PTSD and Complex PTSD in Residential Treatment for Eating Disorders: Moderating Effects on Symptom Severity and Outcome Trajectory

ABSTRACT

Objective

Eating disorders (EDs) and symptoms of trauma commonly co‐occur, yet research is limited on how trauma affects ED treatment outcomes. This is particularly true for complex post‐traumatic stress disorder (CPTSD). Differentiating between the treatment impacts of PTSD and CPTSD (which includes both PTSD symptoms and disturbances in self‐organization [DSO]) may help ED providers address this common comorbidity.

Method

The current study included 95 women (M _age   = 26 years) with EDs (largely anorexia nervosa) who received residential treatment (M = 81 days). Participants completed measures of ED symptoms, anxiety, depression, body mass index (BMI), ED‐specific health‐related quality‐of‐life (ED‐HRQoL) impairment, functional disability, and trauma symptoms at admission, week 4 of treatment, discharge, and 6 months post‐discharge.

Results

All outcomes except BMI were more severe at admission and week 4 of treatment for individuals with comorbid trauma (based on probable CPTSD or exceeding the clinical threshold for PTSD and DSO symptom domains); however, these differences resolved by discharge and remained non‐significant at follow‐up. Some forms of comorbid trauma moderated outcome trajectories for anxiety, depression, and disability (but not ED symptoms), such that individuals with comorbid trauma showed slower improvement early in treatment, steeper improvement later in treatment, and greater resurgence after discharge.

Discussion

These findings highlight that PTSD and CPTSD symptom domains may be associated with more severe ED outcomes early in residential treatment that resolve by discharge, and may predict differential treatment response for secondary outcomes. Implications are discussed for clinical assessment and treatment of comorbid trauma‐related disorders in residential care.

Trial Registration

The study was prospectively registered on the Australian and New Zealand Clinical Trials Registry in November 2021, registration number ACTRN12621001651875

Summary.

• PTSD is recognized as a common comorbidity with eating disorders, but research has been limited on CPTSD and whether trauma moderates treatment outcomes.

• The current study found that trauma‐related symptoms related to both PTSD and complex PTSD were common in a residential setting.

• Comorbid trauma symptoms were associated with more severe eating disorder and other outcomes at admission and early in treatment, but not at discharge or follow‐up.

• Comorbid trauma symptoms may predict slower improvement early in treatment for anxiety, depression, and functioning, but steeper improvement later in treatment.

Trauma‐related symptoms are common among individuals with eating disorders (EDs), especially in higher levels of care. A review of 29 studies found that post‐traumatic stress disorder (PTSD) occurred in approximately one in four individuals with EDs receiving outpatient treatment and a third of those receiving inpatient hospitalization (Ferrell et al. 2022). Other studies suggest half of individuals in ED residential, inpatient, and partial hospitalization settings have comorbid PTSD (Brewerton et al. 2020; Rienecke et al. 2021). High rates of exposure to traumatic events among individuals with EDs have also been well‐demonstrated; in particular, forms of childhood abuse/neglect and sexual trauma in adulthood (Backholm et al. 2013; Molendijk et al. 2017).

Comorbid PTSD has consistently been shown to be associated with more severe ED symptoms (Brewerton et al. 2020; Rijkers et al. 2019; Scharff et al. 2019). A systematic review of 19 studies found that PTSD and trauma exposure were typically associated with more severe ED symptoms at treatment discharge and higher attrition rates (Day et al. 2023). More recently, another study found significantly worse ED symptoms, depression, anxiety, and ED‐related quality of life throughout residential treatment and at a 6‐month follow‐up among individuals who presented with comorbid PTSD (Brewerton et al. 2023). Evidence remains mixed, however, whether PTSD moderates the pattern of change in ED symptoms over the course of ED treatment. One study found that individuals with comorbid PTSD in residential care experienced steeper improvement in ED symptoms during treatment but also greater symptom recurrence over the post‐discharge period (Scharff et al. 2021). In contrast, others have found that PTSD was not associated with the pattern of ED symptom change from residential treatment (Brewerton et al. 2023; Mitchell et al. 2021). These findings from residential services contrast with previous studies showing poorer ED treatment outcomes with comorbid trauma, most of which were conducted in inpatient and outpatient settings (e.g., only 2 of the 19 studies in the review by Day et al. 2023 were in residential services). It is unclear whether these contrasting findings may be due to factors in the residential treatment environment. Unlike other intensive treatment settings such as inpatient hospitalization, residential treatment occurs in a live‐in, home‐like (versus clinical, hospital‐based) environment, focuses on long‐term psychological recovery (versus medical stabilization), and combines evidence‐based psychotherapies with adjunctive interventions (e.g., yoga, art therapy). Aspects of the medicalized hospital environment in inpatient settings have been reported to be restrictive and coercive (such as involuntary refeeding) (Rankin et al. 2023; Touyz et al. 2024), which could be re‐traumatizing for individuals with comorbid PTSD. Residential care also differs from day programs, which could be experienced as insufficiently intensive for the more severe ED psychopathology often associated with comorbid trauma.

While research thus remains inconclusive as to whether comorbid PTSD affects ED treatment outcomes, studies are even sparser for complex PTSD (CPTSD). CPTSD became a new trauma‐related diagnosis in the ICD‐11 (International Classification of Diseases; World Health Organization 2019) to reflect growing evidence of distinct symptom profiles within PTSD (Cloitre et al. 2020; Karatzias et al. 2017). CPTSD includes two symptom domains: PTSD symptoms (heightened threat perception, re‐experiencing, avoidance) and disturbances in self‐organization (DSO; affective dysregulation, negative self‐concept, disturbances in relationships) (World Health Organisation 2019). DSO symptoms may be particularly relevant to individuals with eating disorders given that emotion dysregulation has been shown to mediate the effects of childhood maltreatment on eating disorder psychopathology and poorer treatment outcomes (Cassioli et al. 2022; Dawson et al. 2022). CPTSD reflects symptoms often experienced by survivors of prolonged, interpersonal traumatic experiences, such as childhood abuse and neglect (van der Kolk 2007; Van Nieuwenhove and Meganck 2019). CPTSD is not currently included in the DSM (Diagnostic and Statistical Manual of Mental Disorders; American Psychiatric Association 2013).

A recent study including outpatient, day program, and residential settings established the first prevalence rates for ICD‐11 CPTSD in EDs, finding that 28.4% of the sample met criteria for CPTSD and 3.8% for ICD‐11 PTSD (Day et al. 2024a, 2024b). These prevalence rates support assumptions that previous prevalence studies of DSM‐5 PTSD in EDs have likely captured cases of both PTSD and CPTSD. In addition to finding that CPTSD was significantly more common than PTSD in this population, (Day et al. 2024a, 2024b) also found that the severity of both PTSD and DSO symptoms was independently associated with more severe ED symptoms at pre‐treatment. However, as highlighted by recent reviews (Convertino and Mendoza 2023; Day et al. 2023), there have not yet been any studies that have differentiated between the effects of PTSD, CPTSD, and their symptom domains on ED treatment outcomes. Understanding whether there are distinct treatment effects may help clinicians to provide more targeted treatment of individuals with comorbid EDs and trauma, enhancing the effectiveness of ED higher levels of care for complex presentations.

Building upon recently established prevalence rates for CPTSD in EDs (Day et al. 2024a, 2024b), the current study aimed to investigate how PTSD and CPTSD symptoms affect clinical outcomes from ED residential treatment. It was hypothesized that (1) ED symptom severity and other clinical outcomes (body mass index [BMI], anxiety, depression, and functional outcomes) would be more severe throughout treatment and post‐treatment for individuals with clinical levels of trauma‐related symptoms; and that (2) PTSD and CPTSD symptoms would not moderate the trend of change in clinical outcomes over time (based on recent findings by Brewerton et al. 2023 and Scharff et al. 2021 as described previously).

1. Method

1.1. Participants

This study included 95 participants with an ED who received residential treatment between July 2021 and October 2023. The residential service was Wandi Nerida, located in Queensland, Australia. Wandi Nerida accepts individuals aged 16 years or older with any primary ED presentation and appropriate medical stability. Full admission criteria have been described elsewhere (Day et al. 2024a, 2024b).

1.2. Measures

1.2.1. Demographic Characteristics

Participants were asked to report their age, gender, country of birth, residential postcode, and ED history. Postcode was used to determine socioeconomic index for area (SEIFA) score (Australian Bureau of Statistics 2023), a percentile indicator of relative socioeconomic advantage or disadvantage.

1.2.2. Clinical Outcomes

1.2.2.1. Eating Disorder Symptoms

ED diagnosis was extracted from participants' clinical record based on psychologist and psychiatrist assessment at admission. The Eating Disorder Examination Questionnaire (EDE‐Q v.6; Fairburn and Beglin 1994, 2008) provided a self‐report measure of disordered eating attitudes and behaviors over the past 28 days. The EDE‐Q generates a global score for ED symptoms from 0 to 6 (higher scores reflecting greater symptom severity). The EDE‐Q has been validated in Australian community samples (Mond et al. 2004a, 2004b) and had good internal consistency in the current sample, McDonald's ω = 0.95–0.96 (across survey timepoints). BMI (kg/m²) was measured by clinical staff during treatment and by self‐report at post‐discharge. BMI was only examined as a treatment outcome for participants whose clinical record indicated weight restoration was a treatment goal (91.6% of the sample).

1.2.2.2. Anxiety and Depression

Anxiety and depression were measured using the Generalized Anxiety Disorder Screener‐7 (GAD‐7; Spitzer et al. 2006) and the Patient Health Questionnaire (PHQ‐9; Kroenke et al. 2001). Scores range from 0 to 21 on the GAD‐7 and 0–27 on the PHQ‐9, with higher scores representing more severe anxiety and depression symptoms. These scales have shown acceptable psychometric properties (Costantini et al. 2021; Plummer et al. 2016) and had satisfactory internal consistency in the current sample (McDonald's ω = 0.91–0.95 for GAD‐7; McDonald's ω = 0.82–0.90 for PHQ‐9).

1.2.2.3. Functioning

ED‐specific health‐related quality of life (ED‐HRQoL) impairment was measured using the Clinical Impairment Assessment (CIA; Bohn and Fairburn 2008). Higher scores (ranging 0–48) indicate greater HRQoL impairment due to ED symptoms. The CIA has been validated in clinical and community samples (Raykos et al. 2019; Reas et al. 2010) and had good internal consistency in the current sample, McDonald's ω = 0.95–0.97.

Functional disability was measured using the 12‐item version of the World Health Organization Disability Assessment Schedule (WHODAS‐12; Üstün et al. 2010). Scores range from 12 to 60, with higher scores reflecting greater difficulty engaging in regular everyday activities. The WHODAS‐12 has been validated in community and psychiatric samples (Axelsson et al. 2017; Üstün et al. 2010) and had acceptable internal consistency in the current sample, McDonald's ω = 0.85–0.92.

1.2.3. Trauma‐Related Symptoms

The International Trauma Questionnaire (ITQ; Cloitre et al. 2018) measured PTSD and CPTSD symptoms based on the ICD‐11 diagnostic criteria. Scores on the ITQ are used to indicate probable diagnoses of PTSD and CPTSD, and its validity in distinguishing between these diagnoses has been supported by multiple studies (Camden et al. 2023; Murphy et al. 2020). Scores on PTSD and DSO symptom domains range 0–24, with a clinical cut‐off of 12 on each domain. For a probable diagnosis of PTSD, participants must meet this clinical cut‐off on the PTSD domain (reflecting endorsement of each type of PTSD symptom [re‐experiencing, avoidance, sense of threat] at ‘moderately’ or higher) and report at least moderate functional impairment as a result of these symptoms. For CPTSD, participants must meet the clinical cut‐off on both PTSD and DSO (affective dysregulation, negative self‐concept, disturbances in relationships) symptom domains and functional impairment from both. A diagnosis of CPTSD supersedes one of PTSD such that they are never comorbid. The ITQ symptom domains showed acceptable internal consistency in the current sample, McDonald's ω = 0.92 for PTSD items, ω = 0.80–0.93 for DSO items. All participants were asked whether they had experienced a ‘stressful or traumatic life event/s’ that continued to affect them. However, full history of exposure to traumatic events was not conducted.

1.3. Procedure

The current study was a repeated measures cohort study with outcomes measured at admission, week 4 of treatment, discharge, and 6 months post‐discharge. The study was prospectively registered on the Australian and New Zealand Clinical Trials Registry in November 2021, registration number ACTRN12621001651875. Data collection infrastructure was provided by the TrEAT Registry (ACSQHC‐ARCR‐279), a registry of ED services in Australia and New Zealand that supports the use of outcomes data to inform research and clinical practice (Australian Register of Clinical Registries, 2024; The TrEAT Registry 2024).

Individuals were invited to participate at admission and completed the consent form and measures using online surveys via Qualtrics. Surveys were administered by service staff during treatment and by email at follow‐up. For a description of the treatment approach and clinical effectiveness of the service, see (Day et al. 2024a, 2024b) and Day et al. (2025). Ethics approval was granted by the Western Sydney University Human Research Ethics Committee (approval numbers H14742, H14478).

1.4. Data Analysis

Data were analyzed using R Statistical Software (v4.3.1; R Core Team 2023). Descriptive statistics were calculated for the clinical outcomes at each timepoint by comorbid trauma status (individuals with and without probable PTSD and CPTSD, and threshold symptoms on PTSD and DSO domains). Mann Whitney U‐tests (due to non‐parametric data) assessed significant differences in outcome severity by trauma status at each timepoint.

Complete data was available for 76.8%–100% of cases at admission, 60% at week 4 of treatment, 72.6%–78.9% at discharge, and 65.3%–67.4% of cases at 6‐months post‐discharge. This missing data rate is similar to or lower than other previous studies of residential care (e.g., Brewerton and Costin 2011; Scharff et al. 2021). The rate of complete data varied by outcome at some timepoints because the measure of trauma‐related symptoms was introduced partway through the data collection period as part of a larger evaluation of residential treatment (Day et al. 2024a, 2024b; Day et al. 2025). Mann Whitney U‐tests indicated that average scores on all outcome and trauma‐related variables at admission did not differ based on whether participants were missing data at subsequent timepoints, all p > 0.05.

To identify meaningful covariates, we investigated whether demographic characteristics at admission were associated with baseline outcomes and data missingness. Mann Whitney U‐tests indicated that data missingness was not associated with baseline age, BMI, socioeconomic status, or ED illness duration (see Table S1), except for older age among participants with missing data at week 4 of treatment (n = 34/95). Spearman correlations indicated that age and BMI at admission, socioeconomic status, and ED illness duration were also not significantly associated with baseline outcomes (see Table S2). As the difference in age between participants who had complete versus missing data at week 4 was comparable to mean age differences between other non‐significant comparisons and all correlations including age were also non‐significant, we did not include any of the examined covariates in subsequent analyses.

Mixed effects models examined trauma‐related symptoms as a moderator of change in clinical outcomes over time. Mixed effects models are robust to violations of distributional assumptions (Schielzeth et al. 2020), are able to handle missing data better than alternatives such as analysis of variance, and can assess for non‐linear patterns of change (Mirman 2017). Models were conducted using the R package lme4 (Bates et al. 2015), with outcomes nested within individual participants. Linear, quadratic, and cubic fixed effects of time were added sequentially alongside trauma‐related predictors and their interactions. The level of the outcome variable at baseline was included as a covariate such that findings for the effects of time and trauma status control for the severity of the outcome at treatment admission. Treatment duration was also examined as a covariate but was not found to be significant in any of the models; as such, it was not retained in the final models. A power analysis using the R package pwr (Champely 2020) estimated that a minimum sample size of 93 was required to achieve 80% power in detecting a medium effect size based on including eight predictors (three time polynomials, their interactions with trauma status, the main effect of trauma status, and the baseline level of the treatment outcome).

Based on previous research (Day et al. 2024a, 2024b), we expected small sample sizes for probable PTSD, so we chose to also include threshold PTSD and DSO symptoms at admission as predictors. Model fit was assessed using the likelihood ratio test, with second‐ and third‐order time terms retained when a better fit was indicated, and orthogonal polynomials used to reduce multicollinearity (Mirman 2017). R package lmerTest (Kuznetsova et al. 2017) was used to calculate parameter‐specific p‐values and degrees of freedom (using the Satterthwaite approximation). Effect sizes for each fixed effect were calculated as Cohen's d (R package effectsize [Ben‐Shachar et al. 2020]) and the variance explained by fixed and both fixed and random effects across the entire model was calculated using marginal and conditional R ² (Nakagawa and Schielzeth 2013; R package performance [Lüdecke et al. 2021]). The Benjamini‐Hochberg procedure was used to control the false discovery rate.

2. Results

Participants were women aged 16–50 years (M = 26.02, SE = 0.69). Most participants (90.2%) were born in Australia, and the average SES percentile was 74.97 (SE = 2.60, range = 3–99). ED illness duration ranged from 1 to 30 years (M = 10.22, SE = 0.66) and the average BMI at admission was 19.16 (SE = 0.57). The most common ED diagnosis was atypical anorexia nervosa (40.4%), followed by anorexia nervosa restrictive subtype (38.3%), anorexia nervosa binge‐purge subtype (13.8%), and bulimia nervosa and binge eating disorder (7.5%). The likelihood of meeting criteria for CPTSD did not significantly differ by ED diagnosis, χ ² (4) = 0.58, p = 0.965, nor did the likelihood of exceeding the clinical threshold for PTSD symptom severity, χ ² (4) = 1.51, p = 0.826. Rates of exceeding the DSO symptom threshold significantly differed by ED diagnosis, χ ² (4) = 10.26, p = 0.036. All participants with anorexia nervosa binge‐purge subtype reported threshold DSO symptoms at admission (n = 11/11), as did the majority of those with atypical anorexia nervosa (n = 25/28).

At admission, less than 5% of the sample met criteria for a probable PTSD diagnosis (exact n not reported due to risk of re‐identification) and 39.2% (n = 29) met criteria for probable CPTSD. All participants who met criteria for probable PTSD or CPTSD endorsed having experienced a prior stressful or traumatic life event/s. The number of individuals with probable PTSD was too small to include in subsequent analyses. Larger numbers of participants reported threshold levels on PTSD (39.2%) and DSO (81.1%) symptom domains at admission. Treatment duration ranged from 6 to 209 days (M = 81.18 days, SE = 4.23). The average treatment duration and likelihood of attrition did not differ based on comorbid trauma (probable CPTSD, threshold PTSD or DSO symptoms), all p > 0.05 (based on Chi‐square analyses). The majority (91.4%) of participants reported seeing a mental health provider in the first 3 months' post‐discharge; most commonly, a psychologist (91.2%) and dietitian (89.1%).

2.1. Differences in Clinical Outcomes at Each Timepoint by Trauma Status

Partially consistent with hypotheses, most clinical outcomes were more severe at baseline for individuals with comorbid trauma‐related symptoms, compared to those who fell below trauma‐related symptom thresholds (see Table 1). Some of these differences by trauma symptom status persisted during, but not after, treatment. Participants with threshold PTSD symptoms at baseline had significantly more severe ED symptoms, ED‐HRQoL impairment, anxiety, depression, and functional disability at baseline, with these differences persisting at week 4 of treatment except for ED symptoms. Participants with threshold DSO symptoms reported more severe ED‐HRQoL and anxiety at baseline, and depression and functional.

TABLE 1.

Average clinical outcomes over time by trauma symptom status.

	Scale range		Mean (SE)
		Total sample	PTSD threshold+	PTSD threshold−	DSO threshold+	DSO threshold−	CPTSD+	CPTSD−
			n = 29 (39.2%)	n = 45 (60.8%)	n = 60 (81.1%)	n = 14 (18.9%)	n = 29 (39.2%)	n = 45 (60.8%)
ED symptoms	0–6
Admission		4.57 (0.13)	5.02 (0.18)	4.54 (0.18)	4.85 (0.12)	4.19 (0.42)	4.96 (0.19)	4.57 (0.17)
Week 4		3.61 (0.17)	3.93 (0.29)	3.37 (0.22)	3.63 (0.20)	3.32 (0.41)	3.95 (0.29)	3.36 (0.22)
Discharge		2.93 (0.17)	3.12 (0.30)	2.86 (0.22)	3.00 (0.19)	2.79 (0.43)	3.14 (0.30)	2.85 (0.22)
6mo FU		3.71 (0.19)	3.52 (0.42)	3.82 (0.29)	3.77 (0.26)	3.53 (0.56)	3.60 (0.44)	3.77 (0.28)
ED‐HRQoL impairment	0–48
Admission		40.38 (0.96)	43.90 (1.15)	40.13 (1.30)	43.02 (0.83)	35.57 (2.93)	43.69 (1.23)	40.27 (1.28)
Week 4		35.08 (1.54)	39.10 (2.36)	32.50 (2.10)	35.04 (1.78)	34.40 (4.20)	38.67 (2.35)	32.75 (2.13)
Discharge		27.85 (1.58)	30.84 (2.80)	26.87 (2.23)	28.60 (1.97)	27.60 (3.66)	30.84 (2.80)	26.87 (2.23)
6mo FU		31.67 (1.57)	28.88 (3.39)	33.23 (2.19)	31.78 (2.12)	31.50 (4.00)	29.67 (3.53)	32.71 (2.18)
Anxiety	0–21
Admission		15.32 (0.57)	17.45 (0.88)	14.62 (0.76)	16.45 (0.62)	12.64 (1.47)	17.62 (0.88)	14.51 (0.75)
Week 4		14.11 (0.77)	16.48 (1.18)	12.64 (1.04)	14.11 (0.88)	13.80 (2.29)	16.33 (1.18)	12.72 (1.05)
Discharge		11.29 (0.75)	12.68 (1.32)	10.84 (1.05)	11.64 (0.87)	11.20 (2.45)	12.44 (1.29)	11.00 (1.07)
6mo FU		12.87 (0.81)	14.50 (1.28)	12.47 (1.27)	13.69 (1.04)	11.30 (2.18)	14.87 (1.31)	12.35 (1.23)
Depression	0–27
Admission		19.22 (0.67)	22.79 (0.84)	18.47 (0.88)	21.27 (0.67)	15.43 (1.58)	22.48 (0.89)	18.67 (0.88)
Week 4		16.52 (0.85)	21.19 (1.27)	13.69 (0.94)	17.34 (0.96)	12.30 (1.90)	20.76 (1.31)	13.94 (0.98)
Discharge		13.25 (0.81)	15.28 (1.42)	12.71 (1.10)	14.17 (0.97)	11.40 (1.99)	15.00 (1.41)	12.89 (1.12)
6mo FU		15.73 (0.75)	17.56 (1.52)	14.97 (1.03)	16.22 (1.03)	14.60 (1.52)	17.60 (1.63)	15.03 (1.00)
Functional disability	12–60
Admission		30.34 (0.97)	33.79 (1.55)	28.07 (1.17)	31.38 (1.02)	25.71 (2.58)	33.52 (1.57)	28.24 (1.18)
Week 4		26.70 (1.18)	32.19 (1.75)	23.42 (1.31)	27.66 (1.24)	21.90 (3.14)	31.24 (1.75)	23.97 (1.40)
Discharge		24.50 (1.17)	26.04 (2.04)	23.61 (1.51)	24.77 (1.30)	23.50 (3.57)	25.44 (1.96)	24.00 (1.57)
6mo FU		24.62 (1.01)	26.56 (2.07)	22.87 (1.43)	24.75 (1.31)	22.00 (2.83)	26.07 (2.15)	23.23 (1.43)
BMI a
Admission		18.17 (0.39)	18.37 (0.96)	18.18 (0.45)	18.41 (0.51)	17.47 (0.92)	18.33 (0.93)	18.19 (0.46)
Week 4		19.58 (0.41)	20.03 (0.97)	19.77 (0.46)	19.99 (0.50)	19.20 (1.00)	19.99 (0.93)	19.79 (0.47)
Discharge		21.86 (0.41)	22.25 (0.88)	22.14 (0.49)	22.52 (0.49)	20.53 (0.89)	22.20 (0.85)	22.17 (0.50)
6mo FU		19.40 (1.01)	20.44 (2.31)	18.55 (0.70)	19.75 (1.07)	17.01 (0.93)	20.44 (2.31)	18.55 (0.70)

Note: 6mo FU, 6‐month follow‐up (6‐months post‐discharge). Bolded pairs represent statistically significant differences in scores at timepoint based on Mann Whitney U‐tests at p < 0.05. +/2212 indicates participant subgroups based on whether criteria were met for clinical threshold PTSD or DSO symptoms or a probable CPTSD diagnosis.

Abbreviations: BMI, body mass index; CPTSD, complex PTSD; DSO, disturbances in self‐organization; ED, eating disorder; ED‐HRQoL, eating disorder‐specific health‐related quality of life; PTSD, post‐traumatic stress disorder; SE, standard error.

^a Included for participants with weight restoration as a treatment goal (n = 87/95).

Disability at both baseline and week 4 of treatment. Participants with probable CPTSD had more severe ED symptoms and ED‐HRQoL impairment at baseline, and more severe anxiety, depression, and functional disability at both baseline and week 4 of treatment. ED psychopathology and BMI did not significantly differ at any timepoint based on threshold DSO symptoms. Comorbid trauma was not associated with significantly more severe clinical outcomes at discharge or 6 months post‐discharge, suggesting that any differences in psychopathology and functioning at baseline resolved by end of treatment and did not re‐emerge over the post‐treatment period. Differences were not compared based on PTSD diagnostic status due to insufficient sample size.

2.2. Moderation Effects of Trauma on Clinical Outcome Trajectories

Accounting for baseline levels of the outcome variable, there were no main effects of trauma status on ED symptoms or secondary outcomes, averaged over time (see Tables 2, 3, 4). Trauma status moderated outcome trajectories for some variables, providing mixed support for this hypothesis. Individuals with CPTSD exhibited a greater cubic trend in depressive symptoms over time. As illustrated in Figure 1a, participants with CPTSD showed less improvement in depression in the first 4 weeks of treatment, steeper improvement from week 4 to discharge, and similar symptom resurgence to those without CPTSD from discharge to the 6‐month follow‐up. Threshold DSO symptoms moderated the quadratic effect of time on anxiety, predicting a more pronounced quadratic trend. Individuals with threshold DSO symptoms showed steady improvement in anxiety from admission to discharge followed by a pronounced symptom recurrence post‐discharge, whereas those below the DSO symptom threshold showed slight worsening of anxiety early in treatment but then improvement to discharge that was maintained over the follow‐up (see Figure 1b, Table 3).

TABLE 2.

Mixed‐effects models of time and CPTSD on clinical outcomes.

		Estimate	SE	df	t	p	d	R 2 M	R 2 C
ED symptoms	Time	−7.82	1.50	180.50	−5.23	< 0.001***	−5.04	0.50	0.64
	Time2	10.36	1.48	177.11	7.01	< 0.001***	5.41
	CPTSD	0.08	0.19	75.04	0.41	0.729
	CPTSD $\times$ Time	−2.91	2.53	184.22	−1.15	0.354
	CPTSD $\times$ Time2	−3.53	2.48	182.73	−1.43	0.276
ED‐HRQoL impairment	Time	−77.91	13.51	176.79	−5.77	< 0.001***	−5.48	0.42	0.60
	Time2	60.07	13.37	173.03	4.49	< 0.001***	3.67
	Time3	24.24	13.37	173.86	1.81	0.125
	CPTSD	0.85	1.76	75.83	0.48	0.738
	CPTSD $\times$ Time	−28.98	22.87	180.21	−1.27	0.290
	CPTSD $\times$ Time2	−44.28	22.34	178.32	−1.98	0.114	−2.74
	CPTSD $\times$ Time3	5.50	21.72	173.34	0.25	0.800
Anxiety	Time	−23.52	5.97	179.41	−3.94	< 0.001***	−3.21	0.57	0.66
	Time2	9.53	5.93	174.75	1.61	0.256
	Time3	7.04	5.93	176.12	1.19	0.413
	CPTSD	0.21	0.70	74.89	0.30	0.777
	CPTSD $\times$ Time	−2.86	10.10	183.65	−0.28	0.777
	CPTSD $\times$ Time2	8.81	9.88	181.12	0.89	0.524
	CPTSD $\times$ Time3	16.15	9.63	175.45	1.68	0.256
Depression	Time	−28.86	6.49	178.41	−4.44	< 0.001***	−4.00	0.56	0.67
	Time2	32.57	6.45	173.91	5.05	< 0.001***	3.75
	Time3	4.32	6.45	174.45	0.67	0.504
	CPTSD	1.25	0.81	75.20	1.54	0.226
	CPTSD $\times$ Time	−13.28	10.99	182.04	−1.21	0.313
	CPTSD $\times$ Time2	−11.94	10.75	179.79	−1.11	0.313
	CPTSD $\times$ Time3	25.27	10.47	174.21	2.41	0.039*	2.95
Functional disability	Time	−38.07	9.29	178.34	−4.10	< 0.001***	−3.27	0.54	0.61
	CPTSD	0.60	1.04	66.66	0.57	0.568
	CPTSD $\times$ Time	−26.47	15.45	180.70	−1.71	0.132
BMI a	Time	11.98	2.35	195.12	5.10	< 0.001***	4.02	0.91	0.91
	Time2	−20.57	2.11	185.92	−9.74	< 0.001***	−3.32
	Time3	−13.05	1.81	166.90	−7.09	< 0.001***	−2.71
	CPTSD	0.33	0.20	89.91	1.68	0.137
	CPTSD $\times$ Time	6.48	3.98	198.47	1.63	0.137
	CPTSD $\times$ Time2	4.77	3.53	188.15	1.35	0.178
	CPTSD $\times$ Time3	4.74	3.02	171.22	1.57	0.137

Note: Estimates are unstandardized. Second‐and third‐order polynomials are only included where their addition improved model fit. Models control for baseline levels of the treatment outcome.

Abbreviations: BMI, body mass index; d, Cohen's d effect size (0.2 = small, 0.5 = medium, 0.8 = large); df, degrees of freedom; ED, eating disorder; ED‐HRQoL, eating disorder‐specific health‐related quality of life; R ² _M , marginal R ² ; R ² _M , conditional R ² ; SE, standard error.

^a For participants with weight restoration as treatment goal (n = 77).

^* p < 0.05.

^*** p < 0.001.

TABLE 3.

Mixed effects models of time and threshold PTSD symptoms on clinical outcomes.

		Estimate	SE	df	t	p	d	R 2 M	R 2 C
ED symptoms	Time	−7.37	1.51	180.78	−4.88	< 0.001***	−4.79	0.50	0.64
	Time2	10.17	1.48	177.44	6.85	< 0.001***	5.31
	PTSD‐t	−0.02	0.19	73.93	−0.09	0.925
	PTSD‐t $\times$ Time	−3.97	2.51	183.16	−1.57	0.192
	PTSD‐t $\times$ Time2	−2.94	2.46	181.74	−1.20	0.291
ED‐HRQoL impairment	Time	−74.59	13.59	176.91	−5.49	< 001***	−5.29	0.43	0.60
	Time2	61.84	13.39	173.33	4.62	< 0.001***	3.78
	Time3	23.52	13.32	173.73	1.77	0.139
	PTSD‐t	0.65	1.77	74.66	0.37	0.718
	PTSD‐t $\times$ Time	−36.48	22.54	178.92	−1.62	0.150
	PTSD‐t $\times$ Time2	−48.25	22.09	177.35	−2.18	0.071
	PTSD‐t $\times$ Time3	7.80	21.60	173.13	0.36	0.718
Anxiety	Time	−23.11	6.05	180.49	−3.82	0.001**	−3.22	0.57	0.65
	Time2	12.40	5.99	175.69	2.07	0.139
	Time3	7.69	5.96	176.33	1.29	0.348
	PTSD‐t	0.43	0.69	74.30	0.62	0.751
	PTSD‐t $\times$ Time	−4.63	10.03	183.05	−0.46	0.753
	PTSD‐t $\times$ Time2	0.90	9.86	180.55	0.09	0.927
	PTSD‐t $\times$ Time3	14.49	9.66	175.60	1.50	0.315
Depression	Time	−27.87	6.54	178.99	−4.26	< 0.001***	−3.89	0.56	0.67
	Time2	33.03	6.47	174.41	5.11	< 0.001***	3.80
	Time3	4.19	6.44	174.37	0.65	0.516
	PTSD‐t	1.32	0.82	74.18	1.60	0.201
	PTSD‐t $\times$ Time	−15.81	10.85	181.05	−1.46	0.206
	PTSD‐t $\times$ Time2	−13.19	10.66	178.87	−1.24	0.254
	PTSD‐t $\times$ Time3	25.58	10.43	174.10	2.45	0.036*	2.99
Functional disability	Time	−36.72	9.31	175.63	−3.94	< 0.001***	−3.39	0.55	0.63
	Time2	16.55	9.22	170.14	1.79	0.174
	Time3	−9.94	9.17	170.59	−1.08	0.327
	PTSD‐t	1.52	1.04	66.38	1.47	0.206
	PTSD‐t $\times$ Time	−22.96	15.43	178.39	−1.49	0.206
	PTSD‐t $\times$ Time2	−11.17	15.17	175.51	−0.74	0.463
	PTSD‐t $\times$ Time3	38.44	14.87	169.76	2.59	0.037*	3.33
BMI a	Time	11.94	2.34	195.84	5.10	< 0.001***	4.00	0.91	0.91
	Time2	−20.48	2.10	186.58	−9.75	< 0.001***	−3.31
	Time3	−12.95	1.82	167.13	−7.10	< 0.001***	−2.69
	PTSD‐t	0.35	0.20	89.12	1.76	0.143
	PTSD‐t $\times$ Time	6.54	3.99	198.15	1.64	0.144
	PTSD‐t $\times$ Time2	4.50	3.55	187.79	1.27	0.206
	PTSD‐t $\times$ Time3	4.56	3.04	171.29	1.50	0.158

Note: Estimates are unstandardized. Second‐and third‐order polynomials are only included where their addition improved model fit. Models control for baseline levels of the treatment outcome.

Abbreviations: BMI, body mass index; d, Cohen's d effect size (0.2 = small, 0.5 = medium, 0.8 = large); df, degrees of freedom; ED, eating disorder; ED‐HRQoL, eating disorder‐specific health‐related quality of life; PTSD‐t, threshold post‐traumatic stress disorder symptoms; R ² _M , conditional R ² ; R ² _M , marginal R ² ; SE, standard error.

^a For participants with weight restoration as a treatment goal (n = 77).

^* p < 0.05.

^** p < 0.01.

^*** p < 0.001.

TABLE 4.

Mixed effects models of time and threshold DSO symptoms on clinical outcomes.

		Estimate	SE	df	t	p	d	R 2 M	R 2 C
ED symptoms	Time	−9.73	2.71	184.25	−3.59	0.001**	−5.83	0.50	0.64
	Time2	8.10	2.77	178.42	2.93	0.001**	4.23
	DSO‐t	0.15	0.24	72.64	0.62	0.687
	DSO‐t $\times$ Time	1.23	3.03	183.72	0.40	0.687
	DSO‐t $\times$ Time2	1.36	3.07	178.75	0.44	0.687
ED‐HRQoL impairment	Time	−73.30	24.57	180.74	−2.98	0.023*	−4.86	0.42	0.59
	Time2	27.22	25.03	174.42	1.09	0.597
	Time3	38.44	25.86	175.06	1.49	0.486
	DSO‐t	−1.86	2.26	71.91	−0.82	0.597
	DSO‐t $\times$ Time	−17.41	27.46	180.47	−0.63	0.597
	DSO‐t $\times$ Time2	20.80	27.76	174.67	0.75	0.597
	DSO‐t $\times$ Time3	−15.01	28.37	174.75	−0.53	0.597
Anxiety	Time	−16.06	10.54	184.38	−1.52	0.251		0.59	0.66
	Time2	−13.89	10.82	177.04	−1.28	0.201
	Time3	16.40	11.17	178.80	1.47	0.251
	DSO‐t	−2.08	0.85	76.00	−2.45	0.058
	DSO‐t $\times$ Time	−10.99	11.80	184.07	−0.93	0.412
	DSO‐t $\times$ Time2	32.06	11.99	177.68	2.67	0.049*	4.07
	DSO‐t $\times$ Time3	−4.34	12.26	178.41	−0.35	0.723
Depression	Time	−17.99	11.91	182.22	−1.51	0.328		0.54	0.65
	Time2	30.79	12.18	175.34	2.53	0.086
	Time3	11.44	12.59	176.20	0.91	0.639
	DSO‐t	0.21	1.04	71.39	0.20	0.848
	DSO‐t $\times$ Time	−19.74	13.33	181.78	−1.48	0.328
	DSO‐t $\times$ Time2	−3.13	13.51	175.79	−0.23	0.848
	DSO‐t $\times$ Time3	2.66	13.81	176.07	0.19	0.848
Functional disability	Time	−37.88	16.71	182.47	−2.25	0.076		0.53	0.61
	DSO‐t	0.37	1.28	64.90	0.29	0.776
	DSO‐t $\times$ Time	−12.37	18.78	181.65	−0.66	0.766
BMI †	Time	12.80	4.25	191.73	3.01	0.010*	3.49	0.91	0.91
	Time2	−14.21	3.89	185.20	−3.66	0.002**	−2.38
	Time3	−6.30	3.47	165.43	−1.81	0.167
	DSO‐t	0.32	0.25	82.64	1.31	0.228
	DSO‐t $\times$ Time	1.50	4.74	192.82	0.32	0.751
	DSO‐t $\times$ Time2	−5.81	4.31	185.49	−1.35	0.228
	DSO‐t $\times$ Time3	−6.16	3.82	166.22	−1.61	0.189

Note: Second‐ and third‐order polynomials are only included where their addition improved model fit. Models control for baseline levels of the treatment outcome. Estimates are unstandardized.

Abbreviations: BMI, body mass index; d, Cohen's d effect size (0.2 = small, 0.5 = medium, 0.8 = large); df, degrees of freedom; DSO‐t, threshold disturbances in self‐organization symptoms; ED, eating disorder; ED‐HRQoL, eating disorder‐specific health‐related quality of life; R ² _M , marginal R ² ; R ² _M , conditional R ² ; SE, standard error.

^† For participants with weight restoration as the treatment goal (n = 77).

^* p < 0.05.

^** p < 0.01.

FIGURE 1.

Moderating effect of threshold DSO symptoms and CPTSD at baseline on anxiety and depressive symptoms. CPTSD, complex post‐traumatic stress disorder; DSO, disturbances in self‐organization; DSO‐t, above DSO symptom threshold. Time 1, admission; 2, week 4 of treatment; 3, discharge; 4, 6‐months post‐discharge. Error bars represent standard error.

As individuals with threshold DSO symptoms at baseline also necessarily include participants meeting full criteria for CPTSD (having both threshold DSO and PTSD symptoms), post hoc exploratory analyses were conducted to try to parse out the effects of DSO symptoms and complex trauma. The models with threshold DSO as a predictor of outcomes were repeated, controlling for whether participants exceeded the PTSD symptom threshold at baseline. The same interaction effect was observed as the model that did not control for baseline threshold PTSD symptoms. See (Table S3) for full output and interpretation.

Individuals with threshold PTSD symptoms showed more pronounced cubic time trends for depression and functional disability. Threshold PTSD symptoms were associated with more gradual improvement in depression during the first 4 weeks of treatment, steeper improvement in the latter stage of treatment, and similar recurrence at post‐discharge (see Figure 2a). Participants with threshold PTSD symptoms similarly showed less pronounced improvement in functional disability early in treatment, steeper improvement later in treatment, and steeper recurrence at follow‐up (Figure 2b). All main and moderating effects of trauma status accounted for baseline differences in the severity of the outcome variable between those who were positive versus negative for these trauma‐related symptoms.

FIGURE 2.

Moderating effect of threshold PTSD symptoms at baseline. PTSD, post‐traumatic stress disorder; PTSD‐t, above threshold for PTSD symptoms at baseline. Time 1, admission; 2, week 4 of treatment; 3, discharge; 4, 6‐months post‐discharge. Error bars represent standard error.

Effect sizes for significant time terms and interactions with trauma status were large. The variance accounted for in the treatment outcomes by time, trauma status, their interaction, and baseline severity of the outcome variable ranged from 42% to 91%.

3. Discussion

This study investigated the impact of PTSD and CPTSD on outcomes from an ED residential service. ED psychopathology was more severe at admission for individuals with probable CPTSD and those above the clinical cut‐off on the PTSD symptom domain, but not the DSO domain. Psychiatric comorbidity and functional impairment were more severe at baseline and early in treatment for individuals with CPTSD, as well as those above the threshold on both PTSD and DSO symptom domains. This aligns with previous cross‐sectional research on the association between PTSD and greater ED symptom severity (Brewerton et al. 2020; Scharff et al. 2019), while adding to limited research on associations with CPTSD specifically (Day et al. 2024a, 2024b). However, in contrast to some previous findings (Brewerton et al. 2023), differences in symptom severity by comorbid trauma resolved by end‐of‐treatment and remained non‐significant at the 6‐month follow‐up. That is, although our study corroborated previous research that individuals with trauma‐related symptoms typically represent a more severe subgroup of the ED population, this difference was not equivalent across all symptom domains and clinical outcomes. Moreover, significant differences based on trauma comorbidity dissipated by discharge, suggesting good responsiveness to treatment despite greater psychopathology and impairment at admission. Individuals with comorbid trauma were also not more likely to discontinue treatment early.

Similarly, trauma‐related symptoms did not moderate the pattern of change in ED symptoms over treatment. These findings align with the emerging trend seen in some other residential treatment studies (Brewerton et al. 2023; Mitchell et al. 2021). However, previous studies of PTSD as a moderator of residential treatment outcomes featured services that use integrated therapeutic approaches including the Unified Treatment Model (a modular treatment focused on improving emotional awareness, avoidance, and regulation; Mitchell et al. 2021; Scharff et al. 2021) and ED‐focused therapy combined with principles from CPT (a PTSD‐focused psychotherapy) (Brewerton et al. 2023). The residential service in the current study did not use an integrative or combined trauma and ED treatment approach, or use specific trauma‐focused therapies such as cognitive processing therapy (CPT), trauma‐focused cognitive behavioral therapy (TF‐CBT), prolonged exposure, or eye movement and desensitization therapy (EMDR). As such, it is notable that comorbid trauma did not moderate the trajectory of ED symptom change, despite baseline differences in severity. This lack of moderating effect for ED symptoms contrasts with previous research pointing to detrimental effects of trauma on ED treatment outcomes (Convertino and Mendoza 2023; Day et al. 2023). It is unclear whether there is something particular to the residential treatment environment that may be protective against the broader negative effect of trauma in other settings such as inpatient hospitalization and outpatient care. Recent qualitative findings suggest that individuals find the home‐like nature of the residential treatment environment important in fostering a sense of safety (Rankin et al. 2025), which is likely to be particularly valuable for those who have experienced trauma.

Although trauma‐related symptoms did not moderate outcome trajectories for ED symptoms or BMI, some moderation effects were observed for secondary psychological and functional outcomes. Controlling for baseline outcome severity, for individuals with CPTSD or threshold DSO or PTSD symptoms at admission, a trend emerged of less change in some clinical outcomes (depression, anxiety, functional disability) during the first 4 weeks of treatment, followed by steeper improvement later in treatment, and similar or slightly greater symptom resurgence at discharge. This may suggest that individuals with trauma‐related symptoms may be slower to engage during residential treatment, possibly due to greater hypervigilance after entering a new environment or the theorized challenges trauma‐related symptoms may pose to the development of the therapeutic alliance (Olofsson et al. 2025); however, after the initial period of treatment, this group may show more rapid improvement. PTSD and DSO symptom domains showed some differences in how they moderated outcomes; DSO moderated the symptom trajectory for anxiety, while threshold PTSD symptoms moderated change in depression and functional disability.

3.1. Strengths, Limitations, and Future Directions

This has been the first study to investigate the effects of ICD‐11 CPTSD and its separate symptom domains on ED treatment outcomes. By differentiating between PTSD and DSO domains, our findings provide a more detailed picture of how trauma‐related symptoms affect residential care outcomes, compared with previous studies that have only assessed DSM‐IV/DSM‐5 PTSD. However, the number of participants in our sample who met criteria for ICD‐11 PTSD (which requires that the DSO symptom threshold is not met) was too small to include PTSD diagnosis as a predictor of clinical outcomes, and as such this should be assessed in future studies with larger samples. Similarly, the sample was not large enough to further parse out the effects of threshold DSO symptoms by excluding individuals with threshold PTSD symptoms. As such, models using threshold DSO symptoms at baseline were inclusive of participants who met full criteria for probable CPTSD; however, additional exploratory models controlling for baseline threshold PTSD symptoms did not find any change in findings. Future studies with larger samples should assess threshold DSO symptoms separately from cases of CPTSD to determine whether any of the effects found may be indicative of the importance of general emotion dysregulation rather than complex trauma specifically.

We were not able to conduct a full assessment of participants' lifetime exposure to traumatic events to validate whether the type/s of trauma exposure reflected the events included in the ICD‐11 for PTSD and CPTSD. This should be considered in future studies. Our sample size was insufficient to assess for variations in trauma‐related outcomes and moderation effects by ED diagnosis; these subgroup analyses should be included in future research given that the likelihood of participants presenting with clinical trauma‐related symptoms did appear to differ by ED presentation in the current study. As most participants in the current study had AN, this affects the generalizability of the findings to other ED presentations. We had limited data available on the type of care accessed by participants post‐discharge, which may have contributed to the outcome trajectories seen at follow‐up. As with many treatment studies, outcomes seen at discharge and follow‐up may be limited by attrition bias, such that individuals with less improvement in outcomes may have been more likely to leave treatment prematurely without completing post‐treatment measures (however, attrition was not associated with trauma status).

Other directions for future research include assessing how PTSD and DSO symptoms affect specific ED behaviors rather than just global eating psychopathology. ED attitudinal subscales and behavioral frequency items were not included in the current study due to the relatively small sample size and greater risk of Type I error with multiple comparisons. Our finding that global ED symptoms did not vary at any timepoint based on threshold DSO symptoms was unexpected and could reflect the very high rate of these symptoms (81.1%) in the sample. Further research is needed to determine whether this finding is a function of insufficient statistical power (for the non‐DSO threshold subgroup) or cohort‐specific effects.

Finally, the current study did not account for factors associated with prior exposure to traumatic events. Although a previous study found that the number of different types of adverse childhood experiences did not correlate with DSO symptom severity or baseline ED symptoms in ED treatment seekers (Day et al. 2024a, 2024b), other studies have found detrimental effects of exposure to childhood and sexual abuse on ED treatment outcomes (Convertino and Mendoza 2023; Day et al. 2023). As such, the type or perceived severity—but not necessarily the number of different kinds—of trauma exposure may play a role in whether trauma‐related symptoms go on to affect ED treatment outcomes. We also did not examine change in trauma‐related symptoms themselves in this paper; however, this research question will be addressed in a future paper.

3.2. Implications

Individuals receiving ED residential care may benefit from assessment for comorbid trauma. Validated self‐report tools such as the International Trauma Questionnaire (Cloitre et al. 2018) are quick to administer and score, freely available, and can differentiate between PTSD and CPTSD diagnoses as well as specific symptom subscales on PTSD and DSO domains. Although differences in symptom severity resolved by discharge, some symptoms were found to be slower to improve early in treatment for participants with CPTSD or who scored above the clinical threshold on PTSD and DSO domains. In services or personal circumstances limited to a shorter treatment duration, such individuals could benefit from additional support early in treatment. The average treatment duration of 81 days in the current study may not be feasible for all individuals based on cost (in private services) or interference with work, study, or caregiving responsibilities. Improving the efficiency of eating disorder residential treatment for people with comorbid trauma may help to reduce treatment burden and waiting times to access this resource‐intensive treatment setting.

Tailoring of care early in treatment could include greater focus in individual therapy sessions on evidence‐based strategies for improving anxiety and low mood, alongside ED‐focused therapies, for those with comorbid trauma‐related symptoms. Alternatively, trauma‐related symptoms could be addressed directly using trauma‐focused therapies such as cognitive processing therapy for individuals with predominantly PTSD symptoms (Martin et al. 2021; Trottier and Monson 2021) and dialectical behavior therapy for high DSO symptoms (Bohus 2022). As individuals with comorbid trauma also sometimes showed greater post‐discharge resurgence in anxiety, depression, and functional impairment, this risk could be mitigated by prioritizing more intensive step‐down care options (e.g., a day program rather than weekly outpatient sessions) and close symptom monitoring.

4. Conclusion

There is growing recognition that examining EDs in isolation fails to account for their frequent comorbidities such as trauma‐related disorders (Brewerton 2023; Haynos et al. 2024; Schmidt et al. 2025). The current study illustrates that PTSD and CPTSD symptoms are commonplace in residential care and are often a marker of greater psychopathology at baseline, but do not necessarily predict poorer treatment response, particularly for ED outcomes. Elevated trauma‐related symptoms on both PTSD and DSO domains may be associated with slower responsiveness early in treatment for anxiety, depression, and functional impairment, but also steeper improvement beyond 4 weeks' of care. Although individuals with comorbid trauma did not experience overall poorer treatment gains, these individuals were at risk of greater symptom resurgence in some secondary outcomes after discharge. Further research is needed to establish evidence‐based guidelines for the tailoring of treatment for individuals with comorbid EDs and trauma‐related symptoms in higher levels of care.

Author Contributions

Sinead Day: conceptualization, data curation, formal analysis, visualization, writing – original draft, methodology, investigation, project administration, writing – review and editing, software, resources. Deborah Mitchison: conceptualization, supervision, project administration, writing – review and editing, funding acquisition, methodology. W. Kathy Tannous: conceptualization, methodology, supervision, writing – review and editing, funding acquisition. Phillipa Hay: conceptualization, methodology, supervision, project administration, writing – review and editing, funding acquisition.

Ethics Statement

Ethics approval for this study was granted by the human research ethics committee at Western Sydney University (H14478, H14742). The authors acknowledge that all participants provided informed consent for the use of their collated, deidentified data in this study.

Conflicts of Interest

Phillipa Hay has received sessional fees from the Therapeutic Guidelines publication and the Health Education and Training Institute (HETI, NSW) and royalties/honoraria from Hogrefe and Huber, McGraw Hill Education, Blackwell Scientific Publications, BioMed Central, and PLOS Medicine. She has prepared a report under contract for Takeda (formerly Shire) Pharmaceuticals regarding binge eating disorder (July 2017) and has been a consultant to Takeda Pharmaceuticals. She is currently a consultant to Tryptamine Therapeutics. She was a member of the ICD‐11 Working Group for Eating Disorders (2012–2019). All other authors have no conflicts of interest to declare.

Supporting information

Table S1. Average scores on potential covariates by baseline trauma status and data missingness.

Table S2. Association between potential covariates and baseline outcomes.

Table S3. Mixed effects models of time and threshold DSO symptoms on clinical outcomes, controlling for threshold PTSD symptoms at baseline.

Day, S. , Mitchison D., Tannous W. K., and Hay P.. 2025. “ PTSD and Complex PTSD in Residential Treatment for Eating Disorders: Moderating Effects on Symptom Severity and Outcome Trajectory.” International Journal of Eating Disorders 58, no. 9: 1732–1746. 10.1002/eat.24465.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.