Abstract
Study Objectives
The main objective of this study was to evaluate the role of sham continuous positive airway pressure (CPAP) compared to conservative medical therapy (CMT) as a control arm in the Best Apnea Interventions for Research (BestAIR) study by assessing differences in subjectively and objectively measured outcomes, adverse events, adherence, and retention rates.
Methods
BestAIR is a clinical trial aimed to identify important design features for future randomized controlled trials of CPAP. Participants with obstructive sleep apnea were randomized to one of four groups; two control arms (CMT, sham-CPAP) and two active CPAP arms (with and without behavioral interventions). Blood pressure and health-related quality of life outcomes were assessed at baseline, 6 and 12 months. Study outcomes, retention, and adverse event rates were compared between the two control arms. Sham-CPAP adherence and self-efficacy were also compared to active-CPAP adherence (without behavioral intervention).
Results
Our sample included 86 individuals in the control arms and 42 participants in the active-CPAP arm. There were no differences in longitudinal profiles in blood pressure, health-related quality of life outcomes, dropout rates, or adverse events in sham-CPAP group compared to CMT-only group (all ps > 0.05); standardized differences were generally small and with inconsistent directionality across measurements. When compared to active-CPAP, sham-CPAP was associated with 93 fewer minutes/night of usage over 12 months (p = 0.007) and lower outcome expectations (p < 0.05).
Conclusion
We observed no evidence of differences in objectively or subjectively measured outcomes with the use of sham-CPAP compared to CMT group. The lower adherence on sham-CPAP and poorer self-efficacy compared to active-CPAP may suggest differences in perceived benefit.
Registration
NCT 01261390 Best Apnea Interventions for Research (BestAIR) www.clinicaltrials.gov
Keywords: sleep apnea, CPAP, sham, placebo, adherence, clinical trial
Statement of Significance.
Unlike medications, identifying optimal placebos for use in randomized controlled trials for devices is problematic. Although sham continuous positive airway pressure (CPAP) is often used, it can be burdensome and may negatively affect sleep and quality of life, creating uncertainty in its role in clinical trials. This article investigates the role of sham-CPAP (placebo) compared to conservative medical therapy as a control arm in a randomized controlled trial by assessing the differences in study outcomes, adverse events, adherence, and retention rates. This study finds no evidence of differences in objectively or subjectively measured outcomes between the control arms. These findings support the consideration of alternatives to sham-CPAP in long-term studies where participation and study burden are a concern and end points can be blindly measured.
Introduction
A continuous positive airway pressure (CPAP) device is considered the gold standard treatment for individuals with moderate-to-severe obstructive sleep apnea (OSA) [1, 2]. Several randomized controlled trials (RCTs) have demonstrated effectiveness of CPAP as measured by improvements in sleepiness, blood pressure, and overall quality of life [3–7]. Despite these results supporting the utility of CPAP for a range of intermediate and symptom-based outcomes, there is a paucity of data from RCTs that address long-term clinical outcomes. One challenge to conducting these large-scale trials is implementation of a rigorous and blinded control condition. Although a “placebo” control minimizes biases [8] in evaluating the CPAP effect, there are practical difficulties in finding an ideal placebo for non-pharmaceutical treatments such as CPAP.
Although a few interventions, such as nasal dilator strips [9], have been considered an attractive placebo, sham-CPAP has been advocated as a placebo that most closely mimics active therapy. Some studies have reported sham to be a useful placebo intervention based on the potential for blinding participants [10, 11]. However, sham-CPAP users may be susceptible to discomfort and possible sleep fragmentation resulting from wearing a mask that provides nontherapeutic pressure, biasing the results of a trial by overestimating the true treatment effect. Suboptimal responses to CPAP may lead to low levels of use, as shown in some studies that reported lower adherence of sham-CPAP compared to active-CPAP [12–15], and thus may negatively affect the adoption and use of active-CPAP after trial termination.
The literature is sparse when assessing the placebo effects of sham-CPAP use in an RCT. To the best of our knowledge, there have been no studies analyzing study outcomes, adverse events (AEs), or retention rates of individuals randomized to sham-CPAP versus alternative control interventions. The purpose of this study was to evaluate the role of sham-CPAP compared to conservative medical therapy (CMT) as determined by differences in objectively and subjectively measured outcomes, AEs, and retention rates between the control conditions. We also compare adherence rates and self-efficacy scores between sham-CPAP and active CPAP treatment. Analyses were conducted using data from the Best Apnea Interventions for Research (BestAIR) study, a trial that explicitly aimed to identify the design elements useful for subsequent larger clinical trials using CPAP. We hypothesized that due to the prospective objective measurement of blood pressure, there would be no differences in blood pressure comparing the blinded (sham-CPAP) versus unblinded (CMT) control arm. In addition, we hypothesized that subjectively reported outcomes would differ in the blinded (sham-CPAP) versus the unblinded (CMT) control arm and that sham-CPAP would adversely affect sleep quality compared to CMT.
Methods
Study design
The methods of the BestAIR study have been published previously [16–18]. This study was a parallel-group, partially double-blinded, RCT. Eligible participants recruited from outpatient clinics had an apnea–hypopnea index (AHI) 4% at least 10 events/hour or AHI 3% at least 15 events/hour and were either aged 45–75 years with established cardiovascular disease (CVD) or cardiometabolic disease (coronary artery disease, prior myocardial infarction, coronary artery revascularization procedure, ischemic stroke, and/or diabetes) or were aged 55–75 years with three or more CVD risk factors (male, body mass index ≥ 30 kg/m2, hypertension, dyslipidemia, and/or ≥10 pack-years of smoking). Details of the study design and eligibility criteria are further described elsewhere [16–18]. Before randomization, participants underwent a 2-week run-in period when they were asked to wear a CPAP nasal mask overnight that was open to room air (unconnected to a machine). Individuals who reported wearing the mask on 11 of the 14 nights were eligible for randomization.
To evaluate the relative benefits and limitations of different control and active conditions, participants were randomized into one of four interventions (two control and two active interventions). To evaluate different control conditions, participants were randomized to CMT, which included education on sleep hygiene, sleep positional therapy, and nightly use of nasal dilator strips provided by the study; or to CMT (including nasal strips) plus sham-CPAP. To identify the relative impact of a behavioral intervention added to CPAP, participants were randomized to CMT plus active-CPAP delivered by a respiratory therapist or polysomnologist, or to CMT plus active-CPAP with motivational enhancement therapy delivered by a psychologist. Comparisons of the active arms and the role of motivational enhancement on CPAP adherence have been reported before [18, 19].
The study consisted of a baseline, 6-month, and 12-month evaluation and a phone call every 2 months to address any treatment-related adherence issues. However, participants randomized later in the study (n = 61) were followed only for 6 months.
Outcomes
The primary outcome was 24-hour blood pressure, measured using an automated oscillometric monitor (model 90217; Spacelabs Inc., Redmond, WA), programmed to measure blood pressure every 20 minutes during wakefulness and every 30 minutes during sleep. Measurements were made at baseline, 6-month, and 12-month follow-up visits.
The subjective outcomes included: health-related quality of life outcomes assessed by the Medical Outcomes Study Short Form-36 Questionnaire (SF-36) [20]; sleep quality assessed by the Women’s Health Initiative Insomnia Rating Scale [21]; sleepiness measured by the Epworth Sleepiness Scale [22, 23]; and risk perception of OSA, benefit of CPAP, and self-efficacy measured by the Self-Efficacy Measure for Sleep Apnea (SEMSA) [24].
Participants used either the REMStar Auto (Philips Respironics) or the S9 Autoset (ResMed) CPAP device. Sham-CPAP in Philips Respironics devices was set to reveal a pressure of 10 cm H2O, with a restrictor limiting outflow to 50 l/min and an intentional air leak to provide a peak pressure of approximately 2.4 cm H2O. ResMed sham-CPAP was delivered by restricting mask pressure at less than 1 cm H2O and increasing air leak and vent flow, producing a maximum mask pressure of 4 cm H2O. All CPAP devices were equipped with a modem to transmit nightly objective adherence data, based on run time, to secure servers, thus data collection was not contingent on attendance at visits.
Statistical analysis
Analyses were performed in R, version 3.4 [25], based on intention-to-treat assignment. Baseline characteristics between treatment arms were compared using Wilcoxon rank-sum tests and Fisher’s exact tests for continuous and categorical variables, respectively. To compare the longitudinal profiles of blood pressure and patient-reported outcomes between the sham-CPAP and CMT-only control groups, we implemented mixed-effects linear regression models using functions from the “lme4” package [26]. Observation time (0, 6, and 12 months) was modeled as a fixed effect categorical variable. Treatment effects reported are the averages of the fixed effect beta coefficients corresponding to treatment effects at 6 months and 12 months. We used similar models to compare SEMSA results over time between the sham-CPAP and active-CPAP groups (individuals who additionally received the Motivational Enhancement intervention were excluded from this analysis). Within-group effects, represented by linear functions of the coefficients for observation time (6 and 12 months) and the interaction term between observation time and treatment, were similarly estimated. Participants contribute data to whichever time points they were observed. Models included random intercepts to account for correlation over time and were adjusted for randomization stratification factors (recruitment site, CVD status, and diagnostic study type).
The two control groups were also compared on process outcomes. Dropout proportions were compared using the Fisher’s exact test. Counts of AEs were compared using Poisson regression, adjusting for observation time, and randomization stratification factors. Incidence rate ratios for minor, serious, and combined AEs comparing sham-CPAP to CMT-only groups were calculated by exponentiating the estimated coefficient from the Poisson regression. The estimates represent how many AEs were expected in the sham-CPAP arm for every 1 AE in the CMT-only arm. Because the observed variances were larger than the observed mean AE counts, suggesting the possibility of over-dispersion—violation of the Poisson distribution’s assumption of the mean and variance being equal, robust standard errors were calculated using the “sandwich” package [27] to protect against model misspecification. Time to the first AE was compared using a Cox proportional hazards model [28]. For the analyses of AE counts and time to first event, we also conducted sensitivity analyses adjusting for potential informative censoring (details in Supplementary material).
To further evaluate sham-CPAP as a placebo, we compared adherence between the sham-CPAP and active-CPAP groups (individuals who additionally received the Motivational Enhancement intervention were excluded from this analysis). We fit mixed-effects models with participant-specific intercepts and slopes using adherence data recorded each night. Intervention was included as a fixed effect in the models, and models were adjusted for follow-up duration and randomization stratification factors. Because not all participants were randomized to receive 12 months of treatment by design, we evaluated the adherence data recorded each night for 6 months first and then repeated the analysis using all available data up to 12 months.
Results with two-sided p value of less than 0.05 were considered statistically significant.
Results
A total of 86 individuals were analyzed in the control arms, 44 in the CMT group, and 42 in the sham-CPAP group (Figure 1). The sample had a mean age of 63.7 (SD = 6.9) years; 2.3% were Hispanic/Latino, 5.9% were Black, 88.4% were White, and 3.5% were Other. There were no differences in baseline characteristics between the two control arms (Table 1). Participant characteristics were similar for participants in the active arms (n = 42), as reported previously (data not shown) [2, 18, 19].
Figure 1.
Study flowchart.
Table 1.
Baseline characteristics by study arm
| Control (N = 86) | CMT-only (n = 44) | Sham-CPAP (n = 42) | P value | |
|---|---|---|---|---|
| Age (years), mean (SD) | 63.7 (6.91) | 64.2 (6.35) | 63.1 (7.50) | 0.48 |
| Male sex, n (%) | 55 (64.0) | 32 (72.7) | 23 (54.8) | 0.12 |
| Race/ethnicity (self-report), n (%) | 0.71 | |||
| Hispanic/Latino | 2 (2.3) | 1 (2.3) | 1 (2.4) | |
| Black, not Hispanic/Latino | 5 (5.8) | 1 (2.3) | 4 (9.5) | |
| White, not Hispanic/Latino | 76 (88.4) | 41 (93.2) | 35 (83.3) | |
| Other/multiple | 3 (3.5) | 1 (2.3) | 2 (4.8) | |
| Education, n (%) | 0.45 | |||
| Did not grad high school | 1 (1.2) | 0 (0.0) | 1 (2.4) | |
| HS grad/GED | 24 (27.9) | 11 (25.0) | 13 (31.0) | |
| At least bachelors | 61 (70.9) | 33 (75.0) | 28 (66.7) | |
| Body mass index (kg/m2), mean (SD) | 32.3 (6.5) | 32.4 (7.2) | 32.12 (5.7) | 0.86 |
| Average resting SBP, mean (SD) | 124.4 (16.7) | 125.1 (15.3) | 123.6 (18.1) | 0.47 |
| Average resting DBP, mean (SD) | 69.4 (8.6) | 70.1 (8.6) | 68.6 (8.5) | 0.40 |
| Comorbid illness, n (%) | ||||
| Coronary artery disease | 28 (32.6) | 12 (27.3) | 16 (38.1) | 0.36 |
| Diabetes | 35 (40.7) | 17 (38.6) | 18 (42.9) | 0.83 |
| Hypertension | 73 (84.9) | 34 (77.3) | 39 (92.9) | 0.069 |
| Stroke | 2 (2.3) | 1 (2.3) | 1 (2.4) | 0.99 |
| Smoking history, n (%) | 0.99 | |||
| Current | 5 (5.8) | 3 (6.8) | 2 (4.8) | |
| Former | 40 (46.5) | 20 (45.5) | 20 (47.6) | |
| Never | 41 (58.7) | 21 (47.7) | 20 (47.6) | |
| Apnea–hypopnea index (events/hour), mean (SD) | 32.0 (19.1) | 32.6 (20.1) | 31.4 (18.1) | 0.77 |
| Time at SpO2 <90% (%) of recording, mean (SD) | 9.9 (15.2) | 12.3 (16.9) | 7.4 (12.9) | 0.15 |
| Epworth Sleepiness Scale score, mean (SD) | 8.5 (4.5) | 8.8 (4.9) | 8.1 (4.1) | 0.46 |
Baseline characteristics for the CMT-only and sham-CPAP control arms. Values presented are mean (SD) or n (%). p Values were calculated using Wilcoxon rank-sum tests and Fisher’s exact tests for continuous and categorical variables, respectively. DBP: diastolic blood pressure; SBP: systolic blood pressure; GED: General Educational Development tests.
Systolic blood pressure, diastolic blood pressure, and mean arterial blood pressure were compared at baseline, 6 months, and 12 months between the sham-CPAP and CMT-only control groups. There were no consistent differences in blood pressure observed between the two control groups (Table 2), with average differences between sham-CPAP and CMT across the blood pressure measurements ranging from 2.77 mmHg (95% confidence interval [CI] = –-3.71 to 9.24) for sleep systolic blood pressure to –0.73 (–3.54 to 2.08) for wake diastolic blood pressure. Longitudinal profiles in patient-reported outcomes at baseline, 6 months, and 12 months between the sham-CPAP and CMT-only control groups were also found to have wide confidence intervals and show no consistent differences between the two arms (Table 3). For example, observed differences in longitudinal changes for SF-36 subscales ranged from 4.52 (–4.13 to 13.18) for Emotional Role Functioning to –2.26 (–10.52 to 6.00) for Bodily Pain. Notably, differences in these measured outcomes between two control conditions showed no consistent directionality.
Table 2.
Blood pressure by control arm
| CMT-only (n = 40) | Sham-CPAP (n = 40) | Sham-CPAP–CMT-only: (6M + 12M)/2—Base | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Baseline | 6 months | 12 months | Baseline | 6 months | 12 months | Treatment effect | SE | 95% CI | P value | |
| 24 hour SBP | 125.6 (14.5) | 126.2 (14.1) | 126.8 (14.8) | 128.0 (14.4) | 129.0 (14.8) | 129.4 (14.3) | 1.49 | 2.70 | (–3.81 to 6.79) | 0.58 |
| 24 hour DBP | 72.8 (8.7) | 73.2 (10.0) | 73.7 (8.1) | 73.0 (8.0) | 73.3 (9.9) | 71.2 (6.9) | 0.22 | 1.33 | (–2.38 to 2.81) | 0.87 |
| 24 hour MAP | 90.4 (10.0) | 90.9 (10.8) | 91.4 (9.5) | 91.3 (9.1) | 91.9 (10.5) | 90.6 (8.1) | 0.54 | 1.68 | (–2.76 to 3.84) | 0.75 |
| Wake SBP | 129.8 (14.6) | 131.0 (14.7) | 130.0 (16.3) | 130.2 (15.2) | 132.8 (15.0) | 131.9 (14.3) | 0.72 | 2.85 | (–4.86 to 6.30) | 0.80 |
| Wake DBP | 75.8 (9.3) | 77.0 (10.5) | 76.4 (8.9) | 75.3 (8.2) | 75.7 (9.7) | 73.8 (6.6) | –0.73 | 1.44 | (–3.54 to 2.08) | 0.61 |
| Wake MAP | 93.8 (10.3) | 95.0 (11.3) | 94.3 (10.6) | 93.6 (9.5) | 94.7 (10.2) | 93.1 (8.0) | –0.33 | 1.81 | (–3.87 to 3.21) | 0.86 |
| Sleep SBP | 115.8 (18.5) | 116.8 (13.3) | 119.4 (14.0) | 119.7 (17.9) | 121.5 (18.7) | 121.1 (15.3) | 2.77 | 3.30 | (–3.71 to 9.24) | 0.40 |
| Sleep DBP | 65.8 (9.8) | 66.6 (9.0) | 67.2 (7.7) | 66.0 (10.5) | 66.6 (12.4) | 64.7 (9.0) | 0.83 | 1.74 | (–2.59 to 4.25) | 0.64 |
| Sleep MAP | 82.4 (12.2) | 83.3 (9.8) | 84.6 (8.9) | 83.9 (12.2) | 84.9 (13.7) | 83.5 (9.5) | 1.39 | 2.14 | (–2.80 to 5.57) | 0.52 |
| SBP ratio | 89.3 (10.3) | 89.5 (7.2) | 92.2 (7.3) | 91.2 (10.6) | 91.8 (9.4) | 91.5 (6.6) | 0.011 | 0.019 | (–0.027 to 0.049) | 0.58 |
| DBP ratio | 87.1 (10.5) | 87.1 (7.7) | 88.4 (7.8) | 86.9 (10.1) | 87.4 (10.8) | 87.8 (8.9) | 0.009 | 0.020 | (–0.031 to 0.048) | 0.67 |
| MAP ratio | 88.1 (10.1) | 88.2 (7.1) | 90.1 (7.2) | 88.9 (10.2) | 89.4 (9.8) | 89.6 (7.4) | 0.010 | 0.019 | (–0.027 to 0.048) | 0.59 |
To compare the longitudinal profiles of study end points between the sham-CPAP and CMT-only control groups, we implemented mixed-effects linear regression models using functions from the “lme4” package [26]. Observation time (0, 6, and 12 months) was modeled as a fixed effect categorical variable. The treatment effects reported are the averages of the fixed effect beta coefficients corresponding to treatment effects at 6 months and 12 months. Participants contribute to whichever time points they were observed. Models included random intercepts to account for correlation over time and were adjusted for randomization stratification factors (recruitment site, CVD status, and diagnostic study-type). DBP: diastolic blood pressure; SBP: systolic blood pressure; MAP: mean arterial pressure.
Table 3.
Health-related quality of life outcomes by control arm
| CMT-only (n = 44) | Sham-CPAP (n = 42) | Sham-CPAP–CMT-only: (6M + 12M)/2—Base | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Baseline | 6 months | 12 months | Baseline | 6 months | 12 months | Treatment effect | SE | 95% CI | P value | |
| Epworth Sleepiness Scale | 8.8 (4.9) | 8.2 (4.2) | 8.3 (3.4) | 8.1 (4.1) | 6.8 (4.1) | 7.0 (4.5) | –0.58 | 0.67 | (–1.90 to 0.74) | 0.39 |
| WHI Insomnia Rating Score | 9.4 (3.7) | 7.3 (4.3) | 6.0 (3.6) | 10.2 (4.8) | 7.3 (4.0) | 6.7 (4.6) | 0.12 | 0.76 | (–1.38 to 1.62) | 0.88 |
| Physical Health Summary Score | 44.4 (11.2) | 42.5 (10.6) | 44.2 (9.4) | 44.2 (7.7) | 42.9 (9.7) | 41.5 (7.8) | –0.25 | 1.42 | (–3.04 to 2.54) | 0.86 |
| Mental Health Summary Score | 51.4 (8.8) | 51.5 (9.1) | 53.8 (9.6) | 49.2 (10.3) | 53.0 (10.4) | 54.2 (13.7) | 1.31 | 1.52 | (–1.67 to 4.29) | 0.39 |
| Physical Functioning Scale Score | 69.2 (25.9) | 66.1 (26.3) | 73.6 (20.4) | 69.6 (25.4) | 69.2 (26.8) | 65.3 (25.9) | 0.23 | 3.34 | (–6.31 to 6.77) | 0.95 |
| General Health Scale Score | 60.6 (20.9) | 57.2 (25.8) | 57.3 (26.4) | 55.0 (19.0) | 57.9 (17.0) | 58.2 (14.9) | 4.26 | 3.30 | (–2.21 to 10.74) | 0.20 |
| Vitality Scale Score | 57.6 (20.2) | 57.1 (23.7) | 62.8 (17.3) | 54.4 (19.4) | 60.8 (19.0) | 58.9 (19.1) | 1.83 | 2.89 | (–3.84 to 7.49) | 0.53 |
| Bodily Pain Scale Score | 66.1 (24.0) | 62.0 (26.5) | 59.2 (25.8) | 63.2 (24.0) | 61.6 (25.3) | 53.5 (18.3) | –2.26 | 4.21 | (–10.52 to 6.00) | 0.59 |
| Emotional Role Functioning Scale Score | 82.9 (23.0) | 82.7 (22.6) | 86.1 (21.5) | 76.5 (26.3) | 86.4 (20.6) | 83.3 (28.0) | 4.52 | 4.42 | (–4.13 to 13.18) | 0.31 |
| Physical Role Functioning Scale Score | 72.1 (28.8) | 68.2 (27.8) | 80.4 (22.5) | 73.0 (23.2) | 72.3 (23.6) | 71.8 (24.3) | –0.12 | 4.29 | (–8.53 to 8.28) | 0.98 |
| Social Role Functioning Scale Score | 81.1 (22.2) | 82.1 (23.1) | 86.9 (20.7) | 80.6 (23.2) | 79.5 (22.5) | 85.7 (21.0) | –0.12 | 3.61 | (–7.18 to 6.95) | 0.97 |
| Mental Health Scale Score | 76.0 (15.0) | 75.0 (15.8) | 81.0 (14.8) | 75.4 (18.3) | 78.9 (16.6) | 80.8 (21.0) | 0.22 | 2.54 | (–4.76 to 5.20) | 0.93 |
To compare the longitudinal profiles of study end points between the sham-CPAP and CMT-only control groups, we implemented mixed-effects linear regression models using functions from the “lme4” package [26]. Observation time (0, 6, and 12 months) was modeled as a fixed effect categorical variable. The treatment effects reported are the averages of the fixed effect beta coefficients corresponding to treatment effects at 6 months and 12 months. Participants contribute to whichever time points they were observed. Models included random intercepts to account for correlation over time and were adjusted for randomization stratification factors (recruitment site, CVD status, and diagnostic study-type).
Compared to baseline, during the 6-month and 12-month follow-ups, on average the sham-CPAP arm reported lower scores in perceived risk, outcome expectations, and treatment self-efficacy in the SEMSA questionnaire, with treatment effects of –0.176, –0.329, –0.399, respectively, p values of less than 0.05. Compared to baseline, during the 6-month and 12-month follow-up visits, active CPAP participants reported better outcome expectancies compared to the sham-CAP group at follow-up, with average treatment effect = 0.273 (95% CI = 0.002 to 0.544). Changes in perceived risk (effect = –0.009; 95% CI = –0.215 to 0.197) and treatment self-efficacy (effect = 0.227; 95% CI = –0.058 to 0.512) scores were similar between the two groups during follow-up compared to baseline.
Retention rates and AEs were also compared between the sham-CPAP and CMT-only control groups. Retention was 81.0% and 88.6% in the sham-CPAP and CMT-only groups, respectively (p = 0.38). There was also no difference in AE rates between the two control groups (Table 4). Kaplan–Meier survival curves for time to the first AE in the sham-CPAP and CMT-only groups are shown in Figure 2. Median time to first AE was 3 months for the sham-CPAP group and 4.5 months for the CMT-only group. The hazard ratio for first AE comparing sham-CPAP to CMT-only groups was 1.47 (95% CI = 0.88 to 2.48; p = 0.14). Results of sensitivity analyses adjusting for potential informative censoring (dropout) did not substantially differ from the analyses of AE count or time to first AE (Supplementary material). Figure 3 shows adherence in the sham-CPAP arm and active-CPAP arm over a period of up to 12 months (participants that additionally received the behavioral intervention were excluded from this analysis, as we have previously demonstrated a large effect of Motivational Enhancement on adherence). Sham-CPAP was associated with 89 fewer minutes of usage per night over 6 months (p = 0.011) and 93 fewer minutes of usage per night over 12 months (p = 0.007) when compared to active-CPAP.
Table 4.
Adverse event rate by control arm
| Robust standard errors | ||||||
|---|---|---|---|---|---|---|
| CMT-only | Sham-CPAP | Incidence rate ratio | SE | 95% CI | P value | |
| Combined AE | 1.77 (1.96) | 2.26 (2.12) | 1.288 | 0.245 | (0.887 to 1.870) | 0.183 |
| Serious AE | 0.23 (0.64) | 0.14 (0.52) | 0.636 | 0.439 | (0.165 to 2.455) | 0.511 |
| Minor AE | 1.55 (1.86) | 2.12 (2.04) | 1.384 | 0.282 | (0.929 to 2.063) | 0.110 |
To compare adverse event rates by control arms, incidence rate ratios for minor, serious, and combined AE comparing sham-CPAP to CMT-only were calculated as eβ, where β is the estimated coefficient from the Poisson regression. The estimates represent how many AEs were expected in the sham-CPAP arm for every 1 AE in the CMT-only arm.
Figure 2.
Survival time to first AE by control arm. Kaplan–Meier survival curves for time to the first adverse event in the sham-CPAP and CMT-only groups. Median time to first adverse event was 3 months for the sham-CPAP group and 4.5 months for the CMT-only group. The hazard rate ratio (Sham-CPAP versus CMT-only) was calculated using a Cox Proportional Hazards model, adjusting for randomization stratification factors (p = 0.142). *No CMT-only patients were observed 360+ days AE free; at 356 days (last censoring), survival percentage was still 26.8% for CMT-only.
Figure 3.
Adherence rate by study arm. CPAP adherence in the active-CPAP and sham-CPAP arms over 6 months. Each data point represents the mean adherence for a 7-day period. Lines are fitted local regression curves, with the 95% CI in the shaded regions. Active-CPAP participants who additionally received the Motivational Enhancement intervention were excluded.
At the completion of the study, individuals were asked to guess whether they were on active or sham treatment. Data were available for 128 individuals. Of these, 27 on sham responded: 11 (40.7%) did not know/were unable to guess; 16 (59.3%) guessed sham. Of those on active treatment (n = 63), 60 (95.2%) guessed active CPAP, 1 (1.6%) guessed CMT, and 2 (3.2%) guessed sham. Of those on CMT (n = 38), 38 (100%) guessed CMT (data not shown).
Discussion
This study provides novel data assessing the use of both sham-CPAP and CMT as control arms in a CPAP RCT. The study also evaluates adherence rates of sham-CPAP compared to active CPAP over a period of up to 12 months. When comparing sham-CPAP to CMT as a placebo or control arm, we found no evidence for consistent or clinically important differences for either of our prospectively measured objective outcomes (24-hour blood pressure profiles) or patient-reported outcomes, suggesting that sham-CPAP does not substantively influence a broad range of study outcomes compared to alternative control conditions. Neither did we observe differences in frequency of AEs (including both serious and other AEs). However, although estimates were imprecise, participants in the sham-CPAP arm tended to drop out more frequently (19% versus 11.4%) and experienced AEs at a rate of 1.47 times the rate of the CMT-only arm. Overall, these results suggest that general clinical trial experiences are similar for individuals receiving sham-CPAP or CMT-only group, although those in sham-CPAP may experience early side effects of therapy, such as difficulty sleeping due to discomfort or excessive noise caused by intentional air-leak ports, and/or dryness from the mask.
We observed a large difference in hourly device usage over 6 months between the sham-CPAP and active-CPAP group, which was retained over 12 months in the subset of participants followed for this extended period. The lower adherence in the sham-CPAP group is consistent with some reports in the literature [12–15]. A study assessing CPAP adherence over 6 months found the active-CPAP group’s adherence was 3.7 hours a night whereas the sham-CPAP group’s adherence was 2.9 hours a night [12]. Another study comparing active-CPAP to sham-CPAP found adherence was 4.5 hours compared to 2.5 hours a night over the course of 2 months [14]. However, a study assessing adherence over 1 month found only modest differences in active-CPAP versus sham-CPAP nightly use, 3.9 hours versus 3.6 hours per night, respectively [15]. In comparison, average CPAP use over 12 months in our study was approximately 1.5 hours lower in the sham- compared to active-CPAP group. Differences from prior studies may reflect the longer follow-up period, and selection of a sample with underlying cardiovascular risk factors that may differ from other samples. The difference in adherence between arms was evident from the beginning of the trial, rather than emerging over time.
Low adherence rates for CPAP may arise from many factors, such as the burden of wearing a mask/headgear while asleep, patient characteristics (particularly those related to socioeconomic status), method of CPAP initiation (being in a supportive controlled environment when initially presented the treatment of CPAP), and psychological factors including social support and self-efficacy [24, 29, 30]. The low adherence rates in sham-CPAP likely reflect the lack of perceived benefit coupled with the burden of nightly mask use. Participants assigned sham-CPAP also reported lower expectations over time regarding treatment outcomes and self-efficacy. An important issue that has yet to be addressed is whether sham-CPAP may unfavorably affect CPAP adherence when participants are later offered active treatment. For example, the findings from several cross-over studies show that participants who are first randomized to active CPAP exhibit moderate adherence that drops when they cross to sham-CPAP, whereas those who are first randomized to sham-CPAP exhibit poor adherence, which stays low even after crossing to active therapy [31–36]. These findings suggest the need to further evaluate the potential negative effect of sham-CPAP use on later CPAP adherence, which may have a substantial impact on long-term clinical care following participation in a research study.
Sham-CPAP is currently considered the “gold-standard” placebo for CPAP RCTs because of its similarity to active treatment, which may allow for patient-blinding, as well as its ability to provide objectively measured adherence data [10, 11]. The device modifications required for sham-CPAP, however, means that the investigator or therapist responsible for administering the intervention cannot be easily blinded, despite many publications being referred to as “double-blinded.” Within our study, at study completion, a majority of participants who responded to a question regarding which treatment they were on were able to correctly identify which arm they were assigned to. Sham-CPAP is a more complex intervention than an inert pill, requiring nightly use of a mask that blows air into the nose, potentially causing exposure to noise and physical discomfort. Prior research comparing polysomnograms with and without sham-CPAP suggested that the negative effects on respiratory-related variables, such as AHI and arterial oxygen saturation, are minimal [11]. However, the long-term impact on sleep in the home setting has not been assessed. Although we did not observe significant differences in self-reported sleep quality in the sham-CPAP compared to the CMT group, the tendency for early adverse effects together with the low long-term adherence rates suggests that this control intervention may not be free of burden or side effects.
Although advocates of sham-CPAP may argue that it provides less biased estimates of treatment effects relative to unblinded comparisons, little is known regarding the impact of sham-CPAP on objective and subjective outcomes. We observed no differences in longitudinal changes for objectively and subjectively measured outcomes between the two control arms, sham-CPAP, and CMT-only groups. Moreover, the direction of the differences varied across measurements. Therefore, our data support neither a positive nor a negative impact of use of sham-CPAP on blood pressure or patient-reported outcomes. Given the complexity of sham-CPAP use, including participant burden and cost, alternative control arms may be acceptable, especially when outcomes are prospectively observed and objectively measured, and thus less influenced by lack of blinding. Future CPAP trials may adopt PROBE (prospectively randomized open-label blind end points) designs, which are used when it is cumbersome or not feasible to blind patients and practitioners to the study intervention [37]. These designs are reported to enhance participant’s acceptability and accrual and to reduce cost. PROBE designs have been used in trials testing interventions using anticoagulants (requiring tailored monitoring) and surgical devices, as well as those testing alternative drugs. Study integrity is addressed by focusing on end points evaluated using objective measurements made by an independent party with no knowledge of the treatment assignment. A meta-analysis of studies using the PROBE design versus a randomized, double-blinded, controlled design for an antihypertensive medication showed equivalent trial results [38].
Strengths of this article include the RCT design, which included two control conditions, study duration of up to 12-months, and objective measures of CPAP adherence. The study also had some limitations. The findings from this study may not fully generalize to individuals with the most severe levels of OSA, marked by severe sleepiness and hypoxemia, or to individuals without CVD. However, our sample had a broad range of AHI levels and cardiovascular risk factors are common in OSA. Moreover, our results are likely generalizable to other clinical trials, which similarly impose safety exclusion criteria. Although this study is one of the largest long-term RCTs of CPAP, the sample size calculation was based on our primary analysis of comparisons between combined active CPAP arms and combined control arms whereas the sample size within each arm is relatively small. For this reason, in these analyses of secondary outcomes, we highlight the magnitude and consistency of effect sizes rather than statistical significance. These analyses of multiple outcomes did not provide evidence of consistent differences between the sham-CPAP and CMT-only treatments. In addition, because a majority of those in the sham-CPAP arm correctly guessed to receiving an ineffective treatment, the possibility of placebo effects from CPAP treatment cannot be excluded.
Conclusion
This study provides novel evidence suggesting that although sham-CPAP is not associated with significant adverse effects, it is associated with low adherence and reduced self-efficacy compared to active-CPAP and does not appear to influence study outcomes compared to CMT as an alternative control condition. These data support considering alternatives to sham-CPAP in long-term studies where participant and study burden are a concern, and end points can be blindly measured.
Supplementary Material
Acknowledgments
We would like to thank all study participants as well as research staff including Hannah Buettner, Michael Morrical, Beatriz Oropeza, Erin Reese, Michael Rueschman, Tricia Tiu, and Christina Zenobi. The study is also grateful to members of the Data Safety and Monitoring Board (Drs. Mark Espeland, Mike Sharma, Richard Bootzin, Mark Dyken, Ileana Piña), the external Medical Monitor, Dr. Sergio Waxman, the local Medical Monitor, Dr. Sanjay Patel, and Dr. Dennis Drotar for his expertise in adherence fidelity monitoring.
Funding
This study was supported by U34HL105277 and 5R35HL135818.
Disclosure
Jessie Bakker is a full-time employee of Philips Respironics, which is a company that focuses on sleep and respiratory care. Dr. Bakker also has a part-time appointment at Brigham and Women’s Hospital. Dr. Bakker’s interests were reviewed and are managed by BWH and Partners HealthCare in accordance with their conflict of interest policies.
References
- 1. Loube DI, et al. Indications for positive airway pressure treatment of adult obstructive sleep apnea patients: a consensus statement. Chest. 1999;115(3):863–866. [DOI] [PubMed] [Google Scholar]
- 2. Indications and standards for use of nasal continuous positive airway pressure (CPAP) in sleep apnea syndromes. American Thoracic Society. Official statement adopted March 1944. Am J Respir Crit Care Med. 1994;150(6 Pt 1):1738–1745. [DOI] [PubMed] [Google Scholar]
- 3. Patel SR, et al. Continuous positive airway pressure therapy for treating sleepiness in a diverse population with obstructive sleep apnea: results of a meta-analysis. Arch Intern Med. 2003;163(5):565–571. [DOI] [PubMed] [Google Scholar]
- 4. Pepperell JC, et al. Ambulatory blood pressure after therapeutic and subtherapeutic nasal continuous positive airway pressure for obstructive sleep apnoea: a randomised parallel trial. Lancet. 2002;359(9302):204–210. [DOI] [PubMed] [Google Scholar]
- 5. Doherty LS, et al. Long-term effects of nasal continuous positive airway pressure therapy on cardiovascular outcomes in sleep apnea syndrome. Chest. 2005;127(6):2076–2084. [DOI] [PubMed] [Google Scholar]
- 6. Silva GE, et al. Longitudinal evaluation of sleep-disordered breathing and sleep symptoms with change in quality of life: the sleep heart health study (SHHS). Sleep. 2009;32(8):1049–1057. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Means MK, et al. Changes in depressive symptoms after continuous positive airway pressure treatment for obstructive sleep apnea. Sleep Breath. 2003;7(1):31–42. [DOI] [PubMed] [Google Scholar]
- 8. Friedman LM, Furberg CD, DeMets D, et al. Fundamentals of Clinical Trials. Switzerland: Springer International Publishing Switzerland; 2015. [Google Scholar]
- 9. Yagihara F, et al. Nasal dilator strip is an effective placebo intervention for severe obstructive sleep apnea. J Clin Sleep Med. 2017;13(2):215–221. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Chasens ER, et al. Success in blinding to group assignment with sham-CPAP. Biol Res Nurs. 2013;15(4):465–469. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Rodway GW, et al. Evaluation of sham-CPAP as a placebo in CPAP intervention studies. Sleep. 2010;33(2):260–266. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Batool-Anwar S, et al. Impact of continuous positive airway pressure (CPAP) on quality of life in patients with obstructive sleep apnea (OSA). J Sleep Res. 2016;25(6):731–738. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Berry RB, et al. A novel nasal expiratory positive airway pressure (EPAP) device for the treatment of obstructive sleep apnea: a randomized controlled trial. Sleep. 2011;34(4):479–485. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Paz Y Mar HL, et al. Effect of continuous positive airway pressure on cardiovascular biomarkers: the sleep apnea stress randomized controlled trial. Chest. 2016;150(1):80–90. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Khot SP, et al. Effect of continuous positive airway pressure on stroke rehabilitation: a pilot randomized sham-controlled trial. J Clin Sleep Med. 2016;12(7):1019–1026. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Yaggi HK, et al. Reducing cardiovascular risk through treatment of obstructive sleep apnea: 2 methodological approaches. Am Heart J. 2016;172:135–143. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Gleason K, et al. Challenges in recruitment to a randomized controlled study of cardiovascular disease reduction in sleep apnea: an analysis of alternative strategies. Sleep. 2014;37(12):2035–2038. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Bakker JP, et al. Motivational enhancement for increasing adherence to CPAP: a randomized controlled trial. Chest. 2016;150(2):337–345. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Zhao YY, Wang R, Gleason KJ, et al. Effect of continuous positive airway pressure treatment on health-related quality of life and sleepiness in high cardiovascular risk individuals with sleep apnea: best apnea interventions for research (BestAIR) trial. Sleep. 2017;40(4). doi:10.1093/sleep/zsx040 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Jenkinson C, et al. Comparison of three measures of quality of life outcome in the evaluation of continuous positive airways pressure therapy for sleep apnoea. J Sleep Res. 1997;6(3):199–204. [DOI] [PubMed] [Google Scholar]
- 21. Levine DW, et al. Reliability and validity of the women’s health initiative insomnia rating scale. Psychol Assess. 2003;15(2):137–148. [DOI] [PubMed] [Google Scholar]
- 22. Johns MW. A new method for measuring daytime sleepiness: the Epworth sleepiness scale. Sleep. 1991;14(6):540–545. [DOI] [PubMed] [Google Scholar]
- 23. Johns MW. Daytime sleepiness, snoring, and obstructive sleep apnea. the Epworth Sleepiness Scale. Chest. 1993;103(1):30–36. [DOI] [PubMed] [Google Scholar]
- 24. Weaver TE, et al. Self-efficacy in sleep apnea: instrument development and patient perceptions of obstructive sleep apnea risk, treatment benefit, and volition to use continuous positive airway pressure. Sleep. 2003;26(6):727–732. [DOI] [PubMed] [Google Scholar]
- 25. Team RC. R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing; 2017. [Google Scholar]
- 26. Bates D, Machler M, Bolker B, et al. Fitting linear mixed-effects models using lme4. J Stat Softw. 2015;67(1):1–48. [Google Scholar]
- 27. Zeileis A. Econometric computing with HC and HAC covariance matrix estimators. J Stat Softw. 2004;11(10):1–17. [Google Scholar]
- 28. Cox DR. Regression models and life-tables. J R Stat Soc. 1972;34(2):187–220. [Google Scholar]
- 29. Weaver TE, et al. Adherence to continuous positive airway pressure therapy: the challenge to effective treatment. Proc Am Thorac Soc. 2008;5(2):173–178. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Wells RD, et al. Adherence, reports of benefits, and depression among patients treated with continuous positive airway pressure. Psychosom Med. 2007;69(5):449–454. [DOI] [PubMed] [Google Scholar]
- 31. Phillips CL, et al. Continuous positive airway pressure reduces postprandial lipidemia in obstructive sleep apnea: a randomized, placebo-controlled crossover trial. Am J Respir Crit Care Med. 2011;184(3):355–361. [DOI] [PubMed] [Google Scholar]
- 32. Coughlin SR, et al. Cardiovascular and metabolic effects of CPAP in obese males with OSA. Eur Respir J. 2007;29(4):720–727. [DOI] [PubMed] [Google Scholar]
- 33. Sivam S, et al. Effects of 8 weeks of continuous positive airway pressure on abdominal adiposity in obstructive sleep apnoea. Eur Respir J. 2012;40(4):913–918. [DOI] [PubMed] [Google Scholar]
- 34. Weinstock TG, et al. A controlled trial of CPAP therapy on metabolic control in individuals with impaired glucose tolerance and sleep apnea. Sleep. 2012;35(5):617–625B. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. McEwen BJ, et al. Diurnal changes and levels of fibrin generation are not altered by continuous positive airway pressure (CPAP) in obstructive sleep apnoea (OSA). A randomised, placebo-controlled crossover study. Thromb Haemost. 2012;108(4):701–709. [DOI] [PubMed] [Google Scholar]
- 36. Phillips CL, et al. Effects of continuous positive airway pressure on coagulability in obstructive sleep apnoea: a randomised, placebo-controlled crossover study. Thorax. 2012;67(7):639–644. [DOI] [PubMed] [Google Scholar]
- 37. Hansson L, et al. Prospective randomized open blinded end-point (PROBE) study. A novel design for intervention trials. Prospective randomized open blinded end-point. Blood Press. 1992;1(2):113–119. [DOI] [PubMed] [Google Scholar]
- 38. Smith DH, et al. Prospective, randomized, open-label, blinded-endpoint (PROBE) designed trials yield the same results as double-blind, placebo-controlled trials with respect to ABPM measurements. J Hypertens. 2003;21(7):1291–1298. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.