ICD-11 complex post-traumatic stress disorder and psychiatric comorbidity among UK Armed Forces veterans in Northern Ireland: a latent class analysis

ABSTRACT

Background: There is evidence to suggest that the experience of complex post-traumatic stress disorder (C-PTSD) may be commonly associated with elevated risk for several mental ill-health comorbidities.

Objective: The current study seeks to contribute to the growing literature on C-PTSD comorbidity by examining the relationship between C-PTSD and other mental health disorders in a UK Armed Forces veteran sample.

Method: This study used data from the Northern Ireland Veterans’ Health and Wellbeing Study (NIVHWS). The effective sample consisted of 638 veterans (90.0% male). Tetrachoric correlations examined the relationship between C-PTSD caseness and other mental health outcomes. Latent class analysis was then conducted, determining the optimal number and nature of classes in the sample in relation to C-PTSD, depression, anxiety, and suicidality.

Results: C-PTSD caseness (i.e. probable diagnosis) was found to be significantly associated with positive caseness of depression, anxiety, and suicidality. Overall, four latent classes emerged, with each of these classes characterized by varying degrees of comorbidity: a ‘Resilient/Low Comorbidity’ class, a ‘Lifetime Suicidal’ class, a ‘PTSD Polymorbid’ class, and a ‘C-PTSD Polymorbid’ class.

Conclusions These findings support and extend previous results indicating the highly comorbid nature of C-PTSD. C-PTSD may be considered a highly polymorbid condition, increasing the risk for multiple mental health pathologies concurrently.

HIGHLIGHTS

• The results showed that probable complex PTSD was associated with depression, anxiety, and suicidality in this military veteran sample.

• Latent class analysis revealed that probable complex PTSD was associated with multiple conditions concurrently, suggesting that complex PTSD is not only highly comorbid but polymorbid.

• The findings highlight the importance of screening for multiple pathologies, particularly in cases of probable complex PTSD.

Antecedentes: Existen evidencias que sugieren que la experiencia del TEPT Complejo (TEPT-C) puede estar comúnmente asociada con un riesgo elevado de varias comorbilidades de salud mental.

Objetivo: El presente estudio pretende contribuir a la creciente literatura sobre la comorbilidad del TEPT-C examinando la relación entre el TEPT-C y otros trastornos de salud mental en una muestra de veteranos de las Fuerzas Armadas del Reino Unido.

Método: Este estudio utilizó datos del Estudio de Salud y Bienestarde los Veteranos de NI (NIVHWS por sus siglas en ingles). La muestra efectiva consistió en 638 veteranos (90,0% hombres). Las correlaciones tetracóricas examinaron la relación entre la casuística del TEPT-C y otros resultados de salud mental. A continuación, se realizó un análisis de clases latentes para determinar el número y la naturaleza óptimos de las clases dentro de la muestra en relación con TEPT-C, depresión, ansiedad y suicidalidad.

Resultados: La casuística del TEPT-C (es decir, el diagnóstico probable) se asoció significativamente con la casuística positiva de depresión, ansiedad y suicidalidad. En general, surgieron cuatro clases latentes, cada una de las cuales se caracterizaba por diversos grados de comorbilidad: una clase "Resiliente/Baja comorbilidad", una clase "Suicida de por vida", una clase "Polimorbilidad de TEPT" y una clase "Polimorbilidad de TEPT-C".

Conclusiones: Estos hallazgos apoyan y amplían los anteriores que indican la naturaleza altamente comórbida del TEPT-C. El TEPT-C puede considerarse una condición con alta polimorbilidad, que aumenta el riesgo de múltiples patologías de salud mental concurrentemente.

目的：有证据表明，复杂性 PTSD (C-PTSD) 的经历可能通常与一些精神疾病共病风险增加有关。本研究旨在通过考查英国武装部队退伍军人样本中 C-PTSD 与其他心理健康障碍之间的关系，为不断积累的 C-PTSD 共病文献做出贡献。

方法：本研究利用了来自 NI 退伍军人健康与福祉研究 (NIVHWS) 的数据。 有效样本包括 638 名退伍军人（90.0% 为男性）。四分相关性考查了 C-PTSD 案例与其他心理健康结果之间的关系。然后进行潜在类别分析，确定样本中与 C-PTSD、抑郁、焦虑和自杀相关的最佳类别数量和性质。

结果：发现 C-PTSD 个体（即可能的诊断）与抑郁、焦虑和自杀的阳性个体显著相关。总体而言，出现了四个潜在类别，每个类别都以不同程度的共病为特征：“韧性/低共病”类别、“终身自杀”类别、“PTSD 多病”类别和“C-PTSD 多病”类别。

结论：这些发现支持并扩展了前人那些表明 C-PTSD 高度共病性质的发现。 C-PTSD 可能被认为是一种同时增加了多种心理健康疾病风险的高度多病态疾病。

KEYWORDS:

• ICD-11
• complex PTSD
• military veterans
• comorbidity
• depression
• anxiety
• suicide

PALABRAS CLAVE:

• CIE-11
• TEPT complejo
• veteranos militares
• comorbilidad
• depresión
• ansiedad
• suicidio

关键词:

• ICD-11
• 复杂性PTSD
• 退伍军人
• 共病
• 抑郁
• 焦虑
• 自杀

1. Background

Mental health difficulties have been found to be a serious concern in post-conflict societies, with notably high prevalence rates of depression, anxiety, and post-traumatic stress disorder (PTSD) compared to global means (Charlson et al., 2019). In Northern Ireland (NI), a region with history of armed conflict, an increased risk of suicidal ideation has likewise been found (O'Neill et al., 2014). Such mass exposure to traumatic stressors is thought to contribute to an increased risk of multiple mental health disorders, including PTSD, depression, and anxiety symptomologies (Musisi & Kinyanda, 2020). Indeed, previous evidence has shown a heightened risk of comorbid mental health disorders in the UK and Ireland following conflict-related traumatic stress (Hyland et al., 2021; Murphy et al., 2021; Spikol et al., 2022). These factors are of particular importance to armed forces and veteran populations as these individuals are likely to experience an array of traumatic stressors, placing them at risk for PTSD and other psychological difficulties (Armour, Robinson, et al., 2021; Murphy et al., 2021).

Historically, PTSD has been regarded as highly comorbid with anxiety and mood disorders (Galatzer-Levy et al., 2013). Epidemiological evidence suggests that around half of those who experience PTSD are likely to report comorbid depression (Pietrzak et al., 2012) and 40% are more likely to meet the criteria for comorbid generalized anxiety disorder (GAD) (Milanak et al., 2013). Research has also shown that comorbid presentation with anxiety and depression is more common than PTSD diagnosis alone (Ginzburg et al., 2010). Likewise, the risk for suicidal ideation has been found to be elevated among veterans with PTSD, and further heightened for those experiencing comorbid psychological disorders (Calabrese et al., 2011). It is theorized that the constituent symptoms of PTSD drive such comorbid associations owing to symptom overlap, i.e. difficulties captured by multiple diagnostic criteria, or symptom interaction, i.e. symptoms causing others in a cascading fashion (Duek et al., 2021).

The International Classification of Diseases, 11th revision (ICD-11) revision of ‘Disorders specifically associated with stress’ was made in part with the goal of reducing comorbid statistics, and to accurately quantify specific post-traumatic syndromes (Brewin, 2013; Maercker et al., 2013). In this revision, PTSD was codified to include the experience of three symptoms over the previous month – re-experiencing, avoidance, and sense of threat – along with self-reported functional impairment in at least one domain of life (Cloitre et al., 2018). This revision also included the introduction of a novel diagnostic category, complex post-traumatic stress disorder (C-PTSD), a disorder extending traditional PTSD to include additional symptoms of negative self-concept, affective dysregulation, and interpersonal difficulties (Cloitre, 2020). These disorders are distinguished by these additional symptoms, and common experience of ‘complex trauma’, traumatic stress that is chronic or prolonged in nature leading to additional difficulties through the loss of psychosocial resources (Cloitre et al., 2013).

Consideration of the potentially similar comorbid nature of C-PTSD is warranted owing to its similar aetiology with PTSD, and commonality in veteran populations (Murphy et al., 2021). More specifically for veterans in NI, increased experience of the chronic and complex traumatic stress characteristic of C-PTSD antecedents, e.g. persistent perceived personal threat and reminders of previous traumatic events due to living in a post-conflict society (Armour et al., 2017), may place this group at increased risk for C-PTSD. Moreover, research has indicated that C-PTSD is highly comorbid with other mental health disorders, in the same vein as PTSD, and indeed may be considered a greater risk for comorbidity with the above-mentioned mental health difficulties than traditional PTSD (Hyland et al., 2020; Karatzias et al., 2019; Letica-Crepulja et al., 2020; Moller et al., 2020; Murphy et al., 2021).

The additional disturbance in self-organization (DSO) symptoms characteristic of C-PTSD may be indicative of increased comorbidity of psychopathology and contribute to the risk of positive screening for disorders such as depression and anxiety. These additional symptoms of affect dysregulation (e.g. emotional reactivity and suppression), negative self-concept (e.g. feelings of worthlessness and failure), and interpersonal difficulties (e.g. feeling distant or cut off from others) (Cloitre et al., 2018) overlap as criteria for other mental health difficulties, including anxious and affective disorders (Moller et al., 2020; Resick et al., 2012) and, as such, it is argued that these symptoms may contribute additive risk for mental health comorbidities for those with C-PTSD (Sar, 2011).

To date, studies have primarily considered the relationship between C-PTSD and comorbid mental health difficulties laterally, using analytical approaches involving regression and correlation models to examine C-PTSD comorbidity in relation to each disorder in turn. While this approach provides valuable information on which disorders are associated in a bivariate manner, it is limited as multiple concurrent disorders are not captured by these methods. Given that traditional PTSD is known to be highly and multiply comorbid, i.e. with multiple comorbid mental health difficulties (Ginzburg et al., 2010), there remains a need to understand whether the same patterns exist in relation to C-PTSD symptomology. One such method used to assess this is latent variable modelling, which may identify subpopulations based on individual characteristics, such as the presence of multiple symptomologies/disorders, concurrently (Galatzer-Levy et al., 2013). Such use of a person-centred model is argued to provide a more holistic understanding of the presentation of C-PTSD.

This is exemplified by the work of Fox et al. (2020), assessing the presentation of comorbid common mental health disorders associated with ICD-11 PTSD, and Saraiya et al. (2021), investigating latent comorbid symptom classes with C-PTSD and borderline personality disorder (BPD). In both instances, classes with increased psychopathology across disorder indicators were observed, with the authors suggesting that similarity in symptoms potentially contributes to the observed comorbidity (Fox et al., 2020; Saraiya et al., 2021).

The current study seeks to contribute to the growing literature on C-PTSD comorbidity (Karatzias et al., 2019) and understanding of the relationship between C-PTSD and the above-mentioned psychiatric and behaviour conditions, and to address gaps in the current understanding by examining multiple concurrent comorbidities, contributing the first latent class examination of C-PTSD and diagnostic comorbidity in a sample of UK Armed Forces veterans. This approach has previously been used in a study investigating traditional PTSD comorbidity (Contractor et al., 2015; Galatzer-Levy et al., 2013). This study draws on the previous evidence of PTSD latent class models to examine the comorbid (i.e. co-occurrence of two disorders) and polymorbid (i.e. exhibiting multiple concurrent comorbid mental health disorders) associations with C-PTSD.

It was hypothesized that C-PTSD caseness would be associated with: (1) elevated risk of positive screening for depression, anxiety, and suicidality, and (2) membership of a polymorbid latent class, similarly to the primary findings of previous studies examining traditional PTSD comorbidity (Contractor et al., 2015; Galatzer-Levy et al., 2013). This analysis provides a unique contribution to the literature in this regard, as, to the authors’ knowledge, no previous studies have used this approach to examine polymorbidity with C-PTSD.

2. Method

2.1. Participants

The current study adopted a secondary data analysis design as part of the Northern Ireland Veterans’ Health and Wellbeing Study (NIVHWS) (Armour, McGlichey, et al., 2021). The participants in the current study comprised a community sample of military veterans living in NI surveyed between 2017 and 2019. Data were collected using self-report methods, with online and pen-and-paper survey modes available. The survey was advertised in online and print media, and through local community organizations who work with veterans.

2.2. Procedure

A total of 1329 participants provided informed consent for the study. However, n = 307 were removed as they did not endorse at least one trauma and a further n = 384 were removed as they had 20% or more missing data across the key variables of interest (PTSD, C-PTSD, suicidality, depression, and anxiety). Those with less than 20% missing data across these variables were retained. The final eligible sample size was n = 638. Comparisons between the included sample and cases excluded owing to missing data are presented in Additional file 1 (see Supplementary material). Ethical approval was provided by the Ulster University Research Ethics Committee [REC/17/0031] and ratified by the Queen’s University Belfast Faculty of Engineering and Physical Sciences ethical committee [EPS19_156]. All participants provided written consent.

2.3. Measures

Demographic variables were captured using a bespoke set of items developed for the NIVHWS. These included questions on participants’ gender, age, relationship status, and educational attainment, which were included as covariates in this study.

Trauma exposure was measured using 17 items: 13 items comprising the Stressful Life Events Screening Questionnaire adapted for the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) (SLESQ) (Elhai et al., 2012) and four items from the Life Events Checklist for DSM-5 (LEC-5) (Weathers et al., 2013). This combination of items has been used in previous studies (McGlinchey et al., 2022; Spikol et al., 2022) and was deemed necessary to ensure a fully comprehensive trauma screen within this population, in line with the unique sociopolitical context of NI. Participants were asked whether any of the 17 stressful life events had ever happened to them (Yes/No).

The International Trauma Questionnaire (ITQ) was used to examine ICD-11 PTSD and C-PTSD caseness and symptomatology (Cloitre et al., 2018). The ITQ contains a total of 18 items: six items in relation to PTSD, with two items each allocated to measuring the specific ICD-11 symptom clusters (‘re-experiencing’, ‘avoidance’, and ‘sense of threat’) and six items measuring additional C-PTSD criteria, with two items measuring each of the three ICD-11 symptom clusters (‘affective dysregulation’, ‘negative self-concept’, and ‘disturbed relationships’). The scale contains a further three items measuring functional impairment in relation to PTSD and three to C-PTSD. Participants are asked to answer in relation to their experiences over the past month, and to rate their responses on a five-point Likert scale ranging from ‘not at all’ (0) to ‘extremely’ (4). Meeting caseness was determined using the standardized scoring for the ITQ (Cloitre et al., 2018); PTSD requires endorsement of a score of ≥ 2 (‘moderately’) for at least one of two symptoms from each of the three PTSD clusters, as well as a score of ≥ 2 on at least one functional impairment item. Similarly, meeting caseness for C-PTSD requires meeting the above criteria for PTSD, as well as scoring ≥ 2 for at least one symptom from each of three C-PTSD clusters and for at least one functional impairment item. When these criteria for C-PTSD are met, this supersedes or replaces probable PTSD categorization, and thus these are mutually exclusive. Previous studies have demonstrated the excellent psychometric properties of the scale (Cloitre et al., 2018; Murphy et al., 2021) and enumerated six latent factors in line with the symptom groupings described above in this veteran population (Armour, Robinson, et al., 2021). In the current study, Cronbach’s alpha was .96 for the PTSD items, .96 for the C-PTSD items, and .98 for the total scale.

Depression and anxiety were measured using the Patient Health Questionnaire 9 (PHQ-9) (Kroenke et al., 2001) and the seven-item Generalized Anxiety Questionnaire (GAD-7) (Spitzer et al., 2006), respectively. Both measures are considered valid and reliable self-report tools across general, clinical, and UK veteran populations (Kroenke et al., 2001; Murphy et al., 2016; Spitzer et al., 2006). All items across both scales are scored on a four-point Likert scale, which ranges from 0 to 3 [‘not at all’ (0), ‘several days’ (1), ‘more than half the days’ (2), and ‘nearly every day’ (3)], with participants asked to answer in relation to how they felt in the past 2 weeks. A cut-off score of 10 for each measure is considered to provide an adequate sensitivity and specificity to screen for possible caseness of GAD and major depressive disorder (MDD) (Kroenke et al., 2001; Spitzer et al., 2006). This was the criterion applied in the current study. Cronbach’s alpha was .96 for the GAD-7 and .95 for the PHQ-9.

Suicidality was assessed in this investigation using two items adopted from the World Health Organization Mental Health Survey Consortium (Bunting et al., 2011). These items specifically assessed lifetime suicidal ideation (‘Did you ever in your life have thoughts of killing yourself?’) and lifetime suicide attempt [‘Have you ever made a suicide attempt (i.e. purposefully hurt yourself with at least some intent to die)?]. Each item allowed a binary (Yes/No) response. Suicidal ideation and attempt, while related, are considered precursors of completed suicide and robust indicators of suicidality (Klonsky et al., 2016). Furthermore, it has been noted that these indicators are not inextricably linked and have been observed to have differential associations among veterans with PTSD, suggesting that ideation and attempt should be considered independently (Holliday et al., 2020; Klonsky et al., 2021). Self-reported suicidal ideation and attempt have been demonstrated to be associated with common mental health disorders, both as precursors and as antecedents (de Beurs et al., 2019). The association of lifetime incidence was therefore considered in these analyses.

2.4. Data analysis

The goal of this analysis was to examine the comorbidity of C-PTSD with probable diagnosis of other common mental health disorders. As such, all outcomes were treated as dichotomous, with caseness determined by previously validated scoring (Cloitre et al., 2018; Kroenke et al., 2001; Spitzer et al., 2006). This approach of using categorical indicators and latent class analysis (LCA) was judged to be most appropriate, in line with the self-report and screening instruments used in these data and the study aim of identifying potential polymorbidity across disorders.

First, to confirm the hypothesized positive association between probable ICD-11 C-PTSD caseness and the other indicators of mental health disorders, chi-squared tests of association and tetrachoric correlation (Cramer’s V) were estimated for available data (i.e. using casewise deletion) using Jamovi (Jamovi Project, 2021). Guided by the benchmarks laid out by Cohen (1988), Cramer’s V was interpreted as small (.0–.3), medium (.3–.5), or large (> .5).

Secondly, LCA was applied to examine the potential polymorbid relationship between the mental health disorders under investigation. The latent classes in this study were used as a data reduction technique to group and explain differentiated classes of discrete observations The aim of the LCA approach in the current study was to use a data reduction technique to identify statistically distinct groupings of psychological comorbidity in relation to C-PTSD (Hagenaars & McCutcheon, 2002); in other words, to identify groups of individuals experiencing multiple probable mental health disorders concurrently. To facilitate this caseness for PTSD and DSO, the necessary components of C-PTSD diagnosis were included in the model with probable diagnosis of GAD, MDD, and lifetime incidence of suicidal ideation and attempt. This extends the variable-centred approach of the preceding analysis by estimating whether there exist subpopulations that report different experiences with patterns of one or multiple comorbid conditions (Galatzer-Levy et al., 2013). The latent classes were established in the absence of an assessment of confounding variables, as the focus was on the similarity of responding across participants in particular classes, irrespective of their various sociodemographic profiles.

Model estimations were performed using ‘poLCA’ in RStudio, with the expectation-maximization estimator (Linzer & Lewis, 2011; RStudio Team, 2020). To avoid solutions based on local maxima, 500 random sets of starting values were used with 50 second-stage optimizations. The Akaike information criterion (AIC) (Akaike, 1987) and the Bayesian information criterion (BIC) (Schwarz, 1978) were evaluated, with the class solution that produced the lowest values judged to be the best fitting model. Entropy scores could range from 0 to 1, with higher scores indicating a better fitting model (Gabriel et al., 2015). Differences between classes on sociodemographic variables were assessed using the chi-squared test for categorical data and Kruskal–Wallis rank sum tests for continuous data in the ‘gtsummary’ package (Sjoberg et al., 2021).

3. Results

3.1. Sample characteristics

The sample was predominantly male (90.0%), white (99.4%), and married (73.0%). The majority had served in the Army (86.8%) and achieved a highest rank of non-commissioned officer (51.7%). The mean ± SD number of trauma experiences reported was 6.45 ± 3.05. Of these, the most commonly endorsed was ‘exposure to fire or explosion’ (82.0%), followed by ‘repeated exposure to vivid trauma details’ (70.1%), and ‘experience of a family member or close friend dying’ (54.4%). The most commonly endorsed most distressing (i.e. index trauma) was ‘exposure to fire or explosion’ (17.8%). Full demographic information is presented in Additional File 1 (see Supplementary material).

3.2. Psychiatric morbidity and comorbidity with C-PTSD

Table 1 displays the endorsement rates for all mental health variables measured in this study. C-PTSD caseness was found to be significantly and positively associated with all other mental health outcomes under investigation (Table 1). C-PTSD was found to exhibit a large association with probable GAD and MDD caseness [Cramer’s V > .5] and a medium association with indicators of suicidality [Cramer’s V > .3] (Cohen, 1988).

Table 1. Prevalence of probable mental health disorders and association with probable complex post-traumatic stress disorder (C-PTSD).

Condition	Prevalence, n (%)	Association with C-PTSD Cramer’s V
PTSD	37 (5.8)	–
C-PTSD	148 (23.4)	–
GAD	222 (35.1)	.641***
MDD	268 (42.3)	.578***
Suicidality
Ideation	363 (56.9)	.356***
Attempt	122 (19.1)	.346***

Note: PTSD, post-traumatic stress disorder; GAD, generalized anxiety disorder; MDD, major depressive disorder.

***p < .001.

Given this finding confirming the association between C-PTSD and other aspects of mental ill-health, this investigation proceeded to examine whether statistically identifiable groups, which differed in concurrent comorbid symptomologies, were present in this sample, using latent variable modelling.

3.3. Latent class analysis

Consistent with best practice in latent class estimation, a series of models with increasing numbers of classes was applied to these data and assessed for fit (Cloitre et al., 2013). Specifically, the fit of six models was assessed. The LCA fit statistics for all models specified are presented in Table 2.

Table 2. Fit statistics for two- to six-class solutions.

Model	Log likelihood	df	BIC	ABIC	CAIC	Likelihood ratio	Entropy
2 Class	−1742.299	50	3568.557	3527.283	3581.557	184.36227	.912
3 Class	−1699.841	43	3528.848	3465.350	3548.848	97.13214	.774
4 Class	−1670.311	36	3514.996	3429.273	3541.996	39.27681	.797
5 Class	−1660.389	29	3540.361	3432.413	3574.361	20.23827	.824
6 Class	−1655.861	22	3576.513	3446.342	3617.513	11.87128	.833

Note: df, degrees of freedom; BIC, Bayesian information criterion; ABIC, adjusted Bayesian information criterion; CAIC, consistent Akaike information criterion.

The four-class solution returned the lowest values on absolute indices, and an entropy value approaching .8, suggesting the robust specification of classes. Guided by the interpretation of the sum of fit indices, and by principles of parsimony and interpretability of the classes produced (Collins & Lanza, 2009), the four-class solution was selected as the most meaningful and best fitting model for these data (Table 3).

Table 3. Posterior probability estimates (standard error) for probable diagnoses in each latent class.

Class	PTSD	DSO	GAD	MDD	SI	SA
Resilient (51.41%)	0.006 (0.007)	0.005 (0.006)	0.023 (0.011)	0.053 (0.017)	0.278 (0.035)	0.003 (0.007)
Lifetime Suicidal (14.11%)	0.181 (0.050)	0.157 (0.047)	0.136 (0.054)	0.401 (0.069)	0.971 (0.035)	0.499 (0.089)
PTSD Polymorbid (12.85%)	0.491 (0.076)	0.539 (0.074)	0.826 (0.063)	0.891 (0.046)	0.536 (0.086)	< 0.000 (< 0.000)
C-PTSD Polymorbid (21.63%)	0.925 (0.033)	0.942 (0.029)	0.950 (0.026)	0.990 (0.012)	0.980 (0.020)	0.538 (0.058)

Note: PTSD, post-traumatic stress disorder; DSO, disturbance in self-organization; GAD, generalized anxiety disorder; MDD, major depressive disorder; SI, suicidal ideation; SA, suicide attempt; C-PTSD, complex post-traumatic stress disorder.

The profile plot and probabilities for the four-class solution are shown in Figure 1. Class 1 was the largest class. It contained 51.36% of the sample and was characterized by a low probability of caseness for all mental health outcomes in the current study. This class was labelled the ‘Resilient’ class.

Figure 1. Probability profile plot for latent classes of mental health comorbidity. PTSD, post-traumatic stress disorder; C-PTSD, complex post-traumatic stress disorder; DSO, disturbance in self-organization; GAD, generalized anxiety disorder; MDD, major depressive disorder; SI, suicidal ideation; SA, suicide attempt.

The second class identified was Class 2, ‘Lifetime Suicidal’, and was representative of 14.13% of the current sample. This class was observed to have a low likelihood of meeting caseness for mental health issues, with elevated likelihood for screening positive for MDD (0.401). It is worth noting that this may be attributed in part to consistency between screening for lifetime suicidality and ‘Thoughts that you would be better off dead’ in the previous 2 weeks (PHQ item 9). This class was characterized by a markedly high probability of endorsing suicidal ideation and a moderate probability of endorsing suicide attempt across the lifetime.

Class 3 was representative of 12.85% of the study population. This class was associated with high probabilities of screening positively and concurrently for probable GAD, MDD, and lifetime suicidal ideation. This class also exhibited a moderate probability of screening positive for PTSD. This class was labelled the ‘PTSD Polymorbid’ class.

Class 4 was representative of 21.63% of the total sample. In a similar pattern to the PTSD Polymorbid class, this class was characterized by a high probability of screening positively for probable PTSD, DSO, GAD, and MDD, and an elevated probability of endorsing both indicators of lifetime suicidality. This class was labelled the ‘C-PTSD Polymorbid’ class.

3.4. Class comparisons

The resultant classes were compared on key sociodemographic variables to assess statistically significant differences between subgroups (Table 4). A statistically significant difference was observed between classes, with symptomatic classes being comparatively younger than the Resilient class, less likely to be married or cohabiting with a partner, more likely to be separated or divorced, and more likely to have lower levels of educational attainment. In addition, those in symptomatic classes were more likely to have held lower (non-officer) ranks.

Table 4. Comparison of class membership on sociodemographic variables.

	Resilient (N = 328)	Lifetime Suicidal (N = 90)	PTSD Polymorbid (N = 82)	C-PTSD Polymorbid (N = 138)	Test statistic (p-value)^a
Gender					8.530 (.036)
Female	25 (7.65%)	10 (11.11%)	15 (18.29%)	13 (9.42%)
Male	302 (92.35%)	80 (88.89%)	67 (81.71%)	125 (90.58%)
Age (years)	57.91 ± 11.12	54.35 ± 9.62	51.58 ± 10.44	53.30 ± 8.58	35.131 (< .001)
Relationship status					22.551 (.032)
Single	17 (5.18%)	8 (8.89%)	6 (7.32%)	5 (3.62%)
Married or cohabiting	256 (78.05%)	58 (64.44%)	57 (69.51%)	95 (68.84%)
Separated or divorced	37 (11.28%)	21 (23.33%)	17 (20.73%)	34 (24.64%)
Widowed	17 (5.18%)	3 (3.33%)	2 (2.44%)	4 (2.90%)
Other	1 (0.30%)	0 (0.00%)	0 (0.00%)	0 (0.00%)
Education					67.911 (< .001)
No qualifications	40 (12.94%)	18 (21.43%)	14 (17.72%)	38 (30.16%)
GCSE grade D–G	17 (5.50%)	9 (10.71%)	7 (8.86%)	19 (15.08%)
GCSE grade A*–C	47 (15.21%)	19 (22.62%)	23 (29.11%)	23 (18.25%)
A level	28 (9.06%)	10 (11.90%)	11 (13.92%)	7 (5.56%)
CertHE	26 (8.41%)	7 (8.33%)	4 (5.06%)	10 (7.94%)
DipHE	47 (15.21%)	8 (9.52%)	10 (12.66%)	11 (8.73%)
Bachelor’s degree	53 (17.15%)	5 (5.95%)	5 (6.33%)	10 (7.94%)
Master’s degree	44 (14.24%)	8 (9.52%)	5 (6.33%)	8 (6.35%)
Doctoral degree	7 (2.27%)	0 (0.00%)	0 (0.00%)	0 (0.00%)
Ethnicity					6.402 (.400)
White	327 (99.70%)	88 (97.78%)	82 (100.00%)	137 (99.28%)
Mixed ethnicity	1 (0.30%)	1 (1.11%)	0 (0.00%)	0 (0.00%)
Other	0 (0.00%)	1 (1.11%)	0 (0.00%)	1 (0.72%)
Total number of traumas endorsed	5.55 ± 2.72	6.71 ± 3.04	6.98 ± 2.98	8.07 ± 3.10	64.905 (< .001)
Currently in reserves	36 (10.98%)	8 (8.89%)	6 (7.32%)	5 (3.62%)	6.869 (.076)
Highest rank held					40.715 (< .001)
Other rank	64 (19.57%)	33 (36.67%)	35 (42.68%)	58 (42.34%)
NCO	184 (56.27%)	47 (52.22%)	37 (45.12%)	62 (45.26%)
Officer	79 (24.16%)	10 (11.11%)	10 (12.20%)	17 (12.41%)

^aPearson's Chi-squared test (nominal variables)/Kruskal-Wallis rank sum test (scale variables).

Note: Data are shown as n (%) or mean ± SD.

PTSD, post-traumatic stress disorder; C-PTSD, complex post-traumatic stress disorder; GCSE, General Certificate of Secondary Education; CertHE, Certificate of Higher Education; DipHE, Diploma of Higher Education; NCO, non-commissioned officer.

Finally, the comparison of total trauma endorsements between classes suggested that those in symptomatic classes were likely to report a greater average number of exposures to potentially traumatic events, with those in the C-PTSD Polymorbid group endorsing the greatest mean number of traumatic exposures.

4. Discussion

This study examined the patterns of mental health comorbidity associated with ICD-11 C-PTSD in a UK Armed Forces veteran sample resident in NI, hypothesizing that C-PTSD would be associated with other mental health conditions and would be associated with polymorbid experience of these, i.e. multiple concurrent probable disorder caseness. Consistent with prior research (Karatzias et al., 2019; Murphy et al., 2021), C-PTSD caseness was positively associated with depression, anxiety, and suicidality. This study further contributes a novel examination of the comorbid nature of C-PTSD using LCA methods. The results of this analysis indicated a four-class solution best fitted the data, characterized as ‘Resilient’, ‘Lifetime Suicidal’, ‘PTSD Polymorbid’, and ‘C-PTSD Polymorbid’.

As expected, C-PTSD was associated with polymorbidity, being associated with multiple mental health disorders under investigation concurrently. Similar patterns of probability for screening positive for these issues were observed for both groups; however, the C-PTSD Polymorbid class was observed to have a higher posterior probability of experiencing these issues relative to the PTSD Polymorbid group. This finding supports the study assertion that C-PTSD would be associated with similar patterns of comorbidity to traditional PTSD, finding that those with probable C-PTSD experience a further heightened probability of reporting problems with GAD, MDD, and suicidality.

Similarly, in comparisons between classes, it was found that the classes differed significantly on key variables, with the symptomatic classes being associated with younger age, non-partnered relationship status, lower educational attainments, and greater number of exposures to potentially traumatic events. This pattern of association was, in general, more acutely observed for the C-PTSD Polymorbid class, perhaps unsurprisingly, as these factors are established risk factors for psychological morbidity (Breslau, 2002; Brewin et al., 2000). This finding thus highlights that these established risk factors for mental ill-health should also be considered in screening for polymorbidity, particularly in association with C-PTSD.

Evidence has demonstrated that people reporting C-PTSD symptomology in clinical settings are at heightened risk of experiencing psychological comorbidities (Karatzias et al., 2019). This highly comorbid nature is similarly noted in a general population sample (Hyland et al., 2020). The unique methodology and findings of this study investigating C-PTSD and applied in a sample of military veterans extend this understanding, indicating that beyond being highly comorbid, C-PTSD pathology is unlikely to present in isolation. The examination of latent patterns of comorbidity indicate that C-PTSD is more likely to present with multiple comorbidities concurrently. Previous findings, coupled with those of the current study ,suggest that heightened diagnostic comorbidity may be considered intrinsic to PTSD and C-PTSD presentations, and may be of particular concern for those with C-PTSD (Hyland et al., 2020; Maercker et al., 2013).

This continued presence of diagnostic comorbidity with post-traumatic disorders using ICD-11 criteria should, however, be viewed not as a failure of effective classification, but rather as a by-product of a binary diagnostic classification system. The implications of this are that researchers and clinicians should consider the heightened probability that military veterans reporting C-PTSD pathology are also concurrently experiencing multiple forms of mental ill-health (Hyland et al., 2020). It should also be considered that a similar pattern with elevated comorbidity might be observed for C-PTSD relative to PTSD, as a result of those individuals C-PTSD experiencing greater traumatic stress and more acute internalizing issues spanning multiple diagnostic categories (Spikol et al., 2022).

C-PTSD pathology has been found to be associated with increased comorbidity with other common mental health disorders in a treatment-seeking veteran sample (Murphy et al., 2021). Murphy et al. (2021) also found that dissociation and difficulties with anger were associated with C-PTSD caseness. This finding highlights the potential range of comorbid conditions with C-PTSD and suggests the importance of openness in considering the complexity of C-PTSD expression and epidemiology. C-PTSD may potentially require tailored approaches to treatment (Karatzias et al., 2019) and, likewise, comorbid common mental disorders are recognized as barriers to effective treatment and recovery (Murphy et al., 2021). The results of the current study concur with best practice recommendations in the clinical treatment of patients presenting with PTSD and C-PTSD, highlighting the importance of considering multiple concurrent mental health disorders, particularly for those with C-PTSD in therapeutic case formulations (National Institute for Health and Care Excellence, 2018). However, claims regarding which treatment modalities may best serve this group are outside the remit of this investigation of mental health disorder screening.

There is therefore a call for those working with veterans presenting with C-PTSD to consider the broad psychological burden of polymorbidity regarding mental health disorders, and potentially substance misuse and physical health disorders (Ho et al., 2021; Karatzias et al., 2019). This is of particular importance as veterans subpopulations have been found to be at increased risk for such conditions (Iversen et al., 2007; Morgan & Aldington, 2020), highlighting the need to consider these diverse presentations in the assessment and treatment of C-PTSD.

Another important finding is that of a ‘Lifetime Suicidal’ class, characterized by a heightened likelihood of reporting suicidal ideation and attempt across the life course. This suggests that while C-PTSD is highly comorbid with these indicators of suicidality the association is not ubiquitous, and these may be present in the absence of other mental health difficulties. It should be noted that the co-occurrence of suicidal ideation and attempt in this latent class is unsurprising as often the former precedes the latter, thus capturing some of the variance in its endorsement; however, this is not a ubiquitous relationship (Klonsky et al., 2021). Equally, it is possible that these lifetime indicators of suicidality are indicative of individuals who have experienced mental health difficulties and distress in the past but are not currently experiencing any mental health symptomology (Apter et al., 1993). While suicidality in most commonly observed in conjunction with other mental health disorders, previous research has shown that in a minority of cases of completed suicide no psychological morbidity is observed prior to the event (Apter et al., 1993). This highlights the need for clinicians and those working with veteran groups to be cognizant of diverse presentations of mental ill-health following trauma, and to consider suicidality in the presence and absence of common mental health disorders.

Likewise, the observed moderate probability of MDD caseness in the Lifetime Suicidal class is unsurprising because of the conceptual link between depression and suicidality (Bradvik, 2018) and the presence of current suicidal ideation in the PHQ-9. With measurement delineated by time reference in the current study, it is highlighted that the relationship between probable MDD and suicidality is not ubiquitous, but should be considered a risk factor for comorbidity in this sample.

The highly comorbid and polymorbid nature of C-PTSD observed in this study may likewise be attributed to the shared aetiology of these conditions. The results showed that participants in this sample most often reported multiple traumatic experiences, with, on average, 5.55 different experience types. Coupled with the unique characteristics of this population, comprising former UK Armed Forces personnel living in a post-conflict society, i.e. a context where overt warfare has ceased but tension and the threat of violence resuming still exist (Junne & Verkoren, 2004), it is argued this sample represents a group at heightened risk of exposure to stressful/traumatic life events. Therefore, groups of military personnel and veterans exposed to home service and continuing threats may be considered at risk for C-PTSD polymorbidity.

Previous evidence has shown that the wider population in post-conflict settings is at heightened risk for common mental health disorders (Charlson et al., 2019). Indeed, in NI specifically, research pertaining to the general population estimated that over half (53%) of those who experienced a conflict-related event are likely to also experience some anxiety, mood, or behavioural control problems (Bunting et al., 2011). It is perhaps unsurprising, then, that those who experience complex traumatic events may subsequently develop a range of other mental health symptoms in addition to C-PTSD pathology, thus exhibiting transdiagnostic issues (Hyland et al., 2018). While this argument is made conceptually within the current study, future research is required to explicitly assess trauma complexity, appraisals, and loss of personal resources as antecedents of C-PTSD and comorbid difficulties.

There remains a question as to whether psychiatric caseness categories can be independent of one another when several shared symptoms underpin these categorizations. It is possible that the use of self-report screening assessment used in this study and throughout the discipline contribute further to perceived or misclassified comorbidity, as respondents endorse items across several measures of constructs that are conceptually and/or experientially similar. For instance, the additional DSO symptoms characteristic of C-PTSD have been conceptually linked to other common mental health disorders (Gilbar, 2020). Furthermore, the conceptual and diagnostic overlap between MDD and GAD is well established, with four symptoms being shared between the disorder criteria (Zbozinek et al., 2012). Despite this, latent variable modelling has been widely applied to the study of mental health comorbidity (Contractor et al., 2015; Essau & de la Torre-Luque, 2019; Galatzer-Levy et al., 2013). Additional evidence may be warranted from investigations making use of diverse statistical techniques to complement these findings.

A suggested alternative method would be via the investigation of nuanced symptom-to-symptom interactions (Fried, 2015). In the context of network theory, the concept of bridging symptoms posits that comorbidity occurs when the symptoms of one disorder ‘activate’ the symptoms of a second disorder (Fried & Cramer, 2017). For instance, the association between C-PTSD and MDD may be driven by the negative self-concept symptomology, which links closely with symptoms of MDD, and, indeed, is directly shared in the case of feelings of worthlessness (Gilbar, 2020). Therefore, the results of the current study may therefore be complemented and extended by future investigations adopting a more granular network perspective of comorbidity as a means of understanding the potential for non-specific internalizing symptoms to act as drivers of psychiatric polymorbidity.

4.1. Limitations

These results are derived from self-report screening measures of mental health disorders and the use of cut-off scores to estimate the presence of a disorder; notably, this methodology has the potential to overestimate the prevalence of psychiatric conditions (Charlson et al., 2019; Levis et al., 2020). Similarly, traumatic experiences were screened in a limited format by querying whether an event had ever occurred, lacking full clinician assessment of whether the events met the eligibility and/or definitional criteria for a PTSD diagnosis. The current study did, however, use widely recognized and validated measures of mental ill-health, with well-established cut-off scores defining probable caseness for these conditions. This is indicative of current best practice of screening for mental health conditions. Future studies could seek to use clinician-administered measurements and formal diagnoses of mental health difficulties.

Further to this, the measurement of study constructs and participant experiences was cross-sectional in nature. However, measures in this study assessed past lifetime traumatic exposures and current mental health symptomology (in the past month or past 2 weeks), and therefore, to some extent, temporal ordering between these variables can be suggested. It should also be noted that the indicators of suicidality in this study were reported retrospectively over the lifetime of respondents, and, as such, the sequencing of traumatic experiences and mental ill-health may not be definitively established. Future research may consider adopting longitudinal methodologies to more effectively assess the potential pre- and post-morbid causality of psychiatric comorbidities with C-PTSD (Frewen et al., 2012).

The sample used in the analyses should also be acknowledged. This study focused on a unique population, comprising trauma-exposed military veterans in a post-conflict society. The population and sample are notably skewed in terms of demographic composition. Likewise, the sample included in this study analyses was found to report greater trauma exposure and suicidality than those excluded owing to missing data (see Additional file 1 in the Supplementary material); this may be a result of more psychologically healthy individuals not seeing the relevance of their participation in the study based on the questions being posed. These findings therefore may not be representative of wider populations and thus generalization and interpretation to other contexts should be made cautiously.

Finally, the current study focused on a limited number of comorbid conditions with C-PTSD based on pertinent associations with common mental health problems identified in the existing literature (Hyland et al., 2020; Karatzias et al., 2019; Letica-Crepulja et al., 2020; Murphy et al., 2021). There remains the potential that C-PTSD is linked with additional mental health and behavioural conditions, such as BPD, alcohol use disorders, and physical health conditions that are not assessed in the current study (Ford & Courtois, 2014; Karatzias et al., 2019; Resick et al., 2012). It is likewise acknowledged that the ‘suicide attempt’ variable in our LCA is also likely to capture a proportion of the variance attributable to the ‘suicidal ideation’ variable within our LCA. Future investigations encompassing a broad range of psychological and behaviour comorbidities using alternative and complementary assessments are warranted.

4.2. Conclusions

This study complements and extends the current literature relating to ICD-11 C-PTSD comorbidity, while being focused on a UK Armed Forces veteran sample. The results concur with previous findings, highlighting that C-PTSD is associated with common mental health conditions, and extends those revealing that C-PTSD is associated with multiple concurrent morbidities, or polymorbidity. These findings highlight the importance of considering the potential complexity of C-PTSD pathology and the need to screen for multiple concurrent mental health and behavioural difficulties among military veterans presenting with this disorder, both in research and in clinical contexts.

Ethics approval and consent to participate

Ethical approval for this study was provided by the Ulster University Research Ethics Committee [REC/17/0031] and ratified by the Queen’s University Belfast Faculty of Engineering and Physical Sciences ethical committee [EPS19_156]. All participants provided written consent for participation in the study. Consent for publication was not required.

Authors’ contributions

MR was responsible for the study design, data analysis, and manuscript writing. EMcG was responsible for the literature review, data analysis, and manuscript writing. CA was responsible for manuscript writing and review.

Acknowledgements

The authors acknowledge the support of the members of the NIVHWS team, which supported recruitment and administration throughout the timeline of the project, and would like to thank the Forces in Mind Trust for providing funding.

Disclosure statement

No potential conflict of interest was reported by the authors.

Data availability statement

The data sets generated and/or analysed during the current study are not publicly available as consent was not sought to share this information during original data collection, but are available from the corresponding author upon reasonable request.

Additional information

Funding

This research was supported by the Forces in Mind Trust funding the Northern Ireland Veterans' Health and Well-being Study. The funders had no role in the study design or publication of the results.