Trauma‐focused guided self‐help interventions for posttraumatic stress disorder: A meta‐analysis of randomized controlled trials

Abstract Trauma‐focused guided self‐help (TF‐GSH) is an important alternative to psychological therapy delivered by a therapist. This meta‐analysis evaluates the efficacy of TF‐GSH in reducing posttraumatic stress disorder (PTSD) symptoms and comorbid depressive and anxiety symptoms. A total of 17 trials were included that compared a TF‐GSH intervention (N = 610) to various control comparators (N = 570). Control conditions included treatment as usual (k = 2), waiting list (k = 11), phone monitoring (k = 1), nontrauma writing (k = 1), general support (k = 1), and supportive counseling (k = 1). A moderate‐ to large‐sized effect favouring TF‐GSH was observed for PTSD (k = 17, g = −0.81, 95% confidence interval [CI]: −1.24, −0.39) and a moderate‐sized effect was observed for depressive (k = 13, g = −0.73, 95% CI: −1.16, −0.31) and anxiety (k = 11, g = −0.72, 95% CI: −1.18, −0.27) symptoms, with considerable heterogeneity. Moderator analyses were all not statistically significant. Results indicate that TF‐GSH is a promising treatment for PTSD and comorbid depressive and anxiety symptoms. We discuss the nature, extent, and quality of the literature to provide a point of departure for future research. TF‐GSH (and unguided self‐help) may not be appropriate for certain individuals at certain times. Exploring a broad range of treatment delivery modalities will move the field closer towards a model of evidence‐based care in which the likely appropriate dose and type of intervention can be matched to individuals based on presenting problems and other variables.


| INTRODUCTION
Trauma exposure is near ubiquitous and posttraumatic stress disorder (PTSD) is a common psychological problem following trauma exposure . A large body of research has tested the effectiveness and efficacy of pharmacological, psychological, and other treatments for PTSD. Meta-analytic reviews indicate that some medications (fluoxetine, paroxetine, sertraline, venlafaxine and quetiapine) lead to a small reduction in PTSD symptom severity, (Hoskins et al., 2015(Hoskins et al., , 2021 whereas a number of different traumafocused psychological therapies have demonstrated large effect sizes in reducing PTSD (Mavranezouli et al., 2020a(Mavranezouli et al., , 2020b. Although clinical guidelines for PTSD widely recommend traumafocused psychological therapy as the first-line treatment, a number of barriers limit access to specialty mental health care in general and trauma-focused psychological therapy in particular, including lack of confidence in treatment effectiveness; fear of increasing PTSD symptoms; perceived stigma associated with psychological therapy; practical barriers (e.g., transportation, limited treatment availability, especially in low-and middle-income countries); and long waiting times Smith et al., 2020). Furthermore, not everyone benefits from trauma-focused psychological therapy (around two-thirds of individuals respond adequately (Bryant, 2019); therapy requires significant therapist input, is time consuming and costly (Hedman et al., 2011) to deliver, and places a high emotional demand on therapists; and many clinicians do not feel competent to deliver trauma-focused psychological therapy (Finch et al., 2020).
In response to these barriers and the widespread need for treatment, there has been a growing interest in exploring alternative means of delivering trauma-focused psychological therapies, including via supported and unsupported self-help. Self-help for PTSD that is delivered with active support and monitoring from a trained professional is known as trauma-focused guided self-help (TF-GSH). TF-GSH is a self-administered intervention based on trauma-focused cognitive behavioral therapy in which individuals are guided through written or electronic materials via face-to-face, email, internet, or phone call support. TF-GSH is likely to be cheaper than psychological therapy that is delivered face-to-face by a therapist; requires less staff time, skill, training, and emotional involvement; and may be more accessible, convenient (e.g., fitting around work or school commitments), and appealing for some people, since it does not involve traveling to appointments in formal treatment settings.
A burgeoning literature has tested the efficacy of TF-GSH for PTSD. The present meta-analysis synthesizes the available randomized controlled trials on this topic with a view to obtaining an accurate estimate of the efficacy of TF-GSH and providing a point of departure for future research in this area. Most of the trials included in the review also measured the impact of TF-GSH for PTSD on depressive and anxiety symptoms because PTSD is often comorbid with other mental health problems (Horesh et al., 2017;O'Donnell et al., 2004). We, therefore, also present meta-analyses of the efficacy of TF-GSH for PTSD on depressive and anxiety symptoms.

| METHODS
The meta-analysis was conducted in accordance with best-practice guidelines for conducting systematic reviews (Siddaway et al., 2019) and the Preferred Reporting Items for Systematic Reviews and Metaanalyses standards (Moher, 2009). The systematic review protocol was preregistered with PROSPERO (CRD42015026026).

| Inclusion and exclusion criteria
Randomized controlled trials that employed TF-GSH and measured PTSD pre-and postintervention were potentially eligible for inclusion. To be included, TF-GSH interventions were required to be delivered by a trained clinician on an individual (rather than group) basis, and to be trauma-focused (i.e., incorporating an element of processing trauma memories and working with beliefs regarding the trauma[s]). To provide a comprehensive summary of the available evidence, explore heterogeneity, and increase the generalizability of the findings, no restrictions were applied for TF-GSH medium (e.g., telephone, face to face), publication status, language, age group (children and adults), setting, or comparator group.
Studies were considered for inclusion where PTSD was the primary presenting difficulty. Data from studies that also measured changes in comorbid depressive and anxiety symptoms were included in additional meta-analyses. PTSD status can potentially be determined by a qualified clinician's diagnosis, a standardized diagnostic interview, or a continuous self-report measure of PTSD symptoms.
Diagnostic and continuous approaches to measuring PTSD are both common in the literature and individuals with elevated PTSD symptoms experience significant functional impairment and are often referred for mental health services (Cohen, 2013). If a study included more than one PTSD outcome measure, a primary measure was selected based upon superiority of psychometric properties (i.e., published reliability and validity). If alternative measures had equivalent properties, we extracted data from clinician-rated over self-rated measures and/or the measure most frequently used in other included studies to attempt to reduce this potential source of heterogeneity. Total scores were used to calculate treatment effects where studies include both subscale and total scores. During the data extraction phase, to incorporate all relevant evidence within the review, a decision was made to derestrict the eligibility criteria to include studies where at least 70% of the sample reached clinical levels of PTSD symptoms as defined by a standardized PTSD outcome measure. This approach is consistent with best practice guidelines for conducting systematic reviews (Siddaway et al., 2019).
No restrictions were placed on the type of trauma, the amount of time since the traumatic event, the chronicity of PTSD, or comorbid mental health problems. Studies were excluded if they were sampled from specific groups that would likely significantly affect the effectiveness of TF-GSH or preclude suitability for psychological intervention (eg, personality disorder, neurodevelopmental disorder, learning disability, severe depression, or substance dependency). All identified studies were exported to EndNote X9 for Windows, where duplicates were removed. To determine study eligibility, all titles and abstracts were screened independently by two researchers (C.L. and A.P.S.), who also conducted the second full-text screening independently. Disagreements or uncertainties were discussed with the senior researcher supervising the project (R.M-S.).

| Data analysis
Cohen's d was computed to represent the between-group treatment effect for each study by subtracting the mean postintervention score of the control group from the mean postintervention score of the experimental group and dividing the result by the pooled standard deviation. Cohen's d was transformed into Hedge's g to reduce the bias inherent in d when N is small. An effect size of 0.8 was considered large, 0.5 moderate, and 0.2 small (Cohen, 2013). Study authors were contacted for additional information when an effect size or data to compute an effect size were not reported.
Separate random effects meta-analyses were conducted for PTSD, depression, and anxiety problem effect sizes using R Version 4.1, 2021 (R Core Team, 2014). Heterogeneity was assessed using the Q statistic, τ 2 , and I 2 . τ 2 estimates the amount of total heterogeneity and is measured on the same scale as the effect size itself (in this meta-analysis: g). I 2 is the percentage of the total variability that is due to true (i.e., between-study) heterogeneity rather than sampling error. Percentages of around 25% (I 2 = 25), 50% (I 2 = 50), and 75% (I 2 = 75) indicate low, medium, and high heterogeneity, respectively (Higgins & Thompson, 2002). Subsequently, several substantive and methodological moderators were examined in relation to PTSD effect sizes. Small ks and missing values (see Tables 1 and 2)
The risk of bias for each domain was scored as low (0), high (2), or unclear (1). The risk of bias assessment was performed by C. L and J.
F. with R.M-S consulted in case of uncertainty.
Three strategies were used to assess publication bias. First, a funnel plot was created to visually search for evidence of bias, which would be apparent in an asymmetrical plot. Next, asymmetry was assessed using Egger's weighted regression test (Egger et al., 1997) and Duval and Tweedie's trim and fill method (Duval & Tweedie, 2000).

T A B L E 2 Characteristics of included studies: Treatment modality and duration
Author (

| Moderator analyses and risk of bias
Several outliers (Latif et al., 2021;Sloan et al., 2012) were identified (see Figure 2). See Figure S2 for funnel plots of effect sizes for PTSD, depressive symptoms, and anxiety symptoms with these outliers  Finally, as studies varied widely in how they reported duration of TF-GSH, we examined whether excluding two studies (Knaevelsrud & Maercker, 2009;Knaevelsrud et al., 2015) that delivered the Interapy intervention, which involved a high degree of therapist input, A statistically significant, moderate to large-sized (Cohen, 2013) effect favouring TF-GSH was observed for PTSD and a statistically significant, moderate-sized (Cohen, 2013) effect was observed for depressive and anxiety symptoms, providing evidence of the potential value of TF-GSH for PTSD and comorbid mental health problems. There was no evidence of publication bias, suggesting that the meta-analytic results are unlikely to have been artificially inflated by file-drawer effects. There was also no evidence that methodological rigour affected the strength of effects, furthering underpinning the robustness of the findings. Treatment benefits were demonstrated across severities of PTSD, suggesting that the effects documented here are not limited to "mild" PTSD presentations. That an intervention involving relatively little clinician input can achieve substantial change is consistent with the well-established finding that most people, irrespective of trauma type and exposure, are able to find a way to process and adjust to potentially traumatic experiences without the need for professional involvement (Ehlers & Clark, 2000).
The effect sizes observed here for TF-GSH are similar in magnitude to those observed for internet-delivered cognitive behavioral therapy (iCBT) for PTSD (Kuester et al., 2016;Sijbrandij et al., 2016) and telehealth interventions for PTSD (Sloan et al., 2011).
In contrast to the current review, the meta-analyses of iCBT and telehealth interventions included relatively few studies that used a clinical sample. The present review only included one study that would not be classed as an iCBT and hence might be considered an update or extension of the meta-analyses of iCBT, albeit restricted to iCBT interventions where therapist support was available. While iCBT may have advantages for many populations in terms of accessibility, there is a paucity of research on therapy delivered through low technology means (e.g., printed materials). In some contexts and for some people (e.g., those who are digitally excluded), it may be that simpler, noninternet-delivered TF-GSH interventions are more accessible, easily implemented, and/or disseminated, and potentially cheaper.
A considerable degree of heterogeneity was observed, especially among PTSD effect sizes, potentially suggesting that caution should be applied when interpreting the summary effects, (Deeks et al., 2011) One outlier (demonstrating a much stronger effect of TF-GSH) was identified in each meta-analysis (Latif et al., 2021;Sloan et al., 2012).
Removing the three identified outliers somewhat reduced the strength of associations, substantially reduced the degree of heterogeneity among PTSD effect sizes, and almost eliminated heterogeneity among depressive symptom and anxiety symptom effect sizes. The results with outliers removed are probably the most accurate estimate of the effect of TF-GSH. The small number of included studies means that outliers are not clearly attributable to systematically different methodological features or biases when compared to the other studies included in the meta-analysis and potential reasons for the three outlying effect sizes can only be very tentatively speculated upon (e.g., use of written material [Sloan et al., 2012], study conducted in Pakistan [Latif et al., 2021]).
Moderator analyses were conducted in an attempt to explain the observed heterogeneity; all were not statistically significant. To inform future research, we reported that TF-GSH appeared to be much less effective (but not statistically significantly different) for combat versus noncombat groups and when compared to more active control conditions versus waiting list control conditions. Given the potential for low statistical power because of small numbers of effect sizes and the small sample sizes of included trials, (Hunter & Schmidt, 2004) we note that failure to obtain a statistically significant difference among subgroups was not interpreted as evidence that effect sizes are the same across subgroups (Borenstein et al., 2009).
The average dropout rate from TF-GSH for the studies included in this review is 36% (based on k = 12). This figure is somewhat higher than that found in therapist-administered individual trauma-focused psychological therapies, although we note that a moderator analysis found that dropout rate did not appear to influence outcomes (Lewis et al., 2020). Some of the included studies outlined reasons for dropouts. For example, one study (Lewis et al., 2017) reported that the majority of dropout occurred before participants began TF-GSH, indicating a reluctance to engage in GSH as opposed to an issue related to the tolerability of the treatment itself.
Although the present meta-analytic results suggest that TF-GSH appears to be a promising potential intervention for PTSD, several Further trials are also needed to elucidate whether particular types of TF-GSH are most effective (e.g., comparing written materials with the much more common internet-based materials), and whether TF-GSH is more effective for different populations, trauma types, and individuals meeting criteria for PTSD versus complex PTSD. One possibility raised here which requires further research attention is that TF-GSH may be less effective for individuals who have experienced combat. Although likely contradictions could be speculated upon (e.g., individuals who are experiencing dissociative symptoms or who are at high risk to themselves or others), ultimately, what works for whom and under what circumstances is an empirical question. Our meta-analysis highlighted that the majority of the existing trials used a waiting list or treatment as usual control condition. Adequately powered non-inferiority trials comparing TF-GSH to in-person therapist-led treatment (the standard delivery of individual TF-CBT) in clinical samples, as well as costeffectiveness studies and implementation studies, are much needed to help inform the future planning of services for people with PTSD.

| CONCLUSION
Trials examining the efficacy of GSH interventions for a range of psychological problems are proliferating. The present meta-analysis found that TF-GSH has a moderate to large-sized impact on PTSD and a moderate-sized impact on depressive and anxiety symptoms.
GSH is likely to be cheaper than psychological therapy that is delivered face-to-face to individuals by a therapist; requires less staff time, skill, training, and emotional involvement; and may be more accessible, convenient (e.g., fitting around work or school commitments), and appealing for some people, since it does not involve traveling to appointments in formal treatment settings.
The continued proliferation of research into alternative treatments for PTSD is welcome, as doing so is only likely to increase access to evidence-based interventions and help to continue to refine interventions. Exploring a broad range of treatment delivery modalities will help the field to move closer and closer towards a model of evidence-based care in which the likely appropriate dose and type of intervention can be matched to individuals based on their presenting problems and other variables.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.

PEER REVIEW
The peer review history for this article is available at https://publons. com/publon/10.1002/da.23272