JOEL A. I(IRSW, MS, ALAN V. SAHAKIAN, i-‘,:

STEVEN SWIRYN, MD. FACC

Evanston, Illinois

Irregularity of the ventricuiar rhythm is a hallmark OF patients with atrial fibrilIation, ~/et tk geuesis ~4 the irregularity is not yet fulky understuod. The role oF :he atrioventricular (AV) node in detcrrkning the irregularits

of the ventricular response to atriai fihrilla!ion nas in+Fs.

Ligated by comparing the frecpucncy distributions of lbe

atrial (AA) and the ventricular (RR) intervals.

Atrial electragrams and surface eIPctrocardiograptic leads were recorded during sastained atriai fibr&tion in 12 patients with conduction ovtr the .AV node. RIR xating

factor (mean RR interval/mean AA interval) quantified t% ability of the conduction pathway to scale the atrial mput to

a slower ventricular response and ranged fr+lm Z.5 to 5.92 (mean ZT SD 9.77 2 0.92). The coeffici+:~rt of variatiorr idD! mean) measured the relative variability of the Ah and RR intsrvaf distributions. The atrial and ventricutaa cnefii- cients of variation were not significantly different IO.20 T’C 0.04 Ve?MS 0.21 * 0.03, p > 0.27A

Similar rerordings were analyzed in six patknts with

irregularity of the ventricular rhythm as a ctsn%quencc djf atrial fibrillation has long been recognized I I ! yet ;I\ cxph- nation remains unclear (2). NU!?!CiWS studies focusing on

concealed conduction within the atrmventricular (AV) node

(3-6) and statistical measures of the RR intervals ~3.7-14)

have been undertaken, with conflicting conclusions as to the

factors determining the ventricular responLc (71. More rc-

cently. studies relating the electrophysiologic propcrtw, tit

AV conducting pathways to the ventric&u response during

atrial fibrillation hav\: been perform4 (15-20). ~II! ihere

KIRSH ET AL. JACC Vol. 12. No. 5 VENTRICULAR RESPONSE TO ATRIAL FIBRILLATION November 1988: 1265-72

Table 1. Clinical Characteristics in 18 Patients -

AgelyrY Patient Gender ax Pathway

Atrial Fibrillation Medications

I WV1 CAD AVN Chronic None 2 72M CAD AVN Chronic None 3 6nM CAD AVN Chronic Digoxin 4 7lF CHD AVN Chronic Digoxin, verapamil 5 58M CAD AVN Chronic Digoxin 6 79M CAD AVN Chronic Diltiazem. metoprolol 7 SOM CM AVIS Chronic Digoxin, verapamil 8 tklM AS AVN Chronic Digoxin. diltiazem 9 25M WPW AVN induced None

IO 26M WPW AVN Induced None II 30M WPW AVN Induced None ‘. 47M WPW AVN Induced None 13 19M WPW AP Induced None Ir; 53F WPW AP Induced None I5 25M WPW AP Induced None 16 65F WPW AP Induced None I7 48M WPW AP Induced None 1g 28M WPW AP Induced None

AP = accessory AV ptithway; AS = aortic stenosis; AVN = AV node; CAD = coronary artery disease: CHD = conwnital heart disease: CM = cardiomyopathy: Dx = diagnosis; F = female: M = male: WPW = Wotff-Parkinson-White syndrome.

pared with those from the paiionts with conduction over the AV node.

Methods Criteria for admission to the study. Patients undergoing

either hemodynamic cardiac catheterization or electrophys- iologic study who demonstrated either chronic or induced sustained atrial fibrillation were considered eligible for inclu- aion in this study, The study was limited to those patients who cxhibired either type I or type II atrial fibrillation and therefore had discrete atrial electrograms (25) with AV conduction exclusively over the AV node or exclusively over an accessory AV pathway. Patients with type 111 fibrillation (25) were excluded because of difficulty in accu- rately determining atria1 intervals in the absence of discrete electrograms. The study was approved by the institution’s committee on human research, and written informed consent was obtiined from all patients before study.

Study patients (Table I). A total of I8 patients were studied, and their clinical diagnoses are summarized in Table I. Twelve patients (I I men, I woman), ranging in age from 25 to 79 years (mean 551, had conduction over the AV node during atrial fibrillation. Six patients (four men, two women), ranging in age from I9 to 65 years (mean 40), with Wolff- Parkinson-White syndrome had conduction over an acces- sory AV pathway during atria1 fibrillation. All IO patients with induced atrial fibrillation, including 4 with conduction over the AV node and 6 with conduction over an accessory pathway, were free of all cardioactive medications for at

least five drug half-hves before study. Six of the eight patients with chronic atrial fibrillation (all with conduction over the AV node) were taking various negative dromotropic medications.

Data acquisition. Continuous recordings of 2 to 5 (mean 4) min duration were made during atrial fibrilIation. A bipolar high right atria1 electrogram with a I cm interelectrode spacing (USCI), as well as surface electrocardiographic (ECG) leads 1, II and V,, were amplified and filtered using a physiologic recorder (Honeywell VR-16; Electronics for Medicine, Honeywell Inc.) as previously described for our laboratory (26.27). The intraatrial signals were filtered using a passband of 30 to 500 Hz, whereas the surface leads were filtered using a passband of 0.05 to 5,000 Hz. Signals were recorded on frequency-modulated (FM) tape (Honeywell 101). In three cases the patients were studied at another laboratory, and the signals were amplified and filtered using a similar Honeywell VR-16 recorder, recorded on FM tape using a Honeywell 56OOE and then rerecorded using a Vetter mode1 B FM tape recorder (A. R. Vetter) for transport.

Preprecessing. One surface lead and the high right atrial electrogram were played back simultaneously from tape through an antialiasing filter with a cutoff frequency of 120 Hz. The signals were amplified and then digitized at 1,200 Hz using a Masocomp MCS-563 computer system (Massachu- setts Compute’). Thcsc signals wcrc then reduced to an effective sampling rate of 240 Hz by extracting every fifth data point while stiff satisfying the Nyquist sampling crite- rion (28).

JACC Vul. 12, No. 5 November 1988: 1265-72

IO rcl J T

s r” ,ImV 1

I

I ’ 256 I83 196 192 254 t92 242 267

t---t--+--t---i---+-+----+----l

t I r.ccnr--+

Figure I. High right atrial electrograms during atrial fibrillation with detected AA intervals marked. As defined by Wells et al, (25), type I fibrillation (la) is characterized by discrete complexes sepa- rated by an isoelectric baseline whereas type Ii fibrillation (lb) is characterized by discrete complexes with perturbations cf the baseline between complexes. Type I11 fibrillation fails to demon- strate either discrete complexes or isoelectric intervals. and thus the measurement of AA intervals is less certain. All AA interval measurements are in milliseconds.

Atrial and ventricular event detection. The digitized high right atrial signals were scanned for discrete electrograms with use of an adaptive threshold detection algorithm previ- ously described by our laboratory (a), Patients in the study were restricted to those exhibiting either type I or type II fibrilation as described by Wells et al. (25). and :hus by definition had discrete atriaJ electrograms. Atrial cycle lengths were measured as the intervals between successive discrete complexes (Fig. !). A graphic display of the atrial signal, with detected electrographic complexes marked, was generated on-line and each detected complex was manually

confirmed. R wu:les in the digitized srrrface ECG lead were detected

using a differentiation-squaring algorithm adapted from that

of Hamilton and Tompkins (29). A graphic display of tb* surface lead signal with detected R waves marked

KlRSW ET AL. 1267 vENTRlCULAR RESF’ONSE TO ATRIAL FlBRlLLATION

generated on-line, and each detected R wave was manu&y

:onfirmed.

lnterva! histograms. Frequency distribution histograms

were generated for both atrial and ventricular cyck lengths,

with a class width of 20 and 50 ms. respectively (Fig. 3. The maximal AA and RR intervals were defined as the 95th

percentile points of the &iaI and ven!ricu!ar cycle !engah distributions, respectively.

The frequency of occurrence of RR intervals kmger than the mean has previously been observed to Cdl off a;oong an exponential decay curve reminiscent of the Poisson distri- bution (IO). This result w&d cause a plot of thi: nalural

logarithm of interval frequency versus intmgai length to

appear as a straight iine with a negative slope. To quantify

the shape of the rigtit-hafid tail of both the AA and RR

distributions. the normalized histogram was plotted on a semilogarithmic wale. Regession lines were caicula!ed re-

lating the natural logarithm of mterval frequency to the Interval len$h, and correlation coefficients were measured.

Scaling factors ru;d coe&ien& of variatii. The scaling factor, which was defined as the ratii of mean ventricular to

rr.ean atrial cycle kngth, quantified the ability of the AV

conduction pathway to transform its rapid atrial input into a slower ver&icuiar response. To measure the relative vti- ability of a distribution (i.e., ;lormafized to its mean), the coeflicient of variation was caiculated as the ratio of the standard deviation of the AA or RR interval distribution to its n;can (13,14.30). It was hypothesized that, if the rok of the AV conduction system during atrial fibrillation were solely that of scaling the atria! input iu a slower ventricular resprrdse, then the absolute variability inherent in the atrial rhythm (SD,,, the standard deviation of the AA interval distribution) would be similarly scaled to produce the abso- lute variability in the ventricular rhythm (SD,,, the standard deviation of the RR interval Bstribution), and the co&- cients of variation would nof differ between the atria and ventricles. Similarly, the maximal atrial intervals would be scaled to produce the maxima! ventricular intervals.

T&z term “scaling” has been used by several mthors ,12,23.24) to describe the role of the AV node in determining the ventricular response to atrial tibrilladon. In this study,

Figtare 2. Case 9. Atrial and ventricular interval histo. grams. The ho&cntal axes are not equivalent but were instead set such that the full scale RR interval is equal to the scaling factor multiplied by the futl scale AA interval. The verticaf scales were set autorzatlcaly by the computer to provide maximal vertical resolution. This choice of scales demonstrates the similarity in shape of the two distribu- tions.

Ocl 40 ~120bo200240z8a AA Interval lmsl RR lnldrval (ms)

1268 KIRSH til- At. VEYiRltUL.& RESPONSE TO ATRIAL FtERtLLATtON

JACC Vol. t.2. No. 5 November 1988: t W-72

T&k 2. Xesults in 18 Patients

Patient

I

Pathway

AQN

Mean AA tins1

Hi.3

Mean RR SD** SDRR (ms\ tms) (ms) SF CV** CVFI,

895.0 23.1 l86.9 5.92 0.153 0.209 2

4

AYN i97.7 AVN 217.7 AVN 285.3 AVN 974.0 AVN 252.3 AYN 163.5 AVN 246.4 AVN 146.2 iVN 140.5 AVN 147.2 AVM 160.0 AP 170.0

AP 174.8 AP 155.4 hP 105.7 AP 172.7 AP 183.8

794.2 36.5 148.7 4.02 0.165 0.187 811.7 3?.2 118.5 3.73 0.171 0.146 857.3 61.4 205.6 2.99 0.214 0.240

llKi3.6 39.4 248.5 3.66 0.144 0.248 823.3 45.8 168.6 3.26 0.182 0.205 576.8 25.~ 148.1 3.53 0.156 0.257

1151.8 55.5 254.6 4.67 0,225 0,221 461.2 33.6 97.9 3.!5 0.230 0.212 3ZS.2 39. I 80.2 2.55 0.279 0.224 670.6 35.4 165.6 4.55 0.240 0,247 5117.4 28.0 89.2 3.17 0.175 0.176 38s. 1 53.5 71.8 2.26 0.315 0.186 169.8 35.0 80.8 1.54 0.200 0.300 369.2 2a.9 109.3 2.37 0,186 0.2% 259.9 20.8 37.4 2.46 0.197 0.144 29; 3 63.2 69.9 I.69 0.366 0.240 326.0 37.8 138.3 I .77 0.205 0.424

CQ,, = SD,,/mean AA = atriai coefficient of variation; CV RI = S&,/mean RR = ventricular coefficient of variation; SD = standard deviation; SF = mean RRfmean AA = scaling factor. Other abbreviations as in Table I.

the use of the term “scaling” is meant only to maintain historical consistency rather than imply an assumption us to the precise mechanism of AV node function.

Liif!ar regress~n and statistical tests of 6’best fit” lines. Plots were drawn oftbe ventricular variability (SD,,) versus the scaled atrial variability (scaling factor multiplied by SD,& ailJ of the maximal (95th percentile) RR interval versus the scaled maximal BA interval (scaling factor mul- tipled by the 95th percentile AA interval). Regression lines were calculated using the method of least-squares. The slope and intercept of the best fit line were then tested against the slope and intercept of the identity line (I ,O and 0.0, respec- tively) using the Student’s f test (31).

Results Parknts With Conduction Uver the AV Node (R&/e 2)

By study design, aB atria! clectrograms were indicative of either type 1 (n = 4) or type 11 (n = 8) fibrillation (25), exhibiting discrete complexes of variable timing and config- uration separated by either an isoelectric baseliae or one of varying degrees of perturbation (Fig. I). The mean atrial cycle length ranged from 140 to 286 ms (mean k SD I99 + 54 msj. In all patients, the ventricular response to atrial fibril- lation was mtikedly irregular, with mean cycle lengths W.?ing from 358 to 1,151 ms (743 f 234 ms).

lnlmal hist~rsms (Fig. 2). In all patients, both the AA and RR intervals were &modally distributed. Qualitatively, ptiienls with longer mean ventricular cycle lengths were observed to have wider interval histograms, reflecting an

increase in the absolute variability of the RR intervals. The semilogarithmic plots of the right-hand tail of the distribu- tions were all linear, as demonstrated by correlation coeffi- cients ranging from -0.907 to -0.996 (mean -0.%7) for the AA interval distributions and correlation coefficients ranging from -0.860 to -0,981 (mean -0.940) for the RR interval distributions. All of the correlation coefficients were statis- ticalIy significant (all p < 0.005). This linearity demonstrated that the right-hand tails of both the AA and RR interval distributions had similar shapes, following an exponential decay curve.

Wing factors and cwiWeats of variation. The scaPng factors (mean RR/mean AA) ranged from 2.55 to 5.92 (3.77 * 0.92) 2nd were found to relate both atrial variability and maximal AA intervals to ventricular variability and maximal RR intervals, respectively (see next section). The atrial coefficients of variation (SD,,/mean AA) ranged from 0. I4 to 0.21+ (0.20 f 0.04). The ventricular coefficients of variation (SD&mean RR) ranged from 0.14 to 0.26 (0.21 2 0.03). There was no significant difference between the atrial and ventricular coefficients of variation (p > 0.27) suggesting that scaling of atria! variability can account for ventricular vari- ability.

Maximal intervals. In the 12 patients, with conduction over the AV node, the observed maximal RR interval ranged from 508 to 1,634 ms (1,026 2 338 ms). The product of the scaling factor and maximal AA interval ranged from 510 to 1,597 ms (988 4 313 ms). Similar to the comparison of coefficients of variation described, these were not signifi- cantly different (p > 0.5), suggesting that scaling of the

JACC Vol. 12. No. 5 KIRSH ET AL.. Novcmbcr IYllR:lth5-72 VENTKIC’U t AK RESPONSE TO ATRIAL FINRILLA?l0N

I269

maximal atrial intervals is sufficient to account for the

maximal ventricular intervals,

Patients Wirh Conduction Over un Accrssory Pathw,uy (Table 2)

To further investigate the role of the AV conduction system in determining the irregularity of the ventricular response to atrial fibrillation, the group of six patients with conduction over an accessory pathway was examined and compared with the patients with conduction Over the AV

node. By study design, the atrial electrograms were all indicative of either type I (n = 2) or type 11 (II = 4)

I’ L

fibrillation. The mean atrial cycle length ranged from 106 to I84 ms (I60 + 28 ms). This length was not significantiy Figure 3. Prediction of \cntriG ular variability KY&) by ntrial VT!-

shorter (p > 0.15) than the mean atrial cycle length (I!39 2 54 ability LSD,,,,) related by the scaling factor ISF). The regression line

ms) found in patients with conduction over the AV node. k not significantly different from the line of identity tp Z 0.4,

The ventricular response was irregular, with mean cycle sugge4ng that atrial variability during atrial fibrillation can account

lengths ranging from 260 to 385 ms (317 -c 52 ms). for ventricular variability. AP = accessory AV pathway: AVN = AV node: SD = standard deviation.

Intenai hkbgmms. The frequency histograms of the atria1 and ventricular cycle lengths were similar to those found for the group with conduction over the AV node

o :r the AV node, there was not a statistically significant

except that in one patient the RR distribution was bimodal. difference (p > 0.7) between the scaled maximal AA inter-

The semilogarithmic plots of the right-hand tail of ihe vals and the maximal RR intervals, again suggesting that

distributions were ail linear. The correlation coefficients of scaling of the maximal atrial cycle lengths is sufficient to

these plots for the AA intervals ranged from -0.916 to account for the maximal RR intervals regardless of the

-0.991 (mean -0.9623, and those for the RR intervals ranged nature of the AV conduction pathway.

from -0.834 to -0.984 (mean -0.915). All of the correlation Atrial variability md maxha atria1 cyck BS predicts of

coefficiests were statistically significant (all p < 0.05). This ventricular vrr&&l&y and maximal ventricular cycles. To

illustrate the relation between atrial and ventricular variabil- linearity again demonstrated that the right-hand tail of both the AA and RR interval diskibutions had similar shapes,

ity, a plot of the product of the scaling factor and the standard deviation of the AA interval distribution versus the

following an exponential decay curve. standard deviation of the RR interval distribution for both Sc&isg factors and coefMents of variation. The scaling

factors ranged from 1.54 to 2.46 (2.02 k 0.39). This range groups of patients is shown in Figure 3. The two quantities

was significantly (p < 0.0005) lower than the mean scaling are strongly correlated (r = 0.77, p < 0.0005) and the

factor (3.77 + 0.92) found in patients with conduction over regression line is not significantly different from the line of

the AV node, consistent with the fact that the accessory identity (p r 0.4). This predictability reflects the equality of

pathway is less effective at scaling down the atrial input than the atrial and ventricular coeflkients of variation previously

is the AV node, described. In Figure 3 the data from both groups of patients

The arrial coejicients of variation ranged from 0. I3 to lie along the same line regardless of the nature of the AV

0.37 (0,24 2 0.08). and the ventricular coefficients of varia- conduction pathway.

To illttsmte the relation between he tnnxitnal alrial and tion from 0.14 to 0.42 (0.27 ? 0.10). As for parlents with conduction over the A.V ncde, there was no significant

~ettrrictrlrr cycles, a plot of the product of the scaling factor and the maximal AA intervals versus the maximal RR

difference between the atrial and ventricular coefficients of variation (p S- 0.6), suggesting that scaling of the atrial

interval for both groups of patients is shown in Figure 4.

variability can account for the ventricular variability regard- There is again a strong correlation (; = 0.98, p < O.ooO5) and

less of the nature of the AV conduction pathway. the regression line is not significantly differenr from the line

Maxhnd intervals. The maximal RR interval ranged from of identity (p > 0.4). Again, the data from both groups of

329 to 571 ms (463 2 93 ms). As would be expected, this patients lie along the same line regardless of the nature of the

interval was significantly (p < 0.0#05) shorter than the AV conduction pathway.

maximal RR interval (1,026 ? 338 ms) observed in patients with conduction over the AV node. Tile product of scaling

Discussion factor and maximal AA interval ranged from 327 to 565 ms Previous stud& Early work by Arnoldi (32) and then (444 + 90 ms). As was the case for patients with conduction Soderstrom (7) focused on the existence of a number of

1270 KIRSH ET AL. VENTRICULAR RESPONSE TO ATRlAL flRRILLATlON

JACC Vol. 12, No. 5 Nnvcmhcr 19XR: 1265-72

"0 I , , , , , , , ,

2#) 5w 750 lcol 1250 I500 1750 zoca

SF I mmm~m AA (ms)

Figure 4. Prediction of the maximal (95th percentile) RR interval by the maximal (95th percentile) AA interval related by the scaling factor (SF). The regression line is not significantly different from the line of identity (p > 0.4). suggesting that the long ventricuIar “pauses” associated with atrial fibriliation can be atlributed to scaling of the maximal atrial cycle lengths. Abbreviations as in Figure 3.

discrete “levels” in the interval tachogram (a plot of RR interval against consecutive beat number) corresponding to a set of evenly spaced modes in the interval histogram. Soder- Strom sought IO explain the ordered structure of these modes as being determined by multiples of the refractory period of the AV node, which he was unable lo measure. However, subsequent work (8,10,12) failed to discern any such struc- ture in the patterns of the ventricular response. In particular,

Bootsma (12) and Meijler (24) and their coworkers, using RR interval analysis alone, concluded that the AV node, other than scaling the atria1 impulses, did not play a primary role in determining the irregularity of the ventricular response to atrial fibrillation.

Present study. We found several descriptors of frequency distribution histograms, including unimodality and the shape of the right-hand tail, to be comparable for both atrial and ventricular interval distributions in a group oi’ 12 patients with atrial fibrillation. The predictability of both the ventric- ular variability and the maximal RR interval as scaled functions of the corresponding atrial measures adds to previous data derived from RR interval analysis alone, suggesting that the nature of the distribution of the RR

intervals is primarily a consequence of the nature of the distribution of the AA intervals (l&12,23,24). This predict-

ability would support the suggestion that the role of the AV node during atrial fibrillation is predominantly confined lo that of scaling the atrial activity. These relations were also observed in six patients with atrial fibrillation and conduc- tion over an accessory AV pathway, providing additional evidence that the irregularity and maximal intervals of the ventricular response to atrial fibrillation can be accounted for by the irregularity and maximal intervals of the atrial

JACC Vol. 12. No. 5 KlRSH ET AL.. Novcmhrr :9lilt:lthS-72

1271 VEN rRI(‘U1.AR RESPONSE TO ATRIAL FlBRILLATION

the ventricular response IO atrial fibrillation with conduction acce\5ory AV pathways, further supporting the conclusion over an accessory AV pathway was Iimited to those patients that no special prorzties of the AV node need to be whose QRS complexes Appeared ccinsistently fully pre- considered lo account for the nature of tile RR inrerv;rl excited, refiecting an absence of fusion. and hence the distribution. absence of any evident contribution to the ventricular re- sponse by the AV node. Under these conditions, repetitive retrograde activation of the His bundle by way of the

We sxpre$\ our dpprecialion :o Jose Galh~legu~. MD for atzn~ance In data cnllec!ion. Jo Ellen Womwn. RN for technical anintanre and Shrl!.y J.

accessory pathway might preclude the AV node from par- Hahernan. PhD. Chairman. Depanment of SratiPticc. Nonhuc,tem Univer-

ticipating in any AV conduction (15.40,41). wily for advice regarding the ~!atls!ral techmques employed in ihI,. #.dy and

Determining the inpuUoutpu1 of the AV conduction path- revw nf the maquxripi.

way. The R wave of the s&ace ECG lead identified the output of the AV conduction pathway, but determining the actual input to the pathway was more difficult. It: addition 10 References the probable existence of multiple inputs to both the AV rlode (3.42,43) and the accessory AV pathway (44). atrial

I. Rolhherger CJ, Winterberg H. Vorh+lTimmern ud Arhyrhmia perpctua. Wren Kim Wochenschr Iw9:ZZ:839-U.

activity recorded in the +$II right strium is not necessarily similar to activity simultaneously recorded at other atrial sites (25). The eff& of muliip!e inputs impinging on the AV conduction pathway is unclear; if the inputs respond inde- pendently with no consistent phase relation, then the appar- ent input rate to the conduction pathway would be much more rapid than measured elsswhere in the atrium (3). If, however, the separate inpu.., are not electrically indepen- dent, the effect of multiple inputs is less clear. The effects of multiple inputs to the AV node or accessory pathway would contribute to the behavior of the conducting pathway during atrial fibrillation and would be reflected in the statistical behavior of the output (i.e., the RR intervals). Thus. the

2. Brady DA Ventncular rate panems m arrial fibnllahon (editor& Clrculallon IYirl:4l~i+5.

3. Moe Gk. Ablldskov JA. Observattonr on the .ientricular dyvhytnml. arsoctated with alaal fibnllatton rn the do; hear!. Circ Res 1%4:lJ:447- 60.

J. ! angendorf R. Pick A. Kalr LN. Ventricular rerponx m alrial Cbrifla. timon. role of concealed conduction in Ihe !V junction. Circulatton 1%5X.6%75.

5. Moore EN. Ohrervutions on concealed conduztron m atnal fibnllation. Clrc Re\ 1967.?1:201-8.

h. Yamada K. Okajima M. Hori K. el al. On !he gene+ of rhe absolute venrricutar arrhylhmiaassocialed with slrial fibrillalion. Circ Res t968:??: 707-r.s.

7. Soderstrom N. What is the reason for the venlricular arrhythmia in Casey of duruxlar fibrillation? Am Head I 1950;40:?1?-23.

complication introduced by the multiple inputs was insepa- K. Braunuein JR. Franke EK. Autocorrelalron of vemrxular recpon<e In

rable from other factors determining the behavior of the AV alnal Abnllahon, Circ Res PX&9:3&%4.

conduction pathway and could not be examined mechanis- 9. Hem LG. Kinder JC. Study of ventricular response III alnal fibrdlatmn.

tically in this study. Circ Re\ I9hl.%3OI-I I.

Activity recorded in the high right atrium during atrial fibrillation is not an instantaneous measure of tk input tc the AV conduction pathway. However, there is no reasan to believe that, over long periods of time, the staktical nature of atrial activation intervals would differ between the high right atrium and sites closer lo the atria1 insertion of the AV conduction pathway because of the random nature of atrial fibrillation (3,10,14.45). Therefore, for the purposes of this study, the assumption was made that over a period of minutes the statistical behavior of activation intervals mea- sured in the high right atrium would provide an unbiased estimate of the statistical behavior of activation intervals

IO. Goldcrein RE. Barnelr GO. A rI&lical study of rhe vemrrcular irregu- larity of atrial fibrillation. Comp Biomed Res 1967;l:i46-61.

II. Urbach JR. Grauman JJ. &us SH. Quanlilahve methods for the recognitloo of alriovenlricular junchonnl rhylhmc in atria1 fibrillation. Clrcolialion 1969:39:803-17.

i2. Boolsma BK. Hoelen AJ. Slrdckee J. Meljler FL. Analys~r of R-R intervals tn palienls wilh atrIal fibrlllalion at reel and during exercise. Circularion 1970;41:783-94.

13. Exudero EM. lveli CA. Moreyra AE. Lorenle H. Cingolani HE. The pulce in patlenls wilh atria1 fibrtllalion: iIs irregularity and inequalily. Eur J Cardiol 1976:451-8.

I4 Cohen PJ. Berger RD. Dushane TE. A quannlauve model for the ventncular response during atrial fibrillation. IEEE Trans Biomed Eng 1983.3O:WLMll.

closer lo the AV conduction pathway. Conelusions. This study adds direct observation of arrial

events to a body of data that suggests lhdt the irregularity of the ventricular response to atrial fibrillation is primarily a conseqaence of the irregularity inherent in the atrial activity. Ventricular variability and maximal RR intervals were re- lated lo atnal variability and maximal AA intervals by the scaling factor, and both the atrial and ventricular interval histograms had shapes similar to the Pokon distribution. This relation was observed in patients wi!h and without

IS Wellens HJ, Durrer D. Wolff-Parkinson-Whue syrdrome sod atrral fibril- Ianon: relatton between refrclory period of accessory palhwny and venmcular wte during alrial lihrillaian. Am J Cardiol 1974:34:777-8?.

Ih. B~lletle J. Nadeau RA, Roberge F. Relalion between the mimmum RR inbxvial durme atria! tibrdlallon and the functIonal refraclorv uenod nf Ihe AV Juncllon.?ardiovasc Res 1974:8:347-51

.

17. BIHeae J. Roberge FA. Nadeau RA. Pole> of the AV junction III delerminlng Ihe ventricular response to alrfal fibrdlalion. Can J Phyriol Phdrmitcul 1‘475;53:57!-84.

II!. Rowland E. Curry P. Fox K. Krlkler D. Relarlon between alnovenlriLular pslhway\ and venlncular rerponse durtngarrral fihnllation and Ruller Rr Heart J lY!ll:JS:HF7.

1272 KIKSH ET AL. JACC Vol. 12. No. 5 VENTRICULAR KESPONSE TO ATRIAL FIBRILLATION November 1988~1265-72

IY, Klein GJ. Ycc R. Sharma AD. Concealed conduction in accessory afriovcntricular pathways: an important dctcrminanl of the expression of arrhythmias in patients with Wollf-Parkinson-White syndrome. Circula- lmn 1984;70:402-3 I.

20. Milstein S, Klein GJ. Raltes MF. Sharma AD, Yce R. Comparison of the ventricular response during atrial fibrillation in patients with enhanced arrioventncular node conduction and Wolff-Parkinson-White syndrome. J Am Cdl Cardiol 1987;10:1244-8.

21. Meijler FL. Alrial fibrillation: a new look al an old arrhythmia (editorial). J Am Coll Cardicl l*Q’+791-3. . .,.A.._.<

22. Urbach JR. Ventn’cular response to atrial fibrillr+ion (letter). J Am Coil Cardiol 1984:3:lIOS-6.

23. Scher AM, Heethaar RM. Zimmcrmun ANE. MeiJler FL. Alrral rhylhm during ventricular fibrillation in the dog. Circ Res 1976:38:41-5.

24. Mcijler FL. Kronemau J, Van der Tweell. Herbschleb JN. Hcethaar RM. Borsl C. Nonrdndom ventricular rhythm in horses with atrial fibrillation and its significance fnr pa&Is. J Am Coll Cardiol 1984;4:316-23.

25. Wellr II,, Karp RB. Kouchoukos NT. Maclean WAH, James TN, Waldo AL. Chanctcrizaticln of atria1 fibrillation in man: sludies following open hean surgery. PACE 1976;1:426-38.

26. Slocum J. Sahakian A. Swirvn S. Comoutcr discrimination of atrial fibrillation and regular atrial-rhythms from intra-atrial electrograms. FACE lL88:ll:61C-21.

27. Ropella KM. Sahakian AV. Baerman JM. Swiryn S. The effects of procainamidc on intro-atrial electrograms during atrial fibrillation: impli- cations for detection algorithms. Circulation 1%8;77:1047-54.

28. Jackson LB. IX&al Filters and Signal Processing. Wnnvell. MA: Kluwer Academic Publishers, 1986:70.

29. Hamilton PS. Tompkins WJ. Yet another real-time QRS detection aigo- rithm. Proceedings of the Association for the Advancemcbi of Medical Instrumentation (AAMI) 2ist Annual Meeting. Chicago: Associalion for the Advancement of Medical instrumentation, 1988:36.

30. Spiegel MR. Theory and Problems of Statistics. New York: Schaum Publishing, 1%1:73.

31. Gullman I. Wdks SS, Hunter JS. Inlroductory Engineering Statistics. New York: John Wiley & Sons. 1982360.

32. Amoldi W. Die Ermittlung von dominierenden Rhythmen sowie der Schwcre der Rhythmussti+ung bei Kranken mit Arhythmia perpetua. KIin Wochenschr 1927;h:1846-8.

33. Willkamof FHM. de Joneste MJL. Lie HI. Meiiler FL. Effect of riaht ventricular pacing on ve&icular rhythm duringgtrial fibrillation. J Am Coil Cardiol 1988;11:53~5.

34. Dreifus LS. Mazgalcv T. “Atria] paralysis”: does it explain the irregular ventricular rate during atrial fibrillation (editorial)? J Am Coil Cardiol 1988:11:546-7.

35. Klein GJ. Prystowsky EN. Pritchett ELC, Davis D, Gallagher JJ. Atypical patterns of retrograde conduction over accessory atrioventricu- tar pathways in the Wolff-Parkinson-White syndrome. Circulation 1979;ho: 1477-86.

36. Ward DE. Camm AJ. Ventriculo-atrial conduction over accessory path- ways exhibiting decremental properties. Eur Heart J 1982;3:267-75,

37. Bauemfeind RA, Ayres BF, Wyndham CC, et al. Cycle length in atrioventricular nodal reentrant paroxysmal lachycardia .Gth observa- tions on the LownGanong-Levine syndrome. Am J Cardiol 1980;45: 1148-53.

38. Jackman WM. Prystowsky EN, Naccarelli GV. et al. Reevaluation of enhanced atrioventricular nodal conduction: evidence IO suggest a con- tinuum of normal alriovenlricular nodal physiology. Circulation 1983;67: 441-8.

39. Meljler FL. An “account” of digitalis and atrial fibrillation. J Am Coil Cardiol 1985;5(suppl A):6OA_8A.

40. Durrer D. Wcllcns HJ. The Wolff-Parkinson-White syndrome anno 1973. Eur J Cardiol 1974;1:347-67.

41. Bolli R, Magro SA. Wyndham CRC. Concealed antidromic re-entrance during rapid atrial pacing in the Wolff-Parkinson-While syndrome. PACE 1983;&385-91.

42. James TN. Movhology of the human atriovenlricular node, with remarks pertinent lo its electrophysiology. Am Heart J !%1;62:756-71.

43. Morady F, Scheinman MM. Winston SA, DiCado LA, Davis JC. Disso- ciation ofatrial clcctrograms by right and left atrial pacing in patients with alriovenlricular reciprocating tachycardia. J Am Coil Cardiol 1984;4: 1283-9,

44. Batchelder JE, Miles WM, Mahomed Y. et al. Presence of a broad activation front in Wolff-Parkinson-White patients suggesting arborizing atrial insertion of accessory pathways (abstr). J Am Coil Caniiol 1988:l t: I IOA.

45. Mcilcr FL, Van dcr Tweel 1. Herbschleb JN, Hauer RNW, Roblcs de Medina EO. Role of atria1 fibrillation and atrioventricular conduction (including Wolff-Parkinson-White syndrome) in sudden death. 1 Am Call Cardio! 1985;51suppl B1:178_22B.