Quality in Primary Care Open Access

Keywords

depression, health perception, hypertension, insomnia, obstructive sleep apnoea, sleep

Background

Obstructive sleep apnoea (OSA) is a common sleeprelated breathing disorder characterised by apnoeas and/or hypopnoeas.[1,2] Apnoeas and hypopnoeas are defined as total (apnoeas) or partial (hypopnoeas) obstruction of the upper airway leading to a cessation of airflow over the nose and mouth despite continued respiratory movements. The length of these events should be at least 10 seconds and hypopnoeas (but not apnoeas) have to be associated with blood oxygen desaturation. The severity of OSA is graded using the Apnoea/Hypopnoea Index (AHI) which is the average number of apnoeas and hypopnoeas per hour of sleep.[1] Dominating symptoms are loud snoring and witnessed breathing interruptions. Sleep fragmentation may cause daytime symptoms, such as excessive sleepiness, in which case the disease OSA syndrome (OSAS) occurs.[3] Insomnia, as well as depression have, however, also been linked to the presence of OSA. The prevalence of mild OSA (AHI 5 without daytime symptoms) has been estimated to be as high as 28%, and 4% of men and 2% ofwomen in the generalNorth American population suffer from OSAS.1 OSA has been linked to hypertension4 and cardiovascular disease (CVD).[5] A proposed mechanism is the sympathetic activation and increased levels of catecholamines causing inflammation, arterial stiffness and atherosclerosis, [6] due to apnoea-related oxygen desaturations. Another link between hypertension and OSA is the shared risk factor of obesity.[4] OSA can negatively affect the treatment of hypertension,[7] and has been shown to increase morbidity andmortality.8 A previous study on men with therapy-resistant hypertension showed that as many as 56% had OSA,[9] compared with 19% of successfully treated hypertensive patients matched for age and gender. Continuous positive airway pressure (CPAP) is the treatment of choice and may reduce blood pressure[10] and cardiovascular morbidity and mortality in patients with severe OSAS.[11]

Despite knowledge of this high prevalence, difficulties in identifying patients with OSA have been described in primary care, causing low referral rates to sleep clinics.[12,13] Guidelines that describe flow charts to identify patients in need of sleep evaluation and potential treatment have been published by the American Academy of Sleep Medicine (AASM).[14] The primary step of this flow chart is based on a routine health examination, patient complaints (e.g. sleep history, characteristics of OSA), as well as an evaluation of the occurrence of comorbidities associated with high risk of having OSA. Increased knowledge regarding new or unknown clinical characteristics and symptoms that are easy to collect and measure in a primary care setting may help to identify those who are in need of OSA evaluation/treatment. Early identification of undiagnosed OSA in newly diagnosed hypertension patients may prevent further development of atherosclerosis and future morbidity and mortality.[6] To the best of our knowledge, no studies have evaluated the association of characteristics included in the initial step of the AASM guidelines with the occurrence of undiagnosed OSA in hypertensive primary care patients. The aim of this study was therefore to explore characteristics associated with undiagnosed OSA in hypertensive primary care patients.

Material and methods

Design and selection criteria

A cross-sectional design was used (Figure 1). All patients, 18–65 years of age, with diagnosed hypertension (140/90 mmHg) at four primary care centres in Sweden received written and oral information about the study from a researcher without clinical contact with the patient, and those who gave informed consent were screened. Exclusion criteria were terminal disease, ongoing treatment for OSA or OSAS, severe psychiatric disease, dementia, alcohol or drug abuse, and difficulties reading or understanding the Swedish language. All data were collected during faceto- face interviews or examinations performed by an ear, nose and throat nurse and physician.

Clinical variables

Data regarding clinical variables (e.g. blood pressure), anthropometrics (weight, height, neck circumference, waist circumference), sleep (self-rated total sleep time and estimated sleep need), medication, OSA symptoms and comorbidities were collected. Diagnosis of diabetes mellitus was based on a history of diabetes, current treatment with antidiabetic drugs or repeated measures of fasting blood glucose values 7 mmol/l. Ischaemic heart disease (IHD) was defined as a history of angina pectoris and/or myocardial infarction and/ or coronary angioplasty and/or coronary bypass surgery.

Respiratory disease was defined as a history of asthma or chronic obstructive pulmonary disease, or patients who were on current treatment (2 agonists and/or inhaled corticosteroids). Transient ischaemic attack (TIA)/stroke was defined as a history of TIA and/or stroke. Snoring, morning headache and dry mouth were measured by 10-point scales (1–10, higher scores indicated more symptoms) used at the study site.

Self-rating scales

One question from the Berlin Sleep Apnoea Questionnaire (BSAQ) was used to measure witnessed apnoeas.[15] The respondent rates the frequency of witnessed apnoeas on a five-point scale (almost every night, 3–4 nights/week, 1–2 nights/week, 1–2 nights/ month, never or almost never). The Minimal Insomnia Symptoms Scale (MISS) was used to measure insomnia. [16] The respondent was asked to rate difficulties initiating sleep, difficulties maintaining sleep and difficulties with non-restorative sleep on a five-point scale (0–4). The Epworth Sleepiness Scale (ESS) was used to measure daytime sleepiness.[17] The respondent was asked to assess the chance of falling asleep in eight different situations on a four-point scale (0–3). The Hospital Anxiety and Depression Scale (HAD) was used to measure depressive symptoms.[18] Seven of the items concern depressive symptoms and were scored on a four-point scale (0–3). The first question concerning current health status from the SF-36 was used to measure perceived health.[19] The participants ranked their health as: (1) excellent, (2) very good, (3) good, (4) fair or (5) poor.

Recordings of sleep-disordered breathing

Full-night respiratory recordings with monitoring of nasal airflow, pulse oximetry, respiratory movements and body position were performed in the patients’ homes using polygraphy14 (Embletta, ResMed AB, Trollha¨ttan, Sweden). Apnoeas and hypopnoeas were manually scored by one researcher (OS) who was blinded with regard to other data. An apnoea was scored if the nasal pressure signal amplitude dropped ≥90% for ≥ 10 seconds and 90% of the event met amplitude reduction criteria. A hypopnoea was scored if the nasal pressure signal amplitude dropped≥30%, oxygen-saturation dropped ≥4% for ≥10 seconds and 90% of the events met amplitude reduction criteria. Sleep time was estimated from patient’s sleeplog and respiratory movement patterns. The total number of apnoeas and hypopnoeas was divided by the estimated sleep time giving the AHI. An oxygendesaturation index (ODI) was calculated in the same manner based on desaturations of≥4%. Patients were defined as having mild, moderate or severe OSA if they had AHI scores of 5–14.9, 15–29.9 or ≥30, respectively.

Statistical analysis

Descriptive statisticswere presented in terms of means and standard deviations, or in numbers and percentages. Factor analytical techniques including structural equation modelling (SEM) were used for data reduction and modelling. The rationale for using factor analytical techniques and SEM was to reduce the complexity of data, and analyse complex relational schemes (e.g. look at relationships, direct and/or indirect effects between several variables) in a way that cannot be done with simple multivariate analysis. [20] First, an exploratory principal component analysis with oblique rotation was used to reduce the complexity of a large number of observed variables to create a simpler factor model. Variables easy to collect and measure in a primary care setting were entered in the explorative factor analysis: body mass index (BMI), neck circumference, waist circumference, diagnosis of diabetes, diagnosis of IHD, diagnosis of hypercholesterolaemia, systolic blood pressure, diastolic blood pressure, moderate physical activity, vigorous physical activity, disturbing snoring, dry mouth on awakening, morning headache, global perceived health, depressive symptoms, witnessed apnoeas, excessive daytime sleepiness, difficulties initiating sleep, difficulties maintaining sleep, early morning awakenings and non-restorative sleep. Criteria for a variable to be retained in a factor were that they had to achieve a factor loading of at least 0.3. To determine the number of factors, Eigenvalues > 1, Scree tree plots, as well as a theory-based selection (i.e. that the factors are meaningful and logical) were used.[21] In a second step the factors from the final exploratory factor analysis were incorporated into a measurement model using a confirmatory factor analysis ad modum LISREL.[22] This was done to examine and test the extent towhich the data collected could be represented by the factor model. Finally, SEM analyses20,23 were performed to test and compare a theoretically sound (as judged by the authors based on the existing literature) model of the structural relationships to a dependent continuous variable, in this study AHI (Figure 2). The reason for using AHI as a continuous numerical variable was to avoid effects of position dependent OSA,24 as well as losing power in the analyses. Associations between the factors were derived using maximum likelihood and are described with their standardised coefficients. Standardised effects found between 0.10 and 0.30 were considered to be small, effects found between 0.30 and 0.50 were considered moderate, and effects > 0.50 were considered strong. ‘Goodness of fit’ tests were reported as the 2 value including degrees of freedom (df), root mean square error of approximation (RMSEA) and the comparative fit index (CFI). An overall RMSEA < 0.06 and a confidence interval range from 0.00 to 0.08 indicated a good fit. A CFI value 0.95 was considered a very good fit.25 Alevel of P < 0.05was regarded as significant.

Descriptive and exploratory factor analysis was performed with SPSS version 16.0. Confirmatory factor analysis and SEM analyses were performed with LISREL software.[22]

Results

Study population

Sixty-nine (39 men and 30 women) of the 480 patients who participated in the clinical examination declined respiratory recordings. No significant differences were seen regarding comorbidities and medications compared with those accepting to participate. Of the 411 performed recordings, 17 were lost due to technical problems. Thus, the final study population consisted of 394 patients (186 men/208 women). Fifty-nine percent of the patients had an AHI > 5/hour indicating undiagnosed OSA. Mild (mean AHI 8.8, SD 2.8), moderate (mean AHI 21.8, SD 4.4) and severe OSA (mean AHI 49.3, SD 19.2) occurred among 29, 16 and 14% of the patients, respectively. Population characteristics, comorbidities and medications are given in Table 1.

Exploratory and confirmatory factor analysis

Initially a seven-factor modelwas extracted explaining 64.4% of the variance. Five factors consisting of anthropometrics, blood pressure, comorbidity, health complaints and physical activity were theoretically sound. The two other factors, however, both described OSA-related symptoms with morning headache and dry mouth in one factor and snoring and witnessed apnoeas in the other. Therefore, data were forced into a six-factor solution. Table 2 describes the final sixfactor model with the four variables describing OSArelated symptoms in one factor. The model explained 58.3% of the variance. Four variables, diagnosis of diabetes, global perceived health, excessive daytime sleepiness and morning headache, showed difficulties with factor loadings > 0.3 in two factors. Morning headache loaded 0.49 in health complaints, but also 0.38 in OSA-related symptoms. The other three loaded strongest in the factor to which they logically belonged.

The confirmatory factor analysis established the six factors of the exploratory factor analysis with an acceptable goodness of fit (2 = 153, df 122, P = 0.03; RMSEA = 0.029 [0.011–0.042]; CFI = 0.98). However, after scrutinising the results, minor changes in the included variables were deemed necessary. Morning headache still loaded in two factors and showed high standard residual values and was therefore excluded, which improved the fit (2 = 147.5, df 124, P = 0.074,RMSEA = 0.026). Figure 2 describes the final confirmatory factor analysis and the theoretical paths from the factors to undiagnosed OSA.

The structural model and associations to undiagnosed OSA

The theoretical model was not fully confirmed in the SEM analyses. No associations were found between undiagnosed OSA and comorbidity, blood pressure or physical activity (Figure 3). Moreover the analysis revealed a negative association (–0.31) between health complaints and undiagnosed OSA, implying that fewer health complaints were associated with undiagnosed OSA. Afactor analysis of the factor ‘health complaints’ revealed that the dyssomnia variables (excessive daytime sleepiness, difficulties maintaining sleep, nonrestorative sleep) and the variables describing poor health (global perceived health and depressive symptoms) represented two separate clusters. The factor, ‘health complaints’, was therefore separated into two factors named dyssomnia and poor health. After several options had been tested, the final SEM model (Figure 3) showed moderate significant associations between anthropometrics (0.45), OSA-related symptoms (0.47) and undiagnosed OSA. OSA had no direct effect on dyssomnia or poor health. Indirect significance effect of 0.16, 0.15 and –0.19 were seen on dyssomnia, poor health and decreased physical activity, respectively, mediated by OSA-related symptoms.

Discussion

This study performed on primary care patients with hypertension confirms the associations between undiagnosed OSA, OSA-related symptoms (snoring, witnessed apnoeas, dry mouth) and anthropometrics (BMI, neck circumference, waist circumference). In contrast, there was no direct association between undiagnosed OSA, comorbidities, blood pressure, dyssomnia, physical activity or self-rated poor health.

Treatment of OSA, especially in patientswith severe symptomatic OSA, may be of importance before irreversible vascular damage has taken place.6 Treatment can also decrease morbidity and mortality.[8,10,11]

Despite this, epidemiological studies suggest that the disease is underdiagnosed.1,3 We found that 59% of the study population had an AHI ≥ 5/h. A previous Swedish study found in a stratified sample of hypertensive men that 37% had AHI ≥ 10.26 Furthermore, another study found a prevalence of 83% (AHI ≥10) in patients with both hypertension and diabetes.[27]

One cause for OSA being underdiagnosed might be due to non-specific symptoms, and the fact that patients themselves are unable to describe apnoeas, as they occur during sleep. Another cause might be that the assessments of patients with sleeping problems in clinical practice tend to focus more narrowly on physiological parameters concerning sleep, breathing and circulation. Because a definite diagnosis requires an objective respiratory recording (performed with expensive and technically advanced equipment), and the disorder is fairly common, there is a need in daily clinical practice for methods that may facilitate better detection of patients where such a recording is warranted. Earlier studies have focused on individual risk factors for OSA, such as anthropometric measures, hypertension, daytime symptoms, nighttime symptoms, depressive symptoms and the metabolic syndrome. [14,28] However, most of these studies have also been performed on sleep clinic populations (where patients have already been identified as likely suffering from OSAS) which might have affected the predictive value of the different variables. Amultimodal approach, taking into account both biometric and psychometric aspects easy to assess and collect in a primary care setting may be needed to identify the patients.

Our model demonstrated that anthropometrics (BMI, neck circumference and waist circumference) as well as OSA-related symptoms (snoring, witnessed apnoeas and dry mouth) may be helpful in identifying undiagnosed OSA among hypertensive primary care patients. Furthermore, biometric and psychometric aspects were equally important in the present model. Anthropometrics and OSA-related symptoms showed beta values of 0.45 and 0.47 respectively. Anthropometrics are known associates of OSA and  70% of those with OSA are obese and higher BMI values tend to be associated with a more severe OSA.4 Mean BMI, neck and waist circumference for patients with OSA (AHI≥5) was 29.9 kg/m² (SD 5.0 kg/m²), 40.0 cm (SD 5.0 cm) and 104.0 cm (SD 13.3 cm), respectively, significantly higher than in those without OSA (P < 0.001). From a mechanistic perspective, obesity causes increased fatty deposits that contribute to narrowing of the upper airway, and also leads to an altered shape.[29] Anatomical factors (e.g. maxillo-mandibular retrognathia, enlarged tonsils) can also compromise the size of the upper airway and increase the risk for OSA and other comorbidities over time, but are difficult to assess for a nurse or GP without specific competence. Furthermore, central and upper body fat correlates with occurrence of OSA. Young et al30 found that every increase of 13–15 cm in waist or hip circumference increased the risk for having OSA by a factor of 4. The use of easy assessable anthropometric measures other than BMI (i.e. neck and waist circumference) may therefore be of importance when identifying patients with OSA. The BSAQ15 (a validated tool used to categorise patients as ‘low’ or ‘high’ risk for OSA) may, together with information from partners and simple inexpensive two-channel recording devices,14 be suitable for use at an early follow-up appointment (after diagnosis of hypertension has been established) to detect undiagnosed OSA.

Comorbidities (IHD, diabetes, hypercholesterolemia), blood pressure, dyssomnia (excessive daytime sleepiness, difficulties maintaining sleep, non-restorative sleep) and poor health (perceived health and depressive symptoms) were not directly associated with AHI in the present model. In contrast to our findings, Vgontzas showed that patients with OSA had higher fasting blood glucose, insulin resistance and glycated haemoglobin (HbA1c) than weight-matched controls, [31] and that the severity correlated with severity of AHI. Others have supported the metabolic syndrome as one of the best predictors for OSA.[28] Depression and poor self-rated health are often associated with OSA.[32] The use of a non-sleep clinic population with less severe OSA (41% having no OSA and 29% having AHI 5–15 indicating mild OSA), a low level of daytime symptoms (excessive daytime sleepiness) and already treated and controlled hypertension in the present model might explain these findings. Furthermore, we used SEManalyses that focus on associations of the derived factors with AHI as a continuous numerical variable, not on OSA as indicated by, for example, AHI≥ mild OSA) or AHI≥ 15 (moderate OSA). Identifying the relevant patient is, however, of critical importance because CPAP treatment can decrease morbidity and mortality, thus reducing consumption of healthcare resources.

Limitations

Despite its relatively large sample size, this study has some limitations. A cross-sectional design was used, which limits conclusions of cause and effect in the proposed theoretical model. We performed full-night respiratory recordings with polygraphic equipment in patients’ homes.Guidelines describe polysomnography14 as a preferred method, but polygraphy can be used as a comprehensive sleep evaluation to decrease costs and minimise inconvenience for patients.[33] Well validated self-rating scales15–19 were used to collect self-rated variables in the theoretical model. A limitation, how- ever, was that the model was based on well known characteristics from the existing AASM guidelines14 and did not explore new, unknown characteristics in hypertensive primary care patients. Furthermore, aspects, such as cognitive function (e.g. memory loss, decreased concentration), decreased libido and irritability were not measured and included in the model. Another limitationwas the lack of clear cut-offs for the dependent variable (AHI), anthropometrics, or OSArelated symptoms that we used in the SEM analyses. If data regarding 24-hour blood pressure were collected we might have found different results (i.e. significant differences) for the association between blood pressure and AHI. Furthermore, the results might have been different if only patients with moderate or severe undiagnosed OSA suitable for treatment with CPAP had been included in the SEM model. Further studies are needed to evaluate sensitivity and specificity of this model to identify patients with undiagnosed OSA of different severity levels. Such studies should use a gender perspective to identify men and women that would show high cost benefit for referral to sleep clinics.

Conclusion

Undiagnosed OSA was directly associated with OSArelated symptoms (snoring, witnessed apnoeas, dry mouth) and anthropometrics (BMI, neck circumference, waist circumference). These characteristics could be used by GPs to identify patients who are in need of referral to a sleep clinic for OSA evaluation. No direct associations were found for comorbidities, blood pressure, insomnia or poor health.

Acknowledgments

We would like to thank Anna Sta˚hlkrantz, RN, MnSc, Department of Nursing Science, School of Health Sciences, Jo¨nko¨ping University, Sweden for assistance with collection of clinical variables.

Funding

The Swedish Heart Lung Foundation, Grant 20090547. The Health Research Council in the South-East of Sweden, Grant FORSS-12568 and FORSS-12710.

Ethical Approval

The study protocol was approved by The Ethics Committee at The Faculty of Health Sciences, University of Linko¨ping (Dnr M29–07), Sweden, and is in accordance with the provisions of the Helsinki declaration.

Peer Review

Not commissioned; externally Peer Reviewed.

Conflicts of Interest

None.

References

1. Young T, Peppard P and Gottlieb D. Epidemiology of obstructive sleep apnea: a population health perspective. American Journal of Respiratory and Crititical Care Medicine 2002;165:1217–39.
2. Hrubos-Strøm H, Randby A, Namtvedt SK et al. A Norwegian population-based study on the risk and prevalence of obstructive sleep apnea. The Akershus Sleep Apnea Project (ASAP). Journal of Sleep Research 2011;20(1 Pt 2):162–70.
3. Partinen M and Guilleminault C. Daytime sleepiness and vascular morbidity at seven-year follow-up in obstructive sleep apnea patients. Chest 1990;97:27–32.
4. Wolk R, Shamsuzzaman AS and Somers VK. Obesity, sleep apnea, and hypertension. Hypertension 2003;42: 1067–74.
5. Selim B, Won C and Yaggi HK. Cardiovascular consequences of sleep apnea. Clinincal Chest Medicine 2010; 31:203–20.
6. Buchner NJ, Quack I, Stegbauer J et al. Treatment of obstructive sleep apnea reduces arterial stiffness. Sleep and Breathing 2011;Jan 7 [Epub ahead of print].
7. The Task Force for the Management of Arterial Hypertension of the European Society of Hypertension (ESH) and of the European Society of Cardiology (ESC). Guidelines for the management of arterial hypertension. Journal of Hypertension 2007;25:1105–87.
8. Lavie P. Mortality in sleep apnoea syndrome: a review of the evidence. European Respiratory Review 2007;16:203– 10.
9. Isaksson H and Svanborg E. Obstructive sleep apnea syndrome in male hypertensives, refractory to drug therapy. Nocturnal automatic blood pressure measurements – an aid to diagnosis? Clinical and Experimental Hypertension 1991;13:1195–212.
10. Becker HF, Jerrentrup A, Ploch T et al. Effect of nasal continuous positive airway pressure treatment on blood pressure in patients with obstructive sleep apnea. Circulation 2003;107:68–73.
11. Marin JM, Carrizo SJ, Vicente E and Agusti AG. Longterm cardiovascular outcomes in men with obstructive sleep apnoea–hypopnoea with or without treatment with continuous positive airway pressure: an observational study. The Lancet 2005;365:1046–53.
12. Kramer NR, Cook TE, Carlisle CC, Corwin RW and Millman RP. The role of the primary care physician in recognizing obstructive sleep apnea. Archives of Internal Medicine 1999;159:965–8.
13. Netzer NC, Hoegel JJ, Loube D et al. Prevalence of symptoms and risk of sleep apnea in primary care. Chest 2003;124:1406–14.
14. Epstein LJ, Kristo D, Strollo PJ et al. Clinical guideline for the evaluation, management and long-term care of obstructive sleep apnea in adults. Journal of Clinical Sleep Medicine 2009;5:263–76.
15. Netzer NC, Stoohs RA, Netzer CM, Clark K and Strohl KP. Using the Berlin Questionnaire to identify patients at risk for the sleep apnea syndrome. Annals of Internal Medicine 1999;131:485–91.
16. Broman JE, Smedje H, Mallon L and Hetta J. The Minimal Insomnia Symptom Scale (MISS): a brief measure of sleeping difficulties. Upsala Journal of Medical Sciences 2008;113:131–42.
17. Johns MW. A new method for measuring daytime sleepiness: the Epworth Sleepiness Scale. Sleep 1991; 14:540–5.
18. Zigmond AS and Snaith RP. The Hospital Anxiety and Depression Scale. Acta Psychiatrica Scandinavica 1983; 67:361–70. 298 A Brostro¨m, O Sunnergren, P Johansson et al
19. Ware JE Jr and Sherbourne CD. The MOS 36-item short-form health survey (SF-36). I. Conceptual framework and item selection. Medical Care 1992;30:473–83.
20. Schreiber JB. Core reporting practices in structural equation modeling. Research in Social and Administrative Pharmacy 2008;4:83–97.
21. Nunnally JC and Bernstein IH. Psychometric Theory (3e). New York, NY: McGraw-Hill, 1994.
22. Jo¨reskog K and So¨rbom D. LISREL 8: structural equation modeling with the SIMPLIS command language. Skokie, IL: Scientific Software International, 1993.
23. Muthe´n B. A general structural equation model with dichotomous, ordered categorical and continuous latent variable indicators. Psychometrika 1984;49:115–32.
24. Sunnergren O, Brostro¨m A and Svanborg E. Positional sensitivity as a confounder in diagnosis of severity of obstructive sleep apnea. Sleep Breath 2012 Mar 1. [Epub ahead of print].
25. Hooper D, Coughlan J and Mullen M. Structural equation modelling: guidelines for determining model fit. Electronic Journal of Business Research Methods 2008; 6:53–60.
26. Sjo¨stro¨m C, Lindberg E, Elmasry A et al. Prevalence of sleep apnoea and snoring in hypertensive men: a population based study. Thorax 2002;57:602–7.
27. Hedner J, Bengtsson-Bostro¨m K, Peker Y et al. Hypertension prevalence in obstructive sleep apnoea and sex: a population-based case-control study. European Respiratory Journal 2006;27:564–70.
28. Drager LF, Genta PG, Pedrosa RP et al. Characteristics and predictors of obstructive sleep apnea in patients with systemic hypertension. American Journal of Cardiology 2010;105:1135–9.
29. Pillar G and Shehadeh N. Abdominal fat and sleep apnea: the chicken or the egg? Diabetes Care 2008;3: S303–9.
30. Young T, Palta M, Dempsey J et al. The occurrence of sleep-disordered breathing among middle-aged adults. New England Journal of Medicine 1993;328:1230–5.
31. Vgontzas AN, Zoumakis E, Bixler EO et al. Selective effects of CPAP on sleep apnoea-associated manifestations. European Journal of Clinical Investigation 2008; 38:585–95.
32. Lee IS, Bardwell W, Ancoli-Israel S et al. The Relationship between psychomotor vigilance performance and quality of life in obstructive sleep apnea. Journal of Clinical Sleep Medicine 2011;15:254–60.
33. Collop NA, McDowell Anderson W, Boehlecke B et al. Clinical guidelines for the use of unattended portable monitors in the diagnosis of obstructive sleep apnea in adult patients. Portable Monitoring Task Force of the American Academy of Sleep Medicine. Journal of Clinical Sleep Medicine 2007;3:737.