Feasibility of home sleep apnea testing with NightOwl and SleepImage in patients with type 2 diabetes: an external pilot study

Abstract Background Obstructive sleep apnea (OSA) is more prevalent and substantially underdiagnosed in patients with type 2 diabetes (T2DM), whose regular follow-up offers an opportunity for OSA screening. The current diagnostic approach requires resources not available in that setting, whereas the less complex and less obtrusive wearables NightOwl and SleepImage are promising alternatives. Their diagnostic accuracy must be established in patients with T2DM, in whom cardiac autonomic neuropathy may affect the signals acquired by both devices. This external pilot feasibility study aimed to assess device and study feasibility prior to a diagnostic accuracy study. Methods Adults with T2DM in Denmark were recruited through a tertiary diabetes outpatient clinic, general practice, and Facebook. In this single-arm, non-randomized study, participants underwent home sleep apnea testing on three consecutive nights with the two wearables NightOwl and SleepImage, with the cardio-respiratory monitor Nox T3 added on the third night as a comparator. Failed test nights were repeated. Device feasibility comprised the proportion of analyzable recordings per device and participants' experience with the two wearables, assessed by a questionnaire using visual analog scale (VAS) ranging from 0 to 100 and categorical items. Study feasibility comprised the inclusion rate, recruitment source, and measures of completion. All outcomes were analyzed descriptively. Results Forty-three adults were included. Median age was 61 years, and 51% were male. Most were recruited through Facebook (79%), and the inclusion rate was 3.9 participants per month. Median VAS scores were 83 for NightOwl and 92 for SleepImage. Analyzable recordings were obtained on 75% to 85% of intended test nights across all three devices. Sixty-seven percent completed all planned test nights after a median of three attempted nights, whereas non-completers attempted a median of six nights. Moderate-to-severe OSA was found in 56% of participants with an analyzable Nox T3 recording. Conclusions Participants reported a favorable experience with both NightOwl and SleepImage, and the main diagnostic accuracy study will proceed with active recruitment added. Its design should account for a technical success rate of roughly three in four recordings and a completion rate of two in three. Trial registrations Registered at ClinicalTrials.gov (NCT07778121) on 14 January 2026. Retrospectively registered during recruitment.