Analysis of R-R Intervals in Patients with Atrial
`Fibrillation at Rest and During Exercise
`
`By BERT K. BOOTSMA, M.D., ADRIAAN J. HOELEN, M.SC.,
`JAN STRACKEE, PH.D., AND FRITs L. MEIJLER, M.D.
`
`SUMMARY
`Serial autocorrelation functions and histograms of R-R intervals in patients with
`atrial
`fibrillation, with and without digitalis,
`rest and during exercise, were
`at
`produced by a computer. At rest with and without digitalis the first and higher order
`coefficients did not differ from zero. During exercise
`(also with and without digi-
`talis) only the first autocorrelation coefficient became slightly positive (in the order
`of 0.07) whereas the form of the histograms was profoundly altered by boil
`exercise and digitalis. The change in form of the histograms reveals the change in
`electrophysiologic properties of the A-V conduction system. Since the serial auto-
`correlation functions were not influenced by digitalis and only slightly by exercise,
`the conclusion seems justified that the refractory period of, and the concealed con-
`duction in, the A-V system cannot be (solely) responsible for the random nature of
`the ventricular rhythm in patients with atrial fibrillation. The effect of randomly
`spaced atrial impulses of random strength reaching the A-V node from random
`directions can explain these results.
`
`Additional Indexing Words:
`Concealed conduction
`Digitalis
`Refractory period
`Serial autocorrelation coefficients
`
`Random rhythm
`A-V node
`
`DESPITE increasing knowledge of the
`electrophysiologic properties of the A-V
`the irregular
`of
`junctional
`pattern
`tissue,
`ventricular rhythm in patients with atria]
`fibrillation is still not completely understood.
`(1)
`concealed
`Current opinions hold that
`conduction in and (2) changes of the effective
`refractory period of the A-V junction deter-
`ventricular
`irregular
`of
`mine the
`pattern
`fibrillation.1-3 The
`atrial
`during
`responses
`atrial
`degree of concealed conduction of
`impulses in the A-V junction is related to the
`refractory period of the A-V nodal tissue,3-5
`while the duration of the A-V nodal refractory
`
`From the Department of Cardiology,
`University
`Hogpital, Utrecht, and the Laboratory of Medical
`Physics of the University of Amsterdam, The Nether-
`lands.
`Address for reprints: Dr. Frits L. Meijler, Depart-
`Utrecht,
`Hospital,
`ment of Cardiology, University
`The Netherlands.
`Received November 17, 1969; revision accepted for
`publication January 26, 1970.
`Circulation, Volume XLI, May 1970
`
`783
`
`period is
`related
`to the duration of the
`preceding R-R interval(s).6, 7 At the same
`time the duration of the R-R intervals
`is
`related to the length of refractory period of
`the A-V nodal tissue as well as to the degree
`of concealment of atrial impulses in
`this
`tissue.
`If these mutual relationships are to deter-
`mine the R-R interval behavior in
`atrial
`fibrillation, at least some correlation between
`the duration of a R-R interval and that of its
`successors can be expected. This amongst
`others would imply that at least the first few
`serial autocorrelation coefficients of the R-R
`intervals would differ from zero. In a previous
`paper8 we demonstrated that the ventricular
`rhythm of patients with atrial fibrillation at
`rest who did not receive any medication was
`random. These results were partly confirmed
`by others,9 10 whereas different results have
`also been published."1-'6
`If the electrophysiologic properties of the A-
`V node are responsible for the random pattern
`
`Downloaded from
`
`http://circ.ahajournals.org/
`
` by guest on March 1, 2015
`
`APPLE 1021
`
`1
`
`

`

`BOOTSMA ET AL.
`
`During each part of t-he experimental proce-
`dure approximately 2,000 consecutive complexes
`were recorded. For visual
`inspection
`of
`the
`recorded signals, the tape was played back on an
`oscilloscope and the lead with the most pro-
`nounced R waves was selected
`for
`further
`processing.
`The selected lead signal was band-pass filtered
`(approximately 5 to 30 Hz) in order to suppress
`T waves and noise, and fed into a Schmitt trigger.
`The discrimination level of the trigger and the
`filter
`characteristics
`previously
`selected
`were
`depending on the requirements of the signal but
`remained
`fixed
`during
`the
`actual
`R-wave
`detection.
`The trigger
`one-shot
`initiated
`a
`multivibrator. After the beginning of an output
`pulse the one shot was blocked for 200 msec. The
`
`Table 1
`Patients by Age, Sex, Diseases,
`
`Distribution of
`and Treatment
`
`Digitalis
`treatment
`
`-7+
`
`-7+
`
`-+
`-+
`+/
`
`+ ++++++
`
`Diagnosis
`RHD
`IHD?
`IHD?
`mI
`RHD
`IHD
`IHD?
`RHD
`IHD
`RHD
`IHD
`IHD
`RHD
`IHD
`RHD
`RHD
`RHD
`IHD
`IHD
`RHD
`IHD
`RHD
`RHD
`RHD
`IHD
`RHD
`IHD
`CM
`IHD
`RHD
`IHD?
`
`Sex
`
`M M M M MM M M M M MF M
`
`MF
`
`MF F M M M F M M M M M M M M M
`
`No.
`
`Patient
`
`Age
`(yr)
`
`J.P.
`R.P.
`D.K.
`D.N.
`J.B.
`S.O.
`H.D.
`L.G.
`J.S.
`
`R.N.
`J.B.
`J.K.
`J.F.
`A.D.
`C.K.
`C.M.
`W.J.
`H.L.
`P.R.
`K.Z.
`
`J.M.
`G.K.
`C.G.
`J.E.
`P.O.
`B.H.
`F.V.
`A.S.
`J.V.
`J.B.
`L.K.
`
`49
`47
`42
`57
`43
`52
`43
`35
`62
`45
`58
`70
`49
`62
`50
`26
`50
`62
`63
`43
`49
`31
`37
`36
`58
`47
`45
`45
`53
`37
`56
`
`1 2 3 4 5 6 7 89
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`27
`28
`29
`30
`31
`
`Abbreviations: RHD = rheumatic heart disease;
`IHD = ischemic heart disease; MI = mitral insuffi-
`ciency; CM = cardiomyopathy.
`Circulation, Volume XLI, May 1970
`
`r784
`
`of the ventricular rhythm, interventions like
`treatment with digitalis or exercise, or both,
`which change the electrophysiologic proper-
`ties of the A-V junctional tissue should have a
`distinct
`effect on the R-R interval
`serial
`autocorrelation function.
`This paper deals with the statistical analysis
`of R-R intervals
`atrial
`patients
`with
`in
`fibrillation at rest and during exercise with
`and without digitalis therapy. For comparison
`the effect of exercise detennined by the same
`statistical methods was also studied in healthy
`subjects.
`
`Methods
`
`Patient Population
`The case material consists of 41 subjects: 31
`patients with atrial fibrillation and 10 healthy
`persons. The clinical situation of the patients is
`listed in table 1. Twenty patients were studied
`before, during, and after exercise; three of these
`patients were studied twice, one time without
`treatment and one time during digitalis therapy.
`Of the total of 23 exercise studies, 10 were on
`patients without treatment and 13 on patients
`receiving digitalis. The healthly persons all had
`sinus rhythm. They varied in age from 19 to 40
`years. None of the healthy subjects received any
`medication.
`
`Experimental Procedure
`Before any recording was started, the patients
`rested for 30 min on a bench. Next the recording
`was made during another 30-min period of rest.
`The rest period recording was followed by a
`recording made during exercise. The subjects
`were exercised
`with
`load on a
`constant
`a
`hyperbolic
`bicycle
`the
`sitting
`in
`ergometer
`position for 17 min. The first 2 min of these
`recording periods were discarded. The last 15 min
`of this recording period were used for the analysis.
`The load applied varied according to the validity
`of the patient from 40 to 100 watts. As soon as
`possible
`after
`the
`of
`exercise,
`cessation
`the
`recording
`continued
`the
`recumbent
`in
`was
`position for another 30 min. In summary:
`(1) 30 min rest (without recording)
`(2) 30 min rest (recording)
`(3)
`2 min exercise (without recording)
`(4) 15 min exercise (recording)
`(5) 30 min rest (recording).
`Data Handling
`The electrocardiogram (leads V4, V5, and V6)
`was recorded
`analog
`magnetic
`tape
`on
`an
`(Ampex FR 1300, tape speed 3% in/sec).
`
`Downloaded from
`
`http://circ.ahajournals.org/
`
` by guest on March 1, 2015
`
`2
`
`

`

`R-R INTERVALS IN ATRIAL FIBRILLATION
`
`785
`
`uncontaminated data. So, even small trends give
`rise to large distortions if the variance of the data
`is also small. Undistorted correlograms can be
`computed after detection and classification of the
`trends, if any, and subsequent adjustments of the
`raw data.
`Mostly the classification forms the bottleneck in
`this procedure. For many processes, however, it
`safely be assumed that
`correlation
`the
`can
`coefficient
`coefficients
`range of
`certain
`in
`a
`numbers are zero if trend distortion is not present.
`This property can be used to detect trend after
`the computation of a correlogram, while the
`degree of
`distortion can be used for
`trend
`compensation in the correlogram itself.18 The
`trend compensation is accomplished according to
`the following formula:
`
`r, =
`
`I- i
`
`(2)
`
`where
`r =the i-th coefficient after trend compensa-
`tion
`r* = the i-th coefficient before trend compen-
`sation
`r = the mean value of a number of
`coeffi-
`cients assumed to be zero
`(esti-
`mate for the trend distortion).
`Equation 2 holds for linear trends (if N in
`equation 1 is large) and gives a negligible error
`for monotonic or low frequency trends, if:
`
`100 j
`
`50
`
`0
`
`-50
`
`= BEFORE COMPENSATION
`AFTER COMPENSATION
`a =
`
`coefficient number
`
`20
`
`30
`
`40
`
`50
`
`Atrial fibrillation
`MHR = 115/min
`T
`= 522 msec
`= 134 msec
`SD
`
`Figure 1
`The serial autocorrelogram of the R-R intervals of a
`patient (table 1, R.N.) with atrial fibrillation during
`digitalis treatment in the exercise period. The effect
`of trend compensation on the correlation coefficients,
`bringing aU values back to zero (except the first
`coefficient), is denonstrated. Abbreviations: MHR =
`mean heart rate; SD =standard deviation; 1=aver-
`age interval.
`
`output pulses of the multivibrator were recorded
`on a free channel of the analog tape. This
`procedure prevented spurious signals (e.g., even-
`tual
`to cause the incorrect
`large T waves)
`detection of R waves.
`In this way the original ECG was converted
`into a series of pulses, each pulse starting at the
`onset of an R wave. After inspection the tape was
`sent to the Dutch Scientific Data Center of IBM.
`Here the pulse signal was sampled at a sampling
`rate of 750/sec and a digital tape of the pulse-
`interval durations was produced by an IBM 1401
`tape an IBM 7094-Il
`computer. From this
`computer finally produced 50 serial autocorrela-
`tion coefficients together with the histogram of
`the R-R intervals.
`
`Statistical Methods
`1. Mean Heart Rate
`The mean heart rate is computed from the
`reciprocal of the mean interval.
`II. Serial Autocorrelation
`The serial autocorrelation coefficients are ap-
`proximated17 by:
`
`(Xi -xJ) (Xi + j
`
`X)
`
`N
`
`(xi-x)2
`
`(1)
`
`N-j
`
`N-j
`E.
`i = I
`
`N E
`
`i=1
`
`r=-
`
`where j = coefficient number (0, 1, 2, .
`r = correlation coefficient
`Xi = duration of the i-th R-R interval
`i = mean duration of the R-R intervals
`N = number of subsequent R-R intervals
`(approximately 2000).
`
`., 50)
`
`III. Trend Compensation
`The serial autocorrelation procedure is only
`meaningful for stationary processes. However, the
`parameters of series of R-R intervals are seldom
`time-independent. For example, the heart rate
`may vary as a result of varying psychologic stress
`or in response to exercise.
`For practical reasons these monotonic or more
`or less slow periodic changes of the heart rate, or
`both, have been called "trends" in this study. The
`process with
`autocorrelogram
`of
`serial
`a
`a
`superimposed trend is in general considerably
`different from the correlogram of the process itself
`(fig. 1). The degree of the distortion by the trend
`depends on the relation between the magnitude
`of the trend and the standard deviation of the
`Circulation, Volume XLI, May 1970
`
`Downloaded from
`
`http://circ.ahajournals.org/
`
` by guest on March 1, 2015
`
`3
`
`

`

`BOOTSMA ET AL.
`
`uninfluenced by exercise. After the exercise
`the histogram more or less returns to the
`shape as in the period before exercise. The
`serial autocorrelation function after exercise is
`identical with that before exercise.
`Figure 3 gives the histograms and the
`autocorrelograms of the same patient with
`atrial fibrillation as those in figure 2 now,
`before, and during digitalis treatment. The
`mean heart rate has diminished from 108/min
`refractory
`period as
`to 63/min, and the
`presented by the time between Y axis and
`beginning of the histogram has increased from
`350 to 550 msec. It can be seen that as far as
`the histograms are concerned digitalis has an
`opposite effect to that of exercise. At the same
`time the autocorrelogram has remained virtu-
`coefficients being zero
`ally unchanged, all
`before and during digitalis treatment.
`Although digitalis tended to decrease the
`mean heart rate and had a distinct effect on
`the histograms at rest, the overall results of
`exercise were not influenced by digitalis. This
`is shown in figure 4. The digitalis histogram is
`also profoundly altered by exercise. The
`treated
`of
`autocorrelograms
`with
`patients
`digitalis are identical with those of patients
`without the drug. Although all the histograms
`had a general uniform appearance, being
`rather skew, each histogram differed from the
`other, showing a more or less typical form for
`each patient. In a number of cases
`the
`autocorrelograms showed a low frequency
`component, as shown in figures 1 and 4. The
`low frequency component coincides with a
`gradual increase of the mean heart rate during
`exercise in subjects with a small standard
`deviation.
`The effect of exercise on the mean heart
`rate and the value of the first serial autocorre-
`coefficients of the R-R intervals
`lation
`in
`patients without treatment and during digi-
`talis treatment is shown in table 2A and B and
`figure
`respectively.
`6,
`In all
`cases
`trend
`compensation was applied to compensate for
`this low frequency distortion.
`During exercise at a 5% level of significance
`(sign test) a positive shift in ri has been
`found. The 959
`confidence
`level
`for
`the
`Circulation, Volume XLI, May 1970
`
`786
`
`imax>» n
`
`(3)
`
`where
`imax = the maximum length of lags in the com-
`putation of r
`n = the number of coefficients covering the
`duration of the smallest period in the
`trend.
`It is important that imax be sufficiently large to
`allow for a stable value of i.
`A previous study showed that in patients with
`atrial fibrillation at rest the first and higher serial
`autocorrelation coefficients of the R-R intervals
`did not differ from zero.8 For normal subjects,
`arrhythmia,
`without sinus
`only the first few
`coefficients differ from zero. When sinus arrhyth-
`mia is present, the mean value of a series of the
`higher coefficients
`still
`if the
`series
`is
`zero
`contains one or more complete cycles of arrhyth-
`mia. With the above mentioned constraints in
`mind we have chosen imax = 19, while the value
`is computed as the average of the values of
`of
`the correlation coefficients r* to and including
`
`r I
`
`V. Shift of the First Coefficient
`The sign-test is used to compare the influence
`of the experimental conditions on the first serial
`autocorrelation coefficients. This test is applied to
`the difference of the first correlation coefficients
`for each subject under two different conditions.
`Our null hypothesis is that the coefficients have
`the same value under both conditions. A 5% level
`of significance was chosen.
`
`Results
`The drawings of autocorrelograms in the
`figures are not compensated for trend except
`in figure 1 (see Methods). In figure 2 the
`histograms and the autocorrelograms of a
`representative patient with atrial fibrillation,
`preceding digitalis treatment before, during,
`and after exercise are shown. During exercise
`the mean heart rate increases from 108/min
`(at rest) to 157/min. The refractory period, as
`represented by the time between the Y axis
`and the beginning of the histogram, shortens
`from 350 to 250 msec. The decrease of the
`refractory period by exercise as represented
`by the time between the Y axis and the
`beginning of the histogram varied from 50 to
`350 msec with a mean of 175 msec. Although a
`small trend can be noted (fig. 2B), the
`autocorrelograms of this
`patient remained
`
`Downloaded from
`
`http://circ.ahajournals.org/
`
` by guest on March 1, 2015
`
`4
`
`

`

`R-R INTERVALS IN ATRIAL FIBRILLATION
`A
`
`787
`
`25 n
`
`20
`
`15
`
`10
`
`5-
`
`50 1
`
`40 -
`
`30 -
`
`20 -
`
`102
`
`25
`
`5
`
`[n
`
`l
`
`500
`
`= 22--
`
`I
`
`10Q0
`
`1500
`
`misec
`2000
`
`50 -
`
`-50 -
`
`mSec
`2000
`
`C
`1001
`
`'6
`
`50-
`
`so
`
`0
`
`-50
`
`5W0
`
`1000
`
`1500
`
`2000
`
`msec
`
`PERlOD: REST
`TREATMENT: NONE
`
`coefficient number
`
`10
`
`20
`
`30
`
`40
`
`50
`
`fibrillation
`Atrial
`MHR z 108/min
`T
`a 557 msec
`* 126 miec
`SD
`
`PERIOD: EXERCISE
`TREATMENT: NONE
`
`fibrillation
`Atrial
`MHR R 157/min
`I
`a 382 me
`a 134 msec
`SD
`
`PERIOD: REST AFTER EXERC I SE
`TREATMENT: NONE
`
`10
`
`20
`
`30
`
`40
`
`50
`
`Atrial
`fibrillation
`MHr u 127/min
`T
`* 471 mes
`a 95 mscc
`SD
`
`Figure 2
`(A, B, and C) Histograms and autocorrelograms of the R-R intervals of a patient (table 1,
`R.N.) with atrial fibrillation who was receiving no medication: at rest, during, and after exer-
`cise. The use of a different scale in B should be noted. Despite the considerable change in the
`form of the histogram during exercise (B), the autocorrelogram is almost identical with those
`in A and C.
`
`Downloaded from
`
`http://circ.ahajournals.org/
`
` by guest on March 1, 2015
`
`5
`
`

`

`788
`
`25
`
`20
`
`15 -
`
`10 -
`
`5S
`
`25 %
`
`20 -
`
`15-
`
`10 -
`
`5.
`
`L
`
`A
`1oo I
`
`50 -
`
`0
`
`-50
`
`2000
`
`B
`
`100
`
`50
`
`0
`
`-50-
`
`I
`500
`
`.
`
`1000
`
`~
`
`~.
`1500
`
`BOOTSMA ET AL.
`
`PERIOD: REST
`TREATMENT: NONE
`
`cofficient number
`
`0
`10
`
`2
`20
`
`3
`30
`
`40
`
`So
`
`Arile
`M1Il
`T
`SD
`
`fiibriNtlon
`a 108/min
`a 557 msec
`a 126 msec
`
`PERIOD: REST
`TREATMENT: DIGITALIS
`
`cofficient number
`
`AA.-
`10
`
`v-
`
`J AA v..
`30
`
`20
`
`,a. &
`40
`
`50
`
`Atriel fIbrIlletion
`MNRt
`63/min
`T
`a 948 msec
`SD
`a 233 msec
`
`mIsc
`2000
`Figure 3
`Histogram and autocorrelogram of the R-R intervals of a patient (table 1, R.N.) with atrial
`fibrillation at rest, receiving no medication (A) and during digitalis treatment (B). The auto-
`correlogram is unchanged, despite the change in form and the shifting to the right of the
`histogram.
`coefficient
`distribution
`of the
`median P50
`under this condition is 0.03 < P50 < 0.12.
`In figure 5 the histograms and autocorrelo-
`grams of a healthy man, aged 20 years, with
`sinus rhythm before, during, and after exercise
`is shown. There is a considerable difference
`among the histograms and autocorrelograms
`of the healthy subjects. Figure 5 is a fair
`example of this group. During exercise the
`histograms become rather narrow (indicating
`an equalizing effect on the duration of the R-R
`intervals).
`The effect of exercise on the value of the
`first serial autocorrelation coefficients of the R-
`R intervals in normal subjects with a sinus
`
`rhythm is shown in table 2C. The difference
`between sinus rhythm and atrial fibrillation is
`striking. Finally in
`figure 6 the effect of
`on the
`exercise
`compensated first
`serial
`autocorrelation coefficients and the relative
`standard deviation of the intervals is shown
`for all cases studied. On this basis a computer
`should be able to distinguish sinus rhythm
`from atrial fibrillation by using the R waves
`only.
`
`Discussion
`The serial autocorrelation functions, repre-
`senting the statistical behavior of the R-R
`intervals, indicate a random distribution of the
`Cifrculation, Volume XLI, May 1970
`
`Downloaded from
`
`http://circ.ahajournals.org/
`
` by guest on March 1, 2015
`
`6
`
`

`

`R-R INTERVALS IN ATRIAL FIBRILLATION
`
`789
`
`Table 2
`Data on Patients and Controls: At Rest and During and After Exercise
`
`Rest
`
`Rest after exercise
`Exercise
`1st SAC
`Ist SAC
`1st SAC
`Workload
`(%)
`MHR
`(%)
`MHR
`(%)
`MHR
`Patient
`(watts)
`No.
`A. Data on 19 patients with atrialfibritlation without treatment at rest and on 10 of them also during and after exercise
`0.9
`81
`0.1
`14.4
`75
`130
`100
`J.P.
`1
`-10.6
`63
`-5.5
`-8.7
`60
`109
`75
`R.P.
`2
`-5.0
`102
`11.2
`169
`-0.9
`81
`100
`D.K.
`3
`-6.3
`101
`4.8
`151
`0.2
`D.N.
`91
`50
`4
`1.6
`78
`15.8
`194
`-2.1
`68
`40
`J.B.
`5
`5.1
`108
`9.1
`19.2
`107
`139
`S.O.
`50
`6
`-7.2
`139
`4.7
`-2.9
`50
`139
`182
`7
`H.D.
`6.0
`-1.7
`85
`123
`4.0
`81
`50
`L.G.
`8
`-7.4
`85
`10.0
`124
`-10.2
`80
`J.S.
`50
`9
`3.0
`127
`3.8
`157
`-4.5
`108
`50
`R.N.
`10
`-7.3
`83
`J.B.
`11
`-0.5
`90
`J.K.
`12
`-3.7
`J.F.
`62
`13
`A.0.
`8.4
`89
`14
`13.2
`76
`C.K.
`15
`79
`3.8
`C.M.
`16
`W.J.
`19.2
`59
`17
`-2.8
`N.L.
`63
`18
`1.6
`70
`P.R.
`19
`B. Data on 17 patients with atriat fibrillation during digitalis treatment at rest and on 13 of them also during and after
`exercise*
`7.8
`-17.5
`1.5
`5.2
`-5.1
`-5.4
`-4.1
`-5.1
`-4.2
`-4.5
`-2.0
`5.0
`12.9
`6.5
`0.2
`14.6
`2.9
`
`20
`21
`22
`23
`24
`9*
`25
`10*
`26
`27
`28
`6*
`29
`30
`12*
`31
`13*
`
`K.Z.
`J.M.
`G.K.
`C.G.
`J.E.
`J.s.
`P.O.
`R.N.
`B.H.
`F.V.
`A.S.
`S.O.
`J.V.
`J.B.
`J.K.
`L.K.
`J.F.
`
`50
`50
`50
`50
`50
`50
`40
`50
`50
`50
`40
`50
`50
`
`69
`58
`70
`95
`61
`82
`79
`63
`74
`67
`75
`78
`80
`72
`61
`84
`78
`
`169
`
`111
`
`127
`169
`84
`124
`115
`115
`106
`105
`107
`107
`158
`
`17.8
`3.1
`-26.1
`8.5
`6.6
`7.9
`5.7
`11.0
`-0.3
`-1.2
`8.8
`11.6
`25.8
`
`83
`59
`81
`102
`64
`85
`83
`77
`78
`72
`74
`78
`85
`
`0.1
`-18.6
`0.2
`3.3
`-1.3
`-2.9
`-3.9
`1.1
`0.3
`-9.1
`-2.9
`7.2
`1.2
`
`32
`33
`34
`35
`36
`37
`38
`39
`40
`41
`
`C. Data on 10 normal subjects at rest and during and after exercise
`91.4
`163
`71
`66.9
`100
`W.R.
`87.3
`108
`38.9
`53
`75
`A.S.
`80.5
`129
`62.3
`79
`75
`C.B.
`59.1
`108
`25.9
`70
`75
`J.G.
`67.5
`52.9
`119
`64
`K.B.
`100
`35.3
`119
`-5.3
`51
`75
`M.K.
`84.3
`125
`52.6
`61
`H.B.
`100
`52.3
`123
`81.6
`82
`F.F.
`75
`89.3
`133
`43.1
`69
`75
`J.J.
`84.2
`50.5
`91
`60
`75
`J.H.
`*Patients 6, 9, 10, 12, and 13 were also studied without treatment (see A in this table).
`Abbreviations: lst SAC = first serial autocorrelation coefficient compensated for trend; MHR = mean heart rate.
`Csrculation, Volume XLU, May 1970
`
`95
`60
`86
`71
`74
`55
`72
`89
`74
`60
`
`80.5
`26.5
`68.4
`15.0
`63 8
`10.5
`63.5
`77.5
`46.2
`55.4
`
`Downloaded from
`
`http://circ.ahajournals.org/
`
` by guest on March 1, 2015
`
`7
`
`

`

`790
`
`BOOTSMA ET AL.
`
`25'%
`
`20
`
`15 -
`
`10-
`
`5-
`
`20
`
`15
`
`10
`
`S
`
`25
`
`20
`
`15 -
`
`10
`
`A
`
`100 .-
`
`50-
`
`PERIOD: REST
`TREATMENT: DIGITALIS
`
`coeffici*nt nwumbr
`
`0
`
`IAse.J
`10
`
`-.AA ".J AA A e-
`40
`20
`30
`
`-
`
`5
`so
`
`-50-
`
`I
`500
`
`I
`
`1000
`
`I
`
`1500
`
`mac
`
`2000
`
`100
`
`50
`
`Atrial fibrillation
`MHIR a 63/mm n
`T
`a 948 macc
`a 233 msac
`SD
`
`PERIOD: EXEtCISE
`TtEATMENT: DIGITALIS
`
`500
`
`1000
`
`1500
`
`2000
`
`1W
`
`50-
`
`0
`
`.50-~
`
`10
`
`20
`
`30
`
`40
`
`50
`
`Atrial fibrillatlon
`MHR a 115/nain
`T
`a 522 mac
`SD
`* 134 msec
`
`PERIOD: REST AFTER EXEItCISE
`TREATMENT: DIGITALIS
`
`10
`
`20
`
`30
`
`40
`
`50
`
`Atrial fibrilotion
`77/min
`MHR a
`a 775 msec
`I
`SD
`95 msec
`3
`
`500
`
`1000
`
`1500
`
`mS*C
`2000
`Figure 4
`(A, B, and C) Histograms and autocorrelograms of the R-R intervals of a patient (table 1,
`R.N.) with atrial fibrillation during digitalis treatment at rest and during and after exer-
`cise. The almost identical patterns of the autocorrelograms, despite the change of histogram
`during exercise (B), is demonstrated.
`
`Downloaded from
`
`http://circ.ahajournals.org/
`
` by guest on March 1, 2015
`
`Circulation, Volume XLI, May 1970
`
`8
`
`

`

`R-R INTERVALS IN ATRIAL FIBRILLATION
`A
`
`791
`
`25
`
`20 -
`
`15
`
`10
`
`60
`
`50-
`
`40-
`
`301
`
`10
`
`100 1
`
`50
`
`0
`
`2000
`
`100-
`
`50 -
`
`0
`
`-50 -
`
`;
`
`500
`
`I
`
`1000
`
`I
`
`1500
`
`msoc
`I
`
`2000
`
`C
`100 1
`
`K
`
`50 -
`
`a
`
`-50-
`
`500
`
`1000
`
`1500
`
`2000
`
`m0sc
`
`PERIOD: REST
`TREATMENT: NONE
`
`co*ffkiint numb*r
`
`10
`
`20
`
`30
`
`40
`
`50
`
`Sinus rhythm
`S9/min
`MHR a
`T
`S 871 msec
`a 120 misc
`SD
`
`PERIOD: EXERtCISE
`TREATMENT: NONE
`
`10
`
`20
`
`30
`
`40
`
`50
`
`Sinus rhythm
`MHR a 133/min
`T
`s 451 msec
`SD
`a 25 msec
`
`PERIOD: REST AFTER EXERCISE
`I.- t.iatMENT: NKME
`
`10
`
`20
`
`I
`
`30
`
`9
`
`40
`
`I
`
`50
`
`Sinus rhythm
`MHR a 74/min
`I
`a 809 misc
`SD
`a 176 msce
`
`Figure 5
`(A, B, and C) Histograms and autocorrelograns of the R-R intervals of a normal young man,
`20 years old, with sinus rhythm at rest and during and after exercise. The use of a diferent
`scale in B should be noted. This figure should be compared with figures 3 and 4.
`Circulation, Volume XLI, May 1970
`
`Downloaded from
`
`http://circ.ahajournals.org/
`
` by guest on March 1, 2015
`
`9
`
`

`

`O
`
`1st serial autocorrelation coefficient
`
`BOOTSMA ET AL.
`
`.
`
`a
`
`O
`
`*
`
`o
`
`5
`
`0
`
`a
`
`SINUS RHYTHM
`
`REST EXERCISE
`A
`A
`.
`
`0
`
`PATIENTS
`NORMAL SUBJECTS
`
`0
`
`0
`
`A
`
`.05
`
`I
`
`.10
`
`0
`
`.15
`
`A ATRIAL FIBRILLATION
`
`A
`
`A
`
`IL
`
`IL
`.1
`
`A
`
`A
`
`It A
`
`A
`
`A
`A
`
`A
`
`0 AAA&2
`.20 AyA&
`A25
`,
`
`.30
`
`S'Drel A
`.35
`
`792
`
`80 -
`
`60 -
`
`40 -
`
`20 -
`
`0 -
`
`-20 -
`
`Figure 6
`Scatter diagram of the first serial autocorrelation coefficients at rest and during exercise of
`23 patients and 10 normal subjects. The values on the X axis represent the ratio of standard
`deviation to average interval. All coefficients have been compensated for trend. During exer-
`cise the first serial autocorrelation coefficients of the patients with atrial fibrillation became
`slightly positive.
`
`ventricular responses in atrial fibrillation. As
`an explanation for this random nature of the
`ventricular rhythm in atrial fibrillation we may
`suggest that randomly spaced impulses of
`random strength reach the A-V node from
`random directions. At this moment there is
`insufficient electrophysiologic support to be
`sure about the random spacing and random
`strength of the atrial fibrillation. However, in
`a recent paper Janse'9 has demonstrated that
`the direction from which an atrial wave front
`approaches the A-V node is an additional
`factor deterrnining the successful or unsuc-
`cessful passage of an atrial impulse.
`Our results also demonstrate that the form
`of the histograms of the R-R intervals in
`patients with atrial fibrillation is profoundly
`
`altered by digitalis as well as by exercise. The
`of the R-R intervals
`random distribution
`remained virtually unchanged. The change in
`form of the histograms reveals a change in the
`electrophysiologic properties of the A-V con-
`duction system. Despite the electrophysiologic
`changes of the A-V nodal tissue, the random
`the R-R intervals was not
`of
`behavior
`influenced by digitalis and hardly, if at all, by
`exercise.
`Recently Urbach and co-workers16 have
`described a method of finding A-V junctional
`rhythms in atrial fibrillation. Our patients did
`not show the same high, isolated peaks in their
`histograms expressing the occurrence of a
`large number of R-R intervals
`relatively
`within the same class. In contrast with nine of
`Circulation, Volume XLI, May 1970
`
`Downloaded from
`
`http://circ.ahajournals.org/
`
` by guest on March 1, 2015
`
`10
`
`

`

`R-R INTERVALS IN ATRIAL FIBRILLATION
`
`793
`
`the 10 patients of Urbach and associates,'6 our
`patients were all in a good physical condition.
`Moreover, it should be realized that Urbach
`and associates assumed that a number of
`from
`responses may originate
`ventricular
`nonfibrillatory pacemakers. On the basis of
`this assumption they developed a method of
`detecting eventual sequences of these regular-
`ly spaced ventricular responses.
`In contrast with the findings of Goldstein
`and Barnett,10 who used the same statistical
`methods as we did, we have not been able to
`find any interrelation between R-R intervals in
`patients with atrial fibrillation at rest. This
`may be due to a different way of handling the
`patient. We allowed the patient 30-min rest
`before starting the rest period recording. We
`are inclined to state that one is only allowed
`to speak of an irregular ventricular rhythm
`due to true atrial fibrillation if that rhythm
`indeed has a random nature. In agreement
`with Urbach and associates16 we feel that
`a nonrandom
`histogram
`peaks
`or
`a
`in
`distribution of a part of the R-R intervals may
`point to A-V junctional rhythms, the existence
`of atrial flutter, or other ventricular tachycar-
`dias or rhythms.
`of
`patients
`our
`During exercise
`most
`showed a slightly positive first serial autocorre-
`lation coefficient. These values of the first
`serial autocorrelation coefficients of the R-R
`intervals in patients differ substantially from
`those of the first serial autocorrelation coeffi-
`cients of the R-R intervals in the normal
`subjects during exercise. A definite explana-
`of the first
`slight
`this
`tion for
`positivity
`autocorrelation coefficients during exercise in
`atrial fibrillation cannot be offered. We may
`suggest the possibility of a mutual relationship
`between a small number of sequential atrial
`impulses, which reach the A-V node. Since
`exercise shortens the refractory period of the
`this mutual relationship may
`A-V node,
`influence the randomness of the ventricular
`rhythm. No information is available about the
`influence of exercise on the activation pattern
`of the atrial muscle in patients with atrial
`fibrillation. Shortening of the refractory period
`of the atrial myocardial tissue in a normal
`Circulation, Volume XLI, May 1970
`
`subject during exercise, as shown by Carleton
`and associates,20 would suggest that more
`atrial impulses are formed and may reach the
`A-V node. This effect would increase the
`occurrence of concealed conduction; however,
`the accompanying shortening of the refractory
`period of the A-V nodal tissue during exercise
`would decrease the occurrence of concealed
`conduction.
`The effect of digitalis on the ventricular rate
`in atrial fibrillation is originated by lengthen-
`ing of the short intervals (increase in refrac-
`tory period of the A-V node), but also, and
`this is important, by an increase in numbers of
`long intervals of which the length is
`also
`increased by digitalis.
`Long intervals in atrial fibrillation are con-
`tributed to concealed conduction.1' 2 If this is
`true, during digitalis treatment more atrial
`impulses are being concealed. Digitalis short-
`ens the refractory period of myocardial tis-
`sue.4 Thus digitalis may increase the number
`of atrial impulses that reach the A-V node in
`a given time which may give rise to an in-
`crease of atrial concealed impulses.
`In summary, the short R-R intervals are
`made longer by digitalis by increasing the
`refractory period of the A-V system and the
`long intervals are made longer by digitalis by
`increasing the number of concealed atrial
`impulses that originate from a decrease of the
`refractory period of the atrial myocardium.
`This is an easier explanation of the digitalis
`effect in atrial fibrillation than the so-called
`vagus effect.
`
`Conclusion
`We thus have demonstrated that interven-
`tions that change the electrophysiologic prop-
`erties of the A-V junctional tissue hardly, if at
`all, influence the irregularity of the ventricles
`in atrial fibrillation. This would imply that the
`present explanation for the irregular ventricu-
`lar rhythm using only the refractory period of,
`and the concealed conduction in, the A-V
`node cannot hold for the random pattern of
`ventricular rhythm in atrial fibrillation. The
`conclusion seems inevitable that the cause for
`the ventricular irregularity should be looked
`
`Downloaded from
`
`http://circ.ahajournals.org/
`
` by guest on March 1, 2015
`
`11
`
`

`

`794
`
`for somewhere else than in the A-V system.
`We would like to offer the hypothesis that
`randomly spaced atrial impulses of random
`strength reaching the A-V node from random
`directions are responsible
`for the renewal
`process of the R-R interval sequence, the
`(otherwise essential) role of the A-V system in
`atrial fibrillation being limited to scaling the
`atrial impulses.
`
`Acknowledgment
`These investigations were started in the University
`Department of Cardiology and Clinical Physiology,
`Wilhelmina Gasthuis at Amsterdam (Head: Prof. Dr.
`D. Durrer).
`Our thanks are due to IBM-Holland for providing
`us with computing time. This paper would not have
`been completed without the advice of Prof. Dr. J.
`J.
`Denier van der Gon and the help of D. J. Touw,
`M.Sc., Mrs. Marianne Johansen-Holo, and Mr. W. van
`den Brink.
`
`References
`1. ABILnSKOV JA, MOE GK: Ventricular response in
`atrial fibrillation. In Mechanism and Therapy
`of Cardiac Arrhythmias, edited by LS Dreifus
`and W Likoff. New York, Grune & Stratton,
`Inc, 1966, p 135
`2. MOE GK, ABILDSKOV JA: Observations on the
`ventricular dysrhythmia associated with atrial
`dog
`fibrillation
`the
`heart.
`Circulation
`in
`Research 14: 447, 1964
`3. HOFFMAN BF, CRANEFIELD PF, STUCKEY JW:
`conduction.
`Circulation
`Concealed
`Research
`9: 194, 1961
`4. HOFFMAN BF, CRANEFIELD PF: Electrophysiol-
`ogy of the heart. New York, Blakiston Division,
`McGraw-Hill Book Co, 1960
`5. LANGENDORF R, PICK A, KATz LN: Ventricular
`response in atrial fibrillation: Role of concealed
`conduction in the A-V junction. Circulation
`32: 69, 1965
`T, MASTER AM: Observations
`upon
`6. LEWIs
`conduction
`the mammalian heart: A-V
`in
`conduction. Heart 12: 209, 1925
`
`BOOTSMA ET AL.
`
`7. MENDEZ C, GRUHNIT CC, MOE GK: The influence
`of cycle length upon the refractory period of
`the auricles, ventricles, and AV-node in the
`dog. Amer J Physiol 184: 287, 1956
`8. MEIJLER FL, STRACKEE J, VAN CAPELLE FJL, ET
`AL: Computer analysis of the RR interval-
`contractility relationship during random stimu-
`isolated
`of
`the
`Circulation
`heart.
`lation
`Research 22: 695, 1968
`9. JORDAN H: Die zeitliche Schwankungen der
`Herzschlagintervalle bei absoluter Arrhythmie.
`Arch Kreislaufforsch 21: 40, 1954
`10. GoLDSTEIN RE, BARNETT GO: A statistical study
`of the ventricular irregularity of atrial fibrilla-
`tion. Computers Biom Res 1: 146, 1967
`11. ARNOLDI W: Die Ermittlung von dominierende
`Rhythmen sowie der Schwere der Rhythmus-
`storung bei Kranken mit Arrhythmia perpetua.
`Klin Wschr 6: 1946, 1927
`the reason for the
`12. S6DERSTR6M N: What is
`ventricular
`arrhythmia
`of
`atrial
`in
`cases
`fibrillation? Amer Heart J 40: 212, 1950
`l'irregularite
`d'analyse
`de
`13. MACREZ C:
`Essai
`ventriculaire dans l'arythmie complete. Presse
`Med 68: 607, 1960
`14. HORAN LG, KISTLER JC: Study of ventricular
`Circulation
`atrial
`fibrillation.
`response
`in
`Research 9: 305, 1961
`15. BRAUNSTEIN JR, FRANKE EK: Autocorrelation of
`ventricular response in atrial fibrillation. Circu-
`lation Research 9: 300, 1961
`16. URBACH JR, GRAUMAN JJ, STRAUS SH: Quantita-
`tive methods fo