Online interventions for depression and anxiety – a systematic review

Background: Access to mental health care is limited. Internet-based interventions (IBIs) may help bridge that gap by improving access especially for those who are unable to receive expert care. Aim: This review explores current research on the effectiveness of IBIs for depression and anxiety. Results: For depression, therapist-guided cognitive behavioral therapy (CBT) had larger effect sizes consistently across studies, ranging from 0.6 to 1.9; while stand-alone CBT (without therapist guidance) had a more modest effect size of 0.3–0.7. Even other interventions for depression (non-CBT/non-randomized controlled trial (RCT)) showed modestly high effect sizes (0.2–1.7). For anxiety disorders, studies showed robust effect sizes for therapist-assisted interventions with effect sizes of 0.7–1.7 (efficacy similar to face-to-face CBT) and stand-alone CBT studies also showed large effect sizes (0.6–1.7). Non-CBT/Non-RCT studies (only 3) also showed significant reduction in anxiety scores at the end of the interventions. Conclusion: IBIs for anxiety and depression appear to be effective in reducing symptomatology for both depression and anxiety, which were enhanced by the guidance of a therapist. Further research is needed to identify various predictive factors and the extent to which stand-alone Internet therapies may be effective in the future as well as effects for different patient populations.


Introduction
Although there is an increasing recognition of mental health issues around the world, accessibility to healthcare has been a key problem, with specialist access in psychiatry restricted to only about 10% (Wang et al., 2005). In fact, less than a third of all patients get access to basic care (e.g. seeing a primary-care physician), and the majority (two-thirds) receives no access at all (Wang et al., 2005). Developing appropriate support strategies for the vast majority, especially for highly prevalent problems, such as mood disorders, anxiety disorders and substance-use disorders, is a critical public health challenge. Online interventions have the potential to address this gap for a variety of disorders and problems, including substance abuse, depression, anxiety, lack of social skills and panic disorders (Barak, Hen, Boniel-Nissim, & Shapira, 2008;Marks, Cavanagh, & Gega, 2007). Delivery of interventions through the Internet provides anonymity and easy accessibility, therefore making it a suitable option for clients with psychological problems to receive help. In addition, they can avoid the stigma incurred by seeing a therapist (Gega, Marks, & Mataix-Cols, 2004), and can obtain treatment at any time or place, work at their own pace, and review the material as often as desired.
With Internet-based interventions (IBIs), clients can be supported in a variety of different ways, from screening to structured assessments, and from guided self-help to sophisticated expert-system-based treatments. The level of therapist involvement can vary from no assistance or minimal therapist contact by e-mail or telephone, to the amount of involvement as seen in classic individual therapy. Thus, it may be possible to reduce the therapist involvement time while maintaining efficacy (Wright et al., 2005). Furthermore, it may be possible to reach people through the Internet who might not otherwise receive treatment for their conditions. These advantages outline the popular support that online interventions have received. Yet, the effectiveness of online interventions still needs to be evaluated in order to gain a clear understanding of their potential cost-benefit ratio.
In the past, there have been some excellent systematic and meta-analytical reviews evaluating online interventions. Griffiths and colleagues (Griffiths, Farrer, & Christensen, 2010) evaluated 26 randomized controlled trials (RCTs) on depression and anxiety interventions, excluding all other kinds of studies. They found an effect size difference ranging from 0.42 to 0.65 for depression interventions and 0.29-1.74 for anxiety interventions. Another review which evaluated online interventions for depression and anxiety concentrated on children and adolescents (Calear & Christensen, 2010). This paper compared four intervention programs describing eight different studies. Although it described each of the programs, it did not evaluate for effectiveness. An earlier meta-analysis of 12 RCTs in 2007  evaluated online interventions for depression and anxiety through a mixed effects analysis and found small mean effect sizes for depression (0.32) compared to larger ones for anxiety (0.96). Therapist-supported interventions also had larger effect sizes than the non-therapist ones.
Another quantitative review (Reger & Gahm, 2009) which combined computer-based and Internet-based cognitive behavioral therapy (CBTs) (ICTs) evaluated 19 RCTs and observed ICT was superior to wait-list and placebo assignment across outcome measures (ds = 0.49-1.14), with effects of ICT being equal to therapist-delivered treatment across anxiety disorders. Titov (2011), in his systematic review of 13 studies that dealt with Internet delivered interventions for depression, observed therapist-guided Internet cognitive behavioral therapy (ICBT) to be as effective as face-to-face psychotherapy, although there were no direct comparisons of effect sizes made across studies. A similar meta-analysis of 12 randomized trials of online interventions (which also included two computer-based studies) for depression observed a modest effect size (d = 0.41), with supported CBT having larger effect sizes than unsupported ones (Andersson & Cuijpers, 2009). There have been other reviews which have however been excluded since they focused on only computer-based CBT as opposed to Internet-based CBT or a combination of the two. Since most of the earlier reviews mentioned above were rigid in their definitions of the studies being included, several excellent studies remained excluded. This paper therefore systematically reviewed the existing research on online interventions for depression and anxiety disorders by expanding the scope of the search and including all studies on Internet-delivered interventions. It aimed to qualitatively as well as quantitatively review the efficacy of these interventions.

Method
Articles of potential relevance were identified using PsychInfo and PubMed to search a database of English language abstracts for articles published prior to January 2014. The search was carried out using the keywords "depression", "anxiety", "online", "Internet", "Web" and combinations thereof. The bibliographies of the articles identified via searches revealed additional sources. Studies were included only if they: (i) involved a self-help website intervention or an online intervention that incorporated a self-help component; (ii) described the website application as targeting a depression or anxiety condition; (iii) tested the efficacy or effectiveness of the intervention; (iv) incorporated a measure of symptom outcome for the targeted condition; and (v) had been peer reviewed and published. We specifically excluded computer-based CBT or tele-psychology, as these do not pertain to a direct online intervention. We also excluded studies, for which complete data were unavailable, which were not in English, or which did not primarily target depression/ anxiety. Two independent reviewers reviewed all literature for available interventions and existing research, which was overseen by a senior professor independently. Within-group effect sizes were extracted from either the results section of the individual papers or from the data available using the RevMan 5.0 Meta-analysis calculator (Review Manager 5 (RevMan), 2011). Effect sizes were expressed as Cohen's d values.
Although not a primary aim, the methodological quality of the studies was additionally assessed using three basic criteria: (1) clients did not have prior knowledge of treatment assignment; (2) assessors of outcomes were blinded toward treatment assignment; and (3) complete follow-up data were available (Higgins & Green, 2005).

Results
The exhaustive search yielded more than 840 articles. Abstracts that did not have an online component, described only the program and not the effectiveness of the intervention, or were duplicates were rejected. A total of 43 publications met the inclusion criteria and which reported the results of trials for IBIs were included in Tables 1-5. However, due to the variation in the methodologies of the studies reviewed, such as participant and control group characteristics, the type of intervention delivered, the differing durations of follow ups, it was not possible to compare effect sizes across studies. The studies were therefore systematically reviewed and grouped by treatment target into two categories and presented in Tables 1-5: (a) depression and (b) anxiety disorders. For each, information is provided regarding sample characteristics, intervention conditions, sessions/modules, level of clinician involvement, follow-up periods, and within-group effect sizes.

Online interventions for depression
A total of 33 studies were selected, of which 26 involved the use of various CBT techniques including psycho-education, behavioral activation, cognitive restructuring, social skills training and relaxation, as well as problem solving and relapse prevention. Also, 29/33 studies were RCTs, with or without therapist guidance (which was defined as the presence of trained therapist involvement in the delivery of the intervention). The rest were either non-randomized studies that involved CBT or randomized studies that involved non-CBT techniques.

Online interventions for anxiety
There were 24 studies that met the inclusion criteria and evaluated Internet interventions for anxiety disorders. 18/24 studies again involved the use of various CBT techniques, including psycho-education, principles of CBT, cognitive restructuring, relaxation, exposure hierarchy and graded exposure, communication and assertiveness skills and relapse prevention. Under the broad rubric of anxiety disorders the following were represented: social anxiety disorders, generalized anxiety disorders (GADs), and mixed anxiety disorders. Panic disorders and phobias were not considered for this review since the nature of interventions differed largely from the other anxiety disorders. Also most studies (N = 18) were RCTs, with or without therapist guidance (Andrews, Davies, & Titov, 2011;Berger et al., 2011;Bolier et al., 2013;Carlbring et al., 2011;Carlbring, Nordgren, Furmark, & Andersson, 2009;Hedman, Andersson, Ljótsson, Andersson, Ruck, et al. 2011;Hedman et al., 2011a;Johansson et al., 2013;Johnston, Titov, Andrews, Spence, & Dear, 2011;Lorian, Titov, & Participants with mild to moderate depression were recruited from the general population and randomized to either guided ICBT (n = 33) or to live group treatment (n = 36). The MADRS-S was used as the main outcome measure of depression at the different assessment points. In addition, the BDI was used as a secondary depression measure One hundred participants with diagnoses of mood and anxiety disorders participated in a randomized (1 : 1 ratio) controlled trial of an active group vs. a control condition. Outcome measures were the 9-item Patient Health Questionnaire Depression Scale (PHQ-9) and the 7-item GAD scale (GAD-7). All measures were administered weekly during the treatment period and at a 7-month follow-up The treatment group received a 10-week, psychodynamic, guided self-help treatment based on Affect-phobia therapy (APT) that was delivered through the Internet. The treatment consisted of eight text-based treatment modules and included therapist contact (9.5 minute per client and week, on average) in a secure online environment A large between-group effect size of Cohen's d = 0.77 (95% CI: 0.37-1.18) was found on the PHQ-9. Follow-up at 7 months Substantial within-group effects in the control group make the results harder to interpret Johansson et al.
Ninety-two participants who were diagnosed with Major Depressive Disorder according to the Mini-International Neuropsychiatric Interview were randomized to treatment or an active control. The primary outcome measure was the Beck Depression Inventory-II (BDI-II) that was administered pretreatment, on a weekly basis during the entire treatment phase, at posttreatment and also 10 months after the treatment had ended The psychodynamic treatment was given as guided self-help, with minimal textbased guidance provided on a weekly basis. In all, there were nine treatment modules, totaling 167 pages of text. The treatment modules were largely derived from the self-help book Make the leap that is based on psychodynamic principles. The treatment was called SUBGAP, which stands for (1) Seeing unconscious patterns that contribute to emotional difficulties, (2) Understanding these patterns, (3) Breaking such unhelpful patterns, and (4)  PST subjects described their problems and wrote them down. They then divided these problems into three categories: (a) unimportant problems (problems unrelated to the things that matter to them), (b) solvable problems, and (c) problems which cannot be solved (e.g. the loss of a loved one). Then problem-solving strategies or coping measures were suggested by six-step procedure: describing the problem, brainstorming, choosing the best solution, making a plan for carrying out the solution, actually carrying out the solution, and evaluation. The course took 5 weeks and consisted of one lesson a week.
Between-group effect sizes were modest (0.54)

High attrition rates (30%)
CBT is based on Coping with Depression course Farrer et al.
105 callers to a national helpline service with moderate to high psychological distress (22 or above on the 10-item Kessler Psychological Distress Scale (K10)) were recruited and randomized to receive either Internet CBT plus weekly telephone follow-up; Internet CBT only; weekly telephone follow-up only; or treatment as usual. Assessed using the Center for Epidemiologic Studies Depression Scale at pre, post, and 6 months follow-up The web-only intervention delivered online psycho-education (in Week 1, provided by BluePages: bluepages. anu.edu.au) combined with CBT (in weeks 2-6, provided by MoodGYM: moodgym.anu.edu.au). The telephone assistance was a weekly 10-minute telephone call from a telephone counselor, with the call addressing any issues associated with the participants' use of the online programs Between-group effect sizes were the highest for the Internet CBT plus weekly telephone followup (1.04) followed by ICBT only (0.76), and Telephone calls only (0.38). The effect sizes kept improving at 6 months to become more robust. Follow-up at 6 months Only single assessment tool. Spek, Nyklicek et al. (2007) In a sample of 301 participants aged above 50 years with sub-threshold depression diagnosed as a score of above 12 on the Edinburgh Depression Scale (EDS) were randomized into three groups: Internet-based CBT, group CBT and Wait-listed control group. Assessed using the BDI as outcome measure at the end of 10 weeks The group cognitive behavior therapy protocol was the Coping With Depression (CWD) course consisting of 10 weekly group sessions on psycho-education, cognitive restructuring, behavior change, and relapse prevention. The Internet-based CBT comprised eight modules with text, exercises, videos and figures based on the CWD protocol without any therapist help The group CBT condition had a large improvement effect size of 0.65, while an even larger improvement effect size of 1.00 was found within the Internetbased treatment condition High drop-out rates (52%) among Internet CBT group. Also included people with high education levels, which may have influenced results van der Zanden et al. (2012) 244 young people with depressive symptoms were randomly assigned to the online MYM course or to a waiting-list control condition. The primary outcome measure was treatment outcome after 3 months on the Center for Epidemiologic Studies Depression Scale The online MYM group course is a structured form of CBT for depression. At the core of MYM is the cognitive restructuring of thinking patterns. Course participants are encouraged to detect their own unproductive, unrealistic thoughts, and they are then taught to transform these into realistic, helpful thoughts. Performance of pleasant daily activities is also encouraged, and a mood measure is filled in daily to help understand the connection between pleasant activities and the mood level Between-group effect sizes were moderately large at the end of 12 weeks (d = 0.94; p < 0.001), which persisted at 24 weeks (d = 1.97; p < 0.001) Carlbring et al. (2013) A total of 80 individuals from the general public were randomized to one of two conditions: treatment or control. Both groups completed a weekly mood rating by answering the nine items on the MADRS-S. The primary outcome measure was the beck Depression Inventory II (BDI-II) The treatment material used in this study was a commercially available program called "Depressionshjälpen". The program has a focus on behavioral activation with influences from Acceptance and Commitment therapy (ACT Vernmark et al.
As detailed in Table 1 De Graaf et al.
A 2-armed RCT that compared the effects of access to Psyfit for 2 months (n = 143) to a waiting-list control condition (n = 141) was conducted among mild to moderately depressed adults in the general population. Secondary outcomes were depressive symptoms measured by Center for Epidemiological Studies Depression Scale (CES-D). Online measurements were taken at baseline, 2 months, and 6 months after baseline Psyfit is an online self-help intervention, without support from a therapist. The intervention is based on positive psychological principles and addresses strengths and personal competencies rather than mental problems and deficiencies. It incorporates evidence-based exercises based on positive psychology and elements stemming from mindfulness, CBT, and problem-solving therapy As detailed in Table 1 Ünlü Ince et al.
96 Turkish adults with depressive symptoms were randomized to the experimental group (n = 49) or to a wait-list control group (n = 47) and administered the CES-D for evaluating depression severity. The treatment group received the selfguided, problem-solving intervention -Turkish version (AOC-TR) which was administered in five sessions over five weeks. The control condition was a waiting list comparator. Participants were assessed online at baseline, post-test (6 weeks after baseline), and 4 months after baseline. Post-test results were analyzed on the intention-to-treat sample The AOC-TR consists of five sessions over 5 weeks. During the intervention, participants indicate what they think is important in their lives, they make a list of their problems and worries, and they categorize their problems into three groups: (1) unimportant problems, which are not related to what they think is important in their lives, (2) important and solvable problems, which are approached by a systematic problem-solving approach consisting of six steps, and (3) important but unsolvable problems, such as having lost someone through death or having a chronic general medical disease and making a plan for how to live with it. The core of the intervention is the 6step problem-solving procedure, which teaches to use this technique during the course for several of their important and solvable problems. The idea is that by mastering this technique people will regain mastery of their problems and ultimately their lives Within-group effect size was nonsignificant for the depression group (Cohen's D-0.37 (CI-0.03-0.78).
Follow-up at 4 months Small sample size Using Facebook as major recruitment strategy Table 3. Non-RCT non-CBT interventions in depression.
Lipman et al., 15 lone mothers were recruited and involved in a pilot study to improve coping and mood using Web-based video conference group cognitive therapy. Assessments included Center for epidemiological Studies Depression Scale [CES-D], Rosenberg Self-Esteem Scale, Social Provisions Scale and Parenting Stress Index-Short Form at pre and post was done. In addition, a focus group discussion of seven women was also carried out 19 patients with depression (10 and above on the PHQ-8) received the "moodManager", which was based on cognitive behavioral principles and consisted of six learning modules and four tools and was monitored by a coach. Assessed with the Hamilton Rating Scale for Depression (HRSD), Personal Health Questionnaire (PHQ-9), Perceived Barriers to Psychological Treatment (PBPT), GAD scale (GAD-7), Telephone Interview for Cognitive Status (TICS) and the 10 self-report items from the Positive Affect Scale of the Positive and Negative Affect Scale (PANAS) Learning modules (and associated tools) included the following: (1) "Getting Started", which was an introduction to the basic principles of CBT; (2) "Monitoring Activities", which described the relationship between activities and mood and introduced the "Activity Diary" tool, which allowed participants to track and rate daily activities; (3) "Scheduling Positive Activities", which taught participants to use the "Activity Scheduler", a tool used to plan and schedule positive activities; (4) "Identifying Thoughts", which described the effects of thoughts on mood and taught participants to use the "Thought Diary" tool to monitor automatic thoughts; (5) "Challenging Thoughts", which expanded the Thought Diary tool by teaching participants to develop alternative thoughts; (6) "Maintaining Gains", which summarized the skills learned and encouraged participants to continue using the tools for relapse prevention Within-group effect size was high (1.34 for HRSD, 1.96 for PHQ 9, 1.70 for GAD-7) at the end point.

No comparator group
Small sample  Significant improvements in all groups from pre-to post-treatment and from pretreatment to follow-up (p < 0.001 on every measure), but no group showed additional improvement between posttreatment and follow-up (all p's > 0.05  Significant improvements in all groups from pre-to post-treatment and from pretreatment to follow-up (p < 0.001 on every measure), but no group showed additional improvement between posttreatment and follow-up (all p's > 0.05  Paxling et al. (2011) 89 participants from the general community with DSM IV GAD were randomized into receiving either Internet-delivered CBT or a waitlisted group. Assessed with Penn State Worry Questionnaire (PSWQ), the primary outcome measure, GAD questionnaire-IV (GAD-Q-IV), Montgomery-Asberg Depression Rating Scale -Self-Rated (MADRS-S), Alcohol-Use Disorders Identification Test (AUDIT) and the CGI at pre, post, 1 year and 3 years follow-up

Health Psychology & Behavioural Medicine
The CBT program consisted of therapistguided Internet-delivered eight textbased treatment modules delivered on a weekly basis for 8 weeks. Briefly, they were (1) psycho-education (2) Step 1 of applied relaxation (3) Step 2 of applied relaxation and worry time (4) Step 3 of applied relaxation and cognitive restructuring (5) Step 4 of applied relaxation, more on cognitive distancing and problem solving (6) Step 5 of applied relaxation and worry exposure (7) Step 6 of applied relaxation, interpersonal problem solving, and sleep management and (8) Titov et al. (2008) 98 individuals with social phobia were randomized into three groups: therapist-assisted (CaCCBT), selfguided (CCBT), or to a wait-list control group. Assessments included Social Interaction Anxiety Scale (SIAS), Social Phobia Scale (SPS), the Patient Health Questionnaire Nine-Item (PHQ-9), the Kessler 10(K-10) and the SDS (Wagner et al., 2014) CaCCBT group received the Shyness treatment program consisting of six online lessons; homework assignments; participation in an online discussion forum; and regular e-mail contact with a therapist.
Large effect sizes were observed between the CaCCBT and control groups on the SIAS (0.99) and SPS (1.08). Moderate effect sizes were found between the CaCCBT and CCBT groups on the SIAS (0.64) and SPS (0.67), and small effect sizes were observed between the CCBT and control groups on the SIAS (0.34) and SPS (0.41) Lessons 1 and 2: Psycho-education; Lesson 3: Exposure hierarchy and graded exposure; Lessons 4 and 5: reinforce principles of graded exposure and principles of cognitive restructuring; Lesson 6: Relapse prevention. CCBT group received the Shyness treatment program as described above, but without regular e-mails or forum responses from the therapist (Continued ) Data from three RCTs using the Shyness program to treat social phobia were reanalyzed. The 211 subjects, all of whom met DSM-IV criteria for social phobia, were divided into four groups: (i) social phobia only (SP); (ii) social phobia with elevated symptoms of depression (SP + Dep); (iii) social phobia with elevated symptoms of generalized anxiety (SP + GAD); and (iv) social phobia with elevated symptoms of both generalized anxiety and depression (SP + Dep + GAD) The improvement in social phobia (Social Phobia Scale (SPS) and Social Interaction Anxiety Scale (SIAS), depression (Patient Health Questionnaire nine-item (PHQ-9), and anxiety (GAD-7-Item Scale (GAD-7) following Internet-based CBT (Shyness program) for social phobia was measured Within-group effect sizes were large for all groups on the SIAS (1.1-1.7) and SPS (1.0-1.2). Large effect size differences were also noted for the SP + Dep + GAD group on PHQ 9 and GAD-7. Other results were small or inconsistent Robinson et al., (2010) 138 patients with DSM IV GAD were randomized to receive ICBT which was either clinician assisted (CA; n = 46) or technician assisted (TA; n = 45) or wait-listed (WL; n = 47). Assessed with The PHQ-9, the GAD-7-Item Scale (GAD-7) Penn State Worry Questionnaire (PSWQ), the Kessler 10 (K-10), the SDS, and the Credibility/Expectancy Questionnaire (CEQ) at baseline, one week posttreatment and at three-month posttreatment (follow-up) Treatment group participants received access to the Worry program, an iCBT program consisting of six online lessons, printable summary and homework assignments, automatic emails, and additional resource documents Robust between-group effect size noted compared to control for both groups at post-treatment (1.02-1.25). No significant differences between TA and CA at this point (Cohen's d = 0.07-0.11). At follow-up, effect sizes improved, more for TA than for CA (0.2-0.3). Follow-up at 3 months The TA group received assistance from a techinican employed as an administrator while the CA group did so from a psychiatrist ICBT comprises 15 text modules, each covering a specific theme (e.g. exposure or cognitive restructuring) completed with a homework component. The modules provided the participants with the same knowledge and tools as conventional individual CBT for SAD over a period of 15 weeks.
No control group.
CBGT comprised an initial individual session followed by 14 group sessions over 15 weeks. Johnston et al., (2011).
121 Individuals meeting DSM-IV criteria for a principal diagnosis of GAD, social phobia (SP) or panic disorder with or without agoraphobia (Pan/Ag) were randomized to receive ICBT which was either clinician assisted (CA; n = 42) or technician assisted (TA; n = 39) or wait-listed (WL; n = 40). Assessed with the GAD-7-Item Scale, (GAD-7), Depression Anxiety Stress Scales -21 Item (DASS-21), Penn State Worry Questionnaire (PSWQ), Social Interaction Anxiety Scale and Social Phobia Scale -Short Form (SIAS-6/ SPS-6), Panic Disorder Severity Scale -Self Rating (PDSS-SR), Patient Health Questionnairenine Item (PHQ-9) and SDS Both treatment groups received access to the enhanced Anxiety Program comprising eight online lessons; a summary/homework assignment for each lesson; weekly telephone or email/asynchronous messaging contact with the Clinician or Technician, and regular automated reminder and notification e-mails Within-group effect sizes compared to control were modest to robust, more so for the TA group (1.06-1.3) than the CA group (0.7-0.9). Betweengroup effect size favored TA over CA group (0.3-0.5) Non-blinded assessments (Continued )  The TA group received assistance from a psychologist without specialist postgraduate training while the CA group did so from a psychologist with postgraduate training in Clinical Psychology and experience in ICBT Titov, Andrews, Johnston, Robinson, and Spence (2010) Eighty-six individuals meeting diagnostic criteria for GAD, panic disorder, and/or social phobia (by MINI) were randomly assigned to a treatment group, or to a wait-list control group. Assessments included GAD-7-Item Scale, ( One hundred participants with diagnoses of mood and anxiety disorders participated in a randomized (1:1 ratio) controlled trial of an active group vs. a control condition. Outcome measures were the 9-item Patient Health Questionnaire Depression Scale (PHQ-9) and the 7item GAD scale (GAD-7). All measures were administered weekly during the treatment period and at a 7month follow-up The treatment group received a 10-week, psychodynamic, guided self-help treatment based on Affect-phobia therapy (APT), which was delivered through the Internet. The treatment consisted of eight text-based treatment modules and included therapist contact (9.5 min per client and week, on average) in a secure online environment A moderately large between-group effect size d = 0.48 (95% CI: 0.08-0.87) was found on the GAD-7 Substantial within-group effects in the control group make the results harder to interpret Carlbring et al. (2009)   Did not measure comorbidities Titov, Andrews, Choi, Schwencke, and Johnston (2009) 163 volunteers with social phobia (DSM IV) were randomized to receive either the computerized cognitive behavior therapy (CCBT) only or CCBT + telephone calls weekly. Assessments included Social Interaction Anxiety Scale (SIAS); the Social Phobia Scale (SPS); the PHQ-9; Kessler 10 (K-10) (Choi et al., 2012); and the SDS The CCBT program involved six lessons which were: Lessons 1 and 2: Psycho-education; Lesson 3: Exposure hierarchy and graded exposure; Lessons 4 and 5: reinforce principles of graded exposure and principles of cognitive restructuring; Lesson 6: Relapse prevention. In addition, the other group also were telephoned each week by a research assistant, at a time specified by the participant, when they were commended and encouraged to persevere but no clinical advice was offered   Titov, Andrews, Choi, Schwencke, and Johnston (2009) As discussed in Table 4 Rosmarin et al.   Table 4 Ünlü Ince et al.
96 Turkish adults with depressive symptoms were randomized to the experimental group (n = 49) or to a wait-list control group (n = 47) and administered the Hospital Anxiety and Depression Scale (HADS) for evaluating anxiety severity. The treatment group received the selfguided, problem-solving intervention -Turkish version (AOC-TR) which was administered in five sessions over five weeks. The control condition was a waiting list comparator. Participants were assessed online at baseline, posttest (6 weeks after baseline), and 4 months after baseline. Post-test results were analyzed on the intention-to-treat sample The AOC-TR consists of five sessions over 5 weeks. During the intervention, participants indicate what they think is important in their lives, they make a list of their problems and worries, and they categorize their problems into three groups: (1) unimportant problems, which are not related to what they think is important in their lives, (2) important and solvable problems, which are approached by a systematic problem-solving approach consisting of six steps, and (3) important but unsolvable problems, such as having lost someone through death or having a chronic general medical disease and making a plan for how to live with it. The core of the intervention is the 6-step problemsolving procedure, which teaches to use this technique during the course for several of their important and solvable problems. The idea is that by mastering this technique people will regain mastery of their problems and ultimately their lives  Paxling et al. 2011;Robinson et al., 2010;Titov, Andrews, Choi, Schwencke, & Johnston, 2009;Titov, Andrews, Choi, Schwencke, & Johnston, 2009;Titov, Andrews, Choi, Schwencke, & Mahoney, 2008;Titov, Andrews, Johnston, Robinson, & Spence, 2010;Titov, Dear, Schwenke, et al., 2011). Among the other four studies, one was an open trial, one a follow-up study, and two others RCTs of a spiritually integrated treatment technique or a problem-solving intervention technique.
(a) RCTs (Tables 4 and 5): There were 19 RCTs with the majority involving therapist guidance (15/19) Berger et al., 2011;Carlbring et al., 2009Carlbring et al., , 2011Hedman, Andersson, Ljótsson, Andersson, Ruck, et al. 2011;Hedman, Hedman, Andersson, Ljotsson, Andersson, Ruck, & Lindefors, 2011;Johansson et al., 2013Johansson et al., , 2011Paxling et al., 2011;Robinson et al., 2010;Titov et al., 2008Titov et al., , 2011Titov, Andrews, Choi, Schwencke, & Johnston, 2009;Titov, Andrews, Johnston, Robinson, & Spence, 2010), three without therapist guidance (Bolier et al., 2013;Lorian et al., 2012;Titov, Andrews, Choi, Schwencke, & Johnston, 2009) and one compared therapist with non-therapist guidance) (Titov, Gibson, Andrews, & McEvoy, 2009). Since there were different anxiety disorders being reported, there was no single standard measure of scoring. Modules ranged from 6 to 12 weeks in length with follow-up periods ranging from 3 months to 5 years. Most studies on ICBT intervention in anxiety disorders have been on social anxiety, with a few on GAD or mixed groups. Most studies have also been therapist assisted with robust effect sizes of 0.7-1.7 observed with efficacy similar to face-to-face CBT. Stand-alone CBT studies also were observed to show large effect sizes (0.6-1.7). However, since there were only three non-therapist guided studies, it is difficult to compare them and such conclusions will await further research. All studies showed positive results, and no studies with contrasting results were reported. In addition, one study on the economics of ICBT intervention observed ICBT to be cheaper by about $2000 USD to conventional group-based cognitive therapy (CBGT) in both post-treatment and follow-up costs (Hedman et al., 2011a). (b) Other interventions for anxiety disorders (non-CBT/non-RCT) ( Table 5): There were only four studies in this category Hedman, Furmark et al., 2011;Rosmarin, Pargament, Pirutinsky, & Mahoney, 2010;Ünlü Ince et al., 2013), with two being open trials of CBT in social anxiety Hedman, Furmark et al., 2011) and the other being an RCT of a spiritually integrated treatment for subclinical anxiety (Rosmarin et al., 2010). Both studies observed significant reductions in anxiety scores at the end of the intervention.

Methodological quality of included studies
The quality of the included studies was reasonable to good. Prior knowledge of treatment assignment was presented in all studies. In most studies (>85%), outcome measures were self-reported by participants. Drop-out rates varied between 3% and 50%; hence follow-up data also varied.

Discussion
The findings of this review demonstrate that the Internet is an effective medium for the delivery of interventions designed to reduce the symptoms of depression and anxiety disorders. The effect sizes for both types of conditions were large or at least, modest. In fact, they were at least as large as standard psychological treatment (0.31) in primary care as reported in recent metaanalyses (Titov, Andrews, Choi, Schwencke, & Johnston, 2009). These effect sizes are comparable to the treatment of depression with antidepressant medication (0.37) . Similarly, the effect sizes for anxiety interventions reported here are consistent with controlled effect sizes reported for standardized CBT for various anxiety disorders (Hedman et al., 2011c;Johnston et al., 2011;Titov, Andrews, Johnston, Robinson, & Spence, 2010). However, this comes with the caveat that these effect sizes are present regardless of the timing of follow-up assessments, which range from 3 to 36 months. Most of the studies included here employed an intention-to-treat design. However, recruitment methods varied across studies, as did inclusion criteria. Some studies only included participants with a clinical diagnosis of a depressive or anxiety disorder, while others selected participants on the basis of a clinically significant cut-off score on a self-report measure or questionnaire. Others selected people with elevated but not necessarily clinically significant levels of symptoms, and one study employed a sample of participants with sub-threshold depression, specifically excluding those with a diagnosis of depressive disorder (van der Zanden et al., 2012). Among the studies included, some had assessed effectiveness of oral interventions in clinical samples with diagnosed psychiatric disorders, whereas others had used sub-threshold or only symptomatic diagnosis. A general trend of higher effect size was observed in the clinical, diagnosed samples while a moderate effect size was noted in the group having sub-threshold symptoms.
When it comes to interventions, there are wide variations in delivery as well as in components. Some programs are online versions of self-help manuals with limited or no interaction and others are based more on expert-driven structural frameworks (e.g. Deprexis) and artificial intelligence (AI), tailoring the process based on the experience of the process. When it comes to components of the interventions, there are again wide variations. The techniques that have been used in most depression studies were: cognitive restructuring (used in a total of 13 studies), behavioral activation (8 studies), and psycho-education and relapse prevention (6 studies). It is however important to note that some techniques are referred to differently but are similar to/or are part of, other strategies. For instance, "Thought Diary" and "Challenging Thoughts" are mentioned as individual techniques in one study, but since they are essential components of cognitive restructuring, they were likely used more than evidently shown. On the other hand, some strategies such as assertiveness skills training and sleep management have been used rarely.
With regard to studies about anxiety disorders, graded exposure was a reliable and often used strategy for different anxiety disorders and has been an essential component in 11 studies, while cognitive restructuring and relapse prevention were used in 10 studies each and psycho-education in nine studies, which reflects their importance for the intervention in both depression and anxiety disorders. Comparatively, techniques that were rarely used included self-confrontation and cognitive reappraisal.
In general, the techniques used were reliable and similar to clinical practice, yet some important techniques are used more in real-life CBT than in online interventions; relaxation, for example, has been used in only two studies, probably because it was implied in other techniques such as graded exposure. However, it forms an important component of conventional CBT for anxiety disorders and is one of the techniques used most often, in contrast to its rare use in online interventions. Such variations in the methodology make it impossible to compare across studies and evolve standards for assessment and intervention. Future research therefore needs to focus on effective and standard measures using interactive systems that effectively respond intelligently to the clients.
Based on the current available data, it is not possible to reliably draw conclusions about the factors that predict better outcomes. The effect sizes for anxiety trials appear larger than those for depression trials, but participants in the former trials were more often self-selected volunteers and were typically only included in the trial if they also satisfied diagnostic criteria at screening. There were also high drop-out rates, sometimes reaching 50%, with makes it difficult to comment on the actual effectiveness of the therapies. However, it can also be argued that this reflects treatment settings in the real world, with common drop-out rates reaching 50% in outpatient clinics.
Emerging evidence across trials, especially in depression, clearly suggests that IBIs aim not to replace human interaction, but could help to make much better use of extremely limited expert time. In fact, a direct correlation has been found between the amount of therapist contact in minutes and the between-group effect size (Lorian et al., 2012). However, the picture is not that clear for anxiety disorders, with two studies on CBT without therapist guidance observing near equal effect sizes as those with therapist guidance. Overall, Internet-based treatments are equally efficacious in generating a strong therapeutic alliance (Titov, Andrews, Choi, Schwencke, & Johnston, 2009), while not being strongly associated with outcome. Areas which need to be addressed in the future are the effects of Internet-based treatment related to age and gender, the mechanisms of action and the appropriateness of psychosocial techniques for the online treatment. In addition, it is also essential to explore how these interventions can be integrated into a system of knowledge exchange and assessment strategy.
One of the strengths of this review, which attempted to systematically review the efficacy of online interventions for depression and anxiety, has been the use of original data and outcomes, which allow more clarity and intuitive reading. We believe that transformation of data or outcomes, although important, can sometimes make the review complicated to understand, and we have tried to keep it simple and informative to the reader. In addition, this review has been very exhaustive yet specific, in that it has targeted an understanding of all interventions for anxiety and depression delivered over the Internet. However, the findings of this review have to be understood in light of certain limitations. Despite our broad search, it may be possible that certain good studies may not have been included or been overlooked, due to the search criteria. Also, the desire to be very inclusive may have diluted actual effect sizes or precluded some useful comparisons. Since there were significant differences in study groups, type of interventions and study periods, it was difficult to generate a standardized mean difference (SMD) using pooled data analysis, making any definitive conclusion impossible. Also, we found no trials involving rural residents, older people, or people with low levels of education, which is another limitation of this review. Although we also attempted to analyze the cost-effectiveness of online interventions, the lack of studies prevented us from doing any comprehensive analysis.