| Characteristic | Baseline | Follow-up | ||||
| Adherent | Non-user | p-value# | Adherent | Non-user | p-value¶ | |
| Subjects n | 199 | 118 | 199 | 118 | ||
| Age years | 51.8±10.4 | 52.8±10.0 | 0.996 | |||
| Male | 81.9 | 80.5 | 0.967 | |||
| BMI kg·m−2 | 33.9±5.9 | 33.1±5.9 | 0.975 | 34.9±6.1 | 32.9±5.4 | <0.0001 |
| Current smoker | 20.6 | 20.3 | 0.897 | 18.7 | 28.0 | 0.004 |
| Excessive alcohol | 2.5 | 2.5 | 0.954 | 1.5 | 2.5 | 0.342 |
| Hypertension | 33.2 | 28.8 | 0.892 | 33.8 | 30.5 | 0.953 |
| Cardiovascular disease | 2.5 | 1.7 | 0.949 | 2.5 | 1.7 | 0.688 |
| Diabetes mellitus | 1.5 | 0.9 | 0.987 | 3.1 | 3.5 | 0.460 |
| Participate in exercise | 58.8 | 58.4 | 0.988 | 67.9 | 66.7 | 0.726 |
| Epworth Sleepiness Scale | 12.0±5.0 | 11.5±4.8 | 0.966 | 7.8±4.6 | 9.3±4.9 | 0.002 |
| AHI events·h−1 | 44.8±19.6 | 40.8±19.8 | 0.931 | |||
| ODI events·h−1 | 36.1±19.3 | 32.1±19.0 | 0.990 | |||
| SaO2 nadir | 76.2±7.5 | 77.5±7.3 | 0.993 | |||
| Percentage of time at SaO2<90% | 2.1±1.0 | 1.8±1.0 | 0.942 | |||
| Total cholesterol mg·dL−1 | 204.7±41.6 | 204.9±35.4 | 0.917 | 211.4±41.6 | 213.3±36.0 | 0.528 |
| Total cholesterol ≥200+ | 56.3 | 55.1 | 0.727 | 64.3 | 63.6 | 0.771 |
| LDL cholesterol mg·dL−1 | 149.4±37.3 | 149.7±33.8 | 0.933 | 149.5±30.9 | 151.3±33.3 | 0.475 |
| LDL cholesterol ≥130+ | 70.9 | 72.9 | 0.905 | 75.9 | 72.9 | 0.475 |
| HDL cholesterol mg·dL−1 | 40.0±12.4 | 39.8±10.3 | 0.922 | 46.6±12.8 | 47.3±12.8 | 0.619 |
| HDL cholesterol <40+ | 60.8 | 57.6 | 0.978 | 30.6 | 34.8 | 0.144 |
| Triglycerides mg·dL−1 | 174.8±86.1 | 176.0±102.9 | 0.998 | 174.6±88.9 | 167.9±96.1 | 0.517 |
| Triglycerides ≥150+ | 59.8 | 55.1 | 0.737 | 54.8 | 49.2 | 0.544 |
Data are presented as mean±sd or %, unless otherwise stated. Significant differences are shown in bold. BMI: body mass index; AHI: apnoea/hypopnoea index; ODI: oxygen desaturation index; SaO2: arterial oxygen saturation; LDL: low-density lipoprotein; HDL: high-density lipoprotein. #: adjusted for propensity score subclass from ANCOVA model or conditional logistic regression; ¶: from ANCOVA or conditional logistic examining the difference in follow-up values, adjusted for propensity score subclass and baseline value; +: based on the National Cholesterol Education Programme (Adult Treatment Panel III) published criteria [28].