Pii: s0140-6736(98)08428-

H e a rt-rate turbulence after ventricular premature beats as a
predictor of mortality after acute myocardial infarction
Georg Schmidt, Marek Malik, Petra Barthel, Raphael Schneider, Kurt Ulm, Linda Rolnitzky, A John Camm, J Thomas Bigger Jr, Albert Schömig after ventricular premature beats is a very potent postinfarction risk stratifier that is independent of other B a c k g r o u n d Identification of high-risk patients after acute
known risk factors and which is stronger than other presently available risk predictors.
prophylactic therapy. The predictive accuracy of currently L a n c e t 1999; 3 5 3 : 1 3 9 0 – 9 6
used risk predictors is modest even when several factors are combined. Thus, establishment of a new powerful method for risk prediction independent of the available stratifiers is of considerable practical value.
I n t r o d u c t i o nClinical trials1 , 2 suggest that in high-risk patients with The study investigated fluctuations of sinus- ischaemic heart disease, mortality can be effectively rhythm cycle length after a single ventricular premature beat reduced by implantation of a cardioverter-defibrillator.
recorded in Holter electrocardiograms, and characterised Since the selection of high-risk patients is a crucial part the fluctuations (termed heart-rate turbulence) by two of prophylaxis, risk stratification strategies are important.
numerical parameters, termed turbulence onset and slope.
In patients surviving acute myocardial infarction, the The method was developed on a population of 100 patients predictive value of currently used risk factors, such as with coronary heart disease and blindly applied to the population of the Multicentre Post-Infarction Program (MPIP; 577 survivors of acute infarction, 75 deaths during a median follow-up of 22 months) and to the placebo v a r i a b i l i t y ,8 and mean heart rate9 is modest1 0 even when population of the European Myocardial Amiodarone Trial several predictors are combined and methodological (EMIAT; 614 survivors of acute myocardial infarction, 87 issues of such a combination solved.1 1 Establishment of a deaths during median follow-up of 21 months). Multivariate new risk predictor independent of the presently availablestratifiers is therefore of considerable practical value.
risk stratification was done with the new parameters and We describe a new method for risk stratification based on a simple expression of ventriculophasic sinus F i n d i n g s One of the new parameters (turbulence slope) was
a r r h y t h m i a ,1 2 – 1 4 namely fluctuations of sinus-rhythm cycle the most powerful stratifier of follow-up mortality in EMIAT length after a single VPB. We term such fluctuations and the second most powerful stratifier in MPIP: MPIP risk heart-rate turbulence. In low-risk patients, we observed ratio 3·5 (95% CI 2·2–5·5, p<0·0001), EMIAT risk ratio 2·7 that after a VPB, sinus shythm shows a characteristic (1·8–4·2, p<0·0001). In the multivariate analysis, low left- ventricular ejection fraction and turbulence slope were the deceleration. Such a characteristic pattern does not only independent variables for mortality prediction in MPIP occur in high-risk patients. We propose to characterise (p<0·001), whereas in EMIAT, five variables were this phenomenon by two descriptors, both of which independent mortality predictors: abnormal turbulence contain independent information on the risk ofsubsequent mortality.
onset, abnormal turbulence slope, history of previous The new risk predictors were developed in an open infarction, low left-ventricular ejection fraction, and high study with a training sample of 100 patients accumulated mean heart rate (p<0·001). In both MPIP and EMIAT, the at the medical department of the Technical University in combination of abnormal onset and slope was the most Munich and validated blind, in both univariate and powerful multivariate risk stratifier: MPIP risk ratio 3·2 (1·7–6·0, p<0·0001), EMIAT risk ratio 3·2 (1·8–5·6, populations of myocardial-infarction survivors, namely the population of the Multicentre Post-Infarction Program I n t e r p r e t a t i o n
(MPIP) study4 and in the placebo group of the EuropeanMyocardial Infarction Amiodarone Trial (EMIAT).1 5 Erste Medizinische Klinik (G Schmidt MD, P Barthel MD,
R Schneider MEng, A Schömig MD); and Institut für Medizinische
Statistik und Epidemiologie, Technischen Universität München,
München, Germany (K Ulm PhD); Department of Cardiological
100 patients with coronary artery disease (78 of whom had a Sciences, St George’s Hospital Medical School, London, UK
history of myocardial infarction and 26 a history of multiple (M Malik PhD, A J Camm MD); and Division of Cardiology, Department
infarctions) and presenting with sinus rhythm and more than of Medicine, Columbia University, NY, New York, USA (L Rolnitzky
ten VPBs per hour during 24 h Holter monitoring were used to design the method and to optimise the risk prediction power of Correspondence to: Dr Georg Schmidt, Erste Medizinische Klinik der
the new indices. Characteristics of these patients have Technischen Universität München, Ismaniger Stra␤e 22, 81675 previously been published1 6 and are listed in table 1. During a 2- year follow-up period, 17 of these patients died.
(e-mail: gschmidt@med1.med.tu-muenchen.de) In each patient, a 24 h Holter recording was obtained THE LANCET • Vol 353 • April 24, 1999 Training sample
MPIP population
EMIAT population
Data are mean (SD) or number (%) of patients.
Table 1: Patients’ characteristics
d u r i n g a stable phase of coronary artery disease (at least 3months after acute myocardial infarction). An Oxford ExcelHolter system (Oxford Instruments, Abingdon, UK) was usedto process the Holter recordings. After manual review andrevision, computer files were generated containing the durationof individual RR intervals and morphology classifications ofindividual QRS complexes (normal, supraventricular, andventricular In patients surviving follow-up, we observed a typical pattern of sinus-rhythm RR interval series following singular VPBs. TheVPBs were followed by an early acceleration and a latedeceleration of the sinus rhythm. Figure 1 A gives a typicalexample of this pattern in a long-term survivor. Theacceleration starts immediately after an ectopic beat and lastsfor only a few RR intervals. Subsequent decleration reaches amaximum between the third and seventh sinus cycle; the longestRR interval occurs usually near to the tenth cycle after a VPB.
These variations are subtle and can be recognised only aftercomputer algorithm.
In patients who died during the follow-up period, the extent of this turbulence response to VPBs was substantially smaller Figure 1: Examples of heart-rate turbulence patterns in two
patients from training sample
Numerical descriptors were investigated characterising both A: Typical acceleration-deceleration sequence of RR intervals after phases of the heart-rate turbulence (that is initial acceleration coupling interval and compensatory pause of a VPB recorded in a 64- and subsequent deceleration) with the aim of obtaining year-old woman with anterior myocardial infarction who survived duringfollow-up. B: Almost random pattern recorded in a 77-year-old man with descriptors that were independent each of the other, separated inferior myocardial infarction who suddenly died 7 months after the index patients who did and did not die during follow-up, and were predictors of mortality independent of age, left-ventricular group. The patients were survivors of a recent myocardial ejection fraction (LVEF), and other Holter-based risk factors.
infarction with LVEF of 40% or less, aged 75 years or younger, The numerical factors characterising the chronotropic free of bradyarrhythmia, and free of contraindications to response to VPBs were dichotomised into normal and abnormal amiodarone therapy. Of these, 129 patients were excluded from values. Cut-off points for the dichotomisation were determined our analyses because of atrial fibrillation, no VPB during Holter by the method of maximising the log-rank test statistic for all monitoring, or because the Holter tape was not available. The possible cut-off values within the 10–90 percentiles of each remaining 614 patients were studied (table 1). During median predictor. The approach is identical to recursive partitioning or follow-up of 21 months, 87 patients died.
the use of classification and regression trials as introduced by In the MPIP population, Holter recordings were done in the Breiman and colleagues1 7 and adapted to use for survival data second week after the index infarction; in the EMIAT by LeBlanc and Crowley1 8 in the context of Cox’s proportional- population, the recordings were obtained in the second or third week after the infarction. Initially, the Holter tapes were In each patient, LVEF was assessed either by radionuclide or processed at Columbia University, New York (MPIP data) and by EMIAT investigators. In both populations, a Laser Holter respectively). Other recognised risk factors were obtained from 8000 System (Marquette Medical System, WI, USA) was used the Holter recordings and included mean heart rate, frequency to obtain, after visual inspection and manual editing, computer of ventricular ectopathy, and 24 h heart-rate variability (HRV).
files listing RR interval duration (sampling frequency 128 Hz) HRV was expressed by the so-called HRV triangular index, an and QRS morphological classifications on a beat-to-beat basis.
established measure of global 24 h HRV that is, compared with The RR interval and beat-type files of individual MPIP and other measures, relatively insensitive to the precision with which EMIAT Holter recordings were transferred to the Technical University of Munich for the computation of characteristics ofheart-rate turbulence. The same dichotomies as derived from the training samples were applied without knowledge of The MPIP study4 enrolled 715 survivors of acute myocardial patients’ characteristics and the results provided to the infarction (age р70 years). Of these, 138 patients were excluded collaborating centres (Columbia University, New York, for the from the analysis of the heart-rate turbulence after VPBs MPIP data, and St George’s Hospital Medical School, London, because of atrial fibrillation, no VPB during Holter monitoring, for the EMIAT data) who did the survival analyses.
missing LVEF, or because of technically insufficient or missing To eliminate any possible bias, the centre in Munich has Holter recordings. The remaining 577 patients were used in this never received individual clinical data (clinical variables and study (table 1). The patients were followed up for a median of mortality) of the MPIP and EMIAT populations and the 22 months. During this period, 75 of the patients died.
collaborating centres were not aware of the principle of the The EMIAT trial1 5 randomised 743 patients into the placebo analysis and measurement involved until the statistical analyses THE LANCET • Vol 353 • April 24, 1999 Variable
Training sample
MPIP population
EMIAT population
*Due to inclusion criteria, arrhythmia on Holter was present in every patient of the training sample.
Table 2: Statistical association of risk variables with mortality
intervals immediately after compared with immediately In both MPIP and EMIAT populations, LVEF was assessed before a VPB and is termed here the turbulence onset.
by radionuclide ventriculography,4 , 1 5 mean heart rate was taken The speed of the subsequent deceleration was quantified as the mean of all sinus rhythm cycles in the Holter recordings, by the steepest regression line between the RR interval and HRV triangular index expressing the global 24 h HRV was count and duration. The corresponding factor is termed calculated from the Holter recording by previously described here the turbulence slope. In precise numerical terms, we For the purpose of multivariate analysis, age (dichtomised at <65 years v s ୑65 years), history of previous myocardial Turbulence onset is defined as the difference between infarction, LVEF (dichtomised at ୑30% v s <30%), arrhythmia the mean of the first two sinus RR intervals after a VPB sign on Holter (defined as ten or more VPBs per h or at least and the last two sinus RR intervals before the VPB one non-sustained ventricular tachycardia of three or more divided by the mean of the last two sinus RR intervals beats on the Holter recording), mean heart rate (dichotomised at >75 beats per min v s ୏75 beats per min), and HRV (HRVtriangular index dichotomised at >20 v s ୏20 units) were also assessed. Their cut-off points were based on previous risk stratification investigations.4 , 9 , 1 5 , 2 0 where RR is the i-th sinus rhythm after (i>0) the compensatory pause of the VPB or preceding (i<0) the The endpoint of the study was total mortality. Continuous and coupling interval of the VPB. For convenience, the value categorical variables were compared by the Kruskal-Wallis test of turbulence onset is expressed as a percentage. For and the ␹2 test, respectively. Kaplan-Meier survival functions instance, in figure 1A, the coupling interval of the were calculated to test the association of heart-rate turbulence ectopic is preceded by RR intervals of 1017 ms and 1014 characteristics with total mortality. The main survival analyseswere done with the Cox proportional-hazards model with a ms and its compensatory pause is followed by RR intervals of 974 ms and 963 ms. Thus, in this case, In MPIP and EMIAT datasets, sensitivity, specificity, and positive and negative predictive accuracy of follow-up mortality prediction were also evaluated for conventional and heart-rate- turbulence-based predictors of mortality (with dichotomies asa b o v e ) .
Results of all survival analyses are presented as relative risks These measurements were first performed for each with corresponding 95% CI. A significance level of 0·05 was individual singular VPB and then averaged to obtain the value characterising the patient. Positive values ofturbulence onset mean sinus rhythm deceleration after a VPB, and negative turbulence onset means sinus rhythm Of the number of possibilities tested, two factors were Turbulence slope is defined as the maximum positive selected to characterise the chronotropic response of slope of a regression line assessed over any sequence of five subsequent sinus-rhythm RR intervals within the acceleration was quantified by the relative change of RR first 20 sinus-rhythm intervals after a VPB. The value of Training sample
MPIP population
AMIAT population
Variable
Age >65 years
Combined turbulence onset 7·4 (3·1–17·3) <0·0001
and slope*
MI=myocardial infarction. *Turbulence onset у0 and turbulence slope р2·5 per RR interval; Sen=sensitivity, Spc=specificity, Ppa=positive predictive accuracy, Npa=negative predictiveaccuracy (%).
Table 3: Association of risk variables with total mortality in a univariate analysis
THE LANCET • Vol 353 • April 24, 1999 Figure 2: Kaplan-Meier survival curves in MPIP and EMIAT patients stratified to those with turbulence onset <0 and
୑0 (A, B) and
stratified to those with turbulence slope >2·5 and ୏2·5 per RR interval (C, D)
The numbers of patients of the individual groups involved in the analysis at 0, 6, 12, 18, and 24 months are shown under each graph: the top and
bottom row corresponds to the upper and bottom survival curve, respectively.
turbulence slope is expressed in ms per RR interval and second strongest univariate mortality predictor after low for each recording, it was obtained from the tachogram LVEF (table 3). Simultaneous use of turbulence onset R R , RR , RR . . . , R R , where RR is the average of i- t h and slope provided highest relative risks in both the sinus-rhythm RR interval after the compensatory pause MPIP population (5·0 [95% CI 2·8–8·8]) and the of a singular VPB. The log-rank test statistics for all EMIAT population (4·4 [2·6–7·5], table 3).
possible cut-off points revealed optimal dichotomies of 0 Figure 2 shows Kaplan-Meier cumulative survival for turbulence onset and 2·5 ms per RR interval for curves for turbulence onset and slope in MPIP and turbulence slope. In the training sample, there were EMIAT patients. In the MPIP patients, those with significant associations of turbulence onset and slope turbulence onset of less than zero had a 2-year mortality of 11% versus 20% in patients with turbulence onset ofzero or higher. In the EMIAT patients, the mortality rates were 11% and 24%, respectively. In the MPIP In univariate analyses of both MPIP and EMIAT patients, those with turbulence slope greater than 2·5 ms populations, we noted a strong and significant had a 2-year mortality of 9% versus 27% in patients association of turbulence onset and slope with total w i t h turbulence slope of 2·5 ms or less. In the EMIAT, mortality both when used as continuous variables and these mortalities were 9% and 26%, respectively.
when dichotomised at the predefined cut-off points. In T h e differences in cumulative survival were highly MPIP data, the LVEF, HRV triangular index, and turbulence slope provided the most significant difference Figure 3 shows Kaplan-Meier cumulative survival between numerical values in survivors and non-survivors; curves for the combinations of turbulence onset and in EMIAT data, the differences in numerical values of slope in MPIP and EMIAT. In the MPIP population, turbulence slope were most significant (table 2). In the 2-year mortality rates were 9%, 15%, and 32% in EMIAT, the turbulence slope was the strongest patients with both factors normal, patients with either univariate mortality predictor; in MPIP, it was the factor abnormal, and patients with both factors THE LANCET • Vol 353 • April 24, 1999 Figure 3: Kaplan-Meier survival curves in MPIP and EMIAT patients stratified to three groups
Turbulence onset <0 and turbulence slope >2·5 ms/RR interval (both factors normal); either turbulence onset ୑0 or turbulence slope ୏2·5 ms/RR
interval (one of the factors abnormal); turbulence onset ୑0% and turbulence slope ୏2·5 ms/RR interval (both factors abnormal). The numbers of
patients of the individual groups involved in the analysis at 0, 6, 12, 18, and 24 months are shown under each graph: the order of the rows corresponds
to the order of the survival curves.
abnormal, respectively. In EMIAT, these figures were p<0·0001, respectively). The relative hazard for LVEF 9%, 18%, and 34%, respectively. Again, the differences (୑30% v s <30%) was 2·9; the relative hazard of the in cumulative survival were highly significant.
turbulence onset/slope combination (slope >2·5 and In both MPIP and EMIAT, the combination of onset <0% v s slope ୏2·5 and onset ୑0) was 3·2. In turbulence onset greater than zero and turbulence slope EMIAT, four variables were independent predictors: the o f 2·5 ms per RR interval or less yielded a positive strongest predictor was the combination of turbulence predictive accuracy (33% and 31%, respectively), that onset and slope with a relative hazard of 3·2, while the was higher than the positive predictive accuracy of any other significant predictors were history of previous conventional predictor while maintaining the same level myocardial infarction, LVEF, and mean heart rate with relative hazards between 1·7 and 1·8. Patients with abnormal turbulence onset and abnormal turbulence Table 4 presents the results of the stepwise, slope are not infrequent; in the MPIP and EMIAT multivariate, Cox regression analysis with turbulence population, there were 70 (12·1%) and 80 (13·0%) such onset and slope as separate variables. In the MPIP population, LVEF and turbulence slope were the onlyindependent variables (p<0·001) and their relative hazards were almost identical (3·0 and 2·5). In EMIAT, The results of this study clearly show that heart-rate five variables were independent predictors of mortality, namely turbulence onset and slope, history of previous deceleration of sinus rhythm after a singular VPB) is a myocardial infarction, LVEF, and mean heart rate.
consistent phenomenon in low-risk patients with Table 5 presents the results of the stepwise, multivariate, Cox-regression analysis on a combination phenomenon indicates a significantly increased risk of of turbulence onset and slope. In both MPIP and subsequent mortality. The two measures for quantifying EMIAT populations, the combination of abnormal heart-rate turbulence were developed in one population turbulence onset (୑0) and an abnormal turbulence slope (୏2·5 ms per RR interval) was the strongest mortality prospectively tested with masking in two large and predictor. In the MPIP population, LVEF and the combination of turbulence onset and slope were the only postinfarction trials MPIP and EMIAT. Therefore, we believe that our analysis proves the clinical relevance of MPIP population
EMIAT population
MPIP population
EMIAT population
Table 5: Relative hazards of individual variables in a
Table 4: Relative hazards of significant and independent risk
multivariate analysis involving combination of turbulence onset
variables in a multivariate analysis
and slope
THE LANCET • Vol 353 • April 24, 1999 the new phenomenon. Because of the treatment practice manifestation of this protection may be captured when changes, more patients in EMIAT than in MPIP responding to a potentially proarrhythmic VPB. If the received thrombolysis ␤-blockers, and inhibitors of erratic or absent response to VPBs in patients with high values of turbulence onset and low values of turbulence differences show that our finding is independent of slope is a manifestation of lost antiarrhythmic protection, modern management of postinfarction patients. The the chronotropic response to VPBs might be the MPIP population was also an unselected population of mechanistic link between impaired autonomic balance postinfarction patients, whereas only patients with low LVEF were enrolled in the EMIAT trial. Consequently, The limitations of our approach have to be recognised.
the numbers of VPBs differed in these populations, but We have merely taken measures averaged over 24 h and not investigated the spontaneous variability of the Turbulence onset and slope are both predictors of chronotropic response. We have not made any detailed mortality containing information additional to each other distinction between VPBs with and without retrograde and to other established risk factors. The combination of conduction but, judging from the compensatory pauses turbulence onset and slope was a very strong risk with both present and absent heart-rate turbulence predictor in patients of the MPIP trial and of the placebo (figure 1), the phenomenon we describe is unlikely to be group of EMIAT, even when adjusted for other related solely to such a distinction. We have not established mortality predictors, such as LVEF, investigated the effect of therapy on heart-rate arrhythmia count, heart-rate variability, mean heart rate, turbulence, especially the effects of thrombolytic therapy, ␤-blockade, an ACE inhibition, which are Turbulence onset and slope in combination was by far currently frequent in patients surviving acute myocardial the strongest Holter-based risk predictor.
infarction. However, the observations made in the data It has long been known that a ventricular systole can of the MPIP study in which these therapeutic influence the rate of sinus nodal discharge, even in the absence of retrograde atrioventricular conduction. As observations are not merely a by-product of modern early as 1909, first observations of the so-called therapeutic interventions. Our method is clearly ventriculophasic sinus arrhythmia were made in inapplicable to patients without any VPBs but, as such experimental atrioventricular block.2 1 The first clinical patients are generally at low risk, this limitation is of no description was made in 1914 by Hecht,2 2 who observed practical consequence. We also do not know whether the ventriculophasic sinus arrhythmia in a child with Adams- response to several VPBs needs to be averaged, as was Stokes syndrome. Later on, ventriculophasic arrhythmia the case in this study, to obtain a sensible measure of was observed in patients with ventricular-inhibited heart-rate turbulence. Although the averaging process in p a c i n g .2 3 To our knowledge, however, only one case recordings with multiple VPBs helps to overcome the report exists on ventriculophasic sinus arrhythmia difficulties with precision of RR-interval measurement, the assessment of turbulence onset and slope depends on Various pathophysiological mechanisms have been the sampling frequency of long-term ECGs. Still, our discussed to explain the ventriculophasic mechanisms, results show that even the contemporary precision of including changes in autonomic tone,1 2 – 1 4 , 2 5 , 2 6 traction on Holter reading is sufficient for assessment of turbulence the atrium as well as atrial appendages, atrioventricular onset and slope, possibly because the precision issue junction, and the sinus nodal region,1 2 , 2 3 , 2 7 , 2 8 and transient concerns mainly patients with very few VPBs who are improvement of the blood supply to the sinus node.1 2 , 2 9 , 3 0 Some authors even speculated that the character of Despite the limitations of our approach, the masked ventriculophasic phenomena will eventually gain an tests of this study prove clearly that the absence of the characteristic heart-rate patterns after VPBs is a very Although it is plausible to expect the cardiac potent postinfarction risk stratifier that is independent of autonomic status to influence heart-rate turbulence, it is other known risk factors and is stronger than other also plausible to expect the physiological background of the turbulence to be different from that of heart-rate variability, which reflects, partly, the modulations of the G Schmidt did the conceptual design of heart-rate turbulence, designed cardiac autonomic status. Long-term, such as 24 h, the investigations of this study, supervised the testing sample and data heart-rate variability probably mostly reflects autonomic analyses. M Malik designed the investigations of this study, hadresponsibility for the evaluation of the EMIAT data, data exchange responses to environmental and external stimuli that between the centres, and supervision of the text of the manuscript.
activate a broad variety of physiological reflexes. By P Barthel and R Schneider did the computer implementation and contrast heart-rate turbulence is a phenomenon triggered maintenance of the heart-rate technology. K Ulm did statistical analyses.
L Rolnitzky and J T Bigger had responsibility for the MPIP trial data and by a minimum endogenous stimulus to which the reflex their evaluations. A J Camm had responsibility for the EMIAT data.
responses are possibly more organised and systematic.
A Schomig supervised the study overall.
This might also be the explanation why the risk- predictive power of heart-rate turbulence appears to be This study was supported in part by grants from the Bundesministerium superior to that of heart-rate variability.
für Bildung, Wissenschaft, Forschung und Technologie (13N7073/7, to The mechanisms linking the absence of heart-rate G Schmidt), and the Bund der Freunde der Technischen UniversitätMünchen (to G Schmidt).
turbulence to mortality are not obvious. Probably, theturbulence onset and slope assessment reflects specificaspects of cardiac autonomic status. The preserved vagal R e f e r e n c e s
tone is known to be antiarrhythmic3 1 , 3 2 and probably Moss AJ, Hall WJ, Cannom DS, et al. Improved survival with animplanted defibrillator in patients with coronary disease at high risk constitutes autonomic antiarrhythmic protection. Thus, for ventricular arrhythmia: Multicenter Automatic Defibrillator by measurement of the heart-rate turbulence, a direct Implantation Trial Investigators. N Engl J Med 1996; 3 3 5 : 1 9 3 3 – 4 0 .
THE LANCET • Vol 353 • April 24, 1999 Guidelines for risk stratification after myocardial infarction. A n n premature complexes and mortality risk. Pacing Clin Electrophysiol Intern Med 1997; 1 2 6 : 5 5 6 – 6 0 .
1996; 1 9 : 9 7 6 – 8 0 .
Sanz G, Castaner A, Betriu A, et al. Determinants of prognosis in 1 7 Breimann L, Friedman JH, Ohlsen RA, Stone CJ. Classification and survivors of myocardial infarction: a prospective clinical angiographic regression trees (CART). Belmont, CA: Wadsworth International study. N Engl J Med 1982; 3 0 6 : 1 0 6 5 – 7 0 .
Multicenter Postinfarctions Research Group. Risk stratification and 1 8 LeBlanc M, Crowley J. Relative risk trees for censored survival data.
survival after myocardial infarction. N Engl J Med 1983; 3 0 9 :
B i o m e t r i c s 1993; 4 8 : 4 1 1 – 2 5 .
1 9 Cox DR. Regression models and life-tables. J R Stat Soc 1972; 3 4 :
Bigger JJ, Fleiss JL, Kleiger R, Miller JP, Rolnitzky LM. The relationships among ventricular arrhythmias, left ventricular 2 0 Task Force of the European Society of Cardiology and the American dysfunction, and mortality in the 2 years after myocardial infarction.
Society of Pacing and Electrophysiology. Heart rate variability: C i r c u l a t i o n 1984; 6 9 : 2 5 0 – 5 8 .
Standards of measurement, physiological interpretation, and clinical Moss AJ, DeCamilla JJ, Davis HP, Bayer L. Clinical significance of use. C i r c u l a t i o n 1996; 9 3 : 1 0 4 3 – 6 5 .
ventricular extopic beats in the early posthospital phase of myocardial 2 1 Erlanger J, Blackman JR. Further studies in the physiology of heart infarction. Am J Cardiol 1977; 3 9 : 6 3 5 – 4 0 .
block in mammals: chronic auriculo-ventricular heart-block in the Simson MB. Use of signals in the terminal QRS complex to identify dog. H e a r t 1909; 1 : 1 7 7 .
patients with ventricular tachycardia after myocardial infarction.
2 2 Hecht AF. Das Morgani-Adams-Stokes Syndrome im Kindesalter C i r c u l a t i o n 1981; 6 4 : 2 3 5 – 4 2 .
und seine Behandlung. Wien med Wch schr 1914; 6 4 : 1 7 8 .
Kleiger RE, Miller JP, Bigger JJ, Moss AJ. Decreased heart rate 2 3 Chung EK, Jewson DV. Ventriculophasic sinus arrhythmia in the variability and its association with increase mortality after acute presence of artificial pacemaker induced ventricular rhythm.
myocardial infarction. Am J Cardiol 1987; 5 9 : 2 5 6 – 6 2 .
C a r d i o l o g y 1970; 5 5 : 6 5 – 6 8 .
Copie X, Hnatkova K, Staunton A, Fei L, Camm AJ, Malik M.
2 4 Döhlemann C, Murawski P, Theissen K, Haider M, Forster C, Predictive power of increased heart rate versus depressed left Poppl SJ. Ventriculophasische Sinusarrhythmie bei ventrikulärer ventricular ejection fraction and heart rate variability for risk Extrasystolie. Z Kardiol 1979; 6 8 : 5 5 7 – 6 5 .
stratification after myocardial infarction: results of a two-year follow-
up study. J Am Coll Cardiol 1996; 27: 2 7 0 – 7 6 .
2 5 Roth IR, Kirsch B. The mechanism of irregular sinus rhythm in auriculoventricular heart block. Am Heart J 1948; 3 6 : 2 5 7 – 7 6 .
1 0 Odemuyiwa O, Malik M, Farrell T, Bashir Y, Poloniecki J, Camm J.
Comparison of the predictive characteristics of heart rate variability 2 6 Rosenbaum M, Lepeschkin E. The effect of ventricular systole on index and left ventricular ejection fraction for all-cause mortality, auricular rhythm in auriculoventricular block. C i r c u l a t i o n 1955; 1 1 :
arrhythmic events and sudden death after acute myocardial infarction. Am J Cardiol 1991; 6 8 : 4 3 4 – 3 9 .
2 7 Kappagoda CT, Linden RJ, Saunders DA. The effect on heart rate of 1 1 Redwood SR, Odemuyiwa O, Hnatkova K, et al. Selection of distending the atrial appendages in the dog. J Physiol Lond 1972; 2 2 5 :
dichotomy limits for multifactorial prediction of arrhythmic events and mortality in survivors of acute myocardial infarction. Eur Heart J 2 8 Kappagoda CT, Linden RJ, Snow HM. A reflex increase in heart rate 1997; 1 8 : 1 2 7 8 – 8 7 .
from distension of the junction between the superior vena cava and 1 2 Parsonnet AE, Miller R. Heart block. The influence of ventricular the right atrium. J Physiol Lond 1972; 2 2 0 : 1 7 7 – 9 7 .
systole upon the auricular rhythm in complete and incomplete heart 2 9 Wenckebach KF, Winterberg H. Die unregelmäßige Herztätigkeit.
block. Am Heart J 1944; 27: 6 7 6 – 8 7 .
1 edn. Leipzig: Verlag von Wilhelm Engelmann, 1927.
1 3 Jedlicka J, Martin P. Time course of vagal effects studies in clinical 3 0 Hashimoto K, Tanaka S, Hirata M, Chiba S. Responses of the sino- electrocardiograms. Eur Heart J 1987; 8 : 7 6 2 – 7 2 .
atrial node to change in pressure in the sinus node artery. Circ Res 1 4 Skanes AC, Tang ASL. Ventriculophasic modulation of 1967; 2 1 : 2 9 7 – 3 0 4 .
atrioventricular nodal conduction in humans. C i r c u l a t i o n 1998; 9 7 :
3 1 Lown B, Verrier RL. Neural activty and ventricular fibrillation.
N Engl J Med 1976; 2 9 4 : 1 1 6 5 – 7 0 .
1 5 Julian DG, Camm AJ, Frangin G, et al. Randomised trial of effect of 3 2 Corr PB, Yamada KA, Witkowski FX. Mechanisms controlling amiodarone on mortality in patients with left-ventricular dysfunction cardiac autonomic function and their relation to arrhythmogenesis.
after recent myocardial infarction: EMIAT. L a n c e t 1997; 3 4 9 :
In: Fozzard HA, Haber E, Jennings RB, Katz AN, Morgan HE, eds.
The heart and cardiovascular system. New York: Raven Press, 1986: 1 6 Schmidt G, Morfill GE, Barthel P, et al. Variability of ventricular THE LANCET • Vol 353 • April 24, 1999

Source: http://www.prsa.eu/download/pdf/lancetxxL.pdf