Study Population and Baseline Demographics
The recruitment and follow-up of patients is shown in Figure 1. A total of 901 patients were screened and consented, of which 158 patients were eligible for the study based on the inclusion criteria. Of those eligible, 89.9% of patients had a pre-existing CPAP prescription (Figure 1). Among those with a CPAP prescription, ten patient dropped out of the study due to withdrawal from the study or surgery cancellation.
The baseline characteristics of CPAP adherent and non-adherent patients are presented in Table 1. Among the 132 total participants who completed the study, 88 (62.9%) were female with an average age of 51 ± 12 years, and a body mass index (BMI) of 44.7 ± 12.5 kg/m2. CPAP adherent patients had significantly higher BMI (47 ± 10 vs 41 ± 11 kg/m2, p < 0.001) and greater baseline AHI before CPAP therapy (41.1 (IQR 22.7, 77.0) vs 22.7 (IQR 14.2, 37.0) events/h, p = 0.002) than the CPAP non-adherent patients. There was a greater proportion of patients with severe OSA (AHI ≥ 30 events/h) in the CPAP-adherent vs non-adherent group (54.3 vs 25.0%, p = 0.010). A higher proportion of CPAP-adherent patients underwent bariatric surgery (87.9% vs 61%) and a lower proportion of patients underwent orthopedic (4.5% vs. 24.4%) and spine surgery (14.6 vs 6.1%) (Table 1). There were no significant differences between CPAP adherent and non-adherent patients in ASA physical status, medical history, 48h opioid consumption, and types of anesthesia.
Table 1. Baseline demographic data
Characteristics
|
Adherent
(N=76)
|
Non-adherent
(N=56)
|
P Value
|
Age (years)
|
51 ± 11
|
51 ± 15
|
0.818
|
Sex
Male
Female
|
|
|
0.280
|
23 (30.3%)
|
22 (39.3%)
|
|
53 (69.7%)
|
34 (60.7%)
|
|
BMI (kg/m2)
|
47 ± 10
|
43 ± 11
|
0.017
|
ASA Physical Status
|
|
|
0.594
|
2
|
4 (5.3%)
|
6 (10.7%)
|
|
3
|
68 (89.5%)
|
48 (85.7%)
|
|
4
|
4 (5.3%)
|
2 (3.6%)
|
|
Medical History
|
|
|
|
Cardiovascular Disease*
|
40 (52.6%)
|
30 (53.6%)
|
0.915
|
Diabetes
|
24 (31.6%)
|
21 (37.5%)
|
0.478
|
GERD
|
34 (44.7%)
|
23 (41.1%)
|
0.674
|
Smoker
|
29 (38.2%)
|
15 (26.8%)
|
0.171
|
Asthma/COPD
|
19 (25.0%)
|
13 (23.2%)
|
0.813
|
Hypothyroidism
|
10 (13.2%)
|
8 (14.3%)
|
0.852
|
Arthritis
|
33 (43.4%)
|
21 (37.5%)
|
0.494
|
AHI†, (events/h)
|
36.1 (22.7, 74.0)
|
24.4 (16.2, 38.7)
|
0.031
|
OSA Severity
|
|
|
0.051
|
Mild
|
9 (14.5%)
|
11 (25.6%)
|
|
Moderate
|
18 (29.0%)
|
18 (41.9%)
|
|
Severe
|
35 (56.5%)
|
14 (32.6%)
|
|
Type of Surgery
|
|
|
0.081
|
General
|
8 (10.5%)
|
6 (10.7%)
|
|
Bariatric
|
54 (71.1%)
|
29 (51.8%)
|
|
Orthopedic
|
5 (6.6%)
|
8 (14.3%)
|
|
Spine
|
8 (10.5%)
|
11 (19.6%)
|
|
Urology
|
1 (1.3%)
|
--
|
|
Brain
|
--
|
2 (3.6%)
|
|
Type of Anesthesia
|
|
|
0.441
|
General
|
59 (95.2%)
|
39 (90.7%)
|
|
Spinal/Regional
|
3 (4.8%)
|
4 (9.3%)
|
|
48h opioid consumption (mg) ‡
|
145.0 (90.0, 195.0)
|
105.6 (61.0, 186.3)
|
0.456
|
Supplemental Oxygen Therapy
|
|
|
|
Postoperative Night 1
|
6/61 (9.8%)
|
20/43 (46.5%)
|
<.001
|
Postoperative Night 2
|
1/59 (1.7%)
|
2/42 (4.8%)
|
0.569
|
Data are represented as mean ± SD or median (IQR), or as otherwise indicated. OSA severity defined as mild (5 £ AHI < 15), moderate (15 £ AHI < 30), or severe (AHI ≥ 30) *Cardiovascular disease includes hypertension, angina, myocardial infraction, heart failure, peripheral vascular disease, valvular disease, stroke, coronary revascularization, atrial fibrillation, ventricular tachycardia, supraventricular tachycardia, ventricular premature beats, atrioventricular block, and cardiomyopathy. †Baseline AHI prior to CPAP therapy. ‡Opioid consumption was reported as oral morphine equivalents (mg). ASA = American Society of Anesthesiologists; BMI = body mass index; COPD = chronic obstructive pulmonary disease. CPAP = continuous positivity airway pressure; GERD = gastroesophageal reflux disease. IQR =interquartile range. Two sample independent t-tests or Wilcoxon rank-sum test and chi-squared test, or Fisher’s exact tests were conducted to examine differences in baseline characteristics between adherent and non-adherent patients. Adherence is defined as an average CPAP use ≥ 4h/night.
Perioperative CPAP adherence
Among patients with CPAP prescription, 61% were adherent preoperatively (Figure 2A). In the postoperative period, on N1, 57.6% were adherent; and on N2, 59.0% were adherent.
Among 132 subjects who completed the study, 64 patients (48.5%) were consistently adherent to CPAP, defined as CPAP usage 4h on all pre and postoperative nights (Table 2). Forty-seven patients (35.6%) were consistently non-adherent with CPAP usage < 4 h on all pre- and postoperative nights. Nineteen patients (14.4%) demonstrated partial non-adherence, which was defined as CPAP usage < 4 h on 1 or more nights. Fourteen patients (10.6%) who were preoperatively adherent demonstrated postoperative partial non-adherence. Six (4.5%) patients who were previously non-adherent preoperatively were placed on CPAP postoperatively on one or more nights due to oxygen desaturation by the health care team. Details of the longitudinal perioperative CPAP adherence are further illustrated in Table 2.
Table 2. Longitudinal CPAP adherence patterns by perioperative stage
Pre-op
|
Post-op Night 1
|
Post-op Night 2
|
Number
|
Percent
|
Trend over time
|
Overall pattern
|
Yes
|
Yes
|
Yes
|
56
|
42.4
|
Adherent
|
Consistently adherent
|
Yes
|
Yes
|
n/a
|
8
|
6.1
|
Adherent
|
Consistently adherent
|
No
|
No
|
No
|
30
|
22.7
|
Non-adherent
|
Consistently non-adherent
|
No
|
No
|
n/a
|
17
|
12.9
|
Non-adherent
|
Consistently non-adherent
|
Yes
|
Yes
|
No
|
6
|
4.5
|
Switched to non-adherence on N2
|
Partially non-adherent
|
Yes
|
No
|
No
|
5
|
3.8
|
Switched to non-adherence on N1
|
Partially non-adherent
|
Yes
|
No
|
Yes
|
3
|
2.3
|
Switched on N1 and N2
|
Partially non-adherent
|
No
|
Yes
|
Yes
|
3
|
2.3
|
Switched to adherence N1
|
Partially non-adherent
|
No
|
Yes
|
No
|
2
|
1.5
|
Switched on N1 and N2
|
Partially non-adherent
|
Yes
|
No
|
n/a
|
1
|
0.8
|
Switched on N1
|
Partially non-adherent
|
No
|
Yes
|
n/a
|
1
|
0.8
|
Switched on N1
|
Partially non-adherent
|
|
|
Total
|
132
|
100.0
|
|
|
Abbreviations: n/a = not available due to discharge from hospital. 64 (48.4 %) patients: consistently adherent; 47 (35.6%): consistently nonadherent; 21 (15.9%) partially non-adherent.
Perioperative Overnight Oximetry
Unadjusted, cross-sectional analysis comparing CPAP adherence vs non-adherence is presented in Supplemental Digital Content 2. Overall, CPAP adherence vs. non-adherence was associated with significantly higher preoperative minimum SpO2 (83 vs 79%, p = 0.001), lower ODI (4.3 vs 11.8 events/h, p < 0.001), and lower CT90 (0.5 vs 3.6%, p < 0.001) (Supplemental Digital Content 2).
Perioperative oxygen saturation parameters between CPAP adherent and non-adherent patients were analyzed using a linear regression fixed effects model, which only utilizes within-subject variation, and considers CPAP adherence as a time-varying covariate (Table 3). At the preoperative baseline, CPAP adherence versus non-adherence was associated with significantly lower ODI (6.71 [95% CI 2.60-10.83) vs 18.51 [95% CI 13.57-23.45] events/h, p = 0.0011), but no significant differences in mean SpO2 (93.5% vs 92.3%, p = 0.16), minimum SpO2 (79.5 vs. 78.8%, P = 0.88) or CT90 (5.61 vs 10.68%, p = 0.49) (Figure 2B).
Table 3. Linear regression for perioperative overnight oximetry using a linear fixed effects model
Variable
|
Parameter estimate
|
SEM
|
T-statistic
|
P-value
|
Mean SpO2
|
|
|
|
|
Intercept
|
92.98
|
1.6
|
58.17
|
< 0.0001
|
CPAP Adherence
|
0.54
|
0.78
|
0.69
|
0.49
|
N1 vs. Pre-op
|
1.13
|
0.495
|
2.27
|
0.025
|
N2 vs. Pre-op
|
1.12
|
0.56
|
3.01
|
0.076
|
O2 therapy
|
1.69
|
0.561
|
3.01
|
0.003
|
Minimum SpO2
|
|
|
|
|
Intercept
|
84.45
|
9.6
|
8.8
|
< 0.0001
|
CPAP Adherence
|
2.15
|
4.7
|
0.46
|
0.65
|
N1 vs. Pre-op
|
-2.04
|
2.98
|
-0.69
|
0.49
|
N2 vs. Pre-op
|
-7.45
|
3.76
|
-1.98
|
0.0497
|
O2 therapy
|
-6.17
|
3.37
|
-1.83
|
0.070
|
ODI
|
|
|
|
|
Intercept
|
18.75
|
6.63
|
2.83
|
0.0055
|
CPAP Adherence
|
-11.8
|
3.55
|
-3.33
|
0.0011
|
N1 vs. Pre-op
|
-7.96
|
2.32
|
-3.44
|
0.0008
|
N2 vs. Pre-op
|
-4.51
|
3.49
|
-1.29
|
0.19
|
CPAP adherence × N1
|
9.14
|
2.81
|
3.25
|
0.0015
|
CPAP adherence × N2
|
4.4
|
2.88
|
1.53
|
0.13
|
O2 therapy
|
-0.78
|
2.56
|
-0.3
|
0.76
|
CT90
|
|
|
|
|
Intercept
|
2.95
|
13.53
|
0.22
|
0.83
|
CPAP Adherence
|
-0.35
|
6.62
|
-0.05
|
0.96
|
N1 vs. Pre-op
|
0.22
|
4.22
|
0.05
|
0.96
|
N2 vs. Pre-op
|
0.083
|
5.3
|
0.02
|
0.99
|
O2 therapy
|
-4.53
|
4.75
|
-0.95
|
0.34
|
Statistical analysis using linear fixed effects model where the time-varying covariate CPAP adherence only utilizes within-subject variation. By definition the time invariant covariates are assumed stable over time, and thus are removed from consideration. Abbreviations: CPAP = continuous positive airway pressure, ODI = oxygen desaturation index, CT90 = cumulative time percentage with SpO2 < 90%, N1 = postoperative night 1, N2 = postoperative night 2, SEM = standard error of the mean.
Based on the linear fixed effects regression model, for the parameter ODI we observed a statistically significant interaction between CPAP adherence and N1 (p = 0.0015) but not N2, which suggests differential effects of CPAP adherence in the postoperative period. For the parameters, mean SpO2, minimum SpO2, and CT90, we performed a test of the interaction between compliance and night [i.e. main effects compliance, night, compliance × night interaction and O2 therapy) on 2 degrees of freedom]. There were no significant statistical interactions between CPAP adherence and time for mean SpO2, minimum SpO2, or CT90.This led us to consider the simpler main effects model listed shown in Table 3. Adjusted mean values for ODI, mean SpO2, minimum SpO2, and CT90 are shown in Supplemental Digital Content 3.
Postoperative Oxygen Therapy
On N1, use of supplemental oxygen therapy was much higher in the CPAP non-adherent group vs adherent group (46.5% vs 9.8%, p < 0.001) (Table 1). On N2, no significant differences in supplemental oxygen therapy occurred between the two groups (4.8% vs 1.7%, p = 0.557) (Table 1). Supplemental O2 therapy on any postoperative night predicted an increase in mean SpO2 by 1.69 0.56% (p = 0.003), but no significant differences for minimum SpO2, ODI, or CT90 (Table 3). Since oxygen therapy at baseline was an exclusion criterion, no patients received supplemental O2 therapy preoperatively.
Postoperative complications
There were no significant differences in total postoperative adverse events among CPAP adherent vs non-adherent patients (38% vs 41%) (Supplemental Digital Content 4). No significant differences were observed in cardiovascular, respiratory, or gastrointestinal events. Interestingly, there was a significantly higher proportion of CPAP non-adherent vs adherent patients with inadequate pain control (25.9% vs 11.1%, p = 0.03) (Supplemental Digital Content 4).