The latest developments with internet-based psychological treatments for depression

ABSTRACT

Introduction

Internet-based psychological treatments for depression have been around for more than 20 years. There has been a continuous line of research with new research questions being asked and studies conducted.

Areas covered

In this paper, the author reviews studies with a focus on papers published from 2020 and onwards based on a Medline and Scopus search. Internet-based cognitive behavior therapy (ICBT) programs have been developed and tested for adolescents, older adults, immigrant groups and to handle a societal crisis (e.g. COVID-19). ICBT works in regular clinical settings and long-term effects can be obtained. Studies on different treatment orientations and approaches such as acceptance commitment therapy, unified protocol, and tailored treatments have been conducted. Effects on quality-of-life measures, knowledge acquisition and ecological momentary assessment as a research tool have been reported. Factorial design trials and individual patient data meta-analysis are increasingly used in association with internet intervention research. Finally, prediction studies and recent advances in artificial intelligence are mentioned.

Expert opinion

Internet-delivered treatments are effective, in particular if therapist guidance is provided. More target groups have been covered but there are many remaining challenges including how new tools like artificial intelligence will be used when treating depression.

KEYWORDS:

• Internet delivery
• cognitive behavior therapy
• research designs
• depression
• artificial intelligence

1. Introduction

Depression continues to be a major challenge for societies across the world and has been rated as one of the largest contributor of years lived with disability, with increased rates reported during the last decades [1]. While the diagnosis of major depressive disorder is complicated by comorbidity and heterogeneity, there are several treatment options including both pharmacological and psychological treatments [2]. One way to increase access to psychological treatment is to deliver the treatment using information technology, and in particular internet-delivered psychological treatments have been found to work as a complement and sometimes replacement for face-to-face treatments [3]. Using the internet to provide guided self-help for depression has been around for about 25 years, and there are studies suggesting that guided internet-delivered cognitive behavioral therapy (ICBT) works as well as face-to-face CBT [4] and yields long term effects [5]. Guided treatments usually involve human contact from a distance via e-mail or other means of communication and should be distinguished from pure self-help with no or automated contact [3]. As with psychotherapy in general, not all clients respond to internet-based psychological treatments [6], and there are several challenges for researchers to consider when it comes to understanding mechanisms of change, predicting what works for whom and to target sub-groups such as ethnic minorities and less favored regions in the world.

There is a large literature on ICBT [7] and increasingly on other forms of psychological treatments such as psychodynamic internet treatment [8]. Reading of the literature is complicated by the fact that many different terms are used to refer to ‘internet-based treatments’ including ‘internet-delivered’, ‘digital’, ‘online’, and ‘eMental Health’. Moreover, given that symptoms of depression are very common in other psychiatric conditions such as anxiety disorders and somatic disorders, the literature on internet-based treatments becomes even larger as there are literary several 100s of controlled trials in which measures of depression symptoms have been included. Here I will focus on depression studies but it should be acknowledged that research focusing on other relevant transdiagnostic psychological problems like loneliness [9] and poor self-esteem [10], both of which are associated with symptoms of depression, are also of relevance for how we understand and treat depression.

The aim of this narrative review is to provide examples of recent research and to comment on the latest developments in the field of internet-based psychological treatments for depression. I will focus on new research published from the year 2020 and onwards (based on Medline and Scopus searches), related to different target groups, treatment contents, alternative outcomes, novel research designs and approaches, and finally technical challenges and opportunities.

2. Target groups

There is now much research activity focusing on different target groups. Adolescents with major depression have been in focus of some recent studies even if there are still more studies on adults. Melcher et al. (2022) conducted a randomized controlled non-inferiority trial which included 272 adolescents with major depression aged 15–19 years old [11]. They were randomly assigned to either ICBT (n = 136) or internet-delivered psychodynamic treatment (IPDT) (n = 136). Both treatments were in the format of guided self-help lasting 10 weeks. A non-inferiority margin of Cohen’s d = 0.30 was set and the results showed large within-group effects for both treatments (ICBT: within-group d = 1,75 and IPDT d = 1,93) on the main depression outcome measure. Non-inferiority for IPDT was established with no statistically significant differences. The authors concluded that the range of accessible and effective treatment alternatives for adolescents with depression is not limited to ICBT but could also include IPDT.

While older adults usually are not excluded from trials on adults, there is much less research on older persons with depression than on other age groups. One systematic review focused on ICBT for symptoms of depression in older adults [12]. The authors could include nine studies with a total of 1272 participants with an average age of 66 years. Of these nine studies there was only three controlled trials. The mean within-group effect size was Cohen’s d = 1.27 and for the controlled studies a between-group effect of d = 1.18. Interestingly, authors reported a small but statistically significant negative effect of old age which was associated with smaller effects (b = −0.06). Given the lack of good quality studies the authors recommended more research. Older adults are however often represented in trials on specific somatic conditions and for example in a study on ICBT for depression in association with cardiac disease the mean age at baseline was 63 years [13]. Overall, treating depression in association with somatic disorders is a field that deserves more attention as comorbidity with for example chronic pain is substantial and treatment programs for somatic conditions often include treatment ingredients based on depression protocols [14].

Another challenge for research and clinical practice is to increase diversity by translating, culturally adapting and testing internet interventions for other target groups than traditional Western clients. One example is a pilot RCT conducted in Sweden on Arabic-speaking immigrants [15]. This study tested the effects of individually tailored ICBT for depression and included 59 participants. Compared to a wait-list control group with a between-group effect at post-treatment of Cohen’s d = 0.85 was reported.

We have just recently left the COVID-19 pandemic behind us, but when it occurred increased levels of depression, anxiety and stress were reported [16]. Many researchers focused on this during the pandemic and one example is a trial on tailored ICBT which included 76 participants with mixed psychiatric symptoms [17]. Decreased levels of depression compared to a control group were reported (Cohen’s d = 0.51). The study suggests that ICBT can be rapidly adapted and tested even in times of crisis.

There is now increasing evidence suggesting that ICBT works in regular clinical settings and that long-term effects can be obtained. Etzelmueller et al. (2020) conducted as systematic review of nonrandomized pre-post design studies conducted under routine care conditions [18]. For the 13 studies on depression the authors reported a pooled within-group effect of Hedges’ g = 1.78, and even if there was a significant heterogeneity the effects reported were overall moderate to large (ranging from 0.66 to 1.88). Given that the first studies on ICBT for depression were published more than 20 years ago it is interesting to note that the efficacy trial findings found their way into regular clinics with some like the Mindspot clinic in Australia treating large number of clients [19].

Regarding long-term results, a recent systematic review focused on the long-term effects of ICBT for depression in adults [20]. The time for the follow-ups was set at ≥8 weeks after post treatment measurements, with follow-up periods ranging from 2 months to 1 year post treatment. In contrast to a previous review on the long-term effects of ICBT (2 years or more), this review focused on controlled trials which the previous review did not [5]. Mamukashvili-Delau et al.. (2023) could include 15 studies (with 17 samples) and the overall results across all time points was a medium between standardized mean group difference (SMD) effect size of 0.43 favoring the treatment groups (based on 1689 participants and 9 RCTs).

3. Treatment models

Research usually gives support for CBT as most internet studies have used CBT protocols. I mentioned earlier a trial on adolescents in which IPDT was compared against ICBT showing similar effects [11], and it is likely that therapist-supported IPDT is as effective as ICBT with guidance [8].

When it comes to acceptance commitment therapy (ACT), which often viewed as a form of CBT, there are now an increasing number of internet studies on depressive symptoms. In one recent meta-analysis [21], an analysis of 31 controlled trials (with a total of 4124 participants) was conducted. It is important to note that this analysis included several targets groups and not only clients with major depression. Overall results showed a small between-group effect size on measures of depression (SMD = 0.30), but a subgroup analysis on five studies directly targeting depressive symptoms showed a larger effect (SMD = 0.62).

Overall recent studies tend to focus on different forms of CBT for depression as for example unified protocol or tailored based on patient profiles and preferences. Unified protocol is a transdiagnostic form of CBT for emotional disorders which includes common and often shared treatment components present in for example CBT for depression [22]. One recent control trial of an internet-delivered guided unified protocol treatment included 129 participants who were randomly assigned to treatment or a waitlist condition [23]. On the depression measure PHQ-9 (Patient Health Questionnaire 9) the authors reported a mean Hedges’ g between group effect size of 0.91 in favor of the treatment. They also reported that treatment gains were maintained at a 1- and 6-month-follow-up. Dropout rate was about ¼ and a few more also did not provide follow-up measures. There are previous earlier trials on transdiagnostic internet treatments with similar large effects [24], but this study is among the first to directly test the protocol developed by Barlow and colleagues [22].

A different approach is to tailor the treatment using a selection of treatment modules. This could be referred to a transproblematic rather than transdiagnostic as different comorbid diagnoses can be directly targeted [25]. One recent example will be given later on in this paper. Overall, there are now several different psychological treatment approaches in addition to CBT that are being developed and tested, including unpublished work on internet-delivered interpersonal psychotherapy.

4. Alternative outcomes

While measures of depression symptoms usually become the primary outcome, it is also important to investigate if treatment works in other domains such as quality of life. A recent systematic review and meta-analysis focused on quality of life outcomes in ICBT studies and also reported separate outcomes for depression studies [26]. In this review the effects of ICBT on quality of life in the depression studies (n = 19) was a between group Hedges’ g = 0.343, which is a small effect but in line with another meta-analysis on quality of life treatment effects in depression studies [27].

Studies on knowledge acquisition as a complementary outcome in ICBT trials (see next section) have been reported, which has also been studied using qualitative methodology [28]. Berg et al.. (2020) interviewed 10 adolescents who had completed guided ICBT for depression and identified two themes ‘Active agents of CBT’ and ‘Passive agents of CBT.’ Active agents of CBT relate to remembering and actively applying specific CBT principles in life, whereas the Passive agents of CBT rather reflect vague memories of CBT and more passive/reactive use of CBT strategies. This study can serve as an example of an increasing number of qualitative studies on how ICBT is perceived and used. In addition, the role of the therapist in ICBT has been studied using qualitative methodology [29,30].

Some ICBT studies and projects have included ecological momentary assessment (including smartphone data collection), and one example is a trial derived from the large EU-project E-compared [31]. There are still some uncertainties in intensive data collections such as timing of assessments, but overall and with the use of modern statistical tools important research can be done. In this particular case authors could not find statistically significant cross-lagged coefficients but observed cross-sectional positive associations between activities, pleasure, and mood levels which are of relevance for understanding how ICBT works.

5. Novel research designs and approaches

Previous psychotherapy studies have often been underpowered to test active ingredients and a factorial design approach can be a more effective way to answer more than one question relating to ‘what works for whom,’ and are also possible to conduct given the larger sample sizes in ICBT studies. I will here comment on two recent examples. In one large factorial depression trial, 767 participants were included who were randomly assigned to 32 experimental conditions testing the presence or absence of seven treatment components [32]. While there was a substantial dropout the study was well powered to test effects and interactions. Interestingly, the trial could not detect any large main effects of treatment condition and the authors discussed if it was the case that the shared components explained the lack of differences and also that insufficient module completion may have influenced. On the other hand improvements overall were in line with previous ICBT depression trials.

Another example of a factorial design ICBT depression study is a recent trial testing fewer contrasts and with fewer participants (N = 197). Specifically, the role of own choice versus prescription of tailored treatment modules, support on demand versus scheduled support and finally the role of having the clients being the target for supervision were tested [33]. The treatment period lasted for 10 weeks and in addition to immediate effects a two-year follow-up was included. A knowledge test was also included. Overall, within-group effects were large across conditions (e.g. d = 1.73 on the BDI-II). The authors also reported a small but significant difference in favor of self-tailored treatment over clinician-tailored (d = 0.26). The other two contrasts did not yield large differences, and overall effects were largely maintained at a two-year follow-up. Interestingly, while there were large within-group effects on the knowledge test at post-treatment with a within-group Cohen’s d = 2.14, there was a significant reduction on the knowledge scores at two-year follow-up d = −0.72). This raises interesting questions regarding what relapse prevention should focus on in future research.

Another important methodological development which has benefitted ICBT research is the possibility to conduct individual patient data meta-analyses and also network meta-analyses. I give one example here, in which the authors could include 48 trials with individual data and an additional 28 trials with group data yielding a large number of 23 995 participants [34]. The authors reported strong support for internet-delivered behavioral activation but also that relaxation was not effective. In line with much research baseline depression scores were a strong predictor of outcome. We will most likely see more examples of individual patient data meta-analyses but they require that researchers agree on both how both outcomes and prognostic factors are defined and measured.

6. Technical challenges an opportunities

With more studies and data it easier to inform treatment guidelines about prognostic factors and one recent systematic review focused on predictors and moderators of outcome in ICBT trials [35]. This review included 60 studies and depression studies were well represented (but not separately reported). This is one of the first systematic reviews to focus on a broad set of predictors and moderators (a total of 88). The authors reported that better treatment adherence, treatment credibility and working alliance were clear predictors/moderators of outcomes. In line with expectations higher baseline scores predicted more change but at the same time higher post-treatment symptoms. Overall, it is clear that there are many non-significant predictors which can be explained by the fact that strong predictors of negative outcomes tend to similar to the exclusion criteria in trials but also in regular clinics. For example, as most ICBT trials rely on reading texts not being able to read is an exclusion criteria which then disappears as possible predictor.

Finally, I will say a few words about artificial intelligence and how programs like Chat GPT have stirred up a lot of emotions relating to the pros and cons [36]. There are some concerns about artificial intelligence such as poor data quality and illogical results. The pros involve for example using artificial intelligence for emotion detection and help when writing texts. Machine learning data analytic methods are now increasingly used in research, for example in a depression study in which 894 patients were included and the authors reported that they derived a multi-modal classifier for predicting major depression remission status after ICBT [37]. Among the predictors were demographic, clinical, process, and genetic variables. Other authors have called for use of machine learning tools when personalizing internet treatments [38].

However, when it comes to conversational agents and artificial intelligence, studies on Chat GPT and similar tools are most likely already in progress but ethical issues and preferences are important to consider. It is however already the case that researchers use these tools to refine and update their interventions, and for example translation and cultural adaptation of treatment materials will most likely be made easier with the use of artificial intelligence. However, do we want our therapist to be an AI application?

7. Expert opinion

In this narrative review I selected recent systematic reviews and original studies in order to illustrate new developments in the field of internet-based psychological treatments. It is in my opinion now even more clear that psychological treatments can be delivered from a distance in the guided self-help format [39]. There is less research on real-time video therapy and blended treatments in which face-to-face and online components are mixed. The research on digital interventions overall is hard to grasp given inconsistent and even conflicting definitions [40], and the field would benefit from standardization of terminology not only related to the name of the interventions but also how compliance, treatment completion and outcomes are defined.

In this paper with its focus on very recent papers several observations can be made. First, researchers have been active in widening the number of targets groups for whom internet treatments may work. Not only adults but also adolescents, older adults, immigrant groups are now being studied. The COVID-19 pandemic has been described as a game changer and increased the use of distance technology including internet interventions [41]. However, when we now return to business as usual it will be important to reflect on the pros and cons of remote treatments while retaining the lessons learned during the pandemic.

Another observation made in this review paper is the increased research on different therapy orientations both within and outside of CBT. With regards to ICBT there are now large scale studies suggesting that it works in regular clinical settings and smaller studies have documented long term effects. This has yet to be studied for the other ways to provide internet treatments such as IPDT. Another line of research I mentioned is the use of alternative outcomes and this may be a way to move the field of psychotherapy research forward as even for a problem like major depression there is much we do not know regarding change processes and other outcomes than symptoms. Different research methodologies is another way to move a field forward and both qualitative studies and individual patient data meta-analyses will inform the field and generate new ideas for future research. Perhaps the most striking example of benefits of the internet treatment format is the possibility to run factorial design trials. This has the potential to answer outstanding questions regarding what works for whom as we arguably know by now that getting a bona fide depression treatment is better than being on a waiting list. On the other hand statistical power will be crucial and strong independent variables in order to obtain robust findings.

Finally, given the vast amount of research it is now possible to draw at least some conclusions regarding predictors of outcomes. Most likely this endeavor will be informed even more by recent advances in the use of artificial intelligence tools that have been around for a while but are yet sparsely used.

This narrative review has obvious limitations. In order to provide a broad overview of recent advances in the field a systematic review approach was not used even if literature searches were conducted. It is possible that I have missed important new studies and the selection of papers I highlighted was informed by what I focus on in my own research and my role as editor of a journal that publish much research in the field. It is interesting to note that research now is being conducted in many places in the world and not only Western countries, but they are still not well represented here. Another limitation is that I was alone writing this review. This can also be seen as an advantage but will be more vulnerable for selection bias and personal opinion.

If this paper had been written four years back, I would not have been able to foresee the COVID-19 pandemic and how it affected attitudes toward and use of internet treatments. In the upcoming years studies we will hopefully be able to grasp what actually occurred, but we will also be better prepared to handle new societal crises. Much ongoing work is for example under way dealing with the war situation in Ukraine and refugees. In the depression field the upcoming five years researchers will most likely continue the expansion of evidence-based psychological internet treatments for a broader range of target groups. This includes dissemination and sharing of treatment contents between countries. There are several examples of transfer from one language to another without loss of effects when studies are replicated in the new language (I gave an example of an Arabic language trial earlier e.g. Lindegaard et al., 2020). We will also see more reports of outcomes in regular clinical settings and this is the place to study predictors as efficacy trials tend to be too strict with exclusion criteria and sometimes even not representative for the typical client seen in regular services. We will also see more trials on other psychotherapy orientations than CBT. This will inform us if a treatment works in different formats equally well, which in the case of depression is likely [42], but obviously needs to be tested as most studied have been based on CBT.

Finally, and this is a very uncertain prediction, conversational agents involving programs like Chat GPT will be studied in the upcoming years. It will at least be possible to speed up treatment development but if we can use the technology as a complement or even replacement of a therapist is still uncertain.

Article highlights

• Internet-based psychological treatments can be effective for adolescents and older adults as well as for immigrant groups.

• In particular therapist-supported ICBT appears to work in regular clinical settings.

• The long-term effects of guided ICBT have been documented.

• Different forms of CBT and other treatment orientations like psychodynamic internet treatment have been tested and found to be effective.

• In research symptom measures are now complemented by other outcomes such as knowledge acquisition.

• Different research methodologies are now used more frequently such as factorial design trials, individual patient data meta-analyses and qualitative methods.

• Future studies will probably focus on broadening the scope of internet treatments and discover the pros and cons of recent developments in artificial intelligence and conversational agents.

Declaration of interest

The author has no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Reviewer disclosures

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Additional information

Funding

This manuscript was funded in part by Linköping University.