2008-Issue1.indd Screening of Patients with Snoring and Obstructive Sleep Apnoea using Heart Rate Variability Indices Omar Al Rawas,1 Bazdawi Al-Riyami,1 Christopher Goddard,2 *Mohammed O Hassan3 يّ النَّومِ دادِيُّ االنْسِ سِ النَفَ طاعُ وانْقِ بالشخير املصابني املرضى حتري القلب نبضات ب تغير ناسِ مَ باستخدام عثمان حسن محمد قودارد كرستوفر الريامي، بزدوي عمر الرواس، هو العرض النهار أثناء النعاس بفرط الشخير مصحوبا االنتشار. واسعة االضطرابات هما من يّ النَّومِ دادِيُّ االنْسِ سِ النَفَ طاعُ وانْقِ اخلالصة: الشخير حديثا استخدمت . يّ ــدادِيُّ النَّومِ االنْسِ سِ النَفَ طاعُ انْقِ بشدة القلب نبضات ــرعة س تغير يرتبط . يّ النَّومِ ــدادِيُّ سِ االنْسِ طاعُ النَفَ النْقِ ــارا انتش األكثر عليه معول بشكل املمكن من . يّ النَّومِ دادِيُّ االنْسِ سِ النَفَ طاعُ بانْقِ املرضى املصابني عن للتحري 24 ساعة ملدة القلب نبضات ة رْعَ سُّ طيطُ تَخْ تقنية : الطريقة التحويلي. فورير برنامج باستعمال بالتحليل الطيفي القلب نبضات سرعة في تغيرا تسبب إرادي التي الال العصبي مكونات اجلهاز دراسة مت الشخير يعانون من 10 أشخاص و يّ النَّومِ دادِيُّ االنْسِ سِ طاعُ النَفَ مصابون بانْقِ 13 منهم ، والعمر الوزن في متقاربون 23 شخصا الدراسة شملت العاشرة الساعة من ابتداءا في املنزل 24 ساعة ملدة القلب تخطيط تسجيل مت . قابوس السلطان جامعة مبستشفى النوم مختبر في تشخيصهم السريع. التحويلي ــتعمال برنامج فورير باس باملنزل القلب تخطيط ومن النوم مختبر من ــتخرجة املس القلب لنبضات الطيفي بالتحليل وقمنا . ليال الشخير مرضى من بكثير يّ أقصر النَّومِ دادِيُّ سِ االنْسِ النَفَ طاعُ انْقِ مرضى عند القلب لنبضات الزمنية أن الفترات القلب في تخطيط النتائج: تبني االنسدادي النفس بانقطاع املصابني عند إحصائيا معتد ــكل بش أعلى القلب ــرعة نبضات الطيف لس ض لكثافة فِ ُنْخَ امل التَّرَدُّدُ . (p<0.01) لوحده جهاز استطاع . اموعتني في متساوية الذبذبة العالية حلزم الطيف كثافة نتائج طاقة كانت . (p< 0.0001) بالشخير املصابني في النومي عنه طاقة ــتخدام اس من ــتفادة االس اخلالصة: ميكن ــدادي. االنس النفس وانقطاع ــخير بالش مصابا 13 مريضا على الليل التعرف خالل القلب تخطيط الشخير من يعانون الذين ملعرفة األشخاص ، النوم أثناء القلب نبضات لقياس التحويلي فورير برنامج باستعمال املنخفضة الترددية للحزمة الطيف . االنسدادي النومي النفس بانقطاع املصحوب القلب. نبضات سرعة تغير ، النومي االنسدادي النفس الكلمات: انقطاع مفتاح SULTAN QABOOS UNIVERSITY MEDICAL JOURNAL MARCH 2008, VOLUME 8, ISSUE 1, P. 21-25 SULTAN QABOOS UNIVERSITY© SUBMITTED - 23RD JANUARY 2008 ACCEPTED - 2ND FEBRUARY 2008 Departments of 1Medicine, 2Medical Physics Unit, 3Physiology, College of Medicine & Health Sciences, Sultan Qaboos University, Muscat, Sultanate of Oman *To whom correspondence should be addressed. Email: mhassan@squ.edu.om ABSTRACT Objective: Snoring and obstructive sleep apnea (OSA) are common disorders. Snoring associated with excessive day- time sleepiness is the most prevalent symptoms of OSA. Heart rate variability (HRV) is altered in patients with OSA and the degree of alteration may be linked to the severity of OSA. Alterations in HRV in 24 hour tachograms have recently been used in screening OSA patients. Autonomic components causing HRV can be reliably studied using spectral analysis techniques involving fast Fourier transformation (FFT). Methods: Twenty-three subjects, 3 with severe OSA and 0 controls matched for age and body mass index, were selected from patients who had undergone polysomnography (PSG) for snoring at Sultan Qaboos University Hospital, Oman. A 24- hour electrocardiogram (ECG) Holter recording was done at home, starting at 0am. Spectral analysis of ECG from sleep Holter and PSG recordings was analysed using fast Fourier transformation (FFT). Results: The ECG RR intervals of snorers with OSA were significantly shorter than in snorers without OSA (p<0.0). The low frequency (LF) spectral densities of HRV from polysomnography and Holter were significantly higher in OSA patients than in snorers, (p< 0.000). The power spectral density of the high frequency bands was similar in the two groups. The overnight ECG Holter accurately identified all 3 snorers with severe OSA. Conclusion: The spectral power of the LF band obtained using FFT of sleep HRV from Holter tachograms may be a useful and cost effective test in identifying snorers with severe OSA. Key words: Obstructive Sleep Apnea; Heart Rate Variability. C L I N I C A L A N D B A S I C R E S E A R C H O M A R A L R AWA S , B A Z D AW I A L - R I YA M I , C H R I S T O P H E R G O D D A R D , M O H A M M E D O H A S S A N 22 SNORING AND OBSTRUCTIVE SLEEP APNEA (OSA) are common disorders that affect bothmen and women. Snoring and excessive daytime sleepiness are the most prevalent symptoms of sleep apnea, which in turn is associated with cardiovascular and cerebrovascular complications. 1, 2, 3 The recogni- tion of OSA and snoring as a health problem has grown worldwide and the number of subjects who seek medi- cal help for problems with snoring and with concern about OSA and daytime sleepiness is increasing. In the general population, the greatest challenge for primary care providers lies in determining which patients with these symptoms warrant further evaluation, as most patients with OSA snore, but most snorers do not have OSA.4 In developed countries, the cost of investigating these symptoms has increased considerably during the last decade as the only reliable method for the diagnosis of OSA until now has been overnight polysomnogra- phy (PSG),5 which is a cumbersome, time consuming and expensive procedure requiring specially trained polysomnographers, despite evolving newer software which allows rapid analysis. This, together with long waiting lists of patients, has lead to the search for faster and less expensive diagnosis. Heart rate variability, or short-term changes in the RR intervals, of a continu- ously recorded electrocardiogram (tachogram), is the consequence of various influences of the autonomic nervous system on heart rate.7, 8 Heart rate variability (HRV) is altered in patients with OSA and this altera- tion is evident even in the absence of hypertension, heart failure, or other diseases.9, 10 The degree of altera- tion in HRV variability may be linked to the severity of OSA.11 Alterations in HRV in 24 hour tachograms have recently been used in screening patients with OSA.12 Autonomic components causing HRV can be reli- ably studied using spectral analysis techniques. After Akselrod13 introduced power spectral analysis (PSA) of short-term heart rate fluctuations as a non-inva- sive quantitative probe of beat-to-beat cardiovascu- lar autonomic control, the study of HRV has become a rapidly expanding field in clinical research14, 15 The analysis gives selective information on sympathetic and parasympathetic functions and the reproducibility of HRV findings in OSA have been confirmed by several studies.16, 17 Power spectral density (PSD) provides the basic in- formation of how power, or variance, is distributed as a function of frequency. Fast Fourier transformation (FFT) is the most widely used nonparametric method for the calculation of PSD.16, 17 The advantages of FFT are the simplicity of the algorithm used, the high process- ing speed, and no data pre-processing is required. In addition, FFT software is readily available in most car- diovascular analyses packages and end-users are not required to have advanced computer knowledge. Three main spectral components can be distin- guished in a spectrum calculated from short (2-5 minutes) and long segment (24 hours) of electrocardi- ographic recordings: very low frequency (VLF, 0.0008- 0.04 Hz), low frequency (LF, 0.04-0.14 Hz) and high frequency (HF, 0.15-0.45 Hz). The distribution of the power and the central frequency of LF and HF are not fixed, but may vary in relation to changes in autonomic modulations of heart rate. In normal humans, short- term RR interval variability occurs predominantly at LF which may be due to baroreceptor, sympathetic and parasympathetic modulations and HF which is syn- chronous with the respiratory frequency.16 The physio- logical explanation of the VLF component is much less defined, and the existence of a specific physiological Advances in Knowledge • Learn how to utilize heart rate variability for advanced diagnostics • Perform signal processing on heart rate intervals by using spectral analysis software which is available free in the web • Develop awareness that sleep apnea can affect heart rate variability. • Sleep apnea can also present with snoring. Applications to Patient Care. • Short listing of patients for full sleep studies • Holter ECG can be performed at home • Polysomnography is essential for the definitive diagnosis of sleep apnea S C R E E N I N G O F PAT I E N T S W I T H S N O R I N G A N D O B S T R U C T I V E S L E E P A P N O E A U S I N G H E A R T R AT E VA R I A B I L I T Y I N D I C E S 23 process attributable to these heart period changes has been questioned. However, using FFT and sub-band decomposition of overnight HRV in patients with OSA, we and other workers have shown that the VLF band is augmented in snorers with OSA as compared to snor- ers without OSA.18, 19 The aim of this study was to evaluate the use of FFT for the spectral analysis of HRV as a method for screen- ing patients presenting with snoring as the only symp- tom, in an attempt to identify those who might have OSA and short list them for polysomnography. The accuracy of the identification algorithm de- pends mainly on the accuracy of the electrocardiogram QRS waveform detection (R peak detector), that is used to obtain the R-R intervals (RRI) in milliseconds from the raw electrocardiogram (ECG) data. Data are the normal-to-normal (NN) intervals obtained directly from the QRS detector without any smoothing and fil- tering steps; therefore, it could contain false intervals, missed and/or ectopic beats. The QRS detection is ac- complished by Open Source ECG Analysis Software, an arrhythmia detection software, available on the PhysioNet website.16 The basis of this simple approach to exclude false RRI by binding RRI within lower and upper limits, is typically 400-2000 msec so that all RR intervals beyond these limits are excluded. M E T H O D S Overnight PSG was performed using a digital poly- graph (Heritage Model 15 Astro-Med Inc, MA, USA). The PSG recording included an electroencephalogram (EEG), ocular and leg electromyography, heart rate tachogram, oxygen saturation, snoring and body posi- tion. Sleep stages and apnea/hypopnea index (AHI) iden- tification were verified manually by a qualified polys- omnographer who used 30-second epochs according to standard criteria. Scoring of all computerised studies was done manually in a manner identical to that used for paper studies.20 Subjects with OSA episodes with an AHI > 30/ hour were considered as having severe OSA while those with an AHI of < 5/ hour were con- sidered as normal. The study subjects were snorers with no other symp- toms. They were selected from the pool of all polysom- nography studies performed in our laboratory between the year 1998- 2003 (total 187 studies). The selection of subject was based on the following: A. OSA group a. Severe OSA (AHI>30/hour) b. No OSA treatment c. No evidence of hypertension, heart failures, diabetes or hypothyroidism B. Control group (snorers without OSA) a. No OSA (AHI <5/hour) b. No evidence of hypertension, heart failure, diabetes or hypothyroidism A total of twenty-three subjects were included for this study (13 with severe OSA and 10 age and body mass index matched controls). Subject characteristics are shown in Table 1. All study subjects had a 24 hour ECG recording at home using an ECG analyser with an analogue to digital converter (Pathfinder 4, RR tools, Reynolds). ECG data from the time the patient recorded going to bed until the time of awakening were selected for the analysis. Thirty segments of 5 minutes each of Lead II of the ECG from PSG and from Holter recordings were se- lected by an ECG technologist who was blind to the study. Ectopic beats, movement artifacts and RR inter- vals +100 msec were manually removed. Another set of data was analysed without visual editing. ECG data was then analysed using the Matlab software (Mathworks, USA) with a sampling frequency of 200 msec and a val- idated QRS detector identification algorithm with an automated rejection of ectopic beats and missing data artifacts.21, 22 Student’s one-tailed t-test was used to compare the PSD of the different frequencies while the 95% confi- dence interval was used to compare the LF bands of PSG and Holter records in the two groups. OSA N:13 (8M, 5F) Snorers N:10 (7M, 3F) Age (years) 37.3 (4.2) 38.1 (3.9) BMI (kg/m2) 34.6 (5.2) 33.1 (6.0) AHI/ hour 35.3 (4.5) 4.0 (1.2) R-R (msec) 792 (45) 811 (56) Values expressed as mean (SD); AHI: Apnea hypopnea index Table 1: Characteristics of the study subjects O M A R A L R AWA S , B A Z D AW I A L - R I YA M I , C H R I S T O P H E R G O D D A R D , M O H A M M E D O H A S S A N 24 R E S U L T S Figure 1 shows the power spectral densities in absolute values obtained from RR values of tachograms recorded during sleep using PSG and Holter recordings in snor- ers with and without OSA. The RR intervals of snor- ers with OSA are significantly shorter than in snorers without OSA (p<0.01). There was no difference in the VLF, LF or HF spectral densities in the manually edited and unedited ECGs. The LF frequency spectral densi- ties of HRV from PSG and Holter recordings were sig- nificantly higher in OSA patients than in snorers, (p< 0.0001). The PSD of the HF bands were similar in the two groups. The overnight ECG Holter accurately iden- tified all 13 snorers with severe OSA. The VLF compo- nents, however, had very high spectral densities in both methods, and showed a very wide scatter and overlap with the LF bands in most patients. Nonetheless, the VLF from PSG and Holter recordings accurately identi- fied the same 13 snorers with OSA. D I S C U S S I O N The main findings of this study are that the LF frequen- cy spectral densities of HRV from PSG and Holter re- cordings were significantly higher in OSA patients than in snorers. Second, patients with severe OSA, who are likely to have or develop complications can be detected early for full PSG. Third, complications of OSA such as heart failure, hypertension and obesity have been shown to contribute to the PSD of the LF band in OSA9, 23 and therefore these complications can only increase the sensitivity of the test. Using this method, shorter periods of Holter recordings can be used in patients hospitalized for continuous positive airway pressure (CPAP) titration or other reasons and several patients can be studied per night. The main strengths of this method are that it is simple, cost effective, non-invasive and can be performed at home; in addition, manual ed- iting is not required as the software is capable of remov- ing movement artifacts, ectopic beats and extremes of tachycardia and bradycardia, which are characteristic of OSA episodes and are known to interfere with the spectral analysis. Although the ECGs were recorded using two dif- ferent methods and under different conditions in the laboratory (PSG) and at home (Holter), they clearly demonstrate that the method and the environment of recording do not influence the outcome as both contain ECGs recorded during all sleep stages. The results also show that the LF using FFT and the VLF using sub-band decomposition as shown in our previous study18 were accurate in identifying snorers with OSA. FFT, as com- pared to other algorithms, is commercially available in most HRV software and does not require advanced computer knowledge as does sub-band decomposition. Although the VLF band was found to be augmented in OSA in this and another study, 18, 19 we found it had a very wide scatter as compared to the LF band. Apart from ranking patients for PSG, this method may be useful in centres where there are no PSG facilities and in patients with clearly reported and observed symp- toms, or when some patients refuse this cumbersome procedure. In spite of this valuable scientific observa- tion, this method will not replace the full PSG of the diagnosis of OSA. The study of HRV using Holter ECG after CPAP treatment in OSA patients warrants further investigation. The main limitation of this test is that it has been limited only to small number of severe cases of OSA and controls and this was due to the stringent criteria we used for the selection of subjects. The power spec- tral densities in this study were not converted to nor- malised units so as to show the importance of the VLF component; this may be important in less severe cases of OSA, but it requires lengthy data processing. Further studies, using simple techniques on less severe cases, will be required. C O N C L U S I O N The LF band obtained using FFT for spectral analysis of HRV obtained from overnight Holter tachograms may *P= 0.0001 for difference between the LF band of OSA and snorer subjects. Figure 1: Very low ( VLF), low (LF) and high (HF) frequency bands from polysomnography and Holter ECGs in snorers with and without obstruc- tive sleep apnea (OSA) S C R E E N I N G O F PAT I E N T S W I T H S N O R I N G A N D O B S T R U C T I V E S L E E P A P N O E A U S I N G H E A R T R AT E VA R I A B I L I T Y I N D I C E S 25 be a useful and cost effective test in identifying snorers with severe OSA. R E F E R E N C E S 1. Guilleminault C, Tilkian A, Dement WC. The sleep ap- nea syndromes. Ann Rev Med 1976; 27:465-485. 2. Leung STR, Bradley D. Sleep apnea and cardiovascular disease. Am J Respir Crit Care Med 2001; 164:2147- 2165. 3. Dyken ME, Somers VK, Yamada T, Ren ZY, Zimmerman MB. Investigating the relationship between stroke and obstructive sleep apnea. Stroke 1996; 27:401–407. 4. Young T, Peppard PE, Gottlieb DJ. Epidemiology of ob- structive sleep apnea. Am J Respir Crit Care Med 2002; 165:1217-1239. 5. Taha BH, Dempsey JA, Weber SM, Badr MS, Skatrud JB, Young TB, et al. Automated detection and classification of sleep-disordered breathing from conventional polys- omnography data. Sleep 1997; 20:991-1001. 6. Kirby SD, Eng P, Danter W, George CF, Francovic T, Ruby RR, et al. Neural network prediction of obstructive sleep apnea from clinical criteria. Chest 1999; 116:409-415. 7. Pagani M, Montano N, Porta A, Malliani A, Abboud FM, Birkett C, et al. The relationship between spectral com- ponents of cardiovascular variabilities and direct meas- ures of muscle sympathetic nerve activity in humans. Circulation 1997; 95:1441-1448. 8. van Ravenswaaij-Arts CM, Kollee LA, Hopman JC, Stoe- linga GB, van Geijn HP. Heart rate variability. Ann Intern Med 1993; 118:436-447. 9. Narkiewicz K, Montano N, Cogliati C, van de Borne PJH, Dyken ME, Somers VK. Altered cardiovascular variabil- ity in obstructive sleep apnea. Circulation 1998; 98:1071- 1077. 10. Keyl C, Lemberger P, Pfeifer M, Hochmuth K, Geisler P. Heart rate variability in patients with daytime sleepiness suspected having sleep apnea syndrome: a receiver-oper- ating characteristic analysis. Clin Sci 1997; 92:335-343. 11. Somers VK, Dyken ME, Clary MP, Abboud FM. Sympa- thetic neural mechanisms in obstructive sleep apnea. J Clin Invest 1995; 96:1897-1904. 12. Stein PK, Duntley SP, Domitrovich PP, Nishith P, Carney RM. A simple method to identify sleep apnea using Hol- ter recordings. J Cardiovasc Electrophysiol 2003; 14:467- 473. 13. Akselrod S, Gordon D, Hubel FA, Shannon DC, Barger AS, Cohen RJ. Power spectrum analysis of heart rate variability; a quantitative probe of beat-to-beat cardio- vascular control. Science 1981; 213:220-222. 14. Parati G, Saul P, Di Rienxo M, Manicia G. Spectral analy- sis of blood pressure and heart rate variability in evalu- ating cardiovascular regulation. A critical appraisal. Hy- pertension 1995; 25:1276-1286. 15. Kamath MV, Fallen EL. Power spectral analysis of heart rate variability: a noninvasive signature of cardiac auto- nomic function. Crit Rev Biomed 1993; 21:245-311. 16. Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysi- ology. Heart rate variability: standards of measurement, physiological interpretation and clinical use. Circulation 1996; 93:1043-1065. 17. Macor F, Fagard R, Vanhoof R, Staessen J, Thijs L, Am- ery A. Power spectral analysis of short-term RR interval and blood pressure variability: Comparison of different methods and assessment of reproducibility. High Blood Pressure 1994; 3:15-21. 18. Hossen A, Al-Ghunaimi B, Hassan MO. A new simple algorithm for heart rate variability analysis in patients with obstructive sleep apnea and normal controls. Int J Bioelectromagetism 2003; 5:238-239. 19. Shiomi T, Guilleminault C, Sasanabe R, Hirota I, Maekawa M, Kobayashi T. Augmented very low frequen- cy component of heart rate variability during obstructive sleep apnea. Sleep1996; 19:370-377. 20. Hori T, Sugita Y, Koga E, Shirakawa S, Inoue K, Uchida S, et al. Sleep Computing Committee of the Japanese Soci- ety of Sleep Research Society. A manual of standardized terminology techniques and scoring system for sleep stages of human subjects, the Rechtschaffen and Kales 1968 Standard. Psychiatry Clin Neurosci 2001; 55:305- 310. 21. Pan J, Tompkins WJ. Quantitative investigation of QRS detection rules using the MITBH/I arrhythmia database. IEEE Trans. Biomed Eng 1986; 33:1157-1187. 22. Cuiwei L, Chongxun Z, Changfeng T. Detection of ECG characteristic points using wavelet transformation. IEEE Trans Biomed Eng 1995; 42:21-28.