Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2025 Dec 3.
Published before final editing as: Anxiety Stress Coping. 2025 Nov 9:1–11. doi: 10.1080/10615806.2025.2584314

Bidirectional Associations Between Negative Relationship Events and Suicidal Ideation: An EMA study of Stress Exposure and Generation

Adam G Horwitz a,*, Nadia Al-Dajani b, Kaitlyn McCarthy c, Victor Hong a, Cheryl A King a, Ewa K Czyz a
PMCID: PMC12670516  NIHMSID: NIHMS2125622  PMID: 41207611

Abstract

Background:

Ecological momentary assessments (EMA) have recently enabled the examination of near-term associations between interpersonal stressors and suicide risk. Yet, studies have typically considered the impact of negative relationship events (NREs) on subsequent suicidal ideation (SI) (i.e., stress exposure), with little research examining the impact of SI on subsequent NREs (i.e., stress generation). The present study examined next-day bidirectional associations between NREs and SI, as well as between NREs and interpersonal constructs linked to SI (thwarted belongingness, perceived burdensomeness).

Method:

Young adults (N = 102; Ages 18-25 (M=20.9); 67% cisgender women; 75% White) completed EMAs for 2 months following an emergency department visit for suicide risk-related concerns.

Results:

In multi-level models testing bidirectional relationships with next-day outcomes, within-person main effects were not detected, in either direction, for the constructs under consideration. However, between-person associations were present for NREs and perceived burdensomeness, wherein individuals who generally experience greater burdensomeness endured more daily NREs and those with more NREs reported higher daily-level burdensomeness.

Conclusions:

Results did not suggest that daily fluctuations in SI or NREs correspond to next-day outcomes in this high-risk group. However, person-level differences suggest chronic stressors play a significant role in day-to-day experiences.

Keywords: stress generation, daily diary, suicidal ideation, interpersonal conflict, perceived burdensomeness

1. Introduction

Suicide is a significant public health concern and the second-leading cause of death in the United States for young adults (CDC, 2025). Several prominent theories pertaining to suicidal thoughts and behaviors (STBs) have emphasized the role of interpersonal constructs. For instance, the interpersonal theory of suicide (Joiner, 2005; Van Orden et al., 2010) emphasizes the roles of perceived burdensomeness and thwarted belongingness as social forces contributing to the desire to die. The three-step theory (Klonsky & May, 2015) likewise identifies social connectedness as a key protective factor that can inhibit the transition from suicidal ideation (SI) to suicidal behaviors. While interpersonal constructs have generally been studied cross-sectionally or prospectively over longer time intervals to identify individuals who may be at an elevated risk for suicide, less is known regarding when an individual might be at risk (Franklin et al., 2017; Glenn & Nock, 2014) or how day-to-day negative relationship events might impact near-term STBs. Understanding near-term suicide risk associated with interpersonal factors is especially critical for young people, as there is a heightened sensitivity and experiential significance of these events during adolescence and young adulthood (e.g., Hutchinson et al., 2025).

Fortunately, recent technological advances have allowed for significant growth in the intensive longitudinal study of STBs (e.g., Ammerman & Law, 2022; Auerbach et al., 2023; Horwitz et al., 2022; Rabasco & Sheehan, 2022). Ecological momentary assessment (EMA) and daily diaries of STBs have elucidated the variability and fluctuation of these constructs, as well as associated cognitive, emotional, and social risk factors (e.g., Al-Dajani & Czyz, 2022; Armey et al., 2020; Bentley et al., 2021). Importantly, these intensive longitudinal methods provide opportunities for identifying periods when individuals may be at a particularly elevated risk for STBs, and facilitating an improved understanding of the near-term and dynamic relationship between negative relationship events and STBs.

Despite the expansion of EMA methods to examine near-term associations with SI, relatively few studies have considered the impact of near-term interpersonal factors (Gee et al., 2020). In an EMA study of young women with past-year SI, Victor et al. (2019) found rejection and criticism predicted near-term self-injurious urges, however these relationships were indirect and mediated by increased negative affect. Likewise, an EMA study by Kaurin et al. (2022) found interpersonal conflict to be indirectly related to near-term SI through increased negative affect and decreased positive affect in a sample of individuals with borderline personality disorder. Using timeline follow-back, Bagge et al. (2023) found that negative interpersonal life events were more commonly present for individuals in the 6 hours before they made a suicide attempt relative to a matched 6-hour period the day before. These findings suggest that near-term negative relationship events may spark a cascading effect of affect dysregulation leading to increased SI. Glenn et al. (2022) extended this notion in an EMA study with adolescents, demonstrating that familial thwarted belongingness mediated the relationship between prior-day interpersonal negative life events and next-day SI. Notably, these associations between interpersonal factors and SI have been examined through a stress exposure lens, with studies positing interpersonal conflict and aversive life events predicting or impacting suicidality (e.g., Hallensleben et al., 2019) and social support serving as a protective factor (Coppersmith et al., 2019). However, there is a lack of systematic investigation into the potential inverse association (i.e., stress generation), whereby underlying SI contributes to future interpersonal conflict.

The phenomenon of stress generation was described by Hammen (1991), who demonstrated that individuals with depression were more likely to experience interpersonal negative life events as a consequence of their depressive symptoms (i.e., presence of depressive symptoms like pessimism, irritability, and anhedonia increased the probability for negative interpersonal events). While stress generation theory has been applied and validated for a range of psychological conditions (Rnic et al., 2023), stress generation in the context of suicidality and negative relationship events has been relatively understudied, possibly due to the limited ability of commonly used measures to assess for temporal patterns (Carter et al., 2017; Liu & Spirito, 2019). From a theoretical standpoint, Stellrecht et al. (2006) noted that individuals experiencing suicidal depression may engage in interpersonal behaviors (e.g., seeking excessive reassurance, negative feedback seeking) that can strain relationships and increase interpersonal rejection. Initial work with respect to stress generation resulting from STBs demonstrates suicidal behaviors (Liu & Spirito, 2019) and SI (Uliaszek et al., 2023) predict life stressors over the course of 6 months to 1 year, yet prior studies have not examined these potential bidirectional effects with intensive longitudinal methods. The presence of bidirectional effects would suggest that treatment approaches might benefit from targeting reinforcement cycles that generate additional conflicts, in addition to learning to cope with events after they occur. A better understanding of these near-term associations between negative relationship events and SI, as well as interrelated interpersonal constructs such as perceived burdensomeness and thwarted belongingness, is needed to inform efforts to provide timely and effective interventions targeting STBs.

Taken together, the evidence is clear that interpersonal factors play a critical role in STBs, yet evidence examining the near-term impact of negative relationship events on STBs is in its early stages. Furthermore, EMA studies to date have focused predominantly on a unidirectional pathway between negative relationship events and subsequent STBs, with less attention on the potential generative mechanism of SI contributing to future negative relationship events. This intensive longitudinal study aims to investigate the near-term (i.e., next-day) bidirectional relationships between negative relationship events and SI among young adults at elevated risk for suicide. Given past research indicating potential mediators between interpersonal conflicts and SI (e.g., Glenn et al., 2022; Kaurin et al., 2022), bidirectional associations between negative relationship events and interpersonal constructs linked to SI (i.e., thwarted belongingness, perceived burdensomeness) will also be examined.

2. Methods

2.1. Participants

Participants were 102 Emergency Department (ED) patients between the ages of 18-25 (M(SD) = 20.9 (2.1)). With respect to gender identity, 67% identified as women, 16% as men, and 18% as transgender, non-binary, or gender-queer. Regarding racial/ethnic background composition, 75% of the sample identified as White, 9% reported multiple races, 6% were Asian, 5% were Black or African-American, 2% American Indian or Alaska Native, and 3% other race. 11% reported a Hispanic or Latino ethnicity. With respect to sexual orientation, the sample was 47% heterosexual, 20% bisexual, 17% mostly heterosexual, 8% gay or lesbian, 7% pansexual, 6% queer, 5% asexual, and 6% identified with another sexual orientation (note: individuals could identify with more than one sexual orientation, so percentages are greater than 100%).

2.2. Procedures

Participants were recruited from a psychiatric ED between June 2020 and May 2021. Initial eligibility was based on chart review of inclusion criteria of last-month suicide attempt and/or last-week SI and exclusion criteria of altered mental state (e.g., acute psychosis, mania), moderate-to-severe cognitive impairment, transfer to jail or police custody, or other reasons preventing contact (e.g., no contact information). As described elsewhere (Czyz et al., 2023), 429 individuals met initial eligibility and were contacted by email and/or phone; of those, 289 did not respond to contacts (67.4%), 20 (4.6%) declined or were unable to be enrolled, 120 provided study consent (28.0%), and 110 were enrolled, with a total of 102 participants serving as the data analytic sample (3 withdrew, 1 provided no follow-up data, 4 had no consecutive observations). Following enrollment, participants completed four EMAs between 9:30am and 9:30pm over the course of 8 weeks using the MetricWire (metricwire.com) mobile application; the surveys were randomly sampled within four time-blocks corresponding to morning, early afternoon, late afternoon, and evening time periods. Participants were compensated up to $304 based on the number of study steps completed, which included EMAs, passive data collection, and an end-of-study phone assessment (Czyz et al., 2023; Jiang et al., 2023). The study was approved by the participating university’s Institutional Review Board.

2.3. Measures

2.3.1. Suicidal ideation (SI).

SI was assessed at each EMA, and if SI was present, a follow-up item regarding duration of ideation, modeled after the duration item of the Columbia-Suicide Severity Rating Scale, was delivered, “In the last hour, did you have thoughts of killing yourself? How long did these thoughts last?” Participants responded in reference to the last hour at each EMA (e.g., ranging from a few seconds (1) to full hour/continuous (4); no ideation resulted in a ‘0’). Responses were averaged at the daily level for analyses predicting next-day outcomes (M across observations for 102 participants = 0.24, SD = 0.56).

2.3.2. Burdensomeness and Thwarted Belongingness.

Participants rated their perceived sense of burdensomeness (“The people in my life would be happier without me”) and thwarted belongingness (“I feel close to other people”) on 7-point scales ranging from not at all true for me (1) to very true for me (7) during the past hour at each EMA. These items were based on the Interpersonal Needs Questionnaire (Van Orden et al., 2012). Responses were averaged at the daily level for analyses predicting next-day outcomes (M across observations for 102 participants for burdensomeness = 2.31 (SD = 1.63) and thwarted belongingness = 3.70 (SD = 1.69).

2.3.3. Negative relationship events.

At the evening EMA, participants were asked, “Today, did you have a negative relationship event such as a serious or disruptive argument, separation, or falling out with someone?” for six relationship domains: romantic partner, friend/peer, professional (e.g., teacher, employer/supervisor), parent, non-parent relative, or other. Event examples were informed by prior studies of acute interpersonal suicide warning signs (Bagge et al., 2013; Bagge et al., 2023). A binary variable was computed for presence of a significant negative relationship event on a given day. Presence of a negative relationship event across the 8-week period for the 102 participants in the sample was reported on 682 (13.9%) observation days.

2.4. Data analytic plan

A series of multilevel models examined the bidirectional daily-level associations between: (1) SI and negative relationship event(s), (2) burdensomeness and negative relationship event(s), and (3) thwarted belongingness and negative relationship event(s). This resulted in six models, where: (1) within-person change in SI (i.e., person-mean centered) predicted presence (yes/no) of next-day negative relationship event(s), (2) within-person change in negative relationship event(s) (i.e., person-mean centered) predicted severity of next-day SI, (3) within-person change in burdensomeness (i.e., person-mean centered) predicted presence of next-day negative relationship event(s), (4) within-person change in negative relationship event(s) (i.e., person-mean centered) predicted severity of next-day burdensomeness, (5) within-person change in thwarted belongingness (i.e., person-mean centered) predicted presence of next-day negative relationship event(s), and (6) within-person change in negative relationship event(s) (i.e., person-mean centered) predicted severity of next-day thwarted belongingness. These models were first examined as random intercept only models. To determine whether random slopes should be included, we relied on statistical fit indices (i.e., Akaike information criteria (AIC), Bayesian information criterion (BIC), and log likelihood estimates; Oleson et al., 2022) as well as theoretical rationale (Kleiman, 2017b). Specifically, as the bidirectional associations between these variables have not been investigated in this context, we sought to explore whether allowing the strength and direction of the relationship to vary between people would account for additional variance in the outcome of interest. Thus, we considered models including random intercepts and slopes.

For all models predicting next-day negative relationship event(s), multi-level logistic regression was used to predict the binary variable, coded as presence/absence of negative relationship events. For all models predicting a continuous outcome (SI, thwarted belongingness, or burdensomeness), multi-level linear regression was used. As we considered whether between-person processes (i.e., differences between individuals) contribute to day-level outcomes; all models included both within-person change in the predictor of interest (e.g., person-mean centered burdensomeness) and a between-person variable, which included averaging over the 56-day sampling window (e.g., mean-level thwarted belongingness across the 56 days). All level 1 predictors included in the models were person-mean centered and included the following covariates: continuous time and prior-day levels of the next-day outcome of interest. All level-2 predictors were grand-mean centered. Maximum likelihood estimation was used to manage missing data. All analyses were conducted in R using lme4 package (Bates et al., 2015). Lme4 allows for fitting both multilevel logistic and linear regression models with a default unstructured covariance matrix for the residuals. A test of power indicated adequate power (1 - β > 0.8) to detect moderate (d = 0.5) effect sizes, but limited power to detect small (d = 0.2) effects (Kleiman, 2017a).

3. Results

Throughout the 56-day sampling window, a total of 14,708 observations were recorded out of a possible 22,848 (64.37% adherence rate). At the daily level, 4879 observations were recorded out of a possible 5712 observation days (85.4% adherence rate). 1147 (23.4%) instances of SI were captured at the daily level. Eighty-nine participants reported at minimum one instance of SI at the daily level (87.3%). Significant within-person variability was observed for daily mean burdensomeness (1-ICC = 0.33), daily mean thwarted belongingness (1-ICC = 0.42), daily occurrence of negative relationship events (1-ICC = 0.79), and daily mean SI (1-ICC = 0.55).

3.1. Bidirectional associations between SI and negative relationship events

In random-intercept only models, higher-than-usual SI significantly predicted presence of next-day negative relationship events, while presence of negative relationship events did not predict next-day SI. Model fits were significantly improved when random slopes were included in these models (see Supplement Table 1 for model fit indices), and higher-than-usual SI was no longer a significant predictor of next-day negative relationship events (see Table 1). 1

Table 1.

Bidirectional associations of suicidal ideation and negative relationship events

Intercept only Intercept + Random slopes
Next-day SI1 Est (95% CI) SE t Est (95% CI) SE t
 Intercept 0.24 (0.17, 0.32) 0.04 6.23*** 0.24 (0.16, 0.31) 0.04 6.27***
 SI day 0.40 (0.37, 0.43) 0.02 24.69*** 0.38 (0.35, 0.42) 0.02 23.68***
 NRE 0.01 (−0.03, 0.05) 0.02 0.65 0.00 (−0.05, 0.05) 0.03 0.16
Next-day NRE2 Est (95% CI) SE z Est (95% CI) SE z
 Intercept −1.65 (−2.03, −1.35) 0.17 −9.70*** −1.62 (−2.02, −1.28) 0.17 −9.37***
 SI day 0.32 (0.05, 0.61) 0.13 2.51* 0.02 (−0.50, 0.43) 0.22 0.01
 NRE 0.73 (0.44, 1.04) 0.13 5.61*** 0.70 (0.37, 1.00) 0.13 5.26***
Open in a new tab
1

Linear mixed effects models examined next-day suicidal ideation duration.

2

Logistic mixed effects models examined next-day negative relationship event occurrence.

Note. Est (95% CI) = Estimate and 95% Confidence Interval. SI day = Suicidal ideation duration average across the day. NRE = Negative relationship event. All predictors represent within-person changes (i.e., person-mean centered). ‘Day’ in study was included as control but is not presented. Log likelihood tests indicated that the intercept + random slopes models provided significantly better fit than intercept only.

*

p <.05.

**

p <.01.

***

p <.001.

3.2. Bidirectional associations between burdensomeness and negative relationship events

In random-intercept only models, higher-than-usual burdensomeness significantly predicted presence of next-day negative relationship events, while presence of negative relationship events did not predict severity of next-day burdensomeness. Model fits were significantly improved when random slopes were included in these models, and higher-than-usual perceived burdensomeness was no longer a significant independent predictor of next-day negative relationship events (see Table 2).

Table 2.

Bidirectional associations of perceived burdensomeness and negative relationship events

Intercept only Intercept + Random slopes
Next-day PB1 Est (95% CI) SE t Est (95% CI) SE t
 Intercept 2.24 (1.96, 2.48) 0.14 16.53*** 2.24 (1.96, 2.50) 0.14 16.56***
 PB day 0.49 (0.45, 0.51) 0.02 31.31*** 0.47 (0.44, 0.50) 0.02 30.10***
 NRE 0.00 (−0.07, 0.08) 0.04 0.11 −0.02 (−0.13, 0.11) 0.06 −0.32
Next-day NRE2 Est (95% CI) SE z Est (95% CI) SE z
 Intercept −1.58 (−1.93, −1.22) 0.17 −9.26*** −1.55 (−1.90, −1.19) 0.17 −8.87***
 PB day 0.20 (0.08, 0.32) 0.06 3.57*** 0.03 (−0.12, 0.21) 0.09 0.37
 NRE 0.69 (0.34, 0.99) 0.13 5.25*** 0.63 (0.35, 0.92) 0.14 4.67***
Open in a new tab
1

Linear mixed effects models examined next-day perceived burdensomeness.

2

Logistic mixed effects models examined next-day negative relationship event occurrence.

Note. Est (95% CI) = Estimate and 95% Confidence Interval. PB day = Perceived burdensomeness average across the day. NRE = Negative relationship event. All predictors represent within-person changes (i.e., person-mean centered). ‘Day’ in study was included as control but is not presented. Log likelihood tests indicated that the intercept + random slopes models provided significantly better fit than intercept only.

*

p <.05.

**

p <.01.

***

p <.001.

3.3. Bidirectional associations between thwarted belongingness and negative relationship events

In random-intercept only models, thwarted belongingness did not significantly predict presence of next-day negative relationship events, and presence of negative relationship events did not predict severity of next-day thwarted belongingness. These associations remained non-significant when random slopes were included in these models (see Table 3).

Table 3.

Bidirectional associations of thwarted belongingness and negative relationship events

Intercept only Intercept + Random slopes
Next-day TB1 Est (95% CI) SE t Est (95% CI) SE t
 Intercept 3.63 (3.35, 3.88) 0.13 26.87*** 3.63 (3.34, 3.88) 0.13 26.86***
 TB day 0.30 (0.27, 0.33) 0.02 18.24*** 0.30 (0.27, 0.33) 0.02 18.17***
 NRE 0.03 (−0.07, 0.13) 0.05 0.60 0.03 (−0.08, 0.15) 0.06 0.59
Next-day NRE2 Est (95% CI) SE z Est (95% CI) SE z
 Intercept −1.64 (−1.98, −1.32) 0.17 −9.67*** −1.64 (−2.02, −1.35) 0.17 −9.66***
 TB day −0.04 (−0.14, 0.06) 0.05 −0.83 −0.05 (−0.17, 0.07) 0.06 −0.79
 NRE 0.78 (0.47, 1.11) 0.13 6.09*** 0.78 (0.47, 1.06) 0.13 6.09***
Open in a new tab
1

Linear mixed effects models examined next-day thwarted belongingness.

2

Logistic mixed effects models examined next-day negative relationship event occurrence.

Note. Est (95% CI) = Estimate and 95% Confidence Interval. TB day = Thwarted Belongingness average across the day. NRE = Negative relationship event. All predictors represent within-person changes (i.e., person-mean centered). ‘Day’ in study was included as control but is not presented. Log likelihood tests indicated that the intercept + random slopes models provided significantly better fit than intercept only.

*

p <.05.

**

p <.01.

***

p <.001.

3.4. Secondary analysis: inclusion of level 2 variables

In secondary analyses, we examined a set of multi-level models that included between-level predictor for each predictor of interest, representing the average or aggregate variable across the sampling period. Findings showed that negative relationship events at the between-person level significantly predicted daily burdensomeness (B = 2.36, SE = 0.68, p = .001), suggesting that individuals with generally more negative relationship events have, on average, higher daily-level burdensomeness. Similarly, between-person burdensomeness significantly predicted daily negative relationship events (B = 0.33, SE = 0.10, p = .001), suggesting that those who tend to experience higher burdensomeness in general experienced more daily-level negative interpersonal events. There were no statistically significant associations at the between-person level for SI or thwarted belongingness with negative relationship events.

3.5. Sensitivity Analyses

Due to concerns about reduced power when including multiple controls, we carried out models that did not control for prior-day levels of the outcome (see Supplement Tables 2–4). In these models, there were some significant within-person associations. Specifically, within-person negative relationship events predicted next-day burdensomeness (B(95% CI) = 0.19 (0.04, 0.34), SE = 0.07, p = .013) and next-day suicidal ideation (B(95% CI) = 0.07 (0.01, 0.14), SE = 0.03, p = .025). Within-person burdensomeness, within-person thwarted belongingness, and within-person suicidal ideation remained non-significant predictors of next-day negative relationship events even when removing respective prior-day outcome covariates.

4. Discussion

This was the first study to examine bidirectional associations between negative relationship events and next-day SI. The study also considered bidirectional relationships between negative relationship events and two interpersonal constructs, thwarted belongingness and perceived burdensomeness, that have a well-established relationship with SI. Overall, results did not detect an effect for within-person changes in negative relationship events on next-day SI (i.e., stress exposure) or within-person changes of SI on next-day negative relationship events (i.e., stress generation). Similarly, within-person changes in negative relationship events were not significantly associated with either next-day burdensomeness or thwarted belongingness in this study, nor were within-person changes in these interpersonal constructs linked with next-day negative relationship events.

Despite significant effects for higher-than-usual SI predicting presence of next-day negative relationship events in initial intercept-only models, these associations were no longer significant when random slopes were included. This suggests there are likely person-level differences that explain the associations between SI and negative relationship events. Indeed, additional analyses that incorporated between-level covariates indicated that individuals who tended to experience higher burdensomeness over the course of the study had higher daily-level means for negative relationship events, and conversely, those who generally reported more negative relationship events during the study period had more severe daily-level burdensomeness. Taken together, our findings suggest that despite existing evidence for the association between SI and negative relationship events, these may not necessarily manifest on a day-to-day level over and above trait-level characteristics. These relationships might be better explained by the overall severity of experiences, consistent with prior research demonstrating between-person differences to have stronger effects than within-person changes with respect to the associations between negative life events and depression (e.g., Maciejewski et al., 2021). It is also possible, however, that the bidirectional associations at the within-person daily level manifest in shorter windows of time (i.e., hours apart instead of days apart), or that these effects are small and undetected in the current study’s design. Of note, sensitivity analyses revealed that, when we lessened the model controls of adjusting for prior-day outcomes, some of the within-person relationships were significant in the direction of negative relationship events predicting next-day suicidal ideation and burdensomeness. This suggests that our null findings could potentially be influenced by methodological considerations and should be interpreted with caution. Despite a lack of significance for the hypothesized within-person effect between negative relationship events and SI, these results are in line with previous studies that examined similar near-term associations. Specifically, prior EMA studies suggest that the effects of negative interpersonal events/conflicts may be only indirectly related to suicidal ideation or urges through either negative affect (Kaurin et al., 2022; Victor et al., 2019) or thwarted belongingness (Glenn et al., 2022), though thwarted belongingness lacked a significant association with negative relationship events in our study.

While within-person changes were not detected for next-day outcomes for any of the constructs under consideration, there were some notable trait-level patterns. Specifically, the results of the between-person models suggest that perceived burdensomeness may play a role in the cycle of negative relationship events, such that individuals who tend to experience higher levels of burdensomeness experience more daily-level negative relationship events. Similarly, those who generally experience more negative relationship events are at greater risk of daily-level burdensomeness. While informative for understanding how these more stable characteristics may impact daily-level functioning, the temporality of this association could not be established. It may be that in a high-risk sample, the interpersonal tendencies that contribute to stress generation (e.g., seeking excessive reassurance, negative feedback seeking; Stellrecht et al., 2006) are behaviorally embedded in a cycle of negative relationship events, perceived burdensomeness, thwarted belongingness, and SI that cannot be disentangled. It may also be that within-person changes among these constructs have effects at briefer intervals than those assessed in the current study. Additional research is imperative to better understand the temporal dynamics between negative relationship events and SI, which would inform just-in-time adaptive interventions seeking to disrupt momentary reinforcement cycles that perpetuate SI. Research is especially needed in samples with emerging risk (versus high risk), and into the potential mediators (e.g., negative affect, thwarted belongingness, perceived burdensomeness) and moderators (e.g., personality features, social support network) of these associations between negative relationship events and SI.

4.1. Limitations

While this study had many strengths, including an intensive longitudinal design with high-risk young adults experiencing a recent suicidal crisis, findings should be interpreted in the context of study limitations. Our sample was composed primarily of White women who were recruited from a single study site and consented to participate in an EMA study, which may limit generalizability to individuals of different races, ethnicities, genders, and geographical regions. Generalizability may have also been impacted by COVID-19, as patterns of negative relationship events and/or suicidal ideation might have differed from ordinary times due to various public restrictions and stressors during various points of the study. There are important distinctions between subjective perceptions of connectedness and objective metrics of social integration/network (e.g., Cero et al., 2024) and our study was unable to contextualize a negative relationship event within one’s social network. While SI was measured multiple times per day, the negative relationship event item was only assessed once per day, which restricted the prediction of near-term outcomes to be on the daily-level (versus predicting within-day timepoints). Though this daily assessment is more proximal than most research conducted on this topic to date, multiple daily assessments of negative relationship events would have enabled a narrower examination of near-term effects, particularly if these effects are too transient to capture at a daily level. Furthermore, the primary constructs of interest were from single-item EMAs, which may have negatively impacted reliability due to increased measurement error and vulnerability to bias. Finally, non-significant findings may have been due to inadequate power to detect less-than-moderate effect sizes, particularly due to relatively low daily-endorsement rates for the negative relationship events—it is possible that a continuous measure capturing lower-intensity interpersonal stressors could have identified underlying associations between SI and interpersonal conflict/events.

4.2. Conclusions

While the overarching connections between interpersonal factors and suicide risk have been well-established, relatively few studies have examined the near-term associations between negative relationship events and SI, and potential bidirectional effects have not been fully considered. The current study did not detect significant next-day associations, in either direction, between SI and negative relationship events, though these may be connected through perceived burdensomeness over longer periods of time. Nevertheless, in addition to working on coping skills following stress exposure, potential stress-generative mechanisms that reinforce cycles of SI and interpersonal stress may be valuable treatment targets. Additional research with greater power to detect smaller effects over shorter periods of time is needed to better understand the near-term dynamics of these constructs to best inform the development of timely interventions seeking to disrupt these problematic reinforcement cycles.

Supplementary Material

Supplement

Acknowledgements:

We wish to thank the participants in this study and the clinical and administrative staff at the University of Michigan’s Psychiatric Emergency Services for their support of our research recruitment at their site.

Funding:

This work was supported by grant funding from the American Foundation for Suicide Prevention and the Frances and Kenneth Eisenberg Research Award at the University of Michigan to Dr. Ewa Czyz. Dr. Horwitz was supported by a NIMH award (K23-MH131761). Funders had no role in the analysis, interpretation, or preparation of this manuscript.

Footnotes

CRediT author roles:

Horwitz: Conceptualization, methodology, validation, writing-original draft. Al-Dajani: Conceptualization, methodology, formal analysis, writing-original draft. McCarthy: Conceptualization, writing-original draft. Hong: Conceptualization, resources, writing-editing and reviewing. King: Conceptualization, methodology, supervision, writing-editing and reviewing. Czyz: Conceptualization, methodology, validation, data curation, project administration, funding acquisition, writing-editing and reviewing.

Declaration of competing interest: The authors have no known competing financial interest or personal relationships that might appear to influence the work reported in this manuscript.

1

Given the zero-inflated distribution of the continuous SI score, sensitivity analyses examined SI as a binary variable in the four models. Findings remained the same except in the intercept-only model, the person-centered binary SI predictor was not significantly associated with next-day negative relationship events, though it approached significance (p = .071).

Data availability:

The datasets generated and analyzed for this study are available upon request from the corresponding author.

References

  1. Al-Dajani N, & Czyz EK (2022). Suicidal desire in adolescents: an examination of the interpersonal psychological theory using daily diaries. Journal of Clinical Child & Adolescent Psychology, 1–15. 10.1080/15374416.2022.2051525 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Ammerman BA, & Law KC (2022). Using intensive time sampling methods to capture daily suicidal ideation: A systematic review. Journal of affective disorders, 299, 108–117. 10.1016/j.jad.2021.10.121 [DOI] [PubMed] [Google Scholar]
  3. Armey MF, Brick L, Schatten HT, Nugent NR, & Miller IW (2020). Ecologically assessed affect and suicidal ideation following psychiatric inpatient hospitalization. General hospital psychiatry, 63, 89–96. 10.1016/j.genhosppsych.2018.09.008 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Auerbach RP, Lan R, Galfalvy H, Alqueza KL, Cohn JF, Crowley RN, Durham K, Joyce KJ, Kahn LE, & Kamath RA (2023). Intensive longitudinal assessment of adolescents to predict suicidal thoughts and behaviors. Journal of the American Academy of Child & Adolescent Psychiatry. 10.1016/j.jaac.2023.03.018 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bagge CL, Glenn CR, & Lee HJ (2013). Quantifying the impact of recent negative life events on suicide attempts. J Abnorm Psychol, 122(2), 359–368. 10.1037/a0030371 [DOI] [PubMed] [Google Scholar]
  6. Bagge CL, Littlefield AK, Wiegand TJ, Hawkins E, Trim RS, Schumacher JA, Simons K, & Conner KR (2023). A controlled examination of acute warning signs for suicide attempts among hospitalized patients. Psychological medicine, 53(7), 2768–2776. 10.1017/S0033291721004712 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Bates D, Maechler M, Bolker B, Walker S, Christensen RHB, Singmann H, Dai B, Grothendieck G, Green P, & Bolker MB (2015). Package ‘lme4’. convergence, 12(1), 2. https://doi.org/http://dk.archive.ubuntu.com/pub/pub/cran/web/packages/lme4/lme4.pdf [Google Scholar]
  8. Bentley KH, Coppersmith DL, Kleiman EM, Nook EC, Mair P, Millner AJ, Reid-Russell A, Wang SB, Fortgang RG, & Stein MB (2021). Do patterns and types of negative affect during hospitalization predict short-term post-discharge suicidal thoughts and behaviors? Affective science, 2, 484–494. 10.1007/s42761-021-00058-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Carter G, Milner A, McGill K, Pirkis J, Kapur N, & Spittal MJ (2017). Predicting suicidal behaviours using clinical instruments: Systematic review and meta-analysis of positive predictive values for risk scales. British Journal of Psychiatry, 210(6), 387–395. 10.1192/bjp.bp.116.182717 [DOI] [PubMed] [Google Scholar]
  10. Centers for Disease Control. (2025). Web-based Injury Statistics Query and Reporting System (WISQARS) 2023. Centers for Disease Control. Retrieved 2025 April 23 from https://www.cdc.gov/injury/wisqars/index.html [Google Scholar]
  11. Cero I, De Choudhury M, & Wyman PA (2024). Social network structure as a suicide prevention target. Soc Psychiatry Psychiatr Epidemiol, 59(3), 555–564. 10.1007/s00127-023-02521-0 [DOI] [PubMed] [Google Scholar]
  12. Coppersmith DDL, Kleiman EM, Glenn CR, Millner AJ, & Nock MK (2019). The dynamics of social support among suicide attempters: A smartphone-based daily diary study. Behaviour Research and Therapy, 120, 103348. 10.1016/j.brat.2018.11.016 [DOI] [PubMed] [Google Scholar]
  13. Czyz EK, King CA, Al-Dajani N, Zimmermann L, Hong V, & Nahum-Shani I (2023). Ecological Momentary Assessments and Passive Sensing in the Prediction of Short-Term Suicidal Ideation in Young Adults. JAMA Netw Open, 6(8), e2328005. 10.1001/jamanetworkopen.2023.28005 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Franklin JC, Ribeiro JD, Fox KR, Bentley KH, Kleiman EM, Huang X, Musacchio KM, Jaroszewski AC, Chang BP, & Nock MK (2017). Risk factors for suicidal thoughts and behaviors: a meta-analysis of 50 years of research. Psychological bulletin, 143(2), 187. [DOI] [PubMed] [Google Scholar]
  15. Gee BL, Han J, Benassi H, & Batterham PJ (2020). Suicidal thoughts, suicidal behaviours and self-harm in daily life: A systematic review of ecological momentary assessment studies. DIGITAL HEALTH, 6, 205520762096395. 10.1177/2055207620963958 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Glenn CR, Kleiman EM, Kandlur R, Esposito EC, & Liu RT (2022). Thwarted Belongingness Mediates Interpersonal Stress and Suicidal Thoughts: An Intensive Longitudinal Study with High-risk Adolescents. Journal of Clinical Child & Adolescent Psychology, 51(3), 295–311. 10.1080/15374416.2021.1969654 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Glenn CR, & Nock MK (2014). Improving the short-term prediction of suicidal behavior. Am J Prev Med, 47(3 Suppl 2), S176–180. 10.1016/j.amepre.2014.06.004 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Hallensleben N, Glaesmer H, Forkmann T, Rath D, Strauss M, Kersting A, & Spangenberg L (2019). Predicting suicidal ideation by interpersonal variables, hopelessness and depression in real-time. An ecological momentary assessment study in psychiatric inpatients with depression. European Psychiatry, 56(1), 43–50. 10.1016/j.eurpsy.2018.11.003 [DOI] [PubMed] [Google Scholar]
  19. Hammen C (1991). Generation of stress in the course of unipolar depression. Journal of abnormal psychology, 100(4), 555. [DOI] [PubMed] [Google Scholar]
  20. Horwitz AG, Kentopp SD, Cleary J, Ross K, Wu Z, Sen S, & Czyz EK (2022). Using machine learning with intensive longitudinal data to predict depression and suicidal ideation among medical interns over time. Psychological medicine, 1–8. 10.1017/S0033291722003014 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Hutchinson E, Scott L, Choukas-Bradley S, & Silk J (2025). Interpersonal risk factors for suicide in daily life among young people: A review of intensive longitudinal studies. Development and Psychopathology, 1–21. 10.1017/S0954579424001810 [DOI] [PubMed] [Google Scholar]
  22. Jiang A, Al-Dajani N, King C, Hong V, Koo HJ, & Czyz E (2023). Acceptability and feasibility of ecological momentary assessment with augmentation of passive sensor data in young adults at high risk for suicide. Psychiatry Res, 326, 115347. 10.1016/j.psychres.2023.115347 [DOI] [PubMed] [Google Scholar]
  23. Joiner TE (2005). Why people die by suicide. Harvard University Press. [Google Scholar]
  24. Kaurin A, Dombrovski AY, Hallquist MN, & Wright AGC (2022). Momentary interpersonal processes of suicidal surges in borderline personality disorder. Psychological medicine, 52(13), 2702–2712. 10.1017/s0033291720004791 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Kleiman E (2017a). EMAtools: Data management tools for real-time monitoring/ecological momentary assessment data. In. [Google Scholar]
  26. Kleiman E (2017b). Understanding and analyzing multilevel data from real-time monitoring studies: An easily-accessible tutorial using R. 10.31234/osf.io/xf2pw [DOI] [Google Scholar]
  27. Klonsky ED, & May AM (2015). The Three-Step Theory (3ST): A New Theory of Suicide Rooted in the “Ideation-to-Action” Framework. International Journal of Cognitive Therapy, 8(2), 114–129. 10.1521/ijct.2015.8.2.114 [DOI] [Google Scholar]
  28. Liu RT, & Spirito A (2019). Suicidal Behavior and Stress Generation in Adolescents. Clinical Psychological Science, 7(3), 488–501. 10.1177/2167702618810227 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Maciejewski D, van Sprang E, Spinhoven P, & Penninx B (2021). Longitudinal associations between negative life events and depressive symptoms-A 9-year longitudinal study on between-person and within-person effects and the role of family history. J Pers Soc Psychol, 121(3), 707–721. 10.1037/pspp0000381 [DOI] [PubMed] [Google Scholar]
  30. Oleson JJ, Jones MA, Jorgensen EJ, & Wu YH (2022). Statistical Considerations for Analyzing Ecological Momentary Assessment Data. J Speech Lang Hear Res, 65(1), 344–360. 10.1044/2021_jslhr-21-00081 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Rabasco A, & Sheehan K (2022). The use of intensive longitudinal methods in research on suicidal thoughts and behaviors: A systematic review. Archives of suicide research, 26(3), 1007–1021. 10.1080/13811118.2021.1903635 [DOI] [PubMed] [Google Scholar]
  32. Rnic K, Santee AC, Hoffmeister JA, Liu H, Chang KK, Chen RX, Neufeld RWJ, Machado DA, Starr LR, Dozois DJA, & LeMoult J (2023). The vicious cycle of psychopathology and stressful life events: A meta-analytic review testing the stress generation model. Psychol Bull, 149(5-6), 330–369. 10.1037/bul0000390 [DOI] [PubMed] [Google Scholar]
  33. Stellrecht NE, Joiner TE Jr., & Rudd MD (2006). Responding to and treating negative interpersonal processes in suicidal depression. J Clin Psychol, 62(9), 1129–1140. 10.1002/jclp.20298 [DOI] [PubMed] [Google Scholar]
  34. Uliaszek AA, Hamdullahpur K, Amestoy ME, & Carnovale M (2023). Longitudinal pathways between suicidal ideation and life stress. Anxiety, Stress, & Coping, 36(5), 590–602. 10.1080/10615806.2023.2165646 [DOI] [PubMed] [Google Scholar]
  35. Van Orden KA, Cukrowicz KC, Witte TK, & Joiner TE (2012). Thwarted belongingness and perceived burdensomeness: construct validity and psychometric properties of the Interpersonal Needs Questionnaire. Psychol Assess, 24(1), 197–215. 10.1037/a0025358 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Van Orden KA, Witte TK, Cukrowicz KC, Braithwaite SR, Selby EA, & Joiner TE Jr. (2010). The interpersonal theory of suicide. Psychological Review, 117(2), 575–600. 10.1037/a0018697 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Victor SE, Scott LN, Stepp SD, & Goldstein TR (2019). I Want You to Want Me: Interpersonal Stress and Affective Experiences as Within-Person Predictors of Nonsuicidal Self-Injury and Suicide Urges in Daily Life. Suicide and life-threatening behavior, 49(4), 1157–1177. 10.1111/sltb.12513 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

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

Supplementary Materials

Supplement
NIHMS2125622-supplement-Supplement.docx (23.2KB, docx)

Data Availability Statement

The datasets generated and analyzed for this study are available upon request from the corresponding author.

RESOURCES