Introduction
Childhood trauma, here defined as childhood physical, sexual or emotional abuse, and physical or emotional neglect, has been associated with numerous enduring repercussions on functioning (Baldwin et al., Reference Baldwin, Wang, Karwatowska, Schoeler, Tsaligopoulou, Munafò and Pingault2023; Daníelsdóttir et al., Reference Daníelsdóttir, Aspelund, Shen, Halldorsdottir, Jakobsdóttir, Song, Lu, Kuja-Halkola, Larsson, Fall, Magnusson, Fang, Bergstedt and Valdimarsdóttir2024) and is now considered by many as the most important preventable cause of psychopathology (Teicher et al., Reference Teicher, Gordon and Nemeroff2022). The psychological and intergenerational repercussions of trauma would arise from the complex interactions between the characteristics of trauma and biological (Buss et al., Reference Buss, Entringer, Moog, Toepfer, Fair, Simhan, Heim and Wadhwa2017; Lim et al., Reference Lim, Radua and Rubia2014; Ludmer et al., Reference Ludmer, Gonzalez, Kennedy, Masellis, Meinz and Atkinson2018), genetic (Bradley et al., Reference Bradley, Westen, Mercer, Binder, Jovanovic, Crain, Wingo and Heim2011; Dackis et al., Reference Dackis, Rogosch, Oshri and Cicchetti2012), epigenetic (Radtke et al., Reference Radtke, Schauer, Gunter, Ruf-Leuschner, Sill, Meyer and Elbert2015; Yehuda & Lehrner, Reference Yehuda and Lehrner2018), developmental (Garon-Bissonnette et al., Reference Garon-Bissonnette, Duguay, Lemieux, Dubois-Comtois and Berthelot2022; Sauvé et al., Reference Sauvé, Cyr, St-Laurent, Amédée, Dubois-Comtois, Tarabulsy, Bernier and Moss2022; Toth & Cicchetti, Reference Toth and Cicchetti2013), and environmental (Edwards et al., Reference Edwards, Holden, Felitti and Anda2003) risk factors. Yet, a significant proportion of adults who have experienced trauma have preserved functioning (Racine et al., Reference Racine, Eirich and Madigan2022) and several protective factors would contribute to buffer the effects of trauma on psychological functioning during adulthood and on the next generation, including contextual (Langevin et al., Reference Langevin, Marshall and Kingsland2021), intra-individual (Afifi & MacMillan, Reference Afifi and MacMillan2011) and interpersonal protective factors, such as attachment security with primary caregivers (Busch & Lieberman, Reference Busch, Lieberman, Oppenheim and Gold2007), benevolent childhood experiences (Narayan et al., Reference Narayan, Ippen, Harris and Lieberman2019, Reference Narayan, Merrick, Lane and Larson2023) or enriching family-based care (King et al., Reference King, Guyon-Harris, Valadez, Radulescu, Fox, Nelson, Zeanah and Humphreys2023). This calls for studies on the heterogeneity of profiles in terms of developmental risk and protective factors underlying psychopathology and alterations in functioning in youth and adult populations exposed to childhood trauma (Bonanno & Mancini, Reference Bonanno and Mancini2012; Gee, Reference Gee2021; Yoon et al., Reference Yoon, Pei, Logan, Helsabeck, Hamby and Slesnick2023).
Mentalizing trauma
Research has shown that the extent to which experiences of childhood trauma have been processed or resolved is one of the factors that contribute to buffer the psychological and intergenerational effects of trauma (Jacobvitz et al., Reference Jacobvitz, Leon and Hazen2006; Koren-Karie & Getzler-Yosef, Reference Koren-Karie and Getzler-Yosef2019; Swerbenski et al., Reference Swerbenski, Sturge-Apple, Messina, Toth, Rogosch and Cicchetti2023). In recent years, the resolution of trauma has been operationalized through the concept of Mentalization of trauma and the measure of Trauma-specific reflective functioning (Ensink et al., Reference Ensink, Berthelot, Bernazzani, Normandin and Fonagy2014). Mentalizing trauma refers to the ability to reflect on the psychological and relational impact of trauma and to think of traumatic experiences in a coherent and emotionally-grounded fashion (Berthelot, Savard, et al., Reference Berthelot, Savard, Lemieux, Garon-Bissonnette, Ensink and Godbout2022; Ensink et al., Reference Ensink, Berthelot, Bernazzani, Normandin and Fonagy2014). Available evidence using the Trauma-Specific Reflective Functioning Scale (Berthelot & Garon-Bissonnette, Reference Berthelot and Garon-Bissonnette2023), a coding system applied to attachment interviews (i.e., the Adult Attachment Interview, George et al., Reference George, Kaplan and Main1985, or the Parent Development Interview, Slade et al., Reference Slade, Aber, Berger, Bresgi and K.2003), confirms that mentalizing trauma is an important psychological determinant of resilience and adaptation in adults exposed to childhood maltreatment, namely in terms of parenting. For instance, two Canadian studies using a sample of pregnant women with histories of trauma showed that high levels of mentalization of trauma were associated with a positive investment in pregnancy and with the quality of the parental couple functioning (Ensink et al., Reference Ensink, Berthelot, Bernazzani, Normandin and Fonagy2014), and prospectively predicted the quality of the mother-child attachment relationship at 18 months postpartum (Berthelot et al., Reference Berthelot, Ensink, Bernazzani, Normandin, Luyten and Fonagy2015). The protective role of mentalizing trauma for parenting was further supported by recent findings in an American sample of women showing that the quality of mentalizing processes in relation to trauma was strongly associated with the mothers’ sensitivity to their child’s emotional communication, as measured by “parental insightfulness” (Koren-Karie et al., Reference Koren-Karie, Oppenheim, Dolev, Sher and Etzion-Carasso2002), over and beyond the effect of sociodemographic risk factors and parental mentalization (Berthelot et al., Reference Berthelot, Garon-Bissonnette, Muzik, Simon, Menke, Stacks and RosenblumIn Press). Finally, findings from Borelli et al. (Reference Borelli, Cohen, Pettit, Normandin, Target, Fonagy and Ensink2019) showed that higher reflective functioning in relation to trauma acted as a buffer in the intergenerational cycles of sexual abuse. Interestingly, previous findings (Berthelot et al., Reference Berthelot, Garon-Bissonnette, Muzik, Simon, Menke, Stacks and RosenblumIn Press; Ensink et al., Reference Ensink, Berthelot, Bernazzani, Normandin and Fonagy2014) showed that the quality of mentalizing processes in relation to trauma was not associated with the characteristics of the experiences of abuse or neglect (type, timing, and duration), suggesting that other factors may play a crucial role in the development of mentalization processes following exposure to trauma (Fonagy et al, Reference Fonagy, Campbell and Luyten2023).
To facilitate the assessment of trauma-specific reflective functioning, a self-report instrument, the Failure to Mentalize Trauma Questionnaire (FMTQ; Berthelot et al., Reference Berthelot, Savard, Lemieux, Garon-Bissonnette, Ensink and Godbout2022) was recently introduced and validated in a sample of pregnant women. Contrary to the previous coding system, the FMTQ does not capture the complexity of thought processes in relation to trauma but rather assesses indications that the respondent is unable to maintain coherent thinking when discussing traumatic experiences or adverse relational experiences and indices of definite distortions in the perception of the impact of trauma on the self, mental states, and behaviors (Berthelot, Savard, et al., Reference Berthelot, Savard, Lemieux, Garon-Bissonnette, Ensink and Godbout2022). These indications are organized among seven different types of impairments in mentalizing trauma and adverse relationships: Disorganization of thoughts, Grandiosity, Absorption in trauma, Identification with the victim, Identification with the perpetrator, Avoidance of thoughts, and Justification of trauma. Higher scores at the FMTQ have been associated with intimate partner violence victimization and perpetration as well as with psychiatric symptoms, including dissociative symptoms, post-traumatic stress symptoms and personality dysfunctions, over and above the effect of the severity of trauma (operationalized through the total score on the Childhood Trauma Questionnaire) and other confounding variables (Berthelot, Savard, et al., Reference Berthelot, Savard, Lemieux, Garon-Bissonnette, Ensink and Godbout2022; Gamache et al., Reference Gamache, Savard, Lemieux and Berthelot2021).
Heterogeneity in responses to trauma and in mentalizing processes
Advances in developmental psychopathology and developmental neuroscience have demonstrated heterogeneous patterns of reactions, trajectories and outcomes following traumatic experiences (Bonanno & Mancini, Reference Bonanno and Mancini2012; Gee, Reference Gee2021; Meyer & Lee, Reference Meyer and Lee2023) and shown that these differential patterns were little explained by the characteristics of trauma (Cahill et al., Reference Cahill, Hager and Shryane2023; Yoon et al., Reference Yoon, Pei, Logan, Helsabeck, Hamby and Slesnick2023). Whilst there is a common agreement that the determinants of interindividual variability following exposure to trauma probably lies in the heterogeneity of developmental and sociocultural risk and protective factors (Sroufe, Reference Sroufe2009), the field has been dominated by correlational studies linking trauma to poor outcomes (Berthelot et al., Reference Berthelot, Lemieux and Maziade2019, Reference Berthelot, Savard, Lemieux, Garon-Bissonnette, Ensink and Godbout2020) and there is still a need to move “toward more fine-tuned inquiry into the natural heterogeneity of both trauma outcome and the factors that inform it” (Bonanno & Mancini, Reference Bonanno and Mancini2012, p.81).
The recent literature in developmental psychopathology has paid particular interest into the role of caregiving, attachment, and mentalization in risk and resilience trajectories following exposure to traumatic experiences (Fonagy et al., Reference Fonagy, Allison, Campbell, Bateman and Fonagy2019; Gee & Cohodes, Reference Gee and Cohodes2023). Congruently, research has suggested that trauma would impede the development of mentalization during childhood and adolescence (Ensink et al., Reference Ensink, Bégin, Normandin and Fonagy2016) which would exert a downward impact on later development (Morosan et al., Reference Morosan, Ghisletta, Badoud, Toffel, Eliez and Debbané2020). Yet, much remains to be understood about how trauma influences the development of mentalization and the field seems to be moving toward a deeper understanding of interindividual variability in mentalization processes and of how such variations may lead to different health and functioning outcomes. As a case in point, two recent studies have revealed important distinctions in how individuals attempt to mentalize and in the representational content discussed when interviewed in settings aiming to prompt reflective functioning (Garon-Bissonnette et al., Reference Garon-Bissonnette, Dubois-Comtois, St-Laurent and Berthelot2023; Sleed et al., Reference Sleed, Isosävi and Fonagy2021). More specifically, Garon-Bissonnette and colleagues (2023) showed that the exclusive reliance on a global score when assessing multidimensional constructs such as mentalization may have masked the variability of processes underlying respondents’ scores and limited the detection of important phenomena affecting specific subgroups of the population. Indeed, in a community sample of women, no differences in terms of mentalization capacities were observed between participants exposed to trauma and participants without a history of trauma when using a global score of reflective functioning. However, a more precise inquiry into the types of mentalization impairments exhibited by participants of each group showed that childhood maltreatment was strongly predictive of a disrupted, over-analytical and inconsistent way of thinking about attachment experiences, a phenomenon that was practically absent in adults without history of trauma. Interestingly, the seven-factor structure of the FMTQ may offer the opportunity to capture this heterogeneity of thought processes in relation to trauma and to evaluate whether distinct ways of thinking about traumatic experiences have distinct correlates in terms of psychopathology and functioning.
The present study
The first goal of the present study was to evaluate, by means of a latent profile analysis (LPA), whether distinct profiles of impairments in mentalizing trauma could be identified in a community sample of adults exposed to childhood abuse and neglect. We hypothesized that we would find significant variability among survivors of childhood trauma and that this heterogeneity would reflect qualitatively distinct ways of processing traumatic experiences, that would bear significance for clinical practice and empirical research.
In a second step, we aimed to evaluate whether these heterogeneous profiles were marked by significant differences in internalized (anxiodepressive and post-traumatic stress disorder [PTSD] symptoms) and externalized/relational (anger, intimate partner violence) problems, personality dysfunctions, and antenatal attachment (Table 1). The LPA performed in step 1 identified five distinct profiles of respondents: Profile 1, Identified with the perpetrator; Profile 2, Functionally grandiose; Profile 3, Absorbed in trauma; Profile 4, No impairment in mentalizing trauma; and Profile 5, Global mentalization impairments in relation to trauma. Based on the existing literature, we had different hypotheses for the associations between each of these profiles and external variables. First, given that participants in Profile 1 (Identified with the perpetrator) were characterized by a higher propensity to value aggression and rationalize mean behaviors, we hypothesized that this subgroup would be marked by high levels of externalized and relational problems while showing little internalized problems, as suggested by studies reporting positive associations between perceived acceptability of interpersonal conflicts or aggression and intimate partner violence (Fincham et al., Reference Fincham, Cui, Braithwaite and Pasley2008; Gracia et al., Reference Gracia, Rodriguez and Lila2015) and recent findings linking higher perceived acceptability of childhood maltreatment to lower symptoms of PTSD (Bartoli et al., Reference Bartoli, Wadji, Oe, Cheng, Martin-Soelch, Pfaltz and Langevin2024). Second, since participants in Profile 2 (Functionally grandiose) were characterized by a perception of themselves as being impermeable to the effects of trauma, we expected them to report low levels of subjective distress and alterations in functioning, as it is generally the case for adults with dismissing attachment representations who share with Profile 1 participants a tendency to cut themselves off from negative experiences (Barazzone et al., Reference Barazzone, Santos, McGowan and Donaghay-Spire2019; Martin et al., Reference Martin, Bureau, Lafontaine, Cloutier, Hsiao, Pallanca and Meinz2017; Murphy & Bates, Reference Murphy and Bates1997). Third, since Profile 3 (Absorbed in trauma) participants displayed specific mentalizing impairments (interference of memories of trauma with the regulation of thought, affect and behaviors, depersonalization, and avoidance of thoughts) that evoke some of the core characteristics of PTSD (Moser et al., Reference Moser, Suardi, Schechter, Spalleta, Janiri, Piras and Sani2020), we expected these participants to be characterized by high levels of PTSD symptoms. Fourth, in line with previous studies about the protective role of mentalizing trauma for mental health and parental functioning (Berthelot, Savard, et al., Reference Berthelot, Savard, Lemieux, Garon-Bissonnette, Ensink and Godbout2022; Ensink et al., Reference Ensink, Berthelot, Bernazzani, Normandin and Fonagy2014; Gamache et al., Reference Gamache, Savard, Leclerc, Payant, Berthelot, Cote, Faucher, Lampron, Lemieux, Mayrand, Nolin and Tremblay2021), we hypothesized that Profile 4 (No impairment) participants would show preserved functioning across domains. Contrarily, we expected Profile 5 (Global impairments) participants to present the most severe and widespread symptoms and alterations in functioning.
Comparisons between latent profiles on anxiodepressive symptoms, PTSD symptoms, anger, intimate partner violence, personality dysfunctions, and antenatal attachment
Note. M = estimated marginal means; SE = standard error. BCH = Bolck, Croons, and Hagenaarsé. Sx = symptoms. IPV = intimate partner violence. PD = personality dysfunctions. NS = not significant.
a Age, income, and education were entered as covariates.
b * p < .05; ** p < .01; *** p < .001.
Methods
Participants and procedure
A total of 872 pregnant women having been exposed to childhood trauma according to the validated cut-offs of the Childhood Trauma Questionnaire (see below; Bernstein et al., Reference Bernstein, Stein, Newcomb, Walker, Pogge, Ahluvalia, Stokes, Handelsman, Medrano, Desmond and Zule2003) were recruited using two strategies in the Province of Quebec, Canada. First, 412 women were recruited during their first pregnancy monitoring appointment between April 2018 and January 2023. Second, 460 pregnant women were recruited online through social media advertisements in April 2020. Participants completed self-reported questionnaires during the second or third trimester of pregnancy. Inclusion criteria were being 18 years old or older, having sufficient reading skills in French to complete self-reported assessments, being currently pregnant and having experienced at least one type of abuse or neglect before 18 years old. Of the 876 participants, 47 had incomplete data on all scales of the FMTQ (our main variable of interest) and were thus excluded. Sociodemographic characteristics for the final sample of 825 pregnant women are presented in Table 2. All participants provided written informed consent and studies received ethical approval from our Regional Health Center (CER-2016-016-11) and our University (CER-16-226-10; CER-20-266-10).
Sample characteristics
Note. N = 825. Mean age = 29.59, SD = 4.99.
Measures
Impairments in the mentalization of trauma
Current problems in the way people think of or deal with trauma and adverse relationships were assessed using the French version of the FMTQ (Berthelot, Savard, et al., Reference Berthelot, Savard, Lemieux, Garon-Bissonnette, Ensink and Godbout2022). During the assessment, participants are invited to recall instances of adverse relationships where they felt intense negative emotions, such as betrayal, hurt, abandonment, feeling used or disrespected, fear, or being overwhelmed. Responses are rated on a 5-point Likert scale from 0 (completely disagree) to 4 (completely agree). Higher scores indicate more severe disruptions in mentalizing trauma. The FMTQ comprises seven subscales reflecting specific indicators of mentalizing impairments in relation to trauma. Internal consistency, after applying the Spearman-Brown prophecy formula that allows estimating the reliability of subscales including a low number of items if the number of items was double, was adequate (S-Bα = .70–84).
Childhood trauma
Childhood trauma was evaluated using the French version of the Childhood Trauma Questionnaire (CTQ-28; Bernstein et al., Reference Bernstein, Stein, Newcomb, Walker, Pogge, Ahluvalia, Stokes, Handelsman, Medrano, Desmond and Zule2003; Paquette et al., Reference Paquette, Laporte, Bigras and Zoccolillo2004). This self-report comprises 28 items and examines five types of interpersonal trauma before the age of 18: physical, psychological, and sexual abuse as well as physical and psychological neglect. Participants are asked to rate each item on a 5-point Likert scale, ranging from 1 (never true) to 5 (always true), with higher scores reflecting more severe trauma. Specific cut-offs are validated for each subscale (physical abuse ≥ 8, psychological abuse ≥ 10, sexual abuse ≥ 8, physical neglect ≥ 8 and psychological neglect ≥ 15; Walker et al., Reference Walker, Unutzer, Rutter, Gelfand, Saunders, VonKorff, Koss and Katon1999). Participants are categorized as having experienced childhood trauma if they reach the cut-off on at least one scale. The Cronbach’s alpha for the total score of the CTQ in this study was of α = .81.
Internalized symptoms
Anxiety and depressive symptoms were measured using the French 10-item version of the Kessler Psychological Distress Scale (K10; Gravel et al., Reference Gravel, Connolly and Bédard2003; Kessler et al., Reference Kessler, Andrews, Colpe, Hiripi, Mroczek, Normand, Walters and Zaslavsky2002). Higher scores at the K10 reflect more anxiety and depressive symptoms. A cut-off of ≥ 30 was used since 76.3% of respondents with such elevated scores would meet criteria for a DSM-IV mood, anxiety, or substance use disorder during a diagnostic interview (Andrews & Slade, Reference Andrews and Slade2001). Both the English and French versions have similarly satisfactory psychometric properties (Gravel et al., Reference Gravel, Connolly and Bédard2003). The Cronbach’s alpha for the K-10 in this study was α = .86.
Past-month post-traumatic stress symptoms were assessed using the PTSD Checklist for DSM-5 (Post-traumatic Checklist for DSM-5, PCL-5; Weathers et al., Reference Weathers, Litz, Keane, Palmieri, Marx and Schnurr2013). The PCL-5 has 20 items rated on a 5-point Likert scale ranging from 0 (not at all) to 4 (always). Higher scores reflect more severe symptoms, with a clinical cut-off set at ≥ 33 (Weathers et al., Reference Weathers, Litz, Keane, Palmieri, Marx and Schnurr2013). Both the French and the original versions have equally adequate validity and reliability (Ashbaugh et al., Reference Ashbaugh, Houle-Johnson, Herbert, El-Hage and Brunet2016). The Cronbach’s alpha for the PCL-5 in this study was α = .92.
Externalized symptoms and relational problems
Current intensity of angry feelings and expression of anger was assessed using the French version of the State Anger scale of the State-Trait Anger Expression Inventory-2 (STAXI-2; Borteyrou et al., Reference Borteyrou, Bruchon-Schweitzer and Spielberger2008; Spielberger, Reference Spielberger1999). This specific scale comprises 15 items. Responses are rated on a 4-point Likert scale ranging from 1 (almost never) to 4 (almost always), with higher scores reflecting higher state anger. The Cronbach’s alpha for the State Anger scale in this study was α = .94.
Psychological and physical violence victimization and perpetration during the past year was measured using a 24-item French version of the Revised Conflict Tactics Scale (CTS-2; Godbout et al., Reference Godbout, Daspe, Lussier, Sabourin, Dutton and Hébert2017; Lussier, Reference Lussier1997; Straus et al., Reference Straus, Hamby, Boney-McCoy and Sugarman1996). Responses are rated on an 8-point frequency scale, ranging from 0 (never occurred in the past year) to 6 (more than 20 times in the last year), whereas a score of 7 signifies that the behavior did not occur in the past year but has happened previously. This score was given a zero since the present study aimed to assess current (past year) relational difficulties. A higher score at the CTS-2 reflects higher frequency of victimization and perpetration of psychological and physical violence within the intimate partner relationship. The CTS-2 demonstrated good reliability and validity across various nonclinical samples of adults (Chapman & Gillespie, Reference Chapman and Gillespie2019; Straus et al., Reference Straus, Hamby, Boney-McCoy and Sugarman1996). The Cronbach’s alpha for both scales of the CTS-2 in this study was α = .70.
Personality dysfunction
Personality dysfunctions as defined by Criterion A of the alternative dimensional model of the DSM-V were assessed using the French version of the Self and Interpersonal Functioning Scale (SIFS; Gamache et al., Reference Gamache, Savard, Leclerc and Côté2019). The SIFS has 24 items rated on a 5-point Likert scale ranging from 0 (this does not describe me at all) to 4 (this totally describes me). Higher scores reflect higher personality dysfunctions. In this study, we relied on the two-factor solution of the SIFS assessing self-impairments (i.e., self-direction and identity) and interpersonal dysfunctions (i.e., empathy and intimacy) as well as on the clinical cut-off of ≥ 1.30, indicative of a probable personality disorder characterized as mildly severe according to Gamache et al. (Reference Gamache, Savard, Leclerc, Payant, Berthelot, Cote, Faucher, Lampron, Lemieux, Mayrand, Nolin and Tremblay2021). The SIFS shows good validity across samples (Gamache et al., Reference Gamache, Savard, Leclerc and Côté2019; Gamache et al., Reference Gamache, Savard, Leclerc, Payant, Berthelot, Cote, Faucher, Lampron, Lemieux, Mayrand, Nolin and Tremblay2021; Waugh et al., Reference Waugh, McClain, Mariotti, Mulay, DeVore, Lenger, Russell, Florimbio, Lewis, Ridenour and Beevers2021). The Cronbach’s alpha for the interpersonal dysfunction and self-impairment scales of the SIFS in this study were respectively α = .78 and α = .81.
Antenatal attachment
Prenatal psychological investment toward the unborn child and commitment to the pregnancy was assessed using the Maternal Antenatal Attachment Scale (MAAS; Condon, Reference Condon1993). Responses on the 19 items are rated on a variable 5-point Likert scale. Higher scores reflect greater investment and commitment towards the fetus and pregnancy. The Maternal Antenatal Attachment Scale yields two subscales. In the present study, we focused on the Quality subscale measuring the strength of the emotional bond with the fetus. The instrument has good psychometric properties (Condon, Reference Condon1993). The Cronbach’s alpha for the quality of attachment subscale in this study was of α = .75.
Analytic strategy
In a first step, to identify distinct profiles of mentalizing impairments in relation to trauma, LPAs were conducted using MPlus 8.4 (Muthén & Muthén, Reference Muthén and Muthén2017). Standardized scores of the seven factors of the FMTQ were used. Distributions were truncated at the 98th percentile. Solutions yielding between 2 and 7 profiles were analyzed and compared using the Bayesian information criteria (BIC; i.e., reflecting the parsimony of the model), the entropy (i.e., evaluating the proportion of correct classification within each profile), and the Vuong-Lo-Mendell-Rubin Likelihood Ratio Test (VLMR-LRT; i.e., indicating the loss of fit associated with the removal of profiles). In LPA, the lowest BIC, higher entropy (i.e., value closest to 1) and a significant VLMR-LRT are deemed to indicate the better fitting model, whereas all profiles should include at least 5% of the sample.
In a second step, we first used the DE3STEP command in Mplus to evaluate differences between latent profiles identified in step 1 on potential covariates to control for in further analyses. Associations between latent profiles and exogenous variables (severity of childhood trauma; anxiodepressive symptoms; PTSD symptoms; anger; intimate partner violence; personality dysfunctions; and antenatal attachment) were next evaluated using the manual Bolck, Croons, and Hagenaars method (BCH; Bolck et al., Reference Bolck, Croon and Hagenaars2004). This method allows for the examination of statistically significant mean-level differences by treating exogenous variables as distal variables. It is generally preferred to the three-step method as it allows for the inclusion of control variables, is more robust, and is less perturbed by inequality of variances across latent profiles (Bakk & Vermunt, Reference Bakk and Vermunt2016). This approach also considers each participant’s individual error rate instead of the sample’s average classification error, allowing the capture of an imprecise profile assignment when examining exogenous variables across latent profiles (Asparouhov & Muthén, Reference Asparouhov and Muthén2014; Nylund-Gibson et al., Reference Nylund-Gibson, Grimm and Masyn2019). Finally, the risk (measured using odds ratios) of reaching the clinical cut-off of a probable anxiodepressive disorder, PTSD disorder or personality disorder was calculated for each profile with Profile 4 as category of reference.
Results
Intercorrelations between the subscales of the FMTQ were moderate to high (range .17–.53), showing no sign of collinearity (STable 1 in the electronic supplement). The seven tested solutions are displayed in Table 3. According to Akaike information criterion, BIC, entropy and VLMR-LRT, a 6-profile solution appeared as the best fitting model within the current sample. However, after consideration of the number of participants in each profile, model parsimony, and conceptual meaningfulness, we retained the 5-profile solution. Indeed, as shown in SFigure 1 (see electronic supplement), the 6-profile solution included two profiles with small sample sizes (44 and 64 participants respectively) and some profiles did not differ in a meaningful way. This decision was further supported by the fact that both solutions (5-profile and 6-profile) showed very similar fit indices (Table 3).
Latent profile analysis for solutions 1 through 7 using the seven subscales of the failure to mentalize trauma questionnaire as latent indicators
Note. AIC = Akaike information criteria; BIC = Bayesian information criteria.
The five profiles were labeled Identified with the perpetrator (Profile 1); Functionally grandiose (Profile 2); Absorbed in trauma (Profile 3); No impairment in mentalizing trauma (Profile 4); and Global mentalization impairments in relation to trauma (Profile 5). As shown in Figure 1, Profile 1 participants (Identified with the Perpetrator) were indexed (± .30) by high scores on the Identification with the perpetrator and Justification of trauma subscales of the FMTQ. Profile 2 participants (Functionally grandiose) had high scores on the Grandiosity subscales of the FMTQ and low scores on the Identification with the perpetrator and Disorganization of thoughts subscales. Profile 3 participants (Absorbed in trauma) were especially characterized by high scores on the Identification with the victim, Disorganization of thoughts, and Absorption in trauma subscales of the FMTQ and reported some levels of Grandiosity and Avoidance. Profile 4 participants (No impairment in mentalizing trauma) had low scores on the seven subscales of the FMTQ, whereas Profile 5 participants (Global mentalization impairments in relation to trauma) had elevated scores on all subscales.
Latent profile description using the seven latent indicators from the failure to mentalize trauma questionnaire. Victim = identification with the victim; perpet = identification with the perpetrator; thoughts = disorganization of thoughts; grandios = grandiosity; avoid = avoidance of thoughts; absorb = absorption in trauma; justificat = justification of trauma.
The DE3STEP command revealed significant differences between profiles on age, annual income and education (STable 2, electronic supplement). We thus included these covariates in all further analyses. BCH analyses showed significant differences between profiles on outcome variables (Table 4). As shown in Figure 2, Profile 1 (Identified with the Perpetrator) participants were characterized by high levels of anger and a high likelihood of being involved in bidirectionally violent relationships. Profile 3 (Absorbed) participants showed the highest level of trauma and were characterized by very high levels of PTSD symptoms and some personality dysfunctions, namely in terms of persisting relational problems. They were however very unlikely to show externalized problems such as anger and violence. Profile 5 (Global impairments) participants were not particularly at risk of internalization (anxiodepressive of PTSD symptoms) but displayed the highest levels of anger, intimate partner violence, interpersonal problems, and self-impairments. In contrast, the participants belonging to Profile 2 (Functionally grandiose) and 4 (No impairment) showed very little symptoms and dysfunctions across domains.
Standardized outcome variables across latent profiles.
Summary of participants belonging to each profile, including typical excerpts of narratives about their experience of childhood trauma
a z-score values that extend beyond ±.30 are considered as characteristics features of the profile;
b Excerpts of narratives that illustrate the distinctive impairments in mentalizing trauma presented by participants of each profile were extracted from Trauma Meaning-Making Interviews and Adult Attachment Interviews narratives collected by the authors. Excerpts were expressly chosen to provide illustrations of the specific types of mentalizing impairments in relation to trauma observed in each latent profile.
These results were further supported by categorical analyses relying on previously validated clinical cut-offs indicating probable anxiodepressive disorder, PTSD, and personality disorder. As shown in Table 5, in comparison to Profile 4 (No impairment) participants, Profile 3 (Absorbed) participants were at high risk of probably suffering from any of the three disorders, and 44.6% (OR = 13.44) reached the cut-off for a PTSD. Profile 5 (Global impairments) participants were similarly at risk for the three disorders and more than half (52.5%, OR = 40.55) were classified as possibly presenting a personality disorder. Odds ratios were inferior for Profile 1 and 2 participants, while remaining significant (between 2.12 and 5.53) for PTSD and personality disorders, especially for Profile 2 (Identified with the perpetrator) participants.
odds ratio of reaching the clinical cut-offs of a probable anxiodepressive disorder, PTSD disorder or a personality disorder of each profile of the latent profile analysis in comparison to participants without impairments in the mentalization of trauma (Profile 4)
Note. Odds ratios were computed using Profile 4 participants (No impairment in the mentalization of trauma) as reference. Age, annual income, and education were entered as covariates. The cut-off for a probable anxiodepressive disorder (K10) was of 30 (Andrews & Slade, Reference Andrews and Slade2001). The cut-off for a probable PTSD (PCL-5) was of 33 (Weathers et al., Reference Weathers, Litz, Keane, Palmieri, Marx and Schnurr2013). The cut-off for a probable mild severity personality disorder (SIFS) was of 1.30 (Gamache et al., Reference Gamache, Savard, Leclerc, Payant, Berthelot, Cote, Faucher, Lampron, Lemieux, Mayrand, Nolin and Tremblay2021). Among Profile 4 participants, 12.8% possibly had psychiatric disorder, 7% met the cut-off for a probable anxiodepressive disorder, 11% for a probable PTSD disorder and 4,8% for a possible personality disorder.
Discussion
The study unraveled interindividual variability in trauma-resolution processes among adults who experienced childhood maltreatment. Yet, we were able to identify five types of psychological response to trauma, each manifesting throughout a specific set of symptoms and dysfunctions. Specifically, we identified five distinct profiles of disruptions in the mentalization of trauma: Identified with the perpetrator (Profile 1), Functionally grandiose (Profile 2), Absorbed in trauma (Profile 3); No impairment in mentalizing trauma (Profile 4), and Global mentalization impairments in relation to trauma (Profile 5). Whereas profiles 1, 3 and 5 participants presented specific patterns of symptoms and dysfunctions (see Table 4 and Fig. 2), profiles 2 and 4 participants displayed very few negative outcomes. Interestingly, the vast majority of adults exposed to childhood trauma in our community sample were classified in one of the latter profiles (n = 538; 65%). This encouraging proportion suggests that the most common pattern of response in the face of trauma might be one of resilience, a finding that is consistent with previous observations in youths (Cahill et al., Reference Cahill, Hager and Shryane2023; Martinez-Torteya et al., Reference Martinez-Torteya, Anne Bogat, Von Eye and Levendosky2009; Yoon et al., Reference Yoon, Pei, Logan, Helsabeck, Hamby and Slesnick2023) and adults (Itzhaky et al., Reference Itzhaky, Gelkopf, Levin, Stein and Solomon2017). However, Profile 2 participants, who are characterized by a perception of themselves as being somewhat invulnerable to trauma, are intriguing and call for further research, namely in terms of parenting. Indeed, in the same way as adults who have dismissing attachment representations, these participants may have developed ways to regulate trauma-related mental states that preserve them from experiencing significant distress and symptoms (Barazzone et al., Reference Barazzone, Santos, McGowan and Donaghay-Spire2019; Martin et al., Reference Martin, Bureau, Lafontaine, Cloutier, Hsiao, Pallanca and Meinz2017; Murphy & Bates, Reference Murphy and Bates1997), but that may lead to unsupportive or insensitive caregiving when their child expresses vulnerability or evokes such feelings in themselves. Future research will be needed to evaluate how caregivers in each profile may differ in terms of caregiving behaviors.
Profile 1 participants (Identified with the Perpetrator) consisted of adults who emerged out of maltreating relationships by defensively identifying with hostile attachment figures in an attempt to cope with the fear evoked by trauma (Howell, Reference Howell2014). As we may have expected based on theoretical grounds, these participants reported high levels of anger and were particularly at risk of committing and suffering physical and emotional violence in their partner relationship. Profile 3 participants (Absorbed) were particularly characterized by three types of mentalization impairments in relation to trauma: (1) a tendency to take responsibility for trauma or to consider that abusive behaviors were deserved (measured through the Identification with the victim subscale of the FMTQ), (2) severe problems in the monitoring of reasoning under the forms of depersonalization and destructive behaviors when experiencing trauma-related emotions (measured through the Disorganization of thoughts subscale of the FMTQ), and (3) interference of memories of trauma with the monitoring of thoughts and behaviors (measured through the Absorption subscale of the FMTQ). This specific cluster of disruptions in the mentalization of trauma evokes the PTSD symptoms of Intrusion (Cluster B) and Negative alterations in cognitions and mood (Cluster D). Correspondingly, these participants were indexed by very high levels of PTSD symptoms and some personality dysfunctions, namely in terms of persisting interpersonal problems. They were also the most severely exposed to trauma. However, in contrast to Profile 1 and 5 participants, they displayed very little externalized problems. Finally, Profile 5 participants showed high scores across all subscales of the FMTQ, reflecting widespread impairments in the mentalization of trauma. They were correspondingly the most likely to present high levels of symptoms and dysfunctions.
Analyses using categorical outcomes further illustrated the strong association between psychological processing of trauma and outcomes. Indeed, in comparison to participants for whom trauma-specific mentalization was not compromised (Profile 4), the specific combination of mentalizing impairments that characterized participants of Profile 3 (Absorbed) was associated with a 12.98-fold increased risk of possibly suffering from a psychiatric disorder during pregnancy, whereas the risk was increased 23.61-fold in participants presenting global and extensive mentalization impairments (Profile 5).
Interestingly, the distinct profiles were not associated with a self-reported assessment of the quality of antenatal attachment. This finding is at odds with previous studies using observational measures of parenting during the first two years following childbirth showing that compromised mentalization in relation to trauma was associated with more hostility toward the child, lower sensitivity to the child’s emotional communication, and prospectively predicted disorganized mother-infant attachment relationships, even when considering the effect of important covariates such as the severity of abuse, unresolved/disorganized attachment representations and parental reflective functioning (Berthelot et al., Reference Berthelot, Ensink, Bernazzani, Normandin, Luyten and Fonagy2015; Berthelot et al., Reference Berthelot, Garon-Bissonnette, Muzik, Simon, Menke, Stacks and RosenblumIn Press; Ensink et al., Reference Ensink, Berthelot, Bernazzani, Normandin and Fonagy2014). Our finding of an absence of association between mentalizing impairments in relation to trauma and the quality of maternal-fetal attachment is however in line with the previously documented absence of a direct association between a history of childhood trauma in pregnant women and expecting men and the quality of their thoughts and feelings regarding the child-to-be (Berthelot et al., Reference Berthelot, Lemieux and Maziade2019; Hinesley et al., Reference Hinesley, Amstadter, Sood, Perera, Ramus and Kornstein2020; Sancho-Rossignol et al., Reference Sancho-Rossignol, Schilliger, Cordero, Rusconi Serpa, Epiney, Hüppi, Ansermet and Schechter2018). One possibility is that caregivers who show distortions in the way they reflect about their past traumas may have an idealized picture of their future relationship with their child and not fully recognize that they may encounter challenges as future parents when reporting about this relationship during the prenatal period. These challenges may become more apparent when children are born and especially when they are at the age of expressing hostile aggression triggering maternal trauma (Moser et al., Reference Moser, Graf, Glaus, Urben, Jouabli, Pointet Perrizolo, Suardi, Robinson, Rusconi Serpa, Plessen and Schechter2023; Suardi et al., Reference Suardi, Rothenberg, Serpa and Schechter2017). This would further justify the need to intervene with this specific subgroup of parents with histories of trauma before childbirth. We cannot exclude however the possibility that different findings would have been observed using other measures of antenatal bonding or maternal representations, as suggested by previous evidence linking maternal trauma to lower time spent in attachment mode (Sancho-Rossignol et al., Reference Sancho-Rossignol, Schilliger, Cordero, Rusconi Serpa, Epiney, Hüppi, Ansermet and Schechter2018) and disrupted prenatal representations of the child (Ahlfs-Dunn et al., Reference Ahlfs-Dunn, Benoit and Huth-Bocks2022). Further research will be needed to understand the determinants and correlates of antenatal attachment and maternal representations in pregnant women who experienced childhood trauma.
Whereas the cross-sectional correlational design does not permit us to conclude about the direction of the association between profiles of trauma processing and external variables, the idea that intrapsychic conflicts, cognitive processing of experiences and representations of self and others underly symptoms and dysfunctions is central to developmental and psychological models of psychopathology (Cicchetti, Reference Cicchetti1991; Luyten et al., Reference Luyten, Campbell, Allison and Fonagy2020) as well as to most psychotherapeutic approaches (Brent & Kolko, Reference Brent and Kolko1998) and would suggest that compromised mentalizing in relation to trauma contributes to poor adaptation, rather than being a mere correlate of psychopathology. Accordingly, the findings may provide leads for psychotherapeutic interventions. First, identifying specific areas where individuals struggle to mentalize trauma may help clinicians tailor personalized treatment plans. Second, recognizing and discussing these disruptions with patients contribute to seeing beyond the symptoms and can empower them to gain insight into their own difficulties and engage more actively in the healing process. Finally, awareness of these disruptions can help clinicians to be more attuned to their patients’ needs and emotions, which can strengthen the therapeutic alliance, improve outcomes, and prevent premature termination of treatment. For instance, using the FMTQ may enable clinicians to assess efficiently, in the first stages of a consultation process, a patient’s specific struggles in mentalizing trauma and adjust the intervention consequently. As a case in point, when trauma is evoked in the therapeutic relationship, clinicians should be aware that Profile 1 patients may react with hostility, Profile 2 patients may deny any sense of vulnerability, Profile 3 patients may be overwhelmed by feelings of shame and momentarily lose their capacity to monitor thoughts and regulate emotions, and Profile 5 patients may become highly dysregulated. All these reactions pose important threats to the therapeutic alliance if clinicians are not prepared to deal with such trauma responses.
This study has several strengths including the use of a large sample of over 800 women who experienced childhood maltreatment, the use of well-validated instruments, and its unique focus on interindividual variability in mentalization processes in the context of trauma, thus offering valuable insights and paving the way for further research in this critical area (Ensink et al., Reference Ensink, Bégin, Martin-Gagnon, Biberdzic, Berthelot, Normandin, Fonagy, Bernazzani and Borelli2015, Reference Ensink, Normandin, Target, Fonagy, Sabourin and Berthelot2023; Lorenzini et al., Reference Lorenzini, Campbell, Fonagy and Huppertz2018; Luyten & Fonagy, Reference Luyten, Fonagy, Bateman and Fonagy2019). Despite the strengths of this study, several limitations should be acknowledged which may impact the generalizability and interpretation of the findings. First, the exclusive reliance on self-reported measures for assessing childhood maltreatment, mentalization processes and outcomes may have led to recall and social desirability biases. Further studies using clinical interviews, observational measures (for instance of parenting) and interview-based assessments of trauma-specific reflective functioning will be required to fully capture the nuances of the complex psychological phenomena at play. Second, despite the strong theoretical and clinical grounds of our a priori hypotheses, the cross-sectional nature of this study restricts our ability to make causal inferences. Longitudinal research starting in childhood/adolescence will be necessary to better understand the developmental trajectories of mentalization in individuals with a history of childhood trauma and its role in the onset of psychopathology and dysfunctions. Finally, this study was conducted on a large community sample of female participants, potentially limiting the generalizability of our findings to the broader population. It is crucial to acknowledge that males and females may differ in their experiences of childhood maltreatment and their responses to trauma (Berthelot, Garon-Bissonnette, et al., Reference Berthelot, Garon-Bissonnette, Jomphe, Doucet-Beaupré, Bureau and Maziade2022; Helpman et al., Reference Helpman, Zhu, Suarez-Jimenez, Lazarov, Monk and Neria2017). Future research should aim for balanced sex and gender representations to ensure a more comprehensive understanding of the studied phenomena.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0954579424001019.
Acknowledgments
We thank the mothers and fathers who made this research possible and gratefully acknowledge the contributions of Roxanne Lemieux, Annie Lemieux, Sylvie Moisan, Christine Drouin-Maziade, Vanessa Bergeron, Gabrielle Duguay, and Sabrina Bernier.
Funding statement
This work was supported by the Social Sciences and Humanities Research Council of Canada (grant #430-2017-00994, Principal Investigator, N.B.), the Canada Research Chair in Developmental Trauma (grant #950-232739, Principal Investigator, N.B.), and the Public Health Agency of Canada (grant #1617-HQ-000015, Principal Investigator, N.B.).
Competing interests
None.
Open access
