Systematic review and meta-analyses of psychosocial interventions for veterans of the military

Background The efficacy of psychosocial therapies for common mental health disorders in veterans is unclear and requires further examination. Method Systematic review and meta-analyses of randomised controlled trials (RCTs). Twenty databases were searched. Studies were included if they reported a psychosocial intervention designed to treat or reduce common mental health symptoms in veterans identified as being symptomatic at the time they entered the study. Studies of substance dependency disorders and psychosis were excluded. Eligible studies were assessed against methodological quality criteria and data were extracted and analysed. Results Twenty-nine RCTs were identified. There was evidence for the use of trauma-focused therapies for post-traumatic stress disorder (PTSD) and some evidence for psychological interventions in the treatment of borderline personality disorder, depression, insomnia, and panic disorder co-morbid to PTSD. However, methodological quality of many of the studies was less than optimal. Conclusions Trauma-focused psychological therapies are likely to be effective for combat-related PTSD but there is a need for more research to determine the efficacy of psychological treatments for other mental health disorders in veterans.

M ilitary personnel are considered one of the highest risk occupational groups for exposure to traumatic and adverse events (Hoge et al., 2002). Such events and the demands of being deployed away from family and social support increase vulnerability to a range of mental health problems, in particular to various anxiety and depressive disorders (Browne et al., 2007;Iversen, Waterdrinker, et al., 2007;King's Centre for Military Health Research [KCMHR], 2010; Prigerson, Maciejewski, & Rosenheck, 2001).
Engaging veterans (ex-service personnel who have left the military) into mental health treatment programmes remains challenging due to a variety of factors, including stigma, perceived weakness in acknowledging emotional difficulties, and the military macho culture (KCMHR, 2010). This is compounded by limited evidence to determine what treatment programmes are efficacious, resulting in a lack of consensus as to what should be offered (Creamer, Forbes, Biddle, & Elliott, 2002;Richardson, Naifeh, & Elhai, 2007). More than 60% of US Iraq veterans who screened positive for a mental health problem did not seek treatment (Hoge, 2004). Only 23% of serving UK personnel with common mental health problems were receiving medical professional help, mainly in primary care (Iversen et al., 2010).
The Institute of Medicine (IOM) concluded that the evidence is still ''inadequate to address the specific interventions, settings, and lengths of treatment that are applicable in the veteran population'' (Institute of Medicine Committee, 2007). Studies have shown that veterans with complex psychiatric problems are often difficult to treat with less positive outcomes than nonveterans. For example, in a meta-analysis of 26 posttraumatic stress disorder (PTSD) trials, only five included combat veterans and the overall effect size for veterans was significantly lower for other groups (Bradley, Greene, Russ, Dutra, & Westen, 2005;Chemtob, Novaco, Hamada, & Gross, 1997;Foa, Keane, Friedman, & Cohen, 2009;Glynn et al., 1999). There may be unique aspects to conditions such as PTSD in veterans (Institute of Medicine Committee, 2007).
Early dysfunctional relationships and poor attachments have been associated with mental health problems following exposure to combat and may influence poor outcome in treatment LeardMann, Smith, & Ryan, 2010). The degree of exposure to trauma, experience of continuous threat, serving unit ethos, and moral injury may also be factors that influence psychopathology in veterans (Litz et al., 2009). In order to examine the efficacy of psychosocial interventions for veterans in reducing common mental health disorders, a comprehensive systematic review, and meta-analyses of randomised controlled trials (RCTs) was performed to evaluate the efficacy and relative effectiveness of psychosocial treatments.

Method
A systematic bibliographic search was undertaken to locate and retrieve RCTs of psychosocial treatments for common mental health disorders from 20 databases (including EMBASE, Medline, PsycINFO, PILOTS, CINAHL, and the Cochrane Library). The key themes devised for the search strategy included veterans and military, mental health, psychotherapies, reviews and systematic reviews, psychosocial or care pathway. Each of the 20 databases was searched from inception to January 2012. The search was restricted to papers with English language abstracts. Additional published, unpublished, and in-press studies were found by handsearching (reviewing) the references of retrieved articles, previous systematic reviews, and meta-analyses of psychosocial treatments for common mental health disorders (Bisson et al., 2007), the proceedings of meetings of both the European and International Society of Traumatic Stress Studies between 2006 and 2010 and by contacting a number of international experts within the veteran mental health field to attempt to identify unpublished studies.

Selection
Studies were considered if they reported a psychosocial intervention designed to treat or reduce common mental health symptoms, for example, anxiety or depressive disorders in veterans who were identified as being symptomatic at the time they entered the study. The authors excluded studies of substance misuse and psychosis in order to focus on common mental health disorders within veterans. For the purpose of this review, a psychosocial intervention was defined as: any specific non-pharmaceutical intervention aimed at reducing a range of symptoms, offered by one or more health professional or lay person, with contact between therapist and participant on at least one occasion.
It was decided a priori to include all forms of psychosocial therapy. Studies had to be of randomised controlled design, with adult (16 years old) participants who had previously served in the armed forces regardless of gender, age, and country of origin. The studies had to report at least pre-and post-treatment outcomes and retain at least 50% of the original sample at the posttreatment assessment. The first author undertook the searching. Decisions as to whether individual studies met the inclusion criteria were made independently by three authors (NJK, NPR, and DW). Any disagreements were resolved by discussion between the three reviewers. When consensus could not be achieved, advice was sought from the fourth author (JIB).

Study characteristics
An initial narrative synthesis was undertaken to describe the scope (participants, settings, intervention type, comparators, and measures of effect), quality, and outcomes of the studies. Three main efficacy outcomes were considered: retaining a diagnosis of a common mental health disorder, assessor-rated, and self-reported symptom severity via validated self-report instruments. We decided a priori that our primary outcomes would be clinicianrated symptom severity, although in practice this was not present for many studies.

Quantitative data synthesis
Validity assessment rating tool All published and unpublished papers were assessed against the following quality criteria as described by the Cochrane Collaboration: sequence generation, allocation concealment, blinding of assessors, exclusion criteria and drop out, and completeness of outcome data (Higgins & Green, 2009).

Data extraction
Study details including the disorder of interest, participants' characteristics and type of intervention were entered into Review Manager version five (The Cochrane Collaboration, 2008). The quality criteria and accuracy of outcome data were evaluated independently by three reviewers; disagreement was handled by the same method as disagreements about selection. Quantitative data were used to synthesise posttreatment and follow-up data where appropriate. None of the included studies provided dichotomous data for further analysis. Post-treatment data (or change scores if reported instead) for the psychosocial treatment and control condition were entered in Review Manager tables for analysis through standardised mean difference (SMD). When intention to treat (ITT) data was available, this was reported in the results. Attempts were made to access ITT data by contacting the corresponding author for a complete data set. Reasons given by authors for not being able to provide complete data included the study being old and individuals had changed employment or retired. Completer-only analysis was reported when this was the only data source available.

Heterogeneity
A visual inspection of the forest plots was initially used to explore for possible heterogeneity. It was also measured by observing the I 2 test (used to measure of the consistency between trials in a meta-analysis) (Higgins & Green, 2009). An a priori decision was made to use a random-effects model when the I 2 was 30% or greater. However, because of the degree of clinical heterogeneity of studies included, a post hoc decision was taken to use a random effects model in all analyses undertaken.

Results
Psychosocial intervention vs. treatment as usual/ waiting list to treat depression Dobscha et al. (2006) investigated depression decision supportÁcollaborative care vs. usual care and Hedrick et al. (2003) collaborative care vs. consultÁliaison (usual care). Fortney et al. (2007) tested a collaborative care model within remote isolated clinics via telepsychiatry vs. usual care. Oslin et al., (2003) explored telephone-based disease management in primary care for veterans with depression or at risk drinking vs. usual care. Ross et al. (2008) tested a telephone-based close monitoring program to manage veterans with minor depression in a primary care setting vs. usual care. Figure 2 shows a summary of the outcome effects. There was no significant difference between the collaborative intervention and control in any of the studies or when initial outcomes were combined in meta-analyses (random effects) (K 05, N01271; SMD (0.12, 95% CI (0.25, 0.00). There were concerns about methodological quality in all these trials except Dobscha et al. (2006), with three showing concerns on three or more of the validity assessment criteria.     This was the only study to include a group anger management treatment via teleconferencing with veterans with PTSD and anger control difficulties and therefore was not included within the review.

Psychosocial interventions aimed at reducing insomnia symptoms
Three studies targeted insomnia (primary and co-morbid) and tested either cognitive behavioural therapy (CBT) or a mindfulness-based intervention against an active control. Edinger and Sampson (2003) used a brief (two sessions) CBT-based intervention vs. a sleep hygiene control. Edinger et al. (2009) provided four sessions of CBT vs. sleep hygiene. Nakamura, Lipschitz, Landward, Kuhn, and West (2010) compared two sessions of weekly mindÁbody bridging vs. an active sleep education control. There was no significant difference between the experimental intervention and control in any of the studies or when initial outcomes were combined in meta-analyses (random effects) (K03, N0152; SMD (0.28, 95% CI (0.61, 0.04), see Fig. 3. There were methodological concerns with all three trials with Edinger and Sampson (2003) and Edinger et al. (2009) failing to meet the standard on at least three of the five validity criteria.
Trauma-focused psychosocial interventions vs. usual care or waiting list aimed at reducing PTSD symptoms Self-report data was available from four studies. Carlson et al. (1998) and Devilly et al. (1998) compared EMDR vs. routine clinical care. Keane et al. (1989) compared flooding group techniques vs. wait list control. Monson et al. (2006) tested cognitive processing therapy vs. wait  list control. Figure 4 shows a summary of the outcome effects. At initial follow-up, a difference in favour of intervention was apparent when the outcomes of these studies were combined in meta-analyses (random effects) (K 04, N0128; SMD(0.59, 95% CI (1.09,(0.10).
Validity assessment showed that there were methodological concerns with each of these studies (see Table 1), in particular with Carlson et al. (1998), Devilly et al. (1998 and Keane et al. (1989). Monson et al. (2006) was more robustly controlled and achieved the largest effect size.

Discussion
To our knowledge, this is the first systematic review of psychosocial interventions for veterans presenting with various mental health problems. Twenty-nine RCTs were identified of which 12 were included in meta-analyses. Study quality was variable with significant concerns about methodology in over half of the identified studies, including many entered into meta-analysis. This needs to be taken into account when evaluating the results. There was some evidence for the efficacy of several different trauma-focused psychological therapies delivered on an individual or group basis to treat chronic PTSD with two well-powered and methodologically robust studies reporting particularly positive findings (Monson et al., 2006;Schnurr et al., 2007). There was a lack of evidence to determine the efficacy of trauma-focused therapy delivered by telepsychiatry and of psychosocial interventions for veterans with insomnia. There is evidence from single RCT's for the efficacy of other interventions: dialectical behaviour therapy for female veterans with borderline personality; telephone disease management for depression and at risk drinking; CBT or CBT plus exercise for Gulf War illness and CBT for panic disorder (see Table 2). There is, therefore, limited evidence specific to military veterans on which to base firm recommendations. We were unable to locate any trials of individual psychosocial interventions such as CBT or interpersonal therapy being evaluated in veteran populations with depression or other anxiety disorders. There are several potential reasons for this that could be usefully explored in future research (Fossey, 2010).
Not surprisingly, there was significant clinical and statistical heterogeneity in the included studies. Seventeen of the trials attempted to reduce traumatic stress symptoms, although the nature of the interventions was diverse. In addition, the total number of hours of the interventions provided across studies varied from two to fifty two hours making results from meta-analyses less meaningful and difficult to generalise from. The results should therefore be interpreted cautiously, although it is noteworthy that a meta-analyses limited to the clinically more homogeneous trauma-focused interventions was positive.
A variety of clinical presentations were included and there were also differences with regards to service history, combat exposure, and time of actual service. These issues may have resulted in differences in the way veterans presented, the duration of their symptoms, and their likelihood to respond to treatment. For example, Korean/ Vietnam veterans may be more likely to present with complex, multi-factorial problems and be more difficult to treat than veterans of more recent conflicts or civilians (Beidel et al., 2011). These factors may explain in part the poor outcomes and increase the risk for potential bias (Monson et al., 2006).
The absence of effect of the treatments for depressed veterans is surprising given the evidence for effective psychosocial therapies within primary care for civilians (Butler, Chapman, Forman, & Beck, 2006). This finding may be due in part to civilian therapies not being directly transferable to the veteran population or that other factors such as secondary gain confound results reported in veteran studies (Mossman, 1996). However, it is also arguable that the efficacy of many standard psychosocial interventions has not been adequately evaluated in veteran populations.

Strengths and limitations
This analysis used a structured and systematic approach based on the Cochrane Collaboration guidance (Higgins & Green, 2009). Three of the authors independently rated the methodological quality of each study. Unfortunately, it was only possible to perform meta-analyses of RCTs for depression, insomnia, and chronic PTSD. Only four studies (Beidel et al., 2011;Donta et al., 2003;Monson et al., 2006;Schnurr et al., 2007) provided any information about whether or not any participants experienced side-effects which might have been attributable to their intervention. The dropout rates were no higher in the intervention than the control groups across studies reviewed, however, suggesting that the interventions did not cause major adverse effects. The absence of a tolerability assessment has been noted as a shortcoming in other psychological treatment reviews (Bisson et al., 2007).
The review was limited to studies published in English, which may have meant that a number of relevant studies were excluded, although we did not identify any abstracts for studies that may have been relevant to the review when we conducted our search. Publication bias is always a concern when conducting a systematic review. However, because of the small number of studies included in metaanalyses it was not possible to explore this in this study. Several of the studies not included in the meta-analysis were of a very high quality and should be considered good examples for future researchers to follow in terms of methodological rigour (Donta et al., 2003;Schnurr et al., 2007).

Clinical implications
There is enough evidence for clinical services to treat veterans in a way that is based on the evidence base for certain conditions. The results of this review suggest that veterans respond to out-patient trauma-focused psychosocial interventions for chronic PTSD on a one-to-one or group basis with the therapist within the same room. This is consistent with the evidence from meta-analyses of civilian studies (Bisson et al., 2007) and supports a recommendation that trauma-focused interventions should be offered to all veterans with chronic PTSD.
There is also some evidence for dialectical behaviour therapy for treating borderline personality disorder in female veterans, telephone disease management for depression and at risk alcohol abuse, CBT and exercise for Gulf War illness, and CBT or panic control treatment for panic disorder co-morbid with PTSD, but replication is required.
The lack of efficacy of collaborative psychosocial interventions for veterans with depression suggests that, at present, it is appropriate to offer veterans alternative treatments for depression that have been shown to be effective in civilian populations. Caution should, however be exercised until efficacy studies of treatments such as CBT and interpersonal psychological therapy have been conducted on veterans with depression.
Veterans with mental health problems remain a difficult group for military and civilian mental health services to engage in mental health treatment programmes (Improving Access to Psychological Therapies [IAPT], 2009). Interventions may be more effective for veterans if delivered in a timely fashion post discharge from military service. It would be helpful for future research to consider how best to improve access and uptake of interventions by veterans separated from service at an earlier stage.

Implications for future research
Further well-designed RCTs of existing civilian and veteran-specific psychosocial interventions for common Study or Subgroup Carlson 1998Devilly 1998Keane 1989Monson 2006 Total (95% CI) Heterogeneity: Tau ² = 0.11; Chi ² = 5.24, df = 3 (P = 0.15); I ² = 43%  Fig. 4. Self-reported PTSD symptom severity. mental disorders in veteran populations are required. There is a need for further comparison studies of active psychosocial treatments and the role, if any, of pharmacological treatments in combination with psychosocial therapy. The role of technology that utilises psychosocial therapy in novel formats, for example, manuals, telepsychiatry (video-conferencing), telephone, and websitebased interventions that allow veterans in remote places or who will not enter psychiatric treatment settings due to stigma, and unhelpful beliefs about health providers need to be further developed and thoroughly evaluated.
There is a need for larger effectiveness trials of psychosocial therapies that are conducted in non-military settings and within cultures other than the United States. Further trials should also consider adverse events, tolerability of the treatment provided, carefully controlled for any additional intervention, and evaluate cost effectiveness.