Impacts of Internet-Based Interventions for Veterans With PTSD: A Systematic Review and Meta-Analysis

Background: Veterans who did not seek and complete treatment as intended have been shown to have an elevated risk of experiencing and being exposed to post-traumatic stress disorder (PTSD). Internet-based interventions (IBIs) provide more confidentiality and fewer treatment barriers, and they are regarded as potential treatments to reduce PTSD in veterans. However, the effects of IBI for veterans with PTSD are inconclusive. Objectives: IBI is defined as any internet-based series of psychosocial interventions, of which the internet works as a way of delivery. Psychosocial content and reduction of PTSD symptoms in veterans have been recognized as two core elements of this intervention. This study aimed to (1) examine the effects of IBI on veterans’ PTSD outcomes and (2) distinguish between the elements of IBI that play an important role for veterans with PTSD. Methods: Web of Science, PubMed, EMBASE, PsycINFO, Cochrane, Wanfang Data, CNKI, and CQVIP databases were searched for randomized controlled trials (RCT) in IBI programs for veterans with PTSD, covering all studies in English and Chinese published from January 1990 to November 2020. Also, related studies tracking citations were identified. Studies met the following inclusion criteria of (1) being RCTs; (2) containing IBI in the full text; (3) having IBI conducted on veterans as participants; and (4) being on PTSD. All processes followed PRISMA. The risk of bias of the studies was assessed by the Cochrane Systematic Review Handbook. The confidence of outcomes of this review was valued according to the GRADE (Grading of Recommendations Assessment, Development, and Evaluation). The meta-analysis was done by RevMan 5.13. Two teams of reviewers independently searched the literature, made the assessment, and extracted the data. Results: A total of 1,493 citations were identified after initial searching, of which the full texts of 66 studies were screened. Eventually, six RCT studies met the inclusion criteria. Beneficial effects of IBI were found on the overall PTSD outcome (−0.29; 95% CI–0.48 to −0.11, p<0.01). Particularly, IBI based on cognitive behavioral therapy (CBT) with peer support was found to be effective for PTSD outcomes (−0.36; 95% CI–0.61 to −0.11, p<0.01). The subgroup analysis demonstrated that scores of PTSD outcome measured by a PCL (PTSD Checklist) decreased to an average score of 0.38 (95% CI –0.60 to −0.15, p=0.001). The intervention had a positive effect on the PTSD outcome on veterans with comorbid psychological disorders (−0.30; 95% CI –0.61 to −0.11, p<0.01). Overall, the six studies included were evaluated with a low risk of bias, and the outcomes of the meta-analysis were proven with high confidence. Conclusion: On the whole, IBIs have a positive effect on the overall PTSD outcome of veterans. The results encouraged us to focus on IBI with CBT with peer support for veterans, on specific instruments for veterans with PTSD, and on veterans with comorbid psychological disorders. This study, however, has limits. Only six studies with a Western population were included, which might result in cultural bias on IBI effects. In future, more high-qualified research and diverse cultural background of RCTs is needed to prove the effectiveness of IBI on veterans with PTSD.


INTRODUCTION
Post-traumatic stress disorder (PTSD) is a prevalent, chronic, and disabling psychiatric disorder often diagnosed among military personnel in relation to combat trauma (Possemato et al., 2011;McLean and Foa, 2013). A combination of childhood abuse, combat traumas, and/or sexual assault was suggested as the reason for veterans' PTSD (Morland et al., 2015). The study by Niemeyer et al. (2020) proved that PTSD had a 12-month prevalence rate of 3.7% among the United States population and 2.3% in the German population and other European countries. Among veterans of the wars in Iraq and Afghanistan, PTSD prevalence was estimated to be 10-20%, and female military personnel were more likely than male veterans to experience higher PTSD prevalence rates (Morland et al., 2015).
PTSD is related to many negative effects on patients, such as significant adverse health and life consequences (Morland et al., 2015). PTSD can cause severe distress, chronic suffering, and impairments, of which the core symptoms include persistent avoidance of traumatic content, re-experiencing traumatic content, and negative alterations in cognition, arousal, and reactivity (Haagen et al., 2015). Firstly, PTSD results in altered cognition that includes hypersensitivity to physical sensations or mislabeling these symptoms as common physical health problems (Possemato et al., 2011). Individuals with PTSD report poorer general health and bodily pain (Stevens et al., 2016;Hamblen et al., 2019). Also, veterans with PTSD often have high rates of comorbid mental health disorders (Engel et al., 2015). When patients do not receive timely and effective treatment, PTSD often becomes a chronic and disabling disorder that is often comorbid with substance abuse disorders, major depression, psychological distress, and other anxiety disorders (McLean and Foa, 2013;Niemeyer et al., 2020). What is worse, PTSD is associated with impairment in social/occupational functioning (Hamblen et al., 2019). Veterans with PTSD are believed to be less able to engage in physical activities or fulfill previously held roles (Stevens et al., 2016). They are reported to have increasing impairments in performing life roles (e.g., work and family responsibilities) and daily tasks (e.g., household chores; Stevens et al., 2016). PTSD is also related to violent behavior (e.g., aggression; Morland et al., 2015).
Traditionally, several evidence-based psychotherapeutic approaches to treat PTSD have been recommended, including CBT, cognitive processing therapy (CPT), prolonged exposure therapy (PE), stress management skills training, eye movement desensitization reprocessing (EMDR), and other innovative interventions (Kearney et al., 2012;Rodriguez et al., 2020). Cognitive-behavioral conjoint therapy (CBCT) and integrative behavioral couple therapy (IBCT) were designed to combine acceptance strategies and IBCT techniques to ameliorate the symptoms of PTSD, relationship conflicts, and experiential avoidance (Monson et al., 2009). Structured approach therapy (SAT) as an approach in a stress inoculation therapy (SIT) framework consists of psycho-education, skills training, and coping skills (Sautter et al., 2015). PE has also been shown effective in lessening PTSD symptoms and maintaining treatment gains over time (McLean and Foa, 2013). EMDR produced significant enhancements in decreasing anxiety, anger, depression, isolation, and other symptoms (Koven, 2018). Mindfulnessbased stress reduction (MBSR) is a standardized class series to teach mindfulness to reduce stress by a combination of meditation, body awareness, and mindful breathing, which have been shown to have significant improvements in PTSD symptoms (Kearney et al., 2012;Colgan et al., 2016;Koven, 2018). Mindfulness-based cognitive therapy (MBCT) also belongs to stress-reduction groups involved psychotherapies incorporating mindfulness techniques (King et al., 2013). Yoga has also been used to manage stress via adjusting the cognitive perspective and focusing on inner peace (Reddy et al., 2014). Building spiritual strength (BSS) is an inter-faith, manualized, spiritually integrated intervention to support the participants in their choice of religious affiliation or non-affiliation (Harris et al., 2011). Logotherapy has also been applied as an innovative adjunctive treatment for PTSD by healing through meaning (Koven, 2018). Another complementary intervention is a specially trained psychiatric service dog, offering emotional and therapeutic value (Rodriguez et al., 2020).
Traditional intervention sessions are usually conducted by therapists with the veteran and the veteran's partner (Sautter et al., 2015). While, compared with non-military groups, veterans have a higher need for services addressing PTSD. Perceived stigma of treatment seeking was proven to be negatively related to treatment utilization in a study with a sample of 812 veterans (Kulesza et al., 2015). And for some veterans, logistics and confidentiality concerns also posed problems with treatment utilization (Engel et al., 2015;Krupnick et al., 2017;Niemeyer et al., 2020) Nontraditional online health care pathways have been proposed as possible solutions to cope with the disadvantages of stigma, confidentiality concerns and treatment inefficacy (Parish et al., 2014). It was found that web-based interventions for PTSD were a feasible and helpful intervention for veterans (Belsher et al., 2015). The feasibility, acceptability, and usability of IBI have been well documented (Bush et al., 2011).
According to Barack and Klein, IBIs are defined as therapeutic programs with various health objectives mainly delivered via the internet, described with respect to four major components: content, multimedia, interactive online activities, and guidance and supportive feedback (Barak et al., 2009;Guay et al., 2017). The U.S. Defense Department's National Center for Telehealth & Technology has developed a web through which to provide a set of information on resources, tools, and aids for military personnel, veterans, and their families (Bush et al., 2011). Web-based self-management has been proved an effective online tool for the military (Bush et al., 2011). A web-based video gallery has been provided to share personal stories about PTSD and treatments (Hamblen et al., 2019). In a systematic review and meta-analysis, e-Mental health interventions revealed a significant improvement in PTSD symptoms in the active intervention group among a general population (Simblett et al., 2017). Yet, to our knowledge, there is limited literature on meta-analysis and systematic review of IBI on veterans with PTSD. One review analyzed the prevalence of eight disorders among elderly veterans and found that PTSD prevalence was not markedly high (Williamson et al., 2018). To reduce PTSD symptoms, a meta-analysis suggested that iCBT for PTSD was efficacious, and only one study was analyzed based on a military sample (Niemeyer et al., 2020). One review included 12 studies to access the interventions for supporting partners of military veterans with PTSD, among which only one was based on IBI (Turgoose and Murphy, 2019). However, IBIs in other groups were tested as effective in supporting informal caregivers (Sherifali et al., 2018), which helped us to see the potential of IBI on veterans with PTSD in our study. Therefore, the primary aim of this research was to conduct a systematic review and meta-analysis to examine the effects of IBI on veterans diagnosed with PTSD who were living in a community or absent from military life. The secondary objective was to distinguish the contents of IBI that played an important role for veterans to decrease levels of PTSD.

Population
A total of 622 participants in all six studies included veterans, members of the active military on leave with PTSD, and a small number of civilians diagnosed with PTSD (see Table 1). The average age of the participants in the studies was 38.88.

Interventions
In the definition of IBI, the internet works as a method of delivery assisted by therapists, mental health providers, counselors or specialists, and nurses. The platform aims at conducting PTSD care and relief for veterans by stress-management skills, cognitive reframing techniques, treatment sessions, and CBT lessons with peer support.

Inclusion and Exclusion Criteria
Two teams of researchers screened the literature independently.
Research studies must meet the following inclusion criteria: be RCTs; contain IBI in the full text; have IBI conducted on veterans in trial groups; have no intervention, a delayed intervention, or another psychosocial intervention or usual care but not online conducted on participants in the control group; have PTSD as the outcome; be studies published between January 1990 and November 2020; and be studies published in English or Chinese.
Interventions conducted in closed settings (e.g., medical institutions) or those including the use of drugs were excluded.
To ensure the quality of the study, research not published in a peer-reviewed journal, without data extraction or of detailed information for meta-analysis was also excluded.

Search Strategy
We conducted a detailed automated search and screened relevant literature in the Web of Science, PubMed, EMBASE, PsycINFO, Cochrane, Wanfang Data, CNKI, and CQVIP databases for articles that were RCT in IBI programs for veterans with PTSD. Also, reference lists of related studies were manually tracked. Our search term consisted of three subsets: participant ("veteran" and "military"), intervention ("internet" and "web"), and outcomes ("PTSD" and "mental health"). We connected them with two Boolean operators (AND and OR) to search for relevant research in English and in Chinese published from January 1990 to November 2020 from the databases.

Selection Results
We followed the PRISMA statement to select studies (Moher, 2010). As shown in Figure 1, a total of 1,493 studies in the literature related to the topic were identified from eight databases (n = 1,488) and other resources (n = 5). After the removal of duplicate literature, 1,222 articles remained. Based on the inclusion and exclusion criteria, two teams of reviewers independently screened and excluded a total of 1,153 articles as non-related. The remaining 69 articles were further screened after reading the full-text articles. Considering that there might be topic-related articles or the titles and abstracts were insufficient to judge whether articles were useful. In the remaining 69 studies, we identified six eligible RCTs for our meta-analysis.

Date Extractions
To answer the research questions, we generated a data extraction table. The coding scheme mainly extracted the contents, including participants' characteristics, intervention design, outcome, measurement tool, and the duration of follow-ups. Basic information was extracted, including title, authors, country of authors, publication year, abstract, objective, and the journal. We also addressed the quantitative data: mean and standard deviation (SD). Some literature did not report these data, and we calculated the mean and SD of these studies according to the standard error (SE), d effect size, values of p, 95% confidence intervals (CIs), and other information given in the papers. The detailed information of these studies is reported in Tables 1-4.
All data extraction was completed by two independent reviewers and verified by another reviewer, to ensure the reliability of data extraction. The third reviewer was also responsible for the reconfirmation of inconsistent data that the two reviewers extracted. All reviewers agreed upon the content of the final data extraction table after discussion.

Quality Assessment
Two independent reviewers assessed the risk of bias of each eligible study by the Cochrane Systematic Review Handbook. This quality assessment evaluates the levels of bias in the following content: selection, performance, detection, attrition, reporting, and other biases (Higgins et al., 2011). The confidence  of outcomes of this review was also evaluated by two reviewers and rechecked by the third reviewer according to GRADE, which includes the risk of bias, inconsistency, indirectness, inaccuracy, and publication bias (Grade Working Group, 2004;Moher, 2010). All disagreements on risk bias and quality assessment were discussed to reach a conclusion among the reviewers.

Data Analysis
We used the data (mean, SD) to perform the meta-analysis on the effectiveness of the IBI. I 2 was used to demonstrate the heterogeneity among the studies in quantitative statistics and when the heterogeneity test I 2 < 50%, a fixed-effects model, was used for meta-analysis (Higgins et al., 2003). Accordingly, this study used a fixed-effect model to explore the relationship between the studies. According to the differences in intervention content, measurement tools, and status of comorbidity, we grouped the studies and conducted a subgroup analysis to determine which components were playing key roles in mitigating PTSD symptoms in veterans. We also conducted a sensitivity analysis to test the stability of outcomes. The meta-analysis was done by RevMan 5.3.

Description of Studies
As shown in Table 1, one of the studies was conducted in Germany (Niemeyer et al., 2020) and the other five in the United States. The mean age of the participants was under 50 (range 36.2-48). Two studies did not report the mean age of the experimental group and the control group (Stevens et al., 2016;Possemato et al., 2019). More than half the veterans were male in the majority of the studies (range 81.25-100%),  except for one study in which all participants were female (Morland et al., 2015). All six studies had experimental groups defined as veterans receiving IBI, and five studies had control groups with veterans having optimized usual primary care (OUC; Engel et al., 2015), treatment in person (NP; Morland et al., 2015), web-based supportive treatment (Litz et al., 2007;Possemato et al., 2019) and no treatment (Stevens et al., 2016;Niemeyer et al., 2020). One study had civilian women as participants in the control group (Morland et al., 2015). In all six studies, there were three types of intervention contents as branches of IBI. Two of these studies (33.33%) were categorized as using IBI in psycho-education, which contained self-training and education for stressful situations and self-management treatment (Litz et al., 2007;Engel et al., 2015). Two studies (33.33%) were regarded as using IBI in psychotherapy, which included two kinds of trauma-focused psychotherapy (Morland et al., 2015;Niemeyer et al., 2020). Two studies (33.33%) used IBI on CBT with peer support, which provided a platform for veterans to build social connections (Stevens et al., 2016;Possemato et al., 2019). PTSD outcomes were examined by three subgroups of the IBI, as shown in Table 3. Outcome assessment tools used for the outcome were categorized as CAPS and PCL. Three studies were assessed by CAPS (Morland et al., 2015;Possemato et al., 2019;Niemeyer et al., 2020), and the others were measured by PCL (Litz et al., 2007;Engel et al., 2015;Stevens et al., 2016). In total, five of the six studies were judged as small in sample size (≤100; Litz et al., 2007;Engel et al., 2015;Morland et al., 2015;Possemato et al., 2019;Niemeyer et al., 2020), and one study contained a large sample with 209 recipients in the experimental group and 94 recipients in the control group (Stevens et al., 2016). The length of follow-up in all studies was short (≤6 months). Regarding the recipients' mental health condition, six studies included veterans with PTSD (Litz et al., 2007;Engel et al., 2015;Morland et al., 2015;Stevens et al., 2016;Possemato et al., 2019;Niemeyer et al., 2020). Varying degrees of psychological disorders (depression, anxiety, dysthymia, substance use disorder, and perceived physical health impairment) were represented as comorbidities in five studies (Litz et al., 2007;Engel et al., 2015;Morland et al., 2015;Stevens et al., 2016;Niemeyer et al., 2020). In one study, veterans struggled with hazardous alcohol use in addition to PTSD (Possemato et al., 2019).

Results of Overall Effectiveness
This systematic review and meta-analysis examined the overall effects of IBI on veterans' PTSD outcomes. As shown in Table 2 and Figure 3, the scores dropped an average of 0.29 (95% CI -0.48 to −0.11, p < 0.01). It was also found that there was no heterogeneity between the studies (p < 0.01; I 2 = 0%). The assessment was evaluated as having high quality.

Results of Subgroup Analysis
For the subgroup of intervention measures, three kinds of IBI content have been distinguished. All six studies applied mature IBI training platforms or treatments: online selfmanagement training (Engel et al., 2015), psycho-education via the internet (Litz et al., 2007), online therapy (iCBT; Niemeyer et al., 2020), videotele conferencing (VTC; Morland et al., 2015), CBT with a forum (Stevens et al., 2016), and CBT with peer support (Possemato et al., 2019). Considering the content and targets of treatments, we identified three subgroups of these interventions, which focused on providing psycho-education (Litz et al., 2007;Engel et al., 2015), psychotherapy (Morland et al., 2015;Niemeyer et al., 2020), and CBT with peer support (Stevens et al., 2016;Possemato et al., 2019). As suggested in Table 3, we found that for PTSD symptoms of veterans, the effects of CBT with peer support (SMD = −0.36, 95% CI −0.61 to −0.11) were more helpful than were the effects of psycho-education (SMD = −0.35, 95% CI −0.82 to 0.12) and the influences of psychotherapy (SMD = −0.12, 95% CI −0.47 to 0.24). According to Figure 4 (see Appendix 1), it was also found that the subgroups were not significantly different (p > 0.05; I 2 = 0%), suggesting there was no heterogeneity between the groups. Yet, the result with IBI based on psychotherapy has been found significantly effective on PTSD outcome (p < 0.01). IBI with psycho-education (p > 0.05) and peer support (p > 0.05) had no statistically significant effects on veterans' PTSD outcomes. The results show high quality evaluated by GRADE.
Concerning outcome measurements of PTSD, we split the instruments into two subgroups: CAPS (Morland et al., 2015;Possemato et al., 2019;Niemeyer et al., 2020) and PCL (Litz et al., 2007;Stevens et al., 2016;Engel et al., 2015). It was proven that IBI was more effective on PTSD measured with PCL (SMD = −0.38, 95% CI −0.60 to −0.15)  than was PTSD evaluated with CAPS (SMD = −0.11, 95% CI −0.45 to 0.22). As shown in Figure 5, it was proved that the subgroups were not significantly different (p > 0.05; I 2 = 38.5%), indicating that there was no heterogeneity between the groups. However, IBI has been found to be significantly effective on PTSD outcomes measured with PCL (p = 0.001). IBI had no statistically significant effects on veterans' PTSD outcome accessed with CAPS (p > 0.05).
In the six studies, the veterans with PTSD all had comorbidity. Five had a sample with veterans with depression and anxiety symptoms in addition to PTSD (Litz et al., 2007;Engel et al., 2015;Morland et al., 2015;Stevens et al., 2016;Niemeyer et al., 2020), and one study had a sample with veterans with comorbid PTSD and hazardous alcohol use (Possemato et al., 2019). IBI was shown more effective in veterans having depression and anxiety as comorbidities (SMD = −0.30, 95% CI −0.49 to −0.11) than in veterans with PTSD and hazardous alcohol use (SMD = −0.10, 95% CI −0.99 to 0.78). As presented in Figure 6, the subgroups were not significantly different (p > 0.05; I 2 = 0%), proving that there was no heterogeneity between the two groups.
IBI was found to work significantly on veterans with PTSD, depression, and anxiety (p < 0.01) and have no statistically significant effects on veterans with PTSD and hazardous alcohol use (p > 0.05).

Results of Sensitivity Analysis
In order to verify the stability of the results of the metaanalysis, the exclusion of a single study with a significantly large sample size was conducted to test its effects on the overall results. Based on the results shown in Table 4, of the six studies, the one by Stevens et al. (2016), an RCT from the United States with the largest sample size (166 participants in the experimental group and 90 participants in the control group) was eliminated. The sample size was much larger than that of other included studies. After excluding this study, the results (Figure 7) showed p > 0.05. This suggests that the overall effect was not significant, which is inconsistent with the conclusion before the sensitivity analysis, indicating that the overall results of this combination were unstable.
The study by Morland et al. (2015) was an RCT in the United States with a sample of female veterans and female  civilians. It was also reported as a high-risk study lacking reports on SD. After excluding this study (Figure 8), it showed that the overall effect was significant, which is consistent with the conclusion before the sensitivity analysis, suggesting that the overall results of this combination were stable. Similarly, Niemeyer et al. (2020) conducted an RCT with a sample of servicemen and male veterans in Germany. The trial was reported to be high risk in other biases. After excluding this study (Figure 9), it proved that the overall effect was significant, which is also consistent with the conclusion before the sensitivity analysis, showing that the overall results of this combination were stable.

Principal Findings
Several systematic reviews and meta-analyses of IBI have been conducted on a reduction in symptoms of the mental health of caregivers, including depression, stress or distress, and anxiety (Guay et al., 2017;Hopwood et al., 2018;Sherifali et al., 2018;Zhao et al., 2019). However, less often discussed are the effects on veterans of PTSD. This is the first metaanalysis to include RCT studies, and it is targeted at reducing the symptoms of PTSD of veterans using IBI. We screened the research based on the inclusion and exclusion criteria. A total of six studies were included in the final analysis, and these were assessed with a low risk of bias. All were RCTs done with rigorous research and experimental design and provided high-qualified. The results of these studies reveal the beneficial effects of IBI on veterans with PTSD. A subgroup analysis was conducted to explore the interventions that were the most beneficial for veterans' PTSD, the efficacy differences within diverse outcome measurements, and the differences of the effects on groups with particular comorbidity. The quality of evidence was examined according to the GRADE scores. The quality of evidence was high for the outcomes examined. The research in this review was almost low-risk, indicating the convincing quality of evidence. However, the literature had a common problem of lacking research details and quality reports.
In this meta-analysis, we proved that IBI was indeed beneficial in reducing the effects of PTSD on veterans. This is consistent with the review that IBI is effective in lessening the effects on anxiety and depression in caregivers of persons with disabilities (PwD; Zhao et al., 2019). One possibility is the advantages of technology. The new modality offered strengths over traditional treatment methods, including a reduction in traveling time and transportation costs and enhancement of access to treatment for persons with serious injuries, physical disabilities, or scheduling difficulties (Bose et al., 2001). The efficacy, feasibility, and acceptability of this new approach have proved preferable, according to a review on psychotherapy delivered through VTC among diverse patient populations (Backhaus et al., 2012). Moreover, the effectiveness could come from the recipients feeling safer and more comfortable when using online services in their own homes (Krupnick et al., 2017). All trials showed there were positive improvements by various interventions on  veterans' PTSD status to some degree. Considering sample size, test efficiency, rates of loss in follow-up, lack of follow-up, and the evolution of veterans' illness (Melville et al., 2011;Zhao et al., 2019), we set a rule and chose the reported improved scores among several results of follow-ups. There was no follow-up in several studies (Stevens et al., 2016;Niemeyer et al., 2020). Also, the reduction of PTSD symptoms was greatest at 12-week follow-up but decreased over time.
No benefits remained at 18-week follow-up (Engel et al., 2015). Another study found one significant enhancement at three-month follow-up (Morland et al., 2015). According to Litz et al. (2007), there was a great reduction in PTSD, and the veterans gained high functioning at 6 months. Also, continuing improvements have been observed in posttreatment (12 weeks) and follow-up (24 weeks) in the trial by Possemato et al. (2019). Thus, we chose the reported means and SD of 12-week follow-up (Engel et al., 2015), 3-month follow-up (Morland et al., 2015), 6-month follow-up (Litz et al., 2007), and 24-week follow-up (Possemato et al., 2019).
Regarding the categories and components of IBI, several aspects of IBI interventions showed a promising reduction in PTSD symptoms for veterans. Subgroup analyses were conducted to determine which components were effective in supporting veterans with PTSD (Hopwood et al., 2018). According to the results, CBT with peer support has more positive effects on veterans' PTSD than the other two kinds of IBI. This is highly consistent with one study finding that peer support helped patients manage physical health concerns effectively in peer navigation and support in selfmanagement interventions (Cabassa et al., 2017). Supportive Accountability has clearly pointed out that interpersonal support improves the efficacy of eHealth interventions (Mohr et al., 2011). Only IBI based on CBT with peer support has been found to be significantly effective on PTSD outcome, which indicates that these two types of IBI might be not appropriate for veterans struggling with PTSD. It also shows that IBI is more effective on PTSD measured with PCL than was the outcome evaluated with CAPS. It might be that the study by Stevens et al. (2016) used the PCL-M scale especially for the military, which was more appropriate to measure veterans' situations than was CAPS as a general scale to access PTSD status. In particular, this study has the largest sample (166) of the six studies, which might play a decisive role. This result suggests that IBI is effective in PTSD outcomes measured with PCL, meaning PCL might be a significantly appropriate instrument for testing the effects of IBI on veterans with PTSD. Similarly, IBI is shown more effective in veterans with depression and anxiety as comorbidity than in veterans with PTSD and hazardous alcohol use. This is in accordance with the research finding that online interventions reported a significant decrease in caregivers' depressive symptoms and anxiety (Beauchamp et al., 2005;Hopwood et al., 2018). In particular, the internet proved an effective modality to deliver self-care for depression and PTSD (Christensen et al., 2002;Wagner et al., 2011). IBI has been found to work on veterans with PTSD and depression and anxiety only, suggesting that IBI might be significantly appropriate across veterans with comorbid PTSD and depression and anxiety.
Based on the results of sensitivity analysis, the sample size is a key element that weakens the stability of meta-analysis. It has also been discussed in other reviews that the sample size of studies was insufficient (Sherifali et al., 2018;Zhao et al., 2019). In this review, the studies by Litz et al. (2007) and Possemato et al. (2019) were both based on a small sample size (<20).

Implications for Interventions and Researches
To develop IBI for veterans, some issues need to be addressed. The question of willingness of utilization is first. In one study, it was found out that only 30.9% of the veterans with probable PTSD were willing to use online interventions when faced with all seven kinds of e-Mental health services (Whealin et al., 2015). To solve this problem, familiarsounding language and optimization of intervention should be used in the future, through which the veterans would feel more motivated to apply IBI (Murray et al., 2016). Based on the findings of subgroup analysis, intervention contents of IBI played a distinguished role in the reduction of veterans' symptoms of PTSD. Among diverse contents provided by IBI, CBT with peer support offered the most meaningful support for veterans with PTSD, and more similar online services could be delivered accordingly. Secondly, computer literacy and access to the internet are factors in need of attention, and this is described as a digital divide in the literature (Bernhardt, 2000). This is necessary before participation in IBI. Finally, internet privacy and security are emphasized (Schaller et al., 2016), in order to avoid secondary injury of veterans.
This meta-analysis has demonstrated the demands for more and high-quality RCTs to access the effects of IBI on veterans. Based on the results of subgroup analysis, more details like instruments of outcomes and comorbidity of participants could be evaluated in the research design. Meanwhile, a sufficient sampling size of participants could be set to raise the quality of RCT researches. Future studies should describe more details of interventions to help us understand the elements of IBI that provide the most benefits. Future research should focus on the different types of militaries and indicate the content of IBI that is beneficial for the differentiated groups. Moreover, this review failed to distinguish the treatment trajectory in various stages of PTSD among veterans. All IBI in this review was broad for the course of the entire treatment process, ignoring the special needs during each stage of PTSD, which should be addressed by future studies.

Strength and Limitations
This systematic review and meta-analysis have summarized the most relevant evidence in assessing the benefits of IBI on veterans' PTSD outcomes. All six studies were published between 2007 and 2020, most published from 2015 to 2020, emphasizing the increasing interest in internet technologies to support veterans. This research is the most recent review and meta-analysis which examines the effects of IBI on PTSD outcomes of veterans, using the latest empirical studies and innovative subgroup analysis. In particular, it includes both Chinese and English studies, which take full account of cultural diversity.
However, three qualified research studies were dismissed without the full text in this meta-analysis. Also, only RCTs have been searched. Some of the trial protocols and reports have not been published. The lack of grey literature and limitation of language may lead to selection bias.

CONCLUSION
The basis of evidence for IBI on veterans with PTSD remains limited. This meta-analysis is the first study to discuss the effectiveness of IBI on veterans. IBI has been proved as a positive modality to reduce PTSD symptoms among veterans. Moreover, it encourages us to focus on IBI with CBT with peer support for veterans, specific instruments for veterans' PTSD, and veterans with comorbid depression and anxiety. Our research results highlight the promising potential of IBI to support veterans. It warrants further development and testing within more RCTs with high-quality and diverse cultural backgrounds.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.