The relationship between multiple traumatic events and the severity of posttraumatic stress disorder symptoms – evidence for a cognitive link

ABSTRACT

Background: Previous research has shown that multiple traumatic experiences cumulatively increase the risk for the development of severe symptoms of posttraumatic stress disorder (PTSD). Yet, little is known about the specific psychological mechanism through which this increased risk comes about.

Objective: In the present study, we examined a possible cognitive link between multiple traumatic events and PTSD symptom severity through dysfunctional cognitions and expectations.

Methods: A sample of patients with a diagnosed PTSD (N = 70; M_Age= 42.06; 82% female) and high symptom burden (IES-R M = 79.24) was examined. On average, patients had experienced 5.31 different traumatic events. In a structural equation model, we tested the hypothesis that the relationship between multiple traumatic experiences and PTSD symptom severity is mediated through dysfunctional general cognitions and dysfunctional situation-specific expectations. General trauma-related cognitions were assessed with the Posttraumatic Cognition Inventory (PTCI) and trauma-related situational expectations were assessed with the Posttraumatic Expectations Scale (PTES).

Results: The direct effect of the number of traumatic events on PTSD symptom severity was non-significant. Instead, as hypothesised, there was evidence for a significant indirect effect via dysfunctional general cognitions and situation-specific expectations.

Conclusions: The current results further specify the cognitive model of PTSD by indicating that the relationship between the number of traumatic events and PTSD symptom severity is mediated through dysfunctional cognitions and expectations. These findings emphasise the importance of focused cognitive treatment approaches that seek to modify dysfunctional cognitions and expectations in people with multiple traumatic experiences.

HIGHLIGHTS

• This study shows a cognitive link between the experience of multiple traumatic events and the severity of posttraumatic stress disorder (PTSD) symptoms.

• The aforementioned relationship is mediated through trauma-related general cognitions and situational expectations.

• The results provide further evidence for the cognitive model of PTSD and further specify it by considering different types of trauma-related cognitions.

Antecedentes: Investigaciones anteriores han demostrado que múltiples experiencias traumáticas aumentan acumulativamente el riesgo de desarrollar síntomas graves de trastorno de estrés postraumático (TEPT). Sin embargo, se sabe poco sobre el mecanismo psicológico específico a través del cual se produce este aumento del riesgo.

Objetivo: En el presente estudio, examinamos un posible vínculo cognitivo entre múltiples eventos traumáticos y la gravedad de los síntomas del TEPT a través de cogniciones y expectativas disfuncionales.

Método: Se examinó una muestra de pacientes con TEPT diagnosticado (N = 70; M_edad = 42,06; 82% mujeres) y alta carga de síntomas (M_IES-R= 79,24). En promedio, los pacientes habían experimentado 5,31 eventos traumáticos diferentes. En un modelo de ecuaciones estructurales, probamos la hipótesis de que la relación entre experiencias traumáticas múltiples y la gravedad de los síntomas de TEPT está mediada por cogniciones generales disfuncionales y expectativas disfuncionales específicas de la situación. Las cogniciones generales relacionadas con el trauma se evaluaron con el Inventario de Cognición Postraumática (PTCI en su sigla en inglés) y las expectativas situacionales relacionadas con el trauma se evaluaron con la Escala de Expectativas Postraumáticas (PTES en su sigla en inglés).

Resultados: El efecto directo del número de eventos traumáticos sobre la gravedad de los síntomas del TEPT no fue significativo. En cambio, como se planteó como hipótesis, hubo evidencia de un efecto indirecto significativo a través de cogniciones generales disfuncionales y expectativas específicas de la situación.

Conclusiones: Los resultados actuales especifican aún más el modelo cognitivo del TEPT al indicar que la relación entre el número de eventos traumáticos y la gravedad de los síntomas del TEPT está mediada por cogniciones y expectativas disfuncionales. Estos hallazgos enfatizaron la importancia de los enfoques de tratamiento cognitivo enfocado que buscan modificar las cogniciones y expectativas disfuncionales en personas con múltiples experiencias traumáticas.

背景：前人研究表明，多次创伤经历累积起来会增加出现创伤后应激障碍 (PTSD) 严重症状的风险。 然而，对于这种风险增加的具体心理机制知之甚少。

目的：在本研究中，我们通过功能失调的认知和期望考查了多重创伤事件与 PTSD 症状严重程度之间可能存在的认知联系。

方法：考查了诊断为 PTSD（N = 70；平均年龄 = 42.06；82% 为女性）和高症状负担 (IES-R M = 79.24) 的患者样本。 平均而言，患者经历了 5.31 种不同的创伤事件。 在结构方程模型中，我们检验了以下假设：多重创伤经历与 PTSD 症状严重程度之间的关系是通过功能失调的一般认知和功能失调的特定情境预期中介的。 使用创伤后认知量表 (PTCI) 评估创伤相关的一般认知，并使用创伤后期望量表 (PTES) 评估创伤相关的情境期望。

结果：创伤事件数量对 PTSD 症状严重程度的直接影响不显著。 相反，正如假设一样，有证据表明通过功能失调的一般认知和特定情境预期会产生显著间接影响。

结论：当前的结果进一步明确了 PTSD 的认知模型，表明创伤事件的数量与 PTSD 症状严重程度之间的关系是通过功能失调的认知和期望中介的。 这些发现强调了旨在改变具有多重创伤经历者功能失调认知和期望的集中认知治疗方法的重要性。

KEYWORDS:

• Posttraumatic stress disorder
• expectation
• cognitions
• belief
• appraisal
• structural equation model

PALABRAS CLAVE:

• Trastorno de estrés postraumático
• expectativa
• cogniciones
• creencia
• evaluación
• modelo de ecuaciones estructurales

关键词:

• 创伤后应激障碍
• 期望
• 认知
• 信仰
• 评估
• 结构方程模型

1. Introduction

Many people suffering from a posttraumatic stress disorder (PTSD) have experienced multiple traumatic events rather than a single traumatic event (Kilpatrick et al., 2013). A substantial body of literature suggests that the experience of multiple traumatic events dramatically increases the risk for the severity and chronicity of PTSD symptoms as compared to a single traumatic event (Breslau et al., 1999; Brewin et al., 2000; Gold et al., 1994; Gonzalez et al., 2016; Park et al., 2015). For instance, research has shown that individuals with multiple traumatic events have more severe PTSD symptoms (Follette et al., 1996; Scott, 2007), especially dissociative symptoms, intense feelings of guilt and shame, and interpersonal sensitivity (Hagenaars et al., 2011). Specifying the relationship between the frequency of traumatic events and PTSD, a large study found the number of traumatic events to be linearly associated with the severity of PTSD symptoms, even when controlling for gender, stressful life experiences in the past year, and childhood adversity (Suliman et al., 2009). Similar results have been provided by other research (Cloitre et al., 2009; Gerber et al., 2018; Green, 1994). A population based survey further showed that people who had experienced four or more traumatic events had greater functional impairment, an earlier age of PTSD onset, longer PTSD symptom duration, higher comorbidity with mood and other anxiety disorders, and other psychopathological dysfunctions than people with fewer traumatic events (Karam et al., 2014).

While there is sound evidence of a linear relationship between the number of traumatic events and the severity of PTSD symptoms, relatively little is known so far about the specific mechanism through which this relationship comes about. Better understanding this link would be valuable as it would offer the possibility of targeting the very factors that are related to PTSD symptom severity in the treatment of people with multiple traumas. With reference to the influential cognitive model of PTSD (Ehlers & Clark, 2000), we sought to investigate in the present study whether different types of dysfunctional cognitions explain the relationship between multiple traumatic events and PTSD symptom severity.

Cognitive approaches to PTSD argue that the experience of a traumatic event entails a dramatic change in people’s beliefs of the world, other people, and themselves (Bernardi et al., 2019; Brown et al., 2019; LoSavio et al., 2017). In their influential account, Ehlers and Clark (2000) proposed that people with PTSD differ from trauma survivors without PTSD in their interpretation of the trauma and/or its sequelae, resulting in a persistent sense of threat. In addition, Ehlers and Clark (2000) suggested that traumatic experiences can also lead to a sense of threat through maladaptive memory processing. Numerous studies have provided evidence in support of this cognitive model of PTSD, for example by showing that negative appraisals of the traumatic event and its consequences are associated with – and predictive of – PTSD symptoms (Ehring et al., 2008; Kleim et al., 2013; ter Heide et al., 2017; Woud et al., 2019).

Recently, it has been suggested that expectations might be a particularly important subgroup of cognitions for the understanding of mental disorders (Kube & Rozenkrantz, 2021; Rief et al., 2015; Rief & Glombiewski, 2016) and PTSD in particular (Kube et al., 2020; Linson & Friston, 2019; Wilkinson et al., 2017). More explicitly than other types of cognitions, expectations reflect predictions of future events or experiences and may therefore contribute to substantial suffering if people expect a certain unpleasant state to last for a long time. For instance, recent research has shown that negative expectations about the frequency and severity of intrusive memories (i.e. involuntary distressing thoughts or mental images in relation to traumatic events) influence the actual occurrence of intrusions (Herzog et al., 2022; Kube et al., 2022).

In the present study, we considered PTSD-specific situational expectations and hypothesised that they mediate the relationship between dysfunctional general cognitions and PTSD symptom severity. This hypothesis draws on recent theoretical models from general psychology (Panitz et al., 2021; Rief et al., 2015) proposing that in specific situations, generalised expectations (e.g. ‘I will never be safe again’) are activated and initiate the occurrence of situational expectations (e.g. the anticipation of threat), which are assumed to be accompanied by a corresponding affective reaction (e.g. feeling scared). In line with this assumption, research on depression has shown that depression-specific situational expectations mediate the relationship between depression-specific generalised cognitions and depressive symptoms (Kube et al., 2018; Kube et al., 2018). For PTSD, such a cognitive mediational model has not been tested yet.

Relating this assumed link between general cognitions, situational expectations, and PTSD symptoms to the question of how the association between multiple traumatic events and the severity of PTSD symptoms comes about, we tested the following key hypothesis in a serial mediational model: Multiple traumatic events in the past are associated with more negative general cognitions in relation to the trauma(s); dysfunctional general cognitions are associated with dysfunctional situation-specific expectations, which ultimately relate to the PTSD symptom severity. Testing this mediation hypothesis, we predicted that the direct effect of the number of traumatic experiences on PTSD symptom severity is relatively small (.10 ≤ | β | ≥ .20), with reference to two meta-analyses (Brewin et al., 2000; Ozer et al., 2003). Instead, we hypothesised a significant indirect effect of multiple traumas via dysfunctional general cognitions and situational expectations.

2. Methods

This study was part of a larger research project that was approved by the local ethics committee of the Department of Psychology at the Philipps-University of Marburg (reference number 2018-6k) and was conducted in accordance with the ethical standards as laid down in the Declaration of Helsinki and its later amendments. Recently, first data from this project have been published, in which a novel scale to assess situation-specific expectations in PTSD, the Posttraumatic Expectations Scale (PTES), has been validated (Herzog et al., 2022). Here, we used other data from the same sample to examine the proposed relationships between the cumulative experience of traumatic events, dysfunctional cognitions, situational expectations, and PTSD symptom severity. There are no further studies planned using the same data set.

2.1. Measures

2.1.1. Potentially traumatic events

The Life Event Checklist for DSM-5 (LEC-5; (Weathers et al., 2013) is a screening instrument for potentially traumatic events over the entire life span. The LEC-5 asks for 16 different potentially traumatic events that qualify for the A criterion (i.e. trauma criterion) according to DSM-5. An additional item asks if the person has experienced any other extremely stressful life event, which is not covered by the other 16 items. The German version of this scale (Krüger-Gottschalk et al., 2017), that we used in the present study, uses a five-point scale to assess how each potentially traumatic event was experienced: 1) ‘happened to me’, 2) ‘witnessed it’ 3) ‘learned about it’, 4) ‘part of my job’, 5) ‘not sure’.¹ In order to obtain a quantitative picture of the cumulative number of the traumatic events, a sum score is computed from the responses in the LEC-5. In so doing, the response options happened to me and witnessed it are considered particularly distressing and are therefore coded with one, all other answer options with zero (Ben-Ezra et al., 2018; Karatzias et al., 2016). Based on the binary response system, a total sum score is then calculated with a range from 0 to 17 as successfully used in several previous studies examining people with complex PTSD (e.g. Ben-Ezra et al., 2018; Hyland et al., 2017; Karatzias et al., 2016; Shih et al., 2010).

2.1.2. Dysfunctional general cognitions related to the trauma(s)

To assess dysfunctional general cognitions in relation to the traumatic event(s), we used the Posttraumatic Cognition Inventory (PTCI; Ehlers, 1999). The PTCI is a self-report scale comprising three factors: negative cognitions about the self, negative cognitions about the world and self-blame. As such, an important aspect of the PTCI is how the trauma and its corollaries are appraised. Along with the 33 original items of the PTCI, we used the 11-item Interpretation of PTSD Symptoms Inventory (IPSI; Dunmore et al., 1999), focusing specifically on the interpretation of the PTSD symptoms. The IPSI is often used in combination with the PTCI, which is why we combine them here as well, resulting in a 44-item total scale, which is rated on a seven-point Likert scale from ‘totally disagree’ (1) to ‘totally agree’ (7). As shown in numerous studies, the PTCI and the IPSI have good psychometric properties (Foa et al., 1999; Halligan et al., 2003; Müller et al., 2010). Here, the internal consistency of the entire scale was α = .96. When completing the scale, participants received the following instruction: ‘In this questionnaire you will find a number of thoughts that people may have after traumatic experiences such as acts of violence or serious accidents. We are interested in what thoughts you have had in the last month in relation to your traumatic experience.’

Importantly, both the PTCI and the IPSI cover quite a wide range of different cognitions: from rather general cognitions about the self (e.g. ‘I’m inadequate’) and others (e.g. ‘People are not what they seem’) to cognitions that are also future-directed and thus meet our definition of expectations (e.g. ‘Nothing good can happen to me anymore’). Among these future-directed cognitions, there are also a few items that have quite a high level of situational specificity (e.g. ‘If I cannot control my thoughts and feelings about the event I will go crazy’). Thus, there is some overlap with the measure of situation-specific expectations as presented below.

2.1.3. Situation-specific dysfunctional expectations

The Posttraumatic Expectation Scale (PTES) has recently been developed to assess PTSD-specific expectations with a relatively high level of situational specificity (Herzog et al., 2022). The 81-item self-report questionnaire comprises nine subscales covering different aspects of intra-personal, inter-personal, and treatment-related expectations: general treatment expectations (A; e.g. ‘In psychotherapy I will feel supported’), expectations regarding exposure therapy (B; e.g. ‘If I face the traumatic event in psychotherapy, I will lose control’), expectations about talking about the traumatic event (C; e.g. ‘If I tell my psychotherapist about the traumatic event, she/he will blame me for it’), expectations about re-experiencing symptoms (D; e.g. ‘When I am reminded of the traumatic event, I will feel helpless’), expectations about emotion regulation (E; e.g. ‘When I experience unpleasant feelings (e.g. blame, anger, sadness, disgust …) in relation to the traumatic experience outside of psychotherapy, there will be nothing I can do to feel better’), expectations about confiding in other people (F; e.g. ‘When I confide in other people (e.g. significant others), they will misuse my trust’), expectations about encounters with strangers (G; e.g. ‘When I meet strangers, I will be able to make new acquaintances’, reversely scored), expectations about avoidance behaviour (H; e.g. ‘If I stop avoiding certain things that remind me of the traumatic experience(s) (e.g. places, activities, people), I will reexperience a traumatic event’), and general PTSD-specific expectations (I; e.g. ‘I will be able to live a happy life despite the traumatic experience’). The items are rated on a five-point Likert scale ranging from 1 (‘I don’t agree at all’) to 5 (‘I totally agree’). In the preliminary validation study, the PTES has shown excellent internal consistency with a Cronbach’s α = .96. Before completing the PTES, participants received the following instruction: ‘The following questionnaire is about your personal expectations and fears regarding possible events and experiences in the next four weeks.’

Despite some overlap with the PTCI/IPSI, the PTES is a distinct measure as it focuses exclusively on future-directed expectations and besides a few rather general expectations, most of the items reflect expectations with a high degree of situational specificity. In other words, while the PTCI/IPSI assesses a wide range of cognitions, one of which is expectations, the PTES assesses only expectations. In a previous analysis, the PTES has indeed shown incremental validity over and above the PTCI/IPSI (Herzog et al., 2022).

2.1.4. PTSD symptom severity

To assess the severity of PTSD symptoms, we used the Impact of Event Scale- Revised (IES-R; Maercker & Schützwohl, 1998). This 22-item self-report scale covers three areas of PTSD symptoms: intrusions, avoidance, and hyperarousal. Patients are asked to rate the degree of distress of each item during the past week on a four-point-Likert scale ranging from 0 (‘not at all’) to (‘many times’). The IES-R has shown good psychometric properties (Creamer et al., 2003; Maercker & Schützwohl, 1998). In the present study, the internal consistency of the IES-R was α = .84.

2.2. Participants

The data collection took place from May 2018 to July 2019. Prior to study inclusion, all patients gave informed consent. To be included, participants had to be diagnosed with PTSD according to ICD-10, as assured by structured clinical interviews (often the Structured Clinical Interview for DSM (SCID), but it was not pre-specified what interview type had to be used). The structured clinical interviews were performed by the patients’ treating psychotherapists. Other inclusion criteria were: at least 18 years old and fluency in German. Participants were recruited in several outpatient and inpatient clinics in the surrounding of the study site. In both outpatient and inpatient settings, patients were enrolled after the diagnostic procedure was completed, but before treatment began. This was important because some of the expectations from the PTES refer to some anticipated situations in psychological treatment.

In total, 70 people diagnosed with PTSD took part in this paper-pencil study (M_Age= 42.06 years; SD = 11.82; range 19–64; 82% female; 17% male; and 1% not willing to specify their gender). Further details on the sociodemographic characteristics are presented in the supplement. The average age at the onset of the disease was 23.70 years (SD = 15.45, range 0–52) and PTSD symptoms have been present on average for 17.96 years (SD = 14.64; range 0–51). On average, participants reported 5.31 different traumatic events (SD = 2.88; range 1–14). The most commonly reported traumatic events were sexual assaults (54%) and other unwanted sexual experiences (56%). With a total score of M = 79.24 (SD = 16.58), the results of the IES-R indicate a very high symptom burden. Furthermore, the sum score of the PTCI (M = 155.58, SD = 35.05) indicates the prevalence of highly dysfunctional cognitions in relation to the trauma, as it is clearly above the comparative value of 133 in a PTSD norm sample (Ehlers, 1999). Thus, overall, the sample showed a high degree of suffering.

2.3. Statistical analyses

2.3.1. Pre-processing

In the entire sample, 0.9% of the data was missing, and the MCAR test indicated that data was missing completely at random. To estimate missing values, the Full Information Maximum Likelihood (FIML) method was used. It produces almost identical estimators as multiple imputation (Collins et al., 2001) and is recommended for the analysis of causal models with Amos (Weiber & Mühlhaus, 2014).

2.3.2. Testing the measurement model and estimating the effects

To examine the assumed relationships between the variables of interest, a linear structural equation model (SEM) was tested (Bollen, 1989; Bollen & Long, 1993). We opted for this statistical approach because the influencing structures can be mapped at the latent level. That is, measurement-related errors can be factored out, allowing a more sophisticated analysis of the ‘true’ relationships of the variables of interest. Further, we wanted to test the fit of an entire, theoretically postulated model, which can be more elegantly done with an SEM than in a simple mediation analysis.

The present measurement model consists of 3 latent and 17 manifest variables, as shown in Figure 1. All prerequisites of a SEM were met (Tabachnick & Fidell, 2013). According to methodological recommendations (Kline, 2015), the present sample size is sufficient to perform a SEM, since it has been proposed that at least 10 persons are required for each latent variable. Other research has suggested that the ratio between the sample size divided by the latent variables should be at least 20 (Jackson, 2003), which would result in a minimum sample size of 60 participants in our case. Either way, the current sample size of 70 people is sufficient to perform a SEM according to the literature.

Figure 1. Assumed linear structural equation model with the three latent variables dysfunctional cognitions, dysfunctional expectations and PTSD symptom severity and their manifestations.

After estimating the effects of the model, standardised regression weights can be interpreted as effect sizes similar to correlation coefficients. That is, |β| > .1 corresponds to a small effect, |β| > .3 to a medium effect, and |β| > .5 to a large effect (Cohen, 1988). The indirect effect of the number of different traumatic experiences on PTSD symptom severity was tested in 5,000 bootstrapping samples (one-tailed, given the direction of the hypothesis)

2.3.3. Evaluation of model fit

The quality of the model results from the congruence of the predicted and empirically obtained variances and covariances (Bollen, 1989; Bollen & Long, 1993; Kline, 2015). There is a wide range of different quality criteria, i.e. model fit indices. Following the recommendations for small samples (n ≤ 250) (Hu & Bentler, 1999), we report the ratio between χ² and the degrees of freedom (values ≤ 2 indicate good model fit), the Comparative Fit Index (CFI) (values ≥ 0.96 indicate good model fit), the Standardised Root Mean Square Residual (SRMR; values ≥ 0.09 indicate good model fit).

3. Results

3.1. Correlational results

Table 1 presents the correlations among the PTES, its subscales, the PTCI, PTSD symptoms and depressive symptoms.

Table 1. Results of the intercorrelations of all scales and the PTES subscales.

	PTES_SUM	PTES_A	PTES_B	PTES_C	PTES_D	PTES_E	PTES_F	PTES_G	PTES_H	PTES_I	PTCI_SUM
IES-R_SUM	.54*	.05	.35	.31	.40	.52*	.42*	.36	.28	.52*	.58*
IES-R_Intrus	.43*	.00	.26	.22	.31	.42*	.26	.32	.30	.50*	.51*
IES-R_Avoid	.38	.07	.31	.17	.31	.41*	.32	.15	.17	.29	.37
IES-R_Hyper	.43*	.03	.22	.35	.28	.36	.37	.37	.17	.42*	.48*
PTCI_SUM	.66*	.10	.55*	.44*	.56*	.60*	.55*	.52*	.22	.47*	-
PHQ-9_SUM	.63*	.18	.49*	.47*	.39*	.52*	.49*	.47*	.34	.55*	.57*

Note. IES-R_SUM = Total sum score of the Impact-of-event-Scale-revised; IES-R_Intrus = Intrusion subscale; IES-R_Avoid = Avoidance subscale; IES-R_Hyper = Hyperarousal subscale; PTCI_SUM = Total sum score of the Posttraumatic Cognitions Inventory and the Interpretation of PTSD Symptoms Inventory; PTES_SUM = Total sum score of the Posttraumatic Expectations Scale (PTES); PTES_A = Subscale A: General treatment expectations; PTES_B = Subscale B: Expectations related to the exposure to the traumatic event during psychotherapy; PTES_C = Subscale C: Expectations related to talking about the traumatic event in psychotherapy; PTES_D = Subscale D: Expectations regarding reexperiencing symptoms; PTES_E = Subscale E: Expectations regarding emotion regulation; PTES_F = Subscale F: Expectations regarding openness to other people; PTES_G = Subscale G: Expectations regarding encounters with strangers; PTES_H = Subscale H: Expectations regarding avoidance behaviour; PTES_I = Subscale I: General disorder-specific expectations; PHQ-9_SUM = Sum score of the Patient Health Questionnaire – 9

* p < .05. All p-values were corrected according to the Holm method for multiple comparisons.

3.2. Estimation of parameters and effects

In a first step, the parameters were checked for their significance for the model. Table 2 shows the relevant statistics of the manifest variables pertaining to each latent variable.

Table 2. Estimated effects from the structural equation model.

	β	B	p	S.E. (B)	C.R.
Dysfunctional general cognitions
Neg. self^a	.99	1.00	-	-	-
Neg. world^a	.79	0.82	< .001	.08	10.24
Self-blame	.55	0.76	< .001	.14	5.34
IPSI^b	.86	0.98	< .001	.07	13.13
Dysfunctional expectations
PTES A^c	.38	0.44	.004	.15	2.86
PTES B^c	.72	0.76	< .001	.16	4.78
PTES C^c	.62	0.54	< .001	.12	4.31
PTES D^c	.75	1.04	< .001	.21	4.88
PTES E^c	.81	1.03	< .001	.20	5.26
PTES F^c	.64	1.03	< .001	.24	4.38
PTES G^c	.65	1.03	< .001	.23	4.47
PTES H^c	.61	1.00	.001	.23	4.54
PTES I^c	.77	1.09	< .001	.21	5.18
PTSD symptoms
Hyperarousal^d	.77	1.00	-	-	-
Avoidance^d	.42	0.63	.002	.21	2.97
Intrusions^d	.79	1.02	< .001	.20	5.15

Note. β = standardised regression weight. B = unstandardised regression weight. S.E.(B) = standard error. C.R. = critical ratio. ^aSubscale of the Post- Trauma Cognition Inventory (Ehlers, 1999): Negative Cognitions About the Self (Neg. Self), Negative Cognitions About the World (Neg. World) and Self-Blame.

^b Interpretation of PTSD Symptoms Inventory (IPSI; Dunmore et al., 1999). ^cSubscale of the Post Traumatic Expectation Scale: Scale A general treatment expectations, Scale B expectations regarding exposure to the traumatic event in psychotherapy, Scale C expectations regarding talking about the traumatic event in psychotherapy, Scale D expectations regarding reexperiencing symptoms, Scale E expectations regarding emotion regulation, Scale F expectations regarding openness to other people, Scale G expectations regarding encounters with strangers, Scale H expectations regarding avoidance behaviour, and Scale I general PTSD-specific expectations. ^dSubscales of the Impact of Event Scale- Revised (Maercker & Schützwohl, 1998).

Subsequently, we tested a total of four direct effects and one indirect effect from the SEM (see Figure 2). The direct effect of the number of different traumatic events on symptom severity was non-significant, β = -.15, p = 453. The direct effect of the number of different traumatic events on dysfunctional general cognitions was β = .19 (p = .069). The direct effect of dysfunctional general cognitions on situation-specific expectations was β = .79 (p < .001) and the direct effect of dysfunctional expectations on symptom severity was β = .68 (p < .001). The indirect effect of the number of different traumatic events on PTSD symptom severity via dysfunctional general cognitions and situation-specific expectations was significant with β = .10 (90% CI [.006, .468]). Thus, the relationship between multiple traumatic events on PTSD symptom severity was mediated through dysfunctional general cognitions and situation-specific expectations. An additional ‘small’ mediation analysis examining the direct effect of dysfunctional general cognitions on PTSD symptoms – and its mediation via dysfunctional situational expectations – while not considering the number of traumatic events is presented in the supplement.

Figure 2. Results of the structural equation model. The model illustrates the four direct effects. The direct effect of multiple traumatic events on PTSD symptom severity was not significant. Instead, the indirect effect via dysfunctional general cognitions and situational expectations was significant, speaking to a cognitive mediation model. ⁺ p < .10, *** p < .001.

3.3. Model fit indices

The ratio of between χ² ( =  192.85) and the degrees of freedom ( =  116) was 1.67, which was well below the threshold of 2 and thus indicates a good model fit. However, the CFI was .87, which is below the acceptable value of .96, and the SRMR was .08, which is below the acceptable value of .09; thus, both the CFI and the SRMR do not indicate a sufficiently good model fit. Considering all three fit indices together, the model fit can be considered to be on the borderline between acceptable and unacceptable.

4. Discussion

The aim of the present study was to examine a possible cognitive link between the experience of multiple traumatic events and the severity of PTSD symptoms. We found that the direct effect of multiple traumatic events on PTSD symptom severity, when controlling for the effects of the cognitive factors, was small and failed to reach significance in the given sample. Instead, in support of our hypothesised model, we found a significant indirect effect via different types of cognitions relating to the traumatic experience. Specifically, we found that multiple traumatic experiences were related to a higher presence of dysfunctional general cognitions, which were associated with a higher presence of dysfunctional situation-specific expectations, which were associated with higher PTSD symptom severity. Thus, the present study adds to previous research into the psychological consequences of multiple traumatic events (Brewin et al., 2000; Gonzalez et al., 2016; Park et al., 2015), and is – to our knowledge – the first to provide evidence for a cognitive link between multiple traumas and the severity of PTSD symptoms.

The present study also supports other research highlighting the importance of dysfunctional cognitions in PTSD (Blackwell et al., 2021; Ehring et al., 2008; LoSavio et al., 2017; Ozer et al., 2019; Woud et al., 2021) and further specifies the well-established cognitive model of PTSD (Ehlers & Clark, 2000). Specifically, our findings indicate that the relationship between dysfunctional general cognitions and PTSD symptoms is (partially) mediated through dysfunctional situational expectations and both types of cognitions are the link through which multiple traumatic events are associated with PTSD symptoms. The importance of situational expectations as highlighted in the present research is in line with recent predictive processing approaches to PTSD (Kube et al., 2020; Linson & Friston, 2019; Wilkinson et al., 2017) and with evidence from experimental work, showing that negative expectations about the occurrence of intrusive memories increase their likelihood of occurrence (Herzog et al., 2022; Kube et al., 2022).

Due to the cross-sectional nature of the current data, causal conclusions about the aetiology of PTSD are undue. Yet, it is reasonable to assume that the traumatic event(s) preceded the occurrence of trauma-related negative cognitions, since it is well known that the experience of traumatic events dramatically changes people’s belief system (Bernardi et al., 2019; Brown et al., 2019; LoSavio et al., 2017). Provided that this assumption of the temporal order is correct, the findings from the current clinical sample could imply that the degree to which people hold dysfunctional cognitions increases with the number of traumatic events. Such higher levels of dysfunctional cognitions are then associated with more severe PTSD symptoms. Further speculating on the potential mechanisms of the link between multiple traumatic events, the degree of dysfunctional cognitions, and PTSD symptoms, we suggest that people with a history of multiple traumatic events have – based on their actual experiences – more reason to assume that they are not safe, that other people cannot be trusted, that the world is dangerous, and so forth. Taking into account that such dysfunctional cognitions likely promote avoidance and safety behaviour and contribute to a persistent sense of threat, resulting in hyperarousal and other aversive emotional states, the association with PTSD symptoms as demonstrated in the present research is plausible.

Of note, we also could have considered (the two types of) dysfunctional cognitions as moderator variables instead of a mediation model, similar to research that examined inflexible updating of outcome expectations as a moderator between trauma exposure and PTSD symptoms (Haim-Nachum et al., 2022; Haim-Nachum & Levy-Gigi, 2021). However, in terms of the proposed aetiology model, we found it more plausible to assume that multiple traumatic events lead to a higher degree of dysfunctional cognitions, which then in turn relate to more severe PTSD symptoms, since previous research has shown that the experience of a traumatic events dramatically changes people’s belief system (Bernardi et al., 2019; Brown et al., 2019; LoSavio et al., 2017). This suggested – though not sophisticatedly empirically tested (due to the lack of longitudinal data) – temporal order of the variables of interest was the reason why we opted for a mediation model, rather than a moderation model.

4.1. Limitations

The study has several limitations. First, it should be noted that the study does not allow any conclusions about causality due to its observational nature. Although the temporal precedence criterion was met in the sense that the traumatic events took place before the assessment of patients’ post-traumatic cognitions, this kind of data is still correlational rather than experimental, hence limiting conclusions regarding the underlying aetiology of PTSD. Furthermore, it should be noted that the present study focused only on some selective cognitive aspects assumed to contribute to the psychopathology of PTSD, while not taking into account other reasonable vulnerability factors, such as emotional processing or memory processing. In terms of the structural equation model being tested, we note as an additional limitation that the model fit indices were indicative of an acceptable to unacceptable model fit. In part, this may be related to the relatively low sample size since most approaches to determine the model fit depend on the sample size (Kenny, 2020). With regard to the measure of the number of traumatic events, we note as a limitation that the LEC-5 assesses only the number of different traumatic events. However, it does not reflect whether a person may have experienced the same event multiple times, which likely also contributes to the burden of PTSD symptoms. Relatedly, in the binary scoring approach for the LEC-5 we adapted from previous research (Ben-Ezra et al., 2018; Karatzias et al., 2016), it is not scored as particularly distressing if participants indicated that they learned about a trauma or encountered traumatic events as part of their job. This binary scoring approach may often not be accurate. In addition, it might be criticised that the IES-R, which we used to assess PTSD symptoms, assesses only three symptom clusters (intrusive re-experiencing, avoidance, and hyperarousal), while not considering other symptoms that might be particularly relevant to the new disorder category ‘complex PTSD’ from the ICD-11, i.e. emotional regulation deficits, negative self-concept, and difficulties in interpersonal relationships (Lotzin & Schäfer, 2019). In terms of diagnostics, it might also be criticised that we used the ICD-10 as diagnostic classificatory system (as we used data from routine care and the ICD-10 is the official classificatory system which is used in the German health care system) – and not other systems, such as the Clinician Administered PTSD Scale (CAPS). Furthermore, it would have been interesting to have data on the actual treatment course, in order to be able to examine how pre-treatment expectations influence post-treatment symptoms. This should be examined in future research. In addition, with regard to the PTES, it might be questioned whether all nine subscales are relevant to the present research question, particularly the subscales A-C that refer to expectations for the treatment of the trauma. Our reasoning was that the experience of (multiple) traumatic events dramatically changes people’s belief system and, according to the predictive processing perspective (Kube et al., 2020; Linson & Friston, 2019; Wilkinson et al., 2017), sets up a variety of expectations relating to the anticipation of negative experiences, including experiences in psychotherapy. Yet, acknowledging the aforementioned criticism, we repeated the test of the structural equation model without including the PTES subscales A-C. The results of this analysis are presented in the supplement. Moreover, the diversity of the sample is relatively low, thus limiting the generalizability of the current findings. Relatedly, most of the participants reported to have experienced interpersonal traumas, so a further limitation is that it is unclear how well the findings from this study generalise to other trauma types. Finally, we want to highlight that cumulative trauma can lead to a number of chronic/complex outcomes, not just PTSD, but the present research focused on PTSD only.

5. Conclusions

The present study investigated a possible cognitive link between the experience of multiple traumatic events and PTSD symptom severity. Performing a structural equation model, we found that higher numbers of different traumatic events were associated with more dysfunctional global cognitions. Such dysfunctional global cognitions were associated with dysfunctional situation-specific expectations, which ultimately were related to the severity of PTSD symptoms. Thus, the present study further specifies the cognitive model of PTSD and points to the importance of cognitive treatment approaches for people who experienced multiple traumatic events.

Author contributions (CRediT)

Tobias Kube: Conceptualisation, methodology, project administration, supervision, visualisation, writing (original draft), writing (review & editing). Anna Caroline Elssner: Data curation, formal analysis, investigation, writing (original draft), writing (review & editing). Philipp Herzog: Conceptualisation, investigation, methodology, project administration, supervision, writing (review & editing).

Ethical standards statement

The present study was approved by the local ethics committee of the Department of Psychology at the Philipps-University of Marburg (reference number 2018-6k) and was conducted in accordance with the ethical standards as laid down in the Declaration of Helsinki and its later amendments. All patients gave written informed consent.

Acknowledgements

We would like to thank all bachelor and master graduate students as well as practitioners (especially at the Schön Clinic, Parkland-Clinic, and the Outpatient Clinic Marburg) for their support in recruiting patients for this study.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Data can be made available upon request. Data cannot be made freely available because this was not part of the informed consent that was approved by the institutional review board back in 2018.

Notes

1 The original English version has an additional sixth response option (‘does not apply’), which is not used in the German translation.