1077 Optimizing Message Delivery in Tele-monitoring to Patients at 90 Days: The Tele-OSA Randomized Clinical Trial

Abstract

Introduction: The Tele-OSA study previously reported the impact of CPAP tele-monitoring(TM) with automated patient feedback[U-Sleep; ResMed Corp] on improving 3-month CPAP adherence. We analyzed daily usage patterns to determine patient response to received messages. Methods: Tele-OSA was a 4-arm randomized trial conducted at Kaiser Permanente sleep center(Fontana, CA) studying patients with suspected OSA. The primary outcome(3-month CPAP adherence) revealed improved use with TM but no significant impact with telemedicine- delivered education. TM involved automatic wireless upload of CPAP data followed by automated patient messages(text, email, and/or phone) if CPAP use decreased below a defined threshold or for achieving Medicare compliance. We randomly selected 28 patients with severe OSA that received both TM and tele-education, evaluated daily usage, and tagged specific days messages were delivered. Results: 28 subjects(75.0% male, 50.5 ± 12.0 years, AHI 49.6 ± 19.8, CPAP 90-day usage 285.0 ± 152.0 minutes, 9.2 ± 7.7 messages sent). 10 had “high” adherence(≥70% days with ≥4 hours usage); 11 had “mid” adherence(≥40 to <70%); 7 had “low” adherence(<40%). The “High” group(AHI 53.8 ± 22.5, CPAP usage 456.9 ± 43.5 minutes, 3.2 ± 0.6 messages) showed no response to messaging. CPAP use the night after receiving a message compared to non-messaging nights increased 48%±25% vs. 48 ± 3; stayed same 0%±0% vs. 1%±1%; decreased 52%±25% vs. 51%±3%(All p=NS). The “mid” group(AHI 46.1 ± 15.2, usage 249.1 ± 47.3 minutes, 7.5 ± 2.1 messages) had positive response to messaging. Messaging night usage increased 66%±24% vs. 43%±6%(p=0.02); same 8%±12% vs. 5.6%±4%(p=0.50); decreased 26%±27% vs. 51%±4%(p=0.02). The “low “ group(AHI 49.3 ± 23.8, usage 95.9 ± 63.9 minutes, 20.6 ± 6.4 messages) did not increase use in response to messages( 28%±22% vs. 29%±16%; p=0.91). Rather, patients were more likely to maintain same usage(52 ± 32% vs. 38%±30%; p<0.01) and less likely to decrease(20%±19% vs. 34%±15%; p=0.10). When stratifying message type in the “mid” group, negative messages( for low adherence) increased use(72%±19% vs. 43%±4%; p<0.01) while positive messaging(achieving compliance) had no clear effect(- decreased 75%±42% vs. 51%±4%, p=0.22). Conclusion: Automated patient messaging based on CPAP use results in immediate improvement in CPAP use, primarily in those with moderate adherence. Findings may help guide future messaging protocols.