Partial failure of CPAP treatment for sleep apnoea: Analysis of the French national sleep database

Sébastien Bailly, Najeh Daabek, Ingrid Jullian‐Desayes, Marie Joyeux‐Faure, Marc Sapène, Y. Grillet, Jean‐Christian Borel, Renaud Tamisier, Jean‐Louis Pépin

Respirology · 2019 · 32 citations · 15 references

Concepts

Abstract

ABSTRACT Background and objective Continuous positive airway pressure (CPAP) is the first‐line therapy for obstructive sleep apnoea (OSA). Residual apnoea and/or hypopnoea events, that is an apnoea–hypopnoea index (AHI) > 5, during CPAP contribute to treatment drop‐out. The clinical scenarios triggering residual events during CPAP use are poorly described. Underlying co‐morbidities, especially cardiovascular diseases, lifestyle factors, OSA characteristics at diagnosis and type of mask have been suggested as potential contributors. Methods Patients from the prospective French sleep apnoea registry diagnosed with OSA (AHI ≥ 15 events/h) treated with CPAP were included. Logistic regression analysis identified factors associated with a risk of residual AHI > 5 events/h on CPAP. Results The 12 285 OSA patients were predominantly men ( n = 8715, 70.9%), middle‐aged (58.2 (49.8; 66.1) years) and obese (median body mass index: 31.3 (27.7; 35.6) kg/m 2 ). Most had an AHI ≤ 5 events/h ( n = 9573, 77.9%) versus 22.1% with AHI > 5/h. The latter were less CPAP adherent (5.75 (4.01; 7.00) vs 6.00 (4.53; 7.00) h/night). In multivariable analysis, factors associated with residual AHI >5/h were male sex, age, sedentary lifestyle, OSA severity, cardiovascular co‐morbidities (heart failure and arrhythmia) and type of interface (orofacial mask versus nasal mask: OR = 2.15 (95%CI: 1.95; 2.37)). A subgroup analysis found that patients using pressures above 10 cm H 2 O were 1.43 (95% CI: 1.3; 1.57) times more likely to have residual AHI > 5/h. Conclusion Knowing about risk factors for residual apnoeic–hypopnoeic events may assist in the timely provision of personalized care including the type of PAP therapy, attention to co‐morbidities and choice of interface.

References

15