Sleep as a measurable focus for digital health development

01

Therapeutic content turns digital access into treatment

CBT-I means cognitive behavioural therapy for insomnia: structured work on sleep-related thoughts, time in bed and habits in people with persistent sleeplessness. [1]

Hwang’s meta-analysis found a standardised symptom difference of −0.71 for fully automated CBT-I compared with control conditions. Effects varied substantially; seven studies had high risk of bias. The pooled completion rate across 19 studies was approximately 56%. [1]

A standardised difference puts results from different questionnaires on a comparable scale. It describes a gap between groups, rather than the proportion of people cured. For a development decision, it needs to inform a concrete goal: what change would people actually notice? A statistical difference provides direction; the benchmark for a new service should also reflect the everyday lives of its intended users.

02

Meaningful benefits extend into the day

DIALS compared digital CBT-I with sleep information, both alongside usual care. At 24 weeks, self-reports showed small benefits in functional health and wellbeing, and larger benefits in sleep-related quality of life. Attrition and a predominantly female, white sample limit generalisability. [2]

In DISCO, people with insomnia and cognitive complaints reported less impairment after digital CBT-I than a waiting-list group. Benefits persisted to week 24. Objective cognitive tests showed no between-group difference. Feeling less impaired therefore did not establish improved memory performance. [3]

This suggests a practical decision for service development: the promised benefit should determine the appropriate measurement before the first prototype is built. If the aim is to make everyday life easier, ask about that directly. A claim about better performance requires corresponding performance measures. An attractive presentation of sleep data can support this decision, but cannot replace it.

03

A prevention opportunity for a defined population

In SPREAD, modelled one-year incidence of moderate-to-severe depressive symptoms among initially minimally affected participants was 9.6% with digital CBT-I versus 18.8% with sleep information. High attrition limits interpretation; the outcome was a questionnaire threshold, not a new clinician-confirmed diagnosis. [4]

For a prevention strategy, this distinction is useful: a development goal needs to account for where people are starting. A service for people with existing sleep difficulties requires different communication, support and outcome monitoring from general information about recovery. Defining the intended population therefore also helps determine which professional expertise belongs in the development process.

04

Actual use needs its own development work

The older-adult trial compared adapted digital CBT-I, with or without stepped support, against sleep information. Both treatment groups had fewer insomnia symptoms through 12 months. Overall, 63% completed the programme within nine weeks, rising to 83% with later completions. The sample had limited diversity. [5]

A development process should therefore distinguish between getting started, sustained use and completion. Finishing later can mean something different from dropping out permanently. Feedback on clarity, time demands and practical barriers should sit alongside changes in health. Together, these measures can help establish whether a concept makes sense to people and fits the lives of its intended users.

When evaluating such a service, we would also record the support people actually use. Technical assistance, encouragement and therapeutic care should remain distinguishable in the evaluation. Otherwise, it becomes difficult to decide which skills are needed as the service develops. People who leave also deserve a voice: their feedback may reveal issues that an evaluation of satisfied participants would miss.