Quantity of Research Available
The literature search yielded 757 citations. Upon screening titles and abstracts, 738 citations were excluded and 19 potentially relevant articles were retrieved for full-text review. Three additional reports were retrieved through grey literature searching. Of the 22 potentially relevant reports, 14 did not meet the inclusion criteria. Eight publications were included in this review. The study selection process is outlined in a PRISMA flowchart (Appendix 1). One systematic review, four non-randomized studies, two economic evaluations, and one guideline met inclusion criteria.
Summary of Study Characteristics
Details on study characteristics can be found in Appendix 2.
Study design
One systematic review of data from RCTs was included in this review.3 The systematic review included both parallel and crossover trials, and only the studies using sham-CPAP were double blinded. Of the four non-randomized studies, three were prospective before-and-after studies9–11 and one was a retrospective before-and-after study.12 Two cost-effectiveness analyses13,14 and one evidence-based guideline15 were also included in this review.
Country of origin
The systematic review was performed by a group in the UK and included RCTs from multiple countries.3 Of the non-randomized studies, two were from Australia,10,11 one was from Canada,9 and one was from the United States.12 One economic evaluation each was from the United States13 and the UK.14 The guideline was from the United States.15
Patient population
The systematic review included 48 RCTs of patients ≥ 16 years of age with OSA measured with a standard severity criterion.3 Subgroup analyses were conducted according to OSA severity with moderate OSA defined an AHI of 15 to 30 events per hour, and severe OSA was defined as having an AHI of > 30 events per hour.
Among the observational studies, two prospective studies included patients with at least a moderate severity of OSA, as defined by having an AHI of greater than 15 events per hour.9,11 One prospective study included patients referred to a sleep medicine service with moderate to severe OSA, as defined by having an ESS score of at least 8.11 The retrospective study included patients with OSA that had an AHI of greater than 5: 31% and 36% of patients had moderate and severe OSA, respectively.12 One prospective study included patients with type 2 diabetes.10 In the observational studies, the proportion of male patients ranged from 60% to 86% and the mean age ranged from 41 to 66 years.9–12 The average BMI in the studies ranged from 28.7 to 34.7 kg/m2.9–12
Both economic evaluations used a base case population of males aged 50 years.13,14 One economic evaluation considered the population that had a 50% pretest probability of having moderate to severe OSA, defined as an AHI of greater or equal to 15 events per hour.13 The other economic evaluation did not specify the severity of OSA, but analyzed the cost data according to disease severity.14
The evidence-based guideline focused on adults with OSA.15
Interventions and comparators
The systematic review, non-randomized studies and economic evaluations focused on CPAP treatment as the main intervention. The systematic review looked at best supportive care, sham CPAP and dental devices as comparators.3 There were no comparators used in the prospective observational studies.9–11 The retrospective observational study used oral appliance as a comparator.12 The economic evaluations compared CPAP with either no treatment13 or dental devices and lifestyle advice.14
Outcomes measured
The systematic review3 and prospective observational studies9–11 used ESS as a clinical endpoint, comparing scores before and after CPAP treatment. Other outcomes assessed in the observational studies included the respiratory disturbance index (RDI),9 and oxygen saturation (SpO2),9 the Maintenance of Wakefulness Test (MWT),11 and the reduction in AHI before and after CPAP treatment.12
The economic evaluations estimated the cost-effectiveness of CPAP when compared to no treatment13 or dental devices and lifestyle advice.14 One economic evaluation was from the US third party payer perspective13 and one economic evaluation was from the UK health care payer perspective.14 Data on the effect of untreated OSA on the incidence of motor vehicle collisions and cardiovascular events were estimated from the literature.13,14 One economic evaluation reported costs in USD using prices from 2008.13 The other economic evaluation reported prices in UK pounds using prices from 2005 to 2006.14
The guideline provided recommendations for the treatment of OSA in adults based on an Agency for Healthcare Research and Quality (AHRQ) systematic review of the literature on the effect of OSA treatment on sleep study measures, measures of cardiovascular and diabetes status, and quality of life.15 The AHRQ review included prospective comparative studies and RCTs. The guidelines supplemented the AHRQ review with observational studies in humans reporting death or cardiovascular illness associated with OSA treatment strategies and more recent RCTs.15
Summary of Critical Appraisal
Details on critical appraisal can be found in Appendix 3.
The systematic review employed a comprehensive literature search, screened in duplicate, and assessed the scientific quality of included studies. A summary of the characteristics of the included studies was provided.3 The risk of publication bias was not formally assessed and it was unclear whether grey literature was included in the search strategy. The included RCTs were assessed for whether they had an adequate method of allocation concealment and whether there was blinding of participants. Only in the studies comparing CPAP to sham-CPAP could participants be blinded. It was not possible to have appropriate blinding for studies using dental devices as comparators, and this may bias the outcomes due to the subjective nature of the ESS. The data from the RCTs were pooled using a random effects meta-analysis which was considered appropriate. The sample sizes of the included RCTs were generally small (maximum N = 118).
In the prospective observational studies, the intervention was clearly described.9–11 One study employed a control group of healthy subjects, but this was so that they can compare neuropsychological outcomes of OSA patients to a healthy cohort.9 No observational studies used a control group that received sham-CPAP, so it was unclear whether OSA patients would improve due to a placebo effect over time.9–12 The lack of blinding in all of the studies may be problematic, as the ESS is a subjective assessment of sleepiness. However, sham-CPAP was not always possible due to ethical considerations. CPAP compliance was assessed in all of the prospective observational studies. In the retrospective observational study, the length of CPAP treatment was not specified. Patients who received oral appliance therapy did so after CPAP therapy, so the results may not be generalizable to patients who had not received CPAP therapy in the past.12 There was also no reporting of CPAP compliance and the study was conducted using data from a single clinic, which may limit generalizability of the results.12
Both economic evaluations were conducted in a similar manner in that a literature review of the impact of OSA on sleep outcomes, motor vehicle collisions and cardiovascular events was performed.13,14 The main limitation of the economic evaluations was the availability of data used to generate the model. The majority of patients in the studies were middle-aged men, which may limit generalizability to other patient populations. All of the economic evaluations were conducted in countries outside of Canada and use prices from 2005 to 2008 that may be outdated.
The guideline had was based on a systematic review that had clearly defined selection criteria, methodologies, and a grading system for recommendations.15 Most of the literature identified was of moderate-quality as determined by the study design, risk of bias, and appropriate outcome measures and reporting. Patient preference and views were not taken into consideration and the costs and barriers to guideline implementation were not evaluated or reported. It was unclear whether the guidelines were peer reviewed.
Summary of Findings
Details on study findings can be found in Appendix 4
What is the evidence for the clinical effectiveness and safety of CPAP treatment for adults with moderate to severe obstructive sleep apnea?
One systematic review performed subgroup analyses according to OSA severity to determine the effectiveness of CPAP versus placebo or usual care on daytime sleepiness according to improvements in ESS scores.3 For patients with moderate OSA, pooled results from seven trials found that there was a statistically significant improvement in ESS score in favour of CPAP (mean difference −2.04, 95% confidence interval [CI] −2.99 to −1.09). For patients with severe OSA, pooled results from fourteen trials found that there was a statistically significant improvement in ESS score in favour of CPAP (mean difference −3.41, 95% CI −4.56 to −2.26). The benefit of CPAP compared to placebo or usual care was larger in the trials that included patients with severe OSA. There was a high degree of statistical heterogeneity among the pooled trials (I2 65% to 71%) and five potential sources of this between study variability were baseline disease severity, baseline daytime sleepiness, study design, type of placebo, and study quality. The systematic review also compared CPAP and dental devices and found that for patients with moderate OSA, pooled results from six trials found no statistically significant difference in the impact on ESS scores (mean difference −0.85, 95% CI −2.11 to 0.41).
The three prospective observational studies evaluated the effectiveness of CPAP in patients with moderate to severe OSA using a before-and-after design.9–11 One study found that the mean ESS score after three months of CPAP treatment was statistically significantly better than before treatment (mean difference −6.1, P < 0.001).9 The proportion of patients with an ESS score of greater than 10, which was considered to be clinically significant, dropped from 76% to 30% after CPAP treatment.9 The same study found that there was a statistically significant improvement in mean respiratory disturbance index (P < 0.001) and mean oxygen saturation (P < 0.01) after CPAP treatment compared to before treatment.9 Another prospective observational study found that there was a statistically significant improvement in the median ESS score after one and three months of CPAP treatment compared to before treatment (P < 0.001).10 This study also found an improvement in systolic and diastolic blood pressure after CPAP treatment.10 The third observational study found high variability in patient compliance with CPAP treatment, and analyzed sleep outcomes according to CPAP adherence.11 The proportion of patients that achieved a normal ESS score, defined as a score of less than 10, increased after three months of CPAP treatment.11 There was a clear dose-response relationship for CPAP therapy in that a greater proportion of patients achieved a normal ESS score if they had an average of at least 7 hours of CPAP treatment per night compared to if they had on average less than or equal to 2 hours of treatment per night (80.6% versus 35.9%, respectively).11 Unlike for ESS scores, there was no association between CPAP adherence and MWT scores.11 All three prospective observational studies assessed CPAP compliance using a built-in smart card,9 or a built in device counter.10,11
The retrospective observational study found that CPAP treatment improved the AHI significantly when compared to oral appliance in patients with severe OSA (mean difference −5.88, 95% CI −8.95 to −2.82, P < 0.001), but not in patients with moderate OSA.12 Compliance to CPAP therapy was not measured or reported.
What is the evidence for the cost-effectiveness of CPAP treatment for adults with moderate to severe obstructive sleep apnea?
One economic evaluation estimated the cost-effectiveness of CPAP treatment compared to no treatment.13 A Markov model was created to compare costs and effectiveness over a 10-year interval and the expected lifetime of the patient, assuming that OSA treatment would reduce motor vehicle collisions and cardiovascular events. The base case population was a hypothetical cohort of 50 year old males with a 50% pretest probability of having moderate to severe OSA. Costs were presented in 2008 prices in USD. Using the base case scenario, the incremental cost-effectiveness ratio (ICER) for CPAP treatment was $15,915 per quality-adjusted life year (QALY) gained for the lifetime horizon. Sensitivity analyses found that these results were not sensitive to gender, age or CPAP adherence. The results were most sensitive to the cost of CPAP therapy.
Another economic evaluation estimated the cost-effectiveness of CPAP compared to dental devices or lifestyle advice.14 A Markov model was created to compare these interventions over the expected lifetime of the patient, incorporating the impact of treatments on daytime sleepiness, blood pressure, and health-related quality of life. Daytime sleepiness can increase the risk of road traffic accidents, and blood pressure may affect the incidence of cardiovascular events, and so these events were also included in the model. Costs were presented in 2005 to 2006 prices in UK pounds. The ICER for CPAP treatment compared to dental devices and lifestyle advice was £20,585 per QALY for patients with mild OSA, £9,391 per QALY for patients with moderate OSA, and £4,413 per QALY for patients with severe OSA. The study concluded that CPAP therapy was cost-effective compared to dental devices or lifestyle advice for patients with moderate to severe OSA, but not patients with mild OSA. The results were most sensitive to the cost of CPAP therapy.
What are the evidence-based guidelines regarding CPAP treatment for adults with moderate to severe obstructive sleep apnea?
One evidence-based guideline was identified from the American College of Physicians.15 The American College of Physicians recommends “continuous positive airway pressure as initial therapy for patients diagnosed with obstructive sleep apnea.” [Strong recommendation, moderate-quality evidence]. This guidance was developed from a systematic review of the literature. CPAP was found to be the most extensively studied therapy and improved ESS scores, reduced AHI and arousal index scores, and increased oxygen saturation in patients with at least moderate OSA. However, the literature did not show that CPAP increased quality of life, and evidence on the effect of CPAP on cardiovascular disease, hypertension, and type 2 diabetes was insufficient.15
Limitations
In the systematic review, there was moderate to high heterogeneity among trials in the pooled analyses which limits the confidence in the pooled estimates. However, sensitivity analyses suggested that the benefit of CPAP on daytime sleepiness was robust. In addition, the included studies in the systematic review were not all blinded, particularly in studies that did not use sham-CPAP as a comparator. In the non-randomized studies, a before-and-after design was used with no sham-CPAP control group or blinding and may overestimate the benefit of CPAP treatment on daytime sleepiness symptoms. Although CPAP treatment was associated with statistically significant improvement in ESS scores, the minimal clinically important difference (MCID) for ESS is not known, thus the clinical importance of the differences reported are unclear.
The duration of CPAP treatment in the observational studies was three months, therefore longer-term effectiveness of CPAP treatment in patients with moderate to severe OSA is lacking. The definition of moderate to severe OSA was generally consistent among the included studies (AHI ≥ 15), however in one observational study, the AHI range was not defined in the patients and so it was unclear whether enrolled patients met this criterion.
The majority of patients enrolled in the included studies were male, and this is also reflected in the economic evaluations where the base case scenarios were 50 year old males. This may limit the generalizability of the results to a broader population. There were no economic evaluations identified that were conducted in the context of the Canadian healthcare system, and the included studies used prices from 2005 to 2008 that may be outdated. It may be difficult to generalize results from other countries to the Canadian context.
There was only one American evidence-based guideline that was identified, and the recommendations were not specific to patients with moderate to severe OSA. However, the evidence supporting the recommendation was relevant to patients with at least moderate OSA, and therefore the guideline was included in this review.
