Can digital cognitive behavioral therapy produce durable improvements in real-world settings?

Does digital CBT for insomnia produce lasting results in real-world patients?

Yes, and the evidence is strong. A 2023 randomized controlled trial tested digital CBT for insomnia (dCBT-I) in a diverse, real-world sample of 238 adults aged 19 to 81, recruited from regular care in Germany [1]. The intervention group showed a large reduction in insomnia severity (7.6 points more improvement on the Insomnia Severity Index than the waitlist control group), and these gains were maintained at both 6-month and 12-month follow-ups [1]. This matters because real-world patients often have other health conditions or life stressors that can undermine treatment effects, yet the improvements held up.

Another study on dCBT-I in college students found that sleep quality improved and fatigue decreased, with additional 'spillover' benefits like better cognitive flexibility and a greater ability to get into a focused, immersive state (called 'flow') [5]. While this study did not track long-term durability, it shows that the benefits of digital CBT can extend beyond the primary symptom, which may help sustain motivation and real-world functioning.

Do digital CBT programs for anxiety and depression work as well in everyday life?

The answer is more nuanced. A large retrospective study of over 3,400 people using a 'blended care' model (therapy sessions plus digital video lessons and exercises) found that each therapy session and each 10 minutes of digital video lessons were linked to significant decreases in both anxiety and depression symptoms over 15 weeks [4]. The effects were strongest in the early phase of treatment (weeks 0-7) but continued, though more slowly, in the later phase (weeks 8-15) [4]. This suggests that digital components can contribute to durable improvement when combined with human therapy.

However, a 2026 randomized controlled trial of 540 adults found that a generative AI-powered CBT app increased engagement dramatically (2.4 times more frequent use and 3.8 times longer duration than digital workbooks), but it did not lead to greater overall reductions in anxiety or depression symptoms [2]. This is a key caveat: more engagement does not automatically mean better outcomes. The study did find that participants who actively used the AI personalization features experienced stronger anxiety reduction, hinting that the design of the digital tool matters for real-world effectiveness [2].

How do results from real-world settings compare to controlled research studies?

Real-world studies generally confirm the findings from controlled trials, but with some important differences. A 2023 study of in-person group CBT for tinnitus in a German clinic found that patients showed clinically meaningful reductions in distress that were still present at a 3-month follow-up [3]. The size of the improvement was similar to what has been seen in controlled trials, which supports the 'ecological validity' of CBT — meaning it works outside the research lab [3]. Interestingly, this real-world study also found a reduction in tinnitus loudness, which is not typically seen in controlled trials, suggesting that real-world settings may capture broader benefits [3].

The insomnia trial [1] was specifically designed to mirror regular care by including a heterogeneous sample (wide age range, various comorbidities), and it found effect sizes comparable to or larger than those in more selective trials. This convergence between real-world and controlled evidence strengthens the case that digital CBT can produce durable improvements when deployed in everyday clinical practice.

About These Sources

This answer is built on 5 peer-reviewed studies — published from 2021 to 2026, 1 from 2024 or later, 2 in Q1 journals, collectively cited 61 times — selected as the most relevant from 5 studies that passed quality screening, drawn from 55 papers retrieved from a database of over 500 million.

Sources used in this answer

Related research questions