CONGESTWE HEART FAILURE
Reproducibility and Relation to Mean Heart Rate of Heart Rate Variability in Normal Subjects and in Patients with Congestive Heart Failure Secondary to Coronary Artery Disease Diederik Van Hoogenhuyze, MD, Norman Weinstein, MD, Gary J. Martin, MD, Jerry S. Weiss, MS, John W. Schaad, MS, X. Nader Sahyouni, MS, Dan Fintel, MD, Willem J. Remme, MD, and Donald H. Singer, MD
Before heart rate (HR) variability can be used for predictive purposes in the clinical setting, day-today variation and reproducibility need to be defined as do relations to mean HR. HR variability and mean HR were therefore determined in 2 successive 24-hour ambulatory electrocardiograms obtained from 33 normal subjects (age 34 f 7 years, group I), and 22 patients with coronary disease and stable congestive heart failure (CHF) (age 59 f 7 years, group ii). Three measures were used: (1) SDANN (standard deviation of ail mean i-minute normal sinus RR intervals in successive 5-minute recording periods over 24 hours); (2) SD (the mean of the standard deviation of ail normal sinus RR intervals in -essive S-minute recording periods over 24 hours); and (3) CV (coefficient of variation of the SD measure), a new measure that compensates for HR effects. Group mean HR was higher and HR variability lower in group Ii than in group I (60 f 10 vs 74 f 9 beats/min, p 2 However, becauseall reported HR variability values have been computed from single 24-hour ambulatory electrocardiographic recordings, the day-today variation is unknown. A correlation between mean HR and HR variability within study populations has been described,lT2but the impact of changesin HR on HR variability, both for groups and individual patients, also is not known. To gain insight into these problems we examined (1) the day-to-day variation and reproducibility of frequently used HR variability measurements assessedfrom 2 successive24-hour ambulatory electrocardiographic recordings in healthy young subjects and in patients with congestive heart failure
C
DECEMBER 15, 1991
(CHF) secondary to coronary artery disease,(2) the relation of HR variability and mean HR, and (3) intergroup differences in HR variability and mean HR in normal subjects and patients with CHF. METHODS Study group: Two population groups were studied: Group I consistedof 33 healthy volunteers, 19 men and 14 women without known cardiac disease,systemichypertension or diabetes mellitus. Group mean age was 34 years (range 24 to 54). All had normal results on physical examination and a normal 1Zlead electrocardiogram at rest. Contraceptives and occasionalnonsteroidal antiinflammatory agents were the only medications given. Group II consisted of 22 outpatients, 18 men and 4 women, aged 45 to 71 years (mean 59) with coronary artery diseaseand prior myocardial infarction complicated by stable New York Heart Association class II to III CHF, and treated with digoxin and diuretics. CHF was documentedby history, physical examination, chest x-ray and right-sided cardiac catheterization (cardiac index 15 mm Hg at rest, or both). Patients with diabetes and chronic obstructive lung diseasewere excluded, as were patients with atria1 fibrillation, sick sinus syndrome, atrioventricular block and paced rhythms. Data edlection: All subjectsunderwent two 24-hour ambulatory electrocardiographic recordings using Marquette model 8500 AM 2-channel recorders. Recordings were obtained on 2 consecutivedays for all group II patients (CHF). Recordings from 29 of 33 group I subjects (normal) also were obtained on 2 consecutive days. In the remaining 4 subjects,recordings were obtained 9 to 29 days apart. There were no interim changesin clinical state. Recordings were analyzed using a Marquette 8000 computerized Holter system, which classifies each beat as to site of origin. The beat classification was overread manually and corrected where necessary.Each RR interval during the 48-hour period was determined. The data were then transferred to Digital Equipment VAX computer and Cambridge Digital 11/73 systemsfor further analysis. The software used to measure HR variability was developedin our laboratory.3-6 It determinesthe mean f standard deviation (SD) of the sinus RR intervals for each 5-minute period during the recording period, and subsequently computes their mean value. In instances in which sinus rhythm is interrupted by ectopy, the program eliminates 1 RR interval preceding, and 2 RR intervals following, nonsinus beats. Three HR variability determinations were obtained for each subject: (1) the “SDANN” measure (or the
standard deviation of all mean 5minute “normal” sinus RR intervals, i.e., sinus RR intervals without intervening ectopic beats, over 24 hours2,3);(2) the “SD’ measure (defined as the mean of the standard deviations of all normal sinus RR intervals during successive 5-minute recording periods over 24 hours3); and (3) a new measure, the “CV’ measure (or coefficient of variation of the SD, which is obtained by dividing the SD value for each 5-minute period by the mean RR interval for that period, and then averaging all the 5minute CV values over 24 hours). Data analysis and statistical methods: The following analyseswere performed: First we testedday-to-day variation in, and reproducibility of, HR variability by comparing individual and mean group values for the 2 days. The results were expressedas mean f SD of the percent differencesbetween the days. The overall comparability of the two 24-hour recording periods was assessedby comparing mean RR for days 1 and 2 for each group. The Pearson product-moment correlation coefficient was then calculated to describeday 1 to day 2 relation for HR and HR variability. Second,we examined correlations between mean HR and HR variability for the 2 study groups and usedthe least-squares regression method to define the relation. The 2-way analysis of variance was used to test for differences in mean group values for the RR interval and HR variability, and for differences between men and women. Third, we compared mean HR and HR variability of normal subjects and patients with CHF. The SAS statistical package version 6.03 was used for all statistical analyses.7Normality of distribution of the data was tested using the Shapiro-Will< statistic. When normally distributed, data are presentedas mean f SD. Differences were considered significant at p 20%, the chi-square test for SD versusCV having an approximate p value of 0.05 and for CV versus SDANN p 100 to 30 to 130 ms.
Mean R-R
erable interindividual differences in SD and SDANN valuesfor given mean RR intervals, there was a moderate but significant correlation between the mean 24 hour RR interval and the SD and SDANN values (r = 0.64 and 0.54, respectively) (Figure 3). In contrast, CV values did not correlate with mean RR. GROUPII(PATIENTSWITHCONGESTIVEHEARTFAILURE):
Individual CHF patients with identical or comparable mean 24-hour HR often exhibited considerable differences in HR variability determined by SD and SDANN measures.Although group SD and SDANN values were lower for those with CHF than for normal subjects(Figure 4), personsin the 2 groups exhibited a similar range of HR variability. Relation between 24hour mean RR intervals and mean SD and SDANN values were similar to those in group I (r = 0.52 and 0.50, respectively) (Figure 3). Differences in mean heart rate and heart rate variability between group I and II subjects: Group II pa-
tients (CHF) had a higher mean HR than group I subjects (normal): average mean 48-hour RR interval = 770 f 94 and 8 14 f 93 ms for the 2 groups, respec-
SDANN
Day 2 (msec)
0
Relation between mean heart rate and hearl rate GROUP I (NORMAL SUBJECTS): Despite considvariability:
Day 2 (mrsc)
250
500
750
1000
1250
0
50
100
150
200
Day 1 (mssc)
Day 1 (msec)
SD
uro 1.
Day 2 (mrec)
00
20
40
60
80
0
2
Day 1 (mssc)
1672
FtGURE 2. Variabirty of mean RR interval, and heart rate variabilii determinations using SDANN, SD andCVmasureson2consecutive days in 22 patisnts with congestive heart fake (group II). Sse text for discussion. Abbreviations as in Fig-
THE AMERICAN JOURNAL OF CARDIOLOGY VOLUME 68
4
6
8
Day 1
DECEMBER 15, 1991
10
tively, but the differences did not reach significance (p g 0.09). Day 1 HR variability was significantly lower for patients with CHF than for normal subjects,irrespectiveof the measure used: SD 36 f 16 vs 76 f 25 ms (p
Reproducibility and Relation to Mean Heart Rate of Heart Rate Variability in Normal Subjects and in Patients with Congestive Heart Failure Secondary to Coronary Artery Disease Diederik Van Hoogenhuyze, MD, Norman Weinstein, MD, Gary J. Martin, MD, Jerry S. Weiss, MS, John W. Schaad, MS, X. Nader Sahyouni, MS, Dan Fintel, MD, Willem J. Remme, MD, and Donald H. Singer, MD
Before heart rate (HR) variability can be used for predictive purposes in the clinical setting, day-today variation and reproducibility need to be defined as do relations to mean HR. HR variability and mean HR were therefore determined in 2 successive 24-hour ambulatory electrocardiograms obtained from 33 normal subjects (age 34 f 7 years, group I), and 22 patients with coronary disease and stable congestive heart failure (CHF) (age 59 f 7 years, group ii). Three measures were used: (1) SDANN (standard deviation of ail mean i-minute normal sinus RR intervals in successive 5-minute recording periods over 24 hours); (2) SD (the mean of the standard deviation of ail normal sinus RR intervals in -essive S-minute recording periods over 24 hours); and (3) CV (coefficient of variation of the SD measure), a new measure that compensates for HR effects. Group mean HR was higher and HR variability lower in group Ii than in group I (60 f 10 vs 74 f 9 beats/min, p 2 However, becauseall reported HR variability values have been computed from single 24-hour ambulatory electrocardiographic recordings, the day-today variation is unknown. A correlation between mean HR and HR variability within study populations has been described,lT2but the impact of changesin HR on HR variability, both for groups and individual patients, also is not known. To gain insight into these problems we examined (1) the day-to-day variation and reproducibility of frequently used HR variability measurements assessedfrom 2 successive24-hour ambulatory electrocardiographic recordings in healthy young subjects and in patients with congestive heart failure
C
DECEMBER 15, 1991
(CHF) secondary to coronary artery disease,(2) the relation of HR variability and mean HR, and (3) intergroup differences in HR variability and mean HR in normal subjects and patients with CHF. METHODS Study group: Two population groups were studied: Group I consistedof 33 healthy volunteers, 19 men and 14 women without known cardiac disease,systemichypertension or diabetes mellitus. Group mean age was 34 years (range 24 to 54). All had normal results on physical examination and a normal 1Zlead electrocardiogram at rest. Contraceptives and occasionalnonsteroidal antiinflammatory agents were the only medications given. Group II consisted of 22 outpatients, 18 men and 4 women, aged 45 to 71 years (mean 59) with coronary artery diseaseand prior myocardial infarction complicated by stable New York Heart Association class II to III CHF, and treated with digoxin and diuretics. CHF was documentedby history, physical examination, chest x-ray and right-sided cardiac catheterization (cardiac index 15 mm Hg at rest, or both). Patients with diabetes and chronic obstructive lung diseasewere excluded, as were patients with atria1 fibrillation, sick sinus syndrome, atrioventricular block and paced rhythms. Data edlection: All subjectsunderwent two 24-hour ambulatory electrocardiographic recordings using Marquette model 8500 AM 2-channel recorders. Recordings were obtained on 2 consecutivedays for all group II patients (CHF). Recordings from 29 of 33 group I subjects (normal) also were obtained on 2 consecutive days. In the remaining 4 subjects,recordings were obtained 9 to 29 days apart. There were no interim changesin clinical state. Recordings were analyzed using a Marquette 8000 computerized Holter system, which classifies each beat as to site of origin. The beat classification was overread manually and corrected where necessary.Each RR interval during the 48-hour period was determined. The data were then transferred to Digital Equipment VAX computer and Cambridge Digital 11/73 systemsfor further analysis. The software used to measure HR variability was developedin our laboratory.3-6 It determinesthe mean f standard deviation (SD) of the sinus RR intervals for each 5-minute period during the recording period, and subsequently computes their mean value. In instances in which sinus rhythm is interrupted by ectopy, the program eliminates 1 RR interval preceding, and 2 RR intervals following, nonsinus beats. Three HR variability determinations were obtained for each subject: (1) the “SDANN” measure (or the
standard deviation of all mean 5minute “normal” sinus RR intervals, i.e., sinus RR intervals without intervening ectopic beats, over 24 hours2,3);(2) the “SD’ measure (defined as the mean of the standard deviations of all normal sinus RR intervals during successive 5-minute recording periods over 24 hours3); and (3) a new measure, the “CV’ measure (or coefficient of variation of the SD, which is obtained by dividing the SD value for each 5-minute period by the mean RR interval for that period, and then averaging all the 5minute CV values over 24 hours). Data analysis and statistical methods: The following analyseswere performed: First we testedday-to-day variation in, and reproducibility of, HR variability by comparing individual and mean group values for the 2 days. The results were expressedas mean f SD of the percent differencesbetween the days. The overall comparability of the two 24-hour recording periods was assessedby comparing mean RR for days 1 and 2 for each group. The Pearson product-moment correlation coefficient was then calculated to describeday 1 to day 2 relation for HR and HR variability. Second,we examined correlations between mean HR and HR variability for the 2 study groups and usedthe least-squares regression method to define the relation. The 2-way analysis of variance was used to test for differences in mean group values for the RR interval and HR variability, and for differences between men and women. Third, we compared mean HR and HR variability of normal subjects and patients with CHF. The SAS statistical package version 6.03 was used for all statistical analyses.7Normality of distribution of the data was tested using the Shapiro-Will< statistic. When normally distributed, data are presentedas mean f SD. Differences were considered significant at p 20%, the chi-square test for SD versusCV having an approximate p value of 0.05 and for CV versus SDANN p 100 to 30 to 130 ms.
Mean R-R
erable interindividual differences in SD and SDANN valuesfor given mean RR intervals, there was a moderate but significant correlation between the mean 24 hour RR interval and the SD and SDANN values (r = 0.64 and 0.54, respectively) (Figure 3). In contrast, CV values did not correlate with mean RR. GROUPII(PATIENTSWITHCONGESTIVEHEARTFAILURE):
Individual CHF patients with identical or comparable mean 24-hour HR often exhibited considerable differences in HR variability determined by SD and SDANN measures.Although group SD and SDANN values were lower for those with CHF than for normal subjects(Figure 4), personsin the 2 groups exhibited a similar range of HR variability. Relation between 24hour mean RR intervals and mean SD and SDANN values were similar to those in group I (r = 0.52 and 0.50, respectively) (Figure 3). Differences in mean heart rate and heart rate variability between group I and II subjects: Group II pa-
tients (CHF) had a higher mean HR than group I subjects (normal): average mean 48-hour RR interval = 770 f 94 and 8 14 f 93 ms for the 2 groups, respec-
SDANN
Day 2 (msec)
0
Relation between mean heart rate and hearl rate GROUP I (NORMAL SUBJECTS): Despite considvariability:
Day 2 (mrsc)
250
500
750
1000
1250
0
50
100
150
200
Day 1 (mssc)
Day 1 (msec)
SD
uro 1.
Day 2 (mrec)
00
20
40
60
80
0
2
Day 1 (mssc)
1672
FtGURE 2. Variabirty of mean RR interval, and heart rate variabilii determinations using SDANN, SD andCVmasureson2consecutive days in 22 patisnts with congestive heart fake (group II). Sse text for discussion. Abbreviations as in Fig-
THE AMERICAN JOURNAL OF CARDIOLOGY VOLUME 68
4
6
8
Day 1
DECEMBER 15, 1991
10
tively, but the differences did not reach significance (p g 0.09). Day 1 HR variability was significantly lower for patients with CHF than for normal subjects,irrespectiveof the measure used: SD 36 f 16 vs 76 f 25 ms (p
