Reliability of Postexercise Heart Rate Recovery

Transcription

1 238 Training & Testing Reliability of Postexercise Heart Rate Recovery Authors L. Bosquet 1, 2, F.-X. Gamelin 2, S. Berthoin 2 Affiliations 1 Département de Kinésiologie, Université de Montréal, Montréal, Canada 2 Faculté des Sciences du Sport et de l EP, Université Lille 2, Ronchin, France Key words l " intra-class correlation coefficient l " standard error of measurement l " minimum difference considered real reproducibility l " Abstract Passive postexercise heart rate (HR) recovery is currently used in the assessment of endurance athletes to determine changes in performance or in the clinical setting as a predictor of all-cause mortality. The purpose of this investigation was to assess the reliability of HR recovery. Thirty healthy subjects performed two maximal and two submaximal treadmill exercises, followed by 5 minutes of passive recovery. HR signal was used to compute raw and D (exercise recovery) HR after 1, 2, 3, and 5 minutes of exercise cessation. A mono-exponential function was fitted to the data using the least squares procedure. We found no significant bias between repeated measures. Relative reliability was lower for D HR when compared with raw HR (0.43 <ICC<0.71 vs < ICC < 0.83, respectively). Absolute reliability was relatively constant over time for raw HR (SEM = ~ 8%), while it decreased exponentially from the 1st (SEM = ~ 20%) to the 5th minute of recovery (SEM = ~ 8%) for D HR. The reliability of parameter estimates from exponential curve fitting was less consistent, since both ICC (0.43 to 0.88) and SEM (5.7 to 21.4%) differed from one parameter to the other according to the intensity of exercise. We conclude that passive postexercise HR recovery reliability is heterogeneous. Raw HR is the desired method to describe it. accepted after revision February 13, 2007 Bibliography DOI /s Published online July 5, 2007 Int J Sports Med 2008; 29: Georg Thieme Verlag KG Stuttgart New York ISSN Correspondence Prof. Laurent Bosquet Département de Kinésiologie Université de Montréal CP 6128, Succ. Centre Ville H3C 3J7 Montréal Canada Phone: Fax: laurent.bosquet@umontreal.ca Introduction At the end of both maximal and submaximal exercises, heart rate decreases exponentially toward resting levels [9]. The exponential kinetics is an intrinsic property of the cardiovascular system [25], and is primarily modulated by the autonomic nervous system. In fact, the rapid fall in heart rate appears to be a consequence of the prompt restoration of parasympathetic tone at the heart level, whereas the further decrease is attributed to the progressive decrease of sympathetic tone and hormonal factors [1,16, 23]. The physiological implications of this response make heart rate recovery kinetics a very convenient tool for both clinicians and sport scientists. Whereas a delayed postexercise heart rate recovery has been associated with increased all-cause mortality [7, 17, 21], a faster response reflects a positive adaptation to exercise training and possibly performance capacity in endurance events [13, 30, 33]. Altered kinetics have also been occasionally considered as a marker of overreaching or overtraining [11,14,18], although there are no data available in the literature in support of this contention. The methods used to determine heart rate recovery kinetics can differ dramatically from one study to another. Parameters which are subject to variation between protocols are the exercise mode, which can be cycling on an ergometer [17] or, most often, running on a motorized treadmill [7,8, 21, 26]; the exercise intensity, which can be maximal [5, 7] or submaximal [8,21]; the recovery duration, which vary usually from one [7] to five minutes [5], but sometimes more [33]; the recovery mode, which can be active [21] or most often passive [7, 8,17, 26]. The same diversity applies to the quantification of heart rate recovery kinetics. While exercise training studies usually report raw heart rate at a given time during the recovery period [13, 30, 33], in most clinical studies, heart rate recovery is defined as the difference between heart rate at the end of exercise and heart rate at a given time during the recovery period [5,7,8,17,21, 26]. Finally, in some studies, a mono-exponential model is fit to the heart rate response to derive global parameters of heart rate recovery kinetics such as the time constant or the asymptotic value [3,16, 23].

2 Training & Testing 239 Whatever the purpose of measuring heart rate recovery, clinical or scientific, its reliability is critically important [2]. Such diversity in both the protocols and the way to quantify its kinetics probably impact on its reliability. However, we do not know of any study examining this issue. Thus, the aim of this investigation was to assess the reliability of passive postexercise heart rate recovery kinetics on a treadmill in healthy subjects. Material and Methods Subjects Thirty physically active (3 5 aerobic training sessions per week for at least one year) and healthy subjects (22 men and 8 women) were recruited to participate in this study. Their mean (SD) age, height and body mass were 33 (10) years, 173 (6) cm and 68 (10) kg respectively. The protocol has been reviewed and approved by the Research Ethics Board in Health Sciences of the University of Montreal, Canada. Experimental design Following a thorough briefing and medical screening, all subjects signed a written statement of informed consent. Once included, they completed successively two maximal continuous graded exercise tests and two submaximal square-wave constant duration tests on a motorized treadmill (Quinton, Bothell, WA, USA), which was calibrated at 8 and 16 km h 1 before each session with an in-house system using an optical sensor connected to an acquisition card. All tests were separated by at least 72 hours and were performed at the same time of the day in a laboratory room of constant temperature (228C) and humidity (45%). To avoid any residual fatigue induced by a recent workout, subjects were asked to refrain from strenuous training the day before the tests. Exercise testing Maximal continuous graded exercise test: initial velocity was set at 10 km h 1 and increased by 0.5 km h 1 every minute until exhaustion. The grade was set at zero throughout the test. The passive recovery phase was evaluated over a 5-minute period that consisted of 1 minute in a standing position on the treadmill and 4 minutes in a seated position. The velocity of the last completed stage was considered as the peak treadmill velocity (PTV). Oxygen uptake (V O 2 ) and related gas exchange measures were determined continuously using an automated cardiopulmonary exercise system (Moxus, AEI Technologies, Naperville, IL, USA). Gas analyzers were calibrated before each test, using a gas mixture of known concentration (15% O 2 and 5% CO 2 ) and ambient air. The turbine was calibrated before each test using a motorized syringe (Vacu-Med, Ventura, CA, USA) with an accuracy of ± 1% [15]. The tidal volume was set at 3 l and the stroke rate at 40 cycles per minute. Three minutes after the completion of the test, a blood sample was taken from the fingertip to analyze for blood lactate concentration ([La ] b ) using an amperometric method with an enzyme electrode (Lactate Pro, Arkray, Kyoto, Japan). Heart rate was monitored every 5 seconds during exercise and recovery using a Polar S810 heart rate monitor (Polar Electro Oy, Kempele, Finland). Submaximal square-wave constant duration test: velocity was set at 80% of PTV for a period of 10 minutes. The passive recovery phase was evaluated over a 5-minute period that consisted of 1 minute in a standing position on the treadmill and 4 minutes in a seated position. Again, heart rate was monitored every 5 seconds during exercise and recovery using a Polar S810 heart rate monitor (Polar Electro Oy, Kempele, Finland). Data analysis Determination of maximal oxygen uptake (V O 2max ): mean values of V O 2 (in ml min 1 kg 1 ) were displayed every 30 seconds during the test. The primary criteria for the attainment of V O 2max was a plateau in V O 2 despite an increase in running velocity. In the absence of a plateau, secondary criterion included a respiratory exchange ratio of 1.10 or greater, and a postexercise [La ] b of 8 mmol l 1 or greater [10]. Determination of postexercise heart rate recovery: raw heart rate (in beats per minute, bpm) was defined as the heart rate value at a given time of the recovery period (0, 1, 2, 3, and 5 minutes). D heart rate (bpm) was defined as the difference between heart rate immediately at the end of exercise and after 1, 2, 3 and 5 minutes of passive recovery. The overall kinetics of heart rate during the transition from exercise to rest was described by the following mono-exponential function [23]: HRðtÞ ¼a 0 þ a 1 e ð t=tþ (Eq. 1) where a 0 is the asymptotic value of heart rate (bpm), a 1 is the decrement below the heart rate value at the end of exercise for t= (bpm) and t is the time constant (i.e., the time needed to reach 63% of the gain, in s). Statistical analysis Standard statistical methods were used for the calculation of means and standard deviations. Normal Gaussian distribution of the data was verified by the Shapiro-Wilk test, and homoscedascticity by a modified Levene test. Since most of data sets exhibited heteroscedasticity, a logarithmic transformation was performed before the analysis [2]. Systematic bias, which refers to a general trend for measurements to be different in a particular direction between repeated tests [2], was assessed with a paired Student s t-test. Relative reliability, which represents the degree to which individuals maintain their position in a sample with repeated measurements [2], was assessed with the intra-class correlation coefficient (ICC). Absolute reliability, which is the degree to which repeated measurements vary for individuals, was assessed with the standard error of measurements (SEM). Both the ICC and the SEM were computed from the breakdown of a two-way ANOVA (trials subjects) with repeated measures, using the following equations [32]: MS S MS E ICC ¼ MS S þðk 1ÞMS E þ kðmst MSEÞ n (Eq. 2) where MS S = mean squared subjects, MS E = mean squared error, MS T = mean squared trials, k = number of trials and n = number of subjects. p SEM ¼ ffiffiffiffiffiffiffiffiffi MS E (Eq. 3) where MS E = mean squared error. SEM can also be used to determine the minimum difference to be considered real (MD). MD represents the limit under which the observed difference is within what we might expect to see in

3 240 Training & Testing Heartrate(bpm) (t) HR = EXP 2 R = 0.98 SEE = 3.15 bpm ( t/100) (t) ( t/76) HR = EXP 2 R = 0.97 SEE = 4.29 bpm Time (s) Fig. 1 Heart rate recovery after the maximal continuous graded exercise test (filled circles) and after the submaximal square-wave constant duration test (open circles) for a typical subject, together with the best fit exponential curves. Table 1 Reliability of raw and relative heart rate during the recovery from maximal graded exercise in running Parameter Test 1 Test 2 ICC a SEM b MD c d V O 2max ± ± PTV e ± ± Raw heart rate f HR exercise 185 ± ± HR ± ± HR ± ± HR ± ± HR ± ± Relative heart rate (D) f D 1 40 ± ± D 2 71 ± ± D 3 79 ± ± D 5 84 ± 8 84 ± a Intra-class correlation coefficient; b standard error of measurement; c minimum difference to be considered real; d in ml kg 1 min 1 ; e Peak treadmill velocity, in km h 1 ; f in beats per minute repeated testing just due to the noise in the measurement, and can be calculated as follows [32]: p MD ¼ SEM 1:96 ffiffiffi 2 (Eq. 4) where SEM is the standard error of measurements computed from Eq. 3. Statistical significance was set at p < 0.05 level for all analysis. All calculations were made with Statistica 6.0 (Statsofts, Tulsa, OK, USA). Results We found no difference for V O 2max and PTV between test 1 and test 2 (l " Table 1). Both measures were highly reliable (ICC = 0.92 and 0.95, respectively; SEM = 2.11 ml kg 1 min 1 and 0.38 km h 1, respectively). The minimum difference considered as real was 5.85 ml kg 1 min 1 for V O 2max and 1.04 km h 1 for PTV (l " Table 1). Typical heart rate responses after maximal and submaximal exercise are illustrated in l " Fig. 1. Mean results and reliability measures of heart rate recovery after maximal and submaximal exercises are presented in l " Tables 1 and 2, respectively. Mean parameter estimates from exponential curve fitting and their reliability measures are presented in l " Table 3. We found no difference for all measured or calculated variables between test 1 and test 2, whatever the intensity of exercise (l " Tables 1 3). SEM expressed as a percentage of the mean response was stable over time for raw heart rate (6.65 < SEM < 8.89%), but improved with the duration of recovery for D heart rate (6.67 < SEM < 26.36%; l " Fig. 2). ICC was lower for D heart rate when compared with raw heart rate (0.43 to 0.71 and 0.68 to 0.83, respectively; l " Tables 1 and 2). The reliability of parameter estimates from exponential curve fitting was less consistent, since both ICC and SEM differed from one parameter to the other according to the intensity of exercise (l " Table 3). The minimum difference considered as real (MD) varied from to bpm for raw heart rate (l " Tables 1 and 2); from to bpm for D heart rate (l " Tables 1 and 2); from 9.70 to Table 2 Reliability of raw and relative heart rate during the recovery from submaximal square-wave exercise in running Parameter Test 1 Test 2 ICC a SEM b MD c (bpm ± SD) (bmp ± SD) Raw heart rate HR exercise 165 ± ± HR ± ± HR 2 97 ± ± HR 3 92 ± ± HR 5 87 ± ± Relative heart rate (D) d D 1 45 ± 6 45 ± D 2 68 ± ± D 3 73 ± 9 74 ± D 5 78 ± 8 78 ± a Intra-class correlation coefficient; b standard error of measurement; c minimum difference to be considered real; d computed as the difference between exercise and recovery values Table 3 Reliability of parameter estimates from exponential curve fitting of the heart rate response after maximal and submaximal exercises Parameter Test 1 Test 2 ICC a SEM b MD c Heart rate recovery kinetics from maximal exercise t (s) 82 ± ± a 0 (bmp) 96 ± ± a 1 (bmp) 105 ± ± Heart rate recovery kinetics from submaximal exercise t (s) 59 ± ± a 0 (bmp) 86 ± ± a 1 (bmp) 95 ± ± a Intra-class correlation coefficient; b standard error of measurement; c minimum difference to be considered real bpm (a 0 and a 1 ) and to s (t) for the parameter estimates from exponential curve fitting (l " Table 3).

4 Training & Testing SEM (% mean response) SEM (% mean response) Time (minutes) Time (minutes) 6 Fig. 2 Standard error of measurement (SEM) of passive postexercise heart rate recovery after maximal (left panel) or submaximal (right panel) intensity exercise. Dotted lines are raw heart rate; thick lines are D heart rate. Discussion The aim of this study was to assess the test-retest reliability of heart rate recovery. The main result was that raw heart rate is the desired method to describe the heart rate postexercise passive recovery response. Relative reliability The ICC permits estimation of the percentage of the observed score variance that is attributable to the true score variance [32]. The higher the ICC, the higher the relative reliability, and the lower the influence of measurement error. However, interpreting the magnitude of an ICC is a complex issue. The ICC may be affected by the equation used to compute it [28], but also by the variability of the data [27]. The ICC of raw or D heart rate recovery in the present study was in the range of the ICC reported for heart rate in other measurement conditions, including rest [24] or exercise [31]. A more salient fact was the moderate relative reliability of D heart rate recovery, particularly after 1 minute (0.43 < ICC < 0.58; l " Tables 1 and 2), while it is the main clinical heart rate recovery index used to predict mortality [5, 7, 8,17,21]. The fact that measurement of D heart rate recovery is influenced by the measurement error of both end-exercise and recovery heart rates probably contributes to its decreased reliability. Relative reliability of raw and D heart rate recovery does not appear to be influenced by exercise intensity and recovery duration (l " Tables 1 and 2). It should be kept in mind that recovery mode was not tested in this study, since it was passive in all tests. Additional data are needed to extend this conclusion to active recovery. The ICC of parameter estimates from exponential curve fitting of the heart rate response was in the range of raw and D measures (l " Tables 1 3). However, the results were less consistent since ICC differed from one parameter to the other and was also influenced by exercise intensity. Such heterogeneity has already been reported by Christenfeld et al. [6]. In their study, the heart rate response after 3 minutes of light exercise was fitted with a 3 parameters logistic function (a = amount of recovery; b = rate of recovery; c = recovered level). They found the ICC of these parameters to vary from 0.58 to The interaction between parameters in the nonlinear estimation process together with errors of measurement in the heart rate signal may account for this lack of consistency. Absolute reliability The SEM provides an index of the expected trial-to-trial noise in the data [32]. The lower the SEM, the higher the precision of individual scores. It should be noted that the SEM is independent of the ICC and is not affected by the variability of the data [22]. Although the SEM expressed in a percentage of the mean response was relatively constant over time for raw heart rate, we found it to be very high for D heart rate after 1 minute of recovery, and then to decrease exponentially toward raw heart rate SEM values (l " Fig. 2). In a general manner, the SEM of heart rate recovery, which is around 8% at its best in our study, is relatively high when compared with other measurement conditions. Pitzalis et al. [24] found it to be approximately 4% at rest. Sime et al. [29] reported similar values (~ 4%) when heart rate was measured during a submaximal treadmill exercise. The SEM of parameter estimates from exponential curve fitting of the heart rate response was in the same range as that for raw heart rate with a 0 (~ 5.7%) and a 1 (~ 7.5%). Surprisingly, it was almost 3 times greater for t (~ 21.3%). We are not aware of any comparison data, since Christenfeld et al. [6] did not compute the SEM. This discrepancy between parameters may originate from the nature of the heart rate response during the recovery from exercise. Since t represents the transition between two fairly stable levels (represented by a 0 and a 1 ), it is probably more susceptible to errors of measurements. Practical considerations Describing heart rate recovery as a reliable measure appears to be an oversimplification of the reality. If it is clear that D heart rate is less reliable early in recovery. Improvements in reliability, and particularly absolute reliability, are observed with increasing recovery duration. The same heterogeneity applies to parameter estimates from exponential curve fitting, since a 0 may be considered as a highly reliable measure, while t has a similar reliability to D heart rate after 1 minute of recovery. However, despite this heterogeneity in reliability, our results show that heart rate measured during the recovery from both maximal and submaximal intensity exercise is not as reliable as heart rate measured at rest [24, 29] or during exercise [29, 31]. The consequence is a crucial need for standardization. Monod [19] already highlighted some of the numerous factors that can affect heart rate measurement. Among them we find digestion, temperature, noise, infections and pharmacological or non-pharmacological

5 242 Training & Testing substances known to influence the autonomic nervous system. Minimizing the influence of these variables may decrease the error of measurement. Age [4], physical fitness [3] and occasionally gender [4] (although the influence of the is not systematic [20]) are also known to affect postexercise heart rate recovery, and should be considered in inclusion criteria. More recently, Glynn et al. [12] emphasized the fact that emotional state may influence the recovery from a stressor. The tendency of some subjects to ruminate about stressful episodes potentially decreases the reliability of heart rate during the recovery from exercise. Efforts should be made to provide a laboratory environment that helps the subjects to stay as psychologically neutral as possible during the recovery. This moderate reliability obviously influences the minimal difference needed to be considered real (MD). MD represents the limit under which the observed difference is within what we might expect to see in repeated testing just due to the noise in the measurement [32]. We found MD of raw heart rate to range from 13 to 24 bpm after maximal intensity exercise and from 13 to 17 bpm after submaximal intensity exercise (l " Tables 1 and 2). While this difference may be considered as relatively high, it is well within the ~ 40 bpm decrease in passive postexercise heart rate recovery reported by Hagberg et al. [13] in 8 moderately active subjects after 9 weeks of endurance training, and the ~ 35 bpm difference in D heart rate after 1- to 5-minutes passive postexercise recovery between our data and the data of Shetler et al. [26] in 8000 patients referred for evaluation of chest pain. The same is true for t, since the MD of ~ 21.3 seconds we computed in our study is less than the ~ 44 seconds improvement reported by Hagberg et al. [13] or the ~ 30 seconds difference observed by Bunc et al. [3] between 15 highly trained male rowers and 11 untrained male students. It is striking to note that postexercise D heart rate after 1 or 2 minutes of recovery is poorly reliable, while it is widely used in the clinical literature to prognostic the risk of cardiovascular disease events [4, 7,8,17,21, 26]. The SEM values we computed in our study after maximal intensity (~ 10 bpm) or submaximal intensity (~ 8 bpm) exercise also question the validity of the cut-off values that are often proposed to make the prognostic ( 25 bpm) [4, 7, 8,17, 21,26]. It would be hazardous to generalize our reliability data, particularly the SEM, to other populations than the subjects we tested in this study. However, they point out the urgent need to replicate such a work with patients suffering from cardiovascular diseases in order to determine the weight of random error in these cut-off values, and thus to refine prognostic criteria. Conclusion Describing passive postexercise heart rate recovery as a reliable measure is an oversimplification. Absolute and relative reliability may vary according to the duration of recovery and the way to quantify heart rate. Although moderately reliable, raw heart rate should be preferred to describe the recovery kinetics from both maximal and submaximal intensity exercises. References 1 Arai Y, Saul JP, Albrecht P, Hartley LH, Lilly LS, Cohen RJ, Colucci WS. Modulation of cardiac autonomic activity during and immediately after exercise. Am J Physiol 1989; 256: H132 H141 2 Atkinson G, Nevill AM. Statistical methods for assessing measurement error (reliability) in variables relevant to sports medicine. Sports Med 1998; 26: Bunc V, Heller J, Leso J. Kinetics of heart rate responses to exercise. J Sports Sci 1988; 6: Carnethon MR, Jacobs DR Jr, Sidney S, Sternfeld B, Gidding SS, Shoushtari C, Liu K. A longitudinal study of physical activity and heart rate recovery: CARDIA, Med Sci Sports Exerc 2005; 37: Cheng YJ, Lauer MS, Earnest CP, Church TS, Kampert JB, Gibbons LW, Blair SN. Heart rate recovery following maximal exercise testing as a predictor of cardiovascular disease and all-cause mortality in men with diabetes. Diabetes Care 2003; 26: Christenfeld N, Glynn LM, Gerin W. On the reliable assessment of cardiovascular recovery: an application of curve-fitting techniques. Psychophysiology 2000; 37: Cole CR, Blackstone EH, Pashkow FJ, Snader CE, Lauer MS. Heart-rate recovery immediately after exercise as a predictor of mortality. N Engl J Med 1999; 341: Cole CR, Foody JM, Blackstone EH, Lauer MS. Heart rate recovery after submaximal exercise testing as a predictor of mortality in a cardiovascular healthy cohort. Ann Intern Med 2000; 132: Davies CTM, Di Prampero PE, Cerretelli P. Kinetics of cardiac output and respiratory gas exchange during exercise and recovery. J Appl Physiol 1972; 32: Duncan GE, Howley ET, Johnson BN. Applicability of V O 2max criteria: discontinuous versus continuous protocols. Med Sci Sports Exerc 1997; 29: Fry RW, Morton AR, Keast D. Overtraining in athletes :an update. Sports Med 1991; 12: Glynn LM, Christenfeld N, Gerin W. The role of rumination in recovery from reactivity: cardiovascular consequences of emotional states. Psychosom Med 2002; 64: Hagberg JM, Hickson RC, Ehsani AA, Holloszy JO. Faster adjustment to and recovery from submaximal exercise in the trained state. J Appl Physiol 1980; 48: Hooper SL, McKinnon LT. Monitoring overtraining in athletes :recommendations. Sports Med 1995; 20: Huszczuk A, Whipp BJ, Wasserman K. A respiratory gas exchange simulator for routine calibration in metabolic studies. Eur Respir J 1990; 3: Imai K, Sato H, Hori M, Kusuoka H, Ozaki H, Yokoyama H, Takeda H, Inoue M, Kamada T. Vagally mediated heart rate recovery after exercise is accelerated in athletes but blunted in patients with chronic heart failure. J Am Coll Cardiol 1994; 24: Jouven X, Empana JP, Schwartz PJ, Desnos M, Courbon D, Ducimetiere P. Heart rate profile during exercise as a predictor of sudden death. New Engl J Med 2005; 352: Kuipers H, Keizer HA. Overtraining in elite athletes: review and directions for the future. Sports Med 1988; 6: Monod H. The validity of heart rate measurement in ergonomics. Ergonomics 1967; 10: Morshedi-Meibodi A, Larson MG, Levy D, O Donnell CJ, Vasan RS. Heart rate recovery after treadmill exercise testing and risk of cardiovascular disease events (The Framingham Heart Study). Am J Cardiol 2002; 90: Nishime EO, Cole CR, Blackstone EH, Pashkow FJ, Lauer MS. Heart reate recovery and treadmill exercise scores as predictors of mortality in patients referred for exercise ECG. JAMA 2000; 284: Nunnally J, Bernstein I. Psychometric Theory. New York: MacGraw-Hill, Perini R, Orizio C, Comande A, Castellano M, Beschi M, Veicsteimas A. Plasma norepinephrine and heart rate dynamics during recovery from submaximal exercise in man. Eur J Appl Physiol 1989; 58: Pitzalis MV, Mastropasqua F, Massari F, Forleo C, Di Maggio M, Passantino A, Colombo R, Di Biase M, Rizzon P. Short- and long-term reproducibility of time and frequency domain heart rate variability measurements in normal subjects. Cardiovasc Res 1996; 32: Savin W, Davidson DM, Haskell WL. Autonomic contribution to heart rate recovery from exercise in humans. J Appl Physiol 1982; 53: Shetler K, Marcus R, Froelicher VF, Vora S, Kalisetti D, Prakash M, Do D, Myers J. Heart rate recovery: validation and methodologic issues. J Am Coll Cardiol 2001; 38: Shrout PE. Measurement reliability and agreement in psychiatry. Stat Methods Med Res 1998; 7: Shrout PE, Fleiss J. Intraclass correlations: uses in assessing rater reliability. Psychol Bull 1979; 36:

6 Training & Testing Sime WE, Whipple IT, Berkson DM, MacIntyre WC, Stamler J. Reproducibility of heart rate at rest and in response to submaximal treadmill and bicycle ergometer test in middle-aged men. Med Sci Sports 1972; 4: Sugawara J, Murakami H, Maeda S, Shinya K, Matsuda M. Change in post exercise vagal reactivation with exercise training and detraining in young men. Eur J Appl Physiol 2001; 85: Washburn R, Montoye H. Reliability of the heart rate response to submaximal upper and lower body exercise. Res Quat Exerc Sport 1985; 56: Weir JP. Quantifying test-retest reliability using the intraclass correlation coefficient and the SEM. J Strength Cond Res 2005; 19: Yamamoto K, Miyachi M, Saitoh T, Yioshioka A, Onodera S. Effects of endurance training on resting and post exercise cardiac autonomic control. Med Sci Sports Exerc 2001; 33: