Persistence and Psychological Predictors of ICD-11 Complex PTSD: A Six-Month Longitudinal Study in Hong Kong

Abstract

Recent studies showed that the ICD-11 C-PTSD is associated with considerable individual and social consequences. This study focused on the persistence and psychological predictors of C-PTSD. We analyzed longitudinal survey data from a sample of Hong Kong adults (N = 230) with an interval of approximately six months. Participants completed standardized self-report measures of C-PTSD and various psychological factors (including self-criticism, self-compassion, ruminative response, event centrality, and experiential avoidance) at both timepoints. Over 80% of participants with probable C-PTSD at baseline remained to meet the criteria for PTSD/C-PTSD at follow-up, while over 50% of participants with probable PTSD at baseline remitted at follow-up. Participants with probable C-PTSD were characterized by greater use of mental health services at both timepoints. Additionally, hierarchical multiple regression showed that, among a variety of psychological factors, experiential avoidance was the only significant predictor of PTSD (β = 0.271, p = .001) and disturbances in self-organization (DSO) (β = 0.200, p = .011) symptoms, after controlling for baseline symptom severity and treatment usage. This study shows that C-PTSD persists over time and its symptoms are predicted by experiential avoidance. More efforts are needed to prevent and address C-PTSD in the community. We also offer insights into the management of trauma-related disorders.

Keywords:

• Complex post-traumatic stress disorder (complex PTSD)
• betrayal trauma
• public mental health
• psychological factors

Post-traumatic stress disorder (PTSD) has been recognized as an official mental disorder since 1980 (American Psychiatric Association, 1980). The original formulation of PTSD was designed to describe the psychological symptoms that developed after experiencing a significant traumatic event, such as combat or a natural disaster. However, it has become increasingly evident that individuals who experience complex trauma—especially interpersonal trauma—may develop symptoms that cannot be fully captured by the traditional PTSD diagnosis (Fung et al., 2022b; Herman, 1992; Van der Kolk et al., 2005). After years of research and advocacy, World Health Organization (2019) has introduced complex PTSD (C-PTSD) as a new trauma disorder in addition to PTSD in ICD-11. C-PTSD is characterized by three disturbances in self-organization (DSO) symptom clusters (i.e., affective dysregulation, negative self-concept, and disturbances in relationships), in addition to the three classical PTSD symptom clusters (i.e., reexperiencing, avoidance, and sense of current threat). Just like other trauma-related disorders, such as borderline personality disorder and dissociative disorders, C-PTSD is more related to social-interpersonal trauma and adversities (e.g., betrayal trauma, poor family support) than to non-human trauma (Fung et al., 2022a; Herman, 1992; Sar et al., 2013; Tian et al., 2022). Therefore, from this perspective, C-PTSD can be conceptualized and regarded as a social-interpersonal-related disorder.

While C-PTSD has just been recently recognized as an official disorder, it has received increasing attention in the field (Maercker, 2021). Although one study using both the PTSD Checklist for DSM-5 (PCL-5) and the International Trauma Questionnaire (ITQ) to assess probable DSM-5 PTSD and ICD-11 PTSD/C-PTSD, respectively, showed that ICD-11 PTSD/C-PTSD may be less common than DSM-5 PTSD (Hyland et al., 2017), recent studies showed that ICD-11 C-PTSD might not be rare in the general population of some countries (e.g., United States and United Kingdom), and might be even more common than “simple” PTSD. For example, the prevalence of (probable) C-PTSD was reported to be 3.8–12.9% in two general population studies; both studies found that C-PTSD was more prevalent than PTSD (Cloitre et al., 2019; Karatzias et al., 2019). In a sample of community health service users in Hong Kong, Fung et al. (2023) observed that the past-month prevalence of (probable) ICD-11 PTSD and C-PTSD was 5.6 and 18.4%, respectively. Another prevalence study among trauma-exposed young people also reported that the rates of (probable) PTSD and C-PTSD were 1.5 and 3.4%, respectively (Redican et al., 2022). Additionally, studies found that C-PTSD was closely associated with more betrayal traumatic events, higher levels of impairment, and more clinical symptoms (Fung et al., 2023; Hyland et al., 2020; Karatzias et al., 2019; Webb et al., 2022).

Therefore, C-PTSD is a mental health condition associated with considerable individual and social consequences. However, despite its public health significance, there are major research gaps in the literature. These gaps hinder our understanding of the etiological factors and development of the disorder, limiting the efficacy of prevention and treatment in the community.

One major research gap exists in our understanding of the persistence of C-PTSD over time. While C-PTSD is typically believed to be a chronic condition, we need to empirically examine the extent to which it persists over time. Currently, a few recent studies have investigated the persistence of C-PTSD in the community. In one two-year follow-up study of C-PTSD in adolescents, which only included 66 participants, Kazlauskas et al. (2023) reported that 46% of participants with baseline C-PTSD remained to have PTSD/C-PTSD after two years. In another small sample of traditional Chinese medicine service users in Hong Kong (N = 173), Po et al. (2023) reported that 51.5% of participants with probable baseline ICD-11 C-PTSD (n = 33) remained to have PTSD/C-PTSD even after approximately nine months; in addition, baseline DSO symptoms predicted more somatic symptoms (β = 0.320, p < .001) and lower levels of occupational participation (β = −0.206, p = .019) at follow-up. In addition, Hyland et al. (2020) examined (probable) PTSD and C-PTSD in the general population of Israel over one year and reported that there were no statistically significant changes in the prevalence rates of (probable) PTSD (6.7–5.3%) and C-PTSD (4.9–3.7%) across the two timepoints. Understanding the long-term persistence of C-PTSD is crucial for prevention, advocacy, and policymaking because it can call for more preventive measures and early intervention strategies to address the disorder in the community. In case there would be more evidence showing that C-PTSD is a disabling, common, and persistent condition, more resources should be mobilized to prevent and treat the disorder. Additionally, more data from diverse sociocultural contexts are necessary because cultural factors may influence the effects of trauma (Ni & Hesketh, 2021; Şar, 2021). Therefore, the first goal of the present study was to examine the persistence of probable C-PTSD in our sample of Chinese adults.

Second, in order to inform prevention and treatments, a better understanding of the psychological factors that contribute to the development and maintenance of PTSD and C-PTSD symptoms is required. This research gap could hinder our efforts to prevent trauma-related psychopathology and its associated consequences. As we review the trauma literature, we can identify at least five psychological factors that have a well-documented association with post-traumatic symptoms. These factors include self-criticism (Crapolicchio et al., 2021; Fleming & Resick, 2016), self-compassion (Barlow et al., 2017; Scoglio et al., 2018), ruminative response (Moulds et al., 2020), event centrality (Gehrt et al., 2018), and experiential avoidance (Serrano-Ibáñez et al., 2021; Thompson & Waltz, 2010). While these factors are theoretically and empirically associated with post-traumatic symptoms, it is important to further investigate their longitudinal effects on both PTSD and DSO symptoms in order to confirm their causal relationships with post-traumatic symptoms. Additionally, as most studies have examined these factors in isolation, it is currently unclear which of these factors have the strongest predictive value. Further research is necessary in this regard in order to inform the development of more targeted and effective interventions for individuals exposed to traumatic events.

Against this background, this study aimed to answer two primary research questions: (1) To what extent do ICD-11 PTSD and C-PTSD symptoms persist over a period of six months? (2) Which psychological factors would have the strongest predictive value for PTSD and DSO symptoms? In terms of persistence, we examined how many participants with probable PTSD/C-PTSD at baseline would remain to meet the required criteria after six months. In terms of the psychological predictors, we examined the relationship of the above-mentioned five psychological variables at baseline with PTSD and DSO symptoms at follow up.

In addition to these two primary research questions, we also provided additional exploratory analysis results regarding the differences between participants with and without probable C-PTSD in terms of their trauma histories and use of mental health services in the community. These data can provide insights into the unique features of C-PTSD and the public health importance of preventing this disorder. Given that C-PTSD is often associated with complex trauma (Hyland et al., 2020) and that it may also result in greater impairment (Fung et al., 2022a), we hypothesized that participants with probable C-PTSD would report more traumatic events and greater use of mental health services than those without probable C-PTSD.

Methods

Participants

To answer the above-mentioned research questions, we analyzed longitudinal survey data from a mental health project. This study was a longitudinal study because we conducted the same observations (standardized self-report measures) at two different timepoints, with a relatively long follow-up period, in accordance with the definition provided by White and Arzi (2005). The project was granted ethical approval by the institutional review board at the Chinese University of Hong Kong. The study recruited participants through social networking sites in November and December 2022, using specific inclusion criteria: participants had to be between 18 and 64 years old, provide online written informed consent, voluntarily agree to participate, be a current Hong Kong resident, able to read and write Chinese, and have access to the web-based surveys. Individuals with learning or reading disorders, dementia, or cognitive impairments were not included in the study. Eligible participants completed a web-based survey that included validated psychological measures. After approximately six months, all participants were invited to complete the web-based survey again.

Initially, 412 participants met the inclusion criteria and provided a valid response to the survey at baseline. Nevertheless, some participants (n = 18) did not submit a valid email, and therefore we could not invite them to complete the follow-up survey. After about six months (M = 167.28 days; SD = 10.50), 230 participants provided a valid response to the follow-up survey, resulting in an attrition rate of 41.6% for this longitudinal study.

The demographic and health backgrounds reported at baseline are reported in Table 1. The age range of the included participants was 18–64 years old, with a mean age of 39.73 years (SD = 12.75). There were no statistically significant differences in all demographic variables between participants who completed the follow-up survey (n = 230) and those who did not (n = 182), except for age (Table 1). Participants who submitted the follow-up survey were older than those who did not (p = .041). Additionally, participants who submitted the follow-up survey reported slightly more childhood non-betrayal traumatic events (p = .046) (Table 1).

Table 1. Baseline differences between participants who completed the follow-up survey and those who did not.

	Participants who completed the follow-up survey (N = 230)		Participants who did not complete the follow-up survey (N = 182)
Variables	M	SD	M	SD	t	Effect size (Cohen’s d)^a	p
Age	39.73	12.75	37.11	13.10	2.047	0.20	.041
PTSD symptoms	11.21	6.88	11.23	6.58	−0.033	−0.00	.947
DSO symptoms	12.27	6.65	13.37	6.64	−1.675	−0.17	.095
Childhood betrayal trauma	1.18	1.33	1.35	1.42	−1.247	−0.12	.213
Childhood non-betrayal trauma	0.66	0.88	0.49	0.79	1.997	0.20	.046
Adulthood betrayal trauma	1.06	1.16	0.96	1.08	0.891	0.09	.374
Adulthood non-betrayal trauma	0.73	1.01	0.69	0.96	0.389	0.04	.697
Variables	%		%		χ^2 (df = 1)	Effect size (Phi [φ])^b	p
Gender (female)	83.0		78.6		1.324	0.06	.250
Education level (undergraduate or above)	59.6		52.7		1.922	0.07	.166
Marriage status (married or common-law)	37.4		29.7		2.700	0.08	.100
Seen a psychiatrist in the past 12 months (reported at baseline)	29.1		31.3		0.231	−0.02	.631
Were financially dependent on social welfare (reported at baseline)	10.9		14.3		1.093	−0.05	.296
Rate of probable PTSD/Complex PTSD as screened by the ITQ (assessed at baseline)	36.5		38.5		0.163	−0.02	.686

Notes. PTSD: post-traumatic stress disorder; DSO: disturbances in self-organization; ITQ: the International Trauma Questionnaire.

^a0.2 = small effect, 0.5 = medium effect, 0.8 = large effect.

^b0.1 = small effect, 0.3 = medium effect, 0.5 = large effect.

Measures

In addition to questions about demographic and health backgrounds, participants completed the following validated self-report measures in the web-based surveys. Attention check items (e.g., 5 + 3=?) were also included in the surveys to ensure the validity of the responses. Except for trauma exposure, all variables were assessed at both timepoints.

Exposure to traumatic events was assessed using the Brief Betrayal Trauma Survey (BBRS), which is a 24-item questionnaire that asks about 12 types of childhood and adulthood traumatic events, which can be further categorized as betrayal and non-betrayal trauma (Goldberg & Freyd, 2006). The Chinese version of the BBTS was reported to be reliable and has been used in previous studies (Fung et al., 2022; Wu et al., 2022). The BBTS was only included in the baseline survey and it had acceptable internal consistency (α = 0.696).

PTSD and C-PTSD were assessed using the International Trauma Questionnaire (ITQ), which is a commonly used, well-validated assessment tool for ICD-11 PTSD and C-PTSD (Cloitre et al., 2018, 2021). The ITQ has 18 items and can assess both PTSD and DSO symptoms. The ITQ can generate the PTSD and DSO symptom scores, respectively, as continuous variables. Additionally, the ITQ can be used to make a provisional diagnosis of PTSD/C-PTSD according to ICD-11 rules (for the scoring manual and diagnostic algorithms, see: https://www.traumameasuresglobal.com/itq). To meet the criteria for ICD-11 PTSD, one must endorse (2 = “moderately” or above) at least one symptom from each of the three PTSD symptom clusters, as well as functional impairment associated with these symptoms, and report at least one lifetime traumatic event on the BBTS. To meet the criteria for ICD-11 C-PTSD, one must be positive for PTSD and also endorse (2 = “moderately” or above) at least one symptom from each of the DSO symptom clusters as well as the associated functional impairment. A person can test negative for both conditions, test positive for PTSD, or test positive for C-PTSD (see Cloitre et al., 2019). It is important to note that self-report assessments cannot be used to make official diagnosis. However, self-report assessments are widely used in the field for research purposes. For example, the PCL-5 is a self-report screening measure but it can be used to make a provisional PTSD diagnosis according to DSM-5 rules (Hyland et al., 2017; National Center for PTSD, n.d.); the Self-Report Dissociative Disorders Interview Schedule can also assess probable dissociative disorders (Ross & Browning, 2017). Similarly, the ITQ has been widely used as a self-report measure to make provisional PTSD/C-PTSD diagnosis and estimate the prevalence of (probable) ICD-11 PTSD/C-PTSD in both clinical and nonclinical samples (e.g., Cloitre et al., 2019; Ho et al., 2020; Howard et al., 2021; Hyland et al., 2017; McGinty et al., 2023; 2023). In line with this literature, we also investigated probable PTSD/C-PTSD in our sample using the ITQ. The Chinese version of the ITQ has also been validated, with satisfactory test-retest reliability (r = 0.40–0.75), semantic equivalence (r = 0.51–0.94) and construct validity (Ho et al., 2019).

Self-criticism was measured using the Levels of Self-Criticism Scale (LOSC), which is a 22-item reliable and valid measure of negative self-criticism with two main factors: comparative self-criticism and internalized self-criticism (Thompson & Zuroff, 2004). The LOSC was translated into Chinese using a collaborative approach (Chan et al., 2017; Khosravani & Dastjerdi, 2013), with a team of linguist and social workers; its face validity was confirmed by PhD-level psychology and nursing researchers.

Self-compassion was measured using the Self-compassion Scale (SCS), which is a 26-item measure of self-compassion with excellent reliability with six factors, including self-kindness, common humanity, mindfulness, self-judgment, isolation, and overidentification (Neff et al., 2019). The Chinese version of the SCS was also found to have good internal consistency (α = 0.84) and construct validity, and its six-factor structure was also confirmed (Chen et al., 2011).

Ruminative responses were measured using the Ruminative Responses Scale (RRS), which is a 10-item self-report measure of rumination developed based on the original Response Styles Questionnaire (Nolen-Hoeksema, 1991; Treynor et al., 2003). The Chinese version of the RRS has good internal consistency, test-retest reliability and construct validity (He et al., 2021).

Event centrality was measured using the Centrality of Event Scale (CES), which is a 20-item measure with excellent reliability and good validity (Berntsen & Rubin, 2006). The Chinese version of the CES was found to have excellent internal consistency, good test-retest reliability, and good concurrent and predictive validity (Kung et al., 2022).

Experiential avoidance was measured using the Acceptance and Action Questionnaire II (AAQ-II), which is a 7-item measure with good internal consistency, test-retest reliability and construct validity (Bond et al., 2011). The Chinese version of the AAQ-II was reported to have adequate reliability, a one-factor structure, and good construct validity (Zhang et al., 2014).

In terms of recent service/treatment usages, we asked the participants the following questions at follow up: “In the past 3 months, did you receive medication treatments for emotional or psychological problems?” and “In the past 3 months, did you receive counselling or psychotherapy services for emotional or psychological problems?” Participants could answer “yes” or “no.”

Data analysis

SPSS 22.0 was used for statistical analysis.

To answer the first research question regarding persistence, we reported how many participants with probable C-PTSD at baseline would remain to meet the required criteria at follow up. In addition, we conducted chi-square tests to examine the differences between participants with and without probable C-PTSD at baseline in the rates of meeting PTSD/C-PTSD criteria at follow-up, so as to examine whether C-PTSD would be more persistent than PTSD.

To answer the second research question regarding psychological predictors of PTSD and DSO symptoms, we conducted correlation and regression analyses. First, we reported the descriptive analyses of and correlations among the studied variables. We then conducted hierarchical multiple regression analyses to examine whether the studied psychological factors at baseline would be associated with PTSD and DSO symptoms, respectively, at follow up, after controlling for baseline scores of the dependent variable as well as treatment usage. To do so, we first tested whether all assumptions for multiple regression were met. For example, in both sets of hierarchical multiple regression analyses, there was independence of residuals, as assessed by a Durbin-Watson statistic of 1.786 and 2.066; although there were some correlations larger than 0.7 among the independent variables of interest (Table 2), all the VIF values were below 4, and therefore there were no obvious problems of multicollinearity (Yoo et al., 2014). Also, we assessed the Q-Q Plot and found that the assumption of normality was met. To answer the research question, we first entered the controlled variables into the regression model in Step 1. Next, in Step 2, we entered the five psychological factors into the regression model and examined if there would be statistically significant improvement in the prediction model and which baseline psychological factors would have the strongest association with the dependent variable (i.e., PTSD/DSO symptoms at Time 2).

Table 2. Descriptive analyses of and correlations among variables included in the present study (N = 230).

Description					Correlations
Variables	Measures	Possible range	Mean (SD)	Internal consistency (α) in our sample at both timepoints	1	2	3	4	5	6	7	8	9
1. T1 PTSD symptoms	International Trauma Questionnaire (ITQ)	0–24	11.2 (6.88)	0.884–0.909	1
2. T1 DSO symptoms	ITQ	0–24	12.3 (6.64)	0.904–0.913	0.737***	1
3. T1 Self-criticism	Levels of Self-Criticism Scale (LOSC)	22–154	94.1 (24.0)	0.912–0.927	0.598***	0.754***	1
4. T1 Self-compassion	Self-compassion Scale (SCS)	1–5	2.92 (0.71)	0.921–0.929	−0.558***	−0.695***	−0.826***	1
5. T1 Ruminative response	Ruminative Responses Scale (RRS)	10–40	27.1 (5.54)	0.790–0.839	0.316***	0.379***	0.302***	−0.203**	1
6. T1 Event centrality	Centrality of Event Scale (CES)	20–100	70.2 (15.5)	0.927–0.931	0.358***	0.394***	0.258***	−0.234***	0.440***	1
7. T1 Experiential avoidance	Acceptance and Action Questionnaire II (AAQ-II)	7–49	29.7 (11.5)	0.933–0.939	0.675***	0.800***	0.752***	−0.747***	0.422***	0.371***	1
8. T2 PTSD symptoms	ITQ	0–24	10.8 (6.06)	0.884–0.909	0.723***	0.646***	0.531***	−0.484***	0.313***	0.274***	0.641***	1
9. T2 DSO symptoms	ITQ	0–24	11.9 (6.36)	0.904–0.913	0.631***	0.789***	0.664***	−0.642***	0.344***	0.318***	0.732***	0.721***	1

Notes: PTSD: post-traumatic stress disorder; DSO: disturbances in self-organization.

***p<.001.

Finally, as noted above, we conducted exploratory analyses to examine the potential differences between participants with and without probable C-PTSD. For this purpose, we conducted chi-square tests for categorical variables and one-way ANOVA (with Post hoc analyses using Bonferroni correction) for continuous variables. Effect sizes were calculated to determine the level of group differences using Phi (φ) coefficient for (2 × 2) categorical variables and Cohen’s d for continuous variables (Cohen, 1988; Wearden, 2010).

Results

Research question 1: Persistence of ICD-11 PTSD and C-PTSD

A total of 230 participants were included. At baseline, 66 participants met the ICD-11 criteria for probable C-PTSD on the ITQ, and 18 participants met the criteria for probable PTSD. As reported in Table 3, 81.8% of participants with probable C-PTSD at baseline remained to have probable PTSD/C-PTSD at follow-up, while 55.6% of participants with probable PTSD at baseline remitted at follow-up. The findings indicated that C-PTSD was more persistent than PTSD in our sample.

Table 3. Differences between participants with and without probable complex post-traumatic stress disorder (C-PTSD) at baseline.

	a) Without PTSD (N = 146)	b) With probable PTSD (N = 18)	c) With probable C-PTSD (N = 66)		Effect size (Cohen’s d)^a
Variables	Mean (SD)	Mean (SD)	Mean (SD)	F (post-hoc)		p
Age	41.56 (12.93)	38.00 (12.49)	36.15 (11.74)	4.398 (a > c)	c vs a: −0.44 c vs b: −0.15	.013
Childhood betrayal trauma	0.91 (1.17)	0.78 (1.11)	1.89 (1.45)	15.056 (c > b=a)	c vs a: 0.74 c vs b: 0.86	<.001
Childhood non-betrayal trauma	0.56 (0.85)	0.39 (0.61)	0.95 (0.94)	5.748 (c > b=a)	c vs a: 0.44 c vs b: 0.71	<.001
Adulthood betrayal trauma	0.79 (1.00)	1.06 (1.21)	1.67 (1.26)	14.653 (c > a; c = b)	c vs a: 0.77 c vs b: 0.49	<.001
Adulthood non-betrayal trauma	0.55 (0.89)	0.33 (0.49)	1.24 (1.18)	13.624 (c > b=a)	c vs a: 0.80 c vs b: 1.00	<.001
Variables	%	%	%	χ^2 (df = 2)	Effect size (Phi [φ])^b	p
Gender (female)	80.8	83.3	87.9	1.609	c vs a: 0.09 c vs b: 0.06	.447
Education level (undergraduate or above)	61.0	66.7	54.5	1.185	c vs a: −0.06 c vs b: −0.10	.553
Marriage status (married or common-law)	40.4	44.4	28.8	3.038	c vs a: −0.11 c vs b: −0.14	.219
Seen a psychiatrist in the past 12 months (reported at baseline)	21.2	44.4	42.4	12.106	c vs a: 0.22 c vs b: −0.02	.002
Were financially dependent on social welfare (reported at baseline)	8.2	5.6	18.2	5.226	c vs a: 0.15 c vs b: 0.14	.073
Received medication treatments due to psychological problems in the past 3 months (reported at follow-up)	16.4	33.3	40.9	15.367	c vs a: 0.27 c vs b: 0.06	<.001
Received counseling or psychological treatments in the past 3 months (reported at follow-up)	16.4	33.3	48.5	24.110	c vs a: 0.34 c vs b: 0.12	<.001
Meeting criteria for probable PTSD/C-PTSD at follow-up	16.4	44.4	81.8	83.408	c vs a: 0.63 c vs b: 0.35	<.001

Notes.

^a0.2 = small effect, 0.5 = medium effect, 0.8 = large effect.

^b0.1 = small effect, 0.3 = medium effect, 0.5 = large effect.

Research question 2: Psychological predictors of PTSD and DSO symptoms

As noted in the data analysis subsection, we reported the descriptive analyses of and correlations among the studied variables in Table 2. We then conducted hierarchical multiple regression analyses to examine which baseline psychological factors would predict subsequent PTSD and DSO symptoms after controlling for baseline PTSD/DSO symptoms and treatment usage during the follow-up period. The findings are summarized in Table 4. In both analyses, the addition of the five psychological factors to the prediction of PTSD/DSO symptoms (Step 2) led to a statistically significant increase in R². We found that baseline experiential avoidance was the only significant psychological predictor for both PTSD (β = 0.271, p = .001) and DSO (β = 0.200, p = .011) symptoms at follow-up.

Table 4. Hierarchical multiple regression predicting complex post-traumatic stress disorder (PTSD) symptoms at follow-up (N = 230).

	Classical PTSD symptoms								Disturbances in self-organization (DSO) symptoms
Variables	B	S.E.	95% CI for B	β	p	F	ΔR^2	ΔF	B	S.E.	95% CI for B	β	p	F	ΔR^2	ΔF
Step 1						89.021***	0.542	89.021***						128.814***	.631	128.814***
Baseline scores	0.605	0.042	[0.522, 0.687]	0.686	<.001				0.724	0.722	[0.644, 0.805]	0.756	<.001
Received medication treatments due to psychological problems in the past 3 months (reported at follow-up)	1.668	0.763	[0.164, 3.172]	0.119	.030				0.782	0.710	[−0.642, 2.205]	0.053	.280
Received counseling or psychological treatments in the past 3 months (reported at follow-up)	0.448	0.758	[−0.1046, 1.942]	0.033	.555				0.861	0.722	[−0.539, 2.261]	0.060	.227
Step 2						37.867***	0.037	3.837**						52.931***	.026	3.362**
Baseline scores	0.474	0.054	[0.367, 0.580]	0.537	<.001				0.508	0.070	[0.370, 0.645]	0.530	<.001
Received medication treatments due to psychological problems in the past 3 months (reported at follow-up)	1.388	0.762	[−0.114, 2.889]	0.099	.070				0.448	0.721	[−0.947, 1.869]	0.030	.535
Received counseling or psychological treatments in the past 3 months (reported at follow-up)	−0.007	0.748	[−1.481, 1.468]	0.000	.993				0.320	0.707	[−1.073, 1.714]	0.022	.651
Self-criticism	0.019	0.022	[−0.024, 0.061]	0.074	.390				0.007	0.021	[−0.035, 0.049]	0.028	.729
Self-compassion	0.958	0.739	[−0s.498, 2.414]	0.112	.196				−0.703	0.700	[−2.082, 0.677]	−0.078	.316
Ruminative response	0.037	0.057	[−0.076, 0.150]	0.033	.524				0.037	0.054	[−0.070, 0.144]	0.032	.498
Event centrality	−0.014	0.020	[−0.053, 0.026]	−0.035	.497				−0.004	0.019	[−0.041, 0.033]	−0.009	.840
Experiential avoidance	0.143	0.044	[0.057, 0.229]	0.271	.001				0.111	0.043	[0.025, 0.196]	0.200	.011

T1: Timepoint 1; T2: Timepoint 2.

**p<.01; ***p<.001.

Further exploratory analyses: differences between participants with and without probable C-PTSD

At baseline, compared with participants without PTSD and those with probable PTSD only, participants with probable C-PTSD at baseline reported significantly more childhood betrayal and non-betrayal traumatic events as well as more adulthood non-betrayal traumatic events. In addition, participants with probable C-PTSD also reported significantly more adulthood betrayal traumatic events than those without PTSD. No differences were found between participants without PTSD and those with probable PTSD only at baseline in terms of the numbers of reported traumatic events. The findings are reported in Table 3.

Compared with participants without PTSD, participants with probable PTSD and C-PTSD at baseline were more likely to report visiting a psychiatrist at baseline (p = .002) and receiving medication and psychological treatments at follow-up (p < .001). Participants with C-PTSD also reported a higher rate of financial dependence on social welfare, although the group differences were not statistically significant (p = .073). The findings are reported in Table 3.

Discussion

This study contributes to the C-PTSD literature by examining the persistence and psychological predictors of ICD-11 C-PTSD. The two major findings included: (1) C-PTSD was more persistent than PTSD in our sample; and (2) Although self-criticism, self-compassion, ruminative response, event centrality, and experiential avoidance are all well-documented psychological factors associated with post-traumatic symptoms, our study indicates that experiential avoidance was the only significant, strongest psychological factors for subsequent PTSD and DOS symptoms. The findings and their implications are discussed below.

Addressing C-PTSD as an important public health concern

Since C-PTSD may not be rare in the general population (Cloitre et al., 2019; Karatzias et al., 2019), is associated with higher levels of functional impairments (Po et al., 2023) and is also persistent (as shown in the present study) (also see Redican et al., 2022), more resources should be mobilized to prevent C-PTSD and provide timely support for those in need. We also found that C-PTSD is associated with more needs for medical resources (Table 3). Thus, this study provides further evidence that C-PTSD should be treated as a public health concern. More research should be funded to understand how to better prevent, identify, and treat C-PTSD. Previous studies showed that trauma-related disorders are commonly overlooked, even in clinical settings (Gagnon-Sanschagrin et al., 2022), potentially leading to delays in appropriate interventions. In our convenience, nonpsychiatric sample, less than a half (48.5%) of participants with probable C-PTSD (n = 66) at baseline reported receiving psychological treatment (which is the primary treatment for trauma-related disorders) at follow-up. As C-PTSD symptoms can persist, they should be regularly screened for in community and clinical settings so that timely interventions can be offered and that their long-term individual and social consequences could be reduced.

Identifying factors associated with C-PTSD

The identification of risk and protective factors is vital for the development of targeted preventive and treatment strategies. Our study shows that, among a variety of psychological factors, experiential avoidance was the only prospective predictor of both PTSD and DSO symptoms. The factors that predict post-traumatic symptoms indeed require further investigation. Prior research on psychological factors associated with PTSD mainly focused on simple PTSD rather than C-PTSD. Findings from simple PTSD might not be entirely generalized to the context of C-PTSD. There may also be cultural differences in terms of the psychological predictors of post-traumatic symptoms. Previous findings from the Western populations may not be generalizable to our Chinese sample. These reasons may partly explain why only experiential avoidance but not other psychological factors could predict C-PTSD symptoms in our sample. Moreover, statistically, inclusion of experiential avoidance in the current regression models might reduce the independent predictive effects of other included factors. It is also important to note that many of the PTSD-related psychological factors were identified using a cross-sectional design (e.g., Crapolicchio et al., 2021; Scoglio et al., 2018) and/or relying on correlation analyses without controlling for other confounding variables (e.g., Gehrt et al., 2018; Winders et al., 2020) in previous studies. Therefore, the predictive effects of these psychological factors on C-PTSD symptoms may have been reduced in our longitudinal study where some other variables were also taken into account. Additionally, avoidance is not just a major feature of traumatization; it may result from dissociative phobia, which involves avoiding trauma-related personality parts (Steele et al., 2005). Avoidance can prevent survivors from confronting, processing, and integrating trauma-related aspects of the self, such as traumatic memories and emotions. This can worsen and maintain other trauma-related symptoms, including intrusions and maladaptive thoughts, and reinforce other maladaptive coping strategies (Ross et al., 2017). Our findings indicate that interventions targeting experiential avoidance may have particular potential to change the trajectory of post-traumatic psychopathology. For example, acceptance and commitment therapy, body-oriented interventions, and mindfulness may be beneficial in reducing experiential avoidance (Classen et al., 2021; Li et al., 2023; Na et al., 2022). Replication of our study, however, is necessary to establish the robustness and generalizability of the results. Given the importance of targeted interventions, more research on the mechanism of change and predictors of post-traumatic symptoms is required to inform the advancement of treatment for trauma-related disorders. In order to inform more comprehensive and integrated interventions, future studies should also consider social and biological factors, in addition to psychological factors.

Limitations of this study and the needs for follow-up studies

This study has the strengths of employing a longitudinal design, using reliable and valid measures, demonstrating the persistence of probable C-PTSD, and including a variety of psychological factors in predicting post-traumatic symptoms. However, this study also has some limitations. First, although online methods have been widely used in epidemiological research (Chan et al., 2017; Fung et al., 2019), our convenience sample was not representative of the general population; self-selection bias may be involved in survey studies. Second, this study had a relatively high attrition rate. Third, while our assessment tools are validated, and while the ITQ has been commonly used to investigate probable ICD-11 C-PTSD (e.g., Cloitre et al., 2019; McGinty et al., 2023; Williamson et al., 2021), we relied on self-report data and did not conduct structured interviews to confirm the diagnoses. Traditionally, in the field of psychiatry, structured interviews are regarded as “gold standard” in assessing mental disorders. Having said that, self-report assessments are not less valid than structured interviews; interviews may have more measurement errors (Hyland & Shevlin, 2023). Self-report assessments are also commonly used in epidemiological studies on psychiatric disorders, including trauma-related disorders (Hyland & Shevlin, 2023). Nevertheless, future studies should confirm our findings using data from diverse sources, including chart reviews and diagnostic interviews. Fourth, we only measured trauma histories at baseline, and did not know whether participants had experienced new traumatic events during the follow-up period; these new traumatic events, if any, may be confounding variables that should be considered in the future. Future studies should further investigate the persistence and associated factors of C-PTSD in diverse clinical and nonclinical samples. Additionally, our finding that experiential avoidance is the strongest predictor of PTSD and DSO symptoms should be replicated, as it may have significant implications for the prevention and treatment of trauma-related disorders.

Concluding remarks

This study provides evidence that probable C-PTSD persists over time and is associated with greater use of mental health services in the community. Among various psychological factors, experiential avoidance is a significant predictor of both PTSD and DSO symptoms. The findings offer insights into the prevention and treatment of trauma-related disorders.

Ethical approval

The project was granted ethical approval by Chinese University of Hong Kong (SBRE‐22‐0051). All participants provided online written informed consent before they completed the survey.

Acknowledgements

The last author received The RGC Postdoctoral Fellowship Scheme 2022/2023 from the Research Grants Council (RGC), Hong Kong.

Data availability statement

Data that support the findings of this study are available from the corresponding authors upon reasonable request.

Additional information

Funding

No funding was received for conducting this study.

Notes on contributors

Stanley Kam Ki Lam

Stanley Kam Ki Lam, PhD, RN, is a research assistant professor at Nethersole School of Nursing, Faculty of Medicine, The Chinese University of Hong Kong. His research focuses on mental health and psychological interventions.

Albe Sin Ying Ng

Albe Sin Ying Ng, PhD, is a senior lecture at Department of Psychology, Hong Kong Shue Yan University. Her research focuses on complex trauma, dissociation, post-traumatic growth, resilience, sleep, and psychophysiology.

Chengrui Rachel Zhou

Chengrui Rachel Zhou is a research assistant at Nethersole School of Nursing, Faculty of Medicine, The Chinese University of Hong Kong.

Suet Lin Hung

Suet Lin Hung, PhD, RSW, is a Professor and Department head at Department of Social Work, Hong Kong Baptist University. Her work focuses on women and families, genderbased Violence, community work practice, and narrative therapy.

Hong Wang Fung

Hong Wang Fung, PhD, RSW, is a postdoctoral research fellow at Department of Social Work, Hong Kong Baptist University. His research focuses on trauma, dissociation, social work practice, and social psychiatry.