Internet-Delivered Psychotherapy for Anxiety Disorders and Depression

More than 150 clinical trials evaluating Internet-delivered psychological treatments for anxiety disorders and depression have now been reported, many of which are randomized controlled trials (25). Meta-analyses consistently report the efficacy of these interventions in reducing symptoms of anxiety, depression, and disability relative to control conditions (26–30). One important caveat is that with few exceptions (e.g., 31, 32), therapist-guided iCBT has consistently resulted in superior clinical outcomes relative to self-guided iCBT, which is typically associated with lower effect sizes and higher dropout rates (27, 33–35). For example, one meta-analysis reported an overall effect size (Cohen’s d) of 0.41 for studies evaluating the efficacy of iCBT for depression (27). However, this effect size increased to 0.61 when studies evaluating self-guided versions were removed. Underestimating the important differences between therapist and self-guided iCBT may partially explain discordant results reported in a recent meta-analysis (36), which confounded therapist-guided and self-guided iCBT for depression and concluded iCBT for depression was less effective than reported by previous meta-analyses, which had considered these differences. Notwithstanding the previous literature, we suspect that self-guided interventions will become increasingly effective. Consistent with this, some recent studies which evaluated well-developed iCBT interventions have found similar outcomes when the intervention is offered in therapist-guided and self-guided formats (37, 38).

An inspection of outcomes for iCBT for specific disorders of anxiety and depression is instructive. In a surprisingly short time treatments have been developed and tested for depression (39–43), generalized anxiety disorder (44, 45), social anxiety disorder (46–48), panic disorder (49, 50), obsessive compulsive disorder (51, 52) posttraumatic stress disorder (19, 53, 54), severe health anxiety (55), and specific phobia (56). Although most studies have involved adults, there are also studies on children and adolescents (57, 58), older adults (59–61), and an emerging number of studies with people from non-Western backgrounds (62, 63; unpublished paper of R. Kayrouz, B.F. Dear, L. Johnston, and N. Titov). The majority of these studies have reported large effect sizes (Cohen’s d > 0.80) on symptom measures relevant to the disorders of interest, relative to control conditions. These gains have been sustained at follow-up 5 years after treatment completion (64), and patient satisfaction with treatment is often high, although it should be noted that there is considerable heterogeneity in the outcomes of self-guided iCBT.

The consistent and converging evidence for the clinical effectiveness, acceptability, and cost-effectiveness of iCBT, which includes results replicated independently by multiple research teams, provides considerable confidence in the validity of the results of iCBT. This robust evidence base has led to attempts at implementing iCBT in regular clinical settings. A recent review of iCBT in clinical settings identified four randomized controlled studies and eight open studies conducted in outpatient clinics (65). All studies clearly showed that the promising effects for iCBT in clinical trials appear to be replicated when the treatment is transferred to a regular clinic. Moreover, direct comparisons of face-to-face psychological treatment and iCBT have shown equivalent outcomes (66), with gains sustained in the long term (67).

A key priority for all mental health services is the assessment, monitoring, and management of patient safety before, during, and after treatment. To date, there is no evidence to indicate that iCBT is associated with increased risk to the safety of patients, but negative effects have occasionally been reported (71). Indeed, there are several aspects of iCBT which may facilitate the assessment, monitoring, and management of patient safety (72). First, regular monitoring of symptoms may reduce the risk of significant deterioration, decompensation, or crisis in patients by identifying patients who are beginning to deteriorate before they reach crisis. Second, such monitoring, combined with informed decision rules based on patient characteristics and treatment progress can be used to develop an escalating series of responses and actions by a therapist, designed to help maintain patient safety in an evidence-based manner, and to identify patients who require other or more intensive forms of treatment. Third, iCBT systems can provide a wealth of data about a patient’s use of and engagement with the treatment intervention, which when combined with data about symptoms, can be used to generate automated alerts and systems to inform clinical supervision, especially for high risk patients. Similar to face-to-face mental health services, an outstanding challenge for iCBT is how to manage patients with elevated symptoms who withdraw from treatment and do not respond to attempts at contact by the therapist. Such risks can be mitigated by making safety plans a condition of use of a service and by informing primary health care practitioners about their patient’s engagement or disengagement with the iCBT service. Such safety plans can include consenting that a therapist will contact significant others in situations where a patient’s safety appears to be at risk.

By virtue of its innovative nature and ability to provide services across traditional geographic and jurisdictional boundaries, iCBT often raises questions about clinical and legal governance (73). For example, an iCBT service may provide treatment to patients in different states or even countries, and each jurisdiction may have different requirements or guidelines about clinical governance, duty of care, health professional practice regulations, indemnity, information technology (IT) security, and storage of health records. Although some of these questions and challenges are similar to those experienced by other industries providing services via the Internet, including online sales and marketing services, health services often have more stringent privacy regulations than other industries. For these reasons, readers are strongly encouraged to seek advice and guidance from their legal teams before providing online mental health services.

iCBT also raises questions about the nature of psychotherapy and the validity of assumptions about the role and importance of the therapist for successful psychotherapy. For example, one key but generally unspoken assumption in psychotherapy is that the therapist is essential for effective treatment, and that therapist training, expertise, and skill are the most important predictors of clinical outcomes (74). On the one hand, the results of meta-analyses of outcomes of iCBT are consistent with this assumption. For example, meta-analyses consistently show that iCBT interventions that include therapist-guidance achieve better outcomes than self-guided treatments (33–35). On the other hand, the results of several studies indicate that patients who are able to persevere with self-guided iCBT may obtain significant benefits. Consistent with this, there is an emerging body of literature indicating that self-guided treatments that use automated reminders, which provide encouragement, can result in high adherence rates and good clinical outcomes (31, 75), and that such reminders can be particularly helpful in the early stages of treatment (32). Notwithstanding these results, there are important advantages to guided iCBT. First, a therapist can conduct an assessment, which will help determine the suitability of iCBT for a patient and, if suitable, can help prepare the patient for participating in iCBT. Second, the therapist can tailor an intervention for a patient’s specific symptoms and difficulties and can assist the patient to identify and address barriers to engaging in treatment or address difficulties with applying the intervention to their situations. Related to this, when a patient communicates with a therapist via text messaging or secure e-mail, the therapist can consult colleagues and other experts before answering and providing feedback (76). Third, there are clear indications that support increases adherence and prevents dropout, an important issue given that at least some self-guided interventions have suffered from high dropout rates (e.g., 77). Fourth, therapists can actively recommend and assist patients to access other services, including social, health, and crisis services, prior to, during, or after iCBT. There remain outstanding questions in the literature about the timing of support with one promising line of research indicating evidence of benefit when support is offered either before or after treatment, or on demand (78).

In addition to questions about when therapist support should be provided, there are also questions about the optimum frequency and form of therapist support that should be provided. Although earlier reviews indicated a dose-response relation between the duration of therapist support and the clinical outcome, more recent results indicate that outcomes for treatments in which substantial support is given do not appear to differ from treatments which include 10 minutes or less of therapist time per client per week (79). Studies also indicate that equivalent clinical outcomes can be obtained whether support is provided by an expert therapist or nonexpert coach, providing the latter is under careful clinical supervision and the iCBT is highly structured and has been previously evaluated with expert therapists (39, 80–82). This body of research is important given the shortages of mental health professionals in many parts of the world. It is unclear whether similar outcomes can be obtained with less-structured interventions. These issues are inherently related to the cost-effectiveness of iCBT. Guided iCBT has been reported to be cost-effective relative to traditional treatments administered face-to-face (83, 84). However, the provision of guidance is indeed more costly than self-guided treatments, and self-guided treatments with small clinical effects can still be cost-effective (85). Thus, from a public-health perspective, the minimal costs of providing iCBT without guidance can, in some cases, be justified if they are safe. However, more research is needed; in particular, it would be helpful to conduct comparisons of the effects of less costly forms of guidance such as from carefully trained and supervised coaches, peers, and from models of support that are provided on-demand.

There are other outstanding questions regarding the type of therapist support required to optimize outcomes, including questions about the relevance of therapist factors to iCBT, which are widely held to be important in face-to-face treatments (86). Several studies indicate small or no differences in outcomes between therapists in iCBT (87, 88). However, other studies indicate that some therapist-related behaviors are important. For example, in a study in which the therapist correspondence was coded, it was found that a lenient attitude toward homework was associated with worse clinical outcome (21). Consistent with this, observations from our online research and clinical work indicate that better outcomes are associated with adherence to clinical scripts and workflows, which direct patients to key material and minimize therapist drift.

Therapeutic alliance is another factor that is widely regarded as important in psychotherapy outcome research, and an emerging body of literature has explored this issue in the context of iCBT. Several studies have collected data from patients on how they rate the therapeutic alliance with their online therapists (89), and most show no association with outcome, even if alliance ratings tend to be fairly high (90, 91). There are a few studies in which alliance early in the treatment predicts outcome (92, 93), however, we suspect that therapeutic alliance is likely to be more important in less structured iCBT.

An additional set of questions relate to how therapists might actually use iCBT in their own practice. At least three approaches can be taken. First, therapists can use aspects of technology to create their own iCBT. This might involve the use of the telephone, video conferencing, and e-mail or instant messaging technology, combined with online questionnaires, written treatment materials, and other treatment resources. Second, therapists may elect to simultaneously provide a patient with psychotherapy, while referring that patient to receive iCBT from another service for problems other than those addressed by the therapist. For example, a therapist may elect to provide a patient with psychotherapy addressing relationship distress, while referring the patient to receive iCBT to address a depressive disorder. To our knowledge, the efficacy of this combined approach has not been systematically explored. Anecdotally, our experience is that patients have a threshold for the amount of therapy they can tolerate and, in some cases, participating in more than one therapy at a time may result in conflicting messages between therapy sources, a sense of overload, poor compliance, and therapy burnout. A related approach that we have observed, which has been more successful, is when a therapist provides a patient with face-to-face treatment, but uses the content and structure of an existing iCBT intervention to guide therapy and homework tasks. A third approach is to sequence iCBT and face-to-face treatment within a stepped-care model. For example, where iCBT is used as a first treatment step followed by more intense face-to-face treatments when needed (94). However, this may be more appropriate when self-guided iCBT constitutes the first step and to a lesser extent when guided iCBT is the first step. Our experience is that configuration may also occur in the opposite direction, that is, patients who have failed face-to-face treatments may subsequently improve following iCBT. Such cases are likely to reflect issues associated with treatment readiness rather than about any specific modality of treatment. Nevertheless, more research is needed to address these issues and explore the optimal use of iCBT in stepped-care models.

A final and related set of questions concerns how iCBT can be integrated with existing mental health services. On the one hand, this is a simple question as individual therapists may develop their own iCBT intervention or use an existing one. On the other hand, this is a complex question and requires consideration of the local existing services and the gaps in those services. For example, many mental health systems prioritize resources to the most severe patients, including those with psychotic disorders. In such systems services for people with anxiety disorders and mild to moderate levels of depression are often considered lower-funding priorities and such people may thus remain underserved by existing services. In such situations iCBT represents a promising strategy for increasing the capacity of existing services to provide clinically effective treatments to such people. Two promising examples of iCBT services that have attempted to integrate into regular health care include the Internet Psychiatry Unit in Stockholm, Sweden, and the MindSpot Clinic (www.mindspot.org.au), which provides services nationwide in Australia. Another promising example of using iCBT is following face-to-face therapy as a way to reduce relapse rates (95). This may be a way to prevent relapse, as it is well known that for many psychiatric problems relapse occurs even after successful therapy. Notwithstanding such examples of how iCBT can be implemented, there is a considerable effort required to educate mental health professionals about the strengths and limitations of such services before they are accepted and integrated more broadly into regular health care systems.