In the world of work-from-home (WFH), Zoom meetings and FaceTime catch-ups, it is not surprising that therapy sessions are increasingly done online, in the comfort of one’s home. Despite some controversy around whether the face-to-face element of the therapy-client relationship can be matched through a screen, high-quality research studies have found that therapist-guided, internet cognitive behavioural therapy (also known as iCBT) can be just as helpful as the traditional and trusted in-person version (Hedman B et al., 2012). Even if people seeking support are open to the idea of iCBT, some mental health professionals are still concerned that online therapy is not as effective or not applicable to everyone.
By default, iCBT requires significantly less therapist time, which makes it more cost and resource effective. On the flipside, it is understandable that therapists are concerned that the lower time investment may lead to lower effectiveness.
A group of researchers from Finland (Rosenström TH et al., 2025) noticed that the current evidence, although supportive of iCBT, is based on randomised controlled trials. While these trials are considered high quality evidence, they include participants who volunteered to take part. This means that people sceptical of iCBT or don’t use technology, for example, may not have taken part – therefore, the findings may not be able to paint a real-world picture of iCBT effectiveness. The researchers tried to clarify this by using data from national healthcare system records and advanced machine learning techniques to provide a more realistic picture of the effectiveness of iCBT.
Internet-delivered CBT is a more scalable alternative to traditional in-person CBT.
Methods
The researchers used Finish healthcare records of thousands of adults who received either traditional, face-to-face (392 records) or online therapist-guided CBT (5467 records) to help with depression between 2018 and 2022. They chose a retrospective cohort study design, meaning they used previously collected data from a large number of individuals.
They used a machine learning technique called counterfactual causal inference. This statistical method allows them to compare the effects of CBT by hypothetically asking what would have happened to the same individual if they had received the opposite form of therapy, i.e. face-to-face or iCBT. Changes in depression symptoms were measured using a widely used symptom questionnaire called the PHQ-9 (also known as the Patient Health Questionnaire).
Results
Amongst the total of 5,834 included healthcare records, individuals were slightly younger on average in the iCBT group (34 years) than in the face-to-face CBT group (40 years). 70% of the included patients identified as female, and the majority were Finnish speakers. Those who received iCBT received 5 therapy sessions on average, while the average number of face-to-face sessions was 12.
The key finding was that across both CBT delivery formats, there was a similar improvement in depression symptoms.
Participants’ depression symptoms measured by the PHQ-9 reduced by 4.8 points in iCBT, and 3.0 points in face-to-face CBT. The researchers conducted further statistical checks to make sure the findings were robust and not occurring just due to chance. Following these they were confident that the similar benefits of the two CBT format remained consistent.
Some differences between how individuals reacted to CBT were found, which is likely to happen in the real world. These depended on factors such as the number of sessions and sex, but since the variations occurred regardless of CBT format, the researchers concluded that the type of delivery did not meaningfully affect the benefits.
The study found similar benefits of therapy regardless of the delivery format.
Conclusion
While the benefits of iCBT had already been proven in randomised controlled trials, this study added a new level of insight by showing that its effectiveness remains stable even in real-world settings using healthcare records in Finland. The machine learning techniques used make these findings robust and applicable to healthcare settings.
Given the flexibility, time- and cost-effectiveness of iCBT, these findings will hopefully reassure clinicians, particularly those skeptical about the benefits of iCBT, that clients could experience a similar reduction in depression symptoms even without the in-person element of therapy. This could also mean that more people could receive support more quickly, which is key in the current climate of long waiting lists for mental health support.
The use of healthcare records in this study allowed for real-life, applicable findings about the benefits of online vs. face-to-face CBT.
Strengths and limitations
A key strength of this study is that it used a robust statistical approach that also allowed for practically applicable, real-world findings to be produced. This directly tackled the limitation of previous evidence which was based on a population of participants who chose to engage in studies – this may not have been an accurate representation of those seeking support in routine healthcare settings.
The researchers used sensitive statistical approaches to account for participants stopping therapy early – something that is common in real-life settings – which adds more confidence to the applicability of the findings. They also measured symptoms using a standardised questionnaire, the PHQ-9, which is commonly used by mental health professionals worldwide.
Despite their efforts to minimise bias – the risk of a finding being due to shared characteristics within the participant group – some uncertainties remain. For example, the authors did not report why some participants stopped their sessions, which could have been informative. They also did not include factors specific to the use of digital CBT, such as motivation to use online support, or digital literacy. It is likely that this data was not available or routinely collected in healthcare records, so future studies may be able to investigate these questions specifically.
Finally, because iCBT is easier to deliver on a wider scale, the majority of participants in this study had received it and were spread out nationwide. Face-to-face participants were located in a limited number of regions in Finland, so there may have been geographical and demographic factors that affected how the different types of CBT were received. Since healthcare systems, their capacities in terms of staff training and time, as well as access to digital tools differ across the globe, it would be quite interesting to see whether and to which extent these findings would hold up in different countries and cultures (De Jesús-Romero R et al., 2024).
The study used robust statistical methods and accounted for issues such as drop-out from therapy.
Implications for practice
These findings strengthen the evidence on the effectiveness of iCBT, which will hopefully help increase healthcare providers’ confidence in offering it where possible and appropriate. At a time of increasingly high demand and burden on mental health support services, continuing to roll out the iCBT offer could be a key avenue to narrowing the gap between needing support and receiving it in a timely fashion.
There are people for whom face-to-face therapy may not be an option. We know that travel cost, distances and stigma are some of the barriers people may face. The findings that iCBT can be just as effective offers important reassurance to these individuals that they will receive a similar standard of care even if they can’t attend therapy in person. This therefore has real potential to make therapy more inclusive and accessible.
As someone who does research in the field of digital mental health, I’m optimistic about the wide reach iCBT can have and the solutions it could offer. At the same time, I don’t think it will be for everyone and there still may be individuals and populations for whom the face-to-face element of therapy will be invaluable. It also may be that iCBT is not applicable or as effective across mental health problems, but even offering it as the initial step to someone on a waiting list for further support could potentially be incredibly helpful.
Another aspect to consider is that this study focused on therapist-guided iCBT, which is not to be confused with self-guided iCBT. An interesting avenue for future research would be to further explore the effectiveness of the self-guided version using a similar study design to complement existing evidence (Pelucio L et al., 2024). Since the benefits of self-guided iCBT may not be as significant as those in therapist-guided online therapy, this could potentially suggest that there is something particularly important about the role of a therapist guiding the client, rather than the format of delivery (face-to-face vs. online).
The increasing confidence in the benefits of iCBT could narrow down the need-access gap for mental health support.
Statement of interests
NK research interests focus on digital mental health and interventions. No conflicts of interest to declare.
Edited by
Dafni Katsampa.
Links
Primary paper
Rosenström TH, Saarni SE, Saarni SI, Tammilehto J, Stenberg JH. et al (2025). Efficacy and effectiveness of therapist-guided internet versus face-to-face cognitive behavioural therapy for depression via counterfactual inference using naturalistic registers and machine learning in Finland: a retrospective cohort study. The Lancet Psychiatry 2025 12(3) 189–197.
Other references
De Jesús-Romero R, Holder-Dixon AR, Buss JF, Lorenzo-Luaces L. et al (2024) Race, ethnicity, and other cultural background factors in trials of internet-based cognitive behavioral therapy for depression: Systematic review. Journal of Medical Internet Research 2024 26 e50780.
Hedman E, Ljótsson B, Lindefors N. et al (2012) Cognitive behavior therapy via the Internet: a systematic review of applications, clinical efficacy and cost–effectiveness. Expert Review of Pharmacoeconomics & Outcomes Research 2012 12(6) 745–764.
Pelucio L, Quagliato LA, Nardi AE. et al (2024) Therapist-guided versus self-guided cognitive-behavioral therapy. The Primary Care Companion for CNS Disorders 2024 26(2).
