Cohort characteristics | Non-OSA (unweighted) | PAP group (unweighted) | Standardised difference (unweighted comparison) | Non-OSA (ATT weighted)* | Standardised difference (comparison on an ATT weighted sample) |
N=4 588 200 | N=324 029 | N=330 429 | |||
Demographics at the index date | |||||
Age, median (IQR) | 47 (33–61) | 58 (49–67) | 0.67 | 57.68 | 0.005 |
Sex, male, n (%) | 2 368 385 (51.6) | 211 379 (65.2) | 0.28 | 65.64 | 0.008 |
Rural status: yes, n (%) | 547 452 (11.9) | 40 449 (12.5) | 0.02 | 12.76 | 0.008 |
Neighbourhood income, n (%) | |||||
Quintile 1 | 883 936 (19.3) | 54 792 (16.9) | 0.06 | 16.95 | 0.004 |
Quintile 2 | 898 281 (19.6) | 62 000 (19.1) | 0.01 | 19.04 | 0.002 |
Quintile 3 | 918 096 (20.0) | 66 746 (20.6) | 0.01 | 20.78 | 0.004 |
Quintile 4 | 924 562 (20.2) | 69 379 (21.4) | 0.03 | 21.59 | 0.004 |
Quintile 5 | 949 776 (20.7) | 70 444 (21.7) | 0.03 | 21.65 | 0.002 |
Comorbidities, n (%) | |||||
Prevalent conditions | |||||
Diabetes | 408 683 (8.9) | 96 277 (29.7) | 0.55 | 30.36 | 0.014 |
Hypertension | 899 553 (19.6) | 178 511 (55.1) | 0.79 | 56.03 | 0.019 |
CHF | 42 050 (0.9) | 20 756 (6.4) | 0.30 | 6.49 | 0.004 |
Asthma | 394 682 (8.6) | 67 988 (21.0) | 0.35 | 21.1 | 0.003 |
COPD | 184 450 (4.0) | 56 689 (17.5) | 0.45 | 17.74 | 0.007 |
Immunocompromising conditions | 82 839 (1.8) | 14 537 (4.5) | 0.15 | 4.57 | 0.004 |
Cancer | 219 899 (4.8) | 31 551 (9.7) | 0.19 | 9.93 | 0.006 |
In the last 2 years | |||||
CCI score | |||||
High (≥3) | 13 577 (0.3) | 5387 (1.7) | 0.14 | 1.62 | 0.003 |
Moderate (2) | 19 522 (0.4) | 7020 (2.2) | 0.15 | 2.16 | 0.000 |
Low (1) | 24 926 (0.5) | 8580 (2.6) | 0.17 | 2.81 | 0.010 |
None (0) | 240 723 (5.2) | 30 429 (9.4) | 0.16 | 9.86 | 0.016 |
Non-psychotic mood or anxiety disorders | 476 419 (10.4) | 83 515 (25.8) | 0.41 | 27.01 | 0.028 |
In the past 5 years | |||||
Any CV hospitalisation | 142 505 (3.1) | 49 293 (15.2) | 0.43 | 16.01 | 0.022 |
Prior end-stage renal disease/haemodialysis | 8549 (0.2) | 3480 (1.1) | 0.11 | 1.04 | 0.003 |
Neuromuscular disease | 75 316 (1.6) | 16 382 (5.1) | 0.19 | 5.25 | 0.009 |
Alcohol dependence/intoxication | 78 073 (1.7) | 6672 (2.1) | 0.03 | 2.12 | 0.004 |
Obesity/bariatric surgery | 1425 (0.0) | 5786 (1.8) | 0.19 | 2.28 | 0.035 |
Cardiometabolic morbidity (prevalent diabetes, hypertension or CHF, or hospitalisations for CV conditions in the last 5 years) | 1 108 647 (24.2) | 209 593 (64.7) | 0.89 | ||
Chronic airway disease (COPD or asthma) | 550 303 (12.0) | 103 729 (32.0) | 0.50 |
Individuals with a low probability of OSA (control group) are presented as unweighted (original) and weighted† on the propensity score.
*Estimates presented as mean or prevalence (percentage) as applicable.
†In weight allocation using the ATT approach (used in the main analysis), the exposure group has weight 1, and only the controlled group is weighted. Weight allocation using the average treatment effect approach (used in sensitivity analysis) is presented in online supplemental table E3.
ATT, average treatment effect on the treated; CCI, Charlson Comorbidity Index; CHF, chronic heart failure; COPD, chronic obstructive pulmonary disease; CV, cardiovascular; IQR, interquartile range; OSA, obstructive sleep apnoea; PAP, positive airway pressure.