Introduction

Obstructive sleep apnea affects nearly one in four men and one in ten women between the ages of 30 and 60 years in the United States [1]. OSAS (obstructive sleep apnea syndrome) is present when the number of apneic and hypopneic episodes, longer than 10 s, per hour of sleep (referred to as the apnea-hypopnea index, AHI) is five or more and the patient has excessive daytime sleepiness [2, 3]. Health consequences that may result from chronic sleep disruption or recurrent hypoxemia include neuropsychiatric and cardiovascular sequelae. Neuropsychiatric effects may include depression and cognitive dysfunction that can disrupt professional, family, and social life and increase risks for automobile and industrial accidents. Cardiovascular sequelae can include pulmonary and systemic hypertension, congestive heart failure, arrhythmia, myocardial infarction, and stroke [4–6].

The standard treatment for OSAS is continuous positive airway pressure (CPAP); it has become the treatment of first choice for patients with substantial disease. The effect of CPAP on quality of life (QOL) is unclear [7, 8]. The present meta-analysis reviewed the available QOL evidence from CPAP randomized controlled trials to summarize the best evidence of the effect of CPAP in the treatment of OSAS on QOL.

Methods

Search Strategy and Selection Criteria

Randomized controlled clinical trials were identified via MEDLINE (source PubMed, 1966 to June 2007), EMBASE (1966 to June 2007), the Cochrane Controlled Clinical Trials Register Database (through 2nd quarter 2007), and the ClinicalTrials.gov website. All searches included the keywords and corresponding MeSH terms for Sleep Apnea Syndromes, “Sleep Apnea, Obstructive,” “quality of life,” “Continuous Positive Airway Pressure,” and “randomized controlled trial.” Manual reference checking of the bibliographies of all retrieved articles was also done.

Data Extraction

Data extraction was conducted independently by the two reviewers (J.Y. Jing and T.C. Huang). The following information was extracted from each study: first author, year of publication, study population characteristics, study design (parallel or pilot, crossover study), inclusion and exclusion criteria, number of subjects in each group [CPAP and control treatment including conservative treatment (CT), sham CPAP, oral placebo], quality of study, QOL tool used, domains of QOL, treatment duration, and severity of OSAS.

Inclusion Criteria

Prospective randomized controlled trials that were done in adults and published in English were considered for inclusion in this meta-analysis. The following criteria were used to select studies for analysis: (1) studies comparing CPAP treatment with control treatment, and (2) studies that used validated tools for QOL measurements.

Exclusion Criteria

Studies were excluded from the analysis for the following reasons: (1) outcomes of comparison were not reported or it was not possible to extract the data from the published results; (2) the study that did not use validated tools for QOL measures; and (3) more than one article reported outcomes on the same patient group (in that case either the more recent article or the one of higher quality was included).

Measures of Outcomes of Interest

The tools identified as providing validated measures of QOL following OSAS were Short Form 36 (SF-36), Nottingham Health Profile (NHF), Functional Outcomes of Sleep Questionnaire (FOSQ), European Quality of Life Questionnaire (EuroQOL), and Sleep Apnea Quality of Life Index (SAQLI) [9]. All five systems measure QOL within a range of domains and provide an overall indication of QOL.

SF-36

SF-36 was developed in the United States and is a generic measure of health status and can be used to measure health outcomes of clinical interventions [10]. It has been validated and tested for use in 13 countries [11]. The scoring method for SF-36 uses an algorithm to transform dichotomous and continuous variables into a scale from 0 to 100, with higher scores indicating best possible health.

NHP

The NHP is another generic health-related QOL measure, widely used in Europe. It was designed to reflect the perceived effects of ill health on everyday life from the perspective of a layperson rather than that of the health professional. Part 1 includes 38 yes/no items in six domains: physical abilities, pain, sleep, social isolation, emotional reactions, and energy level. Part 2 includes seven aspects of life affected by health: occupation, ability to do jobs around the house, social life, home relationships, sexual life, hobbies, and holidays [12, 13].

FOSQ

The FOSQ is a sleep-specific questionnaire developed to reflect the impact of sleep disorders and excessive sleepiness on activities of daily life. FOSQ contains 30 items divided into five scales: activity level, vigilance, intimacy and sexual relationships, general productivity, and social outcome. Scale scores were added to compute a global score ranging from 0 (maximal dysfunction) to 120 [14].

EuroQOL

The EuroQOL has been developed with cross-cultural applications in mind. The instrument was developed by the international European Quality of Life Group, which has since grown to include members from the United States, Canada, and Japan. The EuroQOL covers five dimensions of health: mobility, self-care, usual activities, pain/discomfort, and anxiety/depression.

SAQLI

The SAQLI was designed as a comprehensive health-related QOL measure for use in clinical trials with patients experiencing sleep apnea. It was based on broad-based input from sleep apnea patients and their partners as well as expert clinicians and the research literature. The first 35 questions measure four domains: daily function, social interactions, emotional functioning, and symptoms. The fifth domain on treatment-related symptoms is a unique feature, capturing the potential negative QOL impact of a treatment’s side effects [15, 16].

Quality Assessment

The quality of each fully published trial was assessed by means of an established standard of methodologic quality [17, 18]. The quality of each study was evaluated by examining patient selection methods, comparability of the study groups, and assessment of outcome. Total methodologic quality scores were then used to rank the studies. Studies given six or more stars were considered to be of high quality. Methodologic quality assessment was independently performed by two independent reviewers (J.Y. Jing and W. Cui).

Statistical Analysis

The meta-analysis was performed in line with recommendations from the Cochrane Collaboration and the Quality of Reporting of Meta-analyses (QUORUM) guidelines [19, 20]. Statistical analyses of continuous variables such as domain outcome for the QOL scores were analyzed using the weighted mean difference (WMD) and were reported with 95% confidence intervals (CI). The WMD summarizes the differences between the two groups with respect to continuous variables, accounting for sample size. For studies that presented continuous data as means and range values, the standard deviations (SD) were calculated using statistical algorithms and checked by using “bootstrap” resampling techniques. Thus, all continuous data were standardized for analysis. In the tabulation of results, squares indicate the point estimates of the effect of disease (WMD), with 95% confidence intervals indicated by horizontal bars. The diamond represents the summary estimate from pooled studies with 95% confidence intervals.

We used the χ2 and Fisher exact tests to detect significant statistical heterogeneity. Heterogeneity was assessed using two methods. First, graphical exploration with funnel plots and the Egger test were used to evaluate publication bias [21]. Second, sensitivity analysis was undertaken for each of these groups of data. We also analyzed the effects of other covariates on the QOL (i.e., quality score, control type, time of treatment, study design, severity of OSAS ). Meta-regression analyses were conducted to reveal potential sources of heterogeneity. The covariates in the regression analyses included study duration, quality score, mean AHI, and mean ESS. Analysis was conducted using STATA v8.2 (StataCorp., College Station, TX).

Literature Search

Of the original 237 studies identified, 216 were excluded because they did not compare QOL between CPAP and control treatment. We carefully read the full text of remaining 21 studies. Only 16 studies remained, published between 1994 and 2007, that matched the selection criteria and were included in this meta-analysis [7, 22–37]. Analysis was performed on a total of 1256 patients, which included 656 CPAP patients and 600 control patients. The characteristics of these studies are listed in Table 1.

Table 1 Characteristics of included studies
Full size table

Meta-analysis

General QOL Score

Seven studies gave a general (or global) health score for FOSQ (including 4 [28, 29, 31, 34] for absolute scores and 3 [7, 27, 35] for total score), five [7, 22–25] for NHP part 2 (energy domain only), one [27] for SF-36, one [30] for EuroQOL, and one [36] for SAQLI (Table 2). CPAP led to significant improvements in the Nottingham health profile part 2 (WMD = 1.657; 95% CI = −3.005, −0.308; p = 0.016) and in SAQLI (WMD = 0.900; 95% CI = 0.625, 1.175; p = 0.000), but there was no difference in general QOL scores when using other measurement tools. The general QOL score expresses the patient’s personal health evaluation and is not a cumulative total of other domain scores.

Table 2 Meta-analyses of general QOL score for CPAP versus control
Full size table

Individual Domains and Subgroup Analysis

In eight studies [25, 28, 32–37], the SF-36 was used (Table 3). Patients undergoing CPAP scored better in physical function (WMD = 3.457; 95% CI = 0.144, 6.771; p = 0.041), body pain (WMD = 4.017; 95% CI = −0.008, 8.042; p = 0.05), energy vitality (WMD = 6.984; 95% CI = 0.557, 13.411; p = 0.033), and physical component score (PCS) (WMD = 2.040; 95% CI = 0.045, 4.035; p = 0.045) using the SF-36 tool.

Table 3 Stratified meta-analyses on study quality, control type, time of treatment, study design, and severity of OSAS. Outcomes of QOL were measured using SF-36 for CPAP versus control
Full size table

In stratified analyses, the following were found: In parallel studies, physical function (WMD = 7.612; 95% CI = 2.573; 12.65; p = 0.003; four studies) physical problems (WMD = 13.906; 95% CI = 5.413, 22.40; p = 0.001; four studies), body pain (WMD = 7.174; 95% CI = 2.099 = 12.249; p = 0.006; four studies), general health (WMD = 5.671; 95% CI = 0.729, 10.614; p = 0.025; four studies), mental health (WMD = 5.659; 95% CI = 1.371, 9.946; p = 0.01; four studies), and PCS (WMD = 2.237; 95% CI 0.178, 4.296; p = 0.033; three studies), components of SF-36, indicated consistent and significant improvements in health status in favor of CPAP. There was no difference in all domains in crossover studies. There was significant statistical heterogeneity between different study designs for physical function, body pain, vitality, physical problems, emotional problems, and mental health domains (Table 3, Fig. 1, 2). In long-term treatment studies, a significant improvement was demonstrated in physical function (WMD = 6.706; 95% CI = 2.573, 12.65; p = 0.027; three studies).

Fig. 1
figure 1

Study selection process for meta-analysis

Full size image
Fig. 2
figure 2

Forrest plot of outcomes of SF-36 of physical function (a), body pain (b), and physical problems (c) for CPAP versus control, stratified meta-analysis with study design

Full size image

Meta-regression analyses showed that Epworth Sleepiness Scale (ESS), AHI, treatment duration, and study quality score were all not the sources of heterogeneity in most of the SF-36 domain. Only study design and AHI were the sources of heterogeneity in physical problems and body pain domain (Table 4).

Table 4 Meta-regression analysis to identify sources of heterogeneity
Full size table

Using the FOSQ tool, domain scores did not show differences between the CPAP group and the control group in four studies [28, 31, 34, 35]. There also was no significant statistical heterogeneity between subgroups (Table 5).

Table 5 Stratified meta-analyses of study quality and severity of OSAS
Full size table

Ballester [26] reported no difference using NHP. Lam [36] reported CPAP patients had a significantly better score in the symptoms, emotional, and daily function than the control patients using the SAQLI (Table 6). Further studies using these tools could help to derive a more reliable overview.

Table 6 Meta-analyses of quality-of-life domain outcomes measured using NHP and SAQLI for CPAP versus control
Full size table

Discussion

According to Schipper et al. [38, 39], “Quality of life (QOL) in clinical medicine represents the functional effect of an illness and its consequent therapy upon a patient, as perceived by the patient.” Improving the quality of life of the OSAS patient is one of the main targets of treatment [40–44]. When comparing CPAP treatment with control, there was no difference in general health perception and total QOL score, except for Nottingham health profile part 2 (energy domain only).

In individual domains, the results of the present meta-analysis found that there was no difference between CPAP and CT or placebo in improving emotional function, MCS, or mental health domain. However, CPAP improves physical function, energy vitality, and PCS domains of the SF-36. Pichel et al. found that the QOL score and symptom improvement of long-term treatment were better than short-term one [7, 44], and there was significant statistical heterogeneity in physical function improvement between study designs. The decrease in physical function was influenced by objective indices of sleep discontinuity and subjective sleepiness. The physical function and the physical role were related to nocturnal parameters indicating sleep disruption, i.e., amount of stage 1 and slow wave sleep, with additional influence from indices of daytime sleepiness and BMI [45]. Vitality improvement also could be found using NHP part 2 and was not related with control types or the length of treatment period. There was significant statistical heterogeneity in improvement of energy vitality and severity of OSAS between study designs. Hypoxemia, AHI, sleep disruption, and sleep fragmentation appeared to have an impact on physical function and energy vitality.

The improvement in physical scores, especially energy vitality, with CPAP is consistent with studies demonstrating significant improvements in sleepiness with CPAP.

The present meta-analysis facilitated the aggregation of data from a variety of sources using standardized QOL assessment tools, the results of which provided greater statistical power to detect significant differences, with subsequent sensitivity analysis demonstrating the robustness of the pooled analysis. The heterogeneity of the studies was analyzed and the results can be seen in Tables 2–5 and in the funnel plot in Figure 2. There was significant heterogeneity in some of the outcomes of the overall analysis. Sensitivity and publication bias were analyzed and the results are shown in the funnel plots in Figures 3 and 4. The results show that the design of the study affects the result of QOL significantly, as does the control type. However, ESS, the severity of OSAS, the quality of score, and the duration of treatment do not affect the QOL scores, except that the physical function of SF-36 indicated significant improvements in long-term trials but not in short-term ones and mean AHI was the source of heterogeneity in physical problems and body pain.

Fig. 3
figure 3

The influence of individual studies on the summary WMD. The vertical axis at 3.46 indicates the overall WMD and the two vertical axes at 0.14 and 6.77 indicate its 95% confidence interval (CI). Every open circle indicates the pooled WMD when the left study was omitted in a meta-analysis with a random model. The two ends of every broken line represent the respective 95% CI

Full size image
Fig. 4
figure 4

Funnel graph for the assessment of potential publication bias in physical function for CPAP compared with control. This is a scatterplot of the incidence estimated from individual studies plotted on the horizontal line (SE of the WMD) and the vertical axis (WMD). The result of the Egger test for publication bias was not significant (p = 0.805)

Full size image

Study design is important in understanding the relationship between patient characteristics and adherence [46]. The crossover study design reduces the efficiency of capturing the QOL effects of CPAP because the washout period is too short to eliminate the effects of pretreatment. In the included studies [22–25, 28, 29, 31, 34, 37] the washout period ranged from 0 to 2 weeks. Long-term parallel-group trials may be more efficient at capturing the important information regarding the persistence of benefits from CPAP treatment, convenience of continued usage, loss to follow-up, and cardiovascular outcome of CPAP treatment.

Control types include CT, oral placebo, and sham CPAP. The selection of control type is controversial. Engleman et al. [22–25] deemed sham CPAP therapy is not possible because a nasal mask without effective CPAP would make both sleep and gas exchange worse. However, the meta-analysis results do not demonstrate that.

The present meta-analysis results demonstrate that there are some different results among QOL questionnaire tools. As a generic measure, SF-36 does not include questions specific for OSAS. The vitality dimension is the closest proxy for sleep-related disturbances [47]. Thus, SF-36 may successfully discriminate between patients with and without OSAS and be sensitive to treatment-induced changes, but it should be accompanied by an OSAS-specific instrument if the researcher is interested in more besides the eight dimensions and two subscales included in SF-36. For specific health-related QOL instruments, preliminary evidence suggests that the SAQLI and the FOSQ are both potentially useful [9].

Although generic questionnaires are designed to measure all important aspects of QOL, they are less likely to detect change in QOL than a disease-specific questionnaire that focuses on specific areas of QOL [48]. Such a disease-specific questionnaire is clearly needed in OSAS research [49], but the number of randomized controlled trial studies that used the disease-specific questionnaire such as SAQLI was small, so further studies are needed to confirm.

In conclusion, when comparing CPAP with control treatment, our meta-analysis shows little impact of CPAP on general QOL. However, CPAP improves physical domains and vitality. Study designs and validated QOL questionnaire tools are important to capture and evaluate information efficiently. However, generic QOL instruments may not be adequate to detect important changes in the quality of life in patients with OSAS. Future randomized controlled trials in this area should concentrate on a disease-specific questionnaire or on large long-term parallel-group trials.