Skip Navigation


Europace Advance Access originally published online on October 29, 2008
Europace 2009 11(1):94-99; doi:10.1093/europace/eun285
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
11/1/94    most recent
eun285v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Szyszko, A.
Right arrow Articles by Gonzalez-Zuelgaray, J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Szyszko, A.
Right arrow Articles by Gonzalez-Zuelgaray, J.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

Published on behalf of the European Society of Cardiology. All rights reserved. © The Author 2008. For permissions please email: journals.permissions@oxfordjournals.org

Sleep Apnoea and Arrhythmias

Reliability of a Holter-based methodology for evaluation of sleep apnoea syndrome

Ariel Szyszko1,*, Carlos Franceschini2 and Jorge Gonzalez-Zuelgaray1

1 Service of Arrhythmias and Electrophysiology, Instituto Argentino de Diagnostico y Tratamiento, Marcelo T. de Alvear 2346, 1122 Buenos Aires, Argentina; 2 Department of Medicine, Argerich Hospital, Almirante Brown 250, 1155 Buenos Aires, Argentina

Manuscript submitted 31 May 2008. Accepted after revision 22 September 2008.

* Corresponding author. Tel: +5411 4963 9500 ext. 329. E-mail address: aes1974{at}gmail.com


   Abstract
Top
 Abstract
Introduction
 Methods
 Results
 Discussion
 Limitations of the study
 Conclusion
 Funding
 Acknowledgements
 References
 
Aims: Sleep apnoea has significant medical implications. A reliable non-invasive method (as a regular Holter system with a specific software) would be valuable for the screening of this condition in ambulatory patients.

Methods and results: A total of 40 patients were divided into two groups: Group I, 20 patients with clinical suspicion of obstructive sleep apnoea (OSA) and Epworth sleepiness score ≥ 10 and Group II, 20 controls. In Group I, polysomnography was performed simultaneously with Holter (specific software to detect sleep apnoea). In Group II, Holter-based detection was utilized. A cutoff value of 10 for the apnoea–hypopnoea index (for polysomnography) or for the respiratory disturbance index (RDI) (for Holter) was considered abnormal. Sleep apnoea was confirmed by polysomnography in 14 patients (70%) in Group I. Holter recordings correctly identified OSA in 11 patients (r = 0.74 with polysomnography; P = 0.0002). Holter showed 78.5% sensitivity, 83.3% specificity, 91.6% positive predictive value, and 62.5% negative predictive value (with polysomnography as the gold standard). The RDI measured by Holter was 19.5 ± 20 in Group I and 3.9 ± 4.4 in controls (P < 0.005). The measurement between Holter and polysomnography (Bland and Altman method) showed good correlation (mean 4.7 with 39.4 and –30.1 SD) and a Pearson correlation coefficient (r) of 0.74 (P = 0.0002, 95% CI: 0.44–0.89).

Conclusion: Holter-based software may constitute an accessible tool on initial suspicion of OSA.

Key Words: Holter, Diagnosis, Sleep apnoea syndrome


   Introduction
Top
 Abstract
 Introduction
Methods
 Results
 Discussion
 Limitations of the study
 Conclusion
 Funding
 Acknowledgements
 References
 
Polysomnography is the accepted methodology for the diagnosis of obstructive sleep apnoea (OSA),1Go a potentially life threatening disorder with an estimated prevalence of 2–4% in the general population, which increases to 25–40% in patients with heart failure.2Go Despite evidence showing that untreated OSA threatens public health, 93% of women and 82% of men with moderate to severe OSA remain undiagnosed.3Go Polysomnographic recordings are expensive and include full-night evaluation, at a cost of ~$1100 (719 {euro}) per procedure,4Go limiting their utility for initial screening. The limited accessibility to polysomnography in combination with the low suspicion of OSA among physicians and the general public has resulted in the failure to diagnose and treat correctly the majority of patients. Conservatively, ‘~2310 polysomnograms per 100 000 people per year would be required to adequately address the demand for diagnosis and treatment of patients with suspected OSA of at least moderate severity’.5Go This exceeds by a factor of 10 the actual capacity for polysomnography in most countries and underscores the necessity of other reliable and cost-effective methods to detect the population at risk of OSA and its potential consequences and comorbidities.6Go–10Go

Guilleminault et al.11Go postulated that the pattern of heart rate variability (HRV) could be used to diagnose this syndrome. The diagnosis of OSA based on Holter recordings has been suggested by several authors.12Go–14Go

We evaluated a Holter-based methodology that consists of a new software for the detection of OSA in ambulatory patients based on an analysis of HRV and R-wave amplitude.


   Methods
Top
 Abstract
 Introduction
 Methods
Results
 Discussion
 Limitations of the study
 Conclusion
 Funding
 Acknowledgements
 References
 
The protocol was approved by the Institutional Review Board of our institution and the principles of the Declaration of Helsinki were carefully respected. Forty patients who signed an informed form consent were included. In Group I, 20 patients (10 males, aged 49 ± 14.6 years, range 20–76) were included because of high suspicion of OSA, based on the presence of two or more of the following: history of hypertension (50%), smoking, craniofacial dismorphy, family history of OSA, endocrinological disorders, body mass index (BMI) ≥ 30 kg/m2, neck circumference >17 in. for men and 16 in. for women,15Go or an adjusted neck circumference >43 cm if the patient was hypertensive (4 cm was added),16Go and an Epworth sleepiness score (ESS) ≥ 10.17Go This score is derived from a questionnaire designed to quantify daytime sleepiness. A value of 0–9 is considered normal, while a score of 10–24 justifies seeking the advice of a sleep specialist. The ESS (with specificity and sensitivity of 100 and 93.5%, respectively) has been validated primarily in patients with OSA, and is repeated in clinical practice to assess symptomatic improvement after institution of therapy.

Group II was composed of 20 controls (aged 27.3 ± 9.2 years, range 7–52 years) with no suspicion of OSA (absence of hypertension, diabetes, treatment with hypnotics, tobacco or alcohol consumption, BMI below 30 kg/m2, history of disorders of the upper respiratory tract, cardiopulmonary or neuromuscular disease, and usual physical activity at least two times per week).

Patients with ongoing pathological states, autonomic disorders, history of cardiac arrhythmias, or with implanted permanent pacemakers or cardioverter defibrillators were excluded.

In Group I, a polysomnographic study was carried out at night in the laboratory of sleep disorders with a standard software (ATI Delphos PSG 2003) simultaneously with a Holter analysis (for a minimum of 4 h), with a specific software for OSA detection based on the analysis of HRV and modifications of R-wave amplitude (Del Mar Reynolds Lifescreen Version 2.59, Hertford, UK). This method, consisting of a software with 12-bit resolution, is explained in great detail by De Chazal et al.18Go as follows: ‘a QRS detection time is defined as the time of occurrence of the QRS complex in an ECG signal. QRS detection times were generated automatically. Beat to beat (RR) intervals are defined as the intervals between successive QRS detection points. During the breathing cycle, the body-surface ECG is influenced by an electrode motion relative to the heart and by changes in thoracic electrical impedance resulting from filling and emptying of the lungs. The original ECG signal is processed with a median filter of 200 ms width to remove QRS complexes and P-waves. The resulting signal is then processed with a median filter of 600 ms width to remove T-waves. The signal resulting from the second filter operation contains the baseline of the ECG signal, which is then subtracted from the original signal to produce the baseline-corrected ECG signal. A sample point of an ECG-derived respiratory signal (EDR) is obtained by calculating the area enclosed by the baseline-corrected ECG in the region 100 ms beyond the QRS detection point. The RR set contains 52 features derived from the RR intervals, whereas the EDR set contains 36 features’. The RR set provides information about heart beat timing, whereas the EDR set provides information about chest wall movement. The combined set feature results in a better classification performance than either set alone.

Representative examples of Holter recordings in patients from each group are shown in Figures 1 and 2.


Figure 1
View larger version (103K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Figure 1 The Holter analysis in a control patient (Group II) shows predominance of the green vertical lines (probability of sleep apnoea below 50%), with a respiratory disturbance index of 3.6. Although the system automatically reports an apnoea–hypopnoea index, we consider that respiratory disturbance index would be a more accurate term, as explained in the discussion.

 


Figure 2
View larger version (103K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Figure 2 Holter analysis showing predominance of red vertical lines (probability of sleep apnoea over 50%), with a respiratory disturbance index of 50.1 in a patient from Group I. As in Figure 1, the same considerations regarding the use of the terms apnoea–hypopnoea index vs. respiratory disturbance index also apply.

 
The physicians in charge of the analysis of either the polysomnography or the Holter recordings were blinded regarding the findings of the other method and only the times of onset and end of sleep were known to both. In Group II, only Holter-based detection was utilized, as polysomnographic studies were not accepted by the volunteers.

A cutoff test value of ≥10 for the apnoea–hypopnoea index (AHI) (total number of apnoeas and hypopnoeas per hour of sleep time) was considered abnormal. For the calculation of the AHI in polysomnographic recordings, hypopnoea was defined by a reduction of at least 50% in airflow with desaturation of ≥4% and/or arousal during 10 or more seconds (sudden increase in the frequency observed in the electroencephalographic recordings). Apnoea was defined as cessation of oronasal airflow for a minimum of 10 or more seconds. In Holter devices, the quantification of the respiratory disturbance index (RDI)19Go was automated and the same cutoff value of ≥10 was established to discriminate between diseased and non-diseased status. In fact, portable devices quantify ‘respiratory disturbances’ (not distinguishing between apnoeas and hypopnoeas) and use total monitoring time to construct the RDI. The RDI measured by a portable device and the AHI determined by polysomnography are commonly used.19Go

Statistical analysis
Polysomnography with a cutoff value of AHI ≥ 10 was the gold standard for identifying OSA, and the same value was used for Holter-based determination of the RDI. Continuous variables are expressed as mean ± SD. Differences in AHI/RDI scores between both methods were assessed according to Bland and Altman.20Go The receiver operating characteristic (ROC) curves were used to test the diagnostic accuracy of the detection of the sleep apnoea syndrome. Statistical significance was assumed at a value of P < 0.05. Sensitivity and specificity were estimated both in the diseased and non-diseased status with their simultaneous exact 95% CI for sensitivity and specificity, and the Pearson correlation coefficient (r) was estimated. The analyses were performed with a commercially available statistical package (SPSS for Windows 11.1).


   Results
Top
 Abstract
 Introduction
 Methods
 Results
Discussion
 Limitations of the study
 Conclusion
 Funding
 Acknowledgements
 References
 
The mean age in Group I was 48.7 ± 14.6 years, with a BMI of 41.3 ± 11.7 kg/m2. Fifty per cent of the patients were hypertensive, 15% were diabetic (type II), 40% were on β-blocker therapy, 60% received inhibitors of the angiotensin-converting enzyme or angiotensin receptor-blockers, and 10% were treated with amlodipine or thiazides (as a second drug). The oxygen saturation in polysomnography was 93–74% (mean 84.4 ± 5.1%). The desaturation index per hour of sleep was 124.1–2.5 (mean 22.2 ± 32.7). The average maximal and minimal heart rates were, respectively, 87.2 ± 11.9 and 54.6 ± 9.5 bpm. The mean RDI measured by Holter was 19.5 ± 20 and the AHI measured by polysomnograpy was 24.2 ± 26.2 (Table 1). Obstructive sleep apnoea (AHI ≥ 10) was confirmed by polysomnographic recordings in 14 of 20 patients (70%) in Group I. The average age and BMI of patients with diagnosed OSA (AHI 25.4 ± 21.1) were greater than those without confirmed OSA (AHI 4.3 ± 4.7): age 52.5 ± 15.6 vs. 30.1 ± 9.9 years and BMI 41.2 ± 13.2 vs. 26.8 ± 9.2 kg/m2, respectively (P < 0.01 for both comparisons).


View this table:
[in this window]
[in a new window]

  		Table 1 Results in both groups

 
Holter correctly identified 11 of 14 patients in Group I (78.5%). The comparative analysis between Holter and polysomnography showed a sensitivity of 78.5% (95% CI: 48.2–94.2), a specificity of 83.3% (95% CI: 25.8–89.7), a positive predictive value of 91.6% (95% CI: 59.7–99.5), and a negative predictive value of 62.5% (95% CI: 25.8–89.7). The positive and negative likelihood ratios were 4.71 (95% CI: 0.77–28.8) and 0.26 (95% CI: 0.09–075), respectively.

The measurement between Holter and polysomnography (according to Bland and Altman's method) showed a good agreement, with a difference between both measurements of 4.7 (mean) with 39.4 and –30.1 SD (Figure 3). The area under the ROC curve was 0.81 (95% CI: 0.6–1.0). The Pearson correlation coefficient was (r) 0.74 with a level of significance P = 0.0002 (95% CI: 0.44–0.89), indicating that both measurements are related. As an important finding, the RDI analysed by Holter was normal in all patients in Group II (3.9 ± 4.4) with a statistically significant difference with the values obtained in Group I (19.5 ± 20) (P < 0.005).


Figure 3
View larger version (18K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Figure 3 Graphic representation of the measured agreement between polysomnography and Holter-based determination according to Bland and Altman's method.

 

   Discussion
Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
Limitations of the study
 Conclusion
 Funding
 Acknowledgements
 References
 
Guilleminault et al.11Go described a strong relationship between the longest apnoeic events in OSA and the greatest swings in RR intervals, illustrating what they called ‘cyclical variations in heart rate’. These authors also proposed this feature as a screening tool in OSA. Typically, cyclical variations consist of bradycardia during apnoea followed by tachycardia upon its cessation. The detection of episodes based on ‘cyclical variations of heart rate’ was possible by plotting RR intervals vs. time.

The time-domain analysis of HRV appears to be a powerful tool to detect OSAs. Roche et al.14Go used time-domain analysis of HRV, with 90% sensitivity for the diagnosis of OSA. Le Heuzey et al.21Go reported a greater visual enlargement of the R-R trend and a significant difference between minimal and maximal heart rates during the night in patients with OSA when compared with control subjects. In their study, the measurement of variations of the RR intervals was not automated. Quantification of RR-interval variability is currently available in Holter recording systems, allowing the calculation of HRV scores.

In our study, the quantification of RR intervals was automated, as it is currently available in Holter devices, thus permitting the calculation of RDI. The Lifescreen software shows the R-R tachogram, R-R histogram, and the AHI.

The physiological basis of the observed abnormalities lies in the struggle against upper-airway obstruction, with an initial increase in the parasympathetic drive, followed by an abrupt sympathetic activation caused by hypoxia. The alternation between strong successive parasympathetic/sympathetic drives dramatically enhances R-R variability. In contrast, the diurnal HRV is characteristically blunted in patients with OSA.

However, some limitations are still present. First, clinical disorders such as diabetes mellitus, sequelae of myocardial infarction, and chronic heart failure, with reduced autonomic reactivity, are often associated with OSA. These entities may lead to false negative results based on Holter. In our study, three diabetic patients in Group I had both abnormal Holter and polysomnographic recordings, but there was no evidence of autonomic disturbances in any of them. Although we speculate that this method could likely be reliable in the absence of severe autonomic impairment, further trials with a greater number of patients are needed to determine its real utility for the diagnosis of OSA in this population.

Differences in sleep efficiency not detectable by Holter-based methodology can generate an underestimation of the RDI when compared with the values of AHI established by polysomnography. In the present study, this occurred in one patient in Group I, in whom there was an RDI of 62.2 vs. an AHI of 119 measured by polysomnograpy; this constitutes a major difference—still with concordance—probably attributable to the lower accuracy of Holter devices in detecting events.

Also, artefacts in Holter recordings would corrupt the measurement of the R-wave amplitude. Although this did not happen in our study, it would be desirable in newer software developments to include the ability to revert to screening of ‘RR interval only’ in case of artefacts.

Scharf et al.22Go provided information about 22 patients with regard to the analysis of transthoracic impedance signals in pacemakers with minute ventilation sensors. These records were used to score apnoea/hypopnoea events and, when compared simultaneously with polysomnography, there was agreement in 81% between both diagnostic approaches (P = 0.001). Clearly, the diagnosis by polysomnography is the gold standard and should be used to confirm this disease, as this method is able to quantify the severity of desaturation and provides additional information of value to define sleep apnoea. Portable devices can be used to suspect the existence of OSA and, in the presence of a positive result, the patient may be referred to a sleep specialist for further evaluation and appropriate treatment.


   Limitations of the study
Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
 Limitations of the study
Conclusion
 Funding
 Acknowledgements
 References
 
The small number of patients represents the main limitation of this feasibility study. We did not discriminate between hypopnoea and apnoea events because the physiological consequences are similar in both. Finally, polysomnography was not carried out in the control group as this requires night admission (not easily acceptable to volunteers); nevertheless, the RDI measured by Holter recordings was normal in all cases in Group II.


   Conclusion
Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
 Limitations of the study
 Conclusion
Funding
 Acknowledgements
 References
 
A Holter-based software with the analysis of HRV and modifications of the R-wave amplitude may constitute an easily accessible method for suspicion of OSA in the general population. Larger studies are underway to confirm these results.


   Funding
Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
 Limitations of the study
 Conclusion
 Funding
Acknowledgements
 References
 
This study was funded by the Programa de Estudio y Tratamiento de las Arritmias Cardíacas (PRONETAC).


   Acknowledgements
Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
 Limitations of the study
 Conclusion
 Funding
 Acknowledgements
References
 
The authors thank Dr Conor Heneghan for his valuable contributions.

Conflict of interest: none declared.


   References
Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
 Limitations of the study
 Conclusion
 Funding
 Acknowledgements
 References
 
[1] Kushida CA, Littner MR, Morgenthaler T, Alessi CA, Bailey D, Coleman J Jr, et al. Practice parameters for the indications for polysomnography and related procedures: an update for 2005. Sleep (2005) 28:499–521.[Web of Science][Medline]

[2] Bradley TD, Floras JS. Sleep apnea and heart failure. Part II: central sleep apnea. Circulation (2003) 107:1822–6.[Free Full Text]

[3] Young T, Blustein J, Finn L, Palta M. Sleep-disordered breathing and motor vehicle accidents in a population-based sample of employed adults. Sleep (1997) 20:608–13.[Web of Science][Medline]

[4] Iber C, O'Brien C, Schluter J. Single night studies in obstructive sleep apnea. Sleep (1991) 14:383–5.[Web of Science][Medline]

[5] Flemons WW, Douglas NJ, Kuna ST, Rodenstein DO, Wheatley J. Access to diagnosis and treatment of patients with suspected sleep apnea. Am J Resp Crit Care Med (2004) 169:668–72.[Free Full Text]

[6] Peppard PE, Young T, Palta M, Skatrud J. Prospective study of sleep-disorder breathing and hypertension. N Engl J Med (2000) 342:1378–84.[Abstract/Free Full Text]

[7] Coleman J. Complication of snoring, upper airway resistance syndrome and obstructive sleep apnea in adults. Otolaryngol Clin North Am (1999) 32:223–34.[CrossRef][Web of Science][Medline]

[8] Narkiewicz K, Motano N, Cogliati C, van de Borne PJ, Dyken ME, Somers VK. Altered cardiovascular variability in obstructive sleep apnea. Circulation (1998) 98:1071–7.[Abstract/Free Full Text]

[9] Hung J, Withford EG, Parsons EW, Hillman DR. Association of sleep apnea with myocardial infarction in men. Lancet (1990) 336:261–4.[CrossRef][Web of Science][Medline]

[10] Partinen M, Guilleminault C. Daytime sleepiness and vascular morbidity at seven year follow-up in obstructive sleep apnea patients. Chest (1990) 97:27–32.[Abstract/Free Full Text]

[11] Guilleminault C, Connolly S, Winkle R, Melvin K, Tilkian A. Cyclical variation of heart rhythm in sleep apnea syndrome. Mechanism and usefulness of 24 h electrocardiography as screening technique. Lancet (1984) 1:126–31.[CrossRef][Web of Science][Medline]

[12] Ichimaru Y, Yanaga T. Frequent characteristics of heart rate variability produced by Cheyne–Stokes during 24-h electrocardiographic monitoring. Comput Biomed Res (1989) 22:225–33.[CrossRef][Web of Science][Medline]

[13] Roche F, Pichot V, Sforza E, Court-Fortune I, Duverney D, Costes F, et al. Predicting sleep apnea from heart period: a time-frequency wavelet analysis. Eur Respir J (2003) 22:937–42.[Abstract/Free Full Text]

[14] Roche F, Gazpoz JM, Court-Fortune I, Minini P, Pichot V, Duverney D, et al. Screening of obstructive sleep apnea syndrome by heart rate variability analysis. Circulation (1999) 100:1411–5.[Abstract/Free Full Text]

[15] Skomro RP, Kryger MH. Clinical presentations of obstructive sleep apnea syndrome. Prog Cardiovasc Dis (1999) 41:331–40.[CrossRef][Web of Science][Medline]

[16] Flemons WW. Obstructive sleep apnea. N Engl J Med (2002) 347:498–504.[Free Full Text]

[17] Johns MW. A new method for measuring daytime sleepiness: the Epworth sleepiness scale. Sleep (1991) 14:540–5.[Web of Science][Medline]

[18] De Chazal P, Heneghan C, Sheridan E, Reilly R, Nolan P, O'Malley M. Automated processing of the single lead electrocardiogram for the detection of obstructive sleep apnea. IEEE Trans Biomed Eng (2003) 50:686–96.[CrossRef][Web of Science][Medline]

[19] Flemons WW, Littner MR. Measuring agreement between diagnostic devices. Chest (2003) 124:1535–42.[Abstract/Free Full Text]

[20] Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet (1986) 1:307–10.[CrossRef][Web of Science][Medline]

[21] Le Heuzey JY, Romejko P, Fleury B, Scala PJ, Derenne JP, Guize L, et al. Holter monitoring in the diagnosis of sleep apnea syndrome (Abstract). J Am Coll Cardiol (1989) 13:189A.

[22] Scharf C, Cho YK, Bloch KE. Diagnosis of sleep-related breathing disorders by visual analysis of transthoracic impedance signals in pacemakers. Circulation (2004) 110:2562–7.[Abstract/Free Full Text]


Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?



This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
11/1/94    most recent
eun285v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Szyszko, A.
Right arrow Articles by Gonzalez-Zuelgaray, J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Szyszko, A.
Right arrow Articles by Gonzalez-Zuelgaray, J.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?