Cardiac Arrhythmias, Snoring, andSleep ApneaVictor Hoffstein, M.D., Ph.D.; and Susan Mateika, B.Sc.

We investigated the frequency of cardiac arrhythmiasin patients suspected of having sleep apnea, and relatedthem to the severity of apnea, snoring, and nocturnalhypoxemia. We prospectively studied 458 patients whohad nocturnal polysomnography which included objec-tive measurement of snoring (quantified by the numberof snores per hour of sleep [snoring index (SI)] and max-

imum nocturnal sound intensity [(dBmax)], as well as

examination of the electrocardiogram (modified lead 2).We found 58 percent prevalence of arrhythmias in pa-

tients with sleep apnea (apnea/hypopnea index=AHI>10), vs 42 percent in nonapneic controls (x2=16.7,p<0.0001). Patients with arrhythmias had more severe

apnea and nocturnal hypoxemia, but not snoring, thanpatients without arrhythmias. To examine separatelythe relationship between the prevalence of arrhythmiasand snoring, nocturnal oxygenation, and apnea-we

selected subgroups of patients "at the opposite ends ofthe spectrum" with respect to the severity of snoring,hypoxemia, and apnea. We found that 38 percent of lightsnorers had arrhythmias vs 39 percent of heavy snorers,

82 percent of patients with mean nocturnal oxygen sat-

Several studies show a linkage, and perhaps a causalassociation, between sleep apnea and heart dis-

ease.1'6 There is evidence that patients with sleepapnea have higher mortality from cardiovascularcauses than nonapneic individuals. One possibilitythat might explain this association may be increasedprevalence of cardiac arrhythmias in apneic patients.Although there are several investigations dealingwith the mechanisms of altered cardiac rhythm inpatients with recurrent episodes of upper airwayobstruction and hypoxemia at night,5'7-9 there is onlyone large scale clinical study dealing with descriptionof cardiac arrhythmias in patients with obstructivesleep apnea.10 Neither that study nor some of theother ones,11 involving considerably smaller numberof patients, however, had a control group composedof nonapneic individuals. Furthermore, associationbetween snoring alone, without apneas, and cardiacarrhythmias has not been examined, despite epide-miologic evidence identifying snoring as a risk factorfor coronary artery and cerebrovascular disease.12"3

uration <90 percent had arrhythmias vs 40 percent ofpatients with mean nocturnal oxygen saturation >90percent (x2=7.4, p=0.006), and 70 percent of patientswith AHI -40 had arrhythmias vs 42 percent with AHI<10 (X2=9.2, p=0.002). We conclude that patients withsleep apnea as a group have higher prevalence ofcardiac arrhythmias than nonapneic patients and thatsnoring alone, without concomitant sleep apnea, is notassociated with increased frequency of cardiac ar-

rhythmias. (Chest 1994; 106:466-71)

The purpose of the present investigation was tocompare the frequency of cardiac arrhythmias inpatients with and without sleep apnea and to exam-ine separately the relationships between the arrhyth-mias and the severity of apnea, hypoxemia, andsnoring.

METHODSPatients

We studied 458 consecutive and unselected patients referred tothe sleep clinic at St. Michael's Hospital for evaluation of possi-ble sleep apnea. Snoring was the chief complaint in the vast ma-jority of patients, although daytime sleepiness and tiredness werealso common.

Polysomnographic Measurements

In all 458 patients, we carried out nocturnal polysomnography,which included measurements of EEG, submental, and anteriortibial EMG, oronasal flow using thermistors, chest wall andabdominal excursions using inductance plethysomography, oxy-gen saturation using finger oximeter, and single-lead electrocar-diogram using modified precordial lead electrode. Snoring wasmeasured in 388 patients using a microphone-sound meter systemas described before. 14 In short, it was done by placing the micro-phone above the nasion (on the forehead), connecting the micro-phone to the sound meter (model SI120, Pacer Industries; Chip-pewa Falls, Wis) calibrated between 40 and 100 decibels (dB) anddisplaying sound intensity along with other polysomnographicvariables on the strip chart recorder. With this setup normalbreathing registers at less than 50 dB; spikes in sound intensityhigher than 50 dB are counted as snores.

Cardiac Arrhythmias, Snoring, and Sleep Apnea (Hoffstein and Mateika)

AHI=apnea/hypopnea index;COMB=combined atrial andventricular arrhythmias; dB=decibel; dBmax=maximumnocturnal sound intensity; LoO2S=lowest nocturnal oxygensaturation; MnO2S=mean nocturnal oxygen saturation;NSR=normal sinus rhythm; OSA=obstructive sleep apnea;SI=snoring index; SUPRA=supraventricular arrhythmia;TST85%=percent of sleep time spent at oxygen saturationlower than 85 percent; VENT=ventricular arrhythmia

*From the Department of Medicine, St. Michael's Hospital, Tor-onto, Ontario, Canada.

Manuscript received August 30, 1993; revision accepted Decem-ber 9.Reprint requests: Dr. Hoffstein, St. Michael's Hospital, 30 BondStreet, Toronto, Ontario, Canada M5R JWS

466

Downloaded From: http://journal.publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/21698/ on 06/26/2017

All polysomnograms were scored in the usual manner. Apneasand hypopneas were identified as either complete or incomplete(greater than 50 percent) episodes of cessation of breathing last-ing longer than 10 s. The number of such episodes per hour ofsleep is termed apnea/hypopnea index (AHI). Similarly, snoreswere counted and the number of snores per hour of sleep is termedsnoring index (SI); maximum nocturnal sound intensity (dBmax)was also recorded. Three indices of nocturnal oxygenation wereused: lowest nocturnal oxygen saturation (LoO2S), mean noctur-nal oxygen saturation (MnO2S), and percent of sleep time spentat oxygen saturation lower than 85 percent (TST85%). The lattermeasurement was available in 350 patients.

Cardiac rate and rhythm were assessed by examining themodified lead 2 electrocardiogram. Lowest, highest, and meanheart rates were recorded. Arrhythmias were classified accordingto their origin, as is usually done during standard polysomnogra-phy. Supraventricular arrhythmias were divided into sinus andatrial. The former included sinus bradycardia (rate less than40/min), sinus tachycardia (rate >120/min), sinus arrhythmia(irregular rhythm with rate between 60 and 100/min), and sinusarrest (pause lasting longer than 2 s). The latter includedwandering pacemaker, premature contractions, paroxysmal ta-chycardia, flutter, and fibrillation. Junctional rhythm, atrioven-tricular blocks, and bundle brunch blocks were identified andrecorded separately. Ventricular arrhythmias included prema-ture contractions, tachyeardia, flutter, and fibrillation.

Statistical AnalysisThe goal of statistical analysis was to determine whether there

is any relationship between the type of arrhythmia and theseverity of apnea, nocturnal hypoxemia, and snoring. This wasaccomplished as follows.

First, we grouped all patients into the following "arrhythmiagroups": (1) normal sinus rhythm (NSR), (2) supraventricular ar-rhythmia (SUPRA), which included sinus and atrial arrhythmias,(3) ventricular arrhythmia (VENT), which included prematureventricular contractions, atrioventricular block, and junctionalbeats, and (4) combined atrial and ventricular arrhythmias(COMB). We then used Kruskall-Wallis statistics to compare ap-nea (AHI), snoring (SI, dBmax), and oxygenation (LoO2S, MnO2Sand TST85%) among the different arrhythmia groups.

Second, we examined the same problem from a differentviewpoint, based on the hypothesis that patients with severe ap-nea, hypoxemia, and snoring are more likely to have cardiac ar-rhythmias than patients without apnea, hypoxemia, or snoring. Totest this hypothesis we selected subgroups of patients "at the op-posite ends of the spectrum" as far as the severity of their

Table 1-Anthropometric and Sleep Data

Variable* No. Mean + SD Range

Age 458 48+13 14-78Wt 458 89 ± 22 44-182BMI 458 31+ 7 17-65SI 388 384 +353 0-184dBmax 388 89 +10 35-110LoO2 458 81+12 10-95MnO2 458 92 +3 72-98TST85% 350 9 +15 0-88HR mean 458 68+10 42-114HR lowest 458 57 +11 18-95HR highest 458 85+ 14 51-200

*Wt=weight (kg), BMI=body mass index (kg/mi2), HR=nocturnalheart rate.

nocturnal events is concerned.To examine the effect of snoring, we compared nonapneic

nondesaturating light snorers (AHI -<10, SI <100, MnO2S >90%)with nonapneic nondesaturating heavy snorers (AHI -< 10, MnO2S>90%, SI >400).To examine the effect of nocturnal hypoxemia, we compared

nonapneic nonsnoring patients without nocturnal oxygen desat-uration (AHI -'10, SI <100, MnO2S >90%) with nonapneicnonsnoring patients who desaturate at night (AHI C 10, SI <100,MnO2S <90%).To examine the effect of apnea we compared nonapneic non-

desaturating nonsnorers (AHI -<10, MnO2S >90%, SI <100) withapneic nondesaturating nonsnorers (AHI --40, MnO2S >90%, SI<100).

Chi-square statistics with Bonferroni corrections were used tocompare the frequency of arrhythmias within each group. 1 Allstatistical analysis was done using SAS software (The SAS Institute;Gary, NC) version 6.04.

RESULTS

Table 1 summarizes the characteristics of the pa-tient population examined in this study. There were336 men and 122 women. Most patients (244/458)did not have obstructive sleep apnea (OSA) (AHI< 10), 121/458 (26 percent) had AHI between 10 and30, 41/458 (9 percent) had AHI between 30 and 50,

Table 2-List of Medications Used by All Patients

AHI -'10 10<AHI-'30 30<AHI-C50 AHI>50

No. 244 121 41 52Calcium channel blockers 15 4 5 3Beta blockers 12 9 2 3Diuretics 11 8 6 7ACE-inhibitors 9 2 2 4Digoxin 6 3 4 2Antiarrhythmics 2 2 1 0Antihypertensives 2 1 1 6Theophyllines 3 1 2 2Beta-agonists 16 2 2 3Tricyclic antidepressants 11 3 3 2Total 102 40 34 38

CHEST / 106/2/ AUGUST, 1994 467

Downloaded From: http://journal.publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/21698/ on 06/26/2017

Table 3-Cardiac Rhythm During Sleep

AHI<10 10<AHI'30 30<AHI'50 AHI>50

No. 244 121 41 52NSR 141 (58%) 56 (46%) 20 (49%) 13 (25%)SUPRA 54 (22%) 33 (27%) 9 (22%) 15 (29%)VENT 28 (11%) 19 (16%) 6 (15%) 10 (19%)COMB* 21 (9%) 13 (11%) 6 (15%) 14 (27%)ATR+AVB 0 0 0 1ATR+JUN 0 0 1 0ATR+PVC 11 4 2 5AVB+SIN 0 0 0 1PVC+SIN 9 8 2 4ATR+AVB+SIN 0 0 0 1ATR+PVC+SIN 1 1 1 0ATR+AVB+SIN+PVC 0 0 0 2

*This includes combination of the following arrhythmias: ATR=atrial, AVB=atrioventricular block, JUN=junctional rhythm, PVC=prematureventricular contractions, SIN=sinus arrhythmia.

and 52/458 patients (11 percent) had AHI >50.There were 82 patients with history of cardiac

disease (hypertension, 51; previous myocardial in-farction, 6; angina pectoris, 5; congestive heart fail-ure, 4; coronary artery disease, 13; arrhythmia, 3,and 10 patients with chronic obstructive lung disease.The general categories of medications used by thepatients are listed in Table 2. We note that the ma-jority of patients on medications did not have sleepapnea. Overall, there was no significant difference(by x2 statistics) between the proportion of apneicpatients taking medications as compared with non-apneic patients.Of the 458 patients, 228 had various conduction

abnormalities, and of those 125 had sleep apnea (ie,AHI >10). This results in 58 percent (125/214)prevalence of arrhythmias among patients with OSA,which was significantly higher than 42 percent(103/214) prevalence of arrhythmias among patientswithout OSA (X2= 12.0, p<0.001). It should be noted,however, that the prevalence of arrhythmias amongnonapneic patients was relatively high.

There were six patients with episodes of ventricu-lar tachycardia; all of them had significant sleep ap-nea (mean AH1=42) and hypoxemia (LoO2S=71percent, TST85%= 13 percent). None of the patientsrequired urgent interventions.The frequency of various arrhythmias is shown in

Table 3. We note that arrhythmias were significantlymore common (X2=13.4, p<0.005) among patientswith sleep apnea than among nonapneic controls. Ofthe 228 patients with arrhythmias, 174 had a singletype of arrhythmia (ie, atrial, sinus, ventricular, orjunctional), and 54 patients (33 apneic, 21 nonapneic)had combined arrhythmias, with two or more spe-cific disturbances occurring over the course of thenight (Table 3).

Figure 1 compares the mean values of AHI, SI,dBmax, MnO2, and TST85% in patients grouped into

the arrhythmia categories as defined above. Therewas a significant difference in all of these variablesexcept SI and dBmax between the groups, with Kur-skall-Wallis x2 ranging from 17 to 26, p<0.001.

Table 4 shows the effect of snoring, nocturnal hy-poxemia, and apnea on the prevalence of arrhyth-mias.

As may be seen from Table 4, to examine the ef-

APNEA AND SNORING

* AHI E SI D dBmax

100-

ME75-

50-

25-

0INOCTURNAL OXYGENATION

* AnO2sat LoO2sat DTST85%

100-

v I75I

~50-

CARDIAC RHYTHMFIGURE 1. Apnea, snoring, and nocturnal hypoxemia in patientsgrouped according to their arrhythmia.

Cardiac Arrhythmias, Snoring, and Sleep Apnea (Hoffstein and Mateika)468

Downloaded From: http://journal.publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/21698/ on 06/26/2017

Table 4-Subgroups of Patients Used to Examine the Effect of Apnea, Nocturnal Hypoxemia, andSnoring on Prevalence of Arrhythmias

% WithNo. AHI SI Mean 02S Arrhythmia

Effect of snoringSI<100 101 4+2 36+27 94+2 38SI>400 47 5 +3 644+ 246 93 + 2 39

Effect of nocturnal hypoxemiaMnO2S <90 11 5 + 2 409 + 588 88 + 2 82*MnO2S-<90 233 4+3 221 + 249 94 + 2 40

Effect of apneaAHI<10 238 4+3 229+273 94+2 42AHI'>40 33 59 +14 549 + 235 93 + 2 70*

*Significantly (p<0.01) higher than in the control group.

fect of snoring we selected nonapneic, nondesaturat-ing light and heavy snorers, based on the snoring in-dex of less than 100 or greater than 400. We foundno significant difference in the prevalence of arrhyth-mias between these two groups (X2=0.001, p=0.971).To examine the effect of nocturnal hypoxemia, we

should ideally compare nonapneic nonsnoring non-hypoxemic group with nonapneic nonsnoring hypox-emic group. Not unexpectedly, however, there wereno hypoxemic nonapneic nonsnorers-most hypox-emic patients either snore or have sleep apnea.Therefore, since snoring did not affect the prevalenceof arrhythmias anyway, we examined the effect ofhypoxemia on the frequency of cardiac arrhythmiasby comparing nonapneic nonhypoxemic group withnonapneic hypoxemic group without any restrictionon snoring. That is why the snoring index is relativelyhigh in both groups (Table 4), although the differ-ence was not statistically significant by the WilcoxonRank Sum Test (p=0.1787). Hypoxemia was definedbased on mean nocturnal oxygen saturation lowerthan 90 percent. The prevalence of supraventricularand ventricular arrhythmias was significantly higherin hypoxemic patients (x2=7.4, p=0.006).The selection of subgroups to examine the effect of

apnea on the prevalence of arrhythmias followed thesame general guidelines as was used for hypoxemia.We compared nonhypoxemic nonapneic patients(mean nocturnal oxygen saturation >90%, AHI. 10)with nonhypoxemic significantly apneic (AHI240)patients, without any restrictions on snoring. Theprevalence of arrhythmias, particularly the supraven-tricular and combined ones, was significantly higher(X2=9.2, p=0.002) in the apneic group.

DISCUSSION

This study shows that (1) patients with obstructivesleep apnea have higher prevalence of arrhythmiathan nonapneic patients, (2) patients without ar-rhythmias, as a group, have less severe apnea andnocturnal hypoxemia than patients with arrhyth-

mias, and (3) prevalence of arrhythmias is indepen-dent of snoring, but it is linked to sleep apnea andnocturnal hypoxemia. The study design has certainlimitations which could conceivably influence someof our conclusions.One of the drawbacks is the use of only a single lead

(modified lead 2) ECG. This would underestimatethe prevalence of abnormal axis, ST segment, T-wave,and QRS abnormalities, but it is not likely to affectthe detection of conduction abnormalities or calcu-lation of heart rate-which was the primary purposeof our study. Nevertheless, the absence of a full 12-lead ECG prevented us from more detailed classifi-cation of atrioventricular block and grading of pre-mature ventricular contractions, as was done in pre-vious investigations.10"1

Another drawback is that we did not examinenormal (ie, nonsnoring, nonapneic) controls takenfrom the healthy population. Although we identifiedsuch subgroup among our 458 patients, they all camefrom sleep clinic population. It is possible that a pa-tient found to be a nonapneic nonsnorer whensleeping in the laboratory on the night of the sleepstudy is a habitual snorer and has apneas whensleeping at home. This, however, is unlikely given thepublished results regarding the reproducibility ofapnea and snoring.On the other hand, our study has several unique

features not present in previous investigations: (1) itincludes a large group of subjects, all studied pro-spectively, (2) there were many nonapneic patientswho could serve as controls, and (3) we measuredsnoring and related it to arrhythmias without theconfounding effect of apnea and hypoxemia.

Only one large scale study, which could be com-pared with the present one, was performed in thepast.10 The authors examined 400 patients with OSAfor cardiac arrhythmias during sleep. They foundthat 48 percent of patients had cardiac arrhythmiarecorded during the sleep study night. There were atleast 75 patients (20 percent) who had ventricular

CHEST / 106 / 2 / AUGUST, 1994 469

Downloaded From: http://journal.publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/21698/ on 06/26/2017

arrhythmias. No control group was studied, althoughthere were 50 patients who served as their own con-trols and who improved after treatment with trache-otomy.

There is evidence suggesting that disturbances ofcardiac rhythm, particularly ventricular ectopy, inOSA patients may not be related to sleep at all, be-ing also present during wakefulness. Miller'l re-viewed the results of 24 h Holter monitoring in 23patients with OSA and found 78 percent prevalenceof arrhythmias during sleep. Since many patients alsohad cardiac arrhythmias during wakefulness, how-ever, the author concluded that the prevalence ofserious arrhythmias during sleep is low. Although hedid find a significantly higher prevalence of sinusarrhythmias during sleep compared with wakeful-ness, he was unable to show that for ventricular ar-rhythmias.Our results show a significantly higher prevalence

of cardiac arrhythmias, whether supraventricular orventricular, among patients with OSA. Unlike otherauthors,'0"' we found that atrioventricular block, si-nus pauses, and junctional rhythm were uncommonamong our patients, seen in only 2 to 3 percent ofthose with OSA, vs 8 to 11 percent found byothers.'0"'1

Although we found a statistically significant in-crease in the prevalence of cardiac arrhythmiasamong patients with OSA, it is important to note thatnonapneic patients in our study also had a highprevalence of these arrhythmias (42 percent), similarto an observation by Flemons et al.16 One possibilityis that it reflects our particular patient population, allof whom were referred because of sleep disturbanceand many of whom were on medications. Anotherpossibility, however, is that it simply reflects the factthat nocturnal arrhythmias are common in the gen-eral population, occurring in about 50 percent ofhealthy people.17-'9We believe that statistically significant increase in

arrhythmias among patients with OSA found in thecurrent study reflects their underlying disease. Thedata presented in Table 4 supports this hypothesis;furthermore, when we examined our patients withmost serious arrhythmias-bigeminy, trigeminy,quadrigeminy, sinus pauses, asystole-we found thatas a group they had most severe apnea (AHI>40) andhypoxemia (more than 19 percent of total sleep timespent at oxygen saturation lower than 85 percent).Our results provide indirect evidence, but not a

conclusive proof, that there is an independent rela-tionship between cardiac arrhythmias, apnea, andnocturnal hypoxemia. We consider our evidence in-direct because it is based on statistical analysis ofsubgroups matched for apnea and mean oxygen sat-uration, rather than on direct comparison of the fre-

quency of arrhythmias in the same group of apneichypoxemic subjects before and after elimination ofapnea and hypoxemia. There is other evidence sug-gesting that apnea and hypoxemia may have inde-pendent, perhaps an additive, adverse effect on car-diac function. For example, nonapneic patients withnocturnal hypoxemia due to chronic obstructive lungdisease have more ventricular ectopics duringsleep.20'2' Although not directly dealing with ar-rhythmias, there is evidence that hypoxemia alonedoes not explain increases in blood pressure seen insleep apnea,22 suggesting that apnea may have anadditional, independent effect of the regulation ofnocturnal sympathetic activity. Lastly, increasedsympathetic activity, due to hypoxemia, and in-creased vagal activity, due to repetitive episodes ofupper airway obstruction, may both have arrhyth-mogenic effect on the myocardium.23 It appears thatcombination of apnea and oxygen desaturation pro-vide sufficient stress on the heart to predispose thesepatients toward development of ventricular arrhyth-mias.

Measurements of snoring in our large group ofpatients permit us to comment on the possible asso-ciation between snoring and arrhythmias. This issueis currently of interest because of several studies in-dicating that snoring alone, without apneas, may bea risk factor for coronary artery and cerebrovasculardisease.'2"3 We found no difference in the preva-lence of arrhythmias between nonsnorers and heavysnorers. These observations imply that caution mustbe exercised in interpreting results of epidemiologicsurveys comparing cardiac function in snorers andnonsnorers; in the absence of sleep studies, significantdifferences between the two groups may be from"contamination" of the snoring group by individualswith sleep apnea and hypoxemia.We conclude that cardiac arrhythmias are more

common in patients with sleep apnea than in non-apneics, that snoring alone, without sleep apnea orhypoxemia, does not significantly affect cardiacrhythm, and that presence of apnea and nocturnalhypoxemia is associated with increased prevalence ofcardiac arrhythmias.

REFERENCES1 Shepard JW, Jr. Hypertension, cardiac arrhythmias, myocar-

dial infarction, and stroke in relation to obstructive sleep apnea.Clin Chest Med 1992; 13:437-58

2 Parish JM, Shepard JW, Jr. Cardiovascular effects of sleep dis-orders. Chest 1990; 97:1220-26

3 Koskenvuo M. Cardiovascular stress and sleep. Ann Clin Res1987; 19:110-13

4 Koehler U, Dubler H, Glaremin T, Junkerman H, Lubbers C,Ploch T, et al. Nocturnal myocardial ischemia and cardiac ar-rhythmia in patients with sleep apnea with and without coro-nary artery disease. Ktin Wochenschr 1991; 69:474-82

Cardiac Arrhythmias, Snoring, and Sleep Apnea (Hoffstein and Mateika)470

Downloaded From: http://journal.publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/21698/ on 06/26/2017

5 Stoohs R, Guilleminault C. Cardiovascular changes associatedwith obstructive sleep apnea. J Appl Physiol 1992; 72:583-89

6 Fletcher EC, Schaaf JW, Miller J, Fletcher JG. Long-termcardiopulmonary sequelae in patients with sleep apnea andchronic lung disease. Am Rev Respir Dis 1987; 135:525-33

7 Scharf SM, Graver LM, Balaban K. Cardiovascular effects ofperiodic occlusions of the upper airway in dogs. Am Rev RespirDis 1992; 146:321-29

8 Guilleminault C, Connolly S, Winkle R, Melvin K. Cyclicalvariation of the heart rate in sleep apnea syndrome. Lancet1984; 1:126-31

9 Andreas S, Hajak G, von Breska B, Ruther E, Kreuzer H.Changes in heart rate during obstructive sleep apnea. EurRespir J 1992; 5:853-57

10 Guilleminault C, Connolly SJ, Winkle RA. Cardiac arrhythmiaand conduction disturbances during sleep in 400 patients withsleep apnea syndrome. Am J Cardiol 1983; 52:490-94

11 Miller WP. Cardiac arrhythmia and conduction disturbances inthe sleep apnea syndrome. Am J Med 1982; 73:317-21

12 Koskenvuo M, Kaprio J, Partinen M, Langinvainio H, Sarna S,Heikkila K. Snoring as a risk factor for hypertension and anginapectoris. Lancet 1985; 1:893-95

13 Koskenvuo M, Kaprio J, Telakivi T, Partinen M, Heikkila K,Sarna S. Snoring as a risk factor for ischemic heart disease andstroke in men. Br Med J 1987; 294:16-9

14 Hoffstein V, Mateika S, Metes A. Effect of nasal dilation onsnoring and apneas during different stages of sleep. Sleep 1993;16:360-65

15 Altman DG. Practical statistics for medical research. London:Chapman & Hall, 1991; 259-61

16 Flemons WW, McDonald M, Remmers JE, Gillis A. Cardiacdysrhythmias and sleep apnea [abstract]. Am Rev Respir Dis1992; 145:A443

17 Sobotka PA, Mayer JH, Bauernfeind RA, Kanakis C, Jr., RosenKM. Arrhythmias documented by 24-hour continuous ambu-latory monitoring in young women without apparent heartdisease. Am Heart J 1981; 101:753-59

18 Brodsky M, Wu D, Denes P, Kanakis C, Rosen KM. Arrhyth-mias documented by 24-hour continuous electrocardiographicmonitoring in 50 male medical students without apparentheart-disease. Am J Cardiol 1977; 39:390-95

19 Viitasalo M, Halonen L, Partinen M, livanainen M, Kala R,Eisalo A. Sleep and cardiac rhythm in healthy men. Ann Med1991; 23:135-39

20 Douglas NJ. Nocturnal hypoxemia in patients with chronicobstructive pulmonary disease. Clin Chest Med 1992; 13:523-32

21 Flick MR, Block AJ. Nocturnal vs diurnal cardiac arrhythmiasin patients with chronic obstructive pulmonary disease. Chest1979; 75:8-11

22 Ringler J, Basner RC, Shannon R, Schwartzstein R, Manning H,Weinberger SE, et al. Hypoxemia alone does not explain bloodpressure elevations after obstructive apneas. J Appl Physiol1990; 69:2143-48

23 Verrier RL, Kirby DA. Sleep and cardiac arrhythmias. Ann NYAcad Sci 1988; 533:238-51

CHEST/ 106/2/ AUGUST, 1994 471

Downloaded From: http://journal.publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/21698/ on 06/26/2017