Distress Tolerance and Self-Injurious Thoughts and Behaviors: A systematic review and commentary on methodological challenges

Abstract

Impaired distress tolerance (DT) is theorized to be an etiological and maintenance factor for self-injurious thoughts and behaviors (SITBs). However, the literature on DT and SITBs suffers from conflicting results and methodological challenges, and thus, is difficult to interpret.

Objectives:

We conducted a systematic review to synthesize available evidence on DT and the full range of SITBs.

Methods:

We included peer-reviewed, English-language manuscripts presenting empirical data on DT/intolerance in relation to SITB among humans. Studies were identified via a search covering results published through December 31, 2024 at Academic Search Premier, MEDLINE, PsychINFO, and PubMed (most recently executed June 20th, 2025). No formal risk of bias assessments were used. Results were synthesized by DT assessment type (self-report, behavioral, combined) and study approach (observational, longitudinal, intervention). Of the 1058 studies screened, 81 were eligible for inclusion with 30,052 participants. This review was neither funded nor preregistered.

Results:

Self-report measures appear to assess distress intolerance rather than high tolerance/endurance abilities. Findings generally indicate lower scores reflecting greater intolerance are associated with SITB across both suicide-related and non-suicidal self-injury outcomes; however, there are notable exceptions, and significant limitations associated with sample selection and research design that makes interpretation challenging.

Discussion:

The vast majority of results were obtained with a single self-report measure at one point in time. Laboratory paradigms reviewed seem to have poor ecological validity with limited relevance to the contexts in which SITBs emerge, and produced mixed associations with SITB. Within-person variability in DT has largely gone unexamined, and important mediators and moderators are underexplored. We discuss several areas for further research.

Suicide and self-inflicted injury are among our most pressing and devastating public health challenges (World Health Organization, 2025). Each year, over 5 million people present to emergency departments for intentional self-harm in the United States alone, contributing to an annual economic cost exceeding $500 billion (Bommersbach et al., 2024; Peterson et al., 2024). Unfortunately, strong and specific risk factors for self-injurious thoughts and behaviors (SITB) are extremely difficult to discern. Greater synthesis and investigation remain essential to identifying modifiable risk factors which can decrease the burden of SITBs.

The term SITB captures a broad class of interrelated yet distinct experiences pertaining to deliberate bodily harm (Millner & Nock, 2020). These include suicidal ideation (SI; i.e., thoughts about killing oneself, or a general desire to die), death by suicide, and nonfatal self-injurious behaviors (Millner & Nock, 2020). Researchers often distinguish between nonsuicidal self-injury (NSSI; acts in the absence of any intent to die) and suicide attempts (acts with at least some intent to die; Turecki & Brent, 2016). Although distinct, NSSI and suicide attempt frequently co-occur within the same individuals over time, predict one another in longitudinal studies, and share many predisposing factors (Ammerman et al., 2025; Hamza et al., 2012). Furthermore, many individuals who engage in self-injury report ambivalent, fluctuating, or conflicting intentions to die (Fartacek et al., 2024). Although incidents of self-injury can be defined across several dimensions (e.g., intent, lethality, medical outcomes), persons engaging in self-harm typically cannot be neatly categorized as “non-suicidal self-injurers” or “suicide attempters” (Kaufman et al., 2017). Thus, it is critical to examine NSSI and suicide attempts in tandem rather than in isolation.

Despite decades of research, prospectively identifying who will engage in SITB remains exceedingly difficult (Franklin et al., 2017). This is especially true for predicting the transition from ideation to suicidal behavior, including non-lethal suicide attempts and death by suicide (Belsher et al., 2019; Klonsky & May, 2015). Prior engagement in SITB remains one of our best —yet still weak— predictors of future self-harm (Franklin et al., 2017; Ribeiro et al., 2016; Vuagnat et al., 2019). Poor predictive validity stems from many factors, including: low base rates of suicide attempts and suicide, high heterogeneity among persons who engage in SITB, poor specificity of identified vulnerability factors, as well as methodological and ethical barriers to studying these acts in high-risk or ecologically valid contexts (Franklin et al., 2017; Huang et al., 2020).

Historically, researchers have also placed disproportionate focus on distal contributors to SITB and overemphasized disorder-specific predictors, by either constricting samples to specific diagnostic groups (e.g., major depression, borderline personality disorder; BPD) or examining psychiatric diagnosis as the primary predictor of self-injury. Unfortunately, this obscures important transdiagnostic mechanisms of risk that, if targeted by intervention, may yield more potent and wide-sweeping effects (Karnick et al., 2022). SITB prevalence and incidence are elevated across every psychiatric diagnosis compared to the general population (Bentley et al., 2015; Chesney et al., 2014; Orri et al., 2020). However, a sizable percentage of suicide deaths occur on first attempt (Bostwick et al., 2016; Jordan & McNiel, 2020), or without evidence of prior psychiatric diagnosis (Cavanagh et al., 2003; Stene-Larsen & Reneflot, 2019;). Thus, there is a growing call to identify risk processes that cross-cut diagnostic boundaries and better account for the observed variability in SITB presentations and trajectories.

Inability or unwillingness to tolerate distress is one underlying difficulty associated with many forms of psychopathology (Leyro et al., 2010; Zvolensky et al., 2010). Although researchers use a variety of definitions, theorists typically describe distress tolerance (DT) broadly as capturing the degree to which an individual can withstand and remain engaged with aversive experiences—particularly those characterized by negative affect and distressing cognitions (e.g., hopelessness, agitation, frustration; Furnham & Ribchester, 1995; Simons & Gaher, 2005). Scientists theorize that low DT increases persons’ use of dysfunctional or unsustainable coping strategies to relieve discomfort/suffering (Leyro et al., 2010; Zvolensky et al., 2010). Although some can adapt, others are so destabilized by distress that they disengage from larger goals and prioritize immediate relief via avoidant or impulsive behaviors that may ultimately worsen suffering. High DT is considered protective, as it preserves individuals’ capacity to engage in adaptive responses and endure inevitable negative states (there is, however, a smaller literature on the harms associated with distress overtolerance; see Lynch et al., 2011). Importantly, the conceptual distinction between distress tolerance and distress intolerance is not consistently reflected in extant measures using these terms, nor in the empirical literature base. In this review, we use the language that most clearly reflects these terms’ use within published literature (discussing limitations to this approach later in the manuscript).

Though pertinent to many adverse outcomes (e.g., executive dysfunction; Andrés et al., 2021; emotion dysregulation; Conway et al., 2021; lower self-control; Veilleux et al., 2021), impaired DT has particular relevance to SITB for a number of reasons. First, etiological research indicates persons who are most susceptible to developing self-harm are often predisposed to negative affectivity (Guerry & Prinstein, 2010; Linehan, 1993; Nicolai et al., 2016) and exposed to frequent stress and adversity (Howarth et al., 2020; Kaess et al., 2020; McLafferty et al., 2019). Thus, self-injuring individuals are especially likely to experience distress. Second, persons who engage in self-harm typically lack early training in behavior and emotion regulation (Crowell et al., 2009) and are high on impulsivity (Lockwood et al., 2017; McHugh et al., 2019) — elevating their risk for engaging in maladaptive coping. Multiple theorists conceptualize SITBs as responses that emerge in the face of overwhelming distress (Baumeister, 1990; Chapman et al., 2006; Guerry & Prinstein, 2010; Joiner, 2005; O’Connor & Kirtley, 2018; Selby et al., 2013; Shneidman, 1991). Indeed, individuals report using both NSSI and suicidal thoughts as means of escaping emotional pain (Coppersmith et al., 2023; Kleiman et al., 2018; May et al., 2016; Taylor et al., 2018), and poor DT appears to facilitate self-harm even after controlling for acute negative affect (Anestis et al., 2013a; Chapman et al., 2006; Nock & Mendes, 2008). DT is often conceptualized as intentional, effortful use of emotion regulation or other strategies in order to persist through or allow (tolerate) acute distress (Linehan, 1993), and thus may buffer against SITB risk at multiple stages of ideation-to-action frameworks. For example, when faced with negative life events or stressors, DT may facilitate coping that prevents the transition to suicidal thinking. During a period of acute crisis, DT may allow an individual to persist through that distress without acting on suicidal or self-harm urges, interrupting the ideation-to-action transition. Taken together, individuals with SITB may encounter more frequent and intense aversive experiences, display heightened sensitivity to such experiences, and lack the necessary tools to manage their ensuing distress. As a result, leading interventions for SITB such as Dialectical Behavior Therapy (DBT; Linehan, 1993) aim to improve DT by offering targeted skills to tolerate and survive intensely aversive experiences without engaging in SITB.

Despite considerable interest, significant gaps remain in our understanding of DT and SITB, and the available literature suffers methodological limitations. As we would expect, some prior research finds that distress intolerance is associated with greater risk of SITB — particularly NSSI (Akbari et al., 2024). However, other studies suggest paradoxical effects whereby high DT scores are associated with elevated risk of self-harm in certain contexts (especially when DT reinforces self-punishment or prevents adaptive help-seeking; Chung et al., 2025). These mixed findings are further complicated by substantial variability in how researchers define and measure distress intolerance/tolerance, inadequately characterized or selected SITB samples, and a dearth of research examining developmental or causal effects (Ameral et al., 2014).

Systematic reviews summarize a vast literature, synthesize complex and contradictory findings, help identify research barriers, and guide future research efforts (Page et al., 2021). One recent meta-analysis examined DT in relation to NSSI (finding small negative effects; Akbari et al., 2024), yet no comprehensive review has systematically examined DT across the full range of SITB. To take stock of the state of the science and provide recommendations for the future, we have conducted a systematic review of the empirical literature on DT and SITB. We considered whether results would be more clearly summarized using a meta-analytic framework, yet elected to retain our approach given the varied conceptualizations of DT, diverse measures used and samples included, and because of the distinct ways that SITB outcomes were captured. In the sections below, we present results from studies examining distress intolerance through self-report, those using behavioral DT measures, and results from mixed-method designs. Next, we present key theories cited as informing the included literature and discuss evidence related to DT-focused intervention studies. We then highlight challenges affecting the SITB and DT literature, noting strengths and limitations of common DT definitions, measurement tools, and study designs. Finally, we present guidelines for future work in these critical domains.

Method

Study Selection

We performed a literature search in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Page et al., 2021). Studies investigating the relation between SITB and DT/intolerance were identified via Academic Search Premier (January 1, 2001– December 31, 2024), MEDLINE Complete (January 1, 1860 – December 31, 2024), PsycINFO (January 1, 1860 – December 31, 2024), and PubMed (January 1, 1975 – December 31, 2024). Search terms were: (Distress tol* OR Distress intol*); (Frustration tol* OR Frustration intol*); (Mental Pain tol* OR Mental Pain intol*); (Psychological Pain tol* OR Psychological Pain intol*); (Psychache tol* OR Psychache intol*) AND (suicid* OR self-injur* OR selfharm* OR self-mutilation). We did not include DT terms like “uncertainty” or “ambiguity” as these have greater relevance to the anxiety literature than to established SITB theories. Reports such as unpublished manuscripts and conference abstracts were excluded.

After removing duplicate articles, two post-baccalaureate level research assistants screened N = 1,058 study titles and abstracts for any mention of SITB (including one or more of the following: suicidal ideation, death by suicide, and/or non-fatal self-injurious behaviors) and DT or distress intolerance (DI; see Figure 1). If uncertain about article relevancy at this stage, research assistants consulted with the second author and conservatively retained articles for full-text review. Relevant full-length articles (n = 128) were equally divided across the two research assistants and independently evaluated for eligibility. Research assistants reviewed each of their assigned studies multiple times to extract relevant information. Data extracted from full length articles for descriptive and analytic purposes include: date of publication, study aims and hypotheses, sample size and population, study inclusion and exclusion criteria, operationalization of DT/DI, tools used to measure DT/DI and SITB, theoretical framework(s) that informed the included studies, study design, covariates, primary and secondary study results and effect sizes, and study limitations.

Figure 1: Flow Chart for Study Inclusion.

^aRecords were excluded if there was no mention in the title/abstract of both (1) suicidality/SITB/self-harm, and (2) distress (in)tolerance/frustration (in)tolerance/psychache (in)tolerance/mental pain (in)tolerance/psychological pain (in)tolerance/pain persistence

The first and second authors reviewed extracted information from individual studies to confirm eligibility and met jointly with research assistants on multiple occasions to ask clarifying questions, request additional information, and discuss article suitability. We included studies that: (a) investigated DT/DI in relation to SITB among human participants, (b) were published in peer reviewed journals, (c) presented empirical data, (d) and were written in English. After we finalized study selection, the first author synthesized results by DT assessment type (self-report, behavioral, combined) and study approach (observational, longitudinal, intervention) to identify key patterns in the available literature. These were confirmed and expanded upon by the second and third authors.

Results

Of the 1058 studies screened, 81 were eligible for inclusion (see Figure 1). Only 15 studies indexed DT via laboratory paradigms (five of which also included a self-report instrument), whereas 63 studies indexed DT/intolerance exclusively via self-report. Three studies did not use any quantitative DT measure, however DT was integral to the empirical investigation (two studies focused on DT treatment, and one used qualitative methods; Denckla et al., 2015; Goyes, 2024; Martin & Del-Monte, 2022). We summarize the included studies’ key findings in Tables 1–4 of the online supplement; organized by population and type of measure to ease readability.

Self-Report Findings

The Distress Tolerance Scale

The vast majority of reviewed studies relied on a single self-report instrument as their sole measure of DT: the Distress Tolerance Scale (DTS; Simons & Gaher, 2005; n = 51 studies or 62.30%). Among these, most studies were cross-sectional (n = 41 or 80.40%), with a smaller number featuring longitudinal and experimental methods (n = 8 or 15.69%). Samples included undergraduates (n = 19), clinical or treatment-seeking individuals (n = 17), community members (n =7), high school students (n = 5), and veteran/military individuals (n = 3).

The DTS asks respondents to rate their level of agreement with 15 Likert-style items (1= Strongly agree to 5 = Strongly disagree). Four subscales load onto one higher order factor “Distress Tolerance.” Subscales are intended to capture respondents’ perceived ability to tolerate emotional distress (Tolerance; e.g., “Feeling distressed or upset is unbearable to me”), subjective appraisal of distress (Appraisal; e.g., “I can tolerate being distressed or upset as well as most people”), the degree to which their attention is absorbed by distress (Absorption; e.g., “When I feel distressed or upset, all I can think about is how bad I feel), and efforts to regulate distress (Regulate; e.g., “I’ll do anything to stop feeling distressed or upset”). All but one item are reverse-scored, with higher scores indicating greater DT. The most consistent pattern of findings across studies was that lower DTS scores were associated with greater risk of SITB. However, notable variability emerged in the strength and direction of these effects depending on the SITB outcome, methodological characteristics, and inclusion of mediators and moderators.

NSSI findings with the DTS.

Several studies found lower DTS scores were associated with more frequent or severe NSSI, both cross-sectionally (Horgan & Martin, 2015; Kim et al., 2020; Slabbert et al., 2020), and longitudinally (Kang et al., 2018; Lin et al., 2018). One manuscript reported that higher DT scores differentiate individuals who had ceased NSSI in the past year from those who hadn’t, controlling for duration of self-injury history (Kim & Hur, 2023). This finding aligns with theory that developing DT skills may contribute to or reflect broader improvements in coping. However, since this study was cross-sectional, it remains unclear whether DT facilitates NSSI cessation or merely accompanies it. The direction of the effect is also unclear, as it’s possible that ceasing NSSI may change participants’ self-appraisals of how competent they are tolerating distress (yielding improved DTS scores).

Other studies explored how DT may operate alongside other risk factors to predict NSSI engagement. Several investigated DT in mediation models linking distal risk factors to NSSI. Three cross-sectional studies found associations between retrospective reports of early adversity and NSSI through low DT (Erol & Inozu, 2024; Ghorbaninejad & Manesh, 2023; Kang et al., 2018). One longitudinal follow-up study found that depressive symptoms mediated the relation between baseline DT and later NSSI (Lin et al., 2018). However, the authors did not control for baseline depression in their model, making it difficult to determine if DT is a true predictor of later MDD (over and above baseline MDD), or if the result is an artifact. Other investigators examined DTS scores as one of multiple moderators of NSSI (alongside variables like negative urgency, emotion dysregulation, affect intensity, and rumination), with most finding stronger evidence for complex interactive, rather than main, effects (Peterson et al., 2014; Slabbert et al., 2018; 2020). These findings generally align with conceptualizations of DT as a regulatory mechanism that becomes relevant in the context of emotional distress, with low DT amplifying risk when paired with other emotional vulnerabilities. One exception was that Slabbert and colleagues (2018) found rumination predicted NSSI at high levels of DT, suggesting multiple pathways to risk. Unfortunately, most studies in this review were cross-sectional.

Suicide-related findings with the DTS.

Studies focusing on suicide-related outcomes present a more complex picture, with different patterns of association for suicidal ideation and suicidal behaviors. Among those examining suicidal ideation as their outcome, 14 studies found lower DTS scores corresponded with greater suicidal ideation severity or frequency. This pattern held across samples of different ages and with varying degrees of risk (e.g., outpatient vs. community-based samples). A handful of cross-sectional studies showed DTS scores may help explain links between suicidal ideation and other distal risk factors like sleep disturbances, child maltreatment, and posttraumatic stress symptoms (Bartlett et al., 2021; McGrew et al., 2023; Zemestani et al., 2023). Two additional studies examined adolescent self- and parent-reported effects among youth with anxiety (O’Neil Rodriguez & Kendall, 2014) and tic disorders (Johnco et al., 2016). Both research teams found associations between adolescent self-reported suicidal ideation and low DTS scores. However, parent report of youth DT did not differentiate or predict adolescent suicidal ideation. These findings highlight that DT struggles may be experienced privately and support the importance of obtaining self-report. However, results may also be influenced by statistical artifacts, as self-report measures are typically more correlated with one another than a self- and informant-report, all else being equal. Furthermore, adolescent suicidal ideation ratings were dimensional in the Johnco et al. (2016) manuscript, whereas parent report was assessed via a binary item (and thus less powered to detect an effect for parent-reported SI).

One noteworthy result emerged in a manuscript by Gorraiz et al. (2024). The authors found higher reported reasons for living were associated with lower DTS scores among adolescents on an inpatient unit. This counterintuitive pattern may reflect differences in the types of coping resources that youth draw upon. Whereas reasons for living may reflect motivational anchors that keep people alive (even in the context of low DT), DT captures persons’ perceived capacity to endure distress. Important reasons for living may or may not accompany greater DT, highlighting the utility of examining DT alongside other protective and risk processes.

Many studies examining suicidal behaviors showed a similar protective pattern whereby higher DTS scores were associated with lower risk (Capron et al., 2013; Interian et al., 2021; Naifeh et al., 2022; Stanley et al., 2018; Vujanovic et al., 2017a). However, a notable subset of research papers found the opposite: higher DTS scores were associated with greater likelihood of suicidal behaviors (Anestis et al., 2012b; 2013a; Rogers & Joiner, 2018). Some researchers have suggested that higher DT may facilitate engagement in lethal behaviors for certain individuals through a risk process called acquired capability (Anestis et al., 2014). According to the Interpersonal Theory of Suicide, a person’s capacity to enact lethal self-harm is acquired via repeated exposure to painful or provocative experiences (Joiner, 2005). Five studies in this review reported direct or indirect associations between higher DTS scores and acquired capability, suggesting that higher DT was associated with risk for some and protection for others (Anestis et al., 2011b; 2014; Anestis & Joiner, 2012; Bender et al., 2012; Martin et al., 2019).

Anestis and colleagues have examined whether DTS scores could help explain the pathway from NSSI to suicide attempt. In one study, low DTS scores were linked to greater lifetime suicide attempts via NSSI (Anestis et al., 2013b). In another, the authors found the relation between NSSI frequency and suicide risk history was stronger among individuals with higher DTS scores (Anestis et al., 2013a). These studies illustrate different approaches to examining DT from an acquired capability lens. The former considered low DT as a potential driver of the NSSI to suicide risk pathway, whereas the latter suggested high DT may influence suicide risk among those who already engage in self-injury. Since both studies were cross-sectional and relied on lifetime measures of suicide risk, it remains unclear when and for whom DT may meaningfully influence risk. Other limitations include the use of restricted (typically convenience) samples, most evidence being collected from the same research group, and the often modest effect sizes reported overall (i.e., typically small standardized beta coefficients, and small-medium correlation coefficients, Cohen’s d values, and odds ratios).

The acquired capability interpretation of the greater DT corresponding to higher suicide risk finding hinges on DT corresponding to high physical pain tolerance. Yet this has rarely been tested. One study found a small effect for higher DTS scores being associated with higher physical pain tolerance, but no significant relation between DT and acquired capability (Anestis et al., 2011a). Furthermore, acquired capability is typically assessed using the Painful and Provocative Life events scale (Bender et al., 2012), which captures experiences that only loosely relate to risk profiles suggesting lethal self-injury capability (i.e., getting a tattoo, skydiving, adverse experiences). It is also possible that DT facilitates suicidal behavior through toleratance of other types of distress associated with making a suicide attempt (e.g., tolerating fears of death or concerns about causing distress and harm to loved ones), or that high DTS scores reflect emotional resignation or numbness rather than a true, adaptive tolerance of negative affect.

Other DT Self-Report Scales

Although the majority of the reviewed studies used the DTS, researchers have employed other DT self-report measures (see online supplement Tables 1–4). Thirteen studies using the Tolerance of Mental Pain Scale (TMPS) demonstrated that scores indicative of better DT were associated with lower odds of, or less frequent or severe, SITB (for suicidal ideation: Becker et al., 2019; Landi et al., 2020; 2023; Martins et al., 2022; Passos et al., 2025; Yeşiloğlu et al., 2023; and for suicide attempts: Demirkol et al., 2019; Levinger & Holden, 2014; Levinger et al., 2015a; 2015b; Meerwijk et al., 2019; Meerwijk & Weiss, 2018; Soumani et al., 2011). The TMPS distinguishes between managing pain and enduring pain; and the former emerged as a more consistent buffer against suicidal ideation (i.e., actively managing psychological distress rather than merely endure it; Landi et al., 2020; Martins et al., 2022; Passos et al., 2025). Of the remaining scales, scores reflecting higher DI were associated with elevated suicidal ideation and NSSI across all studies (Conway et al., 2021; Denckla et al., 2015; McGrew et al., 2023; Yelpaze, 2023).

Longitudinal Findings

Eight articles in our review reported on longitudinal studies. Multiple manuscripts examined DT as a mediator linking another known risk factor to SITB. For example, Kang et al. (2018) and Wu et al. (2023), each conducted large-scale multiwave self-report studies of high school students in Hong Kong. In both studies, DI itself predicted later NSSI (Kang et al., 2018) or suicidal ideation (Wu et al., 2023), with DI being predicted by childhood abuse or emotional reactivity, respectively. Another study positioned DT as the predictor of NSSI one year later among Taiwanese high school students, and found this relation was mediated by depressive symptoms (Lin et al., 2018). While the longitudinal design strengthens each of these studies, causal relations cannot be determined and shared method variance from relying on self-report data may have inflated the observed associations. Furthermore, although these studies recruited large samples, a relatively small percentage of each endorsed SITB. Despite these limitations, results support DT/DI as related to and even predictive of future SITB among youth.

One additional longitudinal study included measures of both DT and SI; however, the authors did not examine these constructs within the same models (which tested the latent structure of anxiety sensitivity, DT, and psychiatric symptoms following exposure to a natural disaster; Hashoul-Andary et al., 2016). In general, our review indicates that longitudinal research is sparse (particularly mechanistic and intensive repeated-measures studies). Most did not control for baseline SITB; thus, even with the temporal ordering of DT preceding SITB measurement, it is unclear whether the effect is causal or whether it is driven by the un-modeled auto-regressive effects of SITB over time. Thus, these findings cannot directly speak to emergence or maintenance of self-injury as a behavior that results from low DT, as theory would suggest.

Findings from Behavioral DT Tasks

Behavioral measures of DT were initially developed in an attempt to predict substance use treatment dropout, but have since become widely used to mark objective ability to withstand aversive states (e.g., Daughters et al., 2005). These paradigms commonly induce sustained frustration or discomfort via exposure to potent, aversive stimuli such as a loud noise, intense cold, or an unsolvable problem. Participants persist in these tasks until they choose to quit and their time in seconds is typically used as a proxy for DT, with longer latency reflecting greater ability. Our review yielded fifteen studies (18.3%) that examined links between SITB and DT using behavioral measures. The most commonly used paradigms were the Paced Auditory Serial Addition Task Computerized (PASAT-C; Lejuez et al., 2003), the Distress Tolerance Test (DTT; Nock & Mendes, 2008), and the Mirror Tracing Persistence Task (MTPT; Strong et al., 2003).

Findings with the PASAT-C

The PASAT-C is a computerized addition task that presents respondents with single-digit numbers in rapid succession, which the participant must add together (Lejuez et al., 2003). Loud, harsh noises signal incorrect responses. Over time, the task increases in difficulty and culminates in a final phase where participants can discontinue at any point. Validity of the PASAT-C is typically established by examining relations between task performance and outcomes thought to reflect poor DT (e.g., dropout from mental health treatment programs, poor resistance from substance use urges, BPD symptoms, emotion dysregulation; Brown et al., 2009; Daughters et al., 2005; Gratz et al., 2010). The PASAT-C has reliably induced negative affective states across several studies, with larger increases in frustration accompanying earlier quitting (Ameral et al., 2014). In all but one of the five PASAT-C studies included in our review, longer persistence (indicative of higher DT) was linked to SITB (suicidal behavior: Anestis et al., 2012a; 2012b; Viana et al., 2018; suicidal ideation: Viana et al., 2019). Each recruited participants from inpatient or medical-residential settings and examined DT alongside other clinical experiences (i.e., BPD and PTSD symptoms, emotional clarity).

Findings with the DTT

The DTT is a modified version of the Wisconsin Card Sort Test that induces frustration by requiring participants to sort cards with ambiguous instructions (Heaton et al., 1993). Feedback is delivered using a pre-programmed pattern of “correct/incorrect” responses¹. Participants may quit at any point after completing 20 mandatory trials. DT is operationalized as the total number of cards sorted, with fewer operationalized as lower DT. Of the seven studies using the DTT, two employed an enhanced version (DTT-E) that induces and measures physical pain tolerance through applying continuous pressure (Anestis & Capron, 2016; Law et al., 2017).

Studies using the standard DTT yielded mixed associations with SITB. Two found that individuals with a history of NSSI or current suicidal ideation quit earlier than control participants (Barzilai et al., 2024; Nock & Mendes, 2008). Importantly, these effects held even when participants with a NSSI history reported greater distress or showed stronger physiological reactivity—suggesting that low DT reflected difficulty enduring distress rather than these participants experiencing less distress itself (Barzilai et al., 2024; Nock & Mendes, 2008). Other studies have found the opposite pattern. Although some studies showed positive associations between behavioral persistence, acquired capability scores, and past exposure to painful life events (Anestis & Joiner, 2012; Bender et al., 2012), only one study directly examined the relation between task persistence and SITB outcomes. Barzilai et al. (2024) found that individuals with higher suicidal ideation severity persisted longer on an interpersonal distress task (Cyberball), but not on a cognitive distress task (DTT), suggesting that the type of distress or context may matter more than general tolerance ability. This differential response to interpersonal versus cognitive distress challenges assumptions about DT as a unitary construct. Results also raise questions about whether persistence through social rejection represents adaptive tolerance, or a maladaptive decision to remain in harmful interpersonal situations. One study reported no group differences in behavioral DT among participants with and without NSSI, though higher task distress was observed among those with past NSSI (Tuna & Gençöz, 2021).

DT Given that the DTT-E combines affective and physical pain tolerance, this task may tap a broader construct than emotional DT alone, or perhaps multiple constructs. Results align with evidence that individuals who have engaged in self-injury show higher thresholds for experiencing and tolerating physical pain (Hooley & Franklin, 2018; McCoy et al., 2010; Koenig et al., 2016; Risch et al., 2024). The pain induction component of the task may have also functioned to facilitate affect regulation among participants with a history of NSSI or suicidal behavior.

Findings with the MTPT

The MTPT (Quinn et al., 1996) is a computerized paradigm that requires participants to trace a figure using a mirrored cursor that moves in the opposite direction than expected. A loud buzzing sound is emitted when participants deviate from the figure they are tracing. Errors typically become more frequent as shapes increase in complexity. As with the other behavioral tasks reviewed above, longer time spent in the task is argued to signify greater DT.

Only two studies included in this review used the MTPT to assess DT. One recent study of undergraduate students found self-reported DT was associated with self-injury history (NSSI and suicide attempt) as assessed on the Personality Assessment Inventory–Borderline Features Self-Harm subscale, yet task performance was not (Conway et al., 2019). The second study found lower MTPT persistence was associated with a higher number of past suicide attempts among inpatient participants (Vujanovic et al., 2017). However, although self-reported DT was associated with suicidality as the reason for hospital admission, persistence on the MTPT was not. These results underscore a recurring theme in our review: behavioral DT measures may lack the sensitivity or potency needed to capture clinically relevant forms of distress to SITB (Hsu et al., 2023).

Findings from Multimethod Studies

Across the six studies that use both a behavioral task and self-report measures to assess DT, measurement tools appear weakly if at all correlated with one another. Findings were mixed regarding which type of assessment associates with SITB outcomes, yet self-reported DT appears to be more consistently linked with past and current SITB relative to task persistence (Conway et al., 2019; O’Neil Rodriguez & Kendall, 2013; Tuna & Gençöz, 2021). One study suggested that self-reported DT may be a more meaningful marker of current or recent suicidal ideation, whereas behaviourally-indexed DT (i.e., on the MTPT) was associated with lifetime suicide attempt history (Vujanovic et al., 2017b). From this perspective, task persistence may be more relevant to suicidal and self-injurious behaviors as opposed to cognitive indices like suicidal ideation.

Although studies using behavioral measures of DT typically examined these indices in relation to recent or lifetime SITB, one study examined short-term changes in risk over time and suggested a distinct pattern of findings. In one EMA study, poorer persistence on the PASAT-C was a stronger predictor of NSSI across the following week than momentary affect (Ammerman et al., 2017). These results align with hypotheses that low DT increases vulnerability to self-injurious behaviors — specifically in acutely distressing contexts. Thus, while some evidence suggests behavioral DT is less closely tied to current or past SITB than self-report DT, other work indicated task scores may index proximal risk for moving from self-injurious thoughts to behaviors under ecologically valid conditions. Other DT tests indexing persistence through physical pain or aversion (i.e., breath holding, cold pressor task) were also unrelated to self-reported DT and other behavioral tasks, and poorly predicted SITB outcomes (Anestis & Capron, 2016; Vujanovic et al., 2017a). This underscores that DT is multifaceted and our conceptualization of the construct in the context of SITB may be more affective or cognitive, as opposed to physical.

Taken together, extant work on DT and SITB broadly supports an association of low DT with higher incidence or severity of SITB, with notable variation depending on sample type, study methodology, and SITB outcome explored. Theoretical variation in conceptualization and measurement of DT/DI likely also played a role in these varied patterns and effects.

DT-Focused Intervention Studies

Four longitudinal treatment trials were reviewed and generally indicate that teaching DT skills is associated with reductions in SITB. One study presented findings from a cognitive-behavioral treatment trial with a DT skills component (Martin & Del-Monte, 2022). The authors found reductions in SITB among persons with BPD following the intervention, yet did not measure participants’ DT directly or establish whether this was the component of the intervention that yielded beneficial effects. Similarly, Denckla et al. (2015) investigated the effects of teaching DT-only DBT skills to 8 military veterans and found fewer suicide related crisis events in the year post-treatment as compared to the year prior to treatment (i.e., suicide-related emergency room visits, psychiatric visits, and suicide events). However, the authors did not include a DT measure in their study.

Another longitudinal treatment trial administered eight core DT skills from DBT to inpatients with a history of SITB (Booth et al., 2014). The authors found self-reported difficulties with DT and self-harm behavior significantly decreased pre- to post-group with large effect sizes, with gains being maintained at 3-month follow-up. Finally, McMain et al. (2017) conducted a randomized trial comparing brief DBT skills training to waitlist among suicidal patients with BPD. As expected, they found greater improvements in the DBT group compared to waitlist participants on self-reported DT, emotion regulation, anger, as well as frequency of suicidal and self-harm behaviors at 3 month follow-up. Results from our review are consistent with the broader treatment literature indicating changes in self-reported distress intolerance declines as a function of treatment across modalities (see Heiland & Veilleux, 2024 for a review).

Theoretical Frameworks Informing the Literature

Several theoretical frameworks were cited as informing studies included in this review (see online supplement Table 5). Although summarizing each theory is beyond the scope of this paper, we provide a brief description of the most frequently referenced key models below. The most commonly cited framework was Joiner’s interpersonal theory of suicide (IPTS; Joiner, 2005), which claims that SITB emerges from a combination of individuals’ desire to die and their acquired capability to do so. Proximal interpersonal factors contributing to persons’ desire to die (suicidal ideation) are thwarted belongingness (painful social disconnection that arises when a person’s basic needs for belonging go unmet) and perceived burdensomeness (feeling that others would be better off without me; Joiner, 2005). According to IPTS, individuals experiencing suicidal ideation must also overcome self-preservation reflexes to carry out a high lethality suicide attempt. This acquired capability emerges when an individual habituates to physical pain (e.g., through low-lethality self-injury) and sufficiently reduces their fear of death or bodily harm (e.g., through handling weapons). Most articles in our review that referenced IPTS were exploring DT in the context of acquired capability – often with respect to physical pain tolerance.

Linehan’s biosocial model (1993) was the next most commonly referenced theory. This etiological framework provides a testable theory for how emotion dysregulation, self-injury, and BPD develop. Linehan and colleagues propose that emotion dysregulation emerges from an interaction between emotionally invalidating environments and biological vulnerabilities to negative affectivity and trait impulsivity (Linehan, 1993; Crowell et al., 2009). According to Linehan, individuals who are at an increased risk of self-injury tend to experience more intense and frequent negative affect, and are slower to return to baseline after emotion is piqued. When combined with contexts that ignore, minimize, or reject emotional expressions and intermittently reinforce extreme emotional displays, behavior and emotion dyscontrol become entrenched. Over time, SITB may emerge to reduce distress (Nock, 2009; Nock & Mendes, 2008).

Two other theories were cited with some regularity: the emotional cascade model (Selby et al., 2009) and Shneidman’s theory of psychache (Shneidman & Farberow, 1956). The former argues that persons with BPD engage in self-injury to distract from a cycle where intense negative emotions and repetitive negative thinking reciprocally reinforce each other. Individuals with low DT are thought to engage in self-injury to escape aversive states. Shneidman theorizes that psychological pain (including lack of meaning in life, hopelessness, anguish, and psychological distress) leads to internal conflicts and suffering, which can become unbearable and result in suicide. Presumably, less tolerance for and higher sensitivity to these forms of pain are at least partially responsible for increased risk. A number of other important theories were less frequently cited as informing manuscripts included in our review (e.g., the experiential avoidance model [Chapman et al., 2006], three-step theory [Klonsky & May, 2015]; integrated theoretical model of the development and maintenance of self-injury [Nock, 2010]; see online supplement Table 5). Many share a common focus on expansive or undercontrolled negative internal states (even when prompted by external inputs) as a primary driver of SITB.

Critical Evaluation of Key Themes

A number of themes emerged from our review of the literature. Below, we discuss a series of challenges impacting our understanding of DT and SITB, broadly organized into measurement difficulties, conceptual challenges, and methodological considerations.

Measurement Difficulties

Limited Convergence Across Measures

Performance on behavioral DT measures does not generally correspond to self-reported DT (Anestis & Joiner, 2012; Veilleux, 2023; Vujanovic et al., 2017b). Some argue these indices assess distinct dimensions of DT, or capture separate constructs altogether (Ameral et al., 2014; Veilleux, 2023). Whereas self-reported DT taps a (often very general) self-appraisal of one’s capacity to manage emotional distress, DT tasks measure persistence through negative affect within a very specific laboratory context. It is unsurprising that performance on circumscribed tasks like the PASAT-C, DTT, and MTPT may not map onto one’s perceived ability to tolerate real-world distress in day-to-day life. We generally lack a framework for understanding what behavioral persistence on such tasks reveals about SITB risk above and beyond self-report.

Poor Ecological Validity

A related limitation that further complicates study interpretation and generalizability is that most DT measures fail to capture the contexts where SITB actually emerges. Self-report measures assess perceived “general” deficits in DT and fail to parse when DI emerges (discussed further below). Participants who experience SITB may struggle to respond to items assessing overall DT capacity when one’s perceived DT can vary considerably by context. Behavioral tasks use artificial stimuli like loud noises that are unlikely to provoke the type or intensity of distress most relevant to real-world SITB. Key theories of SITB development and maintenance instead emphasize emotional distress and psychosocial stressors like psychological pain, hopelessness, and perceived burdensomeness (Bentley et al., 2016; Klonsky & May, 2015). It is probable that the PASAT-C, DTT, and MTPT succeed in inducing frustration; and these tasks may be able to detect gross deficits in behavioral persistence through irritation. However, the type, intensity, and salience of distress is unlikely to readily generalize to outcomes like SITB, especially when participants with suicidal ideation respond differently to interpersonal versus cognitive distress tasks (Barzilai et al., 2024). Unfortunately, no current DT paradigms mirror the interpersonal conflicts, emotional pain, or life stressors that typically precede or potentiate SITB. Behavioral tasks may also fail to prompt the intense hopelessness and desire to escape that often accompanies SITB, since participants are explicitly told that they may choose to end the task. Laboratory paradigms with higher ecological validity are sorely needed; there are existing paradigms that elicit socially salient distress that could be used to study these experiences (e.g., Collins et al., 2017).

Limitations of the DTS

Self-report is the dominant DT assessment format in the SITB literature, with the majority of reviewed studies administering the DTS as their sole index of this construct. Although widespread use of the same instrument facilitates straightforward comparison across studies, overreliance on any single measure to assess the role of DT/DI in SITB leaves the empirical literature vulnerable to the limitations of the specific tool in question.

Instructions from the original instrument read “Think of times that you feel distressed or upset. Select [the response] that best describes your beliefs about feeling distressed or upset.” Other versions of the scale add, “Please answer regarding your feelings of distress ‘in general’, that is, on average.” Both the original and altered wording are consistent with the assumption that DT is a stable, trait-like factor. Although there are likely individual differences in DT habits, capacity to adaptively engage with or persevere through goal-directed activity in the face of distress is also contextually situated (i.e., variable depending on the nature, intensity, and chronicity of the distress, momentary access to resources, and/or interactions with additional stressors; see Veilleux, 2023 for an articulation of a momentary model of DT). The DTS does not assess DT contexts, variability in DT performance, objective ability, the frequency or time course by which distress is experienced, or how specific DT strategies are employed.

Further, although the questionnaire is titled the Distress Tolerance Scale, only one of the 15 items is worded to align with greater perceived ability (item 6: “ I can tolerate being distressed or upset as well as most people”). The remaining 14 reverse-scored items are worded such that greater agreement with the statements indicate more perceived difficulties tolerating upset. There are a number of potential problems with this. First, the scale may be vulnerable to method effects and response-style biases such as acquiescence effects. Experts in test development also argue it is often unclear what different levels of disagreement mean (leading to difficulties interpreting what respondents intend when marking “disagree” relative to “strongly disagree”; Barchard & Russell, 2020). There may be multiple reasons for disagreement that have relatively little to do with one another, as the “opposite” of the original statement can be unclear². The predominance of negatively framed items on the DTS means that DT ability is only defined by the absence of intolerance as assessed by this tool. DTS scores are almost certainly affected by inadequate domain coverage, as competency in DT is characterized by more than the absence of deficits.

This is important as many studies we reviewed discussed findings related to “high” DT as assessed by the DTS, yet resilience or aptness for tolerating distress is not captured well by the scale. By only measuring a restricted portion of the full range of functioning, we are unable to identify a point at which DT becomes impaired or has particular relevance to SITB outcomes. Furthermore, we cannot detect what degree of DT ability may be associated with SITB resilience. A person who is adequate, but not particularly skilled, at DT may score similarly on the DTS to a person with exceptional ability. This artificial ceiling effect is likely to obscure meaningful variation in DT, inflate the likelihood of nonsignificant findings, and pose a particular challenge for interpreting mixed results.

Some authors interpret positive associations between DTS scores and SITB outcomes as evidence that high DT increases SITB risk via acquired capability, and have even referred to elevated DT on this scale as reflecting a phenomenon described as “overtolerance” (e.g., Akbari et al., 2024; Anestis et al., 2013b). However, there are multiple problems with this interpretation. First, as other scholars have articulated (Veilleux, 2023), overtolerance refers to persons repeatedly enduring distress or negative affect itself to the point of harmful consequences. SITB is typically enacted to avoid or alleviate negative affect rather than sustain it. Second, as discussed above, higher DTS scores capture a range from moderate to high DT ability, compressed into a relatively small range of scores on the scale. Thus, interpreting high DTS scores as overtolerance does not align well with the acquired capability hypothesis.

Another challenge arising from capturing DT via self-report is difficulties in understanding how participants arrive at their responses, and whether responses are influenced by other constructs being examined. For example, some DTS items explicitly ask participants to rate themselves in comparison to “most people.” However, this reference point will differ by respondent, context, and culture. Mental distress, mental illness, and SITBs are not randomly distributed in the population. Persons who experience SITBs may have a high proportion of close connections with others who are struggling with DT; and their view of what “most people” do when in distress will be influenced by their social group. There are also potential feedback loops between SITBs themselves and perceived DT that have not been explored. For instance, some individuals may perceive themselves as having high DT capacity because they see their NSSI as an emotion regulation strategy; conversely, individuals who feel significant shame or guilt surrounding their NSSI may view this behavior as evidence of their own “poor” DT capacity.

Finally, it is noteworthy that two of the three DTS “regulation” subscale items only differ by a single word: “I’ll do anything to [avoid/stop] feeling distressed or upset.” The third statement reads, “When I feel distressed or upset, I must do something about it immediately.” These items seem to tap urgency to avoidant distress rather than regulation strategies.

Conceptual Challenges

Inconsistent Definitions

How scientists define distress and the process of tolerating it impacts study design, sample selection, measurement choices, and their interpretation of findings. One prominent theme that emerged from our review is that research groups operate from inconsistent conceptualizations of DT/DI.³ We found (n = 18) articles defining DT as a person’s perceived capacity to withstand, accept, or function in the context of aversive thoughts, emotions, or “psychological states.” Although likely important for perceived self-efficacy and willingness to deal with distress, persons may not always be skilled at assessing their own DT ability. Other authors include behavioral tendencies or demonstrated ability in their definitions (n = 16 articles; e.g., actual persistence or perseverance through distress). Several groups expand their definitions still further to include tolerating, functioning in, and/or persisting through physical discomfort or pain (n = 10 articles). This definition seems appropriate for studying SITB in the context of chronic pain; however none of the studies we reviewed examined this population. For many with SITB, inducing physical pain is the response to distress rather than the cue.

On the other hand, a number of articles (n = 13) restricted their conceptualization of DT/DI to the context of negative affective states. This definition implies that distress is not inherently driven by exposure to a particular stressor or situation, but is rather filtered through the individual’s negative affect. Perhaps as a consequence, very few studies measure the context that induces distress preceding SITB or how situational factors may help distinguish why the same person may tolerate distress one day and engage in SITB the next. Eight articles defined DT as the ability to withstand mental pain/psychache and 16 did not provide a definition.

Overlap with Emotion Regulation

Some researchers conceptualize DT as an aspect or component of emotion regulation (e.g., Kim et al., 2020), whereas others see it as distinct. For example, from a DBT treatment framework, DT skills help clients survive a crisis without worsening their outcomes⁴; in contrast, emotion regulation skills help the client identify, understand, and communicate about their emotions, and influence their frequency, intensity, and duration. Some argue that DBT DT skills are not particularly generalizable to all persons affected by DI, as not all distress reflects a response to a crisis, and not all persons engage in self-damaging behaviors when in distress (Veilleux, 2023). Further, some preliminary data indicate that DBT emotion regulation skills improve self-reported DT (Dixon-Gordon et al., 2015)—consistent with there being some degree of overlap between the processes that support emotion regulation and DT. Due to the lack of intensive repeated measures data examining both constructs, it is presently unclear how emotion regulation and DT deficits or skills may manifest in the same individuals across contexts.

Methodological Challenges

Poorly Delineated Samples

Many studies examining SITB outcomes fail to adequately assess or describe their samples’ SITB history beyond the outcome of focus, muddying extant literature. Given that SITBs commonly co-occur (Ribeiro et al., 2016), this limits our ability to identify patterns across the full range of outcomes. Other research groups recruited samples where only a subset of participants reported any SITB. One study included participants with at least one of many forms of impulsive/self-damaging behavior, but the authors do not report what percentage of the sample had SITB histories. This makes it challenging to know how relevant reported findings were to SITB, specifically. Relatedly, studies that deliberately recruited SITB samples rarely examined DT/DI as a primary process or outcome, while studies focused on DT or DI rarely had a primary focus on SITBs. DT and SITB were seldom both an explicit focus, but rather, two of many constructs under examination assessed via self-report at a single time point.

Underexplored Mediators and Moderators

Scientists’ interpretation of findings from behavioral DT paradigms typically assume all participants were induced to experience similar levels of distress, and thus the time to discontinue the task is an appropriate indicator of DT. However, intra-individual and environmental factors may drive differences in the level of distress induced by these tasks, which then muddies interpretation of observed results. For example, persons who consistently experience high levels of aversive negative affect may be less distressed during DT tasks due to habituation, and thus persist longer than persons who have less exposure to frustration and failure. This result may not be because these participants are “better” at tolerating distress, but rather, the type of distress induced by the task is not meaningfully different from their typical experiences. It’s also possible that setting may impact task performance in ways that have not been directly measured. Although most studies using the PASAT-C linked longer persistence to SITB, each recruited participants from restrictive environments (e.g., psychiatric inpatient settings). This may be a substantive finding indicative of problematic persistence through unnecessary discomfort among severely impaired populations. However, this pattern may also be driven by differences in environmental context. Participants in inpatient and medical-residential settings may be motivated to persist through these tasks to relieve boredom. Some data show participants are willing to shock themselves in a laboratory paradigm rather than doing nothing (Nederkoorn et al., 2016; Yusoufzai et al., 2022). Individual differences in grit and degree of motivation to persist to the point of true intolerance are also rarely examined as covariates or explanatory factors, despite their potential for confounding results. Few empirically test important mediators such as level of frustration or distress induced by behavioral paradigms.

Within-Person Processes

As with many other risk processes for psychopathology, there appear to be relatively stable between-person differences on DT as well as within-person fluctuations, or state-level variability in individuals’ capacity to tolerate aversive inputs (see Veilleux, 2023). Veilleux (2023) distinguishes between “having” and “doing” lenses for DT (pp. 358). The latter situates DT as an action that may or may not be chosen in a given moment due to immediate availability of resources, perceived self-efficacy for DT, and willingness to experience distress. Researchers who only employ a “having” lens are likely to treat DT as a static ability and assume that performance on a given DT task will generalize. Although some individuals may have gross deficits in DT across most distressing contexts, many may conditionally struggle with DT.

Unfortunately, within-person variability in DT has not been well explored in the SITB literature. This stymies our ability to identify the factors that contribute to when persons are and are not able to tolerate distress and rigorously evaluate DI as a risk factor for SITB. The vast majority of reviewed articles examined DT cross-sectionally (k = 67), precluding any examination of within-person changes. Only two studies employed ecological momentary assessment (EMA) paradigms (Ammerman et al., 2017; Wallace et al., 2024), and only one of these assessed DT repeatedly in relation to SITB. Wallace and colleagues (2024) led a 65-day EMA study examining positive and negative affect, rumination, emotion dysregulation, DI, and suicidal ideation among 95 participants leaving a suicide-related hospitalization. At the within-person level, the authors found that DI, higher daily levels of negative affect, rumination, and emotion dysregulation were each associated with higher next-day suicidal ideation. At the between-person level, greater variability in both rumination and DI scores increased risk for suicidal ideation with intent, suicidal behaviour, and attempt over the 6 month follow-up period. Unfortunately, participants’ EMA adherence was low, resulting in responses being aggregated to the day level. While better able to speak to state-like fluctuations in DT compared with other studies in our review, data likely do not map onto the temporal scale of DT/DI directly influencing momentary SITB. The authors also examined each risk factor in a separate model. It’s thus unclear whether DI would account for variance in SITB over and above other (likely overlapping) constructs.

Although only one study examined DT/DI dynamically, there is some empirical support for variability in vulnerable persons’ perceived DT ability. What’s more, DI and its variability appear relevant to SITB prediction. Further EMA research using more frequent sampling and comprehensive statistical models is needed to clarify the unique predictive value of DT/DI variability relative to other established risk factors.

General Discussion and Future Directions

Varied theoretical models conceptualize DT as protective (and DI as deleterious) in relation to SITB outcomes. Most empirical research in these domains treats DT as a relatively stable trait, which then exerts a causal influence on SITB. This is consistent with the general pattern observed in prior literature, namely, that low DT exhibits between-person relations with increased incidence or severity of SITB. However, variability in published research outputs, including findings in the reverse direction, complicates this picture. Effects are impacted by measurement modality (e.g., self-report versus behavioral task), failure to consider overlapping constructs (e.g., emotion regulation), examination of single SITB domains (even in populations experiencing high rates of co-occurring SITBs), lack of consideration of contextual factors (e.g., baseline distress, environmental context), and almost no data on within-person variability and state-level DT in relation to SITB. We did not include certain types of distressing experiences as search terms in this review, such as tolerance of ambiguity or uncertainty. In spite of these limitations, the present review provides valuable insights into the DT-SITB relationship, which can have useful clinical implications (given evidence that DT is modifiable) and clear theoretical relevance. In order to move this work forward, we provide a number of recommendations for future research in this domain.

Recommendation 1: Improve rigor in cross-sectional research.

Many limitations of cross-sectional research can be addressed by traditional and intensive longitudinal methods. Yet, retrospective, cross-sectional approaches will inevitably endure. Thus, scientists should make efforts to improve their study’s rigor and relevance. When using self-report DT measures, researchers should describe scores as reflecting participants’ self-perceived DT capacity, rather than “true” DT capability. This is important, given lack of concordance with behavioral task outcomes and limited data tying self-report DT to DT practices in daily life. Studies that use both behavioral and self-report measures of DT ought to ensure that neither is presented as the “true” indicator of DT, as each index discrete phenomena. Authors should also directly address the possibility that participants’ experiences with SITB may inform their perceptions of their DT capacity (e.g., asking participants how they interpret self-report items in relation to their SITB). Refining DT self-report also requires nuanced assessment of contextual factors (e.g., designing measures to assess DT in relation to specific triggering events, affective states, etc.), using quantitative, qualitative, or mixed method approaches.

Behavioral tasks ostensibly capturing DT may be less susceptible to self-appraisal biases, but suffer from poorer ecological validity. Future work should, when possible, develop distress inductions that are phenomenologically consistent with the types of distress that can trigger SITB in daily life. This may require idiographic approaches to DT task stimuli. Further, researchers should explicitly measure both participants’ self-rated distress as triggered by the task, and their behavioral indicators of DT (e.g., time to discontinuation), to ascertain whether group differences are driven by one or both of these factors. Researchers should empirically test whether person-specific factors like baseline distress or desire to please the investigator impact both distress induction and behavioral outcomes during these procedures.

When analyzing cross-sectional data on SITB and DT, researchers should make every effort to use robust methods that can explicitly test underlying mechanistic assumptions about their relationship. For example, cross-sectional mediation models suffer limitations in making causal inferences, as it is often possible to “flip” the proposed predictor and mediator and observe similar patterns. Researchers using these models should avoid making causal conclusions and instead consider these results as indicative of whether an observed effect may be explained by or related to another construct, even if they aid in forming causal hypotheses to be tested explicitly in future studies. Lastly, given conceptual and methodological overlap with other constructs of interest in the study of SITB, scientists should empirically test the unique, additive benefit of considering DT in predicting SITB, above and beyond other known correlates. For example, researchers could examine whether DT predicts SITB when controlling for associated phenomena like emotion (dys)regulation.

Recommendation 2: Consider appropriate controls in traditional longitudinal research.

Traditional longitudinal research has the benefit of examining variables of interest reported at different time points, which can contribute to causal inferences in some cases. Yet much extant longitudinal research in DT and SITB is limited by a failure to consider whether these constructs are stable over time, and to what extent covariances at the same assessment interval can create a false impression of temporal precedence. Specifically, if DT itself is at least somewhat stable over time and is associated with SITB when assessed concurrently, a statistically significant relationship between DT at one time point and SITB at a later time point may be driven by the effect of DT predicting itself over time. DT may exhibit a concurrent, but not prospective, relation with SITB at that same later assessment. To make clearer inferences about the actual impact of DT on SITB over time, models should explicitly incorporate the extent to which DT and SITB are stable. This will allow us to determine whether DT itself contributes to later SITB when controlling for prior SITB experiences. When considering developmental perspectives, further work is needed during early life to more accurately model SITB first onset. This work is challenging given the need to follow large cohorts of youth over many years to detect an adequate number of new onset cases to model effectively.

Recommendation 3: Focus on robust, contextually sensitive, intensive longitudinal research.

Without systematically investigating stable and variable elements of DT as they relate to SITB in ecologically valid contexts, we are unlikely to uncover proximal and cumulative risk factors that may be the most promising targets for intervention. Using EMA and other intensive longitudinal data collection methods will allow researchers to address fundamental questions about DT as a construct and its association with SITB. First, basic science research must clarify how much persons’ perceived self-efficacy with DT, and observed DT behavior, vary in daily life (using self-report and mobile behavioral task methodologies, respectively). Evidence of within-person DT consistency over time would support current approaches to conceptualizing DT at the trait level. Evidence of variability, even for a subset of individuals, would clarify the need for greater time-varying investigations of DT in relation to SITB. Subsequent work will need to clarify when and how these fluctuations occur (e.g., state factors influencing DT). This may include consideration of factors such as level and type of distress, affective response (e.g., intensity, specific affective state, duration), access to other coping resources, and the environmental and social contexts. Research is also needed to clarify what, if any, between- persons factors (e.g., trait factors) relate to DT variability or stability, and whether these patterns are protective or harmful in the context of emotion regulation processes. Although one study connected DT variability to increased SITB risk (Wallace et al., 2024), additional work is needed to understand why, and whether DT variability can be associated with improved functioning (e.g., behavioral flexibility, adaptation to context). Lastly, and most critically, research will need to examine time-varying associations between a person’s momentary DT and SITB outcomes. This will clarify when and how DT may play a causal role in specific SITB experiences.

Conclusion

Individuals who struggle with SITB often exhibit patterns of emotional difficulties, including elevated distress and poor emotion regulation, suggesting a potential role of distress tolerance in understanding the development and maintenance of SITB. In general, research supports an association between low DT and SITB outcomes, but patterns vary by study population, measurement approach, and SITB construct under investigation. Further, the majority of this work is cross-sectional and limited in utility for understanding causal relations between DT and SITB. Efforts to improve methodological, psychometric, and statistical rigor of DT-SITB scholarship will improve the interpretability of current and future findings. Most critically, substantial work is needed to expand trait-like conceptualizations of DT into examining state (momentary) DT, as it relates to within-person changes in SITB risk.

Public health significance statement:

Distress intolerance seems weakly to moderately associated with self-injurious thoughts and behaviors (SITBs) when assessed via self-report; however, behavioral measures are less consistently informative. Further research and improved measurement tools are needed to better understand the role of distress tolerance/intolerance in SITBs.