure 4). Patients who sought care from a psychiatric prescriber were more likely to have SI than patients seeking care from a psychotherapist (10,893/50,966, 21.4% vs 24,322/138,259, 17.6%; P<.001; Table 1).

The present findings demonstrate that SI is not a marginal phenomenon among commercially insured outpatients but a common and clinically urgent condition, affecting nearly 1 in 5 individuals presenting for mental health care. This prevalence, observed within a population that is insured, engaged in care, and actively seeking treatment, underscores a critical reality: access to mental health services alone is insufficient to mitigate suicide risk. Instead, these data highlight missed opportunities for earlier detection, more precise risk stratification, and targeted intervention across outpatient settings. Moreover, some areas such as West Virginia and Alaska have the dual burden of elevated prevalence of suicidality coupled with shortages of mental health clinicians. These geographies urgently need increased access, including via telehealth, to effective, convenient, and affordable mental health services.

This study has 3 key findings. First, the results presented challenge the assumption that commercially insured populations represent a lower-risk group for suicidality. Much of the suicide prevention literature and public policy focus on uninsured or publicly insured populations; however, our findings indicate that a substantial burden of SI exists even among individuals with financial access to care. Given that commercial insurance covers most working-age adults and many children in the United States, failure to address suicide risk in this group limits the potential impact of national prevention strategies.

Second, the exceptionally high prevalence of SI among children and adolescents, particularly teenagers, represents one of the most concerning findings of this analysis. The rate of any SI among adolescents in this study (2107/6426, 32.8%) far exceeds prior national estimates [3,14,15] and includes a meaningful proportion reporting nearly daily SI, a marker associated with elevated short-term risk of suicide. Importantly, we observed substantial SI among children aged 6 to 12 years, a group for whom routine screening is not universally recommended [16] and for whom validated screening tools are limited. Notably, the PHQ-9 was used by mental health clinicians in this study among younger children despite not having been validated for use in this population. Given the seriousness of this finding, we urge pediatric mental health professionals to screen children presenting for care with validated instruments. Several factors may explain why the rate of SI among children observed in our study was higher than previously reported. Recent studies have demonstrated that SI in children is more common than historically recognized, challenging the longstanding assumption that young children lack the cognitive capacity for suicidal thoughts [5,17]. The COVID-19 pandemic, social isolation, increased screen time, and exposure to psychosocial stressors may be amplifying SI risk in contemporary cohorts [4,15,18]. Among the age-related findings, we hope clinicians, educators, and policymakers will note the striking finding that SI was higher among female individuals younger than 18 years but consistently higher for male individuals aged 18 years and older. Historically, trends have suggested that female individuals have higher rates of SI and attempts, while male individuals have higher rates of death by suicide (largely due to the use of more lethal means) [19]. Recently, the gap in the use of lethal means of suicide between boys and girls has decreased [20], and the sex differences in death by suicide have narrowed. This suggests that suicide prevention efforts in elementary school settings should be sex-specific: for girls, addressing the rising use of lethal methods and the social and digital risk factors such as cyberbullying and social media [20]; and for boys, improving recognition of atypical depression presentations and increasing help-seeking behaviors [21].

Additionally, increased awareness and reduced stigma around mental health may have increased the likelihood of parents and teachers recognizing a child at risk and encouraging them to present for care. Together, these findings suggest that current screening thresholds and age-based recommendations may be insufficient, especially in clinical populations already presenting for mental health care. These findings urgently underscore the need for pediatricians, school-based care clinicians, and others routinely in contact with children to actively screen both boys and girls for suicidal thoughts and make appropriate psychiatric resources readily available to them. For example, school-based mental health awareness programs such as Youth Aware of Mental Health have demonstrated approximately 50% reductions in suicide attempts in large randomized trials [18] and could serve as a model for future interventions.

Third, the observed sex patterns in suicidality further reinforce the need for nuanced prevention approaches. Higher SI rates among girls younger than 18 years, followed by higher rates among adult men, mirror the well-documented divergence between ideation and suicide mortality. This pattern suggests that prevention efforts must be developmentally tailored, emphasizing early identification and support among youth while also addressing lethal means access, substance use, and help-seeking barriers among adult men. The markedly elevated SI prevalence among nonbinary individuals (644/1719, 37.5%) highlights persistent and profound mental health inequities among sex-diverse populations and reinforces the need for affirming, inclusive models of care that extend beyond traditional diagnostic frameworks.

Several findings from this study point directly toward actionable strategies to reduce suicide risk in the United States. First, the strong and graded association between SI and social vulnerability emphasizes that suicide prevention cannot be disentangled from broader social determinants of health [4,11,14]. Patients living in areas characterized by economic instability, housing insecurity, disability, or limited English proficiency were not only more likely to endorse SI but also more likely to report frequent or near-daily ideation. These patterns argue for integrating social risk screening and navigation services (such as housing support, income assistance, and disability resources) into outpatient mental health care, rather than treating suicidality as an isolated psychiatric symptom. Future studies (especially mixed methods approaches) could be valuable in evaluating whether addressing upstream social needs, such as housing instability, financial stress, or language access to mental health services, reduces the frequency or severity of SI.

Second, the pronounced geographic variation in SI (with rates ranging from approximately 14% to 29% across states), coupled with the SI Care Gap Index, highlights the importance of place-based prevention strategies. States such as Alaska and West Virginia face a dual burden of high SI prevalence and limited access to mental health professionals, conditions that likely contribute to their persistently high suicide mortality rates [22]. These findings support prioritizing telemental health expansion, workforce incentives, and crisis-response infrastructure in high-gap states. At a national level, aligning suicide prevention funding with indicators of unmet need, rather than population size alone, may improve the efficiency and equity of prevention efforts. Moreover, education of primary care clinicians in depression recognition and management represents one of the most scalable and evidence-based suicide prevention strategies with demonstrated reduction in both suicide attempts and deaths [18], and may be particularly relevant in states with high SI care gap indices.

Third, the higher prevalence of SI among patients seeking care from psychiatric prescribers (10,893/50,966, 21.4% vs 24,322/138,259, 17.6% for psychotherapists) underscores the need for universal suicide risk assessment across all outpatient entry points, regardless of clinician type. Whether driven by self-selection, referral patterns [23], or illness severity, this finding suggests that patients at elevated risk may cluster within medication-focused care pathways. Embedding standardized suicide screening, safety planning, and rapid escalation protocols within both psychotherapy and medication management visits could help ensure that risk is identified consistently and addressed proactively.

Our finding that nearly 1 in 5 outpatients endorsed any SI is consistent with prior reports of approximately 19% suicidal-thought prevalence among outpatient populations [4], but extends that literature by demonstrating that this burden persists in a commercially insured, actively engaged population. The 32.8% (2107/6426) any-SI rate among adolescents in our study substantially exceeds national survey estimates of 13% to 22% for this age group [5,15], likely reflecting both the help-seeking nature of the sample and a contemporary increase in adolescent distress documented in recent national datasets [4,17]. The graded association with social vulnerability complements prior facility-level evidence and underscores the need for upstream interventions.

This study has 3 key limitations. First, the population consisted exclusively of commercially insured individuals, which may limit the generalizability of the findings to patients with Medicaid, Medicare, or no insurance. Second, all SI data were derived from a single PHQ-9 item, which, although widely used and clinically informative, may underestimate or overestimate risk compared with structured interviews. Although item 9 on the PHQ-9 has been shown to be a clinically meaningful, predictive marker of subsequent suicide attempts and suicide death [8-10], it does not distinguish between passive death ideation (“thoughts you would be better off dead”) and active self-injurious ideation (“hurting yourself in some way”). Therefore, we recommend that clinicians routinely follow up any nonzero response on item 9 of the PHQ-9 with a validated instrument that assesses SI and behavior, such as the Columbia-Suicide Severity Rating Scale [24]. Moreover, the PHQ-9 has not been validated for screening in children younger than 12 years [12]; thus, we recommend the use of measures, such as the Ask Suicide-Screening Questions tool, which has been validated for use in children as young as 8 years old [25]. Third, cross-sectional measurement precludes assessment of persistence or escalation of SI over time or subsequent suicide attempts. Longitudinal studies are needed to characterize the trajectories of SI over time among commercially insured outpatients to identify which subgroups (particularly children, adolescents, and socially vulnerable populations) are at greatest risk of progression from ideation to attempts. Such studies would help clarify the optimal timing and intensity of interventions.

Despite these limitations, this study provides one of the largest contemporary assessments of SI among commercially insured outpatients in the United States. The findings highlight high and unevenly distributed suicide risk, even among individuals with access to care, and identify clear opportunities for intervention at the clinical, community, and policy levels. Addressing SI earlier, more systematically, and with attention to social and geographic context may be essential to bending the trajectory of suicide rates in the United States.