Coexistence of Post-Traumatic Growth and Distress: Examining Latent Profiles of Homicidally and Naturally Bereaved Adults

Post-Traumatic Growth

The concept of PTG emerged in the 1990s within the field of positive psychology, a branch of psychology which aimed to broaden the scope of inquiry beyond negative outcomes, such as mental disorders, to include positive ones, notably wellbeing. Martin Seligman, the founder of positive psychology, argued that wellbeing is not merely the absence of suffering, but the presence of positive states of mind (Seligman, 2001). To capture the full complexity of human experience, he advocated for a deeper examination of happiness, strengths, fulfillment, and virtue which had received far less attention than distress (Seligman, 2001). Building on these ideas, Tedeschi and Calhoun introduced the term post-traumatic growth (PTG) to describe the positive psychological changes that can follow a traumatic event (Tedeschi & Calhoun, 1996). PTG extends beyond resilience, defined as recovery or return to baseline levels of functioning (B. W. Smith et al., 2008) but encompassing transformation in as many as five domains: discovering new possibilities in life, enhancing relationships, recognizing personal strength, deepening spirituality, or developing a greater appreciation of life (Calhoun et al., 2010). Far from being rare, PTG is reported by an estimated 30% to 70% of people after trauma (Linley & Joseph, 2004). For many, trauma thus becomes a catalyst for increased wellbeing, fostering constructive transformation of perspectives, priorities and fundamental beliefs, and this phenomenon has been observed in bereavement (Hurst & Kannangara, 2024; Michael & Cooper, 2013; Waugh et al., 2018; Wilson et al., 2025). Meta-analytic findings indicate that PTG is not simply the inverse of trauma-related distress. Although these constructs are often related, they are empirically distinct and can co-occur within individuals, reflecting separate underlying processes (Shakespeare-Finch & Lurie-Beck, 2014). In recent years, research has increasingly examined the coexistence of posttraumatic distress and wellbeing, seeking a more nuanced understanding of psychological adaptation to trauma. However, this coexistence has rarely been explored in bereaved populations, particularly those whose loss occurred through homicide. Homicide bereavement represents a unique and complex form of loss due to its violent, sudden, intentional and other-inflicted nature, in addition to possible social stigmatization and trial-related stressors (Connolly & Gordon, 2015; Zinzow et al., 2009). These factors make homicide loss particularly relevant in the study of distress and wellbeing coexistence.

Psychosocial Factors

Several psychosocial factors may influence distress and wellbeing after loss, such as fundamental beliefs. Experiencing a traumatic event such as the death of a loved one has been associated with disruptions in individuals’ fundamental beliefs about themselves, others and the world. Indeed, after a traumatic loss, people may no longer think that the world is a good place, that they are satisfied with who they are, or that others are inherently trustworthy and kind (Ehlers & Clark, 2000; Foa & Rothbaum, 1998; Janoff-Bulman, 1989; Resick & Schnicke, 1992). The association between negative beliefs and posttraumatic distress symptoms has been well documented in studies of homicidally bereaved individuals (HBI) as well as other trauma-exposed populations, showing that endorsement of negative beliefs tends to correlate with higher levels of distress (Bailey & Morris, 2021; Boelen et al., 2006, 2015; Kern & Peterson, 2021; Lenferink et al., 2018; Lyons et al., 2020). Concurrently, the shattering of fundamental beliefs followed by their subsequent rebuilding to accommodate for the traumatic event is described as a necessary process for the emergence of PTG (Calhoun et al., 2010; Janoff-Bulman, 1989; Tedeschi & Calhoun, 1996).

A potential mediator of the association between negative beliefs and distress is trauma coping self-efficacy, defined as one’s perceived capacity to manage recovery demands following trauma (e.g., regulate emotions, resume normal life, seek support, avoid self-blame, etc.; Benight et al., 2015; Benight & Bandura, 2004; Cieslak et al., 2008; Tanguay-Sela et al., forthcoming). Low self-efficacy may foster feelings of helplessness and exacerbate distress symptoms, whereas high self-efficacy has been associated with more favorable outcomes across various populations (Benight et al., 2000, 2004; Benight & Harper, 2002; Bosmans et al., 2015; Chirico et al., 2017; Delahaij & Van Dam, 2017; Lambert et al., 2012).

A large body of literature has also found an association between perceived social support and positive mental health outcomes. Social support is a multi-dimensional construct encompassing both instrumental assistance and emotional care provided by family, friends or significant others (Benkel et al., 2009; Zimet et al., 1988). Numerous studies (including those conducted with bereaved individuals) have confirmed its role as a protective factor against PTSD, MDD and PGD (Al-Gamal et al., 2019; R. Chen, 2022; Vanderwerker & Prigerson, 2004; Zimet et al., 1988). Social support has also been found to facilitate the grieving process (Benkel et al., 2009; Kreicbergs et al., 2007; Laakso & Paunonen-Ilmonen, 2002; Lennon et al., 1990) and to be positively associated with PTG (Kokou-Kpolou et al., 2022; Michael & Cooper, 2013; Prati & Pietrantoni, 2009) and LS (Dimond et al., 1987).

Identifying Bereavement Profiles

Latent profile analysis (LPA) is a statistical method used to identify subgroups of individuals who share similar patterns of response across multiple variables. An advantage of LPA is its ability to assess multiple continuous indicators simultaneously and identify complex, previously unobserved (“latent”) associations. LPA is well suited to bereavement research, given the heterogeneity of psychological reactions to loss. Better understanding the distinct needs of bereaved individuals may inform more targeted and effective interventions.

To our knowledge, only five studies have used latent analyses that combine distress and wellbeing indicators in bereaved samples. These studies have identified subgroups characterized by the coexistence of PGD and PTG (C. Chen & Tang, 2021; Kokou-Kpolou et al., 2022; Li et al., 2021; Valencia et al., 2025; Zhou et al., 2018). However, these samples included few or no HBI.

To build upon this work, we expanded the scope by incorporating a broader set of distress and wellbeing variables, namely PTSD, MDD and LS. We also chose to focus on homicide loss, a form of bereavement that it is distinct in being intentionally caused by another person. A comparison group of naturally bereaved individuals (NBI; i.e., bereaved by physical illness) was included to contrast this prevalent form of loss with the unique grief experience associated with homicide.

Objectives

This first objective of this study was to identify subgroups of HBI and NBI based on their mental health symptomatology. Drawing on previous research, we expected to identify three to four distinct profiles: one profile characterized by low distress and high wellbeing, another by high distress and low wellbeing, and at least one illustrating the coexistence of moderate-to-high levels of both distress and wellbeing. In other words, we anticipated that one or more profiles would show elevated PTG and LS while simultaneously exhibiting considerable levels of PTSD, MDD, and PGD symptoms. The second objective was to determine whether HBI were more likely than NBI to belong to profiles marked by higher distress. Consistent with prior findings, we hypothesized that this would be the case.

Participants and Procedure

Recruitment occurred via Canadian bereavement organizations, social media platforms, newspaper advertisements, and posters displayed in mental health centers and on a university campus. Participants were required to be at least 18 years of age and to have lost a close family member (parent, child, romantic partner, sibling, grandparent or grandchild) to homicide or illness at least 6 months prior to participation. Additional inclusion criteria included residing in Canada. The questionnaire was hosted on the Qualtrics platform (www.qualtrics.com). Participants indicated informed consent via an electronic form. No financial compensation was offered, and participation in an anonymous draw for one of five $70 Amazon gift cards was optional. The Université du Québec à Montréal Research Ethics Board approved the study.

A total of 238 participants were recruited, including 102 HBI and 136 NBI. The average participant age was 53.55 years (SD = 14.89, range: 21.00–88.00). Most participants identified as female (84.5%). Only 6.8% identified as part of a visible minority, with 84.9% reporting their ethnicity as Quebecer, Canadian, or American. Half of participants were married or in a common-law partnership, while others reported being widowed (21.4%), divorced (11.3%) or single (16.4%). Participants were highly educated, with 84.4% having completed a pre-university college or technical diploma, an undergraduate degree or a graduate degree. Participants lost either a parent (29.0%), a child (25.2%), a romantic partner (23.1%), a sibling (16.0%), a grandparent (6.3%), or a grandchild (0.4%). The average age of the deceased was 48.84 years (SD = 23.84, range: 0.00–95.00), with 42.0% identifying as female. The average time elapsed since the death was 10.96 years (SD = 13.02, range: 0.42–60.50).

The majority of homicides were perpetrated by men (78.0%). In most other cases, the participants did not know the murderer’s gender (18.0%). In 86.1% of homicide events, only the participant’s loved one died. Most often, the murderer was reported to be a stranger to the deceased individual (25.0%), a friend, colleague or acquaintance (19.0%) or a former (13.0%) or current (17.0%) romantic partner. Among participants bereaved by illness, most had lost a loved one to cancer (59.3%), with other frequent causes of death including heart disease (13.3%), respiratory disease (4.4%), and dementia or Alzheimer’s disease (3.7%). Compared to NBI, HBI reported lower educational attainment, longer time elapsed since the death, younger ages of the deceased, less frequent presence at the time of the death and a higher number of psychiatric diagnoses in the past year. HBI lost more children and siblings, whereas NBI lost more parents, romantic partners and grandparents. Additional descriptive information about HBI, NBI and the deceased is available in Supplemental Table 1.

Measures

Sociodemographic characteristics and characteristics of the loss were obtained via a custom questionnaire. Quality of the relationship with the deceased was measured using the closeness subscale of the Quality of Relationships Inventory–Bereavement version (QRI-B; Bottomley et al., 2019). The 8-item subscale is rated from 1 (not at all) to 4 (very much). A higher total score (possible range: 8–32) indicates greater closeness with the deceased. The internal consistency was α = .89.

Data Analysis

Data cleaning was performed in R (version 4.1.2). Participants who provided scores for at least 80% of items in each measure were included in the analysis. To obtain total scores for each measure in the case of missing item scores, our procedure was the following: if 20% or fewer item scores were missing, a mean of the existing item scores entered by a given participant was calculated and imputed in the place of their missing item scores. Continuous measures were mean centered (Aiken et al., 1991).

Latent profile analysis (LPA) was conducted to identify the optimal number and composition of profiles among HBI and NBI based on distress (PTSD, MDD and PGD), and wellbeing indicators (LS and PTG). Models ranging from one to six profiles were estimated using Mplus (version 7), with a robust maximum likelihood estimator (Muthén & Muthén, 2017). Model selection was guided by multiple criteria: (1) model fit indices (Akaike information criterion [AIC], Bayesian information criterion [BIC], and sample-size adjusted BIC, with lower values indicating a better fit), (2) comparison of model k versus k-1 using the Lo–Mendell–Rubin likelihood ratio test (LMRT) where significant results suggest improved fit (Lo et al., 2001; McLachlan & Peel, 2004; Nylund et al., 2007), (3) classification accuracy, evaluated using entropy, with higher values indicating better classification, and (4) class size, to ensure adequate subgroup representation, as very small classes may reflect spurious or unstable solutions (Hipp & Bauer, 2006). As part of our model-selection process, we examined the stability of the solution across different model specifications, and the pattern of profiles remained consistent across models, supporting the robustness of the selected solution. All latent profile models were estimated using 1,000 initial random starts and 250 final-stage optimizations (STARTS = 1000 250), with 500 random starts used for the final loglikelihood replication check (STITERATIONS = 500).

The Kruskal-Wallis test was used to examine differences in distress symptoms across profiles (Kruskal & Wallis, 1952). We employed the modified Bolck–Croom–Hagenaars method (BCH; Bolck et al., 2004) within Mplus to test whether profiles differed in age, time since the homicide, fundamental beliefs, trauma-coping self-efficacy, and perceived social support. The BCH procedure was estimated using the MLR estimator, which applies full-information maximum likelihood (FIML) to retain all available data for distal outcomes. In addition, the categorical distal outcome method (DCAT; Lanza et al., 2013) was used to assess differences in death type, gender, and relationship to the deceased across profiles.

Although gender, age and time since death were initially considered as control variables due to their potential influence on the bereavement outcomes, we ultimately chose not to include them as covariates given the inconsistency of these associations across studies, with some studies finding their influence to be non-significant (Burke & Neimeyer, 2013; Djelantik et al., 2020; Killikelly et al., 2019; Lundorff et al., 2017; Zakarian et al., 2019; Zhou et al., 2020).

Latent Profile Analysis

We conducted LPA to identify profiles of HBI and NBI. Model fit indices did not indicate a single optimal solution (see Table 1). Although fit indices favored the five- and six-profile solutions based on AIC, BIC, sample-size adjusted BIC and entropy, closer inspection indicated that the additional classes reflected minor variations in symptom intensity rather than meaningful differences in profile structure. In both models, the smallest classes were very small in size and represented only more extreme versions of existing profiles, raising concerns about overfitting and limiting their interpretability. Inspection of these small classes indicated that they followed the same overall pattern as existing profiles, with uniformly higher symptom levels across all distress indicators, suggesting a severity gradient rather than a qualitatively distinct profile. We therefore focused on the three- and four-profile solutions, evaluating them using LMRT p-values. While both solutions revealed distinct symptom profiles, we prioritized parsimony. We noted that the four-profile solution also included a small class that reflected a severity split rather than a substantively meaningful distinct subgroup. Taken together, these considerations supported the three-profile model as the best-fitting solution, providing a clearer delineation of profiles. See Tables 2 and 3 for mean scores, distributions and between-profile comparisons. Classification quality for the three-profile solution was good, with high average latent class posterior probabilities for each profile, ranging from 0.91 to 0.97 on the diagonal, indicating clear separation between classes (see Supplemental Table 2). To demonstrate that PTG and distress represent empirically distinct dimensions, we examined their bivariate correlations. PTG showed near-zero associations with PTSD (r = .01) and PGD (r = –0.02), and a small negative association with depressive symptoms (r = –0.15).

OPEN IN VIEWER

Table 1.

Model Fit Statistics for Latent Profile Analyses.

Number of profiles	AIC	BIC	Sample-size adjusted BIC	LMRT p-value	Entropy	Share of participants in smallest profile (%)
1	3,392.06	3,426.79	3,395.09	-	-	-
2	3,027.02	3,082.57	3,031.86	<.001	0.90	27.4
3	2,889.30	2,965.69	2,895.96	.004	0.86	18.1
4	2,846.37	2,943.60	2,854.84	.017	0.88	6.7
5	2,824.63	2,942.68	2,834.91	.390	0.89	4.2
6	2,808.23	2,947.12	2,820.33	.539	0.87	3.8

Note. Bold values indicate “best” fit for each respective statistic. AIC = Akaike Information Criterion; BIC = Bayesian Information Criterion; LMRT = Lo–Mendell–Rubin likelihood ratio test.

OPEN IN VIEWER

Table 2.

Means and Standard Errors of Sociodemographic Characteristics, Characteristics of the Loss and Psychosocial Factors by Profile.

Variable	Profile 1 n = 106 (44.5%)		Profile 2 n = 89 (37.4%)		Profile 3 n = 43 (18.1%)		χ2	p
	Low Distress, High LS		Moderate Distress, Moderate LS		High Distress, Low LS
	M	SE	M	SE	M	SE
Age	54.76	1.66	54.00	1.60	49.73	2.11	3.84	.147
Months since loss	168.55	18.03	95.75	14.79	113.06	22.78	9.05**	.011
Relationship quality	27.52	0.52	29.13	0.48	29.43	0.63	7.00*	.030
Social support	65.51	1.56	59.50	1.70	50.38	2.33	29.47***	<.001
Self-efficacy	51.14	0.75	42.79	0.93	29.47	1.41	194.09***	<.001
Fundamental beliefs	114.80	1.11	111.94	1.37	108.37	2.67	6.15*	.046
	N	(%)	N	(%)	N	(%)	χ 2	p
Women	87	82.1	76	85.4	38	88.4	2.28	.320
Death by homicide	38	35.9	34	38.2	30	69.8	18.07***	<.001
Person lost							48.62***	<.001
Child & grandchild	18	17.0	26	29.2	17	39.5
Sibling	24	22.6	8	9.0	6	14.0
Romantic partner	20	18.9	31	34.8	4	9.3
Parent	30	28.3	23	25.8	16	37.2
Grandparent	14	13.2	1	1.1	-	-
Level of education							15.32*	.018
High school or less	12	11.3	14	15.7	11	25.6
College diploma	25	23.6	35	39.3	12	27.9
Undergraduate	42	39.6	29	32.6	13	30.2
Graduate school	27	25.5	11	12.4	6	14.0
Missing	0	0.00	0	0.00	1	2.3

Note. Gender, death by homicide, relationship to the deceased and level of education are expressed as percentages. The “child” and “grandchild” relationship types were merged due to the presence of only one grandchild. The levels of education “less than high school” and “high school” were merged as only five participants reported having less than a high school diploma. The χ² values were derived from the BCH procedure for continuous variables and from DCAT for categorical variables.

*

p < .05; **p < .01; ***p < .001.

OPEN IN VIEWER

Table 3.

Pairwise χ² Comparisons of Sociodemographic Characteristics, Characteristics of the Loss and Psychosocial Factors by Profile Based on BCH and DCAT Analyses.

Variable	Profile 1 vs. Profile 2		Profile 1 vs. Profile 3		Profile 2 vs. Profile 3
	Low Distress, High LS vs. Moderate Distress, Moderate LS		Low Distress, High LS vs. High Distress, Low LS		Moderate Distress, Moderate LS vs. High Distress, Low LS
	χ2	p-Value	χ2	p-value	χ2	p-Value
Months since loss	8.67**	.003	3.67	.056	0.38	.536
Relationship quality	4.64*	.031	5.53*	.019	0.14	.712
Social support	6.05*	.014	29.18***	<.001	9.41**	.002
Self-efficacy	44.17***	<.001	185.23***	<.001	59.13***	<.001
Fundamental beliefs	2.36	.125	4.95*	.026	1.33	.249
Death by homicide	0.06	.810	16.24***	<.001	11.89**	.001
Person lost	24.18***	<.001	28.86***	<.001	13.30*	.010
Level of education	12.01**	.007	6.77	.080	2.46	.483

Note. Only significant comparisons were included.

*

p < .05; **p < .01; ***p < .001.

Individuals in Profile 1, labeled Low Distress–High Life Satisfaction (LS), comprising 44.5% (n = 106) of the sample reported lower levels of distress symptoms (PTSD, MDD and PGD) and higher LS compared to other profiles (see Figure 1). None of the individuals in Profile 1 met the PTSD diagnosis, while only 4.7% met the criteria for PGD, and 2.8% met the threshold for severe MDD. Profile 2, labeled Moderate Distress–Moderate LS (37.4%, n = 89), showed higher distress levels and lower LS than Profile 1. Over a third of individuals in this profile met the PTSD diagnosis threshold (39.3%), the majority (77.5%) met criteria for PGD, and 22.5% met criteria for severe MDD. Lastly, individuals in Profile 3, labeled High Distress–Low LS (18.1%, n = 43), reported the highest levels of distress and the lowest LS. All Profile 3 individuals met criteria for PGD, while 97.7% met criteria for PTSD, and 95.3% met the threshold for severe MDD. PTG did not significantly differ across profiles (χ²(2) = 1.03, p = .596).

OPEN IN VIEWER

Figure 1.

Distributions of standardized scores by predicted profile membership for the three-profile solution.

Latent Profile Comparison

Description of the Identified Profiles

Three profiles were identified: Low Distress–High LS, Moderate Distress–Moderate LS and High Distress–Low LS. These clinically distinct subgroups reflect the heterogeneity of grief responses among bereaved individuals. A 2023 systematic review by Heeke et al. reported that the majority of LPA studies examining mental health in bereaved adult populations identified three profiles, typically including one with low symptom severity and another with high symptom severity. The proportion of participants in each of our profiles is consistent with previous research (Heeke et al., 2023). The intermediate symptom profile in these studies generally displayed high levels of PGD symptoms and low-to-moderate severity of other symptoms. This pattern was also observed in our study: the Moderate Distress–Moderate LS profile was characterized by elevated PGD symptoms, moderate MDD symptoms and low PTSD symptoms. Notably, most of the aforementioned studies did not include wellbeing variables, and those that did generally involved very few HBI (Kokou-Kpolou et al., 2022; Li et al., 2021; Valencia et al., 2025; Zhou et al., 2018). We sought to expand on this prior research by examining not only distress indicators (PTSD, MDD and PGD) but also wellbeing variables (LS and PTG). While previous studies have largely focused on symptom severity, our inclusion of wellbeing variables provided a more nuanced understanding of adaptation after loss and highlighted that distress and wellbeing can coexist in meaningful ways, particularly in the Moderate Distress–Moderate LS profile. Interestingly, this coexistence was characterized by equivalent levels of PTG across profiles. This unexpected finding prompted further analysis of the patterns and potential mechanisms underlying PTG in bereaved individuals. In the following section, we examine the equivalence of PTG between profiles and its implications for theory, research and clinical practice.

Equivalence of Post-Traumatic Growth Between Profiles

Consistent with our hypotheses, our profiles illustrate the coexistence of distress and wellbeing. Indeed, we found no significant difference in PTG between the three profiles, with a mean score of 24.28 (SD = 11.95) on the PTGI-SF. This finding indicates that PTG levels were relatively stable regardless of distress severity. To our knowledge, this is the first study to demonstrate that HBI and NBI experiencing severe distress symptoms may still report PTG levels comparable to those of individuals in lower-distress profiles. It should be noted, however, that no cutoff scores exist to define low, moderate or high levels of PTG. Furthermore, the original PTGI-SF validation study did not report the total score mean of its sample (Cann et al., 2010). As a point of reference, the only study we identified that specifically assessed PTG in an exclusively HBI sample reported mean PTGI-SF scores of 20.3 (SD = 2.94) for men and 25.2 (SD = 2.53) for women (Johnsen & Afgun, 2021). Another study conducted among parents bereaved by a ferry disaster in South Korea reported a mean total PTGI-SF score of 17.22 (SD = 11.21; Huh et al., 2020 ¹ ). Other studies of bereaved individuals either did not report total score means (Zhou et al., 2018), used different measures of PTG (C. Chen & Tang, 2021; Kokou-Kpolou et al., 2022; Patrick & Henrie, 2016; Qian et al., 2025) or both (Li et al., 2021). That said, several studies have used the PTGI-SF in other trauma-exposed populations and may serve as reference points. A recent systematic review in military personnel reported total PTGI-SF means ranging from 17.11 (SD = 14.88) to 20.40 (SD = 11.88), interpreted by the authors as moderate PTG (Mark et al., 2018). While additional studies using the PTGI-SF in HBI and NBI populations are needed to establish robust normative data, our sample’s scores appear to lie on the higher end of distribution observed in the literature, thus representing moderate-to-high levels of PTG. To further explore these findings, future research may benefit from examining qualitative dimensions of PTG within HBI populations, such as the personal meaning attributed to growth experiences or the specific domains of life in which growth is perceived. In addition, it is unclear whether the PTG reported by our sample is constructive or illusory in nature. According to Zoellner and Maercker’s model (2006), PTG may reflect genuine positive psychological change in some cases, but it can also reflect defensive processes such as denial or self-deceptive coping. In other words, self-reported PTG may be subject to bias, as participants may report adaptive change while still experiencing significant distress. This potentially avoidant cognitive strategy may be associated with poorer psychological adjustment to trauma (Zoellner & Maercker, 2006). Future longitudinal research would be needed to better distinguish between these forms of PTG and to clarify their role in psychological adjustment.

Other latent analyses of bereavement that included PTG have also identified the co-occurrence of PTG and distress symptoms (C. Chen & Tang, 2021; Kokou-Kpolou et al., 2022; Li et al., 2021; Valencia et al., 2025; Zhou et al., 2018). However, unlike the present study, these analyses reported that PTG varied across subgroups. This discrepancy may be explained by the type of loss, as prior studies primarily involved deaths due to illness or accidents, with few or no homicide-related losses. According to psychological models, PTG emerges when a traumatic event challenges a person’s fundamental beliefs (Janoff-Bulman, 1989; Tedeschi & Calhoun, 1996). As such, PTG may be particularly likely to occur following severe trauma, such as homicide bereavement (Tedeschi & Calhoun, 1996). Further research is needed to clarify the relationship between homicide-related loss and PTG, and to examine how PTG levels in HBI compare to those observed in individuals bereaved by other causes. Future studies might also explore the mechanisms through which PTG develops in HBI, including the role of meaning making, narrative reconstruction, or access to social and therapeutic support, factors that may play substantially different roles in other types of bereavement.

Coexistence of Life Satisfaction and Distress

The mean LS score in the Low Distress–High LS profile corresponds to a high level of life satisfaction, as defined by Diener (2006). This score is comparable to that observed in a sample of elderly individuals who had lost a partner between 2 and 48 months earlier (Boelen et al., 2023). It is important to note, however, that the vast majority of participants in that study were NBI (Boelen et al., 2023), which limits their suitability as a comparison group, particularly for the High Distress–Low LS profile, which included a higher proportion of HBI. Participants in this profile reported mean LS scores categorized as below average and approaching dissatisfaction (Diener, 2006). The Moderate Distress–Moderate LS profile displays the coexistence of substantial levels of distress symptoms and LS, indicating that despite considerable levels of suffering, people may still evaluate their lives positively. Further research is needed to better understand the factors contributing to SL differences among HBI. To date, few studies have examined SL in bereaved populations. Among those that did, samples generally included very few HBI, did not report LS scores (Prapunoto & Soetjiningsih, 2024), or used non-standardized measures (Xiu et al., 2016).

Prediction of Profile Membership

This study hypothesized that HBI would be more likely to belong to higher distress profiles. This hypothesis was confirmed, as bereavement by homicide significantly predicted membership to the High Distress–Low LS profile. This association between homicide loss and elevated PTSD, MDD and PGD symptoms is consistent with previous findings (Boelen et al., 2015, 2016; Djelantik et al., 2017; Kristensen et al., 2012). For example, Djelantik et al. (2017) identified three classes of bereaved individuals based on symptoms of PTSD, MDD and PGD: a resilient class (minimal symptoms), a class primarily characterized by PGD symptoms, and a combined PGD/PTSD class. Notably, those bereaved by violent deaths were more likely to belong to the latter. Nevertheless, HBI were present in all three of our profiles, suggesting that factors beyond the cause of death contribute to differences in distress and wellbeing levels between subgroups.

Type of death did not predict membership between the Low Distress–High LS and Moderate Distress–Moderate LS profiles. This may be explained by the distribution of this variable across profiles, with the High Distress–Low LS profile containing a substantially higher proportion of HBI (69.8%) than the Low Distress–High LS and Moderate Distress–Moderate LS profiles (35.9% and 38.2%, respectively). Because the latter two profiles contain a similar proportion of HBI and NBI, the type of death can distinguish each of them from the High Distress–Low LS profile, but not from each other.

We assessed several sociodemographic characteristics to better understand profile membership. Neither age nor gender significantly predicted profile assignment. Although some studies have reported gender differences in post-loss mental health, typically with higher symptom levels among women (Boelen et al., 2006, 2015; Soydas et al., 2021; Valencia et al., 2025; Yablon & Itzhaky, 2021), others have found no such association (Boelen et al., 2016, 2019; Taku et al., 2008). Thus, the relationship between gender and bereavement-related distress remains inconclusive, particularly in the context of homicide or natural bereavement. Similarly, age was not associated with profile membership in our sample. While some studies have linked younger (Burke & Neimeyer, 2013; Killikelly et al., 2019; Valencia et al., 2025; Zakarian et al., 2019; Zhou et al., 2020) or older age (Lundorff et al., 2017; Patrick & Henrie, 2016) with greater distress, others found no significant age-related differences (Djelantik et al., 2017). The absence of significant associations between age or gender and profile membership further supported our decision not to include these variables as covariates to preserve model parsimony. Consistent with prior research, we found that educational level was associated with bereavement outcomes. Participants in the Low Distress–High LS profile reported significantly higher levels of education than those in the Moderate Distress–Moderate LS profile. This supports previous findings suggesting that higher levels of education may buffer against psychological distress following a loss (Boelen et al., 2015, 2019; Djelantik et al., 2017; K. V. Smith & Ehlers, 2021). Although individuals in the High Distress–Low LS profile reported the lowest education levels overall, this difference did not reach statistical significance when compared to the other two profiles. This may reflect the relatively small number of participants in this group, or it may indicate that other factors beyond education had a stronger influence on psychological outcomes following the loss. These findings underscore the need to examine interactions between sociodemographic characteristics and psychological resources (e.g., TCSE, social support) to better understand vulnerability and resilience across bereaved subgroups.

Several characteristics of the loss were found to predict profile membership. For example, the Low Distress–High LS profile reported a significantly longer average time elapsed since the death. This finding is consistent with previous research indicating that more time since the loss is associated with reduced distress (Keesee et al., 2008; Kokou-Kpolou et al., 2021; Lundorff et al., 2021; M. O’Connor et al., 2015; Schwartz et al., 2018; Soydas et al., 2021; Zakarian et al., 2019; Zhou et al., 2020). This may be because time allows for emotional adjustment, meaning making and psychological growth following the loss (Keesee et al., 2008).

The relationship with the deceased also varied across profiles. For instance, the High Distress–Low LS profile had the highest proportion of child loss. Prior studies have consistently shown that losing a child is among the most intense forms of grief, often associated with prolonged and severe distress (Boelen et al., 2015; Djelantik et al., 2017, 2020; He et al., 2014; Kersting et al., 2011; K. V. Smith & Ehlers, 2021; Soydas et al., 2021). Some grief studies have even excluded bereaved parents due to the systematic association between child loss and prolonged grief disorder (Fujisawa et al., 2010). This may reflect the perception that child loss is particularly unnatural, as individuals generally expect their children to outlive them, and often hold significant hopes for their future (Brillon, 2025; Kreicbergs et al., 2007). In addition, the loss of a child may challenge the parent’s sense of identity, as the parental role and associated meaning are deeply rooted in emotional and existential dimensions (Brillon, 2025; K. O’Connor & Barrera, 2014; Stroebe & Schut, 1999; Toller, 2008). Another type of loss associated with greater distress is the death of a romantic partner (Boelen et al., 2015; Djelantik et al., 2017; Fujisawa et al., 2010; Kersting et al., 2011; Prigerson et al., 2002; K. V. Smith & Ehlers, 2021; Soydas et al., 2021). In our sample, this loss was more prevalent in the Moderate Distress–Moderate LS profile. The death of a partner may be particularly distressing due to the emotional closeness of the relationship, the loss of a major source of support, and the profound disruption of future plans (Brillon, 2025; Maciejewski et al., 2022). Furthermore, the increased burden of financial, domestic, social, and parenting responsibilities on the surviving partner may compound the psychological toll of the loss (Benight & Bandura, 2004; Brillon, 2025; Stroebe & Schut, 1999).

Beyond the type of relationship, the quality of this relationship with the deceased has also been shown to influence the bereavement process, with closer relationships generally associated with more intense grief reactions (Rheingold et al., 2015; Servaty-Seib & Pistole, 2007). In line with this, we found that relationship quality was significantly lower in the Low Distress–High LS profile compared to the other two. This lower perceived closeness with the deceased may therefore have functioned as a protective factor. While strong relational closeness is often considered a marker of healthy bonds, it may also increase vulnerability to intense grief reactions following the loss (Field & Filanosky, 2009; Stroebe et al., 2007). Conversely, less emotionally intense or ambivalent relationships may buffer the psychological impact of bereavement. These results suggest that, in addition to identifying the type of relationship lost, evaluating its emotional quality is essential to understanding the bereavement experience (Bottomley et al., 2019).

Psychosocial factors also predicted profile membership, with negative fundamental beliefs most prevalent in the High Distress–Low LS profile. This result was expected, as it aligns with the leading cognitive models of trauma and previous studies. Following a traumatic event such as a loss by homicide, individuals may feel more vulnerable than before. Their perceptions of personal safety, trust in others, and self-worth are likely to be disrupted (Ehlers & Clark, 2000; Foa & Rothbaum, 1998; Janoff-Bulman, 1989; Resick & Schnicke, 1992). Similarly, Boelen et al. (2016) found that participants bereaved by violent losses were more likely to belong to a higher-distress profile if they held more negative cognitions about themselves, life, and grief.

Members of the High Distress–Low LS profile also reported lower levels of TCSE than individuals in the other profiles. This was also expected, as higher levels of coping self-efficacy has been linked to better outcomes across several experiences such as cancer loss, natural disasters, military combat, terrorist attacks and sexual assaults (Benight et al., 2001; Benight & Bandura, 2004), and higher TCSE specifically has been associated with more positive fundamental beliefs and better posttraumatic recovery (Benight & Bandura, 2004; Cieslak et al., 2008). In a 2008 study, Cieslak et al. found that TCSE mediated the relationship between negative beliefs about the self and the world, and posttraumatic symptoms among survivors of child sexual abuse and motor vehicle accidents. Such negative cognitions may reduce an individual’s perceived capacity to manage trauma-related challenges, thereby contributing to higher distress levels. We replicated this mediation model in our own study, with results to be presented in an upcoming publication (Tanguay-Sela et al., forthcoming).

Lastly, and consistent with existing literature (Kokou-Kpolou et al., 2022; Li et al., 2021), social support was found to be associated with profile membership. Participants in the Low Distress–High LS group reported the highest level of social support, followed by those in the Moderate Distress–Moderate LS. The High Distress–Low LS group reported the lowest level of social support, highlighting its potential protective role in bereavement-related distress. This underscores the importance of interpersonal connections and support systems, especially for individuals experiencing severe distress and lower LS during bereavement.

It is important to note that certain homicide-specific factors may have affected profile membership and associated psychosocial variables had they been within the scope of this study. For instance, several studies have shown that attending a criminal trial may negatively influence the wellbeing of HBI (Englebrecht et al., 2014; Gekoski et al., 2013; Metzger et al., 2015; Pastia & Palys, 2016). A recent study by Lebel and Brillon (2025) revealed that those who perceived the trial outcome as fair and felt they received adequate information on the criminal justice procedures reported fewer PTSD symptoms and higher life satisfaction, suggesting that perceptions of justice and procedural transparency may interact with profile membership and psychological adjustment. Feelings of distress during the trial, on the other hand, were associated with more PTSD symptoms and lower life satisfaction (Lebel & Brillon, 2025). The media attention received by HBI may also impact their postloss psychological adaptation. HBI have often qualified interactions with media as intrusive, forcing them to grieve in public (Wilson et al., 2025). Furthermore, media reporting is often considered by HBI as intentionally misrepresentative and sensationalistic, amplifying the stigma and shame surrounding their loss (Gekoski et al., 2012; Wilson et al., 2025). Such experiences may influence perceived social support and fundamental beliefs, potentially contributing to more maladaptive profiles. In short, the difficulties of the bereavement process may be exacerbated in HBI by specific stressors which deserve further investigation. These findings also have important implications for service provision across clinical, victim support and criminal justice systems. They suggest the need for coordinated, trauma-informed support systems for HBI, including guidance throughout legal proceedings, protection from harmful media exposure, and interventions aimed at reducing stigma and strengthening access to social support resources.

Theoretical and Clinical Implications

Several implications can be drawn from the results of this study. First, they underscore the heterogeneity of the bereavement experience, emphasizing the importance of tailoring interventions to individuals’ specific psychological and contextual needs. Second, they support Tedeschi and Calhoun’s conceptualization of PTG (1996) by validating the compatibility of PTG and distress symptoms in a population that is still poorly understood, HBI. These findings provide further evidence for distress and wellbeing occupying separate continua, as postulated by proponents of positive psychology (Payton, 2009).

Third, the results indicate that people who exhibit PTG should not be assumed to have healed from the trauma of their loss: they may continue to experience significant posttraumatic symptoms despite positive life changes, as is most clearly shown in the High Distress–Low LS profile. Given the overrepresentation of HBI in this profile, clinicians working with this population should consider the likely coexistence of PTG and distress, monitoring for signs of both and tailoring interventions accordingly. Symptoms of distress should be addressed while educating clients about its common co-occurrence with PTG (Kokou-Kpolou et al., 2022; Liu et al., 2017). Additionally, as distress and growth can co-occur, it may be beneficial to nurture growth early in bereavement interventions, treating symptom reduction and growth fostering as simultaneous rather than sequential goals (Zhou et al., 2018). In this regard, interventions that foster meaning making, TCSE and challenge negative fundamental beliefs may be particularly useful. To promote meaning making, clinicians may integrate narrative approaches or guided expressive writing, enabling bereaved individuals to articulate the significance of their loss and reconstruct their life story in a coherent and adaptive way (Neimeyer, 2001). Cognitive-behavioral interventions may help strengthen TCSE by targeting avoidant behaviors, encouraging mastery experiences, and reinforcing a sense of self-efficacy rather than helplessness in facing trauma-related challenges (Benight & Bandura, 2004). To address negative fundamental beliefs as well as feelings of guilt or shame, cognitive restructuring techniques can be used to identify, examine and modify negative assumptions about the self, others, and the world, which are often activated by traumatic bereavement (Beck & Emery, 1985; Ehlers & Clark, 2000; Foa & Rothbaum, 1998; Janoff-Bulman, 1989; Resick & Schnicke, 1992). Importantly, these strategies should be delivered in a flexible, client-centered manner that respects the pace and readiness of the individual.

Strengths and Limitations of the Current Study

This study has several strengths that enhance its contribution to bereavement research. The inclusion of a large sample (N = 238 total participants) allowed for robust statistical analyses. The use of a wide range of well-validated, reliable psychometric tools to assess psychological outcomes, including PTSD, MDD, PGD, LS, and PTG, adds rigor to the findings. The use of latent profile analysis (LPA) enabled the identification of distinct patterns of mental health, offering a nuanced understanding of bereavement.

However, there are some limitations to consider. The sample was mostly composed of female participants from Quebec who did not consider themselves to be part of a visible minority group. The lack of diversity in the sample likely influenced the emergence and potential stability of latent profiles, as gender and cultural background may shape grief processes, including distress symptomatology and patterns of adaptation. It is therefore possible that different or additional profiles might emerge in more diverse samples. Levels of coping self-efficacy and perceived social support may also differ in a more diverse sample, and thus the strength and nature of the associations between the identified profiles and these variables may be altered. Moreover, the homogeneity of the sample may limit the generalizability of the findings, and future studies should aim to recruit a more diverse sample to capture gender- and culture-based differences in grief. The restrictive inclusion criterion based on close family relationship may also have limited the scope of results. Broadening the criteria to include other relationship types, while continuing to assess relationship quality, would be advisable, as for some, losing a friend may be equally or more impactful than losing a family member. Furthermore, the cross-sectional design limited the ability to infer causality and to track changes in grief over time. Longitudinal methods, such as latent transition analysis (Lenferink et al., 2022) would provide valuable information about the stability of the identified profiles. The reliance on self-report measures introduces potential biases, such as social desirability bias or recall bias, and the recruitment methods (through bereavement organizations, social media, and mental health care centers) may have introduced selection bias. Such methods may attract individuals with higher levels of distress or stronger motivations to seek support, potentially skewing the results. Additionally, the wide range of time since death may have impacted the consistency of the bereavement experience, and the study does not account for other potential confounding factors, such as previous trauma exposure or socioeconomic status, variables which may significantly influence both distress and wellbeing outcomes. These limitations suggest that further research, including more diverse samples and longitudinal designs, could deepen our understanding of the bereavement process.