Health status in subjects with suspected obstructive sleep apnea and comparison with a general population

The purpose of this study was to assess health status (HS) in patients with clinical suspicion of obstructive sleep apnea (OSA) in order to estimate the dose response relationship between HS and OSA severity, and to compare HS in this clinical cohort with a general population sample (N = 5000). Patients referred to an overnight sleep study due to suspected OSA, whom also responded to the SF-36 questionnaire, were included (N = 418). Of these, 194 showed normal findings, while 111, 60 and 53 demonstrated mild, moderate and severe OSA, respectively. Mean age was 47.5 (SD 11.9) and 69% were males. Only the mental health scale (p = 0.015) and mental component summary score (p = 0.023) were associated with OSA severity. This association, however, disappeared in multivariable analysis. All SF-36 scores in the sleep study group were lower than that of the general population sample, in both unadjusted and multivariable linear regression analysis. In this study, there was a lack of association between OSA severity and general HS. However, as a whole, patients in this clinical population referred to an overnight sleep study due to suspected OSA had impaired HS on all scales compared to a general population, with greatest differences in the vitality domain.

and patients with moderate to severe OSA scored lower on vitality compared to patients without respiratory disturbances 14 . In contrast, others have found no significant associations between AHI and HS in clinical populations [15][16][17] . Sleep quality has been shown to broadly influence general HS as assessed by the SF-36 18 , and the vitality domain has also shown associations with sleep disruption in OSA patients 19 .
The estimated prevalence of OSA in middle-aged adults in the general Norwegian population is approximately 25% 20 , but there is limited knowledge of the HS of persons with OSA in Norway 21 . The objective of this study was to use the SF-36 to: (1) estimate the dose response relationship between HS and OSA severity, grouped as mild, moderate or severe OSA, and (2) to compare HS in this clinical cohort with a general population sample.

Methods
Participants. Sleep study group. This cross-sectional study wanted to include all subjects referred to the Department of Clinical Neurophysiology, Akershus University Hospital between October 2003 and December 2007 for a sleep study due to clinical suspicion of OSA. Referrals for the majority of patients came from general practitioners, neurologists or ear-nose-throat specialists. The inclusion process was delegated to neurophysiology technicians at the Department of Clinical Neurophysiology. During this period 1533 sleep studies were performed, 89 subjects declined to participate and 666 were not invited for unknown reasons, resulting in a total of 755 non-participants, and an inclusion rate of 51%. During this inclusion period, PSGs without respiratory assessments were also performed for patients who did not have clinical findings consistent with sleep apnea. These patients were not included in this study. In total, 418 subjects with clinically suspect OSA complaints who also fully completed the SF-36 questionnaire were included for further analysis (Figure 1).
The Regional Committee for Medical Research Ethics, Helse-Sør-Øst approved the project's ethical aspects (210-03096). Informed, written consent was obtained from all participants included in the study.
General population. General population values were collected from a survey conducted as part of the Norwegian Lifestyle Survey by Statistics Norway in 2002 22 . The survey comprised 9698 members of the general population ≥16 years of age that were representative for Norway. The study included home and telephone interviews prior to a postal survey. The postal questionnaire included the Norwegian SF-36 version 1.2 23 mailed between 15 November 2002 and 15 May 2003. We had access to data collected from persons that responded to both postal survey and interview. Of these, 5000 also responded to all SF-36 questionnaire domains, and were used in further analysis.
Questionnaire. Participants in the sleep study group filled out a questionnaire that included items about demographics, smoking status, presence of comorbid conditions (high blood pressure, diabetes mellitus and anxiety or depression), a Norwegian version of the Epworth sleepiness scale (ESS) 24 , and a Norwegian version of the standard SF-36 questionnaire, version 1.2 23 .
The SF-36 is a general HS questionnaire. It assesses eight dimensions of health including physical functioning, role limitations related to physical problems, bodily pain, general health perception, vitality, social functioning, role limitations due to emotional problems and mental health. Raw scores are converted to a scale from 0 to 100. Higher transformed scores indicate better HS according to the SF-36 scoring manual 8 . The scales are also aggregated to provide two summary scales, the physical component summary (PCS) and mental component summary (MCS), which are scored on a standardized scale designed for comparison with a US general population with mean 50 (SD 10) 25 .
Sleep studies. Participants in the sleep study group who were evaluated only for sleep apnea underwent an ambulant PG (n = 343) using the Embla system (EMBLA; Medcare-Flaga; Reykjavik, Iceland), recording airflow with thermistors, oxygen saturation by finger-pulse oximeter, heart rate, abdominal and thoracic respiratory movements, body position and snoring. Participants with symptoms suspect for both OSA and other sleep disorders were referred to a polysomnography (PSG) (n = 75) using the same ambulant Embla system, but with additional surface electrode recordings of the electroencephalogram (C3-A2, C4-A1, O1-A2, O2-A1), electrooculograms (EOG1 and EOG2), electrocardiogram, as well as m. submentalis and bilateral m. tibialis anterior electromyograms. The sleep studies were scored according to the criteria of the American Academy of Sleep Medicine Task Force by physicians specializing in clinical neurophysiology 26 . Statistical analysis. Descriptive statistics were presented using means with SDs or number (%). The characteristics of participants in the sleep study group and the general population sample were compared using the t-test for continuous and chi-square test for discrete variables.
SF-36 scale scores were computed according to standardized criteria 8,25 . We compared SF-36 scale scores when participants were grouped according to OSA severity (normal, mild, moderate, severe) using the Kruskal-Wallis test, and the Wilcoxon-Mann-Whitney test when comparing participants in the sleep study group with the general population sample.
In further analysis, we focused on the SF-36 domains of physical functioning and vitality, which in previous studies have been associated with sleep disorders and therefore expected to be most sensitive to symptoms of OSA 10,12,27 , as well as the two component summary scores, PCS and MCS. We chose to adjust the comparisons of HS for the variables age, gender, education, body mass index (BMI), ESS > 10 and comorbid conditions, which in previous studies have shown associations between OSA and quality of life 10,12,14,27,28 , where available.
In the sleep study group we performed multivariable linear regression analysis for the SF-36 domains physical functioning, vitality, PCS and MCS adjusting for age per 10 years, gender, education level, BMI as well as ESS ≤ 10 or >10, presence of high blood pressure, diabetes mellitus and anxiety or depression for participants grouped according to OSA severity compared to those with AHI < 5 (normal). All independent variables were forced into the models.
When comparing participants in the sleep study group to the general population sample using multivariable linear regression analysis and the same dependent variables as above, we adjusted for age per 10 years, gender, education level, and BMI, again forcing all independent variables into the models.
Some of the residuals in the multivariable linear regression models did not comply with a normal distribution, but log-transformation or square-root transformation of the affected dependent variables did not resolve this situation. We therefore used the untransformed values for the dependent variables, but used bootstrapped 95% confidence intervals with 500 replications in all models. Missing values were not imputed. We chose a 5% significance level, using two-sided tests. The Stata version 14.1 (StataCorp, College Station, TX, USA) was used for analyses.
Data availability statement. The patient data analysed during the current study are not publicly available due to ethical and legal constraints, but anonymized data are available from the corresponding author on reasonable request.
The general population data that support the findings of this study are available from Norwegian Centre for Research Data (NSD) but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request and with permission of NSD.
Ethical approval. All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
Informed consent. Informed consent was obtained from all individual participants included in the study.

Results
Descriptive statistics for the sleep study group. Participants in the sleep study group comprised more males (p < 0.001), had higher BMI (p < 0.001), a lower percentage of smokers (p = 0.044) and a greater percentage of participants with university or college degree than the general population sample (p < 0.001), but there was no difference in age (p = 0.37) between the groups. There was increasing BMI with increasing OSA severity, and participants with severe OSA had the highest percentage of ESS > 10 (Table 1).
Dose-response between health status and OSA severity. Comparison of unadjusted SF-36 scale scores between participants in the sleep study group categorized according to OSA severity showed higher values for participants with mild, moderate and severe OSA on the mental health and MCS scales than those with normal findings (AHI < 5) ( Table 2). Higher transformed SF-36 scale scores indicate better HS.
There was no difference in overall HS between participants in the sleep study group categorized according to OSA severity and those in the normal group (AHI < 5) after adjustment for gender, education, age per 10 years, BMI, ESS > 10, presence of high blood pressure, diabetes mellitus and anxiety or depression. In this multivariable linear regression model, vitality and MCS scales were positively associated with age, but negatively with the presence of anxiety or depression. ESS score > 10 was negatively associated with the vitality scale, and BMI and the presence of diabetes mellitus were negatively associated with the physical functioning scale. The PCS scale was negatively associated with high blood pressure ( Table 3).

Comparison of health status between sleep study group and general population. Unadjusted
SF-36 scores in the sleep study group were lower on all scales than in the general population sample (Table 4).
After adjustment for gender, education, age per 10 years and BMI, participants in the sleep study group scored lower than the general population on the physical functioning, vitality, PCS and MCS scales, with the largest difference found on the vitality scale (unstandardized beta coefficient of −23.1 (CI 95% −25.4 to −20.9) p < 0.001) ( Table 5). All variables used for adjustment, except BMI for the MCS scale, showed associations with the assessed domains.

Discussion
Principal findings. In this study of patients with suspected OSA referred to an overnight sleep study, we could not show an overall difference in SF-36 scale scores according to OSA severity after adjusting for confounders. However, when comparing participants with OSA to those with normal findings, increasing age was associated with higher SF-36 scores for the domains of vitality and MCS, while the presence of anxiety or depression was associated with lower scores in these domains. ESS > 10 was also associated with lower vitality scale scores.
The lack of a dose response relationship between HS and OSA severity found in this study is in line with previous studies assessing OSA severity and HS 29,30 , although associations between OSA severity and certain SF-36 domains have also been found in both clinical and non-clinical populations [12][13][14] . In a population-based study,  mild OSA was related to lower scores on the vitality scale, while more severe OSA was more broadly associated with reduced HS 10 . Unadjusted analysis of the sleep study group showed that patients with OSA had higher SF-36 scores for both the mental health scale and MCS scores compared to patients with normal findings. This supports findings from other studies where patients with severe OSA reported higher mental health scale scores than normative SF-36 data 10 , and less depressive symptoms than patients with normal findings 31 . However, this finding is not universal across studies, as other studies have reported lower mental health scale scores for patients with OSA than normal controls 28,32 . These inconsistencies may be related to differences in sample composition, investigation methods, and varying adjustments of potential confounding variables such as age, BMI, and comorbid medical conditions.
All SF-36 scale scores in the sleep study group were lower compared to the general population sample, both unadjusted and when controlled for age, gender, education and BMI. The largest difference was found on the vitality scale, which may be explained by this domain's responsiveness to sleep disruption 19 . Lower HS in OSA patients compared to general population samples in several SF-36 domains are previously documented 33,34 , with some of them also showing significant improvements in HS after treatment with continuous positive airway pressure (CPAP) therapy, with most obvious improvements in those with severe OSA 34,35 . However, the present study was a cross-sectional study, and none had CPAP therapy.
Strengths of the study. The present study included a relatively large number of patients referred to an overnight sleep study with suspected OSA and could be compared to the SF-36 scores of a general population sample directly. Some previous studies assessing HS in OSA patients have similarly included consecutive patients referred to a sleep study 15,27 , while others have used a selection of OSA patients 9,18 or investigated general population samples 10,12 . To our knowledge, there are no prior studies assessing general HS in a Norwegian clinical OSA population in relation to general population norms. Data from the general population sample used in this study was collected at about the same time and had similar age and smoking status as the sleep study group. Increasing age and smoking are both associated with OSA 36,37 . Limitations of the study. Participant selection. One main limitation of the present study was the recruitment process of participants. The intention of the study was to include all consecutive patients referred to an overnight sleep study. Recruitment to the study was difficult to accomplish during periods of the year and varied over time, possibly leading to a recruitment bias. This non-consecutive sampling may have resulted in a sampling bias, and consequently question the representativeness of the sample and generalizability of the findings. In spite of adjustment for important demographic characteristics, findings from this study, based on a selected clinical population referred to sleep study due to clinical symptoms of OSA, cannot fairly be compared with a general population sample. For example, the selected clinical population will naturally have a higher prevalence of OSA than a random general population sample.
Sleep study recordings and scoring. Another limitation is the difference in measurement methods of OSA. More than 80% of the sleep studies were performed using PG, which may underestimate the AHI compared to a PSG with both respiratory and EEG parameters. In addition, the recordings were performed using only a thermistor, which may have affected correct identification of respiratory events. According to latest scoring guidelines, a thermistor is recommended for identifying apneas and a nasal pressure transducer for identifying hypopneas 38 . The 1999 American Academy of Sleep Medicine (AASM) scoring rules used in this study are more liberal than current criteria, as they do not necessarily distinguish obstructive apneas from hypopneas, and do not require a   hypersomnolence, also have lower general HS compared to general population samples 39,40 . For this reason, clinicians should also consider other sleep disorders as well as physical and mental conditions when assessing these patients. The findings also serve as a reminder that the patients being referred for a sleep study are heterogeneous and present with a number of complaints or unspecific symptoms, which may lead to reduced HS independent of OSA or not. The unspecific nature of the symptoms makes screening for OSA difficult, and there is currently uncertainty about the accuracy or clinical utility of all potential screening tools 41 .

Conclusion
The present study has affirmed the lack of association between OSA severity and general HS, as measured by the SF-36 instrument. In contrast, compared to a general population sample, the general HS of Norwegian patients referred to an overnight sleep study due to clinical suspicion of OSA was impaired, with greatest differences in the vitality domain. These findings suggest that this difference may not necessarily be due to the presence of OSA, but may reflect the unspecific nature of symptoms in OSA compared to other sleep disorders or other confounding factors.