Three-month follow-up after discontinuation of steam-assisted respiratory muscle training in mild-to-moderate obstructive sleep apnea: an exploratory polysomnographic study.

PURPOSE: To explore polysomnographic findings three months after discontinuation of steam-assisted respiratory muscle training (RMT) in mild-to-moderate obstructive sleep apnea.

METHODS

This exploratory secondary follow-up included 33 of 48 intervention entrants who completed polysomnography at baseline, after 12 weeks of training, and three months after cessation. The primary analysis used repeated-measures ANOVA (RM-ANOVA). Distributional diagnostics and exact Friedman sensitivity analyses were conducted for eight sleep-stage and periodic limb movement outcomes.

RESULTS

Conventional respiratory-event indices showed no overall time effect (all p ≥ 0.3). Flow limitation differed across visits (p = 0.02), without an immediate post-intervention reduction. The RM-ANOVA identified visit differences in REM latency (169.9 ± 74.9 min at baseline, 138.8 ± 66.4 min post-intervention, and 134.0 ± 66.5 min at follow-up; p = 0.003), N1 proportion (6.6 ± 4.0%, 10.8 ± 5.9%, and 11.2 ± 6.1%, respectively; p = 0.002), and PLMS-related arousal index (1.3 ± 2.1/h, 0.6 ± 1.1/h, and 0.7 ± 1.4/h, respectively; p = 0.01) and count (9.3 ± 15.2, 4.3 ± 7.9, and 5.2 ± 10.0, respectively; p = 0.02). Exact Friedman analyses yielded visit differences for N1 (p = 0.002) and the PLMS-related arousal index and count (p = 0.01 and 0.02), but not REM latency (p = 0.1).

CONCLUSION

This exploratory follow-up identified differences in N1 proportion and PLMS-related arousal measures across visits, supported by sensitivity analyses, while no overall visit differences were detected in conventional respiratory-event indices. REM latency differed significantly across visits in RM-ANOVA, but not in the sensitivity analysis. These observations warrant further investigation in controlled studies; the single-arm design and absence of an adaptation night prevent attribution to steam-assisted RMT or confirmation of lasting clinical benefit.

CLINICALTRIALS: GOV: NCT05320952.